Host immunity and the colon microbiota of mice infected with Citrobacter rodentium are beneficially modulated by lipid-soluble extract from late-cutting alfalfa in the early stages of infection

Alfalfa is a forage legume commonly associated with ruminant livestock production that may be a potential source of health-promoting phytochemicals. Anecdotal evidence from producers suggests that later cuttings of alfalfa may be more beneficial to non-ruminants; however, published literature varies greatly in measured outcomes, supplement form, and cutting. The objective of this study was to measure body weight, average daily feed intake, host immunity, and the colon microbiota composition in mice fed hay, aqueous, and chloroform extracts of early (1st) and late (5th) cutting alfalfa before and after challenge with Citrobacter rodentium. Prior to inoculation, alfalfa supplementation did not have a significant impact on body weight or feed intake, but 5th cutting alfalfa was shown to improve body weight at 5- and 6-days post-infection compared to 1st cutting alfalfa (P = 0.02 and 0.01). Combined with the observation that both chloroform extracts improved mouse body weight compared to control diets in later stages of C. rodentium infection led to detailed analyses of the immune system and colon microbiota in mice fed 1st and 5th cutting chloroform extracts. Immediately following inoculation, 5th cutting chloroform extracts significantly reduced the relative abundance of C. rodentium (P = 0.02) and did not display the early lymphocyte recruitment observed in 1st cutting extract. In later timepoints, both chloroform extracts maintained lower splenic B-cell and macrophage populations while increasing the relative abundance of potentially beneficially genera such as Turicibacter (P = 0.02). At 21dpi, only 5th cutting chloroform extracts increased the relative abundance of beneficial Akkermansia compared to the control diet (P = 0.02). These results suggest that lipid soluble compounds enriched in late-cutting alfalfa modulate pathogen colonization and early immune responses to Citrobacter rodentium, contributing to protective effects on body weight.


Introduction
Phytochemicals are a class of natural compounds that may improve livestock health through direct and indirect action on the host immune system and intestinal microbiota [1,2]. One dietary phytochemical source is alfalfa, a documented source of bioactive compounds associated with a number of health benefits such as saponins [3,4], phytoestrogens/flavonoids [5,6], and non-cellulosic polysaccharides [7,8]. Alfalfa, a perennial legume forage, allows for multiple harvests to occur within a growing season. While an established ruminant livestock feed ingredient, alfalfa is used less often in non-ruminant diets. Dietary alfalfa inclusion for nonruminant livestock is limited due to the high indigestible fiber content, reducing dietary energy density. While the fiber content has variable animal performance impacts, it may provide a substrate for microbial fermentation to modulate poultry and swine intestinal microbiota in favor of beneficial community members [9][10][11][12][13][14]. Anecdotal evidence from swine producers suggests that feeding late-cutting alfalfa may confer greater health benefits to sows and piglets compared to earlier cuttings. The use of alfalfa extracts can possibly circumvent fiber-associated limitations while still conferring immunological benefits. In poultry and swine, crude aqueous alfalfa extract and protein-rich concentrate increased lymphocyte proliferation while chloroform extracts improved mouse survival during LPS challenge and suppressed production of pro-inflammatory interleukin (IL)-6 in vitro [15][16][17][18].
Compounds that may be enriched in these extracts have documented impacts on the immune system and intestinal microbiota. Alfalfa-derived non-cellulosic polysaccharides isolated from alfalfa reduced pro-inflammatory cytokine expression in cultured murine macrophages, while soybean isoflavones and saponins from other plant sources have been shown to increase peripheral blood lymphocytes, proportions of cytotoxic T-cells, serum immunoglobulin (Ig)G, and intestinal IgA-secreting cells in swine, poultry, and mice [7,8,[19][20][21][22][23]. In rodent models of intestinal and metabolic disease, saponins, flavonoids, and polysaccharides from various sources altered the relative abundance of microbial phyla to profiles more similar to healthy controls [24][25][26][27][28][29]. Factors such as plant maturity, season, cutting, and time of harvest have documented impacts on alfalfa's nutritional and phytochemical profiles, suggesting that feeding different cuttings may differentially impact immunity and the microbiota [6,[30][31][32][33][34].
While alfalfa supplementation has documented benefits, the use of livestock models is limited by the lack of reagents available to descriptively assess immunological and microbiological outcomes. In this study, mice were used to evaluate changes to the immune system and colon microbiota due to their well-characterized microbiome, wide variety of available immunological reagents, and the ability to select dietary treatments of interest for future livestock work [35,36]. To assess alfalfa's effect in both healthy and pathogen-challenged animals, rodentspecific Citrobacter rodentium was administered because it has well-documented intestinal microbiota impacts, a known infection timeline, and colonizes the rodent digestive tract more successfuly than Salmonella or Escherichia coli [37][38][39]. The objective of the work presented here was to assess body weight (BW) and average daily feed intake (ADFI) as general health responses in mice fed early (1 st ) or late (5 th ) cutting alfalfa as hay, aqueous extract (water-soluble), and chloroform extract (lipid-soluble) before and after C. rodentium challenge. Noted trends and responses in BW and ADFI resulted in detailed examination of chloroform extracts and their impacts on immunity and the colon microbiota before and after pathogen challenge.

