se-Exploring the microbial composition of Holstein Friesian and Belgian Blue colostrum in relation to the transfer of passive immunity

For centuries, multicellular organisms have lived in symbiosis with microorganisms. The interaction with microorganisms has been shown to be very beneficial for humans and animals. During a natural birth, the initial inoculation with bacteria occurs when the neonate passes through the birth canal. Colostrum and milk intake are associated with the acquisition of a healthy gut flora. However, little is known about the microbial composition of bovine colostrum and the possible beneficial effects for the neonatal calf. In this prospective cohort study, the microbial composition of first-milking colostrum was analyzed in 62 Holstein Friesian (HF) and 46 Belgian Blue (BB) cows by performing amplicon sequencing of the bacterial V3–V4 region of the 16S rRNA gene. Calves received, 3 times, 2 L of their dam’s colostrum within 24 h after birth. Associations between colostral microbial composition and its IgG concentration, as well as each calf’s serum IgG levels, were analyzed. Colostrum samples were dominated by the phyla Proteobacteria , Firmicutes , Bacteroidetes , and Actinobacteria . The 10 most abundant genera in the complete data set were Acinetobacter (16.2%), Pseudomonas (15.1%), a genus belonging to the Enterobacteriaceae family (4.9%), Lactococcus (4.0%), Chry-seobacterium (3.9%), Staphylococcus (3.6%), Proteus (1.9%), Streptococcus (1.8%), Enterococcus (1.7%), and Enhydrobacter (1.5%). The remaining genera (other than these top 10) accounted for 36.5% of the counts, and another 8.7% were unidentified. Bacterial diversity differed significantly between HF and BB samples. Within each breed, several genera were found to be differentially abundant between colostrum of different quality. Moreover, in HF, the bacterial composition of


INTRODUCTION
The lactation starts around parturition with colostrum production (Lamming, 1994).Colostrum, or "the golden liquid," has a unique composition that reflects the different needs of the newborn.One of these needs is passive immunity, because the bovine placenta limits the transfer of immunoglobulins to the fetus (Peter, 2013).In other words, calves are born without circulating antibodies (Godden, 2008).Colostrum ingestion is a way of passively transferring immunity from mother to dam and providing the calf with antibodies, especially IgG, to secure protection during the first challenging months of life (Chase et al., 2008).These antibodies are absorbed in the calf's intestines and transported into the bloodstream for only a short period, approximately the first 24 h after birth (Bush and Staley, 1980).Due to this time limitation, newborn calves need to ingest a sufficient amount of good quality colostrum as soon as possible after birth.In this context, colostrum quality is defined by its IgG concentration, and is generally accepted to be of good quality if it contains ≥50 g/L IgG (McGuirk and Collins, 2004).Supplying the calf with 4 L of good quality colostrum within 6 h after birth should lead to a sufficient serum IgG level of at least 10 g/L.If this minimum level is not reached, failure of passive transfer (FPT) occurs (McGuirk and Collins, 2004).Calves suffering from FPT are more at risk of becoming sick or even dying in the first months of life (Godden, 2008).Although farmers are paying increasing attention to colostrum management, the prevalence of FPT is still very high (estimated between 20 and 40%; Raboisson et al., 2016).Because the se-rum IgG concentration is highly correlated with the amount of antibodies administered in the first 24 h of life, insufficient colostrum management (late administration or administration of poor quality colostrum) is the most important cause of FPT (Raboisson et al., 2016).In contrast, even though the colostrum protocol implemented at the Flanders Research Institute for Agriculture, Fisheries and Food (ILVO) cattle research farm was standardized and the quality monitored using a Brix refractometer, we observed wide variation in IgG absorption between calves that received equal and sufficient amounts of antibodies.Moreover, despite our efforts, 10% of our calves (n = 13 out of 134) suffered from FPT (our unpublished results).In addition to antibodies, colostrum harbors many other important nutrients and bioactive components, such as growth factors, hormones, vitamins, minerals, microorganisms, and even immune cells and microRNAs (Meganck et al., 2014;Van Hese et al., 2020).It has been hypothesized that several of these factors contribute to the transfer of passive immunity.The potential of these components to improve calf health has been shown in previous research.For example, insulin growth factors were shown to stimulate the growth and development of the gastrointestinal tract and other organs (Georgiev, 2008).In the present study, we were particularly interested in the microorganisms present in colostrum and how they contribute to the development and health of the calf.
Bovine colostrum contains a rich microbial population, which is passed on from mother to neonate through colostral ingestion (Lima et al., 2017;Yeoman et al., 2018).It is well accepted that breast milk has a beneficial effect on the colonization of the neonate's gut (Walker and Iyengar, 2015).Moreover, several bacteria present in a woman's colostrum and milk are associated with the gut health and immune system development of her neonate (Toscano et al., 2017).It has been hypothesized that these beneficial effects of mother's milk on the neonate are applicable to the newborn calf as well (Addis et al., 2016).Multiple studies have researched the association between colostrum feeding and the calf's intestinal colonization (Malmuthuge et al., 2015;Klein-Jöbstl et al., 2019;Hang et al., 2021;Song et al., 2021); however, few studies have focused on the possible benefits for the newborn calf's immunity.
The main goal of the present study was to define and compare the bacterial composition of first-milking colostrum collected from healthy HF and BB cows, the 2 most popular breeds in the Belgian cattle industry.The second objective was to study the effect of cattle management parameters around parturition on the microbial composition of colostrum, the transfer of passive immunity, and health of the calf.Finally, we determined whether specific microbial genera were associated with colostrum quality and IgG absorption in the calf.

