Heat stress and feeding effects on the mucosa-associated and digesta microbiome and their relationship to plasma and digesta fluid metabolites in the jejunum of dairy cows

The intestinal microbiota plays a pivotal role in digestive processes and maintains gut health and intestinal homeostasis. These functions may be compromised by increased environmental heat which in turn reduces feed intake and gut integrity, while activating the intestinal immune system. It remains unknown whether high ambient temperatures, causing heat stress (HS) to dairy cows, disturb the eubiosis of the microbial community and if so, to which extent the reduction in feed intake and the impairment of circulating and intestinal metabolites account for the alterations of the jejunal microbiota. To address these questions, jejunal digesta, mucosa, and plasma samples from cows exposed to heat stress (HS: 28°C, temperature-humidity-index (THI) = 76, n = 10), control conditions (CON: 16°C, THI = 60, n = 10), or pair-feeding (PF: 16°C, THI = 60, n = 10) for 7 d were collected. Digesta fluids were examined for pH, acetate, nonesterified fatty acids (NEFA), glucose, and lactate, while plasma samples were analyzed for glucose, lactate, β-hydroxybutyrate (BHB), triglycerides, NEFA, creatinine and urea. The microbiota of digesta and mucosa samples were analyzed by 16S rRNA sequencing. The α diversity was higher in mucosa than digesta, but not affected by high ambient temperatures. However, the mucosa-associated microbiota appears more responsive to ambient heat than the digesta microbiome. The adaptive responses under HS conditions comprised an increased mucosal abundance of Bifidobacteriaceae , Succinivibrionaceae UCG-001, Clostridia and Lactobacillus . In the digesta, HS has exerted effects on microbial abundance of Coli-dextribacter and Lachnospiraceae


INTRODUCTION
The intestinal tract is colonized by a complex, commensal microbial community that is in permanent interaction with the host's mucosal surface (Wolowczuk et al., 2008).The gut microbiota plays an important role in nutrient digestion, absorption, and innate immune response of the host (Wells et al., 2017).This symbiont coexistence can be disrupted by environmental effects, e.g., antibiotics (Nel Van Zyl et al., 2022), changing portions of feed constitutes such as carbohydrates, fiber or proteins (Kolodziejczyk et al., 2019), and high ambient temperatures (Huus and Ley, 2021), all modifying the living conditions of the gut microbes.High ambient temperatures inducing heat stress (HS) can adversely disrupt the intestinal mucosa and augment the intestinal permeability by altering tight junction proteins and reducing the gut barrier function in pigs (Pearce et al., 2013b, Xia et al., 2022) and cows (Koch et al., 2019).Furthermore, compromising the integrity of the intestinal barrier can result in local inflammation, epithelial leaks, and dysbiosis of the microbial community (reviewed in (Wells et al., 2017).Xia et al. (2022) showed a direct link between impaired barrier function of the ileum and the colon and decreased Lactobacillus and Bifidobacterium in the mucosa of pigs exposed to HS for 72 h (Xia et al., 2022).Previous studies reported Heat stress and feeding effects on the mucosa-associated and digesta microbiome and their relationship to plasma and digesta fluid metabolites in the jejunum of dairy cows that summer compared with spring temperatures negatively influenced the microbial diversity in feces of dairy cows (Li et al., 2020), whereas the microbial α diversity in fecal samples of rats increased after 28 d of heat acclimation (Cao et al., 2021).However, the effect of feed intake reduction with the increase in ambient temperature as well as the change in feed composition from spring to summer were not considered in these studies.Indeed, heat stress has negative effects on feed intake of cows (Wheelock et al., 2010, Rhoads et al., 2011, Lamp et al., 2015), which is a significant factor determining the digesta passage rate and microbial composition of the gastrointestinal tract (Okine and Mathison, 1991).Therefore, the utilization of an additional pair-feeding (PF) group is required to eliminate the confounding effect of dissimilar energy and nutrient intakes of cows fed ad libitum under conditions of thermoneutrality or increased ambient heat.
There is rising evidence that the microbiota composition differs substantially between the luminal and the mucosa-associated site (Walker et al., 2011, Willing et al., 2011).The majority of bacteria occurs mostly in the lumen, while mucin-degrading specialists, e.g., Akkermansia muciniphila, Bacteroides thetaiotaomicron, Bifidobacterium bifidum, B. fragilis, Ruminococus gnavus, and R. torques are also present at the mucosa site, where they promote mucus secretion and prevent pathogen penetration through the gut barrier of humans and mice (Wells et al., 2017).However, it is not known whether high ambient temperatures disturb the eubiosis of the microbiota community in the jejunum of dairy cows and if potential consequences are more pronounced in the mucosa-associated than in the luminal microbiota.
Furthermore, the amount of dietary nutrients in the digesta might also play a pivotal role in nutrient supply not only for the host but also for the microbiota community thus shaping the microbial composition.A recent study demonstrated that gastrointestinal microbiota dysbiosis is linked to the development of feed intake-related diseases such as left displaced abomasum (LDA) in dairy cows (Luo et al., 2022).Moreover, Luo et al. (2022) hypothesized that in the context of LDA, the microbiota dysbiosis is associated with higher serum NEFA and BHB in dairy cows (Luo et al., 2022).Increased plasma NEFA concentrations induced by intravenous lipid infusion reduced the relative abundances of genera belonging to Succinivibrio, Ruminococcaceae, and Ruminiclostridium, and greater relative Bacteroidetes genus abundances in the rumen of dairy cows (Lamp et al., 2018), suggesting that besides dietary effects changes in the concentration of circulatory metabolites exert an influence on the digestal and eventually the epimural microbiome.The hypothesis of the present study was that chronic HS and PF at thermoneutrality have distinct effects on the microbiota composition of the jejunal digesta and mucosa of dairy cows.The study aimed to investigate the relationship between the gut microbiota, digesta fluid, and plasma metabolites under these conditions.

