Long-term growth, feed efficiency, enteric methane emission, and blood metabolite responses to in utero hyperthermia in Holstein heifers

Dairy producers are experiencing production and animal welfare pressures from the increasing frequency and severity of heat stress events due to global climate change. Offspring performance during the pre-weaning and lactating periods is compromised when exposed to heat stress during late gestation ( in utero ). However, knowledge of the lingering impacts of in utero heat stress on yearling dairy heifers is limited. Herein, we investigated the long-term effects of in utero heat stress on heifer growth, feed efficiency, and enteric methane emissions in post-pubertal heifers. During the last 56 d of gestation, 38 pregnant cows carrying heifer calves were exposed to either heat stress (IUHT; n = 17) or artificial cooling (IUCL; n = 21). At 18 ± 1 mo of age, the resulting IUCL and IUHT heifers were enrolled in the present 63-d study. Heifers were blocked by weight and randomly assigned to 3 pens with Calan gates. Body weights (BW) were recorded on 3 consecutive days at the start and end of the trial and used to calculate average daily gain (ADG). Body condition score (BCS), hip width, body length, and chest girth were measured at the start and end of the study. All heifers were fed a TMR comprised of 46.6% oatlage, 44.6% grass/alfalfa haylage, 7.7% male-sterile corn silage, 0.3% urea, and 0.8% mineral/vitamin supplement (DM basis). The TMR and refusal samples were obtained daily, composited weekly, and dried to calculate DMI. During the study, each pen had access to a GreenFeed unit for 8 ± 1d to measure CH 4 and CO 2 gas fluxes. During the last 3 d of measuring CH 4 and CO 2 fluxes, fecal samples were collected, composited by animal, dried, and analyzed to calculate NDF, OM, and DM digestibility. On the last day of fecal sampling, blood samples were also collected via coccygeal venipuncture, and gas chromatography time-of-flight mass spectrometry analysis was performed Residual feed intake (RFI; predicted DMI - observed DMI) and feed conversion efficiency (FCE; DMI/ADG) were calculated to estimate feed efficiency. No differences were found in initial or final BW, hip width, chest girth, or BCS; however, IUCL heifers were longer in body length compared with IUHT heifers. Dry matter intake, ADG, RFI, and FCE were similar between IUHT and IUCL heifers. In utero heat stressed and IUCL heifers produced similar amounts of CH 4 and CO 2 , and no differences were found in the number of GreenFeed visits or latency to approach the GreenFeed. The concentrations of 6 blood metabolites involved in lipogenic pathways were different between in utero treatments. In conclusion, in utero heat stress does not seem to have long-term effects on feed efficiency or methane emissions during the post-pubertal growing phase; however, IUCL heifers maintained a body length advantage over their IUHT counterparts and differed in concentrations of several candidate metabolites that encourage further exploration of their potential function in key organs, such as the liver and mammary gland.


INTRODUCTION
Feed costs incurred by replacement heifers comprise approximately 46% of rearing costs and nearly 25% of dairy farm production costs (Akins et al., 2015;Karszes and Hill, 2020), and improving feed efficiency can provide economic benefits, improve land and water utilization, and reduce greenhouse gas emissions.In 2021, enteric methane emissions from animals raised for agriculture accounted for 3.1% of total greenhouse gas emissions in the US (EPA, 2023); however, few studies have investigated the relationship between enteric methane emissions and feed efficiency.A common measure of feed efficiency in dairy cattle is residual feed intake (RFI), in which dry matter intake (DMI) is regressed on the known energy sinks of body weight (BW), change in BW (ΔBW), and secreted milk energy (NE L ).Residual feed intake represents the difference between observed DMI and predicted DMI for an individual; therefore, an efficient animal consumes less than predicted and has a negative RFI value.Precision management (ben Meir et al., 2019;Fischer et al., 2020) and genetic selection (Li et al., 2020;VandeHaar et al., 2016) are commonly referenced as methods of improving feed efficiency of lactating dairy cows in recent literature; however, early life factors, such as in utero hyperthermia, may have long-term impacts on feed efficiency in post-pubertal dairy heifers (Tao et al., 2019).
When pregnant dams are exposed to severe heat stress in late gestation, the uterine environment also increases in temperature, creating thermal stress on the growing fetus.The fetus undergoes major developmental changes and exponential growth in the final stage of gestation; therefore, the consequences of an unfavorable in utero environment may impact organ development and body growth, cellular proliferation, immune function, and ultimately the performance of dairy heifers throughout the rearing period and subsequent lactations (Dahl et al., 2019;Skibiel et al., 2018;Dado-Senn et al., 2021;Laporta, 2021;Marrero et al., 2021).Specifically, heifers born to heat stressed dams are born earlier, weigh less at birth, and have a less efficient passive immunity (Davidson et al., 2021;Laporta et al., 2017;Monteiro, et al., 2016a).From birth to weaning, heifers born to cooled dams outperformed contemporaries that were born to heat stressed dams in starter intake and average daily gain (ADG; Laporta et al., 2017;Monteiro et al., 2016a).Heifers born to heat stressed dams exhibited reduced cellular proliferation in liver and mammary gland tissues (Dahl et al., 2019;Skibiel et al., 2018) which may induce changes in the composition of systemic metabolites.Until 1 year of age, heifers born to cooled dams were heavier and taller than heifers born to heat stressed dams but had similar ADG from birth to 1 year (Monteiro et al., 2016b).When heifers born to heat stressed dams reach maturity and begin lactation, they produce less milk in the first, second, and third lactations and have reduced productive lifespans, relative to heifers born to cooled dams (Laporta et al., 2020;Monteiro et al., 2016b).Although previous research has demonstrated the negative consequences of in utero heat stress on pre-weaned calves and lactating cows, little is known about the effects of heat-stress in late gestation on the subsequent growth trajectory, feed efficiency, and enteric methane production of post-pubertal heifers.However, changes in the rumen microbial community have been observed in dairy heifers and lactating cows when directly exposed to heat stress (Uyeno et al., 2009;Chen et al., 2018), suggesting the possibility of long-term effects of hyperthermia on the rumen ecosystem.
In the current study, our objectives were: 1) investigate the long-term impacts of in utero hyperthermia on heifer post-pubertal growth and feed efficiency, 2) examine possible impacts of in utero environment on enteric methane emissions, and 3) evaluate differences in systemic metabolite composition between in utero treatments.We hypothesized that dairy heifers born to cooled dams during late gestation would exhibit improved growth, performance, and feed efficiency compared with contemporary heifers born to heat stressed dams.Further, systemic metabolites associated with feed efficiency and in utero environment will be identified and serve as candidates for future exploration.