Ethics Statement
The experimental protocol was approved by the ILVO ethical committee (case number EC2018/313).The methods were carried out in accordance with approved guidelines.

Animals, Feed, and Housing
The present study was conducted at the research farm of the ILVO between December 2017 and June 2019.During this period, the ILVO herd included 253 dairy heifers (minimum 1 yr of age) and lactating cows and 143 beef heifers (minimum 1 yr of age) and cows.Cows and heifers were enrolled in the trial only once and only if first-milking colostrum reached a quality of at least 21% Brix.Cows that had twin births or failed to produce colostrum (or <2 L) at first milking were excluded.Finally, samples of first colostrum from 62 Holstein Friesian (HF) cows, of which 33 were primiparous, and 46 Belgian Blue (BB) cows, of which 11 were primiparous, were analyzed.Calves that could not receive their first meal of colostrum from their own dam (3 BB calves) were excluded from the analysis of calf parameters [serum IgG level and apparent efficiency of absorption (AEA)].Another 2 BB calves were excluded due to missing birthweight records.
Dried-off HF cows were group housed in freestall barns with slatted floors and cubicles covered with rubber mattresses, separated from the lactating herd.The HF cows were dried off approximately 45 d before expected calving and treated with intramammary antibiotics (Virbactan DC, Virbac) and a teat-sealer (OrbeSeal, Zoetis).They were fed a far-off ration based on corn silage (64% of DM content) mixed with straw (29% of DM content) with final CP, crude fiber, and starch levels of 83, 238, and 222 g/kg of DM, respectively.At 14 d before expected calving, cows switched from the far-off to the close-up diet.The close-up ration contained corn silage (34% of DM content), grass silage (26% of DM content), and by-products (28% of DM content) with final CP, crude fiber, and starch levels of 144, 1,815, and 182 g/kg of DM, respectively.Roughage components varied during the trial due to usage of different silages; however, nutrient supply was always balanced to optimally meet the animals' needs.Both rations were supplemented with minerals and vitamins according to CVB recommendations (CVB, 2016).   1 shows the detailed composition and nutritional value of the far-off and close-up rations.Primiparous cows were transferred to the lactating herd 1 mo before expected calving to adapt to the new environment and the milking robot.During this time, heifers received the close-up diet and 1 kg of balanced concentrate when visiting the milking robot.Starting at approximately 5 d before expected calving, body temperature was monitored twice a day.If the cows showed a decline of 0.5°C in temperature in combination with external signs of impending calving (relaxation of the pelvic ligaments and strutting of the teats), they were moved to maternity pens with deep straw bedding.Calving ease was estimated based on the observation and scoring of the animal caretaker and divided in 4 categories: alone (no help needed), normal (minimal help needed), hard (heavy extraction force needed), and unknown.The last category was assigned if no record of calving ease was available.
Primi-and multiparous BB animals were either kept on pasture (grazing group) or in group in pens with deep straw bedding (no grazing group) in separated groups.Grazing was allowed from the beginning of April until mid-November.Cows in the grazing group were kept on pasture without additional feeding and transferred to deep straw bedding pens 1 mo before expected calving.From that time, they received ad libitum a mixed ration with corn silage (40% of DM content), grass silage (40% of DM content), and straw (20% of DM content) supplemented with 0.5 kg of soybean meal and 0.25 kg of concentrates with minerals.Table 1 shows the detailed composition of the ration provided to the BB cows.If body temperature decreased 0.5°C in combination with external signs of impending calving (relaxation of the pelvic ligaments and strutting of the teats), BB cows were moved to a tiestall barn, awaiting elective cesarean section.All calves, both HF and BB, were transferred to individual pens (indoor) immediately after birth and fed colostrum according to the standard procedure described below.After their third meal of colostrum, female calves were housed in individual hutches (outdoor) until 16 wk of age, and male calves were housed in a separate stable in individual pens (indoor) until they left the farm (at approximately 2 wk of age).