Animals and treatments
A total of 30 primiparous, non-pregnant German Holstein cows, [mean ± SD: 169 ± 48 d in milk (DIM)] from the herd of the institution, were genotyped based on HSP70.1 5`UTR SNPs (Koch et al., 2023) and evenly allocated to a heat stress (HS, n = 10), control (CON, n = 10), and pair-feeding group (PF, n = 10).In each of the 10 blocks, a CON, HS and PF cow were examined in parallel.During the adaptation phase, all cows were moved to climate chambers, received feed for ad libitum intake and allowed to acclimate for 6 d to thermoneutrality (TN) at permanent 16°C and relative humidity (RH) of 69% resulting in a temperature-humidity-index (THI) of 60 (NRC, 1971).In the climate chambers, the day-night-rhythm was given by a light cycle ranging from 0600 to 1900 h.During the subsequent experimental phase, the CON group was further kept at 16°C and 69 ± 2% RH (THI = 60) with ad libitum feeding for 7 d.Cows of the HS group were exposed to 28°C with 51 ± 2% RH (THI = 76 ± 0.2) for 7 d.The HS cows had ad libitum access to feed and water, both tempered to 28°C.The PF cows were fed the amount of feed per kg body weight (BW) the HS cows ingested but were exposed to 16°C and 69 ± 2% RH (THI = 60 ± 0.2) for 7 d.Pair-feeding served as control to eliminate the cofounding effect of reduced energy and nutrient intake of the HS relative to the CON group.Cows were fed a total mixed ration (TMR) at 0730 h and 1730 h and milked at 0700 h and 1730 h.During the adaptation and experimental phases, ambient temperatures of the climate rooms were recorded every 10 min by electronic data loggers (testo 174H, Testo AG, Lenzkirch, Germany) in close proximity of the cow.The body surface temperature was assessed by utilizing thermal camera imaging (T620BX, FLIR Systems, Wilsonville, OR, USA) one day before start of the adaptation phase and on d 6 of the experimental phase.Five to 6 pictures of each animal were taken from both body sides (Figure 1A and B).The evaluation of the same area was performed on mean, minimum (min), maximum (max) of the surface temperature and difference between ambient (taken from electronic data loggers) and surface temperatures were calculated.Due to technical issues, only 9 cows per group could be as-sessed for surface temperatures.Milk yield, feed intake, water intake, BW, and rectal temperature were recorded on d 7 of the experimental phase.Details of the total mixed ration, rectal temperature and respiration rates are reported in Koch et al. (2023).All procedures were evaluated and approved by the ethics committee of the State Government in Mecklenburg-West Pomerania, Germany (LALLF M-V/TSD/7221.3-1.1-60/19).All methods were in compliance with the ARRIVE guidelines (Percie du Sert et al., 2020).

Jejunal mucosa and digesta sampling
After blood sampling on d 7, cows were transported to the institutional slaughterhouse, stunned by captive bolt and killed by exsanguination.Within 15 min after death, a 20-cm sample from the mid-jejunum was obtained, carefully rinsed with 0.9% NaCl solution and used to collect mucosal scrapings.The tissue was snapfrozen in liquid nitrogen and stored at −80°C until further analysis.From the same site of tissue sampling, a representative digesta sample was collected.The pH of the digesta was determined (CG 841, Schott, Mainz, Germany) and samples were centrifuged at 15,700 x g for 10 min at 4°C.The supernatant (digesta fluid) was aliquoted and stored at −20°C, whereas the digesta pellet was snap-frozen in liquid nitrogen and stored at −80°C.

