Effect of short-term abomasal corn starch infusions on postruminal fermentation and blood measures

It is possible that some of the systemic responses to subacute ruminal acidosis (SARA) may be caused by increased intestinal starch fermentation. The objective of this experiment was to evaluate the effect of aboma-sal infusion of up to 3 g of corn starch/kg body weight (approximately 1.6 kg of starch/d) on fecal measures of fermentation, plasma acute phase proteins, and white blood cell populations. Six ruminally cannulated cows in late lactation were randomly assigned to duplicate 3 × 3 Latin squares with 21-d periods. Cows were fed a 20.6% starch TMR twice daily and during the last 7 d of each period cows were abomasally infused with corn starch at 0 (CON), 1 (ST1), or 3 (ST3) g/kg body weight split into 2 bolus infusions, provided every 12 h. Fecal samples were collected at 0, 6, 12, and 18 h following feeding on d 21 and were analyzed for pH, VFA, lactic acid, and lipopolysaccharide (LPS). Composite fecal samples were used to estimate apparent total-tract nutrient digestibility using undigested neutral detergent fiber as an internal marker. Blood samples were collected at 0 and 6 h relative to feeding on d 14, 18, and 21 of each period. Concentrations of haptoglobin and serum amyloid A in plasma were measured in all samples, 0 h samples on d 14 and 21 were used to measure white blood cell populations, and 0 h samples from d 14, 18, and 21 were used for flow cytometric analysis of γδ T cells. Data were analyzed in SAS using models that included fixed effects of treatment and period and the random effects of cow and square. For blood measures, d 14 samples collected before the initiation of abomasal infusions were included as covariates. Time (d or h) was added as a repeated measure in variables that included multiple samples during the abomasal infusion period. A contrast was used to determine the linear effect of increasing abomasal corn starch. Abomasal corn starch linearly decreased fecal pH and linearly increased fecal total VFA and LPS, but effects were modest, with fecal pH, total VFA, and LPS changing from 6.96, 57.7 m M , and 4.14 log 10 endotoxin units (EU) per gram for the CON treatment to 6.69, 64.1 m M , and 4.58 log 10 EU/g for the ST3 treatment, respectively. This suggests that we did not induce hindgut acidosis. There were no effects of treatment on apparent total-tract starch digestibility or fecal starch content (mean of 96.9% and 2.2%, respectively). Treatment did not affect serum acute phase proteins or most circulating white blood cells, but the proportion of circulating γδ T cells tended to linearly decrease from 6.69% for CON to 4.61% for ST3. Contrary to our hypothesis, increased hindgut starch fermentation did not induce an inflammatory response in this study.


INTRODUCTION
Lactating cows, particularly those in early and peak lactation, are often fed rations that are relatively high in starch to meet the energy requirements for lactation.High-starch diets not only increase the risk of subacute ruminal acidosis (SARA), but they also tend to increase flow of starch to both the small and large intestines (Sanz-Fernandez et al., 2020).Increased large intestinal starch flow and fermentation lead to similar VFA increases and pH decreases in cecal digesta as observed in the rumen (Li et al., 2012).Increased intestinal starch fermentation also results in changes in fecal appearance (i.e., color and degree of presence of gas bubbles and mucus) and consistency (i.e., looseness or firmness) that can aid in SARA diagnosis (Kleen et al., 2003;Plaizier et al., 2008).
Subacute ruminal acidosis results in a systemic inflammatory response as evidenced by increased blood concentrations of acute phase proteins (Plaizier et al., 2008;Zebeli et al., 2012).This inflammatory response is believed to be the result of SARA-induced breaches in the digestive epithelium, allowing for increased translocation of LPS and other pro-inflammatory microbial compounds (Plaizier et al., 2008;Khafipour et al., 2009;Garcia et al., 2017).Subacute ruminal acidosis damages rumen epithelium as evidenced by decreased epithelial thickness, sloughing and lesions of epithelial layers, and disruption of epithelial barrier function (Penner et al., 2011;Steele et al., 2011).Research suggests that rumen papillae damage and subsequent entry of microbial compounds is the primary driver of the inflammatory response to SARA.However, feeding of high-starch diets to sheep and goats also damages the intestinal epithelium as evidenced by cellular sloughing, decreased tight junction integrity, and mitochondrial damage (Tao et al., 2014;Ye et al., 2016;Wang et al., 2017).It has been proposed that the intestines may be the primary or an additional site of endotoxin entry during SARA, contributing to the inflammatory response (Plaizier et al., 2008;Khafipour et al., 2009;Li et al., 2012).When intestinal epithelial assaults occur, the innate immune system is activated, inducing immune cells to aggregate in the offending location (Little et al., 1996).Among these innate immune cells are the γδ T cells, which are often considered the first line of defense against intestinal epithelial insults such as endotoxin, producing both pro-and anti-inflammatory signals which modulate the inflammatory response (Guerra-Maupome et al., 2019).
Because high-starch diets that induce SARA might affect both the rumen and intestines, experimentally inducing SARA does not allow for differentiation of ruminal and intestinal drivers of downstream inflammatory responses.Determining whether and to what extent these intestinal effects contribute to the inflammatory response is important for future development of mitigation strategies.Providing starch postruminally offers the opportunity to increase intestinal starch fermentation without altering rumen fermentation.In our previous work, we found that 1 g of starch/kg BW (approximately 0.8 kg/d) of starch or oligofructose delivered postruminally 1 or 4 times per day increased colonic fermentation as evidenced by decreased pH and increased VFA and LPS (Mainardi et al., 2011;Gressley et al., 2016).However, there was no effect on plasma acute phase proteins, suggesting the challenge did not affect intestinal epithelial integrity (Mainardi et al., 2011).Bissell and Hall (2010) found that 4 kg/d of abomasal starch continuously infused over 12 h/d resulted in fecal indicators of epithelial damage and extreme sickness response in 2 of the 6 cows infused.If intestinal epithelial disruption contributes to the systemic inflammatory response to SARA, this collective work suggests the intestinal starch flow required to elicit this effect likely occurs at a level greater than 0.8 kg/d when no response was observed but lower than 4 kg/d when overt sickness occurred.However, differences among studies in factors including animals, dietary starch levels, and feed intake likely affect this threshold as well.The objective of this experiment was to evaluate the effect of abomasal infusion of up to 3 g of corn starch/kg BW (approximately 1.6 kg starch/d delivered in 2 pulse doses per day) on fecal measures of fermentation, plasma acute phase proteins, and γδ T cell populations.We hypothesized that abomasal corn starch would induce hindgut acidosis and an inflammatory state as would be evidenced by linear decreases in fecal pH and circulating γδ T cell populations and linear increases in fecal VFA, LPS, and serum acute phase proteins.

