text stringlengths 2k 32k |
|---|
# Emergency department use and Artificial Intelligence in Pelotas: design and baseline results
## RESUMO
### Objetivo:
To describe the initial baseline results of a population-based study, as well as a protocol in order to evaluate the performance of different machine learning algorithms with the objective of predicting the demand for urgent and emergency services in a representative sample of adults from the urban area of Pelotas, Southern Brazil.
### Methods:
The study is entitled “Emergency department use and Artificial Intelligence in PELOTAS (RS) (EAI PELOTAS)” (https://wp.ufpel.edu.br/eaipelotas/). Between September and December 2021, a baseline was carried out with participants. A follow-up was planned to be conducted after 12 months in order to assess the use of urgent and emergency services in the last year. Afterwards, machine learning algorithms will be tested to predict the use of urgent and emergency services over one year.
### Results:
In total, 5,722 participants answered the survey, mostly females ($66.8\%$), with an average age of 50.3 years. The mean number of household people was 2.6. Most of the sample has white skin color and incomplete elementary school or less. Around $30\%$ of the sample has obesity, $14\%$ diabetes, and $39\%$ hypertension.
### Conclusion:
The present paper presented a protocol describing the steps that were and will be taken to produce a model capable of predicting the demand for urgent and emergency services in one year among residents of Pelotas, in Rio Grande do Sul state.
## Objetivo:
Descrever os resultados iniciais da linha de base de um estudo de base populacional, bem como um protocolo para avaliar o desempenho de diferentes algoritmos de aprendizado de máquina, com o objetivo de predizer a demanda de serviços de urgência e emergência em uma amostra representativa de adultos da zona urbana de Pelotas, no Sul do Brasil.
## Métodos:
O estudo intitula-se “Emergency department use and Artificial Intelligence in PELOTAS (RS) (EAI PELOTAS)” (https://wp.ufpel.edu.br/eaipelotas/). Entre setembro e dezembro de 2021, foi realizada uma linha de base com os participantes. Está previsto um acompanhamento após 12 meses para avaliar a utilização de serviços de urgência e emergência no último ano. Em seguida, serão testados algoritmos de machine learning para predizer a utilização de serviços de urgência e emergência no período de um ano.
## Resultados:
No total, 5.722 participantes responderam à pesquisa, a maioria do sexo feminino (66,$8\%$), com idade média de 50,3 anos. O número médio de pessoas no domicílio foi de 2,6. A maioria da amostra tem cor da pele branca e ensino fundamental incompleto ou menos. Cerca de $30\%$ da amostra estava com obesidade, $14\%$ com diabetes e $39\%$ eram hipertensos.
## Conclusão:
O presente trabalho apresentou um protocolo descrevendo as etapas que foram e serão tomadas para a produção de um modelo capaz de prever a demanda por serviços de urgência e emergência em um ano entre moradores de Pelotas, no estado do Rio Grande do Sul.
## INTRODUCTION
Chronic diseases affect a large part of the population of adults and older adults, leading these individuals to seek urgent and emergency care. The implementation in 1988 of the Unified Health System (SUS) resulted in a model aimed at prevention and health promotion actions based on collective activities 1 – starting at Basic Health Units (UBS). There is also the National Emergency Care Policy, which advanced in the construction of the SUS, and has as guidelines universality, integrity, decentralization, and social participation, alongside humanization, the right of every citizen 2.
In a study that evaluated the characteristics of users of primary health care services in a Brazilian urban-representative sample, it was found that the vast majority were women and part of poorer individuals, in addition to almost $\frac{1}{4}$ of the sample receiving the national income distribution program (family allowance) 3. Brazil is a country highly unequal in socioeconomic terms; approximately $75\%$ of the Brazilian population uses the SUS and depends exclusively on it, and do not have private health insurance 4,5.
Individuals with multimorbidity are part of the vast majority who seek urgent and emergency services 6. Multimorbidity is a condition that affects a large part of the population 7, especially older adults 7. In addition, the association of multimorbidity with higher demand for emergency services is a challenge to appropriately manage and prevent these problems 8,9.
Innovative approaches may allow health professionals to provide direct care to individuals who are more likely to seek urgent and emergency services. The use of artificial intelligence can make it possible to identify and monitor a group of individuals with a higher probability of developing multimorbidity. In this context, machine learning (ML), an application of artificial intelligence, is a promising and feasible tool to be used on large scale to identify these population subgroups. Some previous studies have demonstrated that ML models can predict the demand for urgent and emergency services 10,11. Besides, a systematic review showed that ML could accurately predict the triage of patients entering emergency care 12. However, in a search for studies in Brazil, we found no published article on the subject.
In Brazil, urgent and emergency services are a fundamental part of the health care network, ensuring timely care in cases of risk to individuals’ lives 9. Urgent and emergency services are characterized by overcrowding and high demand. In addition, with the current pandemic of COVID-19, updated evidence on the characteristics of the users seeking these services is timely and necessary. The objective of this article was to describe the initial baseline results of a population-based study, as well as a protocol in order to evaluate the performance of different ML algorithms with the objective of predicting the demand for urgent and emergency services in a representative sample of adults from the urban area of Pelotas.
## METHODS
The present cohort study is entitled “Emergency department use and Artificial Intelligence in PELOTAS-RS (EAI PELOTAS)” (https://wp.ufpel.edu.br/eaipelotas/). The baseline was conducted between September and December 2021, and a follow-up was planned to be conducted 12 months later. We utilized the cross-sectional study to measure the prevalence of urgent and emergency care and the prevalence of multimorbidity, in addition to other variables and instruments of interest. The prospective cohort design intends to estimate the risk of using and reusing urgent emergency services after 12 months. Contact information, collected to ensure follow-up, included telephone, social networks, and full address. In addition, we also collected the latitude and longitude of households for control of the interviews.
## Study location and target population
The present study was conducted in adult households in the Pelotas, Rio Grande do Sul (RS), Southern Brazil. According to estimates by the Brazilian Institute of Geography and Statistics (IBGE) in 2020, Pelotas had an estimated population of 343,132 individuals (https://cidades.ibge.gov.br/brasil/rs/pelotas/panorama). Figure 1 shows the location of the city of Pelotas in Brazil.
**Figura 1.:** *Map of Brazil highlighting the city of Pelotas (RS).*
Pelotas has a human development index (HDI) of 0.739 and a gross domestic product per capita (GDP) of BRL 27,586.96 (https://www.ibge.gov.br/cidades-e-estados/rs/pelotas.html). The municipality has a Municipal Emergency Room that operates 24 hours a day, seven days a week, and serves about 300 patients a day, according to data provided by the unit.
## Criteria for inclusion and exclusion of study participants
We included adults aged 18 years or older residing in the urban area of Pelotas. Children and individuals who were mentally unable to answer the questionnaire were not included in the sample.
## Sample calculation, sampling process, and data collection
The sample size was calculated considering three objectives. First, to determine the sample size required to assess the prevalence of urgent and emergency services use, it was considered an estimated prevalence of $9\%$, with±two percentage points as a margin of error and a $95\%$ confidence level 13, concluding that 785 individuals would be necessary. Second, for multimorbidity prevalence, an estimated prevalence of $25\%$, with ± three percentage points as a margin of error and a confidence level of $95\%$ was used 14,15; reaching again, a total of 785 individuals needed. Finally, for the association calculations, similar studies in Brazil were assessed, and the following parameters were considered: significance level of $95\%$, power of $80\%$, exposed/unexposed ratio of 0.1, percentage of the outcome in the unexposed $20\%$, and a minimum prevalence ratio of 1.3. With these parameters, 5,104 individuals would be necessary to study the proposed associations. Adding 10 to $20\%$ for losses and/or refusals, the final sample size would be composed of 5,615–5,890 participants.
The process to provide a population-based sample was carried out in multiple stages. The city of Pelotas has approximately 550 census tracts, according to the last update estimates provided by IBGE in 2019. From there, we randomly selected 100 sectors. Since the sectors vary in size, we defined a proportional number of households for each.
Thus, it was estimated that, in total, the 100 sectors had approximately 24,345 eligible households. To interview one resident per household, we divided the total number of households by the sample size required, which resulted in 4.3. Based on this information, we divided each of the 100 sectors by 4.3 to reach the necessary number of households for each sector. One resident per household was interviewed, resulting in a total of 5,615 households. If there was more than one eligible resident, the choice was made by a random number generator application. Residents were placed in order, a number was assigned for each one, and one of them was selected according to the result of the draw. The first household interviewed in each sector was selected through a draw, considering the selected jump (4.3 households). Trades and empty squares were considered ineligible, and thus, the next square was chosen. Due to a large number of empty houses, it was necessary to select another 50 sectors to complete the required sample size. The additional households were drawn according to the same methodological criteria as the first draw to ensure equiprobability.
## Data collection instrument
We collected the data with the Research Electronic Data Capture (REDCap), a data collection program using smartphones 16,17. Experienced and trained research assistants collected the data. The questionnaire from EAI PELOTAS was prepared, when possible, based on standardized instruments, including questions about chronic diseases, physical activity, food security, use of urgent and emergency services, functional disability, frailty syndrome, self-perception of health, COVID-19, in addition to sociodemographic and behavioral questions. Supplementary Table 1 shows the instruments utilized in the present study.
**Table 1.**
| Characteristics | EAI PELOTAS* | EAI PELOTAS*.1 | PNS 2019† |
| --- | --- | --- | --- |
| Characteristics | Crude % (95%CI) | Survey design % (95%CI) | % (95%CI) |
| Mean age, years | 50.3 (49.9–50.8) | 46.2 (45.5–47.0) | 46.7 (45.9–47.5) |
| Mean number of household people | 2.6 (2.5–2.7) | 2.7 (2.6–2.8) | 3.0 (2.9–3.1) |
| Female (%) | 66.8 (65.6–68.0) | 54.2 (52.4–55.6) | 54.1 (51.7–56.4) |
| Skin color (%) | Skin color (%) | Skin color (%) | Skin color (%) |
| White | 78.2 (77.1–79.2) | 77.3 (74.9–79.5) | 76.8 (74.6–78.7) |
| Black | 15.0 (14.1–16.0) | 15.3 (13.5–17.3) | 8.3 (7.0–9.8) |
| Brown | 6.1 (5.5–6.7) | 6.7 (5.7–7.9) | 14.5 (12.9–16.3) |
| Other | 0.7 (0.5–1.0) | 0.7 (0.4–1.1) | 0.4 (0.2–0.8) |
| Schooling (%) | Schooling (%) | Schooling (%) | Schooling (%) |
| Incomplete elementary school or less | 35.7 (34.5–37.0) | 31.3 (28.6–34.2) | 30.2 (28.1–32.4) |
| Complete elementary school/incomplete high school | 16.2 (15.3–17.2) | 16.4 (15.1–17.7) | 15.7 (14.0–17.5) |
| Complete high school/incomplete higher education | 33.5 (32.3–34.7) | 37.6 (35.6–39.6) | 36.9 (34.6–39.2) |
| Complete higher education or more | 14.6 (13.7–15.5) | 14.7 (12.4–17.4) | 17.2 (15.7–18.9) |
## Dependent variables
The use of urgent and emergency services was assessed on a baseline using the following question: “In the last 12 months, how many times have you sought urgent and emergency services, such as an emergency room?”. This was followed by the characterization of the service used, city of service, frequency of use, and referral after use. One year after the study baseline, we will contact again the respondents to inquire about the use of urgent and emergency care services (number of times and type of service used).
## Independent variables
We assessed multimorbidity as the main exposure using a list of 22 chronic diseases and others (asthma/bronchitis, osteoporosis, arthritis/arthrosis/rheumatism, hypertension, diabetes, cardiac insufficiency, pulmonary emphysema/chronic obstructive pulmonary disease, acute kidney failure, Parkinson’s disease, prostate disease, hypo/hyperthyroidism, glaucoma, cataract, Alzheimer’s disease, urinary/fecal incontinence, angina, stroke, dyslipidemia, epileptic fit/seizures, depression, gastric ulcer, urinary infection, pneumonia, and the flu). The association with urgent and emergency services will be performed with different cutoff points, including total number, ≥2, ≥3, and combinations of morbidities. We will also perform network analyzes to assess the pattern of morbidities.
Other independent variables were selected from previous studies in the literature 18-21, including demographic, socioeconomic information, behavioral characteristics, health status, access, use and quality of health services.
## Data analysis
We will test artificial intelligence algorithms, ML, to predict the use of urgent and emergency services after 12 months. The purpose of ML is to predict health outcomes through the basic characteristics of the individuals, such as sex, education, and lifestyle. The algorithms will be trained to predict the occurrence of health outcomes, which will contribute to decision-making. With a good amount of data and the right algorithms, ML may be able to predict health outcomes with satisfactory performance.
The area of ML in healthcare has shown rapid growth in recent years, having been used in significant public health problems such as diagnosing diseases and predicting the risk of adverse health events and deaths 22-24. The use of predictive algorithms aims to improve health care and support decision-making by health professionals and managers. For the present study, individuals’ baseline characteristics will be used to train popular ML algorithms such as Support Vector Machine (SVM), Neural Networks (ANNs), Random Forests, Penalized Regressions, Gradient Boosted Trees, and Extreme Gradient Boosting (XGBoost). These models were chosen based on a previous review in which the authors identified the most used models in healthcare studies 25. We will use the Python programming language to perform the analyzes.
To test the predictive performance of the algorithms in new unseen data, individuals will be divided into training ($70\%$ of patients, which will be used to define the parameters and hyperparameters of each algorithm) and testing ($30\%$, which will be used to test the predictive ability of models in new data).
We will also perform all the preliminary steps to ensure a good performance of the algorithms, especially those related to the pre-processing of predictor variables, such as the standardization of continuous variables, separation of categorical predictors with one-hot encoding, exclusion of strongly correlated variables, dimension reduction using principal component analysis and selection of hyperparameters with 10-fold cross-validation. Different metrics will evaluate the predictive capacity of the models, the main one being the area under the receiver operating characteristic (ROC) curve (AUC). In a simplified way, the AUC is a value that varies from 0 to 1, and the closer to 1 the better the model’s predictive capacity 26. The other metrics will be F1-score, sensitivity, specificity, and accuracy. As measures of model fit, we will perform hyperparameters and balancing fit, as well as K-fold (cross-validation).
## COVID-19
The current pandemic, caused by the SARS-CoV-2 virus, has brought uncertainty to the world population. Although vaccination coverage is already high in large parts of the population, the arrival of new variants and the lack of other essential measures to face the pandemic still create uncertainty about the effects of the pandemic on people. General questions about symptoms, tests, and possible effects caused by coronavirus contamination were included in our baseline survey. We will also use SARS-CoV-2-related questions to evaluate the performance of ML algorithms. In September 2021, restrictive measures were relaxed due to a decrease in COVID-19 cases in Pelotas, allowing the study to begin. A vaccination passport was required from the interviewers to ensure the safety of both participants and interviewers. In addition, all interviewers received protective equipment against COVID-19, including masks, face shields, and alcohol gel. Finally, the interviewers were instructed to conduct the research in an open and airy area, ensuring the protection of the participants.
## Quality assurance and control
The activities to allow for control and data quality were characterized by a series of measures aimed at ensuring results without the risk of bias. Initially, we developed a research protocol, followed by an instruction manual for each interviewer. Thereafter, interviewers were trained and standardized in all necessary aspects.
REDCap was also important to garanteee the control and quality of responses as the questions were designed using validation checks according to what was expected for each answer. Another measure that ensured the control of interviews was the collection of latitude and longitude of households, which was plotted by two members of the study coordination weekly on maps, to ensure that the data collection was performed according to the study sample. With latitude and longitude data, it is also intended to carry out spatial analysis articles with techniques such as sweep statistics and Kernel.
The database of the questions was checked daily to find possible inconsistencies. Finally, two members of the study coordination made random phone calls to $10\%$ of the sample, in which a reduced questionnaire was applied, with the objective of comparing the answers with the main questionnaire.
## Ethical principles
We carried out this study using free and informed consent, as determined by the ethical aspects of Resolution No. $\frac{466}{2012}$ of the National Council of the Ministry of Health and the Code of Ethics for Nursing Professionals, of the duties in Chapter IV, Article 35, 36 and 37, and the prohibitions in chapter V, article 53 and 54. After identifying and selecting the study participants, they were informed about the research objectives and signed the Informed Consent Form (ICF). The project was referred to the Research Ethics *Committee via* the Brazilian platform and approved under the CAAE 39096720.0.0000.5317.
## Schedule
Initially, we conducted a stage for the preparation of an electronic questionnaire at the beginning of 2021. In February 2021, we initiated data collection after preparing the online questionnaire. The database verification and cleaning steps occurred simultaneously with the collection, and continued until March 2022. After this step, data analysis and writing of scientific articles began.
## First descriptive results and comparison with a population-based study
Of approximately 15,526 households approached, 8,196 were excluded — 4,761 residents were absent at the visit, 1,735 were ineligible, and 1,700 were empty (see Figure 2). We identified 7,330 eligible participants, of which 1,607 refused to participate in the study, totalizing 5,722 residents. Comparing the female gender percentage of the refusals with the completed interviews, we observed a slightly lower prevalence with $63.2\%$ ($95\%$CI 60.7–65.5) among the refusals, and $66.8\%$ ($95\%$CI 65.6–68.0) among the complete interviews. The mean age was similar between participants who agreed to participate (50.3; $95\%$CI 49.9–50.8) and those who refused (50.4; $95\%$CI 49.0–51.9).
**Figura 2.:** *Flowchart describing the sampling process.*
To evaluate the first descriptive results of our sample, we compared our results with the 2019 Brazilian National Health Survey (PNS) database. The PNS 2019 was collected by the IBGE in partnership with the Ministry of Health. The data are in the public domain and are available in the IBGE website (https://www.ibge.gov.br/). To ensure the greatest possible comparability between studies, we used only residents of the urban area of the state of Rio Grande do Sul, aged using the command svy from Stata, resulting in 3,002 individuals (residents selected to interview).
We developed two models to compare our data with the PNS 2019 survey: Crude model (crude results from the EAI PELOTAS study, without considering survey design estimates); Model 1 using survey design: primary sampling units (PSUs) using census tracts as variables and post-weight variables based on estimates of Pelotas population projection for 2020 (Table 1). We evaluated another model using individual sampling weight (i.e., the inverse of the probability of being interviewed in each census tract). These models are virtually equal to the above estimates (data not shown).
The mean age of our sample was 50.3 years (Table 1), 46.2 for model 1, which was similar to PNS 2019 (46.7 years). Our weighted estimates presented a similar proportion of females compared to the PNS 2019 sample. The proportions of skin colors were similar in all categories and models. Our crude model presented a higher proportion of participants with incomplete elementary school or less compared to model 1 and PNS 2019.
Table 2 describes the prevalence of chronic diseases and lifestyle factors in our study and the PNS 2019 sample. Our prevalence of diabetes was higher in the crude model compared to weighted estimates and PNS 2019 sample. In both models, we had a higher proportion of individuals with obesity and hypertension than in PNS 2019. Asthma and/or bronchitis presented similar proportions in our results compared to PNS 2019; the same occurred for cancer. Our study presented a higher proportion of smoking participants in both models than in the PNS 2019 sample.
**Table 2.**
| Chronic diseases and lifestyle factors | EAI PELOTAS* | EAI PELOTAS*.1 | PNS 2019† |
| --- | --- | --- | --- |
| Chronic diseases and lifestyle factors | Crude | Survey design 1 | PNS 2019† |
| Chronic diseases and lifestyle factors | % (95%CI) | % (95%CI) | % (95%CI) |
| Diabetes | 14.2 (13.3–15.1) | 11.5 (10.6–12.4) | 9.0 (8.9–11.1) |
| Obesity | 30.4 (29.2–31.7) | 29.2 (27.7–30.8) | 24.8 (22.6–27.1) |
| Hypertension | 39.0 (37.7–40.3) | 32.4 (31.0–33.9) | 28.1 (25.9–30.5) |
| Asthma or chronic bronchitis | 9.3 (8.6–10.1) | 9.3 (8.4–10.4) | 8.7 (7.3–10.3) |
| Cancer | 4.2 (3.7–4.7) | 3.4 (2.9–4.0) | 3.8 (2.9–4.9) |
| Current smoking | 20.6 (19.6–21.7) | 20.4 (18.9–22.0) | 16.3 (14.6–18.1) |
## DISCUSSION
We described the initial descriptive results, methodology, protocol, and the steps required to perform the ML analysis for predicting the use of urgent and emergency services among the residents of Pelotas, Southern Brazil. We expect to provide subsidies to health professionals and managers for decision-making, helping to identify interventions targeted at patients more likely to use urgent and emergency services, as well as those more likely to develop multimorbidity and mortality. We also expect to help health systems optimize their space and resources by directing human and physical capital to those at greater risk of developing multiple chronic diseases and dying. Recent studies in developed countries have found this a feasible challenge with ML 21,27. If our study presents satisfactory results, we intend to test its practical applicability and acceptance to assist health professionals and managers in decision-making in emergency services among residents of Pelotas.