Digesta fluid metabolite analysis
For the determination of acetate concentrations, thawed digesta fluid samples (200 µL) were mixed with 80 µL iso-capronic acid as internal standard and subjected to a gas chromatograph (GC-2010 Plus, Shimadzu Corp., Kyoto, Japan) equipped with a flame ionization detector and a 25 × 0.25-mm free fatty acid phase column (Roth, Karlsruhe, Germany), similar to the method described by Tummler et al. (2020).For L-and D-lactate analysis, dissolved proteins of digesta fluid were precipitated by 1.5 M HClO 4 , neutralized by 2 M K 2 CO 3 (60:40:20 µL), and centrifuged at 13,000 × g at 4°C for 20 min.Twenty µL of the supernatant were separated isocratically by HPLC (System 1200/1260 infinity, Agilent Technologies, Waldbronn, Germany) on a 150 × 4.6 mm Chirex 3126(D)-penicillamine column protected by a 30 × 4.6 mm guard column and a Krud-Katcher (all: Phenomenex, Aschaffenburg, Germany).Separation was performed using 1 mM CuSO 4 as eluent at a flow of 1 mL/min at 25°C, and analyte detection with refractive index (RI) and UV absorption at 232 nm (HPLC Application ID No. 14094, Phenomenex, modified).The retention times of L-and D-lactate were 21.9 (RI) and 26.6 (UV) min, respectively; the detection limits were 10 µM (RI) and 5 µM (UV), respectively, depending on the detector.The quantification was done by a 4-point-calibration with external standards with RI signals for L-lactate because of overlapping signals with UV detection and with UV signals for D-lactate because of higher sensitivity.Due to technical problems, data from only n = 9 HS, n = 8 PF, and n = 8 CON cows could be obtained.In the digesta fluid, glucose and NEFA concentrations were measured by clinical chemistry analyzer (ABX Pentra C400, HORIBA) as mentioned earlier in the plasma metabolite analysis.

Isolation of microbial DNA
The microbial DNA was isolated from the digesta pellet and the mucosa scrapings using the DNeasy PowerLyzer PowerSoil Kit (QIAGEN, Hilden, Germany) according to the manufacture's instructions, but with additional 10-min incubation steps at 70°C and 95°C, followed by bead beating using a Precellys 24 homogenizer (PEQLab Biotechnology GmbH, Darmstadt, Germany).The concentration and purity of the DNA was determined by a NanoDrop 2000 instrument (Thermo Fisher Scientific, Dreieich, Germany).

Microbial profiling using 16S rRNA sequencing
The obtained DNA was used to amplify PCRfragments of the V4 region of 16S rRNA gene targeting the 515'F (GTGBCAGCMGCCGCGGTAA) and 806R (GGACTACHVGGGTWTCTAAT) sequences (Hugerth et al., 2014).The corresponding primers were designed by combining 16S-specific sequences with adapter sequences of the Illumina flow cell and specific index sequences according to the approach described by (Kozich et al., 2013).The PCR was performed in duplicate using the GoTaq G2 Hot Start polymerase (Promega, Mannheim, Germany).The protocol was as follows: An initial denaturation at 95°C for 2 min, followed by 35 cycles at 95°C for 30 s, 50°C for 60 s, and 72°C for 90 s, and a final extension for 10 min at 72°C.The replicates were pooled, processed over the Sequal-Prep Normalization Plate (Thermo Fisher Scientific), and the resulting products were mixed in the same proportions.The DNA libraries were parallel sequenced for 2 × 250 bp paired-end reads using the HiSeq PE Rapid Cluster Kit v2 and HiSeq Rapid SBS Kit v2 on the rapid-run mode of the HiSeq 2500 system (Illumina, San Diego, CA).Sequence data were de-multiplexed and converted into FASTQ files using the bcl2fastq2 conversion software, v2.19 (Illumina).The sequence reads were processed following the mothur pipeline (version 1.47.0;(Schloss et al., 2009) including references from the Silva database (release v138.V4; https: / / www .arb-silva .de/).Operational taxonomic units were derived from sequences clustered by sequence identity of ≥97% and were subsequently annotated using the Silva database.

Statistical analyses
Data regarding animal characteristics, plasma and digesta metabolite concentrations were analyzed using SAS (Version 9.4, SAS Institute Inc., Cary, NC, USA).The data were analyzed using a linear model implemented in the MIXED procedure.The model included the fixed factors treatments (HS, CON, PF) and block; and DIM served as covariate to account for metabolic variation during lactation (Weber et al., 2016, Koch et al., 2023).Least squares means (LSmean) and their standard errors (SE) were computed and differences of the LSmeans were tested using the Tukey-Kramer procedure.Differences were declared as significant if P < 0.05 and tendencies 0.05 < P < 0.1.
Before analysis of the microbial data, subsampling was performed to the lowest individual number of reads (184,132).At OTU level, α diversity was analyzed by calculating Shannon index, Simpson index, and Abundance-based coverage estimator (ACE) using the phyloseq R package.The significance of indices between groups was analyzed with the Kruskal-Wallis test and pairwise comparisons were performed with the Wilcoxon exact rank sum test.Data of intestinal microbiota were analyzed to identify differentially abundant genera with DESeq2, including all genera with more than 50 counts in at least one third of the experimental population (Love et al., 2014).The statistical analysis employed the negative binomial Wald test including block as fixed effect and DIM as continuous covariate.Differences were considered significant at a Benjamini-Hochberg adjusted P < 0.05 (Supplementary Table 1,2, https: / / doi .org/ 10 .6084/m9 .figshare.24828417).Correlation analyses were performed based on variance-stabilized count data of microbial family abundance using the statistical software R (http: / / www .r-project .org).For each sampling site, the 10 most abundant microbial families were correlated with physiological parameters, Koch et al.: Heat stress and feeding… specific plasma metabolites related to post-absorption effects of the jejunum and jejunal metabolites of digesta fluid.To differentiate between the impact of heat stress and reduction in feed intake, correlation analyses were performed involving HS and PF cows (impact of heat stress only), CON and HS cows (combined impact of heat stress and reduced feed intake), and CON and PF cows (impact of reduced feed intake only).Spearmen correlation coefficients were calculated using the stats R package and visualized with the ggalluvial package (Brunson and Read, 2020).