Animals and Treatments
All work was approved by the University of Delaware's Institutional Animal Care and Use Committee, protocol number 80R, and was completed in summer of 2017.Six ruminally cannulated Holstein cows (4 in lactation 2, 1 in lactation 3, and 1 in lactation 5) in late lactation were used in this study.Cows were selected for this study based on presence of pre-existing rumen cannulas and stage of lactation.The number of cows used was expected to provide sufficient power to observe a 1.0-unit difference in fecal pH.At the start of the study, DIM (mean ± SD) was 219 ± 12, milk yield was 32.6 ± 5.4 kg/d, and BW was 735 ± 49 kg.Cows were housed at the University of Delaware in tiestalls that contained rubber mattresses bedded with wood shavings twice daily.The cows were fed a TMR that contained 20.6% starch (Table 1).The diet was balanced to meet nutrient requirements and consisted of 45.5% corn silage, 13.4% alfalfa silage, and 41.1% grain mix on a DM basis.Cows were fed twice daily (0830 and 1600 h), with 50% of daily feed offered at each time.Cows were fed ad libitum for approximately 5% refusals.Feed offered and refused was measured daily.
Before the start of the experiment, abomasal infusion lines were inserted through the rumen cannula as described by Gressley et al. (2006).The infusion lines were left in the abomasum during the duration of the experiment with placement within the abomasum verified weekly.The study was conducted as a replicated 3 × 3 Latin square with 21-d periods.Cows were randomly assigned using a random number generator first to a square then to a treatment sequence within a square.The treatment sequences across the 2 squares were balanced for first-order carryover effects.The first 14 d of each period were for washout, and during this time all cows were fed the experimental ration and were not abomasally infused.During the last 7 d of each period (d 15-21), cows were abomasally infused with 8 L/d of tap water containing 0 (CON), 1 (ST1), or 3 (ST3) g of food grade corn starch product (Feed Binder GE modified corn starch; Ingredion, Westchester, IL) per kilogram BW.Abomasal infusions were administered as twice daily pulse doses at 12-h intervals (0800 and 2000 h).At each infusion time, half of the daily dose of corn starch was suspended in 4 L of tap water and infused into the abomasal infusion line using a veterinary stomach pump (Nasco, Fort Atkinson, WI).Control cows were infused with only 4 L of tap water at each infusion time.Infusions took approximately 2 min per cow, and 100% of the infusate was always delivered.Researchers were not blinded to treatments due to the obvious physical differences among the treatments, but collected samples were assigned sequential numbers before the start of the trial to prevent bias during sample analysis.