The baseline and methods used to select households resemble the main population-based studies conducted in Brazil, such as the Brazilian Longitudinal Study of Aging (ELSI-Brazil) 28, the EPICOVID 29, and the PNS. The applicability of ML requires suitable predictive variables. Our study included sociodemographic and behavioral variables related to urgent and emergency services, and chronic diseases. EAI PELOTAS study also includes essential topics that deserve particular importance during the COVID-19 pandemic, such as food insecurity, decreased income, physical activity, access to health services, and social support.
We also presented one weighting option in order to obtain sample estimates considering the complex study design. All estimates have their strength and limitation. Each research question answered through this study may consider these possibilities and choose the most suitable one. The estimates were similar without weighting and those considering the primary sampling unit (PSU) and sampling weight. Using the census tract in the PSU is fundamental to consider the sampling design in the estimates of variability (standard error, variance, $95\%$CI, among others). In addition, due to the possible selection bias in the sample, which contains more women and older people than expected, the use of a post-weighting strategy becomes necessary to obtain estimates adjusted for the sex and age distributions of the target population (due to the lack of census data, we used population projections). However, it should be noted that this strategy can produce estimates simulating the expected distribution only by sex and age. Still, we do not know how much this strategy can distort the estimates since the demographic adjustment cannot reproduce adjustment in all sample characteristics, especially for non-measured variables that may have influenced the selection of participants. Thus, we recommend defining the use of each strategy on a case-by-case basis, depending on the objective of the scientific product. Finally, we suggest reporting the different estimates according to the sample design for specific outcomes (e.g., the prevalence of a specific condition) that aim to extrapolate the data to the target population (adults of the city of Pelotas).
In conclusion, the present article presented a protocol describing the steps that were and will be taken to produce a model capable of predicting the demand for urgent and emergency services in one year among residents in Pelotas (RS), Southern Brazil. |
# Alterations in Fecal Microbiota Linked to Environment and Sex in Red Deer (Cervus elaphus)
## Abstract
### Simple Summary
The gut microbiota forms a complex microecosystem in vertebrates and is affected by various factors. Wild and captive red deer currently live in the same region but have vastly different diets. In this study, the 16S rRNA sequencing technology was performed to evaluate variations in the fecal microbiota of wild and captive individuals of both sexes of red deer. It was found that the composition and function of fecal microbiota in wild and captive environments were significantly different. As a key intrinsic factor, sex has a persistent impact on the formation and development of gut microbiota. Overall, this study reveals differences in the in the fecal microbiota of red deer based on environment and sex. These data could guide future applications of population management in red deer conservation.
### Abstract
Gut microbiota play an important role in impacting the host’s metabolism, immunity, speciation, and many other functions. How sex and environment affect the structure and function of fecal microbiota in red deer (Cervus elaphus) is still unclear, particularly with regard to the intake of different diets. In this study, non-invasive molecular sexing techniques were used to determine the sex of fecal samples from both wild and captive red deer during the overwintering period. Fecal microbiota composition and diversity analyses were performed using amplicons from the V4–V5 region of the 16S rRNA gene sequenced on the Illumina HiSeq platform. Based on Picrust2 prediction software, potential function distribution information was evaluated by comparing the Kyoto Encyclopedia of Genes and Genome (KEGG). The results showed that the fecal microbiota of the wild deer (WF, $$n = 10$$; WM, $$n = 12$$) was significantly enriched in Firmicutes and decreased in Bacteroidetes, while the captive deer (CF, $$n = 8$$; CM, $$n = 3$$) had a significantly higher number of Bacteroidetes. The dominant species of fecal microbiota in the wild and captive red deer were similar at the genus level. The alpha diversity index shows significant difference in fecal microbiota diversity between the males and females in wild deer ($p \leq 0.05$). Beta diversity shows significant inter-group differences between wild and captive deer ($p \leq 0.05$) but no significant differences between female and male in wild or captive deer. The metabolism was the most important pathway at the first level of KEGG pathway analysis. In the secondary pathway of metabolism, glycan biosynthesis and metabolism, energy metabolism, and the metabolism of other amino acids were significantly different. In summary, these compositional and functional variations in the fecal microbiota of red deer may be helpful for guiding conservation management and policy decision-making, providing important information for future applications of population management and conservation.
## 1. Introduction
Red deer (Cervus elaphus), which belong to the family Cervidae, order Artiodactyla, distributed in Asia, Europe, North America, and North Africa [1]. The red deer is a typical forest-inhabiting mammal in northeast China and has an important ecological status in the forest ecosystem [2]. Owing to habitat fragmentation, the populations of red deer in the wild are currently in sharp decline [2]. Using captive populations as reintroduction resources is an effective strategy to restore the populations of wild red deer [3].
The complex gut microbiota systems in the mammalian gut are composed of large fractions of microbes [4]. The gut microbiota are a complex product of the long-term evolution of hosts and microbes [4]. Recent studies have shown that not only are gut microbiota a part of the host, but they also have a significant impact on the health of the host, such as promoting immunity, digestion, metabolism, and intestinal endocrine hormones, among others [5,6,7]. Simultaneously, the complex and flexible gut microbiota can be affected by multiple environmental and host genotypes [8]. Many studies have shown that diet is an important factor that affects the structure and function of the fecal microbiota [9,10,11]. For example, changes in diet alter the function and diversity of fecal microbiota as well as the relative abundance of some microorganisms [12]. Moreover, diet-induced loss of microbial function and diversity will increase the risk of diversity loss and extinction through generational amplification [13]. It was necessary to investigate the gut microbiome by comparing differences between wild and captive red deer. However, to date, there has been a lack of studies comparing the gut microbiota between wild and captive red deer [11]. Because of sex differences in behavior and physiology, sex as an important intrinsic factor leads to differences in gut microbiota among individuals within species [14,15,16]. Although the results are inconsistent, animal species with significant sexual dimorphism and human studies have shown sex-related differences in gut microbiota. In mice (Mus musculus), poultry, and forest musk deer (Moschus berezovskii), the composition of the gut or fecal microbiota shows sex differences [17,18,19]. At present, few studies have analyzed the sexual dimorphism of fecal microbiota in red deer.
In order to save endangered populations, artificial breeding of wild populations is carried out. The food types and nutrient intake ratios obtained in captivity and wild environments are very different, especially for endangered cervidae [20]. Therefore, monitoring the digestive system of captive animals and identifying standardized levels of nutritional requirements and fiber composition is critical for captive wild animals to determine whether they have acclimated to artificially provided food and new environments—a part of wildlife conservation’s main problem [21]. Using captive populations as reintroduction resources is an effective strategy to restore the populations of wild red deer. The composition of gut microbiota in wild populations can be a good indicator of the breeding direction of the captive population [9]. Therefore, understanding the impact of dietary differences between wild and captive red deer on the fecal microbiota can help to assess and ensure the long-term viability of this species [9]. At present, the research methods for fecal microbiota have also shifted from traditional methods to 16S rRNA gene sequencing technology, from simple microbial composition, community structure, and core microbiota research to microbial function research, which has become a hot frontier in ungulate research today [22].
The main goal of this study was to characterize the composition of the fecal microbiota of red deer of different sex and feeding plus environment. We used high-throughput 16S rRNA sequencing technology to comprehensively analyze. Thus, we hypothesized that: [1] the fecal microbiota composition and function are different between wild and captive deer; and [2] under the wild or captive environment, the microbiota diversity and evenness are different between females and males.
## 2.1. Study Site, Subjects, and Sample Collection
This study was conducted at the Gaogestai National Nature Reserve in Chifeng, Inner Mongolia (119°02′30″, 119°39′08″ E; 44°41′03″, 45°08′44″ N). The total area is 106,284 hm2. It is a typical transition zone forest-steppe ecosystem in the southern foothills of Greater Khingan Mountains, including forests, shrubs, grasslands, wetlands, and other diverse ecosystems. In February 2019, 75 line transects were randomly laid in the Gogestai protection area. Positive and reverse footprint chain tracking was carried out after the foodprints of red deer were found through line transect investigation. Disposable PE gloves were worn to collect red deer feces. While tracking the footprint chain, set 2 m × 2 m plant quadrate every 200 m to 250 m along the footprint chain, and collect all kinds of plant branches eaten by deer in the quadrate as far as possible [23]. A total of 162 fecal samples were collected and stored at −20 °C within 2 h. The feces of red deer from different areas of the Reserve were identified as coming from different individuals, and 43 feces were identified individually in the laboratory.
In February 2019, the HanShan Forest Farm in Chifeng City, Inner Mongolia, China (adjacent to the Gaogestai Nature Reserve) had a total of 11 healthy adult red deer of similar age and size. Ear tags were used to differentiate each individual red deer. Through continuous observation, feces were collected immediately after excretion by different red deer individuals and stored at −20 °C. We measured crude protein, energy, neutral detergent fiber (NDF), and total non-structural carbohydrates in red deer diets.
## 2.2. Individual Recognition and Sex Identification
We used a qiaamp DNA Fecal Mini-kit (QIAGEN, Hilden, Germany) to extract host deoxyribonucleic acid (DNA) from the fecal samples of red deer as previously described [24]. Microsatellite PCR technology was used with nine pairs of microsatellite primers (BM848, BMC1009, BM757, T108, T507, T530, DarAE129, BM1706, and ILST0S058) [25,26] with good polymorphism that were selected based on the research results of previous studies. These nine pairs of primers can amplify fecal DNA stably and efficiently. A fluorescence marker (TAMRA, HEX, or FAM) was added to the 5′ end of upstream primers at each site (Supplementary Table S1). Primer information, PCR amplification, and genotype identification procedures are described in the literature [27]. Multi-tube PCR amplification was used for genotyping [28], and 3–4 positive amplifications were performed for each locus to determine the final genotype [29]. The excel microsatellite toolkit [30] was used to search for matching genotypes from the data. Samples are judged to be from the same individual if all loci have the same genotype or if only one allele differs at a locus. The microsatellite data were analyzed by Cervus 3.0 software, and the genotyping was completed [31].
Male and female individuals were identified by detecting the existence of genes after the individual identification of red deer was completed. *Sry* gene primers (F:5′-3′ TGAACGCTTTCATTGTGTGGTC; R:5′-3′ GCCAGTAGTCTCTGTGCCTCCT) were designed, and the amplification system was determined. To minimize the occurrence of false positives or false negatives that could affect results, the *Sry* gene was repeated three times to expand and increase during the experiment, and samples with target bands that appeared on the second and third occasions were determined to be male [32].
## 2.3. Fecal Microbiota DNA Extraction, Amplification, and Sequencing
The total microbial DNA of fecal samples was extracted using an E.Z.N.A® Soil DNA Kit (Omega Bio-Tek, Norcross, GA, USA). The DNA integrity of the extracted samples was determined by $1\%$ agarose gel electrophoresis. Targeting a 420 bp fragment encompassing the V4-V5 region of the bacterial 16S ribosomal RNA gene was amplified by PCR using primers 515F (5′-GTG CCA GCM GCC GCG GTA A-3′) and 907R (5′-CCG TCA ATT CMT TTR AGT TT-3′). NEB 154 Q5 DNA high-fidelity polymerase (NEB, Ipswich, MA, USA) was used in PCR amplifications (Supplementary Table S1). A 1:1 mixture containing the same volume of 1XTAE buffer and the PCR products were loaded on a $2\%$ agarose gel for electrophoretic detection. PCR products were mixed in equidensity ratios. Then, the mixture of PCR products was purified using the Quant-iTPicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA). Sequencing libraries were generated using the TruSeq Nano DNA LT Library Prep kit (Illumina, San Diego, CA, USA) following the manufacturer’s recommendations, and index codes were added. The library’s quality was assessed on the Agilent 5400 (Agilent Technologies Co. Ltd., Santa Clara, CA, USA). At last, the library was sequenced on an Illumina NovaSeq 6000 platform, and 250 bp paired-end reads were generated.
Microbiome bioinformatics were performed with QIIME2 2019.4 [33] with slight modification according to the official tutorials (https://docs.qiime2.org/2019.4/tutorials/ (accessed on 30 September 2022)). Briefly, raw data FASTQ files were imported into the format that could be operated by the QIIME2 system using the qiime tools import program. The DADA2 [34] process is to obtain amplified variant sequences through de-duplication. In the process, clustering is not carried out based on similarity, but only de-duplication is carried out. Demultiplexed sequences from each sample were quality filtered and trimmed, de-noised, merged, and then the chimeric sequences were identified and removed using the QIIME2 DADA2 plugin to obtain the feature table of amplicon sequence variants (ASV) [34]. The QIIME2 feature-classifier plugin was then used to align ASV sequences to a pre-trained GREENGENES 13_8 $99\%$ database (trimmed to the V4V5 around a 420bp region bound by the 515F/907R primer pair) to generate the taxonomy table [35]. In order to unify the sequence effort, samples were rarefied at a depth of 25,318 sequences per sample before alpha and beta diversity analysis. Rarefaction allows one to randomly select a similar number of sequences from each sample to reach a unified depth.
## 2.4. Bioinformatics and Statistical Analyses
Sequence data analyses were mainly performed using QIIME2 and R software (v3.2.0). ASV-level alpha diversity indices, such as the Chao1 richness estimator and Pielou’s evenness, were calculated using the ASV table in QIIME2 [36,37], and visualized as box plots (R software, package “ggplot2”). Beta diversity analysis was performed to investigate the structural variation of microbial communities across samples using weighted or unweighted UniFrac distance metrics [38,39] and visualized via principal coordinate analysis (PCoA) (R software, package “ape”). The significance of differentiation of microbiota structure among groups was assessed by PERMANOVA (permutational multivariate analysis of variance) [40]. Random forest analysis (R software, package “randomForest”) was applied to sort the importance of microbiota with differences in abundance between groups and screen the most critical phyla and genera that lead to microbial structural differences between groups using QIIME2 with default settings [41,42]. Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (Picrust2) [43] is software that predicts the functional abundance from the sequencing data of marker genes (typically 16S rRNA). An ASV’s abundance table is used for standardization, and the corresponding relationship of each ASV is compared with the Kyoto Encyclopedia of Genes and Genomes (KEGG) library to obtain the functional information and functional abundance spectrum.
## 3.1. Identification of Individuals and Sex
A total of 22 red deer individuals were identified from 43 fecal samples, including 12 males and 10 females (Supplementary Table S2). The female captive deer were CF1, CF2, CF3, CF4, CF5, CF6, CF7, and CF8. The male captive deer were CM1, CM2, and CM3. We divided all the red deer (22 wild and 11 captive) into four groups: wild females (WF) ($$n = 10$$), wild males (WM) ($$n = 12$$), captive females (CF) ($$n = 8$$), and captive males (CM) ($$n = 3$$). The information about identification, location, sex, and diet is summarized in Supplementary Table S2.
## 3.2. Diet Composition and Nutritional Composition of Wild and Captive Red Deer Winter Diets
The wild red deer were fed on 16 species of plants in the winter. The edible plants belonged to 16 species of 16 genera and 9 families. Since the frequency of occurrence of other edible plants in red deer, such as Mongolian oak (Quercus mongolica) and Chinese maple (Acer sinensis), was less than $7\%$, the nutrient content of these plants was not measured. In addition, we hypothesized that they had little influence on the nutritional strategy of red deer. Therefore, the primary nutrient contents of 14 types of edible plants were determined. The food and nutritional composition of wild red deer are shown in Supplementary Table S3. When the captive red deer were fed, each type of food was fed separately at different times. The nutritional content of the primary food of captive red deer from the farm (adjacent to the Gaogestai Nature Reserve) in winter is shown in Supplementary Table S4. Only one kind of diet were provided to captive deer at each feeding time with all captive deer feeding together. Captive red deer feed on leaves and high protein given by artificial feeding. Compared with captive red deer, wild deer have a wider feeding range and no dietary limitations. Substantial differences exist between these two feeding methods.
## 3.3. Sequencing Analysis and Clustering
A total of 1,561,654 high-quality sequences were obtained from the fresh winter feces of 22 wild deer and 11 captive deer. Rarefaction curves based on the Chao1 diversity index reached asymptotes at 22,500. The results showed that with the increase in amount of sequencing, the curve tended to be flat and no longer changed, indicating that the amount of sequencing in this study basically reflected the diversity of red deer fecal microbiota in this study (Supplementary Figure S1). A total of 15,228 ASVs were obtained using a $100\%$ similarity clustering method. The WF, WM, CF, and CM groups included 3056 ASVs, 3924 ASVs, 6661 ASVs, and 1587 ASVs, respectively.
## 3.4. Microbial Composition and Diversity by Environment and Sex
We found significant differences in fecal microbial composition between wild and captive red deer based on environment. The fecal microbial communities of four groups (WF, WM, CF, and CM) were dominated by the phyla Firmicutes and Bacteroidetes (Figure 1A). The phylum Firmicutes was the most abundant in WF (81.12 ± $2.87\%$), followed by WM (79.03 ± $2.19\%$), CF (58.24 ± $3.17\%$), and CM (59.66 ± $0.47\%$). Secondly, Bacteroidetes was abundant in WF (15.19 ± 2.09), WM (16.89 ± $2.08\%$), CF (33.02 ± 5.48), and CM (31.55 ± $1.61\%$). At the genus level, the genera from the four groups with abundance > $1\%$ were Oscillospira, a candidate genus 5-7N15 from the family Bacteroidaceae, Ruminococcus, Roseburia, Clostridium, and Prevotella (Figure 1B and Table 1).
The chao1 diversity indices demonstrate a significant difference between the WF and WM groups ($p \leq 0.01$). There was no statistically significant difference between the CF and CM groups ($p \leq 0.05$). Pieluo’s diversity index showed that no significant differences occurred between WF and WM groups ($p \leq 0.05$) or CF and CM groups ($p \leq 0.05$) (Figure 2).
Wild and captive red deer also differed in beta-diversity. An PCoA plot based on the Unweighted Unifrac and Weighted Unifrac distance matrix revealed clear separation of the fecal microbiota between wild and captive red deer (Figure 3A). The results of a PCoA analysis showed that the fecal microbial structures of the CF and CM groups were more similar than those of the WF and WM communities ($F = 13.82$, $$p \leq 0.001$$; and unweighted: $F = 5.983939$, $$p \leq 0.001$$; Figure 3A; Supplementary Table S5).
A random forest analysis showed that Firmicutes and Bacteroidetes were the primary microorganisms that had differences between the wild and captive populations by (an importance > 0.1) (Figure 3C, D). This analysis indicated that there were significant differences in the abundances of Firmicute and Bacteroidetes between the four groups (an importance > 0.1), which were the primary phyla that caused differences in the microbial communities between groups (Figure 3C). Ruminococcus, Treponema, Akkermansia, a candidate genus 5-7N15 belonging to family Bacteroidaceae, and a candidate genus rc4-4 belonging to family Peptococcaceae were the main genera that caused differences in microbial communities between sex and environment (importance > 0.04; Figure 3D).
## 3.5. Functional Modules of Fecal Microbial Communities
Metabolism was found to be the most common function prediction performed on fecal microbial communities and included the most important pathways for microbial clustering ($76.67\%$). The second pathway of metabolism included amino acid metabolism ($17.26\%$), carbohydrate metabolism ($17.85\%$), metabolism of cofactors and vitamins ($16.57\%$), and metabolism of terpenoids and polyketides ($12.66\%$) (Figure 4A). A PCoA analysis showed that the WF and WM groups had more similar microbial function clusters (Figure 4B).
It was found that there were significant differences in the three metabolic pathways of glycan biosynthesis and metabolism (GBM), energy metabolism (EM), and metabolism of other amino acids (MAA) ($p \leq 0.05$) (Figure 5).
## 4. Discussion
This is the first study to apply high-throughput sequencing to describe the fecal bacterial microbiota of wild and captive red deer by sex. Analysis of the differences in fecal microbiota is a key step in releasing captive red deer to help expand the wild population. *In* general, the fecal bacterial microbiota of red deer was similar to that of other cervidae, such as elk (Cervus canadensis), white tailed deer (Odocoileus virginianus) [38], and white-lipped deer (Cervus albirostris) [39], at least at the bacterial phylum level, with high proportions of the phyla Firmicutes and Bacteroidetes. In the digestive tract of herbivores, the role of *Firmicutes is* mainly to decompose cellulose and convert it into volatile fatty acids, thereby promoting food digestion and host growth and development. The enrichment of Firmicutes plays an important role in promoting the ability of red deer to obtain abundant nutrients from food and, at the same time, affects the metabolic function of the fecal microbiota. Bacteroidetes can improve the metabolism of organisms, promote the development of the gastrointestinal immune system, participate in the body’s bile acid, protein, and fat metabolisms, and also have a certain regulatory effect on carbohydrate metabolism. It can also produce special glycans and polysaccharides, which have a strong inhibitory effect on inflammation [43]. Differences in microbiota may be explained by changes in diet. Data from previous local and overseas studies have shown that diet is the main factor affecting the gut microbiota in mammals [40]. It is likely that wild deer have a more varied diet, more than captive deer. These phyla, Firmicutes and Bacteroidetes, are involved in important processes such as food digestion, nutrient regulation and absorption, energy metabolism, and host intestinal defense against foreign pathogens [40,41,42].