Materials and methods
All animal protocols were approved by the Iowa State University Institutional Animal Care and Use Committee (Protocol # 11-17-8643-M). All euthanasia procedures used isofluorane gas anesthesia with cervical dislocation as a secondary method. Mice were monitored daily during the infection period with the loss of the righting reflex defined as a humane endpoint requiring euthanasia before a planned sampling timepoint. No animals displayed behaviors related to this humane endpoint and none were found dead during the 35d study.

Experimental design
A total of 163 6-week-old female C57BL/6J mice were obtained from Jackson Laboratories (Bar Harbor, ME) and housed in 45 Innovive cages (2-4 mice/ cage; Innovive Inc., San Diego, CA) using an ear punch to identify individuals within each cage. Following their arrival at Iowa State University, animals were given a 7d acclimation period to allow stabilization of the microbiota after transport and diet change to the Teklad Global 18% Protein Rodent Diet (Envigo Teklad, Madison, WI). After the adaptation period (d0), spleens and colon digesta were collected from 9 mice to establish a baseline and the remaining mice were randomly assigned to 1 of 7 dietary treatments (22 mice/ treatment). These treatments consisted of the Teklad Global 18% Protein Rodent Diet without alfalfa (control) or supplemented with 9% hay, 0.25% aqueous extract, or 0.25% chloroform extract of 1 st or 5 th cutting alfalfa (Table 1). All diets were formulated to be isocaloric and isonitrogenous by Envigo Teklad and offered to mice ad libitum in a pelleted form.
The study was divided into a 14d feed enrichment period and 21d infection period with C. rodentium, totaling 35d overall. Mice weighing 18-20g are most susceptible to C. rodentium and female 6-week-old mice were used to ensure that mice would be in this BW range after the 2-week feed adaptation period [38]. Following the feed adaptation period, 6 mice/ treatment were euthanized for tissue sampling and remaining animals were inoculated by oral gavage with 200μl of 2×10 10 colony forming units (CFU) of C. rodentium. Preparation of the inoculum was done according to methods published by Crepin et al. [38]. Briefly, C. rodentium strain DBS100 (ATCC 51459; Manassas, VA) was cultured overnight in 15ml lysogeny broth (LB) at 37˚C in a shaking incubator (200 rpm). The next morning, cultures were centrifuged (3000 x g for 10 minutes) and bacteria were resuspended in 15ml sterile phosphate-buffered saline (PBS). 100μl of resuspended bacteria was serially diluted and plated on LB to determine CFUs. Following inoculation, 4 mice/ treatment were euthanized for tissue sampling on 4, 8, 14, and 21d post-infection (dpi) in accordance with timelines published by Crepin et al. [38]. The trial concluded at 21dpi when the last subset of mice was euthanized.

Body weight and feed intake monitoring
Individual mouse BW was recorded on arrival, d0, d14, and daily following C. rodentium inoculation. Feeders were replenished when necessary and feed intake for each cage was calculated by subtracting leftover feed from both the feeder and cage bottom from the amount of added feed. The total FI from d0-14, 0-4dpi, 4-8dpi, 8-14dpi, and 14-21dpi was calculated and presented as ADFI per mouse. Following C. rodentium inoculation, the return to pre-inoculation BW was used as a general indicator of recovery and used to identify specific treatments for downstream analysis of the immune system and microbiota.

Flow cytometry
Mouse spleens were gently homogenized in PBS and filtered through a sterile 70μm strainer. Red blood cells were lysed using ACK lysing buffer (Gibco, Fisher Scientific, Hampton, NH) and the obtained cells were washed twice in PBS. Splenocytes were resuspended in RPMI, enumerated using a hemocytometer and stored at -80˚C in cryotubes with heat-inactivated bovine calf serum (BCS) supplemented with 7.5% DMSO until analysis.
Fluorescence minus one (FMO) staining protocols were used with appropriate isotype controls to account for non-specific binding. Following incubation at 4˚C for 30 minutes in the dark, cells were washed in PBS, fixed, and permeabilized according to the manufacturer's instructions using the eBioscience™ Foxp3/ transcription factor staining buffer kit (Cat. No. 00-5523-00; Thermo-Fisher Scientific Corporation, Carlsbad, CA). Cells were stained for intracellular cytokines diluted in permeabilization buffer for 30 minutes at room temperature in the dark. Antibodies for intracellular cytokines were interferon (IFN)-γ APC (clone XMG1.2; rat IgG1,κ), tumor necrosis factor (TNF)-α PE (clone MP6-XT22; rat IgG1,κ), IL-17A Brilliant Violet™ 650 (clone TC11-18H10.1; rat IgG1,κ), and IL-22 Alexa Fluor 1 647 (clone Poly5164; goat polyclonal IgG; BioLegend). After staining, cells were washed twice in permeabilization buffer and resuspended in cell staining buffer. Cells were stored in the dark at 4˚C until immune cell populations could be analyzed using a BD FACSCanto™ cytometer (BD Biosciences). Analysis of cell population data obtained from the cytometer was completed using FlowJo 10.5.0 software.