MATERIALS AND METHODS
All procedures were approved by the Institutional Animal Care and Use Committee of the College of Agricultural and Life Sciences at the University of Wisconsin-Madison (protocol #A006444).

Late gestation in utero treatments
A detailed description of the design and methodology of the late gestation heat stress experiment is provided in Dado-Senn et al. (2020).Briefly, from August to December of 2020, 82 multiparous pregnant Holstein dams were randomly assigned to either heat stressed (THI >68, with access to natural ventilation and shade of a free stall barn; n = 41) or cooled (THI >68, with access to shade of a free stall barn and the addition of active cooling via fans and water soakers; n = 41) treatments during the dry period (54 ± 5 d).Cows were housed at a commercial farm in Florida in adjacent compost-bedded free-stall pens.When the ambient temperature reached or exceeded 22.2°C, water soakers were activated every 5 min for 1.5-min intervals, and fans (Typhoon Power P-51, Seneca Dairy Systems, NY) ran continuously regardless of ambient temperature.
In that study, 36 heifer calves were born to heat stressed dams (in utero heat stressed; IUHT) and 38 heifer calves were born to cooled dams (in utero cooled; IUCL).Heifers were managed identically from birth to weaning at a University of Florida animal facility in individual sand-bedded pens with shade and fans for added ventilation.were euthanized to investigate organ development, and one IUCL heifer was excluded from further data collection due to a congenital condition unrelated to in utero treatment.After weaning, 41 heifers (n = 21 IUCL; n = 20 IUHT) were comingled and housed in groups under identical environmental conditions and until their relocation to the University of Wisconsin-Madison in December 2020, where they were quarantined at the Arlington Agricultural Research Station (Arlington, WI) and subsequently moved to the Marshfield Agricultural Research Station (Marshfield, WI) at 14 to 15 mo of age, where they were housed in pens of 8 and fed a typical heifer ration until the present study was initiated.Before the study, heifers were fed a TMR consisting of 84.2% haylage, 10.5% corn silage, 4.4% liquid whey permeate, 0.13% urea, and 0.8% mineral/vitamin supplement (14.1% CP, 44.1% aNDFom, 6.43% starch, and 2.35 Mcal ME/kg DM).

Growth, feed efficiency, and enteric methane emissions
In the current study, 2 heifers (IUHT) were excluded due to terminal health complications unrelated to in utero treatment.Thirty-eight heifers (569 ± 6.6 d of age; n = 21 IUCL; n = 17 IUHT) were housed at the Marshfield Agricultural Research Station in 3 free stall pens with 16 mattress beds per pen.The average days carried calf (DCC) for pregnant heifers (mean ± SD) were 85.6 ± 42.3 for IUCL and 84.8 ± 38.1 for IUHT.Three IUCL heifers and one IUHT heifer were not pregnant but were included in the present study.Within each in utero treatment, heifers were blocked by initial BW into 3 blocks (mean ± SD; blocks 1 to 3 = 586.4 ± 18.9, 541.6 ± 15.5, 511.7 ± 14.8, respectively) and assigned to pens (1 pen/block) resulting in 12 to 13 heifers per pen, with approximately half from each treatment group.Heifers were assigned randomly to Calan gates (American Calan; Northwood, NH), with one specific gate per heifer.The gates were accessed via neck collar transponders, thereby allowing measurement of individual feed intakes.Heifers were successfully trained to use their specific Calan gates after one week of acclimation to the diet and Calan gates before initiating the 63-d measurement period.