Alpha diversity alterations may be attributed to differential diet or hormonal influences on the gut microbiota. Fecal microbiota richness in wild populations is higher than that in captive animals, such as the Tibetan wild ass (Equus kiang), bharal (Pseudois nayaur), Tibetan sheep (Ovis arise), and yak (Bos mutus) [44,45,46,47,48]. Nevertheless, other studies also found that captivity might increase the alpha diversity of fecal microbiota in most Cervidae compared with other animals, for example, sika deer (Genus Cervus), Père David’s (Elaphurus davidianus), and white-tailed deer (Odocoileus virginianus) [49,50]. It may be that some environmental stresses in the wild or the special structure of the stomach and intestines in these deer lead to decreased alpha diversity of fecal microbiota in wild deer [50]. This phenomenon needs further research to determine its cause. Our results showed that the richness of the fecal microbial community in wild red deer differed by sex (Figure 2). In wild deer, the microbiota diversity was higher for females than males. Microbial community alterations by sex could be attributed to hormonal [51]. The sampling time was during the gestation period of red deer. Levels of female growth hormone during pregnancy may affect the fecal microbiota. Reproductive hormones have also been associated with sex and gut microbial changes in wild animals [17,52,53]. Increased evidence indicates that sex steroid hormone levels are associated with the human gut microbiota [54,55]. Futher, Edwards et al. reported that estrogen and progesterone had an impact on gut function [56]. The captive deer also had the smallest sample size ($$n = 3$$ males and 8 females), which limited our ability to detect these differences.
In this study, the functional pathway composition of wild red deer is more similar (Figure 5B), which is completely opposite to the microbial structure (Figure 3A). The change in microbial structure does not necessarily lead to the change in function, which may be due to the same function in different microbial communities [57]. In recent years, studies have shown that gut microbiota are involved in various metabolic processes such as amino acids, carbohydrates, and energy, confirming their primary role in assisting host digestion and absorption [58]. It has also been found to be involved in environmental information processing, suggesting that the gut microbiota plays an important role in facilitating acclimation to changing environments [59]. The metabolism of gut microbiota is closely related to the feeding habits of the host. In the long-term evolution process, the gut microbiota will respond to changes in diet types or specific diets by adjusting the content of certain digestive enzymes [4,60]. Studies have shown that the decrease of fecal microbial diversity can lead to a reduction in the functional microbiota, in the efficiency of the microbiota, and in the resistance to pathogen invasion [61]. The decrease in fecal microbial diversity in captive populations resulted in a decrease in functional microbiota [61]. Ruminococcaceae and Lachnospiraceae are two of the most common bacterial families within the *Firmicutes phylum* [62]. It has been hypothesized that they have an important role as active plant degraders [63,64]. According to our results, the level of Ruminococcaceae in the captive groups is significantly lower than that in the wild group, which could suggest that the fiber-reduced diet in captivity is modifying the ability of the fecal microbiota to degrade recalcitrant substrates such as cellulose, hemicellulose, and lignocellulose, among others, that are commonly found on the main resources of the wild red deer diet. The captive deer’s consequent reduction of diet resources might trigger the decline of important metabolic pathways associated with nutrient use [64].
16S rRNA analysis constitutes a valuable and cost-efficient approach for surveillance and monitoring wild populations as well as captive individuals. Picrust2 prediction accuracy is dependent on the availability of closely related annotated bacterial genomes in the database and the phylogenetic distance from the reference genome. However, the prediction results are still uncertain, which does not mean that the correlation between the predicted genes and the real metagenome of the microbiota is $100\%$ [65]. At present, due to the difficulty of cultivation, the mechanism by which some functional bacteria exert their effects remain unclear. Therefore, in the follow-up work, it is necessary to repeatedly cultivate the conditions of some intestinal anaerobic bacteria, the most extensive of which are Firmicutes and some Bacteroidetes. The microbiota was cultured in vitro by simulating the gut environment, and its functions were speculated and further verified in combination with multiple groups of studies (metagenomics, meta transcriptome, and proteome, etc.). At the same time, the unknown functional microbiota and its genome sequence information can be explored and studied. These works will help to understand the metabolic activities of the complex microbiota and further explore the host physiological processes involved in gut microbiota.
## 5. Conclusions
In conclusion, our study provided information on the structure and function of the fecal microbiome of red deer through the 16S rRNA gene of fecal samples. Comparing analyses identified significant variations of fecal microbiota composition and functions between captive and wild populations and also indicated that environment and sex have a great influence on these variations. These findings were of great significance for the reintroduction of captive red deer, given that the differences in fecal microbiota composition and functions between captive and wild red deer would greatly impact the ability of captive red deer to adapt to the wild environment. For further study, incorporating novel methods (e.g., transcriptome) to study the functional annotation of gene content and the functional traits of the host would be essential for better understanding the physiology and immunology of red deer. |
# Rumen-Protected Lysine and Methionine Supplementation Reduced Protein Requirement of Holstein Bulls by Altering Nitrogen Metabolism in Liver
## Abstract
### Simple Summary
Excessive protein intake causes dietary nitrogen to be excreted through urine nitrogen and fecal nitrogen, reducing nitrogen use efficiency. The main way to reduce dietary nitrogen loss is to reduce dietary protein content, as well as to meet the nutritional needs of ruminants. Therefore, reducing crude proteins while adding rumen amino acids can achieve a reduction in nitrogen emissions. The results showed that adding RPLys (55 g/d) and RPMet (9 g/d) to the bull diet and low protein diet ($11\%$) could improve the growth performance, increase the level of nitrogen metabolism, and enhance the expression of genes related to nitrogen metabolism.
### Abstract
The aim of this study was to investigate the effect of low-protein diets supplemented with rumen-protected lysine (RPLys) and methionine (RPMet) on growth performance, rumen fermentation, blood biochemical parameters, nitrogen metabolism, and gene expression related to N metabolism in the liver of Holstein bulls. Thirty-six healthy and disease-free Holstein bulls with a similar body weight (BW) (424 ± 15 kg, 13 months old) were selected. According to their BW, they were randomly divided into three groups with 12 bulls in each group in a completely randomized design. The control group (D1) was fed with a high-protein basal diet (CP$13\%$), while bulls in two low-protein groups were supplied a diet with $11\%$ crude protein and RPLys 34 g/d·head + RPMet 2 g/d·head (low protein with low RPAA, T2) or RPLys 55 g/d·head + RPMet 9 g/d·head (low protein with high RPAA, T3). At the end of the experiment, the feces and urine of dairy bulls were collected for three consecutive days. Blood and rumen fluid were collected before morning feeding, and liver samples were collected after slaughtering. The results showed that the average daily gain (ADG) of bulls in the T3 group was higher than those in D1 ($p \leq 0.05$). Compared with D1, a significantly higher nitrogen utilization rate ($p \leq 0.05$) and serum IGF-1 content ($p \leq 0.05$) were observed in both T2 and T3 groups; however, blood urea nitrogen (BUN) content was significantly lower in the T2 and T3 groups ($p \leq 0.05$). The content of acetic acid in the rumen of the T3 group was significantly higher than that of the D1 group. No significant differences were observed among the different groups ($p \leq 0.05$) in relation to the alpha diversity. Compared with D1, the relative abundance of Christensenellaceae_R-7_group in T3 was higher ($p \leq 0.05$), while that of Prevotellaceae _YAB2003_group and Succinivibrio were lower ($p \leq 0.05$). Compared with D1 and T2 group, the T3 group showed an expression of messenger ribonucleic acid (mRNA) that is associated with (CPS-1, ASS1, OTC, ARG) and (N-AGS, S6K1, eIF4B, mTORC1) in liver; moreover, the T3 group was significantly enhanced ($p \leq 0.05$). Overall, our results indicated that low dietary protein ($11\%$) levels added with RPAA (RPLys 55 g/d +RPMet 9 g/d) can benefit the growth performance of Holstein bulls by reducing nitrogen excretion and enhancing nitrogen efficiency in the liver.
## 1. Introduction
Protein, as typically the most expensive macronutrient of diets, plays critical roles in the health, growth, production, and reproduction of animals. However, protein ingredient shortages and nitrogen pollution challenge the livestock farming worldwide, albeit these problems have been alleviated in recent decades due to an increase in demand for animal source food from a fast-growing population with rising incomes [1,2]. Therefore, enhancing the utilization efficiency of dietary protein and reducing excretory losses would be alternative strategies to solve these problems [3].
Low-protein diets have been proven to enhance nitrogen utilization [4,5]. However, restricting N intake also sacrificed the growth performance and productivity of animals [6,7], which has been attributed to limiting amino acid deficiency in low-protein diets [8]. Lysine (Lys) and methionine (Met) are the top two limiting amino acids (LAA) for ruminants [9,10]. Adding rumen-protected Lys and Met in low-protein diets was considered an efficient way to the meet animal amino acids requirement, as they could escape from rumen degradation and increase the supply of amino acids to the intestines, thus improving the N utilization [11]. Incorporating rumen-protected Lys and (or) Met into low-protein diets was reported to increase dry matter intake in transition cows [12,13]. Previous studies also suggested that rumen-protected Lys and (or) Met in low-protein diets promoted milk protein yield in high-producing dairy cows [14,15] and maintained milk production and milk protein yield while reducing the N losses in urine in dairy cows [16]. The question of how to reduce nitrogen emissions of ruminants without affecting their production performance has always been the focus of scholars, and the research in this area has mostly been focused on dairy cows; however, there have been few studies conducted on Holstein bulls.
Nitrogen recycling contributes to effective N utilization in ruminants [17], and ruminal microbiota and the liver play important roles in this nitrogen metabolism [4]. Therefore, the aim of this study was to investigate the effect of low-protein diets supplemented with rumen-protected lysine (RPLys) and methionine (RPMet) on growth performance, rumen fermentation, blood biochemical parameters, nitrogen metabolism, and gene expression related to N metabolism in the livers of Holstein bulls.
## 2. Materials and Methods
This study was conducted between March 2016 and June 2016 at Hongda an animal husbandry in Baoding, P. R. China. The experimental protocol (YXG 1711) was approved by the Institutional Animal Care and Use Committee of Hebei Agricultural University.
## 2.1. Animals, Experimental Design, and Diets
Thirty-six healthy and disease-free Holstein bulls with a similar body weight (BW; 424 ± 15 kg, aged 14 months old) were selected. According to their BW, they were randomly divided into 3 groups with 12 bulls in each group in a completely randomized design. The control group (D1) was fed with a high-protein basal diet (CP$13\%$), while bulls in two low protein groups were supplied diet with $11\%$ crude protein and RPLys 34 g/d·head + RPMet 2 g/d·head (low protein with low RPAA, T2) or RPLys 55 g/d·head + RPMet 9 g/d·head (low protein with high RPAA, T3). Basic diets were prepared according to Japanese feeding standard [2008] for beef cattle [18] (Table 1). The RPAA (Hangzhou Kangdequan Feed Limited Company, Hangzhou, Zhejiang, China) feed was used with a rumen protection rate of $60.0\%$ and was premixed with 100 g of grounded corn which, was used as a carrier for the supplement and was the same amount of grounded corn as that supplied to bulls in the D1 group. All animals were fed ad libitum the basic diets and with free access to clean water. All the experimental animals were housed in tie stalls according to the groups and were fed twice daily at 06:00 and 18:00 h following the removal of the feed refusals before morning feeding. The experiment consisted of 3 periods: a 14-day adaptation period, a 2-month feeding period, and a 7-day sample collection period. Holstein bulls were weighted before morning feeding at the beginning and end of every feeding period.
## 2.2. Sample Collection
The diet offered and refused for individual bulls was weighed every day throughout the trial to average daily dry matter intake (ADMI). Samples of individual feed ingredients, orts, and diets were collected weekly during the experimental period and stored at −20 °C [19]. At the beginning of the experiment, all Holstein bulls were weighed before feeding in the morning to obtain their initial weight. Similarly, at the end of the trial, all Holstein bulls were weighed before morning feeding to obtain the final weight, and the average daily gain (ADG) was calculated as (final weight–initial weight)/test days. Based on the ADMI and ADG, the feed weight ratio (F/G) was calculated. At the end of the feeding period, four Holstein bulls in each group were randomly selected, and a 10-mL blood sample was collected via jugular venipuncture from each bull before morning feeding. The samples were immediately centrifuged at 3000 rpm for 15 min, and the serum samples were collected and stored at −20 °C for further analysis. After 2 h of morning feeding at the end of the feeding period, the ruminal fluid samples of four bulls were collected via an oral stomach tube equipped with a vacuum pump. We discarded the first 100 to 200 mL of fluid collected to reduce the chance that the stomach tube rumen samples were contaminated with saliva. Once again, approximately 200 mL of rumen fluid was collected, and about 20 mL was taken, filtrated with four layers of sterile cheesecloth, and then transferred to 2-mL sterile tubes and stored in liquid nitrogen for further analysis.
Three bulls in each group were randomly selected and euthanized at the end of the feeding experiment after 2 h of morning feeding. The middle part of liver tissue was immediately collected after animal sacrifice and cut into 5-mm fragments; the tissue sample was then placed into sterile tubes and stored in liquid nitrogen for further analysis.
Another three bulls in each group were randomly selected after the feeding period and were transferred to metabolic cages. After a 5-day adaption period, feces and urine were collected during the next 3 days. Total feces and urine were respectively collected daily before morning feeding. The feces of each bull were weighted, mixed, subsampled (100 g/kg), and stored at −20 °C. Each bull fecal sample was evenly divided into two parts, one with $10\%$ (10:1) sulfuric acid solution and the other without acid, before being dried, crushed, sifted, and stored at room temperature for the determination of nutrient content. The urine of each bull was collected using a plastic container with 10 mL of $10\%$ sulfuric acid to prevent the loss of ammonia; then, after the volume was measured, the urine was filtered with four layers of gauze filter, and subsamples (100 mL/individual) were stored at −20 °C for urine nitrogen measurement.
## 2.3. Laboratory Analysis
Offered and refused feed and feces were dried at 55 °C for 48 h, ground to pass through a 1-mm screen (Wiley mill, Arthur H. Thomas, Philadelphia, PA, USA), and stored at 4 °C for analysis of chemical composition. The dry matter (DM, method 934.01), ash (method 938.08), crude protein (CP, method 954.01), ether extract (EE, method 920.39), Ca (method 927.02), and P (method 965.17) contents of the samples were determined according to the procedures of the AOAC [20], and NDF (amylase) and ADF content was analyzed using the methods of Van Soest et al. [ 21]. Lysine and methionine content in the feed was analyzed using an automatic AA analyzer (Hitachi 835, Tokyo, Japan).
Serum alanine transferase (ALT), aspartate transferase (AST), albumin (ALB), total protein (TP), glucose (GLU), and blood urea nitrogen (BUN) were analyzed using an automatic biochemical analyzer (Hitachi 7020, Tokyo, Japan). Serum growth hormone (GH) and insulin-like growth factor-1 (IGF-1) contents were measured with enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturer’s specifications (HZ Bio. CO., Shanghai, China).
The pH value of the rumen fluid was measured immediately by using a digital pH analyzer (PHS-3C, Shanghai, China), and ammonia nitrogen (NH3-N) and microbial protein (MCP) were determined following recommendations provided in previous studies [22]. Volatile fatty acid (VFA) concentrations in rumen fluid were analyzed using gas chromatography (TP-2060F, Tianpu. Co., Ltd., Beijing, China).
The DNA in rumen fluid was extracted using the CTAB method using a commercial kit (Omega Bio-Tek, Norcross, GA, USA), and, after DNA was purified with $1\%$ agarose gel electrophoresis, the library was constructed using a TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, Inc., San Diego, CA, USA). Then, the constructed library was quantified using HiSeq2500 PE250 (Illumina, Inc., San Diego, CA, USA). Sequences data were analyzed using the QIIME2 pipeline according to a previous study [23] and submitted to NCBI with project ID P2016030502-S2-3-1.
The primer of target genes (Table 2) was designed according to the bovine gene sequences reported in NCBI and synthesized by the Shanghai Biotedchnology Technology Corporation Limited Company. The total amount of ribonucleic acid (RNA) was extracted from the liver tissue of Holstein bulls with a miRNeasy kit (Qiagen, Hilden, Germany); then, RNA quality was determined using NanoDrop 2000 (NanoDrop Tec, Rockland, DE) with OD260/OD280 ranging between 1.9 and 2.1. Real-time polymerase chain reaction (PCR) was performed to quantify the expression of target genes, using an SYBR Green PCR Master mix (Takara bio-Co., Shiga, Japan) and following the manufacturer’s protocols. *The* gene expression of liver tissue was calculated using the method of 2-ΔΔCt, where the expression of ACTB was used as referenced D1.
## 2.4. Statistical Analysis
The data management was performed using a spreadsheet program with Excel, and statistical analysis was carried out using R software (version 3.6.3, R Foundation for Statistical Computing, Vienna, Austria.) with a one-way analysis of variance (ANOVA) model: Y = α + Xi + ei, where Y is the observed parameters, α is the overall mean, *Xi is* the ith treatment effect, and ei is the residual error. All data were shown using least squares means, and significant differences among treatments were declared at $p \leq 0.05$ and a tendency if 0.05 < p ≤ 0.10.
## 3.1. Growth Performance
There was no significant difference ($p \leq 0.05$) in ADG, ADMI, and F/G among different groups; however, the F/G in the T2 and T3 groups decreased by $8.45\%$ and $6.67\%$, respectively, compared with D1 (Table 3).
## 3.2. Nitrogen Metabolism
Compared with the D1 group, the intake of nitrogen and the amount of nitrogen excretion by feces and urine were significantly lower in the T2 and T3 groups ($p \leq 0.05$). The ratio of nitrogen excretion by feces and nitrogen intake (FN/IN) was lower in T3 compared with the D1 and T2 groups, while the ratio of nitrogen excretion by urine and nitrogen intake (UN/IN) was lower in the T2 and T3 groups compared to the D1 group. Thus, a significantly higher nitrogen utilization rate was observed in both T2 and T3 groups compared with the D1 group ($p \leq 0.05$; Table 4).
## 3.3. Serum Biochemical Index
Low-protein diet with RPAA supplementation had no effect on concentrations of ALT, AST, ALB, TP, GLU, and GH in serum ($p \leq 0.05$). Concentration of serum BUN significantly decreased; however, the concentration of serum IGF-1 significantly increased in the T3 group compared with the D1 group ($p \leq 0.05$; Table 5).
## 3.4. Rumen Fermentation
No significant difference was detected in the rumen pH, concentration of NH3-N, MCP, propionate, and butyrate, and in the ratio of acetate/propionate among different groups ($p \leq 0.05$). The concentration of acetate in the T3 group was significantly higher than that in D1 and T2 ($p \leq 0.05$; Table 6).
## 3.5. Rumen Microbiota
No significant difference was observed in alpha diversity among the different groups ($p \leq 0.05$; Table 7). The relative abundance of the highest 16 abundant bacteria at the genus level was compared among the different groups. However, the relative abundance of Ruminococcaceae_NK4A214 in the T3 group was lower than that in the D1 group ($p \leq 0.05$), and the abundance of Christensenellaceae_R-7_group in the T3 group was lower than that in both D1 and T2 groups ($p \leq 0.05$). Meanwhile, the relative abundance of Prevotellaceae_YAB2003_group in T3 was higher than that in the D1 group ($p \leq 0.05$), and the relative abundance of Succinivibrio in T3 was higher than that in both the D1 and T2 groups ($p \leq 0.05$; Table 8).
## 3.6. Gene Expression in Liver Tissue
The expression of the CPS-1, ASS, ARG, OTC, and N-AGS genes, which relate to nitrogen metabolism or urea metabolism in liver tissue, are shown in Figure 1. The expression of CPS-1, ARG, and N-AGS was significantly upregulated in the T3 group ($p \leq 0.05$), although no significant difference was observed between the rT2 and D1 groups ($p \leq 0.05$). The expression of CPS-1, ARG, and N-AGS increased by $25\%$, $18\%$, and $13\%$ in the T2 group compared with D1. The expression of ASS and OTC was upregulated in both the T2 and T3 groups compared with D1 ($p \leq 0.05$).
The expression of the SLC3A2, IRS1, PDK, P13K, TSC1, TSC2, mTORC1, eIF4EBP1, S6K1, and eIF4B genes, which are related to the nitrogen metabolism in liver tissue, are shown in Figure 2. The low-protein diet with RPAA supplementation did not affect gene expression of SLC3A2, P13K, TSC2, and eIF4EBP1 ($p \leq 0.05$); however, the expression of IRS1, PDK, S6K1, and eIF4B genes in liver tissue increased significantly ($p \leq 0.05$), and the expression of the mTORC1 gene also increased ($$p \leq 0.09$$), while the expression of TSC1 gene decreased significantly ($p \leq 0.05$).