DNA extraction and sequencing
Colon digesta was collected by incising the tissue and transferring the contents into sterile tubes using sterile forceps before freezing at -80˚C. DNA was extracted from the colon digesta using the DNeasy PowerLyzer PowerSoil Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. A Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA) was used to quantify the extracted DNA before being frozen at -20˚C. Prior to PCR amplification and sequencing, extracted DNA was diluted to approximately 30ng/μl. Microbiota sequencing was conducted using a protocol designed to amplify bacteria and archaea in the DNA facility at Iowa State University (Ames, IA) [40]. Briefly, genomic DNA from each sample was amplified using Platinum™ Taq DNA Polymerase (Thermo Fisher Scientific, Waltham, MA) with one replicate per sample using universal 16S rRNA gene bacterial primers [515F (5 0 -GTGYCAGCMGCCGCGGTAA-3 0 [41], and 806R (5 0 -GGACTACNVGGGTW TCTAAT-3 0 [42]] for the variable region V4, as previously described [43]. All samples underwent PCR with an initial denaturation step at 94˚C for 3 min, followed by 45s of denaturing at 94˚C, 20s of annealing at 50˚C, and 90s of extension at 72˚C. This was repeated for 35 cycles and finished with a 10 min extension at 72˚C. PCR products were purified with the QIAquick 96 PCR Purification Kit (Qiagen Sciences Inc, Germantown, MD) according to the manufacturer's recommendations. PCR bar-coded amplicons were mixed at equal molar ratios and used for Illumina MiSeq paired-end sequencing with 150 bp read length and cluster generation with 10% PhiX control DNA on an Illumina MiSeq platform (Illumina Inc., San Diego, CA).

Statistics
Mouse BW and ADFI data were analyzed using the following statistical model: Where y is the dependent variable (BW or ADFI), μ is the overall mean, Con i is the effect of the control at the i th level (i = 1), Cut (i)j is the fixed effect of the j th level of alfalfa cutting nested within the control (1 st or 5 th ; j = 2), F (i)k is the fixed effect of the k th level of form nested within the control (hay, aqueous extract, or chloroform extract; k = 3), (Cut × F) (i)jk is the fixed effect of the interaction between cutting at the j th level and form at the k th level nested within the control, d0BW (i)jkl is the covariate of d0 BW associated with each observation, and e (i)jkl is the random error. This model was used due to the 2 × 3 + 1 factorial treatment design with the control, 2 cuttings of alfalfa, and 3 supplementation forms.
Data obtained by flow cytometry were analyzed using the following statistical model: In this model, y is the dependent variable (cell population), μ is the overall mean, τ i is the effect of treatment at the i th level (control, 1 st cutting chloroform extract, or 5 th cutting chloroform extract), and e ij is the random error associated with the j th replicate (mouse) from the i th level of treatment.
Both models were analyzed using the MIXED procedure of SAS 9.4 (SAS Institute, Cary, NC). The Satterthwaite method for degrees of freedom and the repeated statement by treatment group were used to analyze data under the assumption of unequal variance between treatments. Significance was denoted at P � 0.05.

Microbiota sequencing data analysis
Sequencing data were assessed for quality and screened using mothur (v.1.40.4) [44]. Prior to clustering into operational taxonomic units (OTUs), paired-end reads were merged and sequences with ambiguous bases were removed. Sequences shorter than 250bp and longer than 255bp were removed, in addition to those with > 8 identical consecutive bases. The remaining sequences were then randomly subsampled to 20,000 sequences/ sample. Unique sequences meeting these criteria were aligned to the SILVA v132 database. Potential chimeric sequences were removed using Vsearch and sequences within 2 mismatches of the aligned sequences were also removed before being clustered into OTUs at 97% similarity, resulting in a total of 484,558 OTUs. The SILVA SSU reference database version 132 was used as taxonomic reference [45]. Whole community comparisons were done using analysis of similarities (ANOSIM), while differences between treatments at an OTU level were analyzed using linear discriminant analysis (LDA) effect size (LEfSe) [46] implemented in mothur.

BW and ADFI
The study arrangement presented here allowed for the assessment of baseline responses to alfalfa supplementation in addition to determining responses during infection with a rodentspecific pathogen. No BW differences were observed during the feed enrichment period (Fig  1). Similarly, no differences in ADFI were observed during the enrichment or later timepoints during C. rodentium challenge ( Fig 2); however, significant reductions in ADFI for mice fed aqueous extract versus hay from 0-4dpi and 4-8dpi were noted (Fig 2A). Regardless of cutting, mice fed hay-supplemented diets ate 0.5g ± 0.1g (18.1%) more than mice given diets with aqueous extract in the first 4dpi (P = 0.005; Fig 2A). From 4-8dpi, feeding hay increased mouse ADFI by 0.5 ± 0.2g (17.2%) compared to mice fed aqueous extracts (P = 0.04). In the early timepoints following inoculation (0-4dpi and 4-8dpi), BW and ADFI were expected to