Body measurements and feed efficiency phenotypes
Initial and final BW were measured for 3 consecutive days (days −2, −1, 0, and 61, 62, 63 of the experiment) with means calculated across the 3 d.Midpoint BW was calculated as the mean of initial and final weights, and midpoint metabolic body weight (MBW) was calculated as BW 0.75 .Hip width, chest girth, and body length measurements were performed on d 1 and 63.Body condition score (BCS) was determined by 2 different qualified personnel on a 5-point scale with half-point increments (1 = emaciated to 5 = extremely obese; Wildman et al., 1982).The average initial BW was subtracted from the average final BW and divided by 63 d to calculate ADG.Feed conversion efficiency was determined by dividing the average DMI over the study period by ADG.Residual feed intake was calculated as the residual of the regression of DMI on midpoint MBW and ADG.The first-stage model is presented as: Where y i is the daily DMI of the ith heifer, β 0 is the intercept, β 1 is the partial regression coefficient of DMI on midpoint MBW, β 2 is the partial regression coefficient of DMI on ADG, and e i is the random error term that represents RFI.Because RFI is calculated as the observed DMI minus the expected DMI, an animal with observed intake less than the modeled expression is represented by a negative RFI value and considered as more efficient.

Diet, fecal sampling, and digestibility
All heifers received a TMR comprised of 46.6% oatlage, 44.6% grass/alfalfa haylage, 7.7% corn silage, 0.3% urea, and 0.8% mineral on a DM basis (Table 1).At 1000h daily, refusals were removed from the Calan gates and weighed, and a fresh TMR was delivered.Diet components were sampled weekly, and daily TMR and refusal samples were collected and composited by week.All samples were dried in a forced air oven at 55°C for 48 h to determine dry matter.Samples were then ground to pass a 1 mm screen (Wiley Mill, Arthur H. Thomas, Philadelphia, PA) and analyzed by a commercial laboratory for ash (ashing at 600°C for 2 h), neutral detergent fiber using α-amylase and sodiumsulfite (aNDF; Van Soest et al., 1991) and corrected for residual ash (aNDFom), in vitro NDF digestibility after 48h incubation (Goering and Van Soest, 1970), starch (Hall et al., 2015), and minerals via inductively coupled plasma analysis of samples digested with 50% nitric acid and diluted (Dairyland Labs, Arcadia, WI; Table 1), and energy concentrations calculated according to NASEM (2001).
Individual TMR and fecal samples were taken on 3 consecutive days for each block (blocks 1 to 3 on d 35 to 37, 42 to 44, and 49 to 51, respectively) to determine nutrient digestibility.To obtain a general diurnal pattern for each heifer, fecal samples were taken at 3 different time points (0800, 1200, and 1600 h) over the 3 d and composited by heifer.The TMR, ort, and fecal samples were dried at 55°C for 48 h in a forced air oven, ground to pass a 1 mm screen (Wiley Mill, Arthur H Thomas) and analyzed for ash (combusted at 500°C for 6h), NDF with α-amylase and sodium sulfite (ANKOM Technology, Method 6, 2017), and undigested NDF after 240h in vitro digestion (ANKOM Technology, Method 3, 2017).

