Effects of heat stress abatement on systemic and mammary inflammation in lactating dairy cows

To examine the effects of evaporative cooling on systemic and mammary inflammation of lactating dairy cows, 30 multiparous Holstein cows (parity = 2.4, 156 d in milk) were randomly assigned to 1 of 2 treatments: cooling (CL) with fans and misters or not (NC). The experiment was divided into a 10-d baseline when all cows were cooled, followed by a 36-d environmental challenge when cooling was terminated for NC cows. The onset of environmental challenge was considered as d 1. Temperature-humidity index averaged 78.4 during the environmental challenge. Milk yield and dry matter intake (DMI) were recorded daily. Blood and milk samples were collected from a subset of cows (n = 9/treatment) on d −3, 1, 3, 7, 14, and 28 of the experiment to measure cortisol, interleukin 10 (IL10), tumor necrosis factor-α (TNF-α), haptoglobin, and lipopolysaccharide binding protein (LBP). Mammary biopsies were collected from a second subset of cows (n = 6/treatment) on d −9, 2, 10, and 36 to analyze gene expression of cytokines and haptoglobin. A subset of cows (n = 7/treatment) who were not subjected to mammary biopsy collection received a bolus of lipo-polysaccharides (LPS) in the left rear quarter on d 30 of the experiment. Blood was sampled from cows and milk samples from the LPS-infused quarter were collected at −4, 0, 3, 6, 12, 24, 48, and 96 h relative to infusion, for analyses of inflammatory products. Deprivation of cooling decreased milk yield and DMI. Compared with CL cows, plasma cortisol concentration of NC cows was higher on d 1 but lower on d 28 of the experiment (cooling × time). Deprivation of cooling did not affect circulating TNF-α, IL10, haptoglobin,


INTRODUCTION
Heat stress is a critical issue in the dairy industry.It not only impairs production but also negatively affects the fertility and health of dairy cows (Bouraoui et al., 2002;Wheelock et al., 2010;Gao et al., 2017).In lactating dairy cows, the reduced milk yield due to heat stress was estimated to cause $1.2 billion annual loss to the US dairy industry in 2012 (Key and Sneeringer, 2014).The impaired productivity in heat-stressed lactating dairy cows is due to a combination of decreased DMI, altered metabolism, immune activation by diseases and heat shock, and altered mammary gland function (Rhoads et al., 2009;Tao et al., 2020;Orellana Rivas et al., 2021).Additionally, there is a seasonal effect on mammary health.During the hot summer months, periods of elevated bulk tank milk SCC occur, as well as a concomitant increase in observed clinical cases of environmental mastitis (Hogan et al., 1989;Olde Riekerink et al., 2007).These suggest an association between heat stress and impaired mammary health.This association is commonly attributed to the impaired immune function of cows under heat stress and to the increased pathogen load in hot and humid environments (Kadzere et al., 2002;Hogan and Smith, 2012).
Heat stress also affects systemic and local inflammatory responses in animals after infection.In a sepsis model induced by intraperitoneal injection of LPS, mice exposed to heat stress displayed increased concentrations of tumor necrosis factor (TNF)-α in blood and peritoneal fluid compared with those under thermoneutrality, suggesting enhanced systemic and local inflammatory responses induced by heat stress (Jiang et al., 1999;Lee et al., 2012).Similarly, in murine models of acute lung injury induced by intratracheal injection of LPS, hyperthermic mice have increased concentrations of inflammatory cytokines (i.e., IL-1β) in bronchoalveolar lavage fluid compared with those exposed to thermoneutrality, indicating a stronger local inflammatory response induced by heat stress (Rice et al., 2005).Heat stress has also been reported to affect the systemic inflammation of dairy cows, but the results are inconsistent.Zhang et al. (2014) reported that circulating TNF-α and IL10 were increased in lactating dairy cows when the temperature-humidity index was high (THI = 80.3) compared with moderate and low THI (73.9 and 56.4,respectively).In lactating dairy cows, hyperthermia induced by an electric blanket increases the circulating LPS binding protein (LBP) and serum amyloid A (SAA) concentrations (Al-Qaisi et al., 2020).Synthesis and release of acute phase proteins, such as haptoglobin, LBP, and SAA, follow acute inflammation (Horadagoda et al., 1999;Eckersall et al., 2001).These data suggest that exposure to heat stress may increase systemic inflammation.In contrast, in early-and mid-lactating dairy cows, circulating concentrations of TNF-α were reduced by deprivation of evaporative cooling during summer (Safa et al., 2019;Marins et al., 2021).In Holstein steers, exposure to the heat stress had no effect on circulating haptoglobin, LBP, or SAA (Opgenorth et al., 2021).The discrepancies between studies may result from the intensities of heat stress, life cycle stages of the animals, distinct experimental models used, degree of heat abatement, length of treatment, inflammatory markers examined, and more.To the best of our knowledge, the effects of heat stress on mammary inflammation of healthy lactating dairy cows remain unclear.
The effects of heat stress on systemic and mammary inflammatory responses during mammary infection in lactating dairy cows are also seldom studied.We (Marins et al., 2019) previously reported that lactating dairy cows deprived of evaporative cooling during summer had a greater reduction in circulating neutrophils and lymphocytes and a greater increase in plasma lactose concentrations following intramammary LPS infusion compared with cows who received evaporative cooling.Neutrophils are the primary immune cells that migrate into the mammary gland during mastitis.Lactose is only synthesized in the mammary gland, and its plasma concentration is an indicator of the disrupted mammary epithelial junction caused by immune cell infiltration during mastitis.These data may suggest that heat stress enhances immune cell migration into the inflamed mammary gland induced by LPS.The degree of neutrophil migration into the inflamed mammary gland is dependent upon both the mammary inflammatory responses and the ability of a neutrophil to migrate in response to an inflammatory chemoattractant or chemotaxis.Because neutrophil chemotaxis is impaired by heat stress in lactating dairy cows (Elvinger et al., 1991(Elvinger et al., , 1992)), our previous data may indicate upregulated mammary inflammatory responses during mastitis by heat stress.Therefore, we hypothesized that heat stress upregulates the systemic and mammary inflammatory responses during LPS-induced mastitis in lactating dairy cows.Our objective was to evaluate the effects of deprivation of evaporative cooling on performance and systemic and mammary inflammation of lactating dairy cows before and following an intramammary LPS infusion.