Collection of Colostrum Samples
Colostrum samples were taken within 1 h after parturition.Before sample collection, the udders were cleaned with a dry cloth and the teats were wiped with a 70% ethyl alcohol gauze.Cows were milked with a portable milking machine.Before milking, the milk bucket was washed with hot water (>70°C).After each usage, the teat cups and tubes were cleaned using the milking parlor's cleaning system, and the milk bucket was washed with hot water and an acid descaler (Parlor Cleaner, DeLaval BV).Colostrum samples were collected straight from the milk bucket (3 sterile centrifuge tubes of 50 mL each) to obtain a representative sample of the colostrum that was administered to the calf.Colostrum was then aliquoted in microcentrifuge tubes of 2.0 mL and stored at −80°C until microbial DNA extraction and radial immunodiffusion (RID) analysis.

Colostrum IgG Concentration Measurement
Colostral IgG concentration was measured using a commercial RID assay (Bovine IgG RID kit, Triple J Farms).Colostrum samples were thawed at room temperature (20 to 24°C), thoroughly mixed, diluted (1:1) with a saline solution (0.9% NaCl), and again mixed well.Following the manufacturer's instructions, each well of the RID test plate was filled with 5 µL of the diluted colostrum samples.Colostrum samples were tested alongside the manufacturer's reference sera containing 180, 1,472, and 2,803 mg/dL IgG, respectively.The RID plates were incubated for 24 h at room temperature, and the precipitating ring diameter was measured with a stereomicroscope (Olympus SZX7, Olympus Corp.) and an ocular micrometer (magnification 10×).For further statistical analysis, colostral IgG concentration was measured with RID.

Colostrum Administration and Serum Collection
Calves were bottle-fed a total of 6 L of colostrum from their own dam in the first 24 h of life (3 feedings

DNA Extraction and 16S rRNA Gene Sequencing
Microbial DNA was extracted from 1.5 mL of firstmilking colostrum with the Powerfood Microbial kit (Qiagen) following the manufacturer's protocol.Samples were thawed at room temperature and centrifuged twice at 13,000 × g for 1 min at room temperature (20°C) to remove fat and supernatant to obtain a concentrated microbial pellet.After DNA extraction, an additional clean-up was performed on the DNA samples with the NucleoSpin gDNA Clean-up kit (Macherey-Nagel GmBH & Co. KG).DNA concentration was measured with the Quantus Fluorometer (Promega), and absorption ratios of 260 nm: 280 nm (A260/280) and 260 nm: 230 nm (A260/230) were measured with the NanoPhotometer N50 (Implen).Library preparation (including quality control) was performed for the V3-V4 region of the 16S rRNA gene using primers 344F (CCTACGGGNGGCWGCAG) and 806R (GAC-TACHVGGGTATCTAATCC) following the Illumina protocol (Illumina Inc.) by Macrogen.Indexed libraries were pooled and sequenced using Illumina MiSeq V3technology (2 × 300 bp; Macrogen).The raw sequence data are stored in the National Center for Biotechnology Information (NCBI) Short Read Archive (project number PRJNA77209).

Sample Categories
Colostrum samples with a quality ≤50 g/L, >50 and <100 g/L, and ≥100 g/L were categorized as bad, good, and excellent quality, respectively (Quigley et al., 2013a).Calves' serum IgG concentrations ≤10 g/L, >10 and <15 g/L, and ≥15 g/L were categorized as low, moderate, and high, respectively, based on Furman-Fratczak et al. (2011).The numbers by category for colostrum and serum IgG concentration are represented in Table 2. Differential abundance analysis of the sequencing data was performed between different categories of colostrum quality and serum IgG concentration.Dry period was categorized as short and long, with a length <42 or ≥42 d, respectively, based on the desired dry period length (Bachman and Schairer, 2003).Gestation was categorized as short and long, with a length <279 or ≥279 d, respectively.This cutoff point for gestation length was the average gestation length of the cows included in this study.Calves with a serum IgG level ≤10 g/L were categorized as having FPT (Beam et al., 2009).The prevalence of FPT was calculated as the number of FPT cases divided by the total number of calves.