Milk and Feed Sampling and Analysis
Cows were milked twice daily at approximately 0430 and 1530 h and milk yield at each milking was recorded throughout the study.Milk samples were collected from both morning and afternoon milkings on d 13, 14, 20, and 21 of each period.Samples were submitted to Dairy One Cooperative Inc. (Ithaca, NY) for individual analysis of lactose, true protein, fat, MUN, and SCC using a MilkoScan FT+ (Foss, Hillerød, Denmark).
Samples of grain mix were collected once weekly, and samples of corn silage, alfalfa silage, and TMR were collected twice each week.A portion of each sample was used for measurement of DM, and subsamples of grain mix, corn silage, and alfalfa silage were frozen until composited by period.Period composites of forages and grain mix were mailed to Cumberland Valley Analytical Services (Waynesboro, PA) for wet chemistry analysis of DM (105°C for 3 h for forages; method 930.15,AOAC International, 2000, for grain), NDF (van Soest et al., 1991), ADF (method 973.18, AOAC International, 2000), N (method 990.03,AOAC International, 2000), starch (Hall, 2009), and ash (method 942.05, AOAC International, 2000).Total mixed ration samples were also collected on d 21 of each period and were analyzed for undigested NDF (uNDF) following 240 h of in vitro digestion (Goering and van Soest, 1970).A composite sample of the corn starch infusate was analyzed at the end of the study and contained 90.5% DM, 80.6% starch, 0.5% CP, 1.2% NDF, 0.0% uNDF, 0.1% ADF, and 1.0% ash.Infusions were delivered on an as fed basis (not corrected for DM), and amount of chemical starch provided by the infusions equated to 0, 0.73, and 2.19 g/kg of BW for the CON, ST1, and ST3 treatments, respectively.

Rumen and Fecal Sampling and Analysis
Rumen pH was continuously measured during d 12 to 14 and d 19 to 21 of each period using weighted, indwelling pH loggers (DASCOR, Escondido, CA) that resided in the ventral rumen.The pH loggers were placed in the rumen at the end of d 11 and removed at the beginning of d 15, then returned to the rumen at the end of d 18 and removed at the start of the next period (after periods 1 or 2) and at the end of the experiment (after period 3).Rumen pH was recorded every 5 min, and data collected from each d were used to determine daily mean, minimum, and maximum pH.
Samples of rumen fluid and feces were collected every 6 h on d 21 of each period, with the first sample occurring immediately before feeding (0 h). Rumen fluid was collected from 4 areas in the ventral rumen sac, mixed, and strained through 2 layers of cheesecloth.A 10-mL sample of rumen fluid was acidified with 0.2 mL of 50% H 2 SO 4 and stored at −20°C until VFA analysis.Fecal samples (~200 g/cow per time point) were collected by rectal palpation.For fecal pH, 20 ± 2 g of feces were added to 20 mL distilled water, shaken for 20 s, strained through 2 layers of cheesecloth, and fecal pH was recorded (Anaheim Scientific P771 pH meter, Yorba Linda, CA).For fecal VFA, 25 ± 2 g of feces were acidified with 10 mL of a 2% H 2 SO 4 solution and mixed.The resulting solution was strained through 2 layers of cheesecloth and frozen at −20°C for later analysis.For LPS measurement, approximately 2 g of feces were placed in LPS-free glass vials and frozen at −20°C until analysis.Remaining feces was stored at −20°C until it was composited by cow and period (100 g of wet feces per cow per sampling time) for analysis of nutrient content and uNDF as described above.
Rumen and fecal content of lactic acid and VFA were evaluated by HPLC as described by Mainardi et al. (2011) following the procedures of Muck and Dickerson (1988).The endotoxin concentration of fecal samples was determined using a commercial chromogenic endpoint amebocyte lysate kit according to manufacturer instructions (Pierce LAL Chromogenic Endotoxin Quantification Kit, Thermo Fisher Scientific, Rockford, IL) as described by Neiderfer et al. (2020).Briefly, fecal samples were thawed and centrifuged, 50 µL of fecal supernatant was pipetted into 450 µL of LPSfree water (Lonza, Basel, Switzerland), the sample was passed through a 0.2-µm filter, further diluted 10-fold, heat-treated at 100°C for 30 min, and stored at −20°C until the assay was conducted.Immediately before conducting the assay, samples were thawed and further diluted with LPS-free water and incubated at a 1:1 ratio with β-glucan blocker to increase assay specificity to endotoxin (Lonza, Basel, Switzerland).The intra-assay coefficient of variation was 5.0% and the interassay coefficient of variation was 16.7%.