Animals and Experimental Design
The experiment was conducted at the Dairy Research Center on the University of Georgia Tifton campus (Tifton, GA) from June to August 2019.Experimental procedures and animal handling were approved by the University of Georgia Institutional Animal Care and Use Committee before initiation of the experiment.
Multiparous lactating Holstein cows (parity = 2.4 ± 0.6, DIM = 156 ± 57d, means ± SD) were randomly assigned to 1 of 2 treatments, cooling (CL, n = 15) or without cooling (NC, n = 15), by tossing a coin.The sample size was determined based on the availability of cows and power tests.Two power tests were performed.In the first analysis, the milk yield of CL and NC cows (37.95 vs. 30.14 kg/d,SD = 3.21 kg/d,respectively) reported by Marins et al. (2021) was used to calculate animal number.Five cows per treatment was calculated using a level of significance of 0.05 and 90% power.In the second analysis, the plasma lactose concentration of CL and NC lactating cows at 3 h following an intramammary LPS infusion (33 vs. 112 μM, SD = 29 μM, respectively; Marins et al., 2019) was used to calculate animal number using a level of significance of 0.05 and 90% power.A sample size of 5 per treatment was determined.Thus, 15 cows/treatment will provide an adequate sample size.
Cows were housed in the same freestall barn and managed in the same manner.The entire experiment was divided into a 10-d baseline period followed by a 36-d environmental challenge.Parity (2.3 vs. 2.5, SEM = 0.1, respectively, P = 0.53) and DIM (163 vs. 149 d, SEM = 15 d, respectively, P = 0.51) were similar between CL and NC cows at the onset of the environmental challenge.During the baseline period, all cows received evaporative cooling provided by fans with misters attached to the front face.Fans (0.9-m diameter) were placed at 6-m intervals above feed bunk and freestalls, and provided a minimum of 9.5 km/h wind speed at the cow level.Fans operated when the air temperature exceeded 20°C, and misters were activated when ambient relative humidity ≤85%.During the environmental challenge, cooling was provided continuously to CL cows but not to NC cows.The onset of environmental challenge was considered as d 1 of the experiment.

Data and Sample Collection
Air temperature and relative humidity in the barn was monitored every 15 min throughout the experiment using Hobo Pro Series Temp probes (Onset Computer Corp., Pocasset, MA).The THI was calculated based on THI = (1.8 × T + 32) − [(0.55 − 0.0055 × RH) × (1.8 × T − 26)], where T = air temperature (°C) and RH = relative humidity (%; NRC, 1971).Vaginal temperature was measured every 5 min for 4 consecutive days each week in all cows using an iButton (Mouser Electronics, Mansfield, TX) attached to a blank intravaginal implant.Respiration rate was measured for all cows by counting flank movements for 1 min, 2 to 3 times each week (1330 h).
Cows were milked twice daily (0500 and 1700 h) every day, and yield was recorded at each milking (Delpro, DeLaval, Kansas City, MO).Milk samples were collected from 2 consecutive milkings each week and stored at 4°C with bronopol-B-14 as a preservative until analysis for milk components (fat, protein, lactose, SNF, MUN, and SCC) at the Dairy One Cooperative (Ithaca, NY).The same diet (Table 1) was fed to all cows as a TMR once daily (1600 h) throughout the experiment.Daily feed intake was recorded using the Calan Broadbent feeding system (American Calan Inc., Northwood, NH).Representative forage samples were collected daily, and remaining ingredients and TMR were sampled 3 times each week.Samples were dried in a forced-air oven at 55°C for 48 h to measure DM.Amounts of each dietary ingredient were adjusted according to changes in DM of each ingredient.Ingredient samples were composited weekly and ground to pass through a 1-mm screen using a Wiley mill (Thomas Scientific, Swedesboro, NJ).Chemical composition (ash, method 942.05, AOAC, 2000;CP, Leco FP-528 Nitrogen Analyzer, St. Joseph, MO;sugar, DuBois et al., 1956;starch, Hall, 2009;ADF, method 973.18, AOAC, 2000;ether extract, method 920.39, AOAC, 2000; and NDF treated using amylase and corrected for ash, Van Soest et al., 1991) were determined for each dietary ingredient to calculate the nutrient composition of the diet (Table 2).Body weight was measured and BCS assessed (Wildman et al., 1982) after the morning milking and before eating (0530 h) every week.
On d −3, 1, 3, 7, 14, and 28 of the experiment, additional milk samples were collected from a subset cows (n = 9/treatment) during the afternoon milking.Samples were stored without preservatives and transported to the laboratory at room temperature within 20 min after collection.Samples were centrifuged at 1,700 × g for 15 min at 4°C to obtain skim milk, which was stored at −80°C until further analysis.Blood was drawn from the same subset of cows as the skim milk collection (n = 9/treatment) on d −3, 1, 3, 7, 14, and 28 (1300 h) of the experiment.Samples were collected from coccygeal vessels into additive-free sodium-heparinized vacutainers and maintained at room temperature for 1 h or on ice before centrifugation at 1,700 × g for 30 min at 4°C to collect serum and plasma, respectively.Samples were stored in −20°C for analysis in the same batch.

