Oligosaccharide equine feed supplement, Immulix, has minor impact on vaccine responses in mice

Several mammalian species are vaccinated in early life, but little is known about the effect of diet on vaccine response. Oligosaccharides are increasingly proposed as dietary supplement for young individuals due to their anti-inflammatory potential elicited through modulation of gut microbiota (GM). Also, diet, e.g. the size of the fat fraction, is known to modulate the GM. We tested if an oligosaccharide diet (Immulix) and/or increased dietary fat content affected antibody titers to a tetanus vaccine in 48 BALB/cJTac mice through GM modulation. Female mice had significantly higher IgG titers with higher variation compared to male mice. The effects of Immulix and/or increased fat content were minor. Immulix negatively affected IgG titers in male mice four weeks after secondary vaccination but upregulated Il1b gene expression in the spleen. Immulix had a downregulating effect on expression of Cd4 and Foxp3 in ileum only if the mice were fed the diet with increased fat. The diet with increased dietary fat increased Il1b but decreased Cd8a gene expression in the spleen. Immulix and diet affected GM composition significantly. Increased dietary fat content upregulated Lactobacillus animalis but downregulated an unclassified Prevotella spp. Immulix decreased Lactobacillales, Streptococcaceae and Prevotellaceae but increased Bacteroides. It is concluded that in spite of some minor influences on immune cell markers, cytokines and IgG titers Immulix feeding or increased dietary fat content did not have any biologically relevant effects on tetanus vaccine responses in this experiment in mice.

www.nature.com/scientificreports/ cytokines IL-10 and TGF-β and the T reg cell transcription factor FOXP 3 as well as downregulation of genes coding for the proinflammatory cytokine IL-12 in ileum and mesenteric lymph nodes (MLN) 16 . Vaccination is an indispensable tool for prevention of infectious diseases, and varying or even lack of efficiency poses problems in the control of global diseases in both human and domestic animals 26,27 . Since Immulix is a product for foals, which are routinely vaccinated against tetanus, influenza and herpesvirus at the age of five to six months 28 , it would be interesting to test how Immulix feeding affects the vaccine response in horses, by using a murine model. Knowledge about effects, as well as possible side-effects, of a product increases its safety in use. Besides that, increased knowledge on the effects of oligosaccharides on GM and the immune system, can be useful for other mammals, including humans, and contribute to further research within this field. Oligosaccharides are known to stimulate production of T reg cells and IL-10 in horses. T reg and IL-10 are among the most important components of immunoregulatory mechanisms aimed to control inflammation i.e. primarily exert anti-inflammatory impact, why it is reasonable to hypothesize that it may have a dampening impact on vaccine response. Oppositely, due to specific positive influences on B cells, IL-10 might exert a positive impact on vaccine responses 29 . A meta-analysis by Lei,et al. 30 concluded that the intake of pre-and/or probiotics, enhance serum immunoglobulin (Ig) titres to influenza vaccines in adult humans, indicating increased immune responsiveness 31 . Similar results have been reported in piglets and broilers 32,33 . A study by van den Elsen, et al. 34 demonstrated that early oligosaccharide administration (either starting from fertilization through the mother, from birth or from weaning), enhanced IgG titers to influenza vaccines in male but not female mice compared to controls. Even though exact mechanisms appear unclear, GM seems to play a critical role due to the fact that oligosaccharides evidently correlate with specific beneficial bacteria that correlate with vaccine responses. Zhang, et al. 10 , reported increased Bifidobacterium, Lactobacillus and Bacteroides abundances in the GM of oligosaccharide fed mice, which correlated positively with serum IgG concentrations.
Basic dietary macronutrient levels may also influence GM, and eventually thereby vaccine responses. Moreno-Indias, et al. 35 showed that the dietary fat content (approx. two times more compared to a standard rodent chow diet) significantly altered GM composition of mice and enhanced the percentage of B cells in Peyer's patches (PP's) and Cd8a expression in ileum 35 . Oil supplements in horse diets have furthermore proven to affect immune functions in horses 36 . It, therefore, seems relevant to be aware that responses to both an oligosaccharide supplement and a vaccine may differ according to the diet fed and that especially the fat proportion of the diet may have an influence on the immunological outcome through GM manipulation.