## 4. Discussion
Protein is one major factor that affects the health, growth, and production of ruminants. Moreover, although people tend to formulate high-protein diets to achieve a better production of ruminants, the global protein shortage is increasing [1], and high-protein diets overload the environment by increasing nitrogen (N) excretion through urine and feces [3], which is harmful for the sustainability of the livestock industry.
By providing bulls with a low-protein diet ($11\%$ CP) supplemented with rumen-protected lysine and methionine, our findings indicate that, compared with a high-protein diet ($13\%$ CP) group which followed the recommended Japanese feeding standard for beef cattle [18], our low-protein diet supplemented with RPAA increased ADG and N utilization and decreased N excretion through urine and feces. These findings were comparable with previous studies in which the feeding of rumen-protected Lys and (or) Met to castrated cattle increased daily gain [24] and reduced urinary nitrogen and urea nitrogen in urine [25]. The World Health Organization (WHO) proved that the addition of RPAA to a low-protein diet increases N utilization, reduces N emission and environmental pollution, and promotes the growth performance of dairy cows [12,14].
Blood biochemical parameters are sensitive to animal health and nutrient condition [26,27]. The serum content of ALT, AST, ALB, TP, GLU, BUN, GH, and IGF-1 was used to assess the nutrient condition of bulls with different treatment groups. From this, we observed that BUN content decreased, and IGF-1 content increased, in bulls provided with a low-protein diet supplemented with RPAA, while other indexes were not affected. The serum BUN content reflects the nitrogen balance of ruminants and negatively correlated with N utilization [17]. When ruminants were provided with low-dietary protein with a higher N utilization, serum BUN decreased [4,28]. The main function of IGF-1 relates to the inhibiting of protein degradation and the promoting of protein synthesis to maintain nitrogen balance and to improve the growth performance of animals [29,30]. These observations further explained the improvement in N utilization and growth performance of bulls on a low-protein diet supplemented with RPAA.
When cattle are fed with low-protein diets, urea N recycling can be considered a high-priority metabolic function because a continuous N supply for microbial growth in the rumen is a strategy for animal survival [31]. The abundance of the microflora reflects its ability to adapt to a particular environment and compete for available nutrients; moreover, it indicates its importance to the overall function of the microbiome as a whole [32]. The ACE (reflecting the richness of bacteria in the sample), Shannon, and PD-whole-tree (reflecting the microbial diversity in feces) indexes were used to assess the alpha diversity of rumen microbiota. Previous studies have demonstrated that rumen fermentation and microbiota are sensitive to protein levels [33,34] or feed ingredients [35] in ruminants, which were also sensitive biomarkers of N utilization [36]. By monitoring the rumen fermentation and microbiota, we observed an increase in the acetate content of rumen; however, other parameters including NH3-N and MCP content were not significant affected, which is similar to the results of a study by Martin et al. [ 37]. The addition of methionine analogue 2-hydroxy-4-methylthiobutyric acid (HMB) and esterified 2-hydroxy-4-methylthiobutyric acid (HMBi) to the diet of dairy cows significantly increased the content of rumen total volatile fatty acids (TVFAs) [37]. Some studies have shown that methionine hydroxy analogue (MHA) can increase the ratio of acetic acid and butyric acid in rumen content [38]. Research has showed that $0.52\%$ of methionine could increase the content of butyric acid in rumen, while $0.26\%$ methionine did not affect the content of VFA [39]. The above results show that the effect of methionine on rumen VFA content is unpredictable. The alpha diversity of microbiota in rumen was not affected by treatment, and only a small portion of bacteria at the genus level (~$5\%$ in abundance) was determined to be significantly different between groups with a decreased relative abundance of Ruminococcaceae_NK4A214_group and Christensenellaceae_R-7_group and increased Prevotellaceae_YAB2003_group and Succinivibrio in bulls on a low-protein diet supplemented with RPAA. These findings hinted that bulls on a low-protein diet supplemented with RPAA would maintain the rumen fermentation and maintain ruminal microbiota homeostasis compared with that from D1.
The liver plays important roles in the utilization efficiency of recycled N. The excess nitrogen in the rumen is usually inhaled into the animal’s blood in the form of ammonia, which is then metabolized by the liver to synthesize urea. All the urea synthesized by the liver, some of which is secreted via saliva into the rumen and intestines of animals, are reused by bacteria, protozoa, and other microorganisms; the other part is filtered by the kidneys and excreted with the urine [28]. The urea cycle plays a key role in maintaining a positive balance of nitrogen in anima, especially at low dietary nitrogen levels. S6K1 and eIF4EBP1 are genes that regulate protein translation downstream of mTORC1. The S6K1 gene can promote protein translation by stimulating the phosphorylation of downstream eIF-4B, RPS6, eIF-2, and PAPB [40], and the SLC3A2, IRS1, PDK, P13K, TSC1, TSC2, mTORC1, eIF4EBP1, S6K1, and eIF4B genes are related to nitrogen metabolism in the liver; moreover, these genes would become overexpressed when blood ammonia increased to increase urea synthesis and balance the blood ammonia [41]. However, unexpected results were observed in the current experiment: when feeding bulls with a low-protein diet supplemented with RPAA, we observed that the serum BUN decreased but the expression of genes associated with urea synthesis in liver increased. This finding can explain why the low-protein diet supplemented with RPAA induced an increase in N efficiency; however, the mechanism behind these upregulated genes in the liver was unclear. Previous studies have demonstrated that AA in diets not only provide animal nutrition but also act as a functional regulator and have ability to stimulate expression altering in multiple tissue cells such as mammary tissue [42], polymorphonuclear cells [43], and adipose tissue [44], as well as liver tissue [45,46]. The influence of RPLys and RPMet on liver genes’ expression requires further study. As the number of samples selected in this study is limited, it is necessary to further test the current data in the future research.
## 5. Conclusions
In summary, providing low dietary protein ($11\%$) with RPLys (55 g/d) and RPMet (9 g/d) to bulls could increase their nitrogen utilization rate, serum IGF-1 content, ruminal acetate content, and expression genes associated with urine metabolism and nitrogen metabolism in liver compared to that with high protein ($13\%$). Our findings indicate that providing a low-protein diet supplemented with RPAA could benefit bulls mainly by increasing liver nitrogen metabolism and utilization; however, the RPAA’s affecting of liver gene expression at a nutrition level or as a signal molecule still requires further study. |
# Epithelial-to-Mesenchymal Transition and Phenotypic Marker Evaluation in Human, Canine, and Feline Mammary Gland Tumors
## Abstract
### Simple Summary
In this study we addressed the analysis of human breast cancer and canine and feline mammary tumors with regard to the expression, at either gene or protein level, of some molecules that are related to the capacity of an epithelial cell to become mesenchymal (epithelial-to-mesenchymal transition), acquiring higher ability to metastasize. In our samples, some typical markers of this transition were not higher at mRNA levels in tumors than in healthy tissues, indicating that some other markers should be investigated. Instead, at protein levels, some molecules such as vimentin and E-cadherin were indeed associated with higher aggressiveness, being potential useful markers. As already described in the literature, we also demonstrated that feline mammary tumors are close to an aggressive subtype of human breast cancer called triple negative, whereas canine mammary tumors are more similar to the less aggressive subtype of human breast cancer that expresses hormonal receptors.
### Abstract
Epithelial-to-mesenchymal transition (EMT) is a process by which epithelial cells acquire mesenchymal properties. EMT has been closely associated with cancer cell aggressiveness. The aim of this study was to evaluate the mRNA and protein expression of EMT-associated markers in mammary tumors of humans (HBC), dogs (CMT), and cats (FMT). Real-time qPCR for SNAIL, TWIST, and ZEB, and immunohistochemistry for E-cadherin, vimentin, CD44, estrogen receptor (ER), progesterone receptor (PR), ERBB2, Ki-67, cytokeratin (CK) $\frac{8}{18}$, CK$\frac{5}{6}$, and CK14 were performed. Overall, SNAIL, TWIST, and ZEB mRNA was lower in tumors than in healthy tissues. Vimentin was higher in triple-negative HBC (TNBC) and FMTs than in ER+ HBC and CMTs ($p \leq 0.001$). Membranous E-cadherin was higher in ER+ than in TNBCs ($p \leq 0.001$), whereas cytoplasmic E-cadherin was higher in TNBCs when compared with ER+ HBC ($p \leq 0.001$). A negative correlation between membranous and cytoplasmic E-cadherin was found in all three species. Ki-67 was higher in FMTs than in CMTs ($p \leq 0.001$), whereas CD44 was higher in CMTs than in FMTs ($p \leq 0.001$). These results confirmed a potential role of some markers as indicators of EMT, and suggested similarities between ER+ HBC and CMTs, and between TNBC and FMTs.
## 1. Introduction
Mammary gland cancer is the most common tumor in women [1] and in female dogs [2], and the third most common neoplasia in cats [3]. Human breast cancer (HBC) is classified into four main subtypes according to the expression of estrogen receptor (ER), progesterone receptor (PR), and epidermal growth factor receptor ERBB2, as follows: (i) Luminal A tumors (ER+ and/or PR+, ERBB2-); (ii) Luminal B tumors (ER+ and/or PR+, ERBB2+); (iii) ERBB2-overexpressing tumors (ER-, PR-, ERBB2+); and (iv) triple-negative (ER-, PR-, ERBB2-) breast cancer (TNBC) [4]. TNBCs are typically high-grade carcinomas characterized by an aggressive behavior and a poor prognosis, with high risk of distant metastasis and death [5]. Canine mammary tumors (CMTs) are classified based on morphologic features [6]. Fifty per cent of CMTs are malignant with a $20\%$ risk of metastasis [7]. The majority (80–$90\%$) of feline mammary tumors (FMTs) are characterized by a highly aggressive behavior that leads to rapid progression and distant metastasis development [8,9]. Typically, FMTs lack the expression of ER, PR, and ERBB2, and have been considered a remarkable spontaneous model for TNBC [10,11,12,13,14,15,16]. In all three species, mammary tumors exhibit both inter- and intra-tumor heterogeneity as a consequence of genetic and non-genetic aberrations [17].
Over the past 20 years, the investigation of cell differentiation/phenotypic markers has been used in both human and veterinary medicine, primarily to improve our knowledge of the histogenesis of mammary tumors [18]. In the normal human, canine, and feline mammary gland, two cell subpopulations are present: luminal epithelial cells, positive for cytokeratin (CK) 7, CK8, CK18, and CK19; and basal/myoepithelial cells, variably positive for CK5, CK6, CK14, CK17, SMA, calponin, vimentin, and p63 [19]. In HBC, the evaluation of cell differentiation proteins is frequently performed in association with routine diagnostic markers (ER, PR, ERBB2, and Ki-67) to better classify this tumor. The identification of HBC subtypes has a diagnostic, prognostic, and therapeutic value, and is associated with the cell differentiation and epithelial-to-mesenchymal transition (EMT) status of the neoplastic population according to a hierarchical model [20].
EMT is a key event that neoplastic epithelial cells use to acquire a mesenchymal phenotype [21]. As a result, tumor cells obtain the ability to detach from the primary tumor mass, invade the surrounding tissue, migrate throughout the body, and eventually give rise to metastases in distant organs [22]. The classical EMT is characterized by a decreased expression of epithelial markers and a complementary upregulation of mesenchymal markers. Classical EMT transcription factors, namely snail family transcription repressor $\frac{1}{2}$ (SNAIL), TWIST, and zinc-finger-enhancer binding protein $\frac{1}{2}$ (ZEB) are known to orchestrate EMT by regulating cell adhesion, migration, and invasion, also interacting with different signaling pathways and microRNAs [22,23]. Although this is a well-de-scribed process that promotes metastasis formation, accumulating evidence suggests the existence of an intermediate state called partial EMT or hybrid E/M, whereby both epithelial and mesenchymal markers are co-expressed in cancer cells [23,24,25].
The aim of this study was to investigate the mRNA expression of classical EMT-related transcription factors SNAIL, TWIST, and ZEB in human, canine, and feline mammary tumors. Additionally, we studied the expression of key proteins involved in the EMT process, including E-cadherin and vimentin, and of proteins related to the tumor phenotype, such as ER, PR, ERBB2, Ki-67, cytokeratin (CK) $\frac{8}{18}$, CK$\frac{5}{6}$, CK14, and CD44.
## 2.1. Tissue Collection
Human samples were collected from the Istituto Oncologico Veneto (IOV, Padua, Italy), whereas canine and feline samples were collected from local veterinary clinics. The human sample collection was approved by the IOV Ethics Committee. All patients or patients’ owners provided informed, written consent to use their samples for this study. Specifically, samples from 5 healthy human mammary gland tissues (MGTs), 5 ER+ HBCs, 5 TNBCs, 4 healthy canine MGTs, 10 canine mammary tumors (CMTs) (5 grade I and 5 grade II), 6 healthy feline MGTs, and 6 grade III FMTs were collected. In this study, to avoid contaminations with other tumor cell subpopulations, we selected only simple tubular carcinomas (STC), which are composed of only one tumor cell subpopulation (luminal epithelial cells) [6]. Healthy MGTs were collected from tumor-bearing patients during the therapeutic/diagnostic surgical procedures, with no additional sampling performed only for the study. Sampling was performed by surgeons. At the time of sampling, most of the tissue was fixed in $4\%$ formaldehyde for histopathology and immunohistochemistry, whereas a peripheral small portion of tumor and normal tissues (approx. 0.5 cm2 each) was collected and preserved in RNALater (Ambion, Austin, TX, USA), according to manufacturer’s instructions. In the lab, before RNA extraction, a small portion of each RNALater-preserved sample was fixed in $4\%$ formaldehyde and embedded in paraffin to check the content of the samples themselves. Four-μm tissue sections were stained with hematoxylin and eosin, and slides were visualized under the microscope to further confirm the presence of healthy tissue in the samples labelled as “healthy” and of tumor tissue in the samples labelled as “tumor”.
## 2.2. RNA Extraction and Real-Time Polymerase Chain Reaction
*For* gene expression analysis, a small portion of each tissue sample preserved in RNALater was used for RNA extraction using Trizol Reagent (Invitrogen, Carlsbad, CA, USA), following the manufacturer’s protocol. The extracted RNA was treated with RNAse-free DNAse I (New England Biolabs, Ipswich, MA, USA). Five-hundred ng of total RNA from each sample was reverse transcribed using the RevertAid First Strand cDNA Synthesis Kit (Invitrogen). The cDNA was then used as a template for quantitative real-time PCR using the ABI 7500 Real-Time PCR System (Applied Biosystem) to evaluate the mRNA expression of the following EMT-related genes: SNAIL1, SNAIL2, TWIST1, TWIST2, ZEB1, ZEB2. All the samples were tested in triplicate. ACTB was used as a house-keeping gene. The primer sequences are reported in Table 1. The primers were designed using NCBI Primer-BLAST. To examine primer specificity, the dissociation curves of qPCR products were assessed to confirm a single amplification peak. The qPCR reactions were then purified using the ExoSAP-IT PCR product cleanup (Applied Biosystems) and sequenced at the BMR Genomics (Padua, Italy). The sequences were then verified using the NCBI BLAST database. For data analysis for each sample, the ΔΔCt value was calculated and expressed as a relative fold change (2−ΔΔCt), as described in [16]. Real-time PCR efficiency was calculated by performing a dilution series experiment and applying the following formula to the standard curve: efficiency = 10(−1/slope) − 1 [26,27]. Real-time PCR efficiency was between 90 and $100\%$ for all the samples.
## 2.3. Immunohistochemistry
Immunohistochemistry (IHC) was performed on the above-mentioned samples as well as on additional human breast tissue samples from the Division of Anatomic Pathology archive of the University of Padua Hospital, and on additional canine and feline mammary tissue samples from the anatomic pathology archive of the Department of Comparative Biomedicine and Food Science of the University of Padua. Specifically, IHC was per-formed on the following tissue samples: 10 ER+ HBC, 11 TNBCs, 11 CMTs grade I, 11 CMTs grade II, 12 FMTs grade III. Sections (4 μm) were processed with an automatic immunostainer (BenchMark XT, Ventana Medical Systems), as previously described [11]. Briefly, the automated protocol included the following steps: a high-temperature antigen unmasking (CC1 reagent, 60 min), primary antibody incubation (1 h at RT, see below for dilutions), an ultrablock (antibody diluent, 4 min), hematoxylin counterstain (8 min), dehydration, and mounting. Negative controls omitted the primary antibody, whereas adnexa, epidermis, and non-tumor mammary gland, when present, were used as positive controls for CK$\frac{8}{18}$, CK$\frac{5}{6}$, CK14, E-cadherin, vimentin, and Ki-67. For ERBB2, an additional technical external positive control was used (ERBB2 3+ HBC), whereas the species-specific cross-reactivity was previously tested in dogs and cats [10,28]. For ER and PR, feline and canine uterus as well as ovary were also stained as positive controls. For CD44, the lymph node was used as positive control. Positive control tissues, typically collected from necropsies, were derived from the same archive as the canine and feline mammary tumor samples. The following antibodies were tested: anti-ER alpha (anti-ERα) (NCL-ER-6F11 1:40, Novocastra in human and feline species—NCL-ER-LH2 1:25, Novocastra in canine species); anti-PR (NCL-PGR-312 1:80, Novocastra in human and feline species); an-ti-ERBB2 (A0485 1:250, Dako in canine and feline species); anti-CK$\frac{8}{18}$ (NCL-L-5D3 1:30, Novocastra); anti-CK$\frac{5}{6}$ (D$\frac{5}{16}$ B4 1:50, Dako); anti-CK14 (NCL-LL 002 1:20, Novocastra); anti-E-cadherin (610182 1:120, BD Biosciences); anti-CD44 (550538 1:100, BD Biosciences); anti-vimentin (M0725 1:150, Dako); and anti-Ki-67 (M7240 1:50, Dako). In the human species, ERBB2 immunolabeling was performed with Bond Oracle HER2 IHC System for BOND-MAX (Leica Biosystems), containing the anti-ERBB2 antibody (clone CB11, ready-to-use). IHC positivity was semi-quantitatively and separately evaluated by ECVP-boarded (V.Z.) and experienced (L.C.) pathologists. Specifically, cytoplasmic and nuclear positivity were measured as a percentage of positive cells for all markers (100 cells per field in 10 high-power fields were counted). ERBB2 was scored as 0, 1+, 2+, and 3+ according to the American Society of Clinical Oncology (ASCO) 2018 recommendations [29] ($10\%$ cut-off), with 2+ and 3+ cases considered weakly and strongly positive for complete membrane immunolabeling, respectively. The protein expression of the studied markers was evaluated in the epithelial/luminal component. Additionally, immunolabeling was observed in healthy/hyperplastic adjacent mammary tissue, and in this case normal basal/myoepithelial cells were also evaluated.
## 2.4. Statistical Analysis
Statistical analyses were performed using Prism version 9.3.1 (GraphPad Software, San Diego, CA, USA). To verify mean differences among groups, either the Student’s t-test or the one-way ANOVA with Tukey’s multiple comparison test was used, when values were normally distributed. A Mann–Whitney test or Kruskal–Wallis test were used when values were not normally distributed. Normality was tested using the Shapiro–Wilk test. The Spearman’s rank correlation analysis was used to analyze associations between variables. The level of significance was set at $p \leq 0.05.$
## 3.1. Gene Expression
We sought to investigate the mRNA expression of the EMT transcription factors SNAIL, TWIST, and ZEB in mammary tumors compared with healthy tissue. In HBC (Figure 1), SNAIL1 showed a higher mRNA expression in TNBCs when compared with ER+ ($p \leq 0.05$). Conversely, the mRNA expression of TWIST1, TWIST2, and ZEB1 in ER+ and TNBCs was significantly lower than in healthy MGTs ($p \leq 0.05$). Additionally, TNBCs had a significantly lower mRNA expression of SNAIL2 and ZEB2 when compared with healthy MGTs ($p \leq 0.05$).
In CMTs (Figure 2), SNAIL1 showed a higher mRNA expression in STC II when compared with healthy MGTs ($p \leq 0.01$) and STC I ($p \leq 0.001$). The mRNA expression of SNAIL2, ZEB1, and ZEB2 was lower in tumors than healthy MGTs, although not statistically significant.
In FMTs (Figure 3), tumors showed a lower mRNA expression of SNAIL1, SNAIL2, TWIST1, TWIST2, ZEB1, and ZEB2 when compared with healthy MGTs, which was significant only for ZEB1 ($p \leq 0.05$).
## 3.2. Immunohistochemistry
Next, we aimed to study the expression of key proteins involved in the EMT process. The expression of the studied markers was evaluated in the tumor epithelial luminal cell population.