Microbiome Data Analysis
The amplicon sequencing data set was demultiplexed and barcodes were clipped by the sequence provider.Removal of primer sequences and filtering as well as trimming were performed using the DADA2 pipeline in R (version 3.6.2;Callahan et al., 2016).Forward and reverse reads were trimmed at a length of 280 and 210 bp, respectively.Forward and reverse estimated errors were set to 2 and 4, respectively.Taxonomy was assigned using the RDP naïve Bayesian classifier method (Wang et al., 2007) and the SILVA database, version 138 (Quast et al., 2013).Reads were classified at multiple taxonomic levels: phylum, class, order, family, and genus.Rarefaction analyses was performed using the R package vegan (Oksanen et al., 2015) and rarefaction curves (see Supplemental Figure S1; https: / / figshare .com/s/ 7234bcd8d550ebc3981f; Van Hese et al., 2022) reached a plateau, which suggests that saturation in sequencing was achieved (Zaheer et al., 2018).Statistical analysis was performed in R (version 3.6.2;https: / / www .r-project .org/ ) using the combined data set of both breeds.In addition, microbiome analysis was performed on separate BB and HF data sets because of the prominent clustering of bacterial composition within breeds, which would overshadow the effect of other variables.α-Diversity (microbial variation within samples) was measured in unfiltered data using the Shannon diversity index, Chao1 richness, and inverse Simpson diversity calculated with the Phyloseq package (McMurdie and Holmes, 2013).Due to a lack of normality, the Wilcoxon rank sum test was used to define statistical difference in α-diversity measures (Shannon, inverse Simpson, and Chao1) between the parameters of interest: breed, gestation length, parity, colostrum quality, serum IgG, season, grazing (only in BB), calving ease (only in HF), and dry period length (only in HF).Bray-Curtis dissimilarities were calculated to assess β-diversity (microbial variation between samples) and visualized using the nonmetric multidimensional scaling (NMDS) method.For subsequent data analysis, only amplicon sequence variants (ASV) with at least 20 counts across all samples were retained, reducing the total number of ASV from 11,599 to 8,418.Read counts were transformed to relative abundances.A Betadisper analysis was performed to check for homogeneity of variances of compared variables.Statistical differences between communities were analyzed by permutational multivariate ANOVA (PERMANOVA) using the Adonis function (Oksanen et al., 2015).Pairwise multivariate ANOVA (MANO-VA) was performed when factors contained more than 2 levels with the RVAideMemoire package (Hervé and Hervé, 2020).Differentially abundant ASV between breed, colostrum quality, and serum level categories were identified using the DESeq2 package with the Wald test (Love et al., 2014).Before the differential abundance analysis, we summed all ASV belonging to the same taxonomic group (e.g., genus) instead of testing each ASV individually, allowing us to study general trends for the complete genus.In the present data set, this led to a decrease in taxa from 8,418 to 815.P-Values were adjusted for multiple testing using the Benjamini and Hochberg method (Benjamini and Hochberg, 1995).A P-value of 0.05 or lower was considered statistically significant.

Statistical Analysis and Modeling
Means of serum IgG level, colostrum IgG concentration at first milking, and total amount of IgG fed within 24 h (i.e., administered in 3 feedings) between HF and BB were compared using Welch's 2-sample ttest.The Pearson chi-squared test was used to compare prevalence of FPT between BB and HF calves.Regression models were built to obtain the coefficient of determination (R 2 ) between serum IgG level and total amount of IgG fed in 24 h.Before performing linear regression modeling, outliers of colostrum and serum IgG concentrations were identified based on the residuals of their univariable linear models.Samples with residuals that deviated from the expected value by more than 20 for serum IgG and by more than 40 for colostrum IgG were excluded.Quantile-quantile (QQ) plots and histograms from the residuals are published in the Supplemental Figures S2 and S3 (https: / / figshare .com/s/ 7234bcd8d550ebc3981f; Van Hese et al., 2022).Linear regression models were used to determine cow, calf, or environmental factors that have significant influence on either colostrum quality, colostral microbiome composition, or calf IgG levels.A forward selection procedure was used to model the association between the recorded variables and colostrum IgG, serum IgG level, and AEA; AEA was calculated as serum IgG (g/L) × BW (kg) × 0.7 (estimated % blood volume)/total amount of IgG administered (g) (Halleran et al., 2017).
To investigate which variables affected IgG concentration, multivariable models were fitted for colostrum IgG concentration and serum IgG concentration separately with a forward stepwise selection procedure.In the colostrum IgG model, the following variables were included: breed, mean environmental temperature on the day of calving, season of calving, gestation length, and parity.In the serum IgG model, the following variables were included: total IgG fed in 24 h, mean environmental temperature on the day of calving, season of calving, IgG concentration of the first feeding, age at the first, second, and third feedings, and parity of the mother.Breed, mean environmental temperature, and parity were included in the final colostrum IgG model.Total IgG fed, mean environmental temperature, IgG concentration of first feeding, and age at first and third feeding were included in the serum IgG model.A Pvalue of ≤0.05 was considered statistically significant.The variance inflation factor was calculated and showed no collinearity between the predictor variables included in the model (Naimi et al., 2014).Mean environmental daily temperature was registered by the ILVO weather station, located close to the ILVO dairy cattle research barn.The analyzed data were considered sufficiently normally distributed based on a graphical examination of the residuals of the model (QQ plots and histograms in Supplemental Figures S2 and S3).