Mammary Biopsy Collections
Mammary biopsies were collected from a subset of cows (n = 6/treatment) that were not enrolled in blood and skim milk collection at d −9 (as baseline samples), 2, 10, and 36 of the experiment, according to procedures described by Weng et al. (2017).Biopsies were collected only from the rear quarters.Tissues were collected from the left rear quarter on d −9 and 10, and from the right rear quarter on d 2 and 36 of the experiment.Before biopsy collection, cows were sedated by intravenous injection of xylazine hydrochloride (20 μg/kg of BW; Phoenix Pharmaceuticals, St. Joseph, MO).The skin area to perform tissue collection was carefully shaved and sanitized 3 times by scrubbing with iodine and 70% ethanol.A 3-mL dose of lidocaine hydrochloride (Animal Rx Pharmacy, Atlanta, GA) was subcutaneously injected above the biopsy region for local anesthesia.A 3-cm incision was made to penetrate the skin and connective tissue.Mammary tissue was collected using a rotating stainless-steel cannula with retractable blade connected to a cordless drill (Farr et al., 1996).Incisions were closed using 18-mm stainless-steel Michel wound clips (GerMedUSA, Garden City Park, NY) and sprayed with aerosol bandage (Neogen Corp., Lexington, KY) to prevent infection.The tissue collected was rinsed with saline, fat trimmed, and placed in 3 mL of RNAlater stabilization solution (Thermo Fisher Scientific, Waltham, MA), and then stored at −80°C until RNA extraction.

Intramammary LPS Infusion
Intramammary LPS infusion (i.m.LPS) was performed on d 30 during the environmental challenge for a subset of cows (n = 7/treatment) that were not enrolled in mammary biopsy collection.All cows enrolled in the intramammary LPS infusion were free of mastitis during the experiment and had SCC below 200,000 cells/ mL in milk samples collected on d 28 of the experiment.During and after i.m.LPS, all cows remained in their respective pens.Two hours after morning milking (0700 h), the left rear quarter of each cow was infused with 10 μg of Escherichia coli O111:B4 LPS (Sigma-Aldrich, St. Louis, MO; Marins et al., 2019) dissolved in 5 mL of pyrogen-free saline solution and massaged to move the inoculum into the gland cistern.The right rear quarter was considered as a control quarter, but no infusion was performed to avoid potential contamination and infection.A previous study suggests that infusion with saline alone has no effect on milk SCC, suggesting minimal influence on mammary inflammation (Shangraw et al., 2020).The vaginal temperature was recorded every 5 min from 24 h before through 48 h after i.m.LPS.Blood was drawn from coccygeal vessels of all cows into additive-free sodium-heparinized Vacutainers (Becton Dickinson) at −4, 0, 3, 6, 12, 24, 48, and 96 h relative to infusion, to collect serum and plasma, respectively.Additional blood samples were collected into Vacutainers containing K3 EDTA (Becton Dickinson) at 0, 3, 6, 12, and 24 h relative to infusion for hematologic profile were analyzed at the University of Georgia Veterinary Diagnostic Laboratory (Tifton, GA) using the ADVIA 2120i Hematology System (Siemens, Tarrytown, NY).Individual milk samples from both left and right rear quarters were collected at −4, 0 (immediately before infusion), 3, 6, 12, 24, 48, and 96 h relative to infusion to measure milk composition, as described previously.Additional milk was sampled from the left rear quarters to collect skim milk at −4, 0 (immediately before infusion), 3, 6, 12, 24, 48, and 96 h relative to i.m.LPS.Before collection, teats were dipped with an iodine-based solution (0.1%) and cleaned using paper towels.The first 10 streams of milk were discarded.After sampling, teats were dipped again in the iodine-based solution.
0-50 ng/mL, Kingfisher Biotech Inc., Saint Paul, MN), and IL10 (Bovine IL-10 DIY ELISA, detection limit: 0-5 ng/mL, Kingfisher Biotech Inc.) were analyzed using commercially available ELISA kits following the manufacturers' instructions.Plasma concentrations of haptoglobin were determined by a colorimetric method described by Cooke and Arthington (2013).Within each plate, the number of cows for each treatment was always balanced and placed in the random order.For samples collected before i.m.LPS, the intra-and interassay CV were, respectively, 5.0 and 6.7% for cortisol, 3.4 and 3.3% for LBP, 3.2 and 15.8% for TNF-α, 3.7 and 8.1% for IL10, and 3.5 and 1.5% for haptoglobin.
After thawing, skim milk samples were centrifuged at 15,000 × g for 45 min at 4°C to collect the supernatant for analysis.Concentrations of TNF-α, IL10, haptoglobin, and LBP were analyzed using the same procedures as described above.For samples collected before i.m.LPS, the intra-and inter-assay CV were, respectively, 2.9 and 10.8% for IL10, 2.5 and 18.6% for TNF-α, and 1.8 and 4.4% for haptoglobin.Because only milk samples collected at d −3, 1, and 28 of the experiment were used to measure LBP, the intra-assay CV of LBP assay was 11.2%.For skim milk samples collected after i.m.LPS, the intra-and interassay CV were 3.8 and 9.2% for IL10, and 3.0 and 18.7% for TNF-α, and 1.6 and 3.6% for haptoglobin, respectively.