CD44 and ERBB2 staining was membranous, whereas CK$\frac{8}{18}$, CK$\frac{5}{6}$, CK14, and vimentin staining was cytoplasmic. E-cadherin staining was present in either or both membrane and cytoplasm and it was separately evaluated. Ki-67, ER, and PR staining was nuclear. As expected, epithelial luminal cells of healthy MGT in all three species were diffusely positive for CK$\frac{8}{18}$, membranous E-cadherin, ER, PR, and occasionally positive for CK$\frac{5}{6}$, CK14, and CD44. The basal/myoepithelial cells of healthy MGT in all three species were diffusely positive for CK$\frac{5}{6}$, CK14, CD44, and vimentin, and occasionally also positive for ER and PR.
Results for the human, canine, and feline mammary tumors are summarized in Table 2, Table S1 and are graphically represented in Figure 4.
In HBC (Figure 4A), ER+ tumors had a high protein expression (roughly $100\%$) of CK$\frac{8}{18}$, whereas they were negative for basal cytokeratins CK$\frac{5}{6}$ and CK14. In TNBCs, the protein expression of CK$\frac{8}{18}$, although fairly heterogeneous, was lower than in ER+ ($p \leq 0.001$) and the protein expression of CK$\frac{5}{6}$ was higher than in ER+ ($p \leq 0.05$). In ER+ tumors the protein expression of E-cadherin was predominantly membranous (Figure 5A), whereas in TNBCs E-cadherin protein expression was often lost from the membrane and pre-dominantly cytoplasmic (Figure 5B). Membranous E-cadherin protein expression was higher in ER+ than in TNBCs ($p \leq 0.001$), whereas cytoplasmic E-cadherin protein ex-pression was higher in TNBCs when compared with ER+ ($p \leq 0.001$) (Figure 4A). Overall, the expression of this protein was quite heterogeneous across the samples. Interestingly, a strong negative correlation between membranous and cytoplasmic E-cadherin protein expression was found in ER+ (r = −1, $p \leq 0.001$) (Figure 4B) and in TNBCs (r = −0.9, $p \leq 0.001$) (Figure 4C). CD44 protein expression was lower in ER+ (Figure 5C) than in TNBCs (Figure 5D), although not statistically significant. Notably, in TNBCs, a strong positive correlation between CK$\frac{5}{6}$ and CK14 expression ($r = 0.8$, $p \leq 0.01$), and a moderate positive correlation between CD44 and vimentin ($r = 0.6$, $$p \leq 0.05$$), were found.
All CMTs (Figure 4D) were positive (>$1\%$) for ER and, therefore, classified as ER+. ER protein expression was lower in STC II than in STC I ($p \leq 0.01$). The protein expression of E-cadherin was quite heterogeneous across the samples. As in HBC, a strong negative correlation between membranous and cytoplasmic E-cadherin protein expression was found in the CMTs (r = −0.974, $p \leq 0.001$) (Figure 4E). In addition, in STC II, a strong positive correlation between CK$\frac{8}{18}$ and membranous E-cadherin ($r = 0.8$, $p \leq 0.01$) and a strong negative correlation between CK$\frac{8}{18}$ and cytoplasmic E-cadherin (r = −0.8, $p \leq 0.01$) were found. Interestingly, in STC II, Ki-67 expression was positively correlated with CK$\frac{8}{18}$ ($r = 0.7$, $p \leq 0.05$) and membranous E-cadherin ($r = 0.8$, $p \leq 0.01$) expression, and negatively correlated with cytoplasmic E-cadherin expression (r = −0.7, $p \leq 0.05$).
All FMTs (Figure 4D) were negative for ER (<$1\%$), PR (<$1\%$), and ERBB2 (either 0 or 1+), and were therefore classified as triple negative. E-cadherin protein expression was quite heterogeneous. As in the HBCs and CMTs, a strong negative correlation between membranous and cytoplasmic E-cadherin protein expression was found (r = −0.984, $p \leq 0.001$) (Figure 4F). In addition, a strong negative correlation between CK$\frac{5}{6}$ and vimentin expression was found ($r = 0.8$, $p \leq 0.01$).
CD44 protein expression was higher in the CMTs (Figure 5E) than in the FMTs ($p \leq 0.001$) (Figure 5F). Vimentin and Ki-67 protein expression was lower in the CMTs than in the FMTs ($p \leq 0.001$) (Figure 6).
The expression of the studied markers was not associated with other histopathological features, such as vascular invasion or regional lymph node metastases (data not shown). Moreover, no significant correlations were found between gene and protein expression of the analyzed markers.
## 4. Discussion
In this study, we investigated the expression of genes and proteins involved in one of the processes thought to play a major role in cancer progression: epithelial-to-mesenchymal transition [22].
EMT is an evolutionally conserved morphogenetic program during which epithelial cells undergo a series of changes allowing them to acquire a mesenchymal phenotype [21]. During classical EMT, epithelial cells lose the expression of tight junction molecules such as membranous E-cadherin and acquire mesenchymal properties such as migration, invasiveness, and elevated resistance to apoptosis. Transcription factors like SNAIL, TWIST, and ZEB regulate this process and are activated by a variety of signaling pathways, including TGF-α, Notch, and Wnt/β-catenin [30,31,32,33].
SNAIL is a classical regulator of EMT that represses E-cadherin transcription in both mouse and human cell lines [34]. In HBC, it has been associated with tumor recurrence and metastasis [35], and with poor patient prognosis [36]. In contrast to the findings of other authors [37], we found that the mRNA expression of SNAIL2 was significantly lower in TNBCs than in healthy MGTs. In CMTs, SNAIL1 expression was higher in STC II when compared with healthy MGTs and STC I, indicating a possible association of EMT with a higher aggressiveness of these tumors. SNAIL2 in CMTs did not show any difference between healthy MGT and tumor tissue, confirming what other authors have also found [38,39,40]. Conversely, in FMTs, there was a trend such that STC III had a lower mRNA expression of SNAIL1 and SNAIL2 when compared with healthy MGTs. To the best of our knowledge, SNAIL has never been investigated in feline tumors.
It is believed that TWIST plays an essential role in cancer metastasis [33]. In HBCs and FMTs, the mRNA expression of TWIST1 and TWIST2 was lower in tumors than in healthy MGTs, which differs from what some authors have found in HBC [41], but is similar to what other authors have found in HBC [42] and in FMTs [43].
ZEB1 has been implicated in carcinogenesis in breast tissue [44] because it enhances tumor cell migration and invasion [45]. In our samples, ZEB1 mRNA expression was lower in tumor than in healthy MGTs, as previously reported by other authors in HBC [42]. Although one study examined the expression of ZEB1 and ZEB2 in five canine mammary carcinoma cell lines [46], to the best of our knowledge, ZEB mRNA expression has never been studied in CMT and FMT tissues.
Overall, our data suggest that these transcriptional factors are often downregulated in tumors compared with healthy MGTs, except for SNAIL1 in TNBCs and in CMTs STC II. The RNA isolated from healthy tissues came from the whole mammary gland, which is composed of different cell populations, namely epithelial cells, connective tissue, and fat. Although these transcription factors are barely detectable in normal mesenchymal cells of adult tissues [47], adipose tissue expresses these genes variably [48]. As a result, the mRNA levels of these genes in healthy samples can be dramatically influenced by the presence of non-mammary gland tissues, such as fat.
Moreover, it is possible that the number of cells undergoing classical EMT is low when compared with the tumor bulk, which is known to be characterized by a remarkable intra-tumor heterogeneity [22]. Furthermore, some authors believe that these genes are regulated post-transcriptionally [35,49,50,51]. Furthermore, accumulating evidence suggests the existence of cell populations with a hybrid E/M state, which exhibit increased plasticity and metastatic potential, characterized by the co-expression of epithelial and mesenchymal markers [23,24,25,52]. However, the expression of some of these markers may be associated with a complete EMT status, whereas others may be associated with a partial EMT status. For example, it is believed that SNAIL1 is a stronger inducer of complete EMT than SNAIL2, which is rather associated with a hybrid E/M state [53,54]. This suggests that the choice of the markers to be analyzed is fundamental and may help in identifying intermediate EMT states more precisely. In addition, in order to study the EMT process, it would be interesting in the future to investigate the expression of these markers at a single cell level, using single-cell omics approaches such as Laser Capture Microdissection or single-cell RNA sequencing.
In the present study, we also assessed the protein expression of several phenotypic as well as EMT-related markers, such as ER, PR, ERBB2, CK$\frac{8}{18}$, CK$\frac{5}{6}$, CK14, E-cadherin, CD44, vimentin, and Ki-67, in a subset of HBCs, CMTs, and FMTs.
The HBC ER+ samples showed a high expression of luminal CK$\frac{8}{18}$, and a negative expression of basal CK$\frac{5}{6}$ and CK14, confirming the strong association between ER+ tumors and highly differentiated glandular cells (CK$\frac{8}{18}$+), as well as null expression of basal CKs (CK$\frac{5}{6}$, CK14). In the TNBCs, the protein expression of CK$\frac{8}{18}$ was highly heterogeneous, whereas the expression of CK$\frac{5}{6}$ and CK14 was low in most of the samples. This result, in concordance with another study [55], supports the idea that the terms “basal-like cancer” and “triple-negative breast cancer” are not interchangeable. Indeed, only a small percentage of TNBCs are basal-like [56]. The CMTs were positive for ER, whereas the FMTs were negative for ER, PR, and ERBB2. Despite only a few samples being analyzed, these data suggest, as already proposed by other authors [11,57], a similarity between CMTs and HBC ER+ and between FMTs and TNBCs. In CMTs and FMTs, the protein expression of CK$\frac{8}{18}$, CK$\frac{5}{6}$, and CK14 was highly heterogeneous, confirming the high inter- and intra-tumor heterogeneity [16,57]. Basal CK14 protein expression was higher in FMTs than in CMTs, confirming that FMTs are more “basal-like” when compared with CMTs [11,12].
E-cadherin is a cellular adhesion molecule, and its disruption may contribute to the enhanced migration and proliferation of tumor cells, leading to invasion and metastasis [58,59,60,61,62]. In our samples, E-cadherin protein expression was evaluated in the membrane and in the cytoplasm of tumor cells, separately. Overall, the expression of E-cadherin was highly heterogeneous across the samples of the three species, confirming once more the high inter-tumor heterogeneity of mammary cancer in the three species. In human ER+ tumors, E-cadherin protein expression was predominantly membranous, whereas in TNBCs it was predominantly cytoplasmic, confirming that the delocalization of the protein is associated with increased tumor aggressiveness [56,63]. These results confirm that it is not only the loss of E-cadherin that correlates with increased tumor aggressiveness, but also the protein translocation from the membrane to the cytoplasm, as already described [64,65,66,67].
Together with E-cadherin, CD44 has been extensively studied in tumor cell differentiation, invasion, and metastasis, and is thought to be involved in the EMT process in HBC [68,69]. Although a few studies on HBC have shown that protein overexpression of CD44 is associated with poor prognosis and metastasis [70], others have shown that downreg-ulation of its expression is correlated with an adverse outcome [68,71]. For this reason, the role of CD44 in the behavior and prognosis of HBC is controversial [71,72]. In our study, CD44 expression was heterogeneous and lower overall in ER+ tumors compared with TNBCs. This trend agrees with study findings by Klingbeil and collaborators, who found high levels of CD44 expression in tumors with a basal-like or triple-negative phenotype, suggesting an association of this protein with an aggressive phenotype in HBC [73]. CD44 was highly expressed (roughly $85\%$) in our CMT samples, regardless of the tumor grading, as well as in the healthy mammary gland tissues. Moreover, other authors found no differences between benign CMTs, malignant CMTs, and normal mammary gland tissues, suggesting that CD44 is not associated with aggressiveness in canine mammary tumors [74,75,76,77,78]. In FMTs, the expression of CD44 was low overall (approximately $5\%$). Sarli and collaborators evaluated the intramammary/intratumoral and extramammary/extratumoral expression of CD44 in feline normal mammary tissues, benign tumors, and malignant tumors in relationship to lymphangiogenesis [79]. They found that CD44 had a significantly higher expression in intramammary/intratumor areas compared with extramammary/extratumor areas in both benign and malignant tumors. Additionally, no statistically significant differences in CD44 expression between normal mammary gland, benign tumors, and malignant tumors were found. To the best of our knowledge, no other studies on CD44 expression in FMT tissues are present within the literature. These data, together with our findings, suggest that CD44 is not a useful marker of malignancy in cats.
Another protein that is well-studied and plays a central role in the EMT process, and therefore in tumor invasion and metastasis, is vimentin [51]. Vimentin is one of the major intermediate filament proteins and is ubiquitously expressed in normal mesenchymal cells [80]. Recent studies have reported that vimentin knockdown causes a decrease in genes linked to HBC metastasis, such as the receptor tyrosine kinase Axl [81]. In our study, we also evaluated the expression of vimentin in HBCs, CMTs, and FMTs. We found a higher expression of vimentin in TNBCs compared with ER+, although not statistically significant. This result suggests that vimentin expression is associated with the triple-negative subtype, aggressive behavior, and a poor prognosis of HBC, as previously reported by many authors [82,83,84,85]. In CMTs, vimentin expression is low (approximately $15\%$), con-firming the low aggressiveness of mammary tumors in dogs, which is in concordance with the findings of other authors [86]. Conversely, in FMTs, the expression of vimentin, although heterogeneous, was quite high (approximately $70\%$), suggesting the high aggressiveness of mammary tumors in this species [9], as well as their similarities with TNBCs [11].
Unfortunately, as a limitation of this study, only grade I and II CMTs were included. No RNALater-sampled canine tumors were diagnosed as grade III. For possible IHC analyses in our archive of paraffin-embedded tissues, a very limited number of grade III simple CMTs were found (14 cases over five years) that were often already vascular/lymph node invasive ($\frac{10}{14}$). This study would not benefit much from adding only IHC analysis of grade III CMTs that already have invaded the vascular system or with metastases. We still believe that the study allowed the collection of some new data on the most frequent FMTs and CMTs in comparison with HBC samples assessing both gene and protein expression.
## 5. Conclusions
In summary, this study showed that most of the classical EMT-related transcription factors SNAIL, TWIST, and ZEB are downregulated in tumor tissues compared with healthy tissues, although additional analyses should be performed to better investigate them in neoplastic clones and in a larger set of samples. IHC analyses indicated a potential role of some markers, namely vimentin and E-cadherin, but not of others (i.e., CD44) as indicators of EMT (including loss of cell differentiation and increased malignancies). Moreover, all the IHC data seem to support the already proposed similarities between FMTs (grade III) and TNBCs, as well as between CMTs (grade I and II) and ER+ HBCs. The two species are widely discussed as potential spontaneous models of specific HBC subtypes [11,12,15,16,57,87,88,89,90]. |
# Effects of Different Phospholipid Sources on Growth and Gill Health in Atlantic Salmon in Freshwater Pre-Transfer Phase
## Abstract
### Simple Summary
Optimal nutrition is important for Norwegian-farmed Atlantic salmon in the challenging early seawater phase, which shows a higher mortality leading to significant economic losses. Phospholipids are reported to enhance growth, survival, and health in the early stages of the fish life. Atlantic salmon (74 to 158 g) were fed six test diets to evaluate alternative phospholipid (PL) sources in freshwater and were transferred to a common seawater tank with crowding stress after being fed the same commercial diet up to 787 g. Krill meal (KM) was evaluated using dose response with the highest $12\%$ KM diet compared against $2.7\%$ fluid soy lecithin and $4.2\%$ marine PL (from fishmeal) diets, which were formulated to provide the same level of added $1.3\%$ PL in the diet similar to base diets with $10\%$ fishmeal in the freshwater period. A trend showing increased weight gain with high variability was associated with an increased KM dose in the freshwater period but not during the whole trial, whereas the $2.7\%$ soy lecithin diet tended to decrease growth during the whole trial. No major differences were observed in liver histology between the salmon that were fed different PL sources during transfer. However, a minor positive trend in gill health based on two gill histology parameters was associated with the $12\%$ KM and control diets versus the soy lecithin and marine PL diets during transfer.
### Abstract
Growth and histological parameters were evaluated in Atlantic salmon (74 g) that were fed alternative phospholipid (PL) sources in freshwater (FW) up to 158 g and were transferred to a common seawater (SW) tank with crowding stress after being fed the same commercial diet up to 787 g. There were six test diets in the FW phase: three diets with different doses of krill meal ($4\%$, $8\%$, and $12\%$), a diet with soy lecithin, a diet with marine PL (from fishmeal), and a control diet. The fish were fed a common commercial feed in the SW phase. The $12\%$ KM diet was compared against the $2.7\%$ fluid soy lecithin and $4.2\%$ marine PL diets, which were formulated to provide the same level of added $1.3\%$ PL in the diet similar to base diets with $10\%$ fishmeal in the FW period. A trend for increased weight gain with high variability was associated with an increased KM dose in the FW period but not during the whole trial, whereas the $2.7\%$ soy lecithin diet tended to decrease growth during the whole trial. A trend for decreased hepatosomatic index (HSI) was associated with an increased KM dose during transfer but not during the whole trial. The soy lecithin and marine PL diets showed similar HSI in relation to the control diet during the whole trial. No major differences were observed in liver histology between the control, $12\%$ KM, soy lecithin, and marine PL diets during transfer. However, a minor positive trend in gill health (lamella inflammation and hyperplasia histology scores) was associated with the $12\%$ KM and control diets versus the soy lecithin and marine PL diets during transfer.
## 1. Introduction
Farmed salmon are typically transferred from early phase production in tanks on land to seawater cages that constitutes a challenging environment, where fish can experience significant mortality before reaching harvest size. For example, mortality in Atlantic salmon ranged from 15 to $16\%$ from 2017 to 2021 in Norway, with approximately $35\%$ of sea cage mortality occurring in the first 0–3 months at sea for the 2010–11 salmon generations in the Norwegian-farmed Atlantic salmon [1]. This mortality in the early sea cage phase leads to significant economic loss [2]. Thus, research on optimal nutrition to produce robust smolts for improved survival and growth after transfer to the sea cage is of interest to the aquaculture industry [3]. Fish meal (FM) and fish oil (FO) dominated early commercial salmon feed formulations and provided essential nutrients, but usage of these marine ingredients has declined over time as they are limited resources at generally higher prices compared to alternative ingredients where sustainability measures are also considered [4]. Antarctic krill meal (KM; Euphausia superba) is a commercially known ingredient in salmon feeds, with potential benefits toward enhancing growth and health in salmonids [5]. The krill fishery in the Antarctic Southern *Ocean is* considered highly regulated and sustainable [6,7]. KM provides a range of nutrients including proteins (similar amino acid profile to FM); water soluble nitrogenous components (free amino acids, peptides, nucleotides, and trimethylamine N-oxide), which can act as potential feed attractants; astaxanthin; marine omega-3 fatty acids (eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA)); and phospholipids (PLs) [5]. Substantial evidence exists showing that dietary PL can improve growth, survival, and health (reduced intestinal steatosis and deformities) in the larval and early juvenile stages of the fish [8,9,10,11]. In addition, KM and krill oil (KO) reduced fat accumulation in the hepatocytes in comparison to soybean lecithin as the PL sources in the diet of seabream larvae [10,12,13]. In addition, there was an indication that seabream juveniles that were fed a diet with $9\%$ KM had lower hepatocyte vacuolization (fat storage) versus a control diet without KM that was higher in fishmeal [12,13], and a non-significant trend for lower hepatocyte vacuolization was indicated for seabream larvae that were fed a diet with krill oil versus soybean lecithin as the PL source [10]. PLs from different sources can have different properties. KM has approximately $40\%$ PL consisting of the total lipid with phosphatidylcholine (PC) at >$80\%$ of the total PL and ca. $18\%$ EPA + DHA of the total lipid [14]. In comparison, fluid soy lecithin can have approximately $46\%$ PL of product (does not include glycolipids and complex sugars) with ca. $35\%$ PC of the total PL and ca. $55\%$ 18:2n-6 of the total FA as the major FA with no EPA + DHA [15]. KM has been documented in the diet of seawater salmon [16,17,18], however, only KO has been documented in the diet of freshwater salmon during the pre-transfer to the seawater phase [19]. The objective of the present study was to document the effect of the KM dose as a source of PL and compare it against other PL sources in the feed of freshwater Atlantic salmon during the pre-transfer phase followed by the early seawater phase by evaluating the growth and histological health parameters. A four-level graded dose response for KM up to $12\%$ of the diet along with a comparison of alternative PL sources (soy lecithin and marine PL from fishmeal) formulated to provide the same level of added $1.3\%$ PL in the diet as $12\%$ KM was evaluated in freshwater diets for salmon during the pre-transfer phase. Fish identified by pit tag with this pre-transfer freshwater feeding history were then transferred to a common seawater tank with crowding stress after transfer and a drop in water temperature at transfer (crowding and water temperature drop can be experienced at transfer commercially) and then were fed the same commercial feed. Gill and liver histology were also compared for salmon that were fed the alternative PL source diets at the end of the freshwater pre-transfer period.