Enteric greenhouse gas emissions and GreenFeed visits
To measure enteric methane (CH 4 ) and carbon dioxide (CO 2 ) emissions of individual animals, a GreenFeed unit (C-Lock; Rapid City, SD) was accessible to each pen for 8 ± 1 d (blocks 1 to 3 on d 30 to 38, 39 to 47, and 47 to 55, respectively) with no acclimation period or training.The order of pen access to the GreenFeed was recorded to account for possible bias in later pens due to visual familiarity with the GreenFeed.When an animal approaches, the GreenFeed senses head proximity, reads a radio frequency identification tag and activates a fan that pulls air and exhalant through the sampling hood and incorporated gas measurement equipment.Heifers voluntarily visited the GreenFeed a maximum of 7 times per day, with a 2-h lockout period between visits, and were encouraged to visit by providing up to 6 drops of 35g of alfalfa pellets (DM 95.2%, NDF 42.0%, CP 18.7%, TDN 56.8% on DM basis) in 40 s intervals at each visit.The GreenFeed unit was calibrated each time it was moved to a new pen.Production of methane (g/d) and carbon dioxide (g/d) were calculated as 8-d averages (g/d).Methane production was divided by ADG and DMI to calculate methane intensity (g/kg ADG) and methane yield (g/ kg DMI), respectively.Only animals that visited the GreenFeed (n = 17 IUCL; n = 15 IUHT) were included in the analysis of methane production, carbon dioxide production, methane yield, and methane intensity.To investigate possible cognitive advantages of IUCL over IUHT heifers, the number of visits to the GreenFeed was averaged over the measurement period for each heifer to calculate visits per day, and heifers that did not visit the GreenFeed were assigned a value of zero visits.Latency to visit the GreenFeed was determined as the difference between the time at a heifer's first visit and the time at which the GreenFeed was made available in the respective pen.Six heifers (n = 2 IUHT; n = 4 IUCL) did not visit the GreenFeed and were assigned the maximum latency value of 216 h (9 d); the next longest observation was 194.4 h.

Blood serum metabolites
Blood was sampled via coccygeal venipuncture using serum vacutainer blood tubes (BD #366430; Franklin Lakes, NJ) on d 37, 44, and 51 for blocks 1, 2, and 3, respectively.Blood samples were allowed to clot for 20 min at room temperature and centrifuged at 3000 X g for 20 min to harvest serum fraction, which was aliquoted into 1.5 mL propylene tubes and stored at −80°C until gas chromatography time-of-flight mass spectrometry (GC-ToF/MS) analysis was performed by the UC-Davis West Coast Metabolomics Center (Davis, CA).Data were normalized using the sum of all peak heights for all identified metabolites (mTIC) and the median of the samples, then auto scaled to achieve normality.MetaboAnalyst version 5.0 (Pang et al., 2021) was used for normalization and identification of differentially expressed metabolites and their respective biological pathways.

Statistical analysis
R version 4.1.1(R Core Team, 2021) statistical software was used to analyze data with packages lme4, lmerTest, and emmeans.This randomized block design experiment had 3 weight blocks and animal as the experimental unit.Days carried calf was considered as a covariate but deemed unnecessary via sequential model selection based on Akaike information criterion.Simple linear regressions with treatment and block as fixed effects were used to analyze initial, change, and final growth data.The DMI model included treatment, ADG, and block as fixed effects.Treatment, mid-point BW, and block were included as fixed effects when modeling ADG.Similarly, the model for FCE included treatment and block as fixed effects.
RFI was analyzed as: Where y ijk is the calculated RFI phenotype for a given heifer; β 0 is the intercept; Trt i is the effect of in utero treatment; Block j is the effect of block; and e ijk is the random error.Methane production and carbon dioxide were modeled as: Where y ijk is the methane production or carbon dioxide production average for a given heifer; β 0 is the intercept; β 1 is the partial regression coefficient of methane production or carbon dioxide production on average daily DMI; β 2 is the partial regression coefficient of methane production or carbon dioxide production on midpoint BW; Trt i is the fixed effect of in utero treatment; Block j is the fixed effect of block; and e ijk is the random error.Fixed effects of treatment and block were used when analyzing methane intensity and methane yield, and regressions on DMI and BW were excluded.Treatment and order of GreenFeed access were included as fixed effects when modeling the number of visits.A Kaplan-Meier survival model was used to analyze latency data with treatment and order of exposure as fixed effects.
Correlations of differentially expressed metabolites with methane production, RFI, and NDF digestibility were assessed using the corr package in R. Supplemental material (https: / / data .mendeley.com/datasets/ spdx3rspdt/ 1) includes correlation coefficients and P-values for all identified metabolites, methane production, RFI, and NDF digestibility.Significance was established at P < 0.05 and tendency at P < 0.10.

Heifer growth and feed efficiency
In utero heat stressed and IUCL heifers did not differ in initial or final BW, hip width, chest girth, or BCS (P ≥ 0.50, Table 2); however, while initial body length was not different (P = 0.11), IUCL heifers were longer in final body length than IUHT heifers (P = 0.02, Table 2).Similarly, BW, hip width, chest girth, body length, and BCS change during the 63-d study period did not differ between IUHT and IUCL heifers (P ≥ 0.29, Table 2).During the 63-d study period, the ADG and daily DMI were similar between IUHT and IUCL heifers (P ≥ 0.44, Table 2).Feed efficiency phenotypes, namely FCE and RFI, were also similar between IUCL and IUHT heifers (P ≥ 0.18, Table 2).To illustrate the concept of RFI, Figure 1 demonstrates that less efficient animals (positive RFI), for which observed intake is more than expected, are plotted above the line while more efficient animals (negative RFI), for which observed intake is less than expected, are plotted below the line.