RNA Extraction, cDNA Synthesis, and Quantitative RT-PCR
Total RNA from mammary biopsies was extracted using TRI-Reagent (Sigma-Aldrich, St. Louis, MO) and PureLink RNA Mini Kit (Invitrogen, Carlsbad, CA) with an on-column DNase treatment (PureLink DNase Set, Invitrogen) according to the manufacturers' instructions.Following extraction, the second DNase treatment was performed using a Turbo DNA-Free Kit (Invitrogen).The RNA samples were stored at −80°C until analyses.The cDNA was synthesized using a High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Applied Biosystems, Foster City, CA).Primers were designed using PrimerQuest (Integrated DNA Technologies, Coralville, IA) and have PCR efficiencies above 90% (Supplemental Table S1).All primers were compared against the bovine genome using BLAST, to ensure that primers were aligned only on the target genes.GAPDH was used as a housekeeping gene.The housekeeping gene was selected from ACTB (actin β), GAPDH (glyceraldehyde 3-phosphate dehydrogenase), and KRT8 (keratin 8) using the GeNorm program.ACTB and GAPDH had smaller M-values (0.052).The threshold cycle (CT) values of both ACTB and GAPDH for all samples were then subjected to statistical analyses to examine the effects of treatment (NC vs. CL), time, or treatment by time interactions.The gene with higher P-values was selected (GAPDH, P > 0.15).The target genes included tumor necrosis factor (TNF1), interleukin-10 (IL10), and haptoglobin (HP).Real-time PCR was carried out using cDNA reversed transcribed from 10 ng of purified RNA and Power SYBR Green Master Mix (Applied Biosystems, Carlsbad, CA) on a StepOnePlus thermocycler (Applied Biosystems).Each reaction used the following conditions: 95°C for 10 min for initial denaturation and enzyme activation, 40 cycles of 95°C for 15 s, and 57°C for 1 min for amplification.Melting curve analysis was performed for each reaction to ensure specificity of the reaction.The "no template controls" were included in all plates to detect contamination, primer dimers, or misprimed products.The 2 -ΔΔCT method, where CT is the threshold cycle, was used to calculate relative gene expression.Samples collected on d −9 of the experiment were used as calibrator samples.

Statistical Analyses
All procedures were performed using SAS 9.4 (SAS Institute Inc., Cary, NC).The PROC UNIVARIATE procedure was used to calculate THI.Repeated measures data, including BW, BCS, vaginal temperature, respiration rate, DMI, milk yield, percentage and yield of milk components, feed efficiency measures (milk yield/DMI, FCM/DMI, ECM/DMI), hematological profiles, plasma, serum or milk concentrations of cortisol, inflammatory cytokines, and haptoglobin, were analyzed using PROC MIXED.The statistical models included treatment, time, and the interaction between treatment and time as fixed variables.Cow nested within treatment was included as a random variable.Because of the frequent mammary biopsy collection, milk yield and composition, DMI, and feed efficiency measure data for cows enrolled in mammary tissue collection were not included in the final analysis.Data collected during the baseline (before the onset of environmental challenge) were included in the models as covariates.For data collected during i.m.LPS, data collected before LPS infusion were included in the SAS models as covariates.Milk composition data collected from control quarters and LPS-infused quarters during i.m.LPS were analyzed separately.Least squares means ± standard error of the means were reported.Significance and tendency were declared when P ≤ 0.05 and 0.05 < P ≤ 0.10, respectively.

Environmental Challenge
During the environmental challenge, THI averaged 78.6 ± 4.5 and 78.1 ± 4.5 (means ± SD) in pens housing NC and CL cows, respectively, suggesting that all cows were exposed to similar intensities of heat stress.Compared with CL cows, NC cows had greater (treatment effects: P < 0.01) vaginal temperature and respiration rate (Table 3).Treatment by time interactions (P < 0.01) occurred for both vaginal temperature and respiration rate, such that the differences between treatments were greater during the first week of the environmental challenge than during the remainder of the trial (Figure 1).Within a day, the vaginal temperature of NC cows was consistently higher than that of CL cows, peaking at 1500 h and reaching nadir at 0800 h (Supplemental Figure S1).Cows without evaporative cooling had lower (P ≤ 0.04) BW, DMI, and yields of milk, FCM, and ECM (Table 3) relative to CL cows.However, no differences (P ≥ 0.11) were observed between treatments for BCS, feed efficiency measures, and concentrations of fat, protein, lactose, SNF, and SCS (Table 3).In contrast, MUN concentrations were greater (P = 0.02) for NC cows compared with CL cows (Table 3).Because of the lower milk yield, yields of milk protein and lactose were lower (P ≤ 0.01) for NC cows, and SNF yield tended (P = 0.07) to be lower compared with CL cows (Table 3).
A treatment by time interaction (P < 0.01) was observed for plasma cortisol concentrations (Table 4).Circulating cortisol concentrations were higher (P = 0.04) at d 1 for NC cows but lower (P = 0.02) on d 28 compared with CL cows (Figure 2).A treatment by time interaction (P = 0.02) was also observed for serum IL10 concentration; however, no treatment effects at individual time points were observed according to the SLICE function of SAS (Supplemental Figure S2).No differences (P ≥ 0.28) were observed for treatment effect or interaction of treatment by time for circulating TNF-α, LBP, and haptoglobin concentrations (Table 4, Supplemental Figure S2).Compared with CL cows, milk concentration of IL10 tended (P = 0.08) to be higher for NC cows (Table 4, Supplemental Figure S3).However, no treatment effect or treatment by time interactions (P ≥ 0.17) were observed for milk concentrations of TNF-α, LBP, and haptoglobin (Table 4, Supplemental Figure S3).Due to the lower milk yield, NC cows had lower (P = 0.03) milk haptoglobin yield per milking and tended (P = 0.09) to have lower milk TNF-α yield per milking relative to CL cows (Supplemental Figure S3).In contrast, the yield of milk IL10 per milking was similar (P = 0.49) between treatments (Table 4, Supplemental Figure S3).No differences (P > 0.10) were observed in mammary gene expression of cytokines and haptoglobin (Supplemental Table S2).