The aim of this study was, therefore, to investigate if Immulix in mice fed diets with standard or increased fat content would influence the immune responses to a tetanus vaccine in a murine model and to assess how the GM were affected for a better understanding of the interplay between GM and the immune system. We used antibody response to the tetanus vaccine as primary read-out. Based on previous results from the literature, the effect of oligosaccharides could be both up-and down-regulatory on the immune system and therefore, a two-sided hypothesis was tested. We tested our hypothesis in mice fed either a diet with an ordinary rodent macronutrient profile (MP = murine profile) or a diet with a more human macronutrient profile (HP = Human profile), i.e. with increased amount of fat.

Materials and methods
This study was approved by the Animal Experiments Inspectorate, Ministry of Food, Fisheries and Agriculture, Denmark. All housing, maintenance and experimental procedures were carried out according to the EU directive 2010/63/EU on the protection of animals used for scientific purposes and The Danish Animal Experimentation Act (LBK 474 from 15/05/2014). This study was reported in accordance with the ARRIVE guidelines 37 .
Study design and sampling. 24 male-and 24 female BALB/cjTac mice (Taconic Europe, Lille Skensved, Denmark), aged three to four weeks at arrival, were randomly allocated in four feeding groups designated MP (mouse profile); HP (human profile); MP + Imx; HP + Imx (Fig. 1B). Each group contained six males and six females co-housed according to sex in polycarbonate 1290D Eurostandard type III cages supplied with Tapvei aspen bedding, Tapvei wooden chewing block, disposable Smart Home shelters, Mini Fun Tunnels, Nestlet nesting material and Enviro-dri (Brogaarden, Lynge, Denmark) at temperatures of 22 ± 2 •C, humidity of 55% ± 10%, air changing 15-20 times/hour and a 12 h light cycle. The sample size was founded on a power analysis, that revealed a sample size of 12 with 90% power and 5% significance level, assuming effect size of 28 U/ml and standard deviation (SD) of 20 U/ml based on comparable results from a previous study of the same type 38 , and a two-sided-test. Taking this into account, in combination with experience from previous comparable studies and practical considerations, a sample size of 12 mice per group was deemed appropriate.
During the 12 weeks experimental period (Fig. 1A), the MP and MP + Imx groups were fed a standard Altromin 1324 diet with a macronutrient composition as normally recommended for mice, i.e. a fat content of 4.3 g % (here designated "MP diet" (mouse profile); Brogaarden, Lynge, Denmark), while the HP and HP + Imx groups were fed a modified Altromin 1320 diet with a macronutrient profile as normally recommended for humans, i.e. a fat content of 10.5% 35 (here designated "HP diet" (human profile); Brogaarden, Lynge, Denmark). In the HP diet, the macronutrient composition was changed to match a diet recommended for humans, similar to the "human profile (HP)" diet previously applied 35 . Especially the fat proportion was increased (from 4.3% g to 10.5% g) in the HP diet (Fig. 1C). The MP + Imx and HP + Imx groups were additionally fed Immulix (6.3 g per 100 g feed; Brogaarden, Lynge, Denmark). During the first 14 days of the experiment, the mice received probiotics (a mix of Bifidobacteria and Lactobacilli; Udo's Choice, Super Infant 3-5 years; FMD, Burnaby, Canada) through drinking water (0.1 g/100 ml water) in order to help the establishment of Bifidobacteria colonies in the mice, which laboratory mice often lack 39 (Fig. 1A) . Gene assays binding to gDNA in the -RT control was excluded from further analysis (Gapdh in spleen). GenEx 6 (MultiD Analyses AB, Gothenburg, Sweden) was used for qPCR data transformation. Cq values were normalized to the reference genes Actb (spleen) and Sdha + Actb (ileum), which was appraised suitable for normalization using the built-in functions geNorm 40 www.nature.com/scientificreports/ and NormFinder 41 . For each gene, normalized expression levels were set relative to the sample with lowest expression to establish relative quantities (RQ) and were log2 transformed before statistical testing.
Gut microbiota characterization by 16S rRNA sequencing. Faecal  Raw data, containing pair-ended reads with corresponding quality scores, were initially merged and trimmed with the settings: -fastq_minovlen 100, -fastq_maxee 2.0, -fastq_truncal 4, -fastq_minlen 130. Quantitative Insight Into Microbial Ecology (QIIME; v 1.9) open source software package 42 was used for identifying unique reads and deconvoluting from chimeric reads as well as construction of de-novo zero-radius OTUs by utilizing the UNOISE pipeline 43 . The EzTaxon-e procaryotic 16S rRNA gene sequence database was used as a reference 44 .
Statistical methods. All data were initially tested for normal distribution (Anderson-Darling test) within feeding groups and equal variances (Levene's test). If criteria for parametric testing were not fulfilled, data were ranked before further statistical testing. Statistics were unless otherwise mentioned performed with Minitab Statistical Software version 19 (Minitab, Coventry, UK).