Descriptive Statistics on Colostrum Quality and Passive Transfer of Immunity
The results of the descriptive statistical analysis are summarized in Table 2.There was a clear difference in colostrum quality at first milking between the 2 breeds: BB cows produced colostrum with an average IgG concentration 19.96 g/L higher than that of HF cows (87.86 vs. 67.90g/L, respectively; P < 0.001).Because, according to our protocol, all calves had to receive 2 L of colostrum at each of 3 feeding times, BB calves received, on average, more antibodies in 3 feedings than HF calves (350.9 vs. 292.8g of IgG, respectively; P < 0.001), which consequently led to higher IgG serum levels in BB than in HF calves (IgG 5.7 g/L higher in BB vs. HF calves; 23.5 vs. 17.8 g/L IgG, respectively; P < 0.001).Despite the positive correlation (R 2 = 0.15 for HF and R 2 = 0.26 for BB) between the amount of IgG fed to calves and their serum IgG concentration, we observed wide variation in serum IgG concentration between calves (Figure 1).The latter suggests that the calf's serum IgG level is influenced by variables other than the amount of antibodies delivered.In addition, the prevalence of FPT was 15% higher in HF calves than in BB calves (11/62 vs. 1/41, respectively; P < 0.05).The fact that BB calves absorbed the administered antibodies more efficiently than HF calves (i.e., BB calves had higher AEA) might explain the higher prevalence of FPT in HF than in BB calves.
Linear regression modeling showed that breed, environmental temperature, and parity were significant predictors of colostrum IgG concentration: colostrum IgG (g/L) = −17.67× breedHF -0.51 × environmental temperature + 2.89 × parity + 86.35 (F 4,104 = 20.75,P < 0.001), with an adjusted R 2 of 0.42.As mentioned above, BB cows produced colostrum of better quality than HF cows.In BB and HF cows, colostrum IgG concentration decreased with increasing environmental temperature.In BB and HF, cows of parity 3 or more produced colostrum with significantly higher IgG concentration compared with cows of parity 1 and 2.
The following variables were significant predictors for serum IgG concentration in calves: serum IgG (g/L) = 0.06 × total IgG fed + 0.28 × environmental temperature + 0.13 × colostrum IgG first feeding + 98.39 × age first feeding -8.88 × age third feeding − 8.49 (F 5,103 = 18.05,P < 0.001).The more total antibodies administered to the calf, the higher its serum IgG level (Figure 1).Only in HF did serum IgG levels tend to increase with increasing environmental temperature (r = 0.23, P = 0.067).In both BB and HF calves, serum IgG level increased with increasing age at first feeding.The age at first feeding ranged between 10 and 150 min, with an average of 53 min.In HF calves only, age at third meal was negatively correlated with serum IgG level (r = −0.44,P < 0.001).The age of HF calves at third feeding ranged between 8 h 30 min and 27 h, with an average age of 18 h 37 min.Finally, environmental temperature, breed, and age at first and third feedings were significant predictors for AEA: AEA = −0.09× breedHF -0.14 × age third feeding -0.01 × environmental temperature + 0.83 × age first feeding + 0.0001 × environmental temperature × colostrum IgG first feeding + 0.44 (F 6,100 = 15.58,P < 0.001).