Intramammary LPS Infusion
Regardless of the treatments, i.m.LPS reduced DMI and milk yield (time effect: P < 0.01, Figure 3).Compared with CL cows, milk yield and DMI for NC cows remained lower (P < 0.01) after i.m.LPS (Figure 3).However, relative to the data collected before i.m.LPS, NC cows had greater (P = 0.01) reduction in DMI on d 4 and tended (P = 0.09) to have greater reduction in DMI on d 2 after i.m.LPS compared with CL cows (treatment × time: P = 0.08, Figure 3).Both NC and CL cows had similar (P ≥ 0.78) changes in milk yield relative to pre-infusion level (Figure 3).A treatment by time interaction (P < 0.01) was observed for vaginal temperature during i.m.LPS (Figure 4).Vaginal temperature reached a similar peak at 6 h after i.m.LPS for both NC and CL cows; however, vaginal temperature was higher (P < 0.01) before and after the peak for NC cows compared with CL cows (Figure 4) Infusion of LPS decreased concentrations of fat, lactose, and SNF in milk but increased concentrations of protein, MUN, and SCC in LPS-infused quarters (time: P < 0.01, Table 5, Figure 5, Supplemental Figure S4).No treatment or treatment by time interactions (P ≥ 0.37) were observed for milk concentrations of fat and lactose (Table 5, Supplemental Figure S4).Compared  with those of CL cows, the milk samples collected from LPS-infused quarters of NC cows had greater (P ≤ 0.05) concentrations of SCC and MUN (Figure 5, Supplemental Figure S4).Tendencies (P ≤ 0.08) for treatment by time interaction were observed for milk concentrations of protein and SNF.These are because the milk collected from LPS-infused quarters of NC cows had greater (P < 0.01) concentration of protein at 12 h and greater (P = 0.02) concentration of SNF at 48 h following i.m.LPS compared with milk collected from CL cows (Table 5, Supplemental Figure S4).Concentrations of milk fat, protein, lactose, MUN, and SNF decreased, but SCC increased in the control quarter during the collection period (without LPS infusion; time: P < 0.01, Table 5, Supplemental Figure S5).Compared with CL cows, the control quarters of NC cows had lower (P < 0.01) concentrations of lactose and SNF, and tended (P = 0.09) to have lower milk fat concentration.An interaction of treatment by time was observed for concentration of MUN of the control quarters.Milk urea nitrogen tended (P ≤ 0.09) to be higher at 0, 3, and 6 h but lower (P = 0.03) at 24 h in the non-infused quarters of NC cows compared with CL cows (Table 5, Supplemental Figure S5).Intramammary LPS infusion reduced (time: P < 0.01) circulating concentrations of white blood cells, lymphocytes, neutrophils, monocytes, basophils, and eosinophils without regard to treatment over the collection period.However, no treatment or treatment by time interactions (P ≥ 0.11) were observed for hematological profiles (Supplemental Table S3).Following i.m.LPS, circulating TNF-α concentration tended (P = 0.07) to be higher at 48 h compared with 0 h (time effect: P = 0.07), and serum IL10 concentrations were lower (P < 0.05) at 12 and 24 h relative to 3 and 6 h (time effect: P < 0.01, Figure 6).Relative to pre-infusion level (0 h), i.m.LPS increased (P < 0.01) plasma haptoglobin concentrations from 12 to 48 h after infusion (time: P < 0.01), and the peak concentration was observed at 48 h following infusion (Figure 6).No treatment by time interaction was observed for circulating TNF-α, IL10, or haptoglobin concentrations, but NC cows had lower (treatment: P < 0.01) serum TNF-α and IL10   concentrations and tended (treatment: P = 0.09) to have lower plasma haptoglobin concentration compared with CL cows during i.m.LPS.Following LPS infusion, milk concentrations of IL10, TNF-α, and haptoglobin increased over time (time: P < 0.01) regardless of treatment.The peak milk concentrations were observed at 6 h for TNF-α, 12 h for IL10, and 48 h following i.m.LPS for haptoglobin (Figure 6).Compared with CL cows, milk concentrations of IL10 and haptoglobin of NC cows tended (treatment: P ≤ 0.07) to be higher during the i.m.LPS (Figure 6).Although no treatment by time interaction (P ≥ 0.12) was observed for milk Table 5. Milk composition of lactating Holstein cows provided evaporative cooling (CL, n = 7) or not (NC, n = 7).Milk samples were collected from LPS-infused and non-infused (control) quarters at 3, 6, 12, 24, 48, and