Animal characteristics, plasma and digesta fluid concentrations
All cows had a similar BW on d 7 of the experimental phase (Table 1), but HS cows showed higher rectal temperature than CON and PF cows (P < 0.001).Thermographic measurements showed a significant increase of the surface temperatures of the side area (HS mean 36.0°C vs. TN 31.5°C,P < 0.001) on d 6 (Table 2).The difference between ambient temperature and mean temperature of the side area was significant lower in HS than TN animals (HS = 11.4K and TN = 18.0 K, P < 0.01) on d 6.Furthermore, the milk yield was higher in CON than HS and PF cows (each P < 0.05), while it tended to be lower in HS than PF cows (P = 0.06; Table 1).The water intake/BW ratio was higher in HS than CON cows (P < 0.05) and a trend for a lower water intake/BW ratio was found in PF compared with HS cows (P = 0.06).The DMI/ BW ratio was higher in CON than HS and PF cows (P < 0.05, respectively), but this was according to the experimental design demonstrating no differences between the HS and PF group.The plasma glucose concentration was lower in HS than CON cows (P < 0.05), whereas a tendency for higher glucose concentration was found in PF compared with HS cows (P = 0.06), however, plasma lactate concentration did not differ between groups.The plasma triglyceride concentration was lower in CON than HS and PF cows, whereas the NEFA concentration was higher in PF than CON and HS cows (P < 0.05, respectively).The plasma creatinine and urea concentrations were significantly higher in HS than PF cows (P < 0.05, respectively), while the creatinine concentration was higher in HS than CON cows (P < 0.001).The plasma BHB concentration did not differ among groups.In the digesta fluid, pH, and lactate, glucose, NEFA, and acetic acid concentrations were comparable among the treatment groups.

Microbial diversity in intestinal mucosa and digesta of dairy cows
The high-throughput analysis generated a total of 59,827,258 reads for the 30 digesta and 30 mucosa samples after filtering and processing of raw sequencing data.These sequences were aggregated at OTU level resulting in 52,790 OTU assigned to 1,280 genera.According to the sample with the lowest number of processed reads, the subsequent subsampling resulted in 184,132 reads per sample.Thus, sets of 14,646 OTUs representing the digesta and 13,876 OTUs representing the mucosa-associated microbial community was obtained.The non-metric multidimensional scaling (NMDS) of the entire data set based on Bray-Curtis similarity index revealed the clustering of digesta and mucosa-associated microbiota into 2 separate groups (Figure 2A, P = 0.001).The comparison of the digesta and mucosa forming OTUs between HS, CON and PF cows is displayed in Figure 2B and 2C.The 3 groups commonly shared 2844 digesta OTUs and 2858 mucosaassociated OTUs.The annotation of all OTUs at the genera level resulted in a total set of 930 taxa that were considered for further analysis.
The α-diversity was investigated by 3 diversity indices assessing species richness, species diversity, and evenness of species.The Shannon and Simpson indices reveal that the microbiota of the mucosa was more diverse than in the digesta (P < 0.001), although the species richness as indicated by ACE values was higher in the digesta than in the mucosa (Figure 3A-C).According to the Shannon and ACE indices, the mucosaassociated microbiome of PF cows had a significantly lower α diversity compared with the CON group (P < 0.05), whereas this effect was not present in the digesta microbiota.Furthermore, effects of heat stress on microbial α diversity were not observed (Figure 3A-C).The NMDS representation and PERMANOVA based on Bray-Curtis distances revealed no significant overall dissimilarities between groups within sampling site (Figure 3D,E).
Annotation of all OTU tags to the Silva reference database revealed that the mucosa and digesta samples contained 930 different genera assigned to 368 families from 36 phyla.At phylum level, Firmicutes, Euryarchaeota and Actinobacteriota were the most abundant taxa in the 2 sampling sites, followed by mucosal Bacteroidota and Proteobacteria and digestal Verrucomicrobiota (Figure 4).The taxa plots at family level also indicated a considerable overlap in the overall microbial profiles between mucosa and digesta samples, with highest abundance of microbes assigned to Lachnospiraceae, Methanobacteriaceae and Peptostreptococcaceae (Figure 5).