PLOS ONE
Late-cutting lipid alfalfa extract beneficially impacts immune response and microbiome drop with recovery occurring as C. rodentium was cleared during later timepoints (8-14dpi and 14-21dpi).
Post-inoculation, ADFI was unchanged by alfalfa-supplemented diets compared to the control; however, varying significant responses and trends in BW were observed (Figs 2-4). In the first 4dpi, feeding 5 th cutting alfalfa positively impacted BW, regardless of supplementation form (Fig 3) and significant differences between the two cuttings were observed within the first 7dpi. Compared to 1 st cutting alfalfa, BW was increased in 5 th cutting-supplemented mice by 0.5g ± 0.2g (2.6%) and 0.4g ± 0.2g (2.0%) at 5 and 6dpi, regardless of form (P = 0.002 and 0.01; Fig 4A). In the later timepoints of the infection (14-21dpi), the main effects of both aqueous and chloroform extracts positively impacted BW (Fig 4B and 4C). At 15dpi, mice fed aqueous and chloroform extracts weighed 1.4g ± 0.7g (7.1%) and 1.9g ± 0.7g (9.5%) more than mice fed hay, respectively, regardless of cutting (P = 0.03; Fig 4B).
In examining recovery post-inoculation, mice fed control diets recovered to pre-infection BW (18.9g) at 4dpi but experienced additional losses in BW. Full recovery to pre-infection BW in mice fed the control diet occurred at 13dpi followed by an increase to a final BW of 19.7g (4.1% increase over pre-infection; Fig 5). Mice fed 5 th cutting chloroform extract displayed an expedited recovery to pre-infection BW at 2dpi, followed by a BW that stayed numerically higher than the control until 9dpi (Fig 5F). For 4 days in the middle of the infection (10-13dpi),

PLOS ONE
Late-cutting lipid alfalfa extract beneficially impacts immune response and microbiome the BW of mice fed 5 th cutting chloroform extracts dropped to 1g (5.3%) below the pre-infection BW before achieving a final recovery at 14dpi and increasing to a final BW 0.2g (1.0%) above the control. In contrast, mice fed diets with 1 st cutting chloroform extract did not show recovery to pre-infection BW until 9dpi but did not experience BW loss following recovery and increased to a final BW that was 0.6g (2.7%) greater than control. Notably, mice fed the chloroform extracts from either cutting had final BW that were greater than the control (Fig 5E  and 5F).

Cytokine-producing cells
The observed patterns in BW and ADFI suggested that chloroform extracts had the potential to improve these parameters as general health indicators during C. rodentium infection. Analysis of the immune system and intestinal microbiota in these treatments was conducted in order to investigate some of the underlying physiologies contributing to these observations. Pro-inflammatory interferon (IFN)-γ is a cytokine produced by a number of cell types including antigen-presenting cells and both CD4 + and CD8 + T-cell subpopulations [47]. In healthy mice at d14, chloroform extracts from 1 st cutting alfalfa increased the percentage of IFN-γ-producing cells over the control by 14.7%, respectively (P = 0.002; Fig 6A). Over the course of infection with C. rodentium, mice fed 1 st and 5 th cutting chloroform extracts displayed 50.0 and 28.2% reductions in IFN-γ + cells from 0-4dpi to levels 44.9 and 37.9% below the control, respectively (P < 0.0001). At 14dpi, splenic populations of IFN-γ-producing cells in mice fed the control diet were reduced by 51.6% to levels 28.8 and 26.2% below 1 st and 5 th cutting chloroform extracts, respectively (P = 0.003). Notably, at this point in the infection, splenic IFN-γ + cells increased by 18.9% in diets supplemented with 1 st cutting chloroform extract but remained constant in mice fed diets with 5 th cutting extract. While no differences between each treatment were observed in the last infection timepoint (21dpi), mice fed 1 st cutting chloroform extracts had splenic percentages of IFN-γ + cells 32.1% below pre-infection levels while both the control and 5 th cutting chloroform extract diets showed recovery in this cell population (Fig 6A).
Co-expression of IFN-γ with CD4 or CD8 was measured to identify splenic T-cell populations underlying the observed IFN-γ responses. Of the measured IFN-γ + cells, T-helper 1 (T H 1; IFN-γ + CD4 + ) cells comprised 20-35% of cytokine production at a baseline health state (Fig 6B), while IFN-γ + CD8 + cells accounted for a lower percentage (4-7%; S1A Fig). In healthy mice, 1 st cutting chloroform extract-supplemented diets increased T H 1 cells by 12.1% compared to control (P < 0.0001). Throughout the infection, the percentage of splenic T H 1 cells in the control diet remained relatively constant, while shifts in these populations were observed in both chloroform extract diets. From 0-4dpi, mice fed 5 th cutting chloroform extract-diets showed increases in T H 1 cells by 22.3% to levels 21.9% above the control (P = 0.0009). At 14dpi, splenic T H 1 cells decreased by 31.3% in mice fed 1 st cutting chloroform extract to levels approximately 25% less than both the control and 5 th cutting chloroform extract diets (P < 0.0001). In the final days of the infection from 14-21dpi, T H 1 cells increased in both 1 st and 5 th cutting chloroform extract diets by 30.4 and 16.1%, respectively. At 21dpi, all treatments showed percentages of T H 1 slightly elevated above pre-infection levels with 5 th cutting chloroform extracts having 15.4% greater percentages of this cell type compared to the control (P = 0.001; Fig 6B).
Another pro-inflammatory cytokine, TNF-α, was also measured using intracellular cytokine staining. This cytokine is produced by both innate and adaptive immune cell types [48][49][50]. Compared to IFN-γ, the percentage of TNF-α + cells in the spleens of healthy mice was considerably lower (~1% compared to~50-60%; Fig 6C). Diets supplemented with 1 st cutting chloroform extract had the greatest percentage of TNF-α + cells at d14 with levels 45.8 and 43.5% above both the control and 5 th cutting chloroform extract diets (P < 0.0001). From 4-14dpi, while control and 1 st cutting chloroform extract diets maintained TNF-α + populations, 5 th cutting chloroform extracts displayed a 38.0% reduction in these cells to levels 28.1% below the 1 st cutting chloroform extract diet (P = 0.003). In the last phase of the infection from 14-21dpi, all treatments showed increases in the percentage of TNF-α + cells, with the greatest increase observed in 5 th cutting chloroform extracts (75%) to levels 47.4 and 36.2% above both the control and 1 st cutting chloroform extract diets (P < 0.0001; Fig 6C).
Production of TNF-α in two different cell types was conducted to determine which populations were responsible for the observed changes: macrophages and CD3 + CD8 + T-cytotoxic (T C ) cells. Production of TNF-α is often associated with macrophage activity, among other outcomes; however, only a small percentage of splenic macrophages stained positively for TNF-α (1-2%) and showed consistent changes over the course of infection across all treatments (S1B Fig). At a baseline health state, TNF-α + staining was observed in approximately 5% of T C cells with 5 th cutting chloroform extracts having 35% lower percentages of this cell type compared to 1 st cutting chloroform extracts (P = 0.01). Notably, percentages of TNF-α + T C cells increased in the spleen at later timepoints during the infection in a pattern similar to overall TNF-α + cells (Fig 6D). At 21dpi, 1st and 5 th cutting chloroform extracts had 43.5 and 20.8% more TNF-α + T C cells than the control, respectively, with 5 th cutting chloroform extracts having 28.6% more of these cells than 1 st cutting chloroform extract (P < 0.0001).