DISCUSSION
Heat stress caused by a combination of high ambient temperature and humidity leads to significant productive and physiological responses in dairy cows, such as increased body temperature and respiration rate and decreased DMI and milk yield (West, 2003).In this experiment, we employed a widely used experimental model to compare the responses of cows provided with evaporative cooling (soakers, misters, and fans) against those that are not.Using this experimental model, a previous study reported that lactating cows without evaporative cooling had 0.91 ○ C (2.3%; 39.94 vs. 39.03○ C, respectively) higher vaginal temperatures relative to cows provided with evaporative cooling (Weng et al., 2018).Similarly, in the present study, NC cows had higher vaginal temperatures and respiration rates than CL cows, indicating that deprivation of evaporative cooling increased the heat load of cows.Thus, the experimental model used in this study is suitable for examining the influence of heat stress on lactating dairy cows.However, it is also important to note that, despite less heat load carried, CL cows were also exposed to similar intensity of heat stress as the NC cows, and experienced mild hyperthermia (>38.5°C).
The differences in vaginal temperatures and respiration rates between CL and NC cows were greater in the first week of the experiment compared with following weeks.This suggests that cows have stronger physiological responses under acute exposure to heat stress than under prolonged and chronic heat stress, due to acclimation.Consistent with previous research using a similar experimental model (Weng et al., 2018;Marins et al., 2021), NC cows had lower DMI and milk yields compared with CL cows.The decreased nutrient intake is thought to account for 35 to 50% of the decrease in milk yield with heat stress, because cows under heat stress tend to utilize energy and nutrients (e.g., glucose) for body functions rather than for milk synthesis (Rhoads et al., 2009;Wheelock et al., 2010).However, in the current study, feed efficiency measures were not different between CL and NC cows, suggesting that the reduced DMI may account for most of the reduction in milk yield.
The effects of heat stress on milk protein concentrations have been inconsistent.In similar studies to our current experiment, deprivation of evaporative cooling from lactating dairy cows during summer had no impact on milk protein concentrations (Chen et al., 1993, Chan et al., 1997, Weng et al., 2018, Marins et al., 2021).In contrast, when housed in the environmentally controlled chambers, cows under heat stress are reported to have decreased milk protein concentrations relative to those housed under thermoneutrality (Rhoads et al., 2009, Gao et al., 2017).These discrepancies between studies emphasize the importance of experimental models on the impact of heat stress on production responses.Seasonality of milk fat concentration has been reported and is typically lowest during the summer (Kadzere et al., 2002).In contrast, milk fat concentration was unaltered by heat stress or deprivation of evaporative cooling in controlled studies (Shwartz et al., 2009, Weng et al., 2018, Marins et al., 2021), consistent with the result observed in our current experiment.This discrepancy suggests that factors other than heat stress contribute to the reduced milk fat concentration observed during summer.Milk lactose concentration was not different between treatments in this experiment which is consistent with Gao et al. (2017) but differs from Weng et  2018) who reported lower milk lactose concentration for NC cows compared with CL cows.Different experimental models, diets and management systems may contribute to the inconsistent impact reported on milk lactose concentration by heat stress.In the current study, MUN was higher for NC than CL cows which is consistent with previous studies and results from altered N metabolism in the rumen or increased amino acid deamination in the liver, or both (Ríus, 2019).
Circulating cortisol concentrations were higher for NC cows than for CL cows immediately after deprivation of cooling, suggesting increased cortisol release under acute heat stress, consistent with previous studies (Hall et al., 2018;Marins et al., 2021).Following the initial increase, no differences were observed in plasma cortisol concentrations between NC and CL cows until d 28 of the experiment, when circulating cortisol concentrations were lower for NC cows compared with CL cows.Christison and Johnson (1972) reported that cortisol secretion increased when dairy cows were exposed to acute heat stress but was depressed during chronic heat stress.Tumor necrosis factor-α is a proinflammatory cytokine and is commonly used as a marker for inflammatory status.Compared with those raised during the period with temperate weather conditions, lactating dairy cows raised during the period under heat stress conditions have greater circulating TNF-α concentration, suggesting upregulated systemic inflammation (Min et al., 2016;Chen et al., 2018).In contrast, Safa et al. (2019) and Marins et al. (2021) reported that deprivation of evaporative cooling reduced circulating TNF-α in lactating dairy cows.The anti-inflammatory cytokine IL10 is primarily synthesized by monocytes, Th2 lymphocytes, and B lymphocytes (de Waal Malefyt et al., 1992;Opal and DePalo, 2000).Consistent with Marins et al. (2021), deprivation of evaporative cooling did not affect circulating IL10 in lactating dairy cows in the present study.In contrast, Zhang et al. (2014) reported that mid-lactating cows experiencing high-THI (80.3) conditions had higher circulating IL10 than cows during low-THI (56.4) conditions.Acute phase response occurs when animals experience inflammation and is characterized by the increased hepatic synthesis of acute phase proteins such as LBP and haptoglobin (Ceciliani et al., 2012).In lactating dairy cows, circulating LBP and SAA concentrations increased during hyperthermia induced by electric blankets (Al-Qaisi et al., 2020).However, in our currently study, neither plasma haptoglobin nor LBP concentrations were affected by treatments.Coupled with the unchanged circulating cytokine concentrations, our results suggest that deprivation of evaporative cooling did not affect systemic inflammation in healthy lactating dairy cows.
Acute phase proteins are produced not only in the liver but also in the mammary gland (Hiss et al., 2004).Concentrations of acute phase proteins in milk increase during mastitis, making them useful indicators for mammary infections or inflammation (Jawor and Stefaniak, 2011).In this present study, milk collected from NC cows tended to have higher IL10 concentrations but had similar TNF-α, haptoglobin, and LBP concentrations compared with milk collected from CL cows.Our results indicate that deprivation of evaporative cooling did not induce inflammation in the lactating mammary gland.These data are consistent with the similar mammary gene expression of cytokines and haptoglobin and the unchanged milk SCS between CL and NC cows.The lower yield of milk TNF-α and haptoglobin reflects the lower milk yield of NC cows compared with CL cows.
Consistent with previous studies (Silanikove et al., 2011;Marins et al., 2019), milk yield and DMI were reduced by i.m.LPS.Reduced milk yield following mammary inflammation results from both systemic effects (e.g., reduced DMI) and local effects such as apoptosis and cell damage of the mammary gland (Zhao and Lacasse, 2008;Ballou, 2012).Relative to pre-infusion, NC cows had a similar reduction in milk yield but a greater decrease in DMI at 2 and 4 d following i.m.LPS compared with CL cows.This indicates that heat stress did not affect milk yield response when mammary inflammation is induced in lactating dairy cows by LPS but results in stronger systemic responses (e.g., reduced DMI).