Enteric greenhouse gas emissions and GreenFeed visit behavior
Table 3 includes results from the analysis of emissions data from IUCL (n = 17) and IUHT (n = 15) heifers that visited the GreenFeed.There were no differences between IUHT and IUCL heifers in methane production, carbon dioxide production, methane yield, or methane intensity (P ≥ 0.47, Table 3).Visits per day were also similar between IUHT and IUCL heifers (P = 0.39, Table 3).Latency to approach the GreenFeed for the first time did not differ by treatment, and 50% of heifers visited the GreenFeed within approximately 3 d of access (Figure 2).The shortest latency among IUCL heifers was 3 min and 19 s, and the shortest latency among IUHT heifers was 30 min and 51 s.

Nutrient intake, fecal output, and digestibility
Results from digestibility analysis are presented in Table 4.In utero heat stressed and IUCL heifers had similar intake, output, and digestibility of dry matter (DM), organic matter (OM), and NDF (P ≥ 0.45, Table 4).

Blood serum metabolites
Correlations and P-values of all 143 identified metabolites with RFI, methane production, and NDF digestibility can be found in Supplemental Table 1 (https: / / data .mendeley.com/datasets/ spdx3rspdt/ 1).Of the 143 identified metabolites, 7 were significantly correlated with methane production, 9 were significantly correlated with residual feed intake, and 7 were significantly correlated with NDF digestibility (Supplemental Table 1).Six of the 143 identified metabolites were significantly differentially expressed between IUCL and IUHT heifers (P-value <0.048,Table 5).Least squares means of normalized, differentially expressed metabolites are shown in Table 5.Of the metabolites differen-   tially expressed by treatment, a moderate positive correlation was found between inositol-4-monophosphate concentration and RFI (P = 0.03, Figure 3A), whereas inositol-4-monophosphate concentration and NDF digestibility tended to be negatively correlated (P = 0.09, Figure 3B), and glycerol-α-phosphate concentration was found to be negatively correlated with NDF digestibility (P = 0.01, Figure 3C).Impact factors and significance levels of differentially expressed metabolites on respective pathways are presented in Figure 4.In the pathway analysis, 3 out of the 6 differentially expressed metabolites were associated with 6 pathways: purine metabolism, glycerophospholipid metabolism, inositol phosphate metabolism, phosphatidylinositol signaling system, glycerolipid metabolism, and arginine biosynthesis.Urea is involved in purine metabolism and arginine biosynthesis; however, because urea is a byproduct of these pathways, it had an impact factor of zero (Figure 4).