Blood Sampling and Analysis
Blood samples were collected at 0 and 6 h relative to feeding on d 14, 18, and 21 of each period into 10-mL evacuated blood tubes (BD Vacutainer; Becton, Dickinson and Co., Franklin Lakes, NJ).At the 0 h time points, blood was collected from the jugular vein into a single blood tube without anticoagulant and multiple blood tubes with K 2 EDTA anticoagulant.At the 6-h time points, a single tube of blood without anticoagulant was collected from a coccygeal vessel.Following each sampling time, the tubes without anticoagulant were centrifuged at 1,000 × g for 30 min at 25°C and serum was collected.Serum was stored at −80°C until analyzed in duplicate for haptoglobin and serum amyloid A using commercial kits (Tridelta Development Ltd., Ireland).Intra-and interassay coefficients of variation were 3.2% and 3.4% for haptoglobin and 5.5% and 12.4% for serum amyloid A, respectively.At 0 h on d 14 and 21, one K 2 EDTA tube for each cow was placed on ice and transported to New Bolton Center (Kennett Square, PA) within 1 h of collection and analyzed for complete blood count using the ADVIA 120 Hematology System (Siemens Healthcare, Erlangen, Germany).
At 0 h on d 14, 18, and 21, blood collected into K 2 EDTA tubes was additionally used for flow cytometric analysis of γδ T cell populations.Blood was centrifuged, buffy coats were collected, and erythrocytes were removed with hypotonic lysis.Lymphocytes were adjusted to 1 × 10 7 cells/mL in PBS, and triplicate 100-µL aliquots were incubated with 100 µL of 1.87 µg/mL mouse anti-bovine primary monoclonal antibody N24 (γδ T cell receptor, clone number GB21A0, IgG2b, Monoclonal Antibody Center, Pullman, WA) in the dark at 4°C for 30 min.Following incubation, plates were washed, treated with 50 µL of secondary antibody (goat anti-mouse IgG2b conjugated with Rphycoerythrin, Jackson ImmunoResearch, West Grove, PA) diluted 1:100 in PBS containing acid citrate dextrose and 2% goat serum, and incubated in the dark at 4°C for 30 min.Following incubation, plates were washed and cells were fixed in PBS with 1% buffered formaldehyde.Cells were stored at 4°C until the end of each period when analysis of marker expression was performed by flow cytometry (FACS Calibur, Becton Dickinson, Franklin Lakes, NJ).Populations were gated for lymphocytes based on forward and side scatter, and 10,000 events within each gate were counted for each sample.To control for background fluorescence, results were corrected for cells stained with irrelevant, isotype matched control antibodies.

Statistical Analyses
All analyses were conducted using the GLIMMIX procedure of SAS (version 9.1; SAS Institute, Cary, NC).Mean intake and milk yield were determined for the last 3 d of each period.Weighted average milk composition was calculated for the last 2 d of each period.For the washout periods, mean intake and milk yield were calculated for the last 7 d, and weighted milk composition was calculated for the last 2 d to be included in the model as covariates.A preliminary analysis revealed that serum amyloid A and haptoglobin were not affected by hour of sampling (0 or 6 h after morning feeding), so mean values for the 0 and 6 h time points on each day were calculated before formal statistical analysis.
All variables were evaluated using a model that included the fixed effects of treatment and period and the random effects of cow and square.For variables that included pretreatment data collected at the end of the washout period (milk yield and composition, intake, rumen pH, acute phase proteins, and white blood cells), washout period data were included as covariates.Repeated measures models were used for those variables where samples were collected at multiple time points on d 21 of each period (rumen and fecal VFA and lactate, fecal pH and LPS) or on multiple days of the treatment period (rumen pH, acute phase proteins, γδ T cells).For those repeated measures, day or sampling time as well as its interaction with treatment were included in the model, the RANDOM _RESIDUAL_ statement was used to specify the repeated measure, the subject was period × cow, and a first-order autoregressive covariance structure was used.
For all models, a CONTRAST statement was used to determine the linear effect of increasing abomasal infusions from 0 to 1 to 3 g/kg of BW.Because these treatments were unequally spaced, the IML procedure was used to determine contrast coefficients, which were −0.6172, −0.1543, and 0.7715 for the CON, ST1, and ST3 treatments, respectively.When an interaction of treatment by time was significant, the linear contrast was used to assess the effect of treatment within each time point.Significance was declared at P ≤ 0.05 and trends at 0.05 < P ≤ 0.10.Because periods were relatively short, we also tested for carryover effects in the models described above for our primary variables of interest (fecal pH, fecal total VFA, fecal LPS, plasma acute phase proteins, and γδ T cells).To test for these effects, we added 2 coefficients to each model that coded for treatment in the previous period.No carryover effects were detected for any of the variables tested (P > 0.10 for effects of coefficients used to test for carryover effects).