Regardless of the treatment, milk fat concentrations in both LPS-infused and non-infused quarters decreased to their lowest concentration at 6 h after i.m.LPS, suggesting reduced mammary fat synthesis.Following i.m.LPS, plasma nonesterified fatty acid concentration declines (Pires et al., 2019).This may inhibit incorporation of preformed fatty acids into milk fat in the mammary gland.The fact that no differences were observed among treatments in milk fat concentrations following LPS infusion suggests that deprivation of evaporative cooling did not affect milk fat synthesis in the inflamed mammary gland.The increased milk protein concentration in LPS-infused quarters following infusion in both treatments is consistent with previous studies (Aditya et al., 2017;Marins et al., 2019).As discussed by Marins et al. (2019), the increased milk protein concentration after mammary inflammatory challenge is due to a combination of the dilution effect by lower milk yield, increased serum protein influx because of disrupted mammary epithelial junction, and enhanced synthesis of antibacterial proteins and inflammatory products such as inflammatory cytokines and acute phase proteins.Greater milk protein concentrations in the LPS-infused quarters of NC cows compared with CL cows at 12 h following infusion may be partially due to the greater mammary synthesis of acute phase proteins, such as haptoglobin.
Intramammary LPS infusion reduced milk lactose concentration in LPS-infused quarters until 12 h following infusion regardless of treatments.This is partially because of the disrupted mammary epithelial junction and lactose leakage into circulation during mastitis (Kobayashi et al., 2013).Marins et al. (2019) reported that plasma lactose increases and peaks at 3 h after i.m.LPS.Additionally, Waldron et al. (2006) reported that i.m.LPS resulted in ~850 g glucose consumption by tissues other than the mammary gland within 14 h after infusion in early-lactating dairy cows.This reduces lactose synthesis within the mammary gland due to limited substrate (glucose) supply.Marins et al. (2020) reported that NC cows had lower circulating glucose concentrations but a more rapid and prolonged increase in serum insulin concentration following i.m.LPS compared with CL cows.This may suggest greater usage of glucose by tissues rather than by the mammary gland in NC cows and may partially explain the lower milk lactose concentration in the non-infused quarters of NC cows compared with CL cows.In this current study, NC cows had increased MUN concentrations in LPSinfused quarters.The result was consistent with a previous report (Marins et al., 2019) and suggested altered ruminal and peripheral protein metabolism by heat stress during mammary inflammation induced by LPS.As expected, milk SCC increased in the LPS-infused quarter.Interestingly, NC cows tended to have higher SCC in the LPS-infused quarter than CL cows.It is important to note that 2 NC cows had milk samples collected at 12 h after infusion that had SCC higher than the detection limit and were not included in the final analysis.These data suggest a greater immune cell At 6 h following i.m.LPS, body temperature, as measured by vaginal temperature, peaked at 41.1°C, confirming upregulated systemic inflammation.Compared with CL cows, NC cows had higher vaginal temperature before and after i.m.LPS due to the greater heat load.However, both treatments reached a similar peak of vaginal temperature following i.m.LPS, indicating similar fever development.This is consistent with our previous report that the body temperatures of CL and NC cows reach a similar peak following an intravenous LPS infusion-induced systemic inflammation (Marins et al., 2021).Regardless of treatment, i.m.LPS slightly increased circulating TNF-α only at 2 d following infusion, and decreased serum IL10 concentration at 12 to 48 h after infusion.It is common to observe no changes or only minor changes in circulating concentrations of inflammatory cytokines during i.m.LPS.Lehtolainen et al. (2004) reported that circulating TNF-α concentration was too low to be detected after i.m.LPS in lactating dairy cows.Similarly, Johnzon et al. (2018) reported no significant changes in plasma concentration of TNF-α following i.m.LPS.In contrast, systemic inflammation induced by intravenous LPS infusion in lactating dairy cows increased circulating concentrations of TNF-α by 75-fold and of IL10 by 10-fold at 1 h following infusion (Marins et al., 2021).Plasma haptoglobin concentration increased and peaked at 48 h following i.m.LPS, consistent with previous reports (Suojala et al., 2008;Jawor and Stefaniak, 2011).Vels et al. (2009) and Zarrin et al. (2014) also reported upregulated hepatic haptoglobin gene expression following i.m.LPS.These data suggest an upregulated systemic acute phase response during LPS-induced mammary inflammation.Interestingly, NC cows had lower circulating TNF-α, IL10, and haptoglobin concentrations, indicating a weaker systemic inflammatory response after i.m.LPS compared with CL cows.However, the mechanisms and impacts of this effect are not clear.This deserves further investigation.
In contrast to circulating concentrations, milk concentrations of TNF-α and IL10 profoundly increased and peaked at 6 and 12 h after i.m.LPS, respectively.Coupled with the increased milk haptoglobin concentration, these data confirm the upregulated mammary inflammation.Similarly, Perkins et al. (2002) and Suojala et al. (2008) reported increased milk inflammatory cytokine and acute phase protein concentrations after intramammary endotoxin or E. coli infusion.Schmitz et al. (2004) and Zarrin et al. (2014) also reported upregulated mammary gene expression of TNF-α, IL10, haptoglobin, and milk amyloid A after i.m.LPS.Interestingly, the milk collected from NC cows had greater concentrations of TNF-α and IL10 at 3 h following i.m.LPS, suggesting more rapid development of mammary inflammation compared with CL cows.Coupled with the greater milk haptoglobin concentration, these data suggest that the mammary gland of NC cows develops faster and stronger inflammatory responses to LPS compared with the mammary gland of CL cows.These changes explain the greater concentration of milk SCC for NC cows compared with CL cows following i.m.LPS observed in this experiment.Mammary inflammatory responses to infection are initiated by resident macrophage or mammary epithelial cells that recognize and bind pathogen-associated molecular patterns of bacterial molecules (Schukken et al., 2011;Wellnitz and Bruckmaier, 2012).Marins et al. (2021) reported that peripheral blood mononuclear cells isolated from lactating dairy cows without cooling had greater TNF-α and IL10 production when stimulated with LPS in vitro compared with CL cows.Similarly, Molinari et al. (2023) reported that, during summer, circulating immune cells collected from early-lactating cows who did not receive evaporative cooling during the prepartum period had greater inflammatory cytokine production (IL1β, IL10) when challenged by LPS in vitro compared with cells from cows who were cooled before calving.In miniature pigs, heat stress upregulates the gene expression of tolllike receptor-4 in peripheral blood mononuclear cells (Ju et al., 2014).Consistently, peritoneal macrophages isolated from heat-stressed mice produced more TNF-α after in vitro LPS stimulation compared with mice under thermal neutrality (Lee et al., 2012).The data suggest that heat stress upregulates the inflammatory responses of immune cells to LPS, potentially including the resident macrophages within the mammary gland.It is important to note that bovine mammary epithelial cells also express toll-like receptor-4 and are able to produce inflammatory cytokines and chemokines (Pareek et al., 2005;McClenahan et al., 2006;Ibeagha-Awemu et al., 2008).Thus, the possibility that the inflammatory responses of mammary epithelial cells to LPS or bacterial stimulation is also upregulated by heat stress cannot be excluded.
In conclusion, results from our current experiment demonstrate that deprivation of evaporative cooling during summer significantly impairs lactating dairy cows' performance.However, similar circulating and milk concentrations of cytokines and acute phase proteins between NC and CL cows suggest unchanged systemic and mammary inflammation.In contrast, deprivation of evaporative cooling depressed the systemic inflammation but resulted in a faster and stronger mammary inflammatory response following i.m.LPS, resulting in greater increase in milk SCC.The results indicated that heat stress might lead to a stronger Chen et al.: HEAT STRESS ABATEMENT AND INFLAMMATION Chen et al.: HEAT STRESS ABATEMENT AND INFLAMMATION