DISCUSSION
To optimize performance and ensure animal welfare on dairy farms, it is important to investigate how in utero heat stress, a major environmental stressor, may influence the growth, performance, and environmental sustainability of heifers later in life.Previous studies have shown that heifers born to heat stressed dams, including the heifers used in the present study, are smaller at birth, have weaker passive immunity, decreased grain starter intake, and lower ADG from birth to weaning (Davidson et al., 2021;Laporta et al., 2017;Monteiro et al., 2016a), a result that was confirmed in the heifers used in the present study earlier in life (Dado-Senn et al., 2021).However, few studies have followed the performance of IUHT and IUCL heifers beyond weaning.The present study is the first to assess the long-term impacts of disparate in utero thermal environments during late-gestation in post-pubertal dairy heifers, more than 1.5 years after the initial prenatal insult.Monteiro et al. (2016b) gathered retrospective growth data from IUHT and IUCL heifers until one year of age and revisited their performance at parturition.The authors reported that IUCL heifers were taller and heavier than IUHT heifers at one year of age, with similar ADG, but no differences in BW were observed at parturition.In the present study, which occurred at 18 mo of age there were few residual effects of the in utero treatments on heifer growth, feed efficiency, and methane emssions.Final body length was the only significant difference observed; IUCL heifers were significantly longer than IUHT heifers at the end of the trial.This difference in body length between IUHT and IUCL heifers was also observed at birth (Dado-Senn et al., 2021).The lack of statistical differences in BW, chest girth, hip width, and ADG on the current study appears to be consistent with conclusion of Monteiro, et al. (2016b) that IUHT heifers may undergo compensatory growth between one year of age and first calving.If compensatory gain occurs in IUHT heifers, our results suggest it occurs before 18 mo of age.While preweaning ADG has been associated with improved milk production, excessive ADG postweaning may reduce future milk production (Zanton and Heinrichs, 2005).It is possible that compensatory postweaning gain in IUHT heifers may be associated with losses in future milk production, as the latter was observed in previous studies using our IUHT model (Minteiro et al., 2016;Skibiel et al., 2018;Laporta et al., 2020).
Previous research has identified management strategies and dietary manipulations that may improve feed efficiency in heifers.Hoffman et al. (2007) and Kruse et al. (2010) offered control or limit-fed diets to gravid heifers of similar age and weight to those in the current trial, and heifers fed ad libitum in the control group exhibited similar growth, DMI, ADG, and FCE as those reported in the current study.Furthermore, heifers fed diets diluted by high-fiber, low-energy forages exhibited similar ADG as those observed in the current trial (Coblentz et al., 2015;Williams et al., 2022).While precision feeding and diet manipulation in previous research successfully improved heifer feed efficiency, the present study did not detect differences in DMI, FCE, or RFI between IUHT and IUCL heifers, despite DMI differences of starter grain concentrate observed at 42 to 63 d of age by Dado-Senn et al. (2021).Under the conditions of the current study, IUHT and IUCL heifers performed equally well, indicating feed efficiency at 18 mo of age may not be a long-term carry-over consequence of IUHT.Differences in feed efficiency, nutrient digestibility, or methane emissions may be influenced by variation in rumen microbial communities.Low RFI cows were found to have improved NDF digestibility relative to high RFI cows in studies by Potts et al. (2017) andOlijhoek et al. (2018), who hypothesized that this difference may have been due to longer rumen retention time in low RFI cows.Rumen retention time and diet digestibility may also be impacted by heat stress (Bernabucci et al., 1999;Nonaka et al., 2008); however, there is no literature investigating the effects of IUHT on diet digestibility.In the current study, no differences were observed in DM, NDF, or OM digestibility, intake, or output between IUCL and IUHT heifers.Values for intake, output, and digestibility of DM, NDF, and OM observed in this study were consistent with previous studies that fed a similar control diet (Li et al., 2019;Williams et al., 2022).Future research involving rumen sampling, characterization of rumen microbes, and passage rate is needed to further unravel the potential impact of IUHT on diet digestibility.
Potential alterations to the rumen environment may suggest a relationship between heat stress and enteric methane emissions.Sheep under direct heat stress exhibited increased enteric methane emissions, perhaps as a result of poor diet digestibility, increased rumen retention time, or altered rumen environment (Mbanzamihigo et al., 2002;Ulyatt et al., 2002;Ulyatt et al., 2005).To date, no studies have investigated the effects of IUHT on enteric methane emissions.Several studies have suggested that improving feed efficiency can decrease methane emissions (Capper et al., 2009;Connor, 2015).In growing beef steers, more efficient animals emit 24 to 28% less daily methane production (Nkrumah et al., 2006).Ryan et al. (2022) evaluated GreenFeed visit behavior and methane emissions of 361 beef heifers of similar age and weight to those in the present study and reported similar values for methane production and methane intensity.In the present study, we found no differences in methane produciton, carbon dioxide production, methane yield, or methane intensity in IUCL heifers relative to IUHT heifers.Although our sample size was limited, similarities in methane emissions would be expected given the lack of differences in body weight, DMI, and DM digestibility between IUCL and IUHT heifers in the present study.
Latency to the first GreenFeed visit was calculated for each heifer as a measure of cognitive development and response to novel objects.A previous study indicated that IUCL heifers had larger head circumference than their IUHT contemporaries (Dado-Senn et al., 2021), and Shiota and Kayamura (1989) discovered in utero heat stressed mice were slower learners than control counterparts in cognitive tests.In the present study, IUCL and IUHT heifers did not differ in latency to approach the GreenFeed or number of GreenFeed visits.Approximately 50% of heifers visited the Green-Feed within 75 h of access and only 13% of heifers did not visit the GreenFeed, with no differences observed between treatments.In our previous study at the same facility, 50% of 18-mo-old heifers approached the GreenFeed within 8 h of access (Riesgraf et al., submitted).The difference in latency may be due to increased competition for the GreenFeed, as there were 12 to 13 heifers per pen in the current study and only 6 to 8 heifers per pen in the previous study.Further investigation is necessary to understand if latency response to other novel objects might be impacted in IUHT post-pubertal heifers.In addition, since visiting the GreenFeed is voluntary in free stall barns, more visits may be achieved with an acclimation period or extended measurement period (Manafiazar et al., 2017;Renand and Maupetit, 2016).
In addition to cognitive development, severe and chronic heat stress in utero for over 50 d may induce physiological and molecular cellular changes in the developing fetus, altering cellular and tissue function and ultimately compromising systemic metabolism (Skibiel et al., 2018).Published literature regarding the effects of heat stress on the blood metabolome is limited to lactating cows under direct exposure to heat stress (Joo et al., 2021), and little is known about the long-term impacts of IUHT on the blood metabolome of dairy heifers.Herein, we identified inositol-4-monophosphate and glycerol-3-phosphate as 2 differentially expressed metabolites between IUHT and IUCL heifers that were also correlated with RFI or NDF digestibiltiy.Inositol monophosphate is an intermediate in the inositol phosphate metabolism and phosphatidylinositol signaling system pathways.In human cells, inositol is essential for cell proliferation and viability, and while cells produce some inositol de novo, depletion of external inositol significantly handicapped cell proliferation (Suliman et al., 2022).Phosphorylated inositols are also involved in cell membranes as polyphosphoinositides, critical for many cell signaling responsibilities (Suliman et al., 2022).Mammalian cells exposed to direct severe heat-shock undergo protein denaturation and interruption of plasma membranes, impeding cell proliferation and compromising cell viability (Balogh et al., 2013).Animals exposed to IUHT exhibited long-term reduced cellular proliferation in multiple tissues, including liver and mammary glands, compared with their IUCL counterparts (Skibiel et al., 2018).The current study found decreased systemic concentrations of inositol in IUHT heifers, which may be a residual effect of suboptimal in utero metabolism, possibly explaining the negative impact on cell proliferation in these heifers.
The moderate, positive correlation between RFI and inositol-4-monophosphate may be explained by an increase in DMI with higher RFI, therefore providing more substrate for the synthesis of inositol.The tendency for Blood samples for metabolite analysis were collected on d 37, 44, and 51 for blocks 1, 2, and 3, respectively, of the study.Bars represent the impact factor of associated metabolite on the respective pathway.Black dots represent the significance of the pathway analysis as -log 10 P-values.The impact factor is calculated as the sum of importance attributed to statistically significant metabolites normalized by the sum of importance of all metabolites in each pathway.Purine metabolism and arginine biosynthesis have an impact factor of zero because urea, the involved metabolite, is a byproduct of the pathway.a negative correlation between NDF digestibility and inositol-4-monophosphate has not been reported previously; however, it is possible that increased circulation of inositol is synthesized de novo to compensate for depleted dietary precursors.Furthermore, glycerolipid and glycerophospholipid metabolism may be impacted by in utero environment.In the current study, it was found that serum from IUHT heifers had a higher concentration of glycerol-α-phosphate, also called glycerol-3-phosphate, which is an important intermediate in the creation of triglycerides and complex lipids (Prentki and Madiraju, 2008).In monogastrics, glycerol-3-phosphate is derived from an intermediate of glycolysis; however, glycolysis in ruminants is limited, as the majority of glucose is utilized by rumen microbes to produce volatile fatty acids, such as acetate and propionate.Propionate is then used to make glucose via the citric acid cycle and gluconeogenesis, with glucose-3-phosphate as an intermediate (Aschenbach et al., 2010).Increased concentrations of glucose-3-phosphate in IUHT relative to IUCL heifers may result from increased lipolysis in IUHT heifers or potential differences in microbial production of propionate.The later hypothesis may provide a better explanation for the negative correlation between NDF digestibility and glycerol-α-phosphate; as NDF becomes less digestible, fatty acids are catabolized for energy.Within the limitations of this study, we can only speculate on the downstream impact of the differentially expressed metabolites.Further research is necessary to determine the long-term carry-over consequences of the metabolomic differences observed between IUHT and IUCL heifers.