Microbial Composition of the Colostrum
Sequencing Results.After preprocessing of the sequences, 10,572,225 reads remained for downstream data analysis.The average number of reads per sample was 97,981, with a range of 46,478 to 138,261.Sequencing depth was sufficient, as nearly all rarefaction curves reached a plateau (Supplemental Figure S1).
Phylogenetic Profile.The 10 most abundant phyla and genera per breed are shown in Figure 2A and B, respectively.Proteobacteria was the most abundant phylum, followed by Firmicutes, Bacteroidetes, and Actinobacteria.The 15 most abundant genera in the α-Diversity.The average colostral bacterial diversity (inverse Simpson) and evenness (Shannon) differed significantly in HF and BB cows, with higher diversity and evenness in HF colostrum samples (Figure 3).In BB cows, α-diversity did not differ between colostrum quality categories, serum IgG level categories, or season of calving.Bacterial richness (Chao1) was significantly higher in BB cows of parity ≥3 compared with cows of parity 2 (P = 0.049).In the HF data set, calves with a low serum IgG level received colostrum with a significantly lower Chao1 index compared with that administered to calves with a moderate (P = 0.029) or high (P = 0.0074) serum IgG level.In addition, colostral bacterial richness (Chao1) and evenness (Shannon) from cows that calved in spring were significantly lower than that of cows that calved in winter (Figure 4).In contrast to BB colostrum, α-diversity of HF colostrum did not differ significantly between parity 1, 2, and ≥3.
β-Diversity.β-Diversity is a way to express the similarity between bacterial communities; for example, between colostrum from HF and BB cows (Xia et al., 2018).The following parameters were tested for differences in their microbial community: breed (complete data set), calving season, grazing, colostrum quality, calf serum IgG, parity, and calving ease (for BB and HF separately).Colostrum microbial composition differed significantly between HF and BB cows (P = 0.001), which can be observed in the NMDS plot showing separate clustering around BB and HF samples (Figure 5).However, there was no homogeneity of variance between these 2 breeds, meaning that it is difficult to determine whether the difference in β-diversity is due to a true difference or to nonhomogeneity.In both BB and HF, microbial communities were significantly different between cows with different seasons of calving.In BB, the microbial community of colostrum differed only between spring and autumn (P = 0.009), spring and summer (P = 0.009), and summer and autumn (P = 0.044), whereas in HF, the microbial community differed among all seasons (spring-autumn: P = 0.0015; spring-summer: P = 0.0015; spring-winter: P = 0.0024; summer-autumn: P = 0.01; summer-winter: P = 0.0015; autumn-winter: P = 0.0015).β-Diversity of colostrum was significantly different between the grazing and no-grazing groups in BB cows (P = 0.001).The  bacterial community of colostrum was similar between colostrum samples with different IgG concentrations in both breeds.We also analyzed the microbial composition of colostrum in relation to serum IgG level in calves.Only in HF did the colostral bacterial community differ significantly between colostrum leading to low versus high serum IgG levels in calves (P = 0.048).Bacterial communities were similar between cows of different parity in both breeds.In HF, there was no difference in β-diversity for different ease of calving; however, colostral microbial composition of cows that calved alone tended to differ from cows that needed intervention during extraction (i.e., calving ease being normal or hard).
Differential Abundance Analysis.The previous analyses were performed on the community level, considering differences of the entire bacterial community between different categories.Differential abundance analysis makes it possible to screen for taxa that are more or less abundant between categories of interest.This analysis was performed on filtered data, and low abundant taxa with <20 counts across all samples were excluded (reduction from 11,323 to 8,341 ASV).Next, counts of ASV belonging to the same genus were summed to find general trends for the complete genus (reduction from 8,341 to 792 taxa).Out of 792 taxa, 103 were differentially abundant, with 42 having higher and 61 having lower abundance in HF versus BB cows. Figure 6 shows the log-transformed base mean (log-baseMean) and log2 fold change (log2FC) of the 10 differentially abundant genera with highest baseMean between HF and BB.A positive log2FC means a higher abundance for that genera in BB versus HF, whereas a negative log2FC means a lower abundance for that genera in BB versus HF.Only results from the differential abundance analysis based on colostrum quality categories and serum IgG categories will be discussed.Although we found no significant difference between the overall bacterial communities of colostrum of different quality, several genera were differentially abundant between the bad, good, and excellent colostrum categories.Figures 7 and 8 show the logbaseMean and log2FC of the 10 differentially abundant taxa with highest baseMean between colostrum of different quality in BB and HF cows, respectively.As mentioned above, in the HF breed, we found a significant difference in microbial composition of colostrum leading to low versus high serum IgG levels in calves, with 41 taxa that were less abundant and 5 taxa that were more abundant in the low serum IgG group compared with the high serum IgG group.The 10 differentially abundant genera with highest baseMean are represented in Figure 9.We analyzed which genera were differentially abundant in bad quality colostrum (leading to low serum IgG levels) in calves (bad-low, n = 3) versus colostrum of bad quality leading to high IgG levels in calves (bad-high, n = 2), and then in colostrum of good quality leading to low serum IgG levels (good-low, n = 8) versus colostrum of good quality leading to high IgG levels (good-high, n = 37).This analysis was only performed in the HF data set as there were insufficient combinations in the BB data set (only 1 calf with a low serum IgG level).Between the bad-low and bad-high categories, 14 taxa were differentially abundant, with 8 having lower and 6 having higher abundance in the bad-low versus the bad-high group.Figure 10A represents the 10 most common differentially abundant genera between the bad-low versus bad-high group.Between the good-low and good-high category, 63 taxa were differentially abundant, with 56 having lower and 7 higher abundance in the good-low versus good-high group.Figure 10B represents the 10 most common differentially abundant genera between the good-low and good-high groups.

DISCUSSION
The basic goal in colostrum administration is feeding the calf sufficient antibodies as soon as possible after birth.The first step is to administer colostrum of good quality (i.e., colostrum with an IgG concentration of at least 50g/L; Quigley et al., 2013a).Several studies show high variation in the IgG concentration of colostrum (Pritchett et al., 1991;Quigley et al., 2013a), which was confirmed in this study.Only 42% of the variation could be explained by breed, environmental temperature, and parity.Other possible factors influencing the variation in IgG concentration could be the volume produced, the time between calving and milking, and the prepartum diet (Puppel et al., 2019).Unfortunately, in this study, the volume produced was not recorded, making it hard to assess the effect of volume on the IgG concentration of colostrum.The prepartum diet and time between calving and milking were standardized in our study, meaning that these parameters can be ruled out as causes of variation between our samples.Logically, when more antibodies are delivered, more will be absorbed.Indeed, many studies explain the wide variation in serum IgG levels between calves by differences in IgG concentration of the administered colostrum (Bush and Staley, 1980;Johnsen et al., 2019).We also found a positive correlation between the serum IgG level in the calf and the total amount of administered antibodies, which can partly explain the difference in IgG absorption between HF and BB calves because BB cows produced, on average, colostrum with higher IgG levels compared with HF cows.However, this positive correlation between serum IgG and administered IgG only accounted for 15% of the observed variation between calves.Still, the most important question to answer is which colostrum-or calf-related factors could explain the other 85% of the variation in serum IgG absorption.
In this study, we focused on colostrum-related factors.The presence of (certain) bacteria might be a factor that interferes with the ability to absorb antibodies in the calf's intestines.Calves that received pasteurized colostrum were able to absorb more antibodies than calves that received raw colostrum (Johnson et al., 2007).This can be explained by lesser availability of antibodies as they will attach to present bacteria (Johnson et al., 2007).Competition between bacteria and IgG for the epithelial receptors, present on neonatal enterocytes, can occur as well (Stewart et al., 2005).Colostrum can harbor a wide variation of harmful bacteria, and pasteurization of colostrum is therefore a common practice in cattle husbandry.For a long time, colostrum was considered to be sterile, and isolated bacteria were ascribed to either the mother (udder and skin) or to environmental contamination.Based on extensive research on human breast milk in the last decade, the existence of a unique milk microbiome is widely accepted (Lyons et al., 2020).Bovine colostrum also contains beneficial bacteria that contribute to the colonization of the neonate's gut (Addis et al., 2016).Whether these beneficial bacteria also contribute to the efficient absorption of antibodies in the gut remains unexplored.
The present study revealed 4 major findings: (1) β-diversity measures showed a distinct difference in the microbial composition between colostrum from HF and BB cows; (2) multiple differentially abundant bacterial genera were observed between colostrum from different quality categories in both breeds; (3) differences in microbial composition were observed in colostrum categorized according to serum level categories of the corresponding HF calves; and (4) in both HF and BB cows, there was a seasonal effect on the microbial composition of colostrum.