Innate immune cells
Two innate immune cells measured in the spleen were neutrophils and macrophages based on the expression of CD11b (S1C Fig) with other extracellular markers. Neutrophils (CD11b + Ly6G + ) were detected in the spleens of healthy mice at low percentages (3-4%). Healthy mice (d14) mice fed 5 th cutting chloroform extracts had 28.0 and 20.3% fewer neutrophils than mice fed the control and 1 st cutting chloroform extract diets, respectively (P = 0.004; Fig 7A). Throughout the course of infection, mice fed the control and 5 th cutting chloroform extract diets did not show notable changes to neutrophil populations, despite observed differences in the percentages of these cells between the two treatments. In the early timepoints of the infection, mice given 1 st cutting chloroform extract showed a 49.4% reduction in neutrophils to levels 52.4 and 20.5% below the control and 5 th cutting chloroform extract diets, respectively, at 4dpi (P = 0.0002). Mice fed 1 st cutting chloroform extract displayed a notable 81.8% increase in neutrophils from 4-14dpi to levels 54.1 and 58.9% above the control and 5 th cutting chloroform extract diets, respectively (P < 0.0001). In the final timepoint of the infection, both mice fed the control and 1 st cutting chloroform extract diets had levels of splenic neutrophils above preinfection levels and had populations of this cell type approximately 46% above mice fed 5 th cutting chloroform extract (P < 0.0001; Fig 7A).
Compared to neutrophils, populations of macrophages (CD11b + Ly6G -F4/80 + ) in the spleen were greater, with detected populations accounting for approximately 15-17% of all CD11b + cells with no differences detected between treatments (Fig 7B). Over the course of the infection, mice fed control diets experienced a 66.1% reduction in macrophages at 4dpi. From 0-4 dpi, 1 st and 5 th cutting chloroform extracts showed a similar decrease in macrophages but to a lesser degree than the control with 42.6 and 43.6% reductions, respectively, to levels 45.2 and 36.5% above the control (P < 0.0001). Mice fed the control diet had a 57.1% increase in macrophage populations from 4-14dpi. During this time, both chloroform extract diets maintained percentages of these cells at levels that were 31.8 and 42.8% below the control, respectively (P = 0.002). From 14-21dpi, control and 1 st cutting chloroform diets showed 23.6 and 13.2% reductions in splenic macrophages, whereas 5 th cutting chloroform extracts maintained this cell population at approximately 7.5% (23.8% less than control; P = 0.02). None of the treatments displayed a return to pre-infection levels of macrophages by the end of the infection period (Fig 7B).