CONCLUSIONS
Despite the well-established effects of in utero heat stress on feed intake, growth performance, and physiology of young calves, we did not observe differences in feed intake, feed efficiency, methane emissions, or diet digestibility between IUHT or IUCL heifers during their post-pubertal growing phase.Final body length was greater for IUCL heifers at the end of the study, but other growth parameters did not differ at the beginning or end of the study, nor did DMI and ADG differ during the study.Because previous research has established substantial differences in BW, ADG, and DMI during the preweaning phase, our results suggest IUHT heifers undergo compensatory growth after weaning.We uncovered metabolites primarily involved in lipid metabolism that were differentially expressed.Additional research is necessary to interpret the relationships between in utero environment and subsequent long-term alterations to the blood metabolome.Although in utero heat-stress does not seem to cause significant differences in feed efficiency between IUHT and IUCL heifers later in life, heat abatement remains a crucial feature of cattle comfort and welfare.
Riesgraf et al.: LONG-TERM CONSEQUENCES OF IN UTERO HYPERTHERMIA Riesgraf et al.: LONG-TERM CONSEQUENCES OF IN UTERO HYPERTHERMIA Riesgraf et al.: LONG-TERM CONSEQUENCES OF IN UTERO HYPERTHERMIA

Figure 1 .
Figure 1.Residual feed intake (RFI) from heifers (18 ± 1 mo of age) exposed to heat stress in utero (IUHT, n = 17 red squares, ■) or cooled in utero (IUCL, n = 21 blue circle, •).In utero heat-stress heifers were born from dams under environmental heat-stress during the last 56-d of gestation, while IUCL heifers were born from dams under environmental heat-stress but actively cooled via access to fans and water soakers.The RFI did not differ between IUHT or IUCL treatments.Heifers are considered more efficient if below the line of identity and less efficient if above the line of identity (dotted line).
Riesgraf et al.: LONG-TERM CONSEQUENCES OF IN UTERO HYPERTHERMIA