Changes in the composition of the mucosaassociated and digesta microbiome
The comparison of the microbiota composition between cow groups revealed lower abundance of species belonging to Succinivibrionaceae UCG-001 in digesta and mucosa of PF than HS and CON cows (each p adj < 0.01; Table 3).Colidextribacter was less abundant in the mucosa and digesta of HS compared with CON cows, and was also less abundant in the digesta samples of PF compared with CON cows (p adj < 0.05).Furthermore, digestal Lachnospiraceae UCG-008 tended to be less abundant in HS and PF than CON cows (p adj < 0.077).Microbes assigned to an unclassified genus of Bacillales were less abundant in the PF than CON group (p adj = 0.049), whereas unclassified Oscillospirales tended to be more abundant in PF than CON cows (p adj = 0.079).More pronounced differences between groups were found regarding the mucosa-associated microbiome.In detail, the microbes belonging to Gastranaerophilales (p adj = 0.022), Clostridia (p adj = 0.05) and Bifidobacteriaceae (p adj < 0.05) were more abundant in HS than PF cows.Tendencies for higher abundances of Methanobrevibacter, Denitrobacterium, and Lactobacillus were found in HS compared with PF cows (p adj < 0.1).Furthermore, Cellulosilyticum showed a significantly higher abundance in PF than CON and HS cows (p adj < 0.007).Trends for lower abundances of Anaerofustis, Turicibacter, Clostridium sensu stricto 1, Sharpea and unclassified Peptostreptococcaceae were found in HS relative to PF cows (p adj < 0.1).Higher abundances of 28.2 ± 1.2 a 24.7 ± 1.1 b,A Water intake/BW (kg/kg) 0.15 ± 0.01 a 0.13 ± 0.01 b,A 0.11 ± 0.01 b,B DMI/BW (kg/kg) 0.021 ± 0.001 b 0.031 ± 0.001 a 0.021 ± 0.001 b Plasma metabolites Glucose (mmol/L) 3.17 ± 0.10 b,B 3.52 ± 0.10 Turicibacter, Clostridium sensu stricto 1, Peptostreptococcaceae and Sharpea were found in PF compared with CON cows (p adj < 0.05), while unclassified microbes of Chloroflexi showed a trend for a higher abundance in PF than CON cows (p adj = 0.058).The microbes belonging to Clostridia UCG-014 were less abundant in PF than in CON (p adj < 0.01), and a trend for a lower abundance of unclassified Clostridia was also found in PF compared with CON cows (p adj = 0.073).

Correlations between gut microbiota, plasma metabolites, physiological parameters and digesta fluid metabolites
Because the gut microbiota composition may alter in response to the digesta milieu or be affected by metabolite concentrations of the circulation, correlations between the most abundant microbial families and the metabolite concentration of the respective body fluid were examined (Figure 6A-C, Supplementary Table 3, https: / / doi .org/ 10 .6084/m9 .figshare.24828417).Correlation analyses involving HS and PF cows were performed to assess heat stress-specific microbiota and metabolite changes.They revealed that the digesta fluid NEFA concentrations were negatively correlated with the abundance of Bifidobacteriaceae, Erysipelotrichaceae, Lachnospiracea, and positively correlated with Anaerovoracaceae.The abundance of Bifidobacteriaceae was further negatively correlated with plasma BHB and urea concentrations.The inclusion of CON and HS cows in the correlation studies yielded significant correlation coefficients between different microbial families and plasma concentrations of BHB, lactate, and urea, as well as the digesta fluid pH, acetate, D-lactate, L-

Animal characteristics, metabolite profile in plasma and jejunal fluid
High ambient temperatures significantly increased body surface and rectal temperatures after 7 d of heat stress.Furthermore, chronic heat exposure reduced feed intake and milk yield in dairy cows, while water consumption increased.The drop in milk yield was more pronounced in HS than PF counter partners.These findings are in line with earlier results demonstrating that ambient heat decreases milk production by 35-40% (West, 2003).In addition, heat stress suppresses appetite and reduces the production of metabolic heat, thereby altering the utilization of nutrients (Mani et al., 2012).The metabolic status of HS cows studied herein showed lower plasma glucose concentrations compared with PF and CON cows.This results implies a lower availability of ruminal propionate (Rhoads et al., 2009), although the capacity for whole-body glucose production is maintained under heat stress (Baumgard and Rhoads, 2012).Of note, PF cows showed increased plasma NEFA concentrations due to the consequence of reduced energy intake.However, HS cows did not respond with an increase in plasma NEFA concentration indicating that fat mobilization is inhibited at high ambient temperatures (Baumgard and Rhoads, 2012).as indicated by the higher plasma TG concentration after heat exposure.The higher plasma urea and creatinine concentrations revealed that prolonged heat stress induced protein catabolism.Recent studies support our results because they showed higher plasma 1,3-methylhistidine, creatinine, urea concentrations and milk urea concentration in HS early lactating (Lamp et al., 2015) and mid-lactating cows (Gao et al., 2017).Consequently, the metabolism of lactating HS compared with PF cows relies more on protein degradation and amino acid catabolism, presumably to serve as an energy resource and to enter into a glucose sparing mode (Baumgard and Rhoads, 2013).
In contrast to the differential metabolic profile in plasma, we also expected to find differences in the metabolic profile of the jejunal fluid.However, there were no differences in jejunal D-and L-lactate, glucose, NEFA, and acetate concentration, indicating that the plane of feed intake and heat exposure are not major factors determining the intestinal milieu.A previous study described a series of changes in fecal metabolites in heat-stressed dairy cows exposed to a cycling THI of 72-82 for 14 d (Ruiz-Gonzalez et al., 2022).These included histidine, p-hydroxyphenylacetic acid, vitamin B7, tridecylic acid, myristic acid, arginine, and acetyl ornithine.The metabolite panel indicated an adaptation process of the microbial metabolic pathways to heat stress when compared with pair-fed cows kept at a THI of 61 to 64 (Ruiz-Gonzalez et al., 2022).Notably, fecal volatile fatty acid concentrations were not affected by ambient heat (Ruiz-Gonzalez et al., 2022), which is in parallel to our findings in the jejunal fluid.Nevertheless, the digesta metabolite profile in different parts of the gut (small intestine vs. feces) should not be directly compared as the morphology, biological function, and microbiota colonization change from the proximal to the distal gut.More studies along the gastrointestinal tract are required to uncover the different effects of heat stress.