Figure 1 .
Figure 1.Vaginal temperature (VT, A) and respiration rate (RR, B) of cows exposed to evaporative cooling (solid circles •) or not (open circles ○) during environmental challenge.The solid line without marks represents the daily average temperature-humidity index (THI) of the pen that housed non-cooled cows, and the dashed line without marks represents the daily average THI of pen that housed cooled cows.**P ≤ 0.01, *P ≤ 0.05.

Figure 4 .
Figure 4. Vaginal temperature (VT) of cows exposed to evaporative cooling (solid circles •) or not (open circles ○) during intramammary LPS infusion.The solid line without marks represents the daily average temperature-humidity index (THI) of the pen that housed non-cooled cows, and the dashed line without marks represents the daily average THI of the pen that housed cooled cows.
cytokine concentrations, a separate analysis of data collected only from the 3 h after the infusion suggested greater (P < 0.05) milk concentrations of TNF-α and IL10 of NC cows compared with CL cows (insets, Figure6D, 6E).

Figure 5 .
Figure 5. Milk SCC concentration of cows exposed to evaporative cooling (solid circles •) or not (open circles ○) during intramammary LPS infusion.Data were transformed to natural log (Ln) for statistical analysis.Data presented as LnSCC (A) and SCC (B, back-transformed from LnSCC).BL = data collected prior to intramammary LPS infusion.
Chen et al.: HEAT STRESS ABATEMENT AND INFLAMMATIONinflux into the mammary gland of NC cows than CL cows following i.m.LPS.

Table 1 .
Chen et al.: HEAT STRESS ABATEMENT AND INFLAMMATION Ingredient composition of the experimental diets

Table 2 .
Chemical composition of the experimental diet 1 Mean ± SD.

Table 4 .
Blood concentrations of cortisol and inflammatory products and milk concentrations and yields of inflammatory products of lactating Holstein cows exposed to evaporative cooling (CL, n = 9) or not(NC, n Chen et al.: HEAT STRESS ABATEMENT AND INFLAMMATION Chen et al.: HEAT STRESS ABATEMENT AND INFLAMMATION