## 2.1. Feed Formulation and Composition
Three different sources of PL were tested in pre-transfer freshwater feeds: (i) krill meal (QrillTM Aqua; Aker BioMarine Antarctic ASA) at four levels for dose response ($4\%$, $8\%$, and $12\%$ of diet), (ii) fluid soy lecithin as a vegetable PL source, and (iii) marine phospholipid-rich oil sourced from North Atlantic fish species from Triple 9 (TripleNine, Trafikhavnskaj 9, DK-6700 Esbjerg, Denmark))., and a control diet. The trial diets are referred to as Control, KM4, KM8, KM12, VegPL, and MarPL, respectively. Trial feeds were formulated using a commercial formulation program with external oil mix calculations and produced by extrusion at Cargill Innovation Center (Dirdal, Norway) for ca. 74 g fish with lipid nutrients and then adjusted for purposes of the trial. The 4-mm pre-transfer freshwater trial feeds were formulated and analyzed to have similar digestible energy (22.1–23.6 MJ/kg gross energy), protein (46–$49\%$ range), and fat (22–$24\%$ range) (Table 1) and with similar calculated $1.1\%$ EPA + DHA in diet, 15–$16\%$ saturated in total FA, and 1.3 n-6/n-3 fatty acid (FA) ratio across trial feeds. Protein was analyzed by the Dumas principle using the Elementar Rapid Max N system. Fat was analyzed by low-field nuclear magnetic resonance scan using the NMR Analyzer Bruker minispec mq10 system (Cargill Innovation Center, Dirdal, Norway). Gross energy was analyzed by the Leco gross energy bomb calorimetry system (Cargill Innovation Center, Dirdal, Norway). Moisture was predicted by the NIR FOSS DS2500 system (Cargill Innovation Center, Dirdal, Norway) by using the feed model at Cargill. A similar $1.3\%$ PL in diet across pre-transfer freshwater diets was calculated from the addition of $12\%$ krill meal, fluid soy lecithin, and marine PL test ingredients to base formulations with the same $10\%$ fishmeal level across the diets. There was variation in the other ingredients (added oil, plant ingredients, and micronutrients) needed for balancing or reaching nutrient targets. The choline level was formulated to be the same for control and VegPL diet with MarPL and KM12 providing additional choline to these diets in the form of phosphatidylcholine (PC). However, formulated choline levels for control diet and fluid soy lecithin diets were in excess of the NRC 2011 requirements for salmonids and in excess of the lowest choline level used by Hansen and coworkers [20] with no growth differences observed (1340 to 4020 mg choline/kg diet dose response trial for 456 g initial weight salmon). Lipid accumulation in the gut was reduced for salmon (456 g initial weight) at increased choline levels [20]. The formulation and composition of feeds are given in Table 1.
## 2.2. Fish Trial Conditions
The experiment was performed according to the guidelines and protocols approved by the European Union (EU Council $\frac{86}{609}$; D.L. 27.01.1992, no. 116) and by the National Guidelines for Animal Care and Welfare published by the Norwegian Ministry of Education and Research.
Atlantic salmon (Salmo salar) with an initial weight of ca. 67 g were used for the trial. The fish were pit-tagged and randomly distributed into 24 freshwater flow-through tanks (1 m diameter and 0.45 m3 volume) to have 40 fish per tank at the start of trial diet feeding. These fish after 15 days of tank acclimation were 74 ± 12 g (average ± SD for all 960 fish in 24 tanks at the start of trial feeding) and then were fed the freshwater pre-transfer trial diets (Table 1) over a 53-day period. Water temperature averaged 14.3 °C (13.3–15.3 °C range) with $107\%$ average oxygen saturation at the inlet and $90\%$ oxygen saturation at the outlet during the freshwater acclimation and trial diet feeding period. Fish were fed the six trial diets to four replicate tanks during the 53-day freshwater pre-transfer period using an automatic belt feeder with continuous feeding for 20 h per day in excess of satiation level. Feed intake was calculated on a weekly basis by collecting and weighing uneaten pellets as well as by weighing the amount fed. There was a 12 h light: 12 h dark photoperiod regime from Day 0 at freshwater tank acclimation to Day 33 after which a 24 h light regime was used to initiate smoltification. After this freshwater pre-transfer feeding period, fish from all the tanks (17–20 fish per tank from the 24 freshwater tanks) were transferred to a larger common seawater flow-through tank (5 m diameter and 21.6 m3 volume with 28.5 ppt salinity, and no acclimation time from 0 ppt freshwater to 28.5 ppt seawater) with a water temperature drop at transfer (ca. 14 to 9 °C) and crowding stress after transfer (lowered water level to ca. 20 cm for one hour with supplemental oxygen for all 459 fish of ca. 167 g within a ca. 0 to 30 h period after transfer) in the common seawater tank after all 17–20 fish per tank from the 24 freshwater tanks fish were transferred over and then were fed a common commercial extruded salmon diet (EWOS AS) for a further 98 days. Daily water temperature was lower during the seawater phase averaging 9.4 °C (8.5–11.1 °C range).
## 2.3. Fish Growth
The 40 fish per tank were weighed individually with pit-tag identification on acclimation to the freshwater tanks (Day 0), at the start of trial diet feeding (Day 15), at intermediate weighing (Day 33), and after 53 days of trial feeding in the freshwater (Day 68). The fish weight gain in the freshwater pre-transfer period from Day 15 (start of freshwater trial diet feeding) to 68 were compared statistically between diets. A total of 17–20 fish from each of the 24 freshwater tanks were transferred to the common seawater tank on Day 68 with fish weighing performed on Days 35, 73, and 98 after transfer to seawater. There were 9 to 17 fish representing the original tanks in the freshwater period with 50 to 58 fish representing each of the test diets from the freshwater period at final weighing in seawater at 98 days after transfer to the common seawater tank. The fish weight gain over the whole trial period in freshwater and seawater from Day 15 to 166 days were statistically compared between diets.
## 2.4. Hepatosomatic Index
Hepatosomatic index (HSI) is the liver weight percent of the whole body weight. HSI was measured on 10 fish randomly sampled per tank (four tank replicates per diet) to study 40 fish per diet at the end of the freshwater pre-transfer period when fed test diets and 40 fish per diet (identified by pit-tag) at the end of the seawater phase when fed the common commercial diet.
## 2.5. Histology
Gill and liver histology were performed on the fish involved in the dietary phospholipid source comparison (KM12, VegPL, and MarPL) and on fish fed the Control diet at the end of the freshwater pre-transfer period. Liver (half tissue section) and gill (left gill arch 2) tissues were randomly sampled from five fish per tank to give a total of 20 liver and 20 gill tissues per diet group for histological analysis. The tissues were fixed in formalin ($4\%$ formaldehyde) and stored at room temperature until sent to Pharmaq Analytiq AS (Harbitzallée 2A, 0275 Oslo, Norway) for histological analysis.
## 2.6. Statistical Analysis
The weight gain for the different periods was modelled by computing the weight gain of each tagged individual and then using a hierarchical generalized additive model (GAM) with the spline function to describe the possibly non-linear dose-response. A random effect of tank was added to the model to account for the multiple individual observations per experimental unit. The total feed intake over the periods of interest was modelled with a single level GAM with a spline function describing the dose-response function. Hepatosomatic index (HSI) was modelled by a hierarchical GAM model using a spline function to describe the dose-response function, mean-centered round weight of the fish as a covariate, and a random effect of tank to account for the multiple individual observations per tank. From this model the expected liver weight was solved for an average-sized sampled fish and expressed as HSI by dividing the expected liver weight with the mean round weight of the sample. Gill and liver histology scores are ordinal variables for which common arithmetic operations, such as sum or mean, are not defined and therefore scores require an ordinal model returning the score probability for evaluation. A hierarchical GAM for ordinal data was set up by using a spline function to describe the dose-response function, and a random effect of tank was included to account for multiple individuals observed per tank. The models for weight gain, feed intake, and HSI assumed the error distribution is the normal distribution, and the model for gill and liver scores assumed the model is ordinal and the errors followed the ordered categorical family. All data processing and statistical modelling was conducted with the R language [21]. The GAMs were estimated with the “gam” function of the R language add-on package “mgcv” [22].
The outcomes from the fitted statistical models are presented graphically by showing the mean response and the $95\%$ credible intervals. The mean (median) response and the $95\%$ credible intervals were computed with the help of a parametric bootstrap (with 10,000 random draws per parameter) by taking the $25\%$, $50\%$, and $97.5\%$ quantiles of the computed response vector. In the case of a categorical predictor variable (for comparing the different PL sources), the graphs show the mean and an error bar of the $95\%$ credible interval. In the case of a continuous predictor (for the dose-response of krill meal inclusion), the mean response is shown as a median dose-response curve and the $95\%$ credible interval is shown as a confidence band around the mean curve. This way both the magnitude of any potential effect (biological significance) and the uncertainty of any effect estimate (statistical significance) can be shown in the same graph for all the results independent of the response following the normal, binomial, or ordered categorical distribution.
## 3.1. Growth Performance
Atlantic salmon of 74 g (overall tank average) were fed the six test diets up to 158 g (overall tank average), growing 2.1-times the initial fish weight to the end of the freshwater pre-transfer period. There was no clear trend for increased feed intake with KM dose in the FW pre-transfer phase (Figure 1). A trend for increased feed intake was indicated for the Control and KM12 diets compared to the MarPL and VegPL diets in the PL source comparison for the FW pre-transfer phase (Figure 2). There was overall high variability for the feed intake comparisons. A trend for increased fish weight gain with high variability was indicated with increased KM dose in the FW phase (Figure 3). There was similar weight gain during the whole trial with feeding the KM dose in the FW pre-transfer phase followed by feeding the same commercial diet in a common tank for the SW phase (Figure 4). Fish fed the KM12 diet had increased weight gain compared to the VegPL diet with the MarPL and Control diets having intermediate weight gains in the PL source comparison for the FW pre-transfer phase (Figure 5). Weight gain was similar for the fish that were fed KM12, MarPL, and Control diets, with a trend for higher indicated weight gain than the VegPL group during the whole trial, with feeding the KM dose in the FW pre-transfer phase followed by feeding the same commercial diet in a common tank for the SW phase (Figure 6, Tables S1 and S2).
## 3.2. Hepatosomatic Index
A trend for decreased hepatosomatic index (HSI; liver% of fish weight) was indicated for the fish that were fed increased KM dose from 0 to $12\%$ of diet at the end of the freshwater pre-transfer feeding phase (Figure 7). There was no decrease in HSI with feeding KM dose at the end of the whole trial after the FW pre-transfer phase followed by feeding the same commercial diet in a common tank for the SW phase (Figure 8). A lower HSI was indicated for the fish that were fed the KM12 diet compared with the fish that were fed the MarPL, VegPL, and Control diets at the end of the freshwater pre-transfer feeding phase (Figure 9) with a similar minor HSI trend observed over the whole trial (Figure 10).
## 3.3.1. Gill Histology
An increased probability for very mild to mild gill lamella inflammation and hyperplasia score was indicated for the salmon that were fed the VegPL and MarPL diets compared to the Control and $12\%$ KM diets at the end of the freshwater pre-transfer phase after 53 d of feeding the trial diets (Figure 11a,b). Other following gill histology responses were evaluated with no major differences between the diets: vascular lesions, filament inflammation, necrosis of respiratory epithelium, necrosis affecting deeper tissues, fusion of lamella,and other lesions noted as present or absent.
## 3.3.2. Liver Histology
No major differences were observed in liver histology between the control, $12\%$ KM, soy lecithin, and marine PL diets at the end of the FW pre-transfer phase after 53 d of feeding the trial diets (data not shown). The following liver histology responses were evaluated: total amount of abnormal tissue, inflammation, necrosis, inflammation in liver tissue or capsule (peritonitis), peribiliary or perivascular inflammation, neoplasia, fibrosis, lipid deposition, other degenerative changes, vascular lesions, and other lesions noted as absent or present.
## 4. Discussion
The present study evaluated the effect of different phospholipid sources fed over 53 d in the freshwater pre-transfer phase followed by feeding the same commercial diet over 98 d in a common seawater tank on growth performance and health parameters of Atlantic salmon. KM was evaluated in dose response ($4\%$, $8\%$, and $12.0\%$ of diet), and diets with $2.7\%$ fluid soy lecithin (VegPL) and $4.2\%$ MarPL as alternative PL sources were formulated to provide the same level of added $1.3\%$ PL in diet as $12\%$ KM. All the test diets contained $10\%$ fishmeal in the FW phase. A trend was indicated for increased fish weight gain (high variability) with increased KM dose in the FW pre-transfer phase but a carry-over effect on growth was not observed for the same salmon fed the same commercial diet after seawater transfer. Salmon (104 g initial weight) that were fed krill meal at 7.5 and $15\%$ of diet for higher fishmeal diets (40–$52\%$ of diet range) than the current trial had increased growth after transfer to sea cage [16]. Fishmeal provides PL, so higher fishmeal diets may reduce the need for KM as a PL source [23]. However, KM also provides amino acids (protein), water-soluble nitrogenous components (potential feed attractants), astaxanthin, and EPA + DHA, hence, it is more than a PL source. KM feeding may need to continue after sea water transfer to have a positive effect on growth at the end of the trial, noting the positive effects of KM on salmon growth observed in other but not all trials, which can depend on life stage and challenges, diet composition, KM refining (de-shelling etc.), and inclusion level [5].
A trend for decreased fish weight gain was indicated for the VegPL diet in the FW phase and over the whole trial compared with the control diet, whereas the MarPL diet showed more similar growth to the control diet over the whole trial, noting that only one PL level tested for MarPL and fluid soy lecithin matched that provided by KM12, so optimal dose was not evaluated. The choline level was formulated to be the same for the control and VegPL diets with KM12 and MarPL providing additional choline to these diets in the form of phosphatidylcholine (PC). Formulated choline levels for the control diet and fluid VegPL diets were in excess of the NRC 2011 requirements for salmonids and in excess of the lowest choline level used by Hansen et al. in 2020 with no growth differences observed (1340 to 4020 mg choline/kg diet dose response trial for 456g initial weight salmon) [20]. Lipid accumulation in the gut was reduced for these salmon (456 g initial weight) at increased choline levels [20]. Effects of increased choline with KM inclusion cannot be ruled out and further research would be needed to separate choline from PL effects for these smaller pre-transfer salmon (74 to 158 g fish weight) that were fed lower fat pre-transfer diets (22–$24\%$ fat) than during the seawater growth with choline requirements for reducing the lipid accumulation in the intestine, potentially dependent on dietary fat level [20]. Higher growth was generally observed for PL provided by KO over soy lecithin at various PL doses for the first-feeding stage of salmon, but this growth trend was not consistent at various PL doses over the whole trial from the first-feeding to smolt [19]. PL from KO was indicated to be more effective than fluid soy lecithin for reducing intestinal steatosis in smaller salmon (2.5 g salmon, but no steatosis observed across diets for 10–20 g salmon) and low level of vertebral deformities [19]. Marine PL sources (FM and KO) were also compared against soy lecithin at a similar ca. $3.5\%$ PL of diet level for the first-feeding Atlantic salmon (0.14 g initial weight) with these PL sources, giving similar growth to ca. 2.4 g final fish weight with no conclusive mortality or intestinal histology differences between PL sources but these parameters were generally improved for the PL source diets with higher PL compared to the control diets with lower PL. An uncertain observation of higher average growth was indicated for the marine PL sources over soy lecithin at intermediate weighing for salmon at ca. 0.6 g [24]. Effects of PL cannot be isolated from KM but the increased growth for KM12 over the VegPL diet in the pre-transfer phase may be due to PL, choline, water soluble nitrogenous components, etc., noting that there was also an indicated trend for decreased growth of VegPL versus the control diet in the pre-transfer phase.
Addition of KM did not give a clear increase in feed intake compared to the control diet and there was an indicated trend of decreased feed intake for the MarPL and VegPL diets, but strong conclusions cannot be made due to the high variability. Feed intake can only be measured on a tank basis, so it was not possible to estimate feed intake of fish with different pre-transfer freshwater feeding histories in a common tank that were fed the same diet in the seawater phase.
A trend for decreased hepatosomatic index (HSI) was indicated with increased KM inclusion and for the $12\%$ KM diet versus the other PL sources added to provide the same PL level in the pre-transfer phase, but the effect of KM on decreasing HSI was not carried over into the seawater phase with fish that were fed the same diet in a common tank (Figure 7, Figure 8, Figure 9 and Figure 10). There was no difference in the liver lipid droplet accumulation based on histology (normal scores only) for salmon that were fed the diets containing different PL sources at the end of the freshwater pre-transfer period. The lower HSI in KM12 could be due to the positive effects from krill PL (and choline) on the lipid transport and deposition in organs, with this effect of feeding $12\%$ KM to Atlantic salmon documented by [17] with less pale livers and reduced liver fat. The authors further supported this observation with a significantly higher expression of the cadherin 13 (Chd) gene in the $12\%$ KM group associated with circulating levels of the adipocyte-secreted protein adiponectin that has potential anti-inflammatory effects and plays an important role in metabolic regulation and is associated with the fatty liver index in humans [25]. However, Chd expression was not studied in the current study, and hence, further studies are warranted to explore the association between Chd expression, his, and absolute fat accumulation in the liver in salmon. Increased choline, which KM provided in this trial, was shown to reduce fat accumulation in the intestine of Atlantic salmon [20]. Choline supplementation was also indicated to reduce HSI in Atlantic salmon, but this was not reflected in lower liver fat or histological vacuolization, noting that there are variable trends of dietary choline deficiency on the liver fat level of fish reported in the literature [26]. PL from KO was indicated to be more effective than fluid soy lecithin for reducing intestinal steatosis in smaller salmon (2.5 g salmon but no steatosis observed across diets for 10–20 g salmon). Further studies are required to associate higher liver fat with welfare in salmon.
Gills are one of the most vital organs of fish, due to their function in respiration, osmoregulation, excretion of nitrogenous waste, pH regulation, and hormone production [27]. Gill health has become one of the most significant health and welfare challenges in the salmon aquaculture industry in Norway, Scotland, and Ireland [28,29,30]. The gill disorders are generally complex and multifactorial and are related to both biological factors, such as parasites and pathogens, handling stress, treatments, or due to the environmental factors, such as temperature, salinity, algal blooms, etc. Hence, the gill diseases are challenging to prevent and control and lead to high mortality, reduced production performance, and impaired fish welfare, cumulating in huge economic losses [31]. There were no differences reported for histological parameters investigated except in the presence of ectopic epithelial cells containing mucus in the lamina propria in the hindgut (potential inflammatory marker) of salmon (grown from 2.3 to 3.9 kg in sea cages) that were fed $15\%$ fishmeal diet but not for $12\%$ KM of diet in a $5\%$ fishmeal diet, which may suggest anti-inflammatory effects of KM [17]. KM provides astaxanthin (166 mg/kg in the KM used for the present study) to the diet as a natural antioxidant with potential anti-inflammatory properties [32]. KM and MarPL also provide EPA + DHA attached to PL, which may affect bioavailability of EPA + DHA for use in cell membranes and inflammatory response [33] but this is not documented in fish. In the current study, there was decreased probability for very mild to mild gill lamella inflammation and hyperplasia scores indicated in salmon that were fed $12\%$ KM compared to the soy lecithin and marine PL diets but gill histology for salmon that were fed the $12\%$ KM diet was similar to the control diet without KM (Figure 5).
## 5. Conclusions
Overall, increased KM tended to increase growth (high variability), whereas the VegPL diet tended to decrease growth compared to the control diet in the FW pre-transfer phase. The positive growth trend indicated for KM fed pre-transfer was not carried over into the seawater phase for fish fed the same diet. A minor positive trend in gill health (lamella inflammation and hyperplasia histology scores) was indicated for the $12\%$ KM and Control diets compared with the VegPL and MarPL diets in the FW pre-transfer phase. Hepatosomatic index tended to decrease with KM fed in the pre-transfer phase, noting that all livers evaluated by histology were considered normal for lipid droplet accumulation. Only one VegPL and MarPL dose was tested, so dose effect of these PL sources and comparison with krill oil to better isolate the PL effect from other nutrients in KM as well as a post-transfer feeding comparison of these PL sources could be areas to research further in transfer diets for salmon. |
# Effects of Bacillus licheniformis and Combination of Probiotics and Enzymes as Supplements on Growth Performance and Serum Parameters in Early-Weaned Grazing Yak Calves
## Abstract
### Simple Summary
This study was conducted to investigate the effects of dietary supplementation with *Bacillus licheniformis* and a combination of probiotics and enzymes on the growth and blood parameters of grazing yak calves. The body weight, body size, serum biochemical parameters, and growth hormone levels of grazing yaks were assessed. We found that supplementation with probiotics alone or with a combination of probiotics and enzymes significantly increased the average daily gain, compared to the controls, and the combination of probiotics and enzymes showed a better performance. Supplementation with the complex of probiotics and enzymes significantly increased the concentration of serum growth hormone, insulin-like growth factor-1, and epidermal growth factor, which may be the main reason for the higher daily weight gain. The findings of this study may help improve the growth efficiency of yak calves on the Qinghai–Tibetan Plateau.