Microbial diversity
There is a huge spatial heterogeneity in microbial composition, diversity and species abundance along the gastrointestinal tract, including the digesta and mucosa of dairy cows (Mao et al., 2015).In our study, we found a greater α-diversity of the mucosa-associated compared with digesta microbiota, whereas the species richness was higher in the digesta than mucosa across cow groups.These findings are in line with earlier observations in non-challenged cows, in which the mucosal microbiota of the small and large intestine exhibited a higher Shannon index than the digesta microbiome (Mao et al., 2015).The mucosa-associated bacterial community contains oxygen-tolerant species utilizing the host's mucin as substrate, which luminal microbes cannot (Marteyn et al., 2011).It has been speculated that the diversity of the mucosa-associated microbiota is more sensitive to microenvironmental changes, e.g., oxygen (Marteyn et al., 2011) and nutrient concentrations (Albenberg and Wu, 2014) or calorie restriction (Kokten et al., 2021) than the luminal microbiota.Because chronic heat exposure may reduce oxygen and nutrient supply to the gastrointestinal tract and compromise the intestinal barrier function in pigs (Pearce et al., 2013a, Pearce et al., 2013b, Xia et al., 2022) and cows (Koch et al., 2019), it was tempting to speculate that heat stress differentially affects the composition of the mucosa-associated microbiome and the digesta microbiome.However, neither the intestinal digesta nor the mucosa microbiome α diversity was affected by heat stress in the present study.Our finding agrees to earlier studies reporting no change in the intestinal microbial α diversity in response to heat stress in Jersey cows and pigs (Kim et al., 2020, Xia et al., 2022).Relative to CON cows, microbial diversity changed in PF but not HS cows.Although feed intake was reduced to the same extent in both groups, absent of diversity change in HS cows might be explained by a higher apparent digestibility and altered absorption capability during heat stress maintaining microbial diversity (Al-Mamun et al., 2008).On the other hand, the reduction in feed intake in PF relative to CON cows led to a significant decrease in some of the microbial α diversity indices (Shannon, ACE richness).In agreement with a recent study (Patra and Kar, 2021), our finding demonstrates that the reduction in feed intake is the major driver of taxonomic perturbations of the intestinal microbial communities, and that additional factors compensate for the diversity loss during heat stress.