Weight data (AUC), IgG, Log2(RQ) and OTUs were analysed by a multifactorial general linear model ANOVA on factors Immulix, diet and sex including interactions. Interactions were compared with Fisher's least significant difference test. Whether Immulix feeding resulted in significantly more or less high responders (cut-off defined as median + 2 × SD) in IgG titers was tested with Fisher's exact test. Relative abundance at each taxonomic level was based on rarefied OTU-tables and was only included for statistics if at least one of the groups mean (+ Immulix vs. -Immulix; MP vs. HP) was > 1%. One OTU was analysed with Kruskal-Wallis test with Dunn's post hoc test due to unequal variances of ranked data (Graphpad Prism8; GraphPad Software, San Diego, CA, USA). Each set of p-values (Immulix or diet) was corrected for False Discovery Rate (FDR) (Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli, desired FDR = 5%) generating "q-values" based on the newly defined statistical significance level: p < 0.003 (Graphpad Prism 8; GraphPad Software, San Diego, CA, USA). Bray-Curtis dissimilarities (beta-diversity) were calculated from rarefied OTU-tables (10,000 sequences/sample) and visualized with three-dimensional principal coordinate analysis (PcoA) plots. Differences in beta-diversity (analysis of similarities), and alpha-diversity (t-test) were calculated using functions implemented in QIIME; v 1.9 toolbox 42 .
It was additionally tested if correlations existed between OTUs and remaining data. p-values were obtained with Pairwise Pearson Correlations and were subsequently FDR-corrected as described above.

Immulix feeding increased
Diet and Immulix altered gut microbiota composition. Beta diversity (Bray-Curtis dissimilarities) between faecal samples from week 12 was significantly altered by diet (p = 0.001; Fig. 5C) and Immulix (p = 0.010; Fig. 5D). No effect of Immulix (p = 0.521) or diet (p = 0.738) on alpha-diversity was found. When investigating differences in relative abundance at different taxonomic levels, it was found that Immulix feeding downregulated the order Lactobacillales (q = 0.035) driven by specific downregulation of the family Streptococcaceae (q = 0.000; Table 1and Fig. 6). While Immulix did not affect genus or species level of Lactobacillus, the HP diet upregulated L. animalis (q = 0.034). Immulix upregulated the phylum Bacteroidetes (q = 0.047), mainly driven by an increase of the family Bacteroidaceae (q = 0.010) and genus Bacteroides (q = 0.010). An unspecified Prevotella sp. was decreased by both Immulix (q = 0.006) and the HP diet (q = 0.034). Akkermansia and Bifidobacteria were not abundant in the mice, except in four mice from the HP + Imx group, that harboured Bifidobacterium animalis (0.31-14.5%). No significant correlations between bacterial abundances and remaining data were identified when p-values were FDR-corrected, although some correlations between some taxa and gene expression of Tnfα and Cd4 in spleen tended to be significant (p < 0.05; q < 0.1; Table 2).
A human profile diet increased weight gain. Three male mice (no. 29, 30 and 36) from the MP + Imx group from the same cage were euthanized at week 8 due to weight loss and impairment, probably caused by aggressive behaviour. There were no significant effects on body weight if these mice were not included in the data evaluation (Fig. 7).