Colostral Microbial Composition
In both BB and HF cows, Proteobacteria, Firmicutes, Bacteroidetes, and Actinobacteria were the dominant phyla of the microbial community in colostrum.This is consistent with other studies conducted on bovine colostrum and milk (Zhang et al., 2015;Lima et al., 2017).Interestingly, these phyla are also dominant in human milk (Kumar et al., 2016;Moossavi et al., 2019) and in sow milk (Chen et al., 2018).The origin of these milk bacteria has been widely investigated but is still not fully understood.Suggested sources of bacteria present in the mammary gland and milk are contamination from the environment and transfer of bacteria present on the maternal skin and in the neonate's oral cavity (Addis et al., 2016).However, the presence of strictly anaerobic bacterial species related to the intestinal microbiota has led to the discovery of a bacterial translocation from the gut to the mammary gland in late gestation and during lactation, the so-called entero-mammary pathway, in mice (de Andrés et al., 2017) and humans (Perez et al., 2007), and has been suggested in cows (Young et al., 2015) and other ruminants (Li et al., 2017).It is believed that vertical transmission of these bacteria is an important mechanism in the colonization of the neonate's gut and maturation of the neonate's immune system (Jost et al., 2014).
The study of Lima et al. (2017) shared several common genera with our study, such as Acinetobacter, Pseudomonas, and Staphylococcus, albeit in different proportions.In contrast to the present study and despite the similar management of primiparous and multiparous cows, Lima et al. (2017) observed a difference in microbial diversity of colostrum between primiparous and multiparous cows.However, this difference In Belgian Blue (BB) colostrum, of the top 10 differential abundant taxa (i.e., highest baseMean), 9 were less abundant and 1 was more abundant in colostrum of (A) bad versus good quality colostrum; and (B) bad versus excellent quality colostrum.Base mean was log-transformed (logbaseMean) for better visualization.Log2 fold changes (log2FC) indicate the differential abundance of the particular genera, where a positive (negative) log2FC indicates a higher (lower) abundance in bad versus good or excellent quality.If taxonomy could not be assigned on the genus level, the most specific taxonomy level available was assigned (F = family).
was only shown in Chao1 richness, not in the Shannon diversity index, and only with a P-value between 0.05 and 0.1 (Lima et al., 2017).the species level is important to attain greater certainty on the effect of its presence.For example, Staphylococcus aureus is associated with mastitis, whereas Staphylococcus saprophyticus is more prevalent in milk of healthy cows (Rall et al., 2014).An alternative for 16S rRNA sequencing is whole (meta)genome sequencing (WGS), which looks at the complete genome of microbes.Not only does this technique allow accurate classification at the species level, it can also give more insight in the metabolic potential of these microbes (Iyer, 2016).Another limitation of 16S rRNA sequencing (as well as of WGS) is the lack of functional analysis of the colostrum microbiota, making it difficult to determine potential functions of the identified bacteria.Although 16S rRNA metabarcoding tells us which bacteria are present and WGS gives insight in the potential of the microbial community, metatranscriptomics shows which bacterial genes are expressed (Iyer, 2016).However, these more sophisticated techniques are more challenging in practical terms and more expensive compared with 16S rRNA metabarcoding.Additionally, colostrum is a very specific matrix, making it challenging for high-quality DNA isolation.