How heat stress shapes the microbial composition
Although no overall differences in microbial β diversity, represented by Bray-Curtis dissimilarity, were found between treatment groups, the detailed analysis of differentially abundant genera in HS cows revealed several differences compared with the CON and PF groups.The higher abundance of microbes assigned to Gastranaerophilales, Bifidobacteriaceae and Clostridia shaped the microbiota in the mucosa of HS compared with PF cows.In earlier studies, ruminal Gastranaerophilales and Clostridia were negatively associated with dry matter intake (Wang et al., 2022b), milk yield, and feed efficiency (Bach et al., 2019).Garcia et al. (2013) reported that some species belonging to the Clostridium family are conditional pathogens and induce intestinal diseases in sheep, goats and other ruminants (Garcia et al., 2013).In the present case, a higher abundance of Clostridium spp.may release higher amounts of toxins that may affect the gut barrier function and, or independently thereof penetrate the 'leaky' gut to induce inflammatory responses in HS cows (Koch et al., 2019, Koch et al., 2021, Koch et al., 2023).The inflammatory response of the intestinal mucosa is characterized by activated NF-κB signaling and antioxidative defense systems (Koch et al., 2019), suggesting that the immune system act to neutralize potentially invaded pathogens as a consequence of diminished gut barrier function (Koch et al., 2021).However, to what extent toxins released from Clostridium spp.penetrate the gut barrier and affect gut health in HS cows requires additional studies.
The abundance of Bifidobacteriaceae and Lactobacillus in fecal samples were shown to be positively correlated with ambient temperature (Nguyen et al., 2020), and this finding is comparable to our results obtained from intestinal samples.Microbes assigned to Bifidobacteriaceae correlated negatively with plasma BHB and plasma urea levels in our study.Interestingly, a link was found between plasma BHB levels and the abundance of Bifidobacterium and Lactobacillus in the intestine, although the effect of the ketogenic diet used is not clear (Ang et al., 2020).Moreover, there are several indications that these bacteria have an important function in maintaining the intestinal integrity (Wang et al., 2022a, Xia et al., 2022).This suggests that the increased abundance of Lactobacillus and Bifidobacteriaceae in the mucosa of HS cows could act to reduce the harmful consequences of heat stress on the gut barrier function.Furthermore, it has been shown that some NEFA exert a cytotoxic effect on microbes (Hazell and Graham, 1990, Bergsson et al., 2001, Qi et al., 2023).Medium-chain saturated, long-chain unsaturated fatty acids and their monoglycerides are able to kill Gram-positive cocci bacteria by destroying the cell wall (Bergsson et al., 2001).In our study, Bifidobacteriaceae correlated negatively with the NEFA concentration in the digesta fluid, suggesting that NEFA may negatively affect Bifidobacteriaceae abundance.However, further studies are required to discriminate between the different interactions of bacterial genera and different metabolites in the digesta fluid.
While high ambient temperatures reduce rumen and gut motility, namely the contraction amplitude or frequency (Calamari et al., 2018) as well as the ruminal pH (Kim et al., 2022), the jejunal pH as an indicator of changes in overall microbial metabolism was not affected by ambient temperatures in our study.Nevertheless, some correlations were found between the intestinal pH and the abundance of microbial families, including those previously described to be responsive to pH changes, such as Ruminococcaceae and Erysipelotrichaceae (Firrman et al., 2022).
Colidextribacter were less abundant in the digesta and mucosa of HS cows compared with the levels found in CON cows.Interestingly, in a previous study performed in mice, a positive correlation between the fecal Colidextribacter abundance and the expression of thermogenic genes in white adipose tissue was detected (Li et al., 2022).Gryanznova et al. (2021) reported that a higher abundance of Colidextribacter in the milk of dairy cows was linked to a higher risk of mastitis and thus inflammation in the mammary gland (Gryaznova et al., 2021).It seems that anaerobic Colidextribacter, a recently proposed new genus (Ricaboni et al., 2017), plays a general role eliciting inflammatory processes, but its specific function during heat stress is largely unclear.
At genus level, HS and PF cows showed a tendency for lower digesta Lachnospiraceae UCG-008 abundance compared with CON cows.This finding is in line with earlier observations demonstrating that the fecal Lachnospiraceae abundance was lower in summer than in spring (Li et al., 2020).Lower Lachnospiraceae abundance were also found to be related to increased gut inflammation in inflammatory bowel disease (Lobionda et al., 2019).Thus, the lower Lachnospiraceae UCG-008 abundance might reflect an increased inflammatory state in the small intestine of HS cows, and this conclusion agrees with earlier findings showing infiltrated macrophage-like cells in the jejunal submucosa of HS cows reflecting a locally activated inflammatory state (Koch et al., 2019).

Microbial metabolism and nutrient provision to the host
An earlier proteomic study reported downregulation of enzymes involved in jejunal glycolysis in HS compared with PF cows (Koch et al., 2021).However, the latter study did not clarify if this difference was due to divergent plasma or digesta glucose concentrations.In the present study, we found no differences in intestinal glucose or other metabolite concentrations between cow groups, suggesting that the reduction in jejunal glycolysis during heat stress is rather regulated by the glucose concentration in the circulation.On the other hand one might expect that genera differentially abundant between HS and PF cows could affect the digesta glucose, lactate and further metabolite concentrations.The lack of the difference between digesta metabolite concentrations could be explained by crossfeeding between microbial species.For example, the lactate producers Bifidobacteriaceae and Lactobacillus can provide the substrate for Firmicutes and Clostridia (Frolova et al., 2022), thereby maintaining lactate homeostasis.Furthermore, the tending lower abundance of Anaerofustis, which is involved in the fermentation of complex carbohydrates yielding glucose (Lawson, 2015), and the greater abundance of the carbohydrate utilizing Bifidobacteria (Milani et al., 2016) may act to maintain digesta glucose concentration in HS cows.It should be noted that fermentation of complex carbohydrates occurs predominantly in the rumen, however, the abundance of Anaerofustis but also of the carbohydrate fermenting Cellulosilyticum (Kwon et al., 2021) suggests that some minor fermentation occurs also in the small intestine.The abundance of Cellulosilyticum in the mucosa was higher in PF compared with CON and HS cows, suggesting an intensified fermentation of cellulose after reduction in feed intake, but not in combination with heat stress.
Previous studies illustrated a link between the presence of Succinivibrionaceae, which facilitates succinate accumulation in the gut, and the induction of inflammation in irritable bowel diseases (Fernandez-Veledo and Vendrell, 2019).Moreover, succinate possess the ability to act as a messenger molecule to signal to distant organs enlarging the centers of inflammation in the body (Fernandez-Veledo and Vendrell, 2019).In our study, we found higher Succinivibrionaceae UCG-001 abundance in the digesta and mucosa of HS than PF cows, suggesting higher succinate concentrations and thus potential deleterious effects to the small intestine and further distantly located organs.In an earlier study, we found increased pro-inflammatory cytokine expression and tending higher Toll-like receptor 2 (TLR2) protein expression in mesenteric lymph nodes of HS compared with PF and CON cows (Koch et al., 2023).However, if increased intestinal succinate concentrations account for the immune response in mesenteric lymph nodes of HS cows remains to be investigated in future studies.
The abundance of Succinivibrionaceae UCG-001 is not only affected by heat stress as shown herein, but may also decline in the rumen during feed restriction (McCabe et al., 2015).In the present study, the abundance of Succinivibrionaceae UCG-001 did not differ between CON and PF cows suggesting that ruminal and intestinal Succinivibrionaceae are differentially affected by the reduction of feed intake.