Adaptive immune cells
The spleen is characterized as a secondary lymphoid organ and maintains discrete populations of lymphocytes [51]. Approximately 40-50% of the live cells detected in the spleens of healthy mice were identified as B-cells (B220 + ; Fig 8). After the 14d feeding enrichment period, mice fed both 1 st and 5 th cutting chloroform extracts had 23.2 and 30.7% more B-cells than the control, respectively (P < 0.0001; Fig 8A). At 4dpi, control diets showed a 33.2% increase in B-cells, while 5 th cutting chloroform extracts maintained splenic B-cell populations. Most notably, 1 st cutting chloroform extracts showed a 58.2% reduction in B-cells from 0-4dpi to levels 63.7 and 60.0% below the control and 5 th cutting chloroform extract diets, respectively (P < 0.0001). At 14dpi, B-cell populations in the spleens of mice fed the control diet were reduced by 45.7% to levels below both chloroform extracts (P < 0.0001). Mice fed 5 th cutting chloroform extracts showed a similar decrease in splenic B-cells from 4-14dpi, but to a lesser degree than observed in the control diet (27.4%). In contrast, mice fed 1 st cutting chloroform extract displayed a 53.1% increase in splenic B-cells at this timepoint. In the final timepoint of infection, B-cell populations remained below pre-infection levels in the spleens of mice fed both chloroform extracts, while the control diet showed recovery to pre-infection levels ( Fig 8A).
Changes to overall populations of CD3 + T-cells were similar to those observed in B-cells. At a baseline health state, both chloroform extracts increased splenic T-cell populations compared to control, with 1 st cutting chloroform extracts having the greatest T-cell presence at 14.1% (P < 0.0001; Fig 8B). At 4dpi, mice fed the control diet had a 43.4% increase in the percentage of T-cells, while those fed 5 th cutting chloroform extract maintained populations of these cells. Similar to patterns observed in B-cells, 1 st cutting chloroform extract diets showed early recruitment of T-cells characterized by a 40.7% reduction to levels below both the control and 5 th cutting alfalfa diets (P < 0.0001). At 14dpi, mice fed the control diets experienced a 49.9% reduction in T-cells while those fed 5 th cutting chloroform extract showed a 35.1% reduction in this population. Mice fed 1 st cutting alfalfa showed further reductions to levels maintained below 5 th cutting chloroform extract (P = 0.04). In the last day of the infection period (21dpi), all treatments showed increases in T-cell populations, but only 1 st cutting chloroform extract diets did not display recovery to T-cell populations above pre-infection levels and remained lower than those seen in mice fed 5 th cutting chloroform extract (P < 0.0001; Fig 8B).
T-cell populations were further divided into CD3 + CD4 + T H and CD3 + CD8 + T C subpopulations to identify which were responding to alfalfa supplementation and C. rodentium infection. Notable differences in T H populations between treatments were not observed after the feeding enrichment period and changes over the course of the infection were consistent across treatments (Fig 8C). In contrast, changes to T C cells roughly corresponded to changes in overall Tcell populations (Fig 8D).