Discussion
The overall aim of this study was to investigate whether immune-regulating effects of Immulix influence immune responses to vaccination, as foals, which are the targets of Immulix, are routinely vaccinated. Significantly lower tetanus-specific IgG titers four weeks after secondary vaccination were observed in Immulix fed male mice www.nature.com/scientificreports/ compared to males not fed Immulix (p = 0.036). Testosterone has previously shown immunosuppressive properties on vaccine responses 45 , but van den Elsen, et al. 34 reported increased trivalent influenza vaccine (TIV)specific IgG titers in male mice four weeks after primary vaccination (p < 0.01). However, it is doubtful whether the observed difference in IgG titers between male mice fed Immulix and males not fed Immulix has any biological relevance, because they were approximately twofold, while a booster vaccination increased the response tenfold and lowered the difference. When both sexes were included in the statistical tests, no effect of Immulix was detected. In agreement with van den Elsen, et al. 34 , our study revealed significantly higher and more diverse IgG titers for females compared to males, which might explain why significant effects were only detected in males. The higher variance in females might be explained by cycle fluctuations, since oestrogens can promote B cell proliferation and Ig production 46 . In 2014, The US National Institutes of Health (NIH) introduced policies about sex inclusion, to promote equal use of males and females in cell and animal research, as present with NIH-funded clinical research 47 . However, conclusions in animal experiments are often solely based on results from either sex or results are not corrected for the sex-effect. The sex-effects observed in our study underline the importance of equally including both males and females in immunological animal experiments. The current literature supports the findings by van den Elsen, et al. 34 , that oligosaccharides have a positive effect on vaccine responses. Therefore, our findings, that oligosaccharides had a negative effect on vaccine responses in male mice, are scientifically novel. Nonetheless, it is worth to note that antibody titers are a measure of immunogenicity and do not definitely tell if the animal is fully protected against disease. Therefore, more exact conclusions about disease prevention can be obtained with challenge experiments. Since the purpose of this project have been to study the role of oligosaccharides on antibody titers as well as some ethical aspects, challenge experiments have been disemployed. www.nature.com/scientificreports/ The fact that Immulix feeding upregulated Il1b in the spleen (p = 0.001), but did not affect IL-10 and T or B cell surface markers, substantiates the lack of effect on vaccine responses. An HP diet has previously proven to increase the appearance of T-cell subsets in ex-germ-free mice inoculated with a human microbiota 35 , but our study seems to indicate that in mice with their own microbiota, Immulix may dampen that effect, as Immulix fed in combination with the HP diet downregulated the expression of Cd4 (Fig. 5A) and Foxp3 (Fig. 5B) in ileum. As prevailing with the IgG results, sex was a considerable factor with higher expression levels of Cd4 (p = 0.008) and Cd8a (p = 0.023) in ileum as well as Cd19 (p = 0.005) and Foxp3 (p = 0.016) in the spleen of female mice, compared to males. This was moreover the case in the study by van den Elsen, et al. 34 , in which female mice expressed significantly higher levels of IL-2, IL-6 and IL-10 in splenocyte supernatants harvested 72 h after ex vivo TIV stimulation. The T cell markers (Cd4 and Cd8a) and the B cell marker (Cd19) might be upregulated as a result of stronger vaccine responsiveness in females compared to males. Dietary fat content affected gene expression in the spleen, with higher expression of Il1b (p = 0.001), but lower levels of Cd8a (p = 0.035) in mice fed the HP diet, compared to MP diet fed mice. Moreno-Indias, et al. 35 reported lower Cd8a expression levels in colon of ex-germ-free mice inoculated with mouse microbiota, than with human microbiota 35 , which supports that our results are specific for the outcome in mice with a murine microbiota. Also, the results by Moreno-Indias, et al. 35 were detected in ileum and colon, indicating local responses, whereas this present study only revealed significant results in the spleen, which is more the indication of a systemic response. Interestingly, dietary fat content had an upregulating effect on il6 and il1b in females ( Fig. 4A and B), but a downregulating effect on il6 and cd4 in males ( Fig. 4D and E), possibly as a result of oestrogen-driven sex differences on lipid metabolism 48 . In accordance, Wallace, et al. 49 found reduced proinflammatory cytokines in spleen lymphocytes (either Concanavalin A stimulated or incubated with various fatty acids) from male mice fed high-fat diets (21%; coconut, safflower, or fish oil) compared to a low-fat diet (2.5%; maize oil). Unfortunately, no females were included in that experiment. As an extension to the sex differences found in our experiments, it is enviable to investigate the hormones' role on the immune response. Oligosaccharides are known modulators of the GM and have proven to increase specific beneficial bacteria. In horses, Immulix has previously proven to increase abundances of Akkermansia spp. 12 , which has an immune modulating effect 50,51 . In our study, Akkermansia were not present in the mice, which was also the case in the characterization of faecal microbiota of BALB/c mice in the study by Krych, et al. 52 . Hänninen, et al. 51 transferred GM from non-obese diabetic (NOD) mice with a low diabetes incidence to NOD mice with a high diabetes incidence without