Heat stress effects on intestinal methanogenic archaea
It is well known that methane is produced by methanogenic archaea in the rumen but to a minor extent also in the small and large intestine.The intestinal Methanobrevibacter from the order Methanobacteriales showed a trend for a higher abundance in the mucosa of HS compared with PF cows.Consequently, the lower DMI under hot ambient temperatures seemed to play only a minor role for the abundance of this taxa.Notably, the relative abundance of Methanobacteriaceae in the jejunum mucosa and digesta was found to be consistently high in all cows of the current study.Methanobacteriaceae are the dominant archaea in the rumen of cows and are strongly influenced by diet (Furman et al., 2020).However, the relative abundance of archaea in the small intestine has been described only recently (Lin et al., 2023), making it difficult to draw specific conclusions about the importance of archaea in this gut section.Yet, the Methanobacteriaceae abundance correlated positively with plasma BHB and digesta glucose concentrations, whereas digesta acetate concentrations correlated positively with the abundance of mucosal and digesta Erysipelotrichaceae and mucosal Peptostreptococcaceae.As stated above, the role of altered intestinal methanogenic abundances during heat stress is still largely unclear, although some studies reported earlier an increase in the amount of Methanobrevibacteria in the rumen of heat-stressed cattle (Yadav et al., 2016, Correia Sales et al., 2021).

CONCLUSIONS
The reduction in feed intake is a major driver for shifts in the composition of the mucosa-associated microbiota.On the genera level, the mucosa-associated microbiota appears to be more sensitive to heat stress than the digesta microbiota.Although, no losses in microbial α diversity were observed during heat stress, changes in microbial composition provided some implications for the underlying adaptive responses.Specifically, increased abundance of Bifidobacteriaceae and the trend toward a higher abundance of Lactobacillus, especially in the mucosa of HS cows, suggests measures strengthening the gut barrier.On the other hand, higher abundance of species belonging to unclassified Clostridia and Succinivibrionaceae UCG-001 potentially provide a source of pathogens and messenger molecules, respectively, that facilitate the activation of immune responses in HS cows.Similarly, Colidextribacter and Lachnospiraceae UCG-008 seem to have a role in controlling acclimation to heat stress, and as such the abundance of certain species as well as the concentration of their released toxins should be analyzed in near future.Correlations between microbial changes and post-absorptive metabolites were also found, but whether the shift in the microbial profiles affects plasma metabolite concentrations or if an altered metabolite concentration in the circulation affects the intestinal microbiota under HS conditions, should be subject in future studies.
Koch et al.:  Heat stress and feeding… Table1.Animal characteristics, plasma metabolites, and jejunal metabolites of digesta fluid after 7 d of heat-stressed (HS) cows and control (CON) and pair-fed (PF) cows at thermoneutral conditions.n = 10 cows per group, except for D-lactate: HS n = 9; PF n = 8, and CON n = 8.Data are presented as LSmean ±

Figure 3 .
Figure 3. Alpha diversity assessed by (A) Shannon, (B) Simpson diversity index, and (C) ACE richness.Beta diversity of the (D) digesta and (E) mucosa-associated microbiome as assessed by PERMANOVA based on the Bray-Curtis distance.Anosim and multi-response permutation procedure tests show no significantly difference among groups (red -HS cows, white -CON cows, blue -PF cows).* P < 0.05

Figure 4 .
Figure 4. Histogram of species relative abundance at the phylum level of heat-stressed (HS) cows, and control (CON) and pair-fed (PF) cows at thermoneutral conditions.

Figure 6 .
Figure 6.Correlation of physiological parameters, plasma metabolites and jejunal metabolites of digesta fluid with microbial family abundance in mucosa (clear boxes) and digesta (dotted boxes) samples.Correlation analyses were performed across (A) CON and HS cows, (B) HS and PF cows, and (C) CON and PF cows.Connecting lines between microbial taxa and traits indicate significant correlations (P < 0.05).Plus and minus symbols indicate positive and negative correlation coefficients, respectively.The colors of boxes and connecting lines represent the different microbial families.
Koch et al.: Heat stress and feeding…