RESULTS AND DISCUSSION
All animals completed the study and there were no adverse events.This study abomasally infused corn starch product at 0, 1, or 3 g/kg BW, split into 2 bolus infusions, provided every 12 h, corresponding to on average 0, 0.7, or 2.2 kg/d product and 0, 0.5, and 1.6 kg/d starch.In their meta-analyses, Owens et al. (1986) and Offner and Sauvant (2004) observed a mean ruminal starch disappearance of 72% of starch intake, although results were highly variable across studies.More recent work by Sanz-Fernandez et al. (2020) demonstrated that greater starch intakes were associated with decreased ruminal starch degradation, with an average of 72% ruminal starch degradation with less than 10 g/d starch per kilogram of BW body (~28% dietary starch) but 55% for diets with greater than 10 g/d starch per kilogram of BW body.Thus, high-starch diets can increase both the total flow and proportional flow of starch to the duodenum.In this study, all treatments provided less than 10 g/d starch per kilogram of BW body from the combination of the diet and infusion that were 7.1, 7.6, and 9.0 g/d per kilogram of BW for the CON, ST1, and ST3 treatments, respectively.Because enzymatic digestion of starch in the small intestine is limited, accounting for a mean of 53% (Owens et al., 1986) or 66% (Offner and Sauvant, 2004) of duodenal starch flow, high-starch diets also increase starch flow to the large intestine.The diet for the present study was formulated to contain 16.7% CP, 30.5% NDF, 19.9% ADF, and 25.0% starch.Assuming TMR intake of approximately 25 kg/d and consequent starch intake of 6.3 kg/d, ruminal starch digestion of 55% to 72% would result in duodenal starch flows of 1.8 to 2.8 kg/d.On average, the ST1 and ST3 treatments provided 0.5 and 1.6 kg/d abomasally infused starch.Based on the above predictions of duodenal starch flow from the diet, the infusions should have increased total duodenal starch flow by 19% to 29% for the ST1 treatment and 57% to 89% for the ST3 treatment.However, analyzed dietary nutrient content (Table 1) differed from formulated, most notably for starch (20.6%).The difference in starch was due to a lower analyzed starch content of the grain mix compared with formulated (21.4% vs. 24.1%)as well as a lower analyzed starch content in the corn silage (25.5%) compared with a preexperiment sample (32.6%) that was used to formulate the ration.The mean DMI from the diet across treatments was 24.7 kg/d.With the dietary starch content of 20.6%, this resulted in a mean actual starch intake of 5.1 kg/d.Again, assuming ruminal starch digestion of 55% to 72%, this would have been expected to result in actual duodenal starch flows from the diet of 1.4 to 2.3 kg/d, approximately 20% lower than targeted.
Experimental periods were 21 d, with 14 d of washout and 7 d of abomasal infusions.The 7-d infusion periods were intentionally short due to our concerns at the time that the greater infusion level might cause adverse health effects.In additional, we expected most changes in our collected variables to be stabilized by the end of the 7-d infusion period.Our previous work found that a single pulse dose of oligofructose or corn starch rapidly induced fecal changes that returned to baseline within 24 h (Gressley et al., 2016).Danscher et al. (2015) similarly found that fecal pH stabilized within 1 d of fully transitioning cows to a SARA-inducing ration, and their companion publication noted changes in fecal microbial communities despite sampling only 7 d after complete transition to the SARA-inducing ration (Plaizier et al., 2017).Similarly, Li et al. (2012) observed changes in cecal samples and feces 7 d after feeding a SARA-inducing diet.A study in steers sampled every 2 d following initiation of a continuous 1.35 kg/d duodenal corn starch infusion found that small intestinal starch digestion had stabilized by d 6 (Brake et al., 2014).Following completion of our work, van Gastelen et al. (2021a) published that 5 d of continuous abomasal starch infusions was sufficient to induce changes in measures including apparent total-tract digestibility, with no carryover effects observed between periods despite only 2-d washout periods.This is similar to our findings of no carryover effects.However, studies designed to evaluate temporal responses up to 28 d following initiation of a SARA challenge have noted continued responses beyond 7 d (Wang et al., 2017;Neubauer et al., 2020).Thus, we expect that most of our sampled variables should have stabilized by the end of the infusion period, but it is possible responses may have been different if the challenge had continued for a longer period.Another limitation of this study was that cows were in late lactation.This challenge model may have been more effective in early lactation cows, as early lactation cows would likely have had greater dietary starch flow to the intestines from greater passage rate.
By design, DMI from the abomasal infusate increased linearly in response to treatment (Table 2; P < 0.01), but DMI from the diet or from the diet plus infusate did not differ among treatments (Table 2; P > 0.10).Treatment also did not affect milk yield or milk composition (Table 2; P > 0.10).Studies using early lactation cows have found that up to 3 kg/d continuous abomasal ground corn or corn starch increased yields of milk and milk protein (Reynolds et al., 2001;van Gastelen et al., 2021b), likely as a result of increased energy supply.However, the lack of a treatment effect on milk yield in the present study is not unexpected, as cows were in late lactation and total DMI from the diet and infusion did not differ among treatments.Increased intestinal fermentation should increase the conversion of blood urea into fecal microbial protein and thus decrease MUN (Gressley and Armentano, 2007;van Gastelen et al., 2021a).However, that was not observed in the present study and in fact was numerically greatest in the ST3 treatment.The reason for this lack of effect is unknown but may be related to the modest effects of abomasal starch on fecal indicators of hindgut fermentation as described below, the short treatment periods, or the relatively small contribution of the hindgut to total urea recycling.Rumen pH values were relatively high, with treatment means ranging from 6.26 to 6.39 and minima ranging from 5.85 to 6.02, and rumen VFA were relatively low, with total VFA ranging from 83.2 to 88.4 mM (Table 3).However, these data match with the expected relationship between rumen pH and VFA presented by Dijkstra et al. (2012).The high pH and low VFA were likely due to dietary starch content being only 20.6%.There were no treatment effects on rumen VFA or lactate concentrations.There was a time effect on most rumen VFA, and for acetate, propionate, butyrate, and total VFA, this was due to lower concentration at 0 h compared with one or more of the other 3 sampling times (data not shown).There was a tendency for abomasal starch to linearly increase rumen pH (P = 0.08), though no overall treatment effect was observed (P = 0.18).The tendency for the linear effect was unexpected and may be related to the numeric decrease in voluntary DMI with increasing abomasal starch.Our primary interest in measuring rumen pH and VFA was to indirectly confirm that corn starch was delivered postruminally, as decreased rumen pH and increased VFA would suggest backflow into the reticulorumen.Because neither of these occurred, our data suggest there was not significant backflow.
Abomasal starch linearly decreased fecal pH (P < 0.01), and there was also a treatment by time interaction on fecal pH (P = 0.04; Table 4; Figure 1).The interaction was further evaluated by testing for the linear effect of treatment at each time point.Abomasal starch linearly decreased fecal pH at 6 h (P < 0.01), but the linear effect was not observed at 0, 12, or 18 h (P = 0.96, P = 0.30, and P = 0.11, respectively).The magnitude of the decrease in fecal pH in response to abomasal starch infusion (0.27 pH unit difference between CON and ST3) was smaller than expected.Continuous abomasal infusions of 1.2 kg/d of wheat starch decreased fecal pH from 6.64 to 6.26 (Reynolds et al., 2001), continuous infusions of 3 kg/d of ground corn decreased fecal pH from 6.86 to 6.00 (van Gastelen et al., 2021b), and increasing from 0 to 1.5 to 3 kg/d abomasal corn starch decreased fecal pH from 6.49 to 6.00 to 5. 