affecting the diabetes incidence because Akkermansia muciniphila was one of few taxa not successfully transferred. However, when A. muciniphila was orally administered, diabetes incidence decreased significantly 51 . This indicates that despite major alterations in GM, the absence of one single taxon www.nature.com/scientificreports/ can have considerable consequences. Especially Lactobacilli and Bifidobacteria have extensively been reported to increase as a result of oligosaccharide feeding 10,11,53,54 . In this study, Bifidobacteria were not found, except in four mice from the HP + Imx group, where Bifidobacterium animalis were abundant (0.31-14.5%), and Immulix actually had a downregulating effect on expression of Cd4 and Foxp3 in ileum of mice fed the HP diet. Two weeks of probiotic supplementation (containing various amounts of Bifidobacteria and Lactobacilli spp.) did not seem to be sufficient to establish colonies of Bifidobacteria. Several Lactobacilli spp. were abundant in this study, but unexpectedly Immulix decreased the order Lactobacillales (q = 0.035), whereas no significant changes were observed for Lactobacillus animalis. Interactions and synergistic effects of Bifidobacteria and Lactobacilli are known to occur, and these bacteria are often upregulated simultaneously 11,55,56 . Therefore, the absence of Bifidobacteria might have affected abundances of Lactobacilli as well. The genus Bacteroides was in this study upregulated by Immulix feeding, which might have outcompeted Bifidobacteria, Lactobacillales and Prevotella in those mice, which is in compliance with Hansen, et al. 57 , who found that xylooligosaccharide feeding tended to decrease Prevotella to the detriment of upregulated Bacteroides and especially Parabacteroides. GM composition has previously been correlated with vaccine responses. For instance, vaccine specific antibody responses have been positively correlated with Bifidobacteria and Lactobacilli 10,58 and negatively correlated with Clostridiales 59 . None of those bacteria were affected by Immulix feeding or correlated with vaccine responses in our study. The Proteobacteria/Bacteroidetes ratio have furthermore proven to affect vaccine responses, with Proteobacteria being a stronger immune stimulator than Bacteroidetes 16,60 . In our study, Proteobacteria were unaffected, and Bacteroides spp. were upregulated in Immulix fed mice. Additionally, Immulix decreased Prevotella, which have been correlated with T h 1 and T h 17 response 61 . Hence, the differences in GM of mice fed Immulix and those www.nature.com/scientificreports/  Figure 6. Heatmap presenting taxa affected significantly (q < 0.05) or tended to be affected (p < 0.05) by either diet or Immulix. p-values and q-values (FDR-corrected p-values) obtained by multifactorial general linear model ANOVA on factors Immulix, diet and sex, based on rarefied (10,000) OTU-tables of faecal samples from week 12. Scale on the right indicates relative abundance (%). www.nature.com/scientificreports/ not fed Immulix, do not indicate that Immulix fed mice should respond better to a vaccine. With this in mind, it is not surprising that vaccine responses, immune cells and cytokine profiles were not affected dramatically by the treatments. Therefore, it might be argued that this BALB/c murine model may not be a highly appropriate model for testing oligosaccharides, and at least this is very dependent on its gut microbiota, which varies substantially between vendors 62 . The GM of laboratory mice often has a low diversity and is very different from feralanimals 63 . Therefore, when using this model for testing an equine oligosaccharide-based feed supplement, it could be considered to inoculate the mice with microbiota from foals as demonstrated by Lindenberg et al. 12 , and to orally administer specific target bacteria, which was successfully demonstrated by Hänninen et al. 51 . In Table 2. Correlation coefficients between bacterial abundances (based on rarefied (10,000) OTU-tables) and remaining data that tended to be significant (p-value < 0.05; q-value (FDR-corrected p-values) < 0.1).  Figure 7. Weight gain within feeding groups over the experimental period from week 0 to week 12. Three animals from theMP + Imx group were euthanized at week 8 due to weight loss and impairment. (A): Body weight, females. (B): Body weight, males. (C): Body weight, males, where mouse no. 29, 30 and 36 were excluded. A multifactorial general linear model ANOVA on factors Immulix and diet revealed significantly higher weight gain (AUC) of mice fed the HP diet, compared to the MP diet (p = 0.034), but no significant effect of Immulix (p = 0.475). However, if data from mouse no. 29, 30 and 36 were excluded, this effect disappeared (Immulix p = 0.992; diet p = 0.149), indicating that the increased weight gain was caused by the abnormal weight loss from those three mice, rather than a factor-effect. www.nature.com/scientificreports/ order to achieve better inoculation efficiency, the mice diet could have been modulated to resemble a horse diet either with respect to macronutrient content or source, as demonstrated by Moreno-Indias, et al. 35 for human microbiota colonization. However, it should be noted that xenografted microbiotas do not stimulate the immune response very well 64 .
In conclusion, Immulix feeding or increased dietary fat content significantly modified the GM of mice, but neither of these had any biological relevant effects on tetanus vaccine responses in the mice, despite minor influences on immune cell markers, cytokines and IgG titers. Additionally, significant sex differences existed, with female mice exhibiting higher and more diverse immune responses than males, and the antibody response of the males in contrast being negatively affected by the oligosaccharide feeding.