### Abstract
Early weaning is an effective strategy to improve cow feed utilization and shorten postpartum intervals in cows; however, this may lead to poor performance of the weaned calves. This study was conducted to test the effects of supplementing milk replacer with *Bacillus licheniformis* and a complex of probiotics and enzyme preparations on body weight (BW), size, and serum biochemical parameters and hormones in early-weaned grazing yak calves. Thirty two-month-old male grazing yaks (38.89 ± 1.45 kg body weight) were fed milk replacer at $3\%$ of their BW and were randomly assigned to three treatments ($$n = 10$$, each): T1 (supplementation with 0.15 g/kg Bacillus licheniformis), T2 (supplementation with a 2.4 g/kg combination of probiotics and enzymes), and a control (without supplementation). Compared to the controls, the average daily gain (ADG) from 0 to 60 d was significantly higher in calves administered the T1 and T2 treatments, and that from 30 to 60 d was significantly higher in calves administered the T2 treatment. The ADG from 0 to 60 d was significantly higher in the T2- than in the T1-treated yaks. The concentration of serum growth hormone, insulin growth factor-1, and epidermal growth factor was significantly higher in the T2-treated calves than in the controls. The concentration of serum cortisol was significantly lower in the T1 treatment than in the controls. We concluded that supplementation with probiotics alone or a combination of probiotics and enzymes can improve the ADG of early-weaned grazing yak calves. Supplementation with the combination of probiotics and enzymes had a stronger positive effect on growth and serum hormone levels, compared to the single-probiotic treatment with Bacillus licheniformis, providing a basis for the application of a combination of probiotics and enzymes.
## 1. Introduction
Yaks (Bos grunniens) occur on the Qinghai–Tibet Plateau at high altitudes and with long cold seasons and limited pasture resources. This species is a unique product of long-term natural selection, providing local herders with the most basic living materials and livelihood resources, such as meat, milk, shelter (hides and furs), and fuel (dung), and is an indispensable part of the ecology and economy of the Qinghai–Tibetan Plateau [1]. However, the low reproductive rate of yaks seriously restricts their production and utilization. The cold season on the Tibetan Plateau lasts for eight months (October to the following May), during which time the quantity and quality of pasture decrease below the nutritional requirements of lactating yaks [2]. The deficiency of feed intake results in a negative body energy balance and metabolic stress [3]. On the other hand, under traditional grazing management, plateau-grazing yak calves are weaned naturally or artificially under various conditions at an age of 18–24 months [4], rather than the weaning age of domestic beef cattle (<6 months). The slow recovery itself and the late weaning of yak calves, which result in a poor postnatal physical condition, severely affect the onset of the next estrous cycle in the cow. Most yaks exhibit a long postpartum anestrous period and calve twice every 3 years or once every 2 years [5]. Therefore, the early weaning of yak calves may help mitigate these adverse effects.
Early weaning has become more popular in recent years for various reasons, including the better use of limited feed resources and alleviating grazing pressure on pastures by reducing the nutritional needs of cows [6]. Weaning calves before the start of the breeding season improves the reproductive performance of cows [7,8] because the cows can regain their weight faster, thus accelerating the onset of postpartum estrus. The use of milk replacer in early weaning is common in livestock production [9,10]. The milk replacer has demonstrated positive benefits in animal experiments, such as improved immunity and relieved weaning stress response [11]. Increasing evidence suggests that enhanced milk replacer feeding is beneficial for improving gut microbial development and growth performance in early-weaned lambs [12,13].
Over the past few decades, probiotics have been widely used in livestock and poultry production for their ability to enhance animal disease resistance, improve feed utilization, and improve growth performance [14]. In ruminants, yeasts and bacteria, including Lactobacillus, Bifidobacterium, Bacillus, Propionibacterium, and Enterococcus, alone or in combination, are used as additives in diets [15,16]. Probiotics can decrease diarrhea, improve production and feed utilization efficiency, and strengthen the immunity system in young ruminants [17,18,19]. Moreover, supplementation with probiotics improves the rumen and intestinal epithelial cell growth, which enhances the gastrointestinal tract development and health status of calves [17,20,21]. Oral administration of *Bacillus licheniformis* can increase ruminal digestibility and total volatile fatty acid concentrations in Holstein cows [22] and growth performance in Holstein calves [23]. In vitro inoculation with *Bacillus licheniformis* also improves ruminal fermentation efficiency of forage of various qualities [24]. However, no information is currently available on the effect of *Bacillus licheniformis* on the growth performance of yak calves.
Compound enzyme preparations are produced from one or more preparations containing a single enzyme as the main entity, which is mixed or fermented with other single enzyme preparations to form one or more microbial products [25], including saccharylases, amylases, cellulases, proteases, phytases, hemicellulases, and pectinases. Depending on the differences in digestive characteristics and diet composition, specific enzyme preparations can be used for livestock [26]. Specific enzyme complex preparations can degrade multiple feed substrates (antinutrients or nutrients), and different types of enzymes can work synergistically to maximize the nutritional value of feed [27]. In buffalo calves, cellulase and xylanase are more effective with regard to average daily weight gain (ADG) and feed efficiency [28]. Further, the addition of exogenous fibrolytic enzymes to wheat straw has no effect on starter feed intake and increases nutrient digestibility and recumbency, but decreases the ADG of weaned Holstein dairy calves [29].
The effects of probiotics or compound enzyme preparations on the production performance and biochemical blood indexes of calves are not consistent [29,30,31,32,33]. The respective discrepancies may be due to differences in the amounts of added probiotics and exogenous enzymes, the strains of probiotics, diets, and animal management strategies. Therefore, this study was conducted to compare the effects of *Bacillus licheniformis* and a combination of probiotics and enzymes on the growth performance and serum parameters in yak calves, so as to provide a theoretical basis for the application of probiotics in grazing yak calves.
## 2.1. Animals and Treatment
This study was performed in accordance with the Chinese Animal Welfare Guidelines, and the experimental protocols were approved by the Animal Care and Ethics Committee of the Institute of Animal Husbandry and Veterinary Medicine, Tibet Academyof Agriculture and Animal Husbandry Science (No. # TAAAHS-2016–27).
The feeding trial was conducted at Damxung Co., (Lhasa, China; 30.5° N, 91.1° E) from July to October. The average altitude was 4200 m, the average annual temperature was 1.3 °C, and the average annual precipitation was 456.8 mm. Thirty two-month-old male yaks (38.89 ± 1.45 kg body weight (BW)) were fed milk replacer solution at $3\%$ of their BW every day and were randomly assigned to three dietary supplementation treatments ($$n = 10$$, each), according to BW and age, as follows: T1, supplemented with 0.15 g/kg *Bacillus licheniformis* (2 × 1010 CFU/g); T2, supplemented with a 2.4 g/kg combination of probiotics and enzymes (containing 0.4 g/kg Bacillus licheniformis, 2 × 1010 CFU/g; 1.0 g/kg yeast, 1 × 1010 CFU/g; 1.0 g/kg mixture of xylanase, cellulase, and glucanase in a 1:1:1 ratio, xylanase, 20,000 U/g, cellulase, 1500 U/g, glucanase, 6000 U/g); and a control treatment. The milk replacer, probiotics, and enzyme preparations were provided by the Chinese Academy of Agricultural Sciences (Beijing, China). All yak calves were allowed to graze on an alpine meadow during daytime for the 60-day trial, and they were individually fed milk replacer before and after grazing (0800 and 2000 h, respectively). The forage of the alpine meadow was mainly composed of Kobresia tibetica, and the nutrient composition (dry matter basis) was analyzed in our previous study [34], i.e., $10.4\%$ crude protein, $2.1\%$ ether extract, $67.8\%$ neutral detergent fiber, $34.2\%$ acid detergent fiber, and $4.6\%$ ash. The powdered milk replacer was weighed and mixed with warm water (approximately 40 °C) at a ratio of 1:7 (w/v) to obtain milk replacer solution, according to our previous study [35]. Based on preliminary assessments, the feeding amount of milk replacer was calculated so that all yak calves were able to feed without surplus [35]. The nutrient composition of the milk replacer is shown in Table 1.
## 2.2. Sample Collection and Analysis
The BW of each yak calf was recorded before morning feeding on d 0, 30, and 60 using a platform scale, and the ADG was calculated accordingly. The body size indexes of all yak calves were determined using a linen tape at the beginning (d 0) and end (d 60) of the experiment, as previously described [36].
Blood samples (approximately 10 mL) were collected from the jugular vein of the yak calves using a vacuum tube before morning feeding on d 0 and 60. The blood samples were centrifuged at 1100× g for 10 min to obtain serum, which was then aliquoted in 1.5 mL centrifuge tubes and stored at −20 °C.
The serum biochemical parameters, including blood urea nitrogen (BUN), globulin (GLB), blood glucose (GLU), and non-esterified fatty acids (NEFAs), were analyzed using an automatic biochemical analyzer 7020 (Hitachi, Tokyo, Japan). Metabolic hormones in the serum, including insulin-like growth factor-1 (IGF-1), epidermal growth factor (EGF), cortisol, insulin (INS), and growth hormone (GH), were determined using commercial ELISA kits (Jiahong Technology Co., Ltd., Beijing, China) according to the manufacturer’s instructions. Briefly, 50 μL of each five-fold diluted serum sample was added to each well of a 96-well ELISA plate. After 30 min of incubation at 37 °C, the plate was washed five times using PBS (Servicebio, Wuhan, China) to remove unbound proteins. Then, 50 μL of HRP-conjugated antibodies was added to allow them to bind with their corresponding antigens. The 3,3′,5,5′-tetramethylbenzidine working solution was added to each well, followed by stop solution. Absorbance was measured using a multi-plate reader (Varioskan LUX, Thermo Fisher Scientific, Waltham, MA, USA) at a wavelength of 450 nm.
## 2.3. Statistical Analysis
All experimental data of this study were statistically analyzed using a one-way analysis of variance followed by Duncan’s post hoc test with SPSS 26.0 software (SPSS Inc., Chicago, IL, USA). Each yak calf was considered an experimental unit. Data are expressed as means ± standard error. $p \leq 0.05$ was considered statistically significant.
## 3.1. Body Weight
The three treatments did not differ significantly in terms of BW on d 0, 30, and 60 (Table 2). The ADG was higher ($p \leq 0.05$) in the calves under T2 treatment than those under the control treatment, from d 0 to 30, d 30 to 60, and d 0 to 60, and higher ($p \leq 0.05$) than that of those calves under the T1 treatment from d 0 to 60, indicating that the supplementation of *Bacillus licheniformis* and the combination of probiotics and enzymes could improve the growth performance of early-weaned grazing yak calves. The ADG of calves under T1 treatment was higher ($p \leq 0.05$) than that of those under the control treatment from d 0 to 60.
## 3.2. Body Size
The body size parameters did not differ significantly among the three treatments on d 0 and 60 (Table 3), indicating that the supplementation of *Bacillus licheniformis* and the combination of probiotics and enzymes did not affect the body size of yak calves within 60 d.
## 3.3. Serum Biochemical Parameters
The concentrations of serum GLB, BUN, GLU, and NEFAs did not differ significantly among the three treatments on d 0 and 60 (Table 4).
## 3.4. Serum Hormone
As shown in Table 5, the concentrations of serum IGF-1 on d 60 were higher in T2-treated calves than in the T1- and control-treated calves ($p \leq 0.05$, each). The concentrations of serum EGF and GH on d 60 were higher in the T2-treated calves than in the controls ($p \leq 0.05$). The concentration of serum COR on d 60 was higher in the control calves than those under the T1 treatment ($p \leq 0.05$).
## 4. Discussion
Early weaning may have various benefits for cows; however, early weaned calves generally perform poorly compared to naturally weaned calves [37]. Early weaned calves without breastfeeding grew at a lower rate and subsequently took longer to reach their target weight than breastfed calves [38]. To improve the growth performance of early-weaned calves, several improvements were made to the composition of milk replacer or additional feeds were added [39,40,41]. Moreover, the addition of probiotics to the diets of calves significantly improved the ADG [29,30,33]. Dietary supplementation with compound enzyme preparations also improved growth performance in weaned piglets [42,43] and growing-finishing pigs [44]. However, previous studies also reported that supplementation with probiotics, yeast cultures or enzymes had no effect on the growth performance of calves [31,32,45]. In the current study, the addition of *Bacillus licheniformis* alone or a complex of probiotics and compound enzyme preparations to the milk replacer significantly improved the performance of grazing yaks and calves compared with milk replacer alone. Further, the addition of probiotics is beneficial for the regulation of the intestinal microbiota community structure, improving intestinal health and fecal consistency, and reducing diarrhea prevalence [19,31,46,47,48]. The supplementation of fibrolytic enzyme to the diet of crossbred calves improved their nutrient digestibility with a positive effect on daily gain [49]. Calves typically exhibit high metabolism and fast growth; however, their growth performance is susceptible to environmental stress and nutrient absorption and digestive problems, especially in the period after weaning [50]. Under natural grazing conditions on the Qinghai–Tibet Plateau, due to the long-term lack of pasture and harsh environmental conditions, the normal growth of yak calves is severely restricted [48]. In the present study, none of the study animals died, which may be attributed to the supplementation with milk replacer. Therefore, the addition of probiotics and compound enzyme preparations was beneficial for the growth of grazing yak calves.
In most cases, calf weight is positively correlated with body length, and body length can be used to predict calf live weight [51,52]. Supplementation with *Bacillus subtilis* results in an increased body length and BW in Barki lambs at the third and fourth week, as observed in a four-week continuous feeding trial [53]. In the present study, neither body size nor BW differed among the treatments, which may be due to insufficient trial duration and individual differences in animals. Therefore, more time may be required to elucidate whether the probiotic and compound enzyme preparations affected the calves’ body size.
To a certain extent, blood biochemical parameters reflect the metabolism and the acid–base balance of the animal body, and they vary within a certain range [54,55]. The results of the current study revealed that supplementation with *Bacillus licheniformis* and the complex of probiotics and enzyme preparations had no effect on the blood biochemical parameters of grazing yak calves, which is consistent with previously reported results in crossbred and Holstein calves [56,57]. The blood biochemical values of calves vary with the growing stage and are strongly influenced by weaning [58,59], and these possible factors may be stronger than the influence of diet on blood biochemical indicators.
Insulin-like growth factors (IGFs) are small polypeptide hormones mainly synthesized and secreted from the liver, and they are structural homologs of insulin, with similar activities. These consist in binding to specific carrier proteins in the blood to form a composite factor that stimulates systemic body growth and has growth-promoting effects on almost every cell in the body [60,61]. As mediators of GH action, the synthesis of IGFs is also affected by the blood level of GH [62]. EGF is a member of the growth factor family, a single polypeptide of 53 amino acid residues that is involved in regulating cell proliferation [63]. We found that the addition of probiotics and a combination of probiotics and enzymes significantly increased the concentration of serum IGF-1, EGF, and GH, whereas supplementation with *Bacillus licheniformis* alone did not achieve this effect. These results are consistent with the ADG results. GH and IGF-1 are important controllers in regulating amino acid metabolism in calves, where GH promotes the entry of amino acids in muscle tissue into cells and increases protein synthesis, and IGF-1 increases protein deposition by promoting protein synthesis [63,64]. Cortisol is commonly used as a marker of stress responses (such as weanling stress) in animals, and it occurs at high serum levels for a period of time after calves are weaned [65]. In line with our results, oral supplementation with probiotics markedly decreases the concentrations of serum cortisol in neonatal and weaned calves [66,67]. Interestingly, we found that the concentrations of serum cortisol were lower in the T1 than in the T2 group, which was, however, not statistically significant. This suggested that the addition of *Bacillus licheniformis* alone may better alleviate weaning stress in grazing yak calves. However, the respective mechanisms remain to be resolved in more detail.
A limitation of this study is that the T2 group did not strictly control a single variable compared to the T1 group, and the factors (yeast or xylanase, cellulase and glucanase) that contributed to the difference were unclear. This was due to the initial intention of this study to improve the milk replacer by adding probiotics or compound enzyme preparations, and ultimately promote the growth performance of yak calves on the Qinghai–Tibet Plateau. Further, we were unable to collect data on diarrhea and determine nutrient digestibility in grazing calves, which would have further improved our understanding of the weight gain of yaks under the various treatments.
## 5. Conclusions
Our results suggest that supplementation with *Bacillus licheniformis* alone or with a complex of probiotics (*Bacillus licheniformis* and yeast) and compound enzyme preparations (xylanase, cellulase, and glucanase) can improve the ADG of grazing yak calves, and the complex had a better effect on the ADG. The addition of the complexes of probiotics and complex enzyme preparations also increased the concentrations of serum GH, IGF-1, and EGF, which may have led to a higher ADG. Thus, the addition of a combination of probiotics and enzymes to milk replacer may serve as an effective strategy to improve the production of yak calves. |
# Growth Performance, Antioxidant and Immunity Capacity Were Significantly Affected by Feeding Fermented Soybean Meal in Juvenile Coho Salmon (Oncorhynchus kisutch)
## Abstract
### Simple Summary
Fish meal has been the main aquatic feed protein source for aquaculture. However, global fish meal is lacking, and the price of fish meal continues to rise, which has been unable to meet the needs. Soybean meal is currently recognized as the best choice to replace fish meal in aquatic feed, but soybean meal contains anti-nutritional factors which can affect the health of aquatic animals. Microbial fermentation is a commonly used biological method for treating soybean meal antigens and palatability. In this study, juvenile coho salmon were fed a diet with replaced $10\%$ fish meal protein with fermented soybean meal protein supplementation for 12 weeks. The results indicated that the diet with replaced $10\%$ fish meal protein with fermented soybean meal protein supplementation could significantly ($p \leq 0.05$) influence the expression of superoxide dismutase, catalase, glutathione peroxidase, glutathione S-transferase, nuclear factor erythroid 2-related factor 2, tumor necrosis factor α and interleukin-6 genes, the growth performance, the serum biochemical indices, and the activity of antioxidant and immunity enzymes.
### Abstract
This study aims to investigate the effects of partial dietary replacement of fish meal with unfermented and/or fermented soybean meal (fermented by Bacillus cereus) supplemented on the growth performance, whole-body composition, antioxidant and immunity capacity, and their related gene expression of juvenile coho salmon (Oncorhynchus kisutch). Four groups of juveniles (initial weight 159.63 ± 9.54 g) at 6 months of age in triplicate were fed for 12 weeks on four different iso-nitrogen (about $41\%$ dietary protein) and iso-lipid (about $15\%$ dietary lipid) experimental diets. The main results were: Compared with the control diet, the diet with replaced $10\%$ fish meal protein with fermented soybean meal protein supplementation can significantly ($p \leq 0.05$) influence the expression of superoxide dismutase, catalase, glutathione peroxidase, glutathione S-transferase, nuclear factor erythroid 2-related factor 2, tumor necrosis factor α and interleukin-6 genes, the growth performance, the serum biochemical indices, and the activity of antioxidant and immunity enzymes. However, there was no significant effect ($p \leq 0.05$) on the survival rate (SR) and whole-body composition in the juveniles among the experimental groups. In conclusion, the diet with replaced $10\%$ fish meal protein with fermented soybean meal protein supplementation could significantly increase the growth performance, antioxidant and immunity capacity, and their related gene expression of juveniles.
## 1. Introduction
Coho salmon (Oncorhynchus kisutch) has become one of the most promising fish in China because of its fast growth rate, high economic value, rich nutrition, containing a variety of minerals, and delicious meat [1,2,3]. At present, the feed needed by the salmon aquaculture industry is mainly fish meal, and fish meal has been the main aquatic feed protein source for aquaculture because of its high protein content, balanced amino acid composition and rich nutrition [4]. However, due to the continuous growth of the modern aquaculture industry, global fish meal is lacking, and the price of fish meal continues to rise, which has been unable to meet the needs [5]. Therefore, it is urgent to find a suitable protein source to replace fish meal in the aquaculture industry.
Soybean meal is a plant protein with high digestive protein content, wide source, and low price, so it is currently recognized as the best choice to replace fish meal in aquatic feed [6]. However, the soybean meal contains unbalanced amino acids and soybean antigen protein, urease, trypsin inhibitor, soybean lectin, phytic acid, saponins, phytoestrogens, anti-vitamins and allergens, and other anti-nutritional factors [7,8,9], which can affect the palatability, and inhibit the digestion and absorption of nutrients, and cause the damage of tissue and organ, and seriously affect the health of aquatic animals [10,11]. Microbial fermentation is a commonly used biological method for treating soybean meal antigens and palatability, and soybean meal after microbial fermentation can reduce most of the anti-nutritional factors, produce carbohydrates, digestive enzymes and other nutrients, degradation of macromolecular protein, produce small active peptides, organic acids, thereby enhancing its nutritional value and enhance the digestion and absorption of nutrients [12,13,14]. In addition, fermented soybean meal can also provide animals with probiotics, prebiotics and flavonoids and other active substances [15,16] and increase the antioxidant properties of free amino acid content and the concentration of phenolic compounds [17].