15, respectively (van Gastelen et al., 2021a).Piantoni et al. (2023) found that only 0.5 kg/d of single pulse-dose abomasally infused enzymatically resistant tapioca starch was sufficient to reduce fecal pH from 6.91 to 6.26 at 8.5 h following the infusion.Our ST3 treatment delivered approximately 2.2 kg/d product that contained 1.6 kg/d starch.Based on the relationship between fecal pH and abomasal corn starch presented by van Gastelen et al. (2021a) (fecal pH = 6.74 − 0.59 × kg of corn starch abomasally infused), our model should have been sufficient to induce hindgut acidosis and reduce fecal pH below 6.0, but that was not observed.However, we used a twice daily pulsedose model, whereas van Gastelen et al. (2021a) used a continuous infusion, so some of the difference may have been due to differences in delivery method.Regardless, the relatively small change in fecal pH coupled with the lack of treatment effect on apparent total-tract starch digestibility (Table 5) suggest relatively more of the abomasally infused corn starch may have been enzymatically digested in the small intestine compared with similar experiments, resulting in relatively less starch flow to and fermentation in the large intestine.The lower than formulated level of starch in the ration likely contributed to this finding by reducing dietary starch flow to the small intestine.In addition, differences among studies in chemical composition of the infused starch, starch content of the base diet, frequency of abomasal starch delivery (pulse dose vs. continuous), and intake of the base diet may lead to differences in studies in intestinal starch digestion.Total fecal VFA was linearly increased by abomasal starch (P = 0.05), but the effects were small, with only an 11% increase between CON and ST3 (Table 4).In addition, the overall treatment effect on total fecal VFA was not significant (P = 0.14).Fecal pH and VFA are strongly negatively correlated (Dijkstra et al., 2012), so it fits that we observed only modest effects on both measures.The increase in total fecal VFA was driven by tendencies for linear increases in acetate and propionate (P < 0.10) and a significant linear increase in butyrate (P < 0.01).These findings are similar to those of others who have observed increases in fecal acetate, propionate, and butyrate with feeding of a high-starch diet (Li et al., 2012;Tao et al., 2014;Neubauer et al., 2020).However, responses to abomasal corn starch are a little more variable.Literature shows consistent increases in fecal butyrate in response to continuous abomasal corn starch, but effects on acetate and propionate differ by study (Robbers et al., 2019;van Gastelen et al., 2021a,b).In our study, isovalerate linearly decreased with infused starch (P = 0.03), but there were no effects on isobutyrate, valerate, or lactate (P > 0.10).Decreased isovalerate has been previously observed in response to abomasal starch infusion (Gressley et al., 2016;van Gastelen et al., 2021a,b).As branched short chain fatty acids are produced by protein and AA fermentation (Zarling and Ruchim, 1987), it is possible that the increased availability of intestinal starch may have reduced protein fermentation and led to decreased isovalerate.
Despite the modest effects on fecal pH and VFA, we did observe a linear increase in fecal LPS with increasing abomasal corn starch (P < 0.01; 1 Treatments were abomasal infusion of corn starch product at 0 (CON), 1 (ST1), or 3 (ST3) g/kg BW, supplying starch at a mean of 0, 0.5, and 1.6 kg/d starch, respectively.The total amount was split into 2 bolus infusions, provided every 12 h.
3 P-value for the linear effect of increasing abomasal corn starch. 4EU = endotoxin unit.
Figure 1.Effect of abomasal infusion of corn starch product at 0 (CON), 1 (ST1), or 3 (ST3) g/kg BW, supplying a mean of 0, 0.5, and 1.6 kg/d starch, respectively, on fecal pH.The infusions were split into 2 bolus infusions provided every 12 h.Error bars represent standard error of the difference for the interaction of treatment and time.There was an interaction of treatment by time (P = 0.04).Abomasal starch linearly decreased fecal pH at 6 h (P < 0.01), but the linear effect was not observed at 0, 12, or 18 h (P = 0.96, P = 0.30, and P = 0.11, respectively).
back transformed means of our data indicate that LPS increased 23% between CON and ST1 (13,803 to 16,982 endotoxin units [EU] per gram) and 124% between ST1 and ST3 (16,982 to 38,019 EU/g).An increase in fecal LPS has been noted to occur as a result of a grain-based SARA challenge in both cows and goats (Li et al., 2012;Tao et al., 2014).This is likely due to increased intestinal availability of carbohydrates stimulating the growth of LPS-producing bacteria (Li et al., 2012).In our previous work, abomasal infusion of 1 g of starch/kg of BW delivered in 1 to 4 pulse doses per day increased fecal LPS to a similar extent as observed in the ST1 treatment in the present experiment (Gressley et al., 2016).However, 1 g of starch/kg of BW provided as twice daily pulse doses resulted in similar LPS as observed for the ST3 treatment (Neiderfer et al., 2020), and a grain-based SARA challenge increased fecal LPS approximately 3-fold relative to the ST3 treatment (Li et al., 2012).Thus, there appears to be some variability in the fecal LPS response to increased intestinal starch supply, but the fecal LPS observed in this study is likely within normal physiological ranges for cows fed moderate starch diets.
It was expected that abomasal starch would linearly increase fecal starch, as increased duodenal flows of starch are correlated with increased fecal starch content (Sanz-Fernandez et al., 2020).However, fecal starch content was relatively low and was not affected by treatment (range 1.73% to 3.10% DM, P = 0.23; Table 5).Interestingly, Li et al. (2012) found that a feeding challenge to induce SARA increased starch content in cecal digesta but did not significantly increase fecal starch.Thus, it is likely that in the present study some infused starch reached the large intestine and was fermented, as reflected by the decreased pH and increased VFA.However, this fermentation was fairly complete, resulting in no effects on fecal starch.Our results differ from those of Abeyta et al. (2023), where 4 kg/d abomasal pulse-dose corn starch dramatically increased fecal starch from 2.2% to 9.6%.Similarly, Piantoni et al. (2023) found that fecal starch numerically increased from 2.72% to 6.73% at 8 h after a single pulse-dose abomasal infusion of only 0.5 kg of an enzymatically resistant tapioca starch.Thus, the ability of abomasally infused starch to reach the large intestine and the extent of large intestinal fermentation of that starch seems to vary quite a bit among studies and may be related to the quantity infused, characteristics of the starch, frequency of infusion, infusion technique (continuous vs. pulse dose), and other animal or dietary factors.
Abomasal starch infusion did not affect total-tract apparent digestibility of DM, CP, NDF, or starch (Table 5; P > 0.10).Abomasal starch infusions often decrease apparent CP digestibility (Reynolds et al., 2001;van Gastelen et al., 2021a) due to large intestinal starch fermentation increasing fecal microbial protein (Westreicher-Kristen et al., 2018), but that was not observed in the current study.Effects are more variable for apparent total-tract starch digestibility, with some reporting a decrease (Robbers et al., 2019;van Gastelen et al., 2021a), and others finding no effect (Knowlton et al., 1998) or increased starch digestibility (Reynolds et al., 2001;van Gastelen et al., 2021b).Collectively, our results suggest that, even at the greatest infusion level, relatively more of the infused starch was enzymatically digested in the small intestine compared with other studies, resulting in no effects on digestibility or fecal starch and smaller effects on fecal pH and VFA.The extent of enzymatic digestion of starch in the small intestine and fermentation in the large intestine seems to vary across studies, making the development of challenge models difficult.
Treatment did not affect blood cell populations (Table 6) or haptoglobin (Table 7).For both serum amyloid A and haptoglobin, our reported concentrations would be considered within normal range.The typical serum amyloid A and haptoglobin concentrations for a healthy dairy cow range from 0 to 70 µg/mL (Trela et al., 2022) and 0 to 1 g/L (Huzzey et al., 2009), respectively.There were no effects of treatment on serum amyloid A, but there was a tendency for an interaction of treatment by day (P = 0.09; Table 7).Despite this tendency for an interaction, there were no linear effects of treatment at d 18 (P = 0.13) or d 21 (P = 0.87; data not shown).Similar to the present results, Rodríguez-Lecompte et al. ( 2014) found that a SARA challenge did not affect white blood cell populations.However, Abeyta et al. (2023) found that pulse-dose abomasal starch increased both total circulating white blood cells and neutrophils and tended to decrease lymphocytes.Grain induced SARA challenges consistently induce systemic inflammation as indicated by an increase in blood concentrations of acute phase proteins (Plaizier et al., 2008;Zebeli et al., 2012).It has been proposed that a downstream breach in intestinal barrier function may be driving at least some of this systemic response (Plaizier et al., 2008;Khafipour et al., 2009;Li et al., 2012).However, we and others have been unable to replicate this acute phase protein response when using an abomasal infusion model to generate hindgut acidosis independent of ruminal acidosis (Mainardi et al., 2011;van Gastelen et al., 2021b;Piantoni et al., 2023).Although this is not particularly unexpected in the current experiment given the modest response in fecal pH and VFA, this lack of response was also noted in studies that induced a dramatic decrease in fecal pH (van Gastelen et al., 2021a;Abeyta et al., 2023).Thus, at least in most challenge models to date, hindgut acidosis without accompanying ruminal acidosis does not appear to drive an acute phase response.
Despite the lack of effects on acute phase proteins and overall blood cell populations, abomasal starch tended to linearly decrease the proportion of γδ T cells (P = 0.06; Table 7), although there was no overall treatment effect (P = 0.13).The γδ T cells mature in the thymus through recognition of autoantigens, then migrate to the epithelial surfaces where they play roles in both immune regulation and activation (Cheroutre et al., 2011;Guerra-Maupome et al., 2019).Specifically, resident γδ T cells respond to different environmental stimuli by secreting anti-inflammatory cytokines that allow for protective and restorative effects or inflammatory cytokines and chemokines that recruit and activate immune cells (Chen et al., 2002;Guerra-Maupome et al., 2019).The γδ T cells can respond directly to Treatments were abomasal infusion of corn starch product at 0 (CON), 1 (ST1), or 3 (ST3) g/kg BW, supplying a mean of 0, 0.5, and 1.6 kg/d starch, respectively.The total amount was split into 2 bolus infusions, provided every 12 h.