Colon microbiota
In terms of whole community comparisons, 1 st cutting chloroform extracts did not differ from control as determined by ANOSIM. Feeding diets supplemented with 5 th cutting chloroform extract altered the intestinal microbiota at whole-community levels compared to control and 1 st cutting chloroform extract at both a baseline health state (P = 0.04and 0.03, respectively) and 4dpi (P = 0.03). Chloroform extracts of 5 th cutting alfalfa did not alter the microbiome at a whole-community level at later timepoints of C. rodentium infection (14 and 21dpi; Table 2). Principal coordinates analysis (PCoA) showed variations in the clustering pattern of different dietary treatments throughout the course of the study, with 5 th cutting chloroform extracts tending to cluster differently from both the control and 1 st cutting chloroform extracts (Fig 9). In addition, a clustering of samples according to sampling days was observed indicating that the microbial communities changed over time independent of the dietary additives. At the genus level, OTUs associated with Muribaculaceae and Lachnospiraceae families were among the most represented in the colon microbiota throughout the study (Fig 10). Citrobacter rodentium was detected in the colon microbiota of inoculated mice and identified as OTU 47. In healthy animals (d14), 5 th cutting chloroform extracts resulted in 273 significantly different OTUs compared to control, versus the 164 different OTUs observed between 1 st cutting chloroform extract and the control at this timepoint. Of these altered OTUs, 1 st cutting chloroform extract reduced the relative abundance of Faecalibaculum (OTU 74; P = 0.02), while feeding 5 th cutting chloroform extract increased the relative abundance of more highly abundant OTUs associated with Muribaculaceae (OTUs 1, 14, 26, 34, and 80) out of the 100 most abundant OTUs (S1 and S2 Tables). In particular, 5 th cutting chloroform extract increased the relative abundance of Muribaculaceae OTUs 1, 26, 34, and 80 over both the control and 1 st cutting chloroform diets, which may be contributing to the significantly altered overall community (S2 and S3 Tables).
In the earliest infection timepoint (4dpi), feeding 1 st cutting chloroform extract resulted in 157 different OTUs from the control whereas 5 th cutting extract resulted in 553 significantly different OTUs from the control. These changes are likely due to the greater relative abundance of 13 OTUs associated with Muribaculaceae and three OTUs associated with Lachnospiraceae in mice fed 5 th cutting chloroform extracts (S5 Table). In contrast, feeding 1 st cutting chloroform extract increased 7 OTUs association with Lachnospiraceae and only 1 associated with Muribaculaceae, while also increasing the relative abundance of the genera Lactobacillus (OTU 6; P = 0.03) while reducing the relative abundance of Roseburia (OTU 58; P = 0.03; S4 Table). In addition to increasing OTUs associated with Muribaculaceae, 5 th cutting chloroform extract reduced the relative abundance of Parasutterella (OTU 21; P = 0.02) and Bifidobacterium (OTU 19) compared to control (P = 0.02; S4 Table). The most notable change observed at 4dpi was the significant reduction in the relative abundance of C. rodentium (OTU 47) in the colon of mice fed 5 th cutting chloroform extracts compared to the control (P = 0.02; Fig 11; S5 Table).
Compared to the control, the number of significantly different OTUs between both chloroform extracts did not vary greatly at 14dpi, with 291 different OTUs in mice fed 1 st cutting chloroform extract and 261 in those fed 5 th cutting extract. Both chloroform extracts increased the relative abundance of Turicibacter (OTU 13) compared to control (P = 0.02), with 5 th cutting chloroform extracts having 4.9-fold greater abundance of this OTU compared to 1 st cutting (P = 0.02; Fig 11; S7 and S8 Tables). Additionally, feeding 1 st cutting chloroform extract increased the relative abundance of Bifidobacterium (OTU 19; P = 0.02) and Romboutsia (OTU 49) compared to both the control and 5 th cutting extracts (P = 0.04 and 0.02, respectively; S7 and S9 Tables). Similar to observations at 4dpi, feeding 5 th cutting chloroform extract had greater impacts on highly abundant OTUs associated with Muribaculaceae at this timepoint, with increases to OTUs 4, 45, 57, and 62 compared to control and 1 st cutting (OTUs 4, 57, and 62 only; S8 and S9 Tables). Feeding 5 th cutting chloroform also decreased the relative abundance of Roseburia (OTU 58) compared to control and 1 st cutting extract (P = 0.01 and 0.05, respectively).
At the conclusion of the study, corresponding with C. rodentium resolution (21dpi), mice fed 1 st cutting chloroform extracts had 263 significantly different OTUs compared to the control whereas mice fed 5 th cutting extracts had 466 different OTUs. At this timepoint, mice fed 1 st cutting chloroform extract had increased relative abundance of Roseburia (OTU 58) compared to control and 5 th cutting (P = 0.04), with additional increases in Mollicutes (OTU 67) compared to control (P = 0.05; S10 and S12 Tables). Feeding 5 th cutting chloroform extract increased the relative abundance of Akkermansia (OTU 8; Fig 11) compared to the control (P = 0.02) along with reducing the relative abundance of Oscillibacter (OTU 36) compared to control and 1 st cutting (P = 0.02; S11 and S12 Tables).