At present, there are relatively few studies on the replacement of fish meal with fermented soybean meal in coho salmon. The antibacterial substances produced by *Bacillus cereus* have the effects of promoting growth, regulating immune function, and treating diseases in livestock and poultry [18]. Therefore, coho salmon was selected as the research object, and *Bacillus cereus* was used as a fermentation strain to explore the effects of replacing part of fish meal with fermented soybean meal on the growth performance, muscle composition, antioxidant and immunity capacity, and their related gene expression of juvenile coho salmon in this study. The results provide a theoretical basis for the development and optimization of coho salmon compound feed and the healthy development of the artificial breeding industry.
## 2.1. Experimental Diets
Four different iso-nitrogen (about $41\%$ dietary protein) and iso-lipid (about $15\%$ dietary lipid) experimental diets were designed and based on the references [19,20,21], in which the soybean meal could replace $10\%$ fish meal protein. The G0 diet contained $28\%$ fish meal protein (control group). Three other diets (G1, G2 and G3) were replaced $10\%$ fish meal protein with unfermented and/or fermented soybean meal: The G1 diet replaced by $10\%$ unfermented soybean meal protein, the G2 diet replaced by $5\%$ unfermented soybean meal protein and $5\%$ fermented soybean meal protein, and the G3 diet replaced by $10\%$ fermented soybean meal protein, based on per kg of dried feed, as shown in Table 1.
All the feed materials were provided by Conkerun Ocean Technology Co., Ltd. in Shandong, China, and they were animal food-grade. The soybean meal was fermented by Bacillus cereus, and the bacterial strain was collected from mangrove root soil in Maowei Sea, Qinzhou, Guangxi, China (21°81′66″ N, 108°58′46″ E). The experimental strains and fermentation conditions were derived from preliminary experiments in our lab. The inoculation amount of *Bacillus cereus* was $10\%$ (v/m), the ratio of material to water was 1:1.4, and the fermentation was cultured at 37 °C for 60 h. The fermented soybean meal was dried for 24 h in a blast drying baker at 37 °C. A hammer mill was used to grind raw all the dry materials into a fine powder (80-μm mesh), then all the dry materials were mixed in a roller mixer for 15 min and added some water to make a hard dough. Floating pellets with a diameter of 2.0 × 3.0 mm were obtained by a single screw extruder, and they were dried in the air flow at 37 °C until the water content was below 100 g/kg. Then the dry floating pellets were sealed in plastic bags and stored at −20 °C until use.
## 2.2. Experimental Fish and Culture
Six hundred juvenile coho salmon at the age of 6 months were from a hatchery located in Benxi rainbow trout breeding farm in Liaoning, China. Outdoor feeding and breeding experiments of juvenile coho salmon were carried out at a rainbow trout breeding farm in Nanfen District, Benxi City, Liaoning, China.
After being disinfected using a concentration of $\frac{1}{100}$,000–$\frac{1}{50}$,000 potassium permanganate, the juveniles were acclimatized for 14 days, using water temperature at 10–18 °C, water intake ≥ 100 L/s, surface velocity ≥ 2 cm/s, dissolved O2 ≥ 6.0 mg/L, pH 7.8–8.3 and natural light. The juveniles were fed three times a day at 08:00, 12:00 and 16:00 h, using a control diet ($28\%$ fish meal protein), and the daily feeding quantity was fed until the fish was no feeding behavior at the feeding time.
After being acclimatized for 14 days, 390 juvenile coho salmon (initial weight 159.63 ± 9.54 g) were selected for the formal experiment, and 30 of the selected juveniles were freely taken for initial samples. The remaining 360 of them were assigned randomly into 4 groups in triplicate, making a total of 12 net cages (1.0 × 1.0 × 0.8 m, L × W × H) with 30 fish in each net cage. The juveniles were cultured in the same breeding environment, and they were fed for 12 weeks using one of the 4 diets above (Table 1) and the daily feeding quantity was fed until the fish was no feeding behavior at the feeding time.
## 2.3. Sampling
The juvenile coho salmon were sampled at day 0 and the end of 12 weeks, respectively, after being starved for 24 h. All sample fish were separately anesthetized using 40 mg/L of 3-aminobenzoic acid ethyl ester methane sultanate (MS-222, Adamas Reagent, China). Then, their body weight and length were individually measured. At day 0, 20 juveniles were taken for dissecting liver samples and the other 10 juveniles for the sampling of whole fish. At the end of 12 weeks, 9 fish per net cage were randomly taken for the samples, 3 of which were for whole fish samples and 6 for the samples of serum, viscera mass, and liver.
A sterile syringe was used to collect blood from the tail vein of juvenile coho salmon; then, the blood was transferred to a 2 mL sterile enzyme-free centrifuge tube. At 3000× g and 4 °C, the blood was centrifuged in a centrifuge for 15 min, and the supernatant was serum. The liver weight and visceral mass weight were weighed and recorded separately for analysis of the growth performance. All the experimental samples were stored at −80 °C for subsequent analysis.
## 2.4.1. Growth Performance
The survival rate, weight gain rate, specific growth rate, condition factor, hepatosomatic index, viscerosomatic index, feed conversion ratio, and protein efficiency ratio are calculated according to the following formulas. Survival rate (SR, %)=100 ×final amount of fishinital amount of fish Weight gain rate (WGR, %)=100 ×final body weight (g) − initial body weight (g)initial body weight (g) Specific growth rate (SGR, %/d)=100 ×ln(final body weight (g)) − ln(initial body weight (g))days Condition factor (CF, %)=100 × body weight (g)(body length (cm))3 Hepatosomatic index (HSI, %)=100 ×liver weight (g) body weight (g) Viscerosomatic index (VSI, %)=100 ×viscera weight (g) body weight (g) Feed conversion ratio (FCR)=total diets weight (g) final body weight (g) − initial body weight (g) Protein efficiency ratio (PER, %)=100 ×final body weight (g) − initial body weight (g) total intake of crude protein weight (g)
## 2.4.2. Determination of Feed and Whole Fish Composition
The compositions of feed and whole fish were analyzed following the standard methods of the Association of Official Analytic Chemists (AOAC, 2005) [22]. The samples were dried at 105 °C until constant weight in an oven to determine moisture content. The muffle furnace at 550 °C for 24 h was used to determine ash. Kjeldahl method was used to determine crude protein. Soxhlet method by ether extraction was used to determine crude lipid.
## 2.4.3. Determination of Serum Biochemical Parameters
The indicators in serum were measured using the kit produced by Nanjing Jiancheng Bioengineering Institute (Nanjing, China) and referred to the instructions in the kit for specific operation steps. All the instructions can be found and downloaded at http://www.njjcbio.com (accessed on 1 March 2023). The total protein (TP) content was determined by the Coomassie brilliant blue method. The glucose (GLU) content was determined by the glucose oxidase method. The total cholesterol (T-CHO) content was determined by the cholesterol oxidase (COD-PAP) method. The albumin (ALB) content and alkaline phosphatase (AKP) vitality were determined by the microplate method.
## 2.4.4. Determination of Liver Antioxidant Capacity
The indicators in the liver were measured using the kit produced by Nanjing Jiancheng Bioengineering Institute (Nanjing, China) and referred to the instructions in the kit for specific operation steps. All the instructions can be found and downloaded at http://www.njjcbio.com (accessed on 1 March 2023). The superoxide dismutase (SOD) was determined by the water-soluble tetrazole salt (WST-1) method. The catalase (CAT) was determined by the visible light method. The malondialdehyde (MDA) was determined by the thiobarbituric acid (TBA method). The total antioxidant capacity (T-AOC) was determined by the ferric-reducing ability of plasma (FRAP) method. The glutathione peroxidase (GSH-PX), glutathione S-transferase (GST), hydroxyl radical clearance ratio (OH·-CR) and superoxide radical clearance ratio (O2·-CR) were determined by the colorimetric method. The reduced glutathione (GSH) was determined by the microplate method.
## 2.4.5. Expression of Antioxidant and Immunity Genes
The method of Ding et al. [ 23] was applied to determine the expression of sod, cat, gsh-px, gst, nrf2, tnf-α and il-6 mRNA in the liver of the juvenile coho salmon. Briefly, the Steady Pure Universal RNA Extraction Kit and the Evo M-MLV reverse transcription kit (Accurate Biology Biotechnology Engineering Ltd., Changsha, China) were used to extract 500 ng of total RNA from samples and reverse-transcribe it into cDNA. The polymerase chain reaction (PCR) conditions were 50 °C for 30 min, 95 °C for 5 min, and 5 °C for 5 min.
The forward and reverse primers of sod, cat, gsh-px, gst, nrf2, tnf-α and il-6 genes for reverse transcription were designed by referencing the corresponding genomic sequences of coho salmon in the National Center for Biotechnology Information (NCBI) database. The primers were synthesized by Sangon Biotech (Shanghai) Co., Ltd. (Shanghai, China). The primers were shown in Table 2, and β-actin was chosen as the nonregulated reference gene.
The real-time quantitative polymerase chain reaction (RT-qPCR) was conducted using an RT-qPCR System (LightCycler® 96, Roche, Switzerland) and SYBR Green Pro Taq HS qPCR kit (Accurate Biology Biotechnology Engineering Ltd., Changsha, China). The RT-qPCR conditions were as follows: initial denaturation at 95 °C for 30 s, 40 cycles of denaturation at 95 °C for 5 s, annealing at 60 °C for 30 s and extension at 72 °C for 20 s.
The 2−ΔΔCT method [24] was applied to calculate the relative expression levels of sod, cat, gsh-px, gst, nrf2, tnf-α and il-6 mRNA.
## 2.5. Statistical Analysis
All the data were analyzed using IBM SPSS Statistics 25 (Chicago, IL, USA) and one-way analysis of variance (ANOVA) and tested for normality and homogeneity of variance. Duncan’s test was used for multiple comparison analysis when it was significantly different ($p \leq 0.05$). Statistics are expressed as means ± standard deviation (SD).
## 3.1. Effect of Replacing a Portion of Fish Meal with Unfermented and/or Fermented Soybean Meal on the Growth Performance of Juvenile Coho Salmon
The WGR, SGR, CF, and PER of the juveniles in G3 and the HSI, VSI, and FCR of the juveniles in G1 and G2 were significantly higher ($p \leq 0.05$) than those of the juveniles in G0. The HSI, VSI, and FCR of the juveniles in G3 and the WGR, SGR, CF, and PER of the juveniles in G1 and G2 were significantly lower ($p \leq 0.05$) than those of the juveniles in G0. However, there was no significant difference in the SR of the juveniles between the groups ($p \leq 0.05$), as shown in Table 3.
## 3.2. Effect of Replacing a Portion of Fish Meal with Unfermented and/or Fermented Soybean Meal on the Whole-Body Composition of Juvenile Coho Salmon
No significant difference ($p \leq 0.05$) was found in the moisture, crude protein, crude lipid, and ash of juvenile coho salmon fed diets of replacement of fish meal with unfermented soybean meal and/or fermented soybean meal, as shown in Table 4.
## 3.3. Effect of Replacing a Portion of Fish Meal with Unfermented and/or Fermented Soybean Meal on the Physiological and Biochemical Indices in Serum of Juvenile Coho Salmon
The TP, GLU, ALB, AKP, and T-CHO of the juveniles in G3 were significantly higher ($p \leq 0.05$) than those of the juveniles in G0. The TP, GLU, ALB, AKP, and T-CHO of the juveniles in G1 and G2 were significantly lower ($p \leq 0.05$) than those of the juveniles in G0, as shown in Table 5.
## 3.4. Effect of Replacing a Portion of Fish Meal with Unfermented and/or Fermented Soybean Meal on the Antioxidant Capacity in the Liver of Juvenile Coho Salmon
The SOD, CAT, GSH-PX, GSH, GST, OH·-CR, O2·-CR, and T-AOC of the juveniles in G3, and the MDA of the juveniles in G1 and G2 were significantly higher ($p \leq 0.05$) than those of the juveniles in G0. The MDA of the juveniles in G3 and the SOD, CAT, GSH-PX, GSH, GST, OH·-CR, O2·-CR, and T-AOC of the juveniles in G1 and G2 were significantly lower ($p \leq 0.05$) than those of the juveniles in G0, as shown in Table 6.
## 3.5. Effect of Replacing a Portion of Fish Meal with Unfermented and/or Fermented Soybean Meal on the Expression of Antioxidant and Immune Genes in the Liver of Juvenile Coho Salmon
The expression of the sod, cat, gsh-px, gst, and nrf2 genes in the liver of the juveniles in G3 and the expression of the il-6 and tnf-α genes in the liver of the juveniles in G1 and G2 were significantly higher ($p \leq 0.05$) than those of the juveniles in G0. The expression of the il-6 and tnf-α genes in the liver of the juveniles in G3 and the expression of sod, cat, gsh-px, gst, and nrf2 genes in the liver of the juveniles in G1 and G2 were significantly lower ($p \leq 0.05$) than those of the juveniles in G0, as shown in Figure 1.
## 4. Discussion
The growth performance of fish can be used to reflect growth and health status, and it is affected by many factors, such as fish species, growth stage, nutrient deficiency, metabolic disorders, anti-nutritional factors, and toxic and harmful substances [25]. The results of this study showed that partial replacement of fish meal with fermented soybean meal could significantly increase the growth performance of juvenile coho salmon. However, partial replacement of fish meal with unfermented soybean meal could significantly decrease the growth performance of juvenile coho salmon. The reasons are supposed to be: First, unfermented soybean meal had adverse factors such as poor palatability, essential amino acid imbalance, low phosphorus utilization, high anti-nutritional factors, and easily cause lipid metabolism disorder, which will lead to decreased growth performance [26]. Second, fermented soybean meal could reduce and even eliminate anti-nutrient factors, and the protein could be degraded into easily digestible peptides or amino acids; thus, fermented soybean meal could improve the nutritional quality of feed and the digestibility of fish [27]. Third, the active bacteria, organic acids, and vitamins in fermented soybean meal would also play a positive role in growth performance [28]. Similar studies had shown that feeding largemouth bass (Micropterus salmoides) [21] and Macrobrachium nipponense (Macrobrachium nipponense) [29] with the diet with partial replacement of fish meal with fermented soybean meal significantly improved their growth performance.
Serum biochemical indexes of fish are closely related to metabolism, nutrient absorption, and health status. They are important indexes to evaluate physiology and pathology and are widely used to measure metabolism and health status [30,31]. TP and ALB in the blood are synthesized by the liver, and the increase of TP and ALB content indicates that the ability of the liver to synthesize protein is enhanced. AKP is one of the important indicators of fish physiological activity and disease diagnosis, which can reflect the anti-stress ability of biological organisms [32]. T-CHO is an important index to reflect the body’s lipid metabolism [33]. GLU is the main functional substance of the body, and its content is affected by nutrition and feed intake [34]. The results of this study showed that partial replacement of fish meal with fermented soybean meal could significantly increase the serum biochemical indexes of juvenile coho salmon, indicating that fermented soybean meal could be used as a protein substitute for fish meal to improve the health of juvenile coho salmon. The reasons are supposed to be: First, fermented soybean meal could improve the intestinal structure and function of fish, increase the activity of digestive enzymes, and increase the absorption and utilization of dietary proteins and lipids [35]. Second, compared with macromolecular proteins, the small peptides in fermented soybean meal are more easily absorbed by fish, which could improve the diet protein utilization rate, consequently enhancing the serum protein content of fish [12]. Third, fermented soybean meal could decrease the content of soybean saponins, increase the activity of α-glucosidase, and improve the absorption of glucose [36]. Fourth, fermented soybean meal could not only reduce the inhibitory effect of soy isoflavones on serum T-CHO levels but also stimulate the antioxidant system of the body, thereby inhibiting the process of lipid oxidation and increasing the content of T-CHO in the serum [37]. In addition, bioactive peptides during fermentation can act as immune stimulants to enhance AKP activity [38].
Nuclear factor erythroid 2-related factors (nrf2) is an important nuclear transcription factor and can be involved in a variety of cellular processes, including maintaining intracellular redox balance, cell proliferation/differentiation, metabolism, protein homeostasis and inflammation regulation, and disease development [39,40]. The activation of the nrf2 signaling pathway can initiate the expression of multiple downstream target proteins, such as SOD, CAT, GPX, glutathione ligase (γ-GCS), glutathione catalase (GR), glutathione S-transferase (GST) and glucose-6-phosphate kinase (G-6-PDH) [41]. The expression of these genes is an important way for the body to resist oxidative stress damage [42]. Nrf2 signaling pathway can negatively regulate various cytokines (TNF-α, IL-1 and IL-6), chemokines, cell adhesion factors, matrix metalloproteinases, cyclooxygenase-2, inducible nitric oxide synthase, and other inflammatory mediators, which plays a protective role in the dysfunction caused by inflammation [43]. IL-6 and TNF-α are often used as indicators of the inflammatory response [44]. MDA content has been used by many researchers to evaluate the effect of protein replacement sources on the antioxidant capacity of fish, which can be used as an important marker of endogenous oxidative damage in organisms [45]. The results of this study showed that partial replacement of fish meal with fermented soybean meal could significantly increase the antioxidant capacity and the expression of their related gene in the liver and significantly decrease the expression of il-6 and tnf-α gene in the liver of juvenile coho salmon. However, partial replacement of fish meal with unfermented soybean meal could significantly decrease the antioxidant capacity and the expression of their related gene in the liver and significantly increase the expression of the il-6 and tnf-α genes in the liver of juvenile coho salmon. The reasons are supposed to be: First, the soybean globulin and β-conglycinin in soybean meal could destroy the antioxidant system of fish and cause oxidative damage [46]. Previous studies have shown that soybean meal in feed may cause oxidative stress in fish such as gilthead sea bream (Sparus aurata) [47]. Second, a high concentration of soybean peptides and phenols in fermented soybean meal could up-regulate nrf2 gene expression, induce the expression of the sod, cat, gsh, and gsh-px genes, and improve the antioxidant ability of the body [48,49]. Lee et al. found that an appropriate proportion of fermented soybean meal in a diet can increase the activities of SOD, GSH-Px, and GSH in the liver [50]. Third, *Bacillus could* stimulate the production of antioxidant enzymes and antioxidants, thereby scavenging free radicals, maintaining homeostasis, improving antioxidant capacity, and activating the Nrf2 pathway [51]. Fourth, the replacement of fish meal protein with $10\%$ fermented soybean meal protein was insufficient for causing a change in the body’s ability to recognize foreign bodies and did not lead to an inflammatory reaction [52]. In addition, after soybean meal fermentation, a unique fragrance could be formed, which can promote the feeding of aquatic animals and increase their immunity [53].
However, the results of this study showed that partial replacement of fish meal with unfermented and/or fermented soybean meal had no significant effect on the survival rate and whole-body composition of juvenile coho salmon. The reasons are supposed to be: First, the energy required by fish to maintain normal life activities mainly depends on the breakdown of protein and fat, and fish meal contains a complete set of essential amino acids that meet the protein requirements of most aquatic animals [54,55]. Second, the crude protein and crude fat contents of the four diets in this study were the same and were enough to satisfy the daily needs of juvenile coho salmon. Third, fish body composition is affected by external conditions such as feed nutrients, food composition, aquaculture water environment and season, but fish body composition was not affected by plant protein levels [56]. Similar results were obtained in pompano (Trachinotus ovatus) [53] and Florida pompano (Trachinotus carolinus) [56] fed with fermented soybean meal partially replacing fish meal. However, studies have shown that a high proportion of fermented soybean meal instead of fish meal significantly increased the whole-body moisture and reduced crude protein and crude lipid content of Japanese seabass (Lateolabrax japonicus) [57]. In giant grouper (Epinephelus lanceolatus), high levels of fermented soybean meal replacement also significantly increased whole-fish moisture and decreased crude protein and crude lipid content [58]. The above inconsistent results might be related to the strains of fermented soybean meal, the basic feed formula, the substitution ratio of fermented soybean meal, the types of aquatic animals, the breeding cycle, and the growth stage.
## 5. Conclusions
In conclusion, the diet with replaced $10\%$ fish meal protein with fermented soybean meal protein supplementation can significantly influence the expression of superoxide dismutase, catalase, glutathione peroxidase, glutathione S-transferase, nuclear factor erythroid 2-related factor 2, tumor necrosis factor α and interleukin-6 genes, the growth performance, the serum biochemical indices, and the activity of antioxidant and immunity enzymes of juvenile coho salmon. The results provide a theoretical basis for the development and optimization of coho salmon compound feed and the healthy development of the artificial breeding industry. |
"# Association of Computed Tomography Measures of Muscle and Adipose Tissue and Progressive Changes (...TRUNCATED) |
"# Comparison of Fecal Microbiota Communities between Primiparous and Multiparous Cows during Non-Pr(...TRUNCATED) |
"# Bamboo Plant Part Preference Affects the Nutrients Digestibility and Intestinal Microbiota of Ger(...TRUNCATED) |
End of preview. Expand
in Data Studio
No dataset card yet
- Downloads last month
- 6