2
For haptoglobin and serum amyloid A, samples were collected at 0 and 6 h following feeding on d 18 and 21 of each period.Mean values across the 0-and 6-h sampling times for each cow on each day were analyzed.For γδ T cells, samples were collected at 0 h on d 18 and 21 of each period.gut antigens including LPS, and they exhibit homing capabilities, accumulating in the intestines in response to inflammatory conditions (Yeung et al., 2000;Hedges et al., 2005).The tendency for a linear decrease in the proportion of peripheral circulating γδ T cells together with an increase in fecal LPS could be related to increased migration of γδ T cells to the intestinal epithelium, though this would need to be validated by further study.

CONCLUSIONS
Abomasal infusions of up to 1.6 kg/d starch resulted in a modest linear decrease in fecal pH and a linear increase in fecal VFA.Fecal LPS was also linearly increased.Despite this, abomasal corn starch did not affect apparent total-tract starch digestibility or fecal starch concentration, demonstrating that we did not overwhelm the capacity for starch digestion in the small and large intestine.Abomasal starch did not increase serum concentrations of acute phase proteins, but the proportion of circulating γδ T cells tended to be linearly decreased with increasing abomasal starch.This could suggest aggregation of immune cells in the intestinal epithelium in response to the starch challenge.Overall, abomasal infusion of up to 1.6 kg/d corn starch split into 2 pulse doses per day did not produce an inflammatory response in this experiment.