Figure 2 .
Figure 2. Latency to approach the GreenFeed from heifers born from environmentally heat stressed (IUHT, n = 17, red line) and actively cooled (IUCL, n = 21, blue line) dams during the last 56 d of gestation.In utero heat stressed and IUCL heifers visited the GreenFeed unit for the first time equally as quickly.Heifers that did not visit the GreenFeed were assigned a latency value of 216 h (9 d).Dotted red and blue lines represent 95% confidence intervals for IUHT and IUCL heifers, respectively.
Riesgraf et al.: LONG-TERM CONSEQUENCES OF IN UTERO HYPERTHERMIA

Figure 3 .
Figure 3. Significant correlations of (A) inositol-4-monophosphate with residual feed intake (RFI), (B) inositol-4-monophosphate with NDF digestibility, and (C) glycerol-α-phosphate with NDF digestibility.Red squares represent heifers born from dams under environmental heat-stress (IUHT, n = 17), while blue circles represent heifers born from dams under environmental heat-stress but actively cooled via access to fans and water soakers (IUCL, n = 21) during the last 56 d of gestation.Blood samples for metabolite analysis were taken on d 37, 44, and 51 for blocks 1, 2, and 3, respectively, of the study.Correlation coefficients (r) and P-values are depicted in each graph for IUHT (red) and IUCL (blue) groups and regardless of treatment (black).
Figure 4. Metabolic pathways associated with 6 differentially expressed metabolites identified between heifers born from dams under environmental heat-stress (IUHT, n = 17) or heifers born from dams under environmental heat-stress but actively cooled via access to fans and water soakers (IUCL, n = 21) during the last 56 d of gestation.Blood samples for metabolite analysis were collected on d 37, 44, and 51 for blocks 1, 2, and 3, respectively, of the study.Bars represent the impact factor of associated metabolite on the respective pathway.Black dots represent the significance of the pathway analysis as -log 10 P-values.The impact factor is calculated as the sum of importance attributed to statistically significant metabolites normalized by the sum of importance of all metabolites in each pathway.Purine metabolism and arginine biosynthesis have an impact factor of zero because urea, the involved metabolite, is a byproduct of the pathway.
All heifers were fed 0.87 kg/d DM milk replacer (UF Special 28/20 Bova Medicated, Southeast Milk, Okeechobee, FL) until the start of weaning at 49 d of age, when calves were fed 0.23 kg/d DM milk replacer for one week until fully weaned at 56 d of age.Ad libitum calf starter (Ampli-Calf Starter 20 Warm Weather, Purina Animal Nutrition LLC, Shoreview, MN) was provided starting at 4 d of age.At birth and weaning, 16 heifers from each treatment Riesgraf et al.: LONG-TERM CONSEQUENCES OF IN UTERO HYPERTHERMIA

Table 1 .
Riesgraf et al.: LONG-TERM CONSEQUENCES OF IN UTERO HYPERTHERMIA Ingredient composition of diet and mean nutrient composition of TMR and individual diet components based on weekly analysis 1Based on NASEM (2001) calculations.

Table 2 .
Growth performance and feed efficiency of post-pubertal heifers (18 ± 1 mo of age) that were born to heat stressed (IUHT) or cooled (IUCL) dams during the last 56 d of gestation (LSM ± SEM) 1 IUHT = heifers were born from dams under environmental heat stressed during the last 56 d of gestation, IUCL = heifers were born from dams under environmental heat stressed but actively cooled via access to fans and water soakers.

Table 3 .
Enteric gas emissions from GreenFeed measurements of post-pubertal heifers (18 ± 1 mo of age) that were born to heat stressed (IUHT) or cooled (IUCL) dams during the last 56 d of gestation (LSM ± SEM) All animals, including those that did not visit the GreenFeed, were included in the calculation of visits per day and assigned a value of zero if they did not visit (n = 21 IUCL; n = 17 IUHT). 2

Table 4 .
Nutrient intake, fecal output, and digestibility of postpubertal heifers (18 ± 1 mo of age) that were born to heat stressed (IUHT) or cooled (IUCL) dams during the last 56 d of gestation (LSM ± SEM)

Table 5 .
Differentially expressed metabolites of post-pubertal heifers (18 ± 1 mo of age) that were born to heat stressed (IUHT) or cooled (IUCL) dams during the last 56 d of gestation (LSM ± SEM) 1 IUHT = heifers were born from dams under environmental heat stressed during the last 56 d of gestation, IUCL = heifers were born from dams under environmental heat stressed but actively cooled via access to fans and water soakers.