Discussion/Conclusions
Body weight and ADFI outcomes were used to focus further investigation into underlying changes to immunity and the colon microbiota in response to the varying forms and cuttings of alfalfa supplemented. In livestock, performance measurements like BW gain and ADFI are commonplace, but these measures are rarely reported in mouse models due to differences in husbandry objectives and difficulties in accurately measuring rodent FI. While not typical for the model used here, results from the BW and ADFI measurements showed no differences in ADFI and BW due to alfalfa supplementation in healthy mice; however, post-inoculation responses varied considerably based on diet. The consistently greater ADFI observed in mice fed hay diets was likely a function of increased dietary fiber reducing dietary energy availability causing mice to increase ADFI to fulfill energy requirements while post-inoculation BW results suggest an inability to adequately compensate during C. rodentium infection (Fig 5A and 5B) [52]. Observed responses throughout the infection period suggest that late-cutting alfalfa had a protective effect on BW in the early timepoints of C. rodentium infection, while supplementation form, particularly chloroform extracts, had greater impacts at later timepoints. Notably, both outcomes were observed in mice fed 5 th cutting chloroform extracts, suggesting that enriched lipid-soluble phytochemicals in the later cutting may have a protective effect on mouse BW during health challenge. These observations led to in-depth analysis into how underlying changes to the immune system and microbiota in mice fed chloroform extracts may be contributing to the observed BW responses before and after infection.
Healthy mice fed chloroform extracts from both 1 st and 5 th cutting alfalfa displayed shifts in splenic immune cell profiles and colon microbiota. After 14d of feeding enrichment, 1 st cutting chloroform extracts contribute to a pro-inflammatory environment as evidenced by the significant increases in both IFN-γ and TNF-α-producing cells compared to the control diet (Fig 6A  and 6C). While 1 st cutting chloroform extracts had greater impacts on cytokine-producing cells, extracts from both cuttings increased lymphocyte populations similar to observations reported in non-ruminant livestock [13,16,17]. At this timepoint, the microbiota of mice fed 5 th cutting extracts differed from both the control and 1 st cutting extracts, likely due to changes in the relative abundance of highly represented Muribaculaceae OTUs. While these changes were determined to be significant, similarities in BW and ADFI at d14 indicate that alterations to the immune system and microbiota were not substantial enough to negatively impact these indicators of general health.
Implementation of a rodent-specific challenge with C. rodentium provided functional insights into physiological changes as they translated to general health responses. Citrobacter rodentium is often limited to the colon and is not known to colonize the spleen, which limits our understanding of the spleen's role during infection [39]. Given the spleen's role in initiating innate and adaptive immune responses to infection, reductions in splenic cell populations may indicate recruitment to sites of infection in the peripheral tissues [51]. Infection with C. rodentium is characterized by a strong IFN-γ and its associated T H 1 response in addition to dysbiosis as pathogen overgrowth displaces the resident microbiota [37,53]. Citrobacter rodentium was not present in the colons of healthy mice and was identified at 4dpi before returning to low or undetectable levels by the later infection timepoints. Both chloroform extracts displayed early recruitment of IFN-γ-producing cells to peripheral tissues as evidenced by reductions in splenic populations at 4dpi versus 14dpi in the control diet ( Fig 6A). Although changes in T H 1 populations were observed in the spleen (Fig 6B), overall T H populations did not display evidence of recruitment to the site of infection ( Fig 8C). Together, these results suggest that recruitment of IFN-γ-producing cells to peripheral tissues were not associated with a T H 1 response. Instead, the observed fluctuations of splenic T H 1 cells were likely a function of shifts in overall IFN-γ + populations and any T H 1 activity during infection was localized to the colon.
During the early timepoints post-inoculation, mice fed 5 th cutting chloroform extracts rapidly recovered to their pre-infection BW within 2d of inoculation, and some of the underlying physiological responses may have contributed to this response. In terms of immunity, 5 th cutting chloroform extracts maintained elevated levels of lymphocyte populations at 4dpi while control diets showed evidence of rapidly expanding cell populations and 1 st cutting chloroform extracts exhibited early recruitment of lymphocytes to peripheral tissues 10d before expected. Additionally, 5 th cutting chloroform extracts significantly reduced the presence of C. rodentium in the colon. Combined, these results suggest that 5 th cutting chloroform extracts reduced the ability of C. rodentium to colonize the murine colon and did not contribute to potentially detrimental early recruitment of adaptive immune cells, which may have contributed to the observed BW responses.
During resolution of C. rodentium infection at later timepoints both chloroform extracts maintained reduced splenic macrophage populations while control diets showed fluctuations in this cell type in later timepoints (Fig 7B). Similarly, both chloroform extracts showed signs of prolonged B-cell responses compared to the control. While splenic B-cells and macrophages did not show recovery to pre-inoculation levels at later timepoints in mice fed both chloroform extracts, both extracts notably increased the relative abundance of potentially beneficial Turicibacter at 14dpi and only 5 th cutting extracts increased Akkermansia at 21dpi (Fig 11). Increased relative abundance of Turicibacter in chickens has been associated with low residual feed intake (RFI), which roughly corresponds with observations that mice fed chloroform extract-supplemented diets ate the lowest amount of feed and weighed more than mice fed the control diet in later timepoints post-inoculation [54]. This suggests that increased Turicibacter abundance may contribute to a similar phenotype in different species; however, the exact similarities remain unclear as RFI is not typically assessed in mouse models. Akkermansia is a genus that is inversely correlated with inflammatory bowel disease and generally regarded as anti-inflammatory [55]. In the final days of C. rodentium infection, both chloroform extracts resulted in numerically greater BW compared to the control, which may be the result of prolonged immune cell responses, increases in beneficial microbial genera, or a combination of the two.
The identity of compounds that may be responsible for these effects remains unclear as utilization of chloroform alfalfa extracts in animal research is limited, with one study examining the effects of chloroform alfalfa extract to date. Work published by Choi and colleagues (2013), found that chloroform alfalfa extract reduced in vitro pro-inflammatory cytokine production during LPS challenge and suggested that palmitic, linoleic, linolenic, and a number of phenolic acids may be causing the observed effects [18]. The structural similarity of alfalfa-derived phytoestrogens to lipid-soluble steroid hormones, suggests that they may also be present in lipidsoluble chloroform extracts; however, their presence was identified in aqueous extract used by Dong and colleagues (2007) and isoflavones are generally regarded as being water-soluble [16,56]. The lack of published literature concerning chloroform alfalfa extracts and the fact that this study did not specifically characterize the phytochemicals present in chloroform extract leaves the identity of these compounds largely unknown.
The concurrent assessment of the immune system and microbial communities of both healthy and pathogen-challenged mice provided insights into changes underlying general measurements of BW and ADFI; however, the assays used to analyze these systems do not provide specific functional insights. While measuring spleen immune cell profiles showed systemic responses to C. rodentium, site-specific responses to C. rodentium were difficult to measure due to the low yield of immune cells obtained from the colon. Despite these limitations, the results obtained by this work demonstrate that lipid-soluble compounds present in alfalfa modulate host immunity and the intestinal microbiota to improve BW in the late stages of an immune challenge, while those derived from late-cutting alfalfa have greater impacts immediately following inoculation. These insights gained from well-characterized mouse model can guide future livestock applications to focus more specifically on the combined benefits of lipid-soluble compounds from late-cutting alfalfa in production animals.