3P
-value for the linear effect of increasing abomasal corn starch.4 PBMC = peripheral blood mononuclear cells.

Table 1 .
Ingredient and nutrient composition (±SD) of the TMR, as percent of DM unless otherwise stated

Table 2 .
Cronin et al.: ABOMASAL STARCH AND BLOOD MEASURES Effect of abomasal infusion of corn starch on milk yield, intake, and milk composition 2P-value for the linear effect of increasing abomasal corn starch.3 DMI from diet only.4 DMI from corn starch infusate.5 DMI from diet and corn starch infusate.6 SCS = log 2 (SCC/100,000) + 3.

Table 3 .
Cronin et al.: ABOMASAL STARCH AND BLOOD MEASURES Effect of abomasal infusion of corn starch on rumen pH, VFA, and lactate

Table 4 )
. The Cronin et al.: ABOMASAL STARCH AND BLOOD MEASURES

Table 4 .
Effect of abomasal infusion of corn starch on fecal pH, LPS, VFA, and lactate

Table 5 .
Cronin et al.: ABOMASAL STARCH AND BLOOD MEASURES Effect of abomasal infusion of corn starch on apparent total-tract nutrient digestibility and fecal starch

Table 6 .
Cronin et al.: ABOMASAL STARCH AND BLOOD MEASURES Effect of abomasal infusion of corn starch on white blood cell populations

Table 7 .
Effect of abomasal infusion of corn starch on serum acute phase proteins and γδ T cells