Colostral Microbial Composition Differs Between HF and BB
The bacterial community of colostrum from HF cows was significantly different from that of BB cows.Moreover, colostral microbial diversity was higher in HF cows than in BB cows.A first explanation of these findings could be the different genetic background of the 2 breeds.Differences in microbial composition between cattle breeds have been observed before.Gonzalez-Recio et al. (2018) showed differences in the microbial composition of the rumen between HF and Brown Swiss cows and found an association between the host's genetics and the rumen microbiome.Cremonesi et al. (2018a) reported that milk microbiota of HF cows was distinct from that of Rendena cows (an Italian dual-purpose breed), with a more diverse microbial community in milk of HF cows.
Another explanation of the interbreed variation could be the effect of dissimilar environments (Vacheyrou et al., 2011).As mentioned before, at the ILVO, HF and BB cows are group housed per breed in separate barns, so we cannot rule out the possibility that colostrum from HF is exposed to different environmental contamination from that of BB (Raboisson et al., 2016).
A third possible explanation is different management.First, HF cows receive a different diet than BB cows.In human research, a dietary effect on the milk microbiome has been found (Zimmermann and Curtis, 2020).In the study of Zhang et al. (2015), several milk bacteria were differentially abundant between HF cows that received a diet with high versus low concentrate.Second, there is a major difference in antimicrobial therapy between the 2 breeds.Prophylactic intramammary treatment with antibiotics at drying-off is a common practice in dairy farming.At the ILVO, when HF cows are dried off, at approximately 45 d before expected calving, they are treated with long-acting intramammary antibiotics.Several studies reported no significant effects of intramammary antibiotic treatment on the milk microbiome (Ganda et al., 2016;Bonsaglia et al., 2017;Biscarini et al., 2020).This is supported by the fact that we did find differences in the colostral microbial composition between primiparous and multiparous cows, the former not being systematically treated with intramammary antibiotics.Belgian blue cows, on the other hand, are treated with antibiotics during their elective cesarean section.In our study, time between this antibiotic treatment and colostrum collection in BB cows was approximately 30 min.Whether antibiotic treatment can affect the colostrum microbiome in this short period has yet to be determined.

Differentially Abundant Taxa Between Colostrum of Different Quality
Colostrum was assigned to 1 of the following 3 quality levels: bad (≤50 g/L IgG), good (>50 and <100 g/L IgG) and excellent (≥100 g/L IgG) based on common thresholds for IgG concentration used in literature and practice (Quigley et al., 2013a).Differential abundance analysis was performed separately for the 2 breeds, and the results showed multiple differentially abundant bacterial genera between the different quality categories.For example, in BB, an unclassified Enterobacteriaceae was 97 times higher in abundance (log2FC = 6.6) in colostrum of bad quality compared with that of good and excellent quality (Figure 7).Several members of this family belong to the coliforms (e.g., Escherichia) and are indicators of fecal and environmental contamination (Martin et al., 2016).The study of Saldana et al. (2019) reported higher coliform counts in colostrum of low quality, and reducing these counts by heat treatment leads to higher efficiency of IgG absorption in calves.Interestingly, according to the results reported in Lima et al. (2017), an unclassified Enterobacteriaceae was part of the core microbiome of bovine colostrum.Although this genus was highly differentially abundant between BB colostrum of different quality, this was not the case for HF colostrum (Figure 8).
In BB colostrum, 8 of the top 10 differentially abundant bacteria were lower in abundance in the bad versus good and bad versus excellent quality categories (Figure 7).Several of these differentially abundant bacteria have previously been associated with dairy products.The genera Dermacoccus and Macrococcus were previously found in bulk tank cow milk with high SCC but have not been described as mastitis pathogens (Rodrigues et al., 2017).Well-known species from the genus Mycobacterium include pathogens such as M. paratuberculosis and M. bovis; however, several species are nonpathogenic and have been isolated from drinking water (Vaerewijck et al., 2005).Species from the family Microbacteriaceae have been detected in cow milk, feed, and air and dust in the barn (Vacheyrou et al., 2011;Montel et al., 2014).When milk samples of HF cows showing signs of acute clinical mastitis were compared with samples from healthy cows, the genus Ralstonia was significantly more abundant in milk of healthy cows (Kuehn et al., 2013).The Lactococcus and Carnobacterium genera were more abundant in excellent versus bad BB colostrum, but not in good versus bad colostrum.Lactococcus has also been isolated from milk samples (Vacheyrou et al., 2011).Moreover, in the study of Cremonesi et al. (2018b), Lactococcus was a member of the core microbiome of Rendena (but not of HF) cow milk.Carnobacterium spp.are associated with antibacterial activity against Listeria monocytogenes (Leisner et al., 2007).Between BB and HF colostrum, only Mycobacterium was commonly less abundant in bad versus good colostrum.This could mean that other bacteria are associated with colostrum quality in HF and BB or that the presence of Mycobacterium plays an important part in determining the amount of antibodies in colostrum.Interestingly, the top 10 differentially abundant genera between bad and good colostrum quality were all less abundant in bad colostrum compared with good colostrum, whereas the top 10 differentially abundant genera between bad and excellent colostrum were all more abundant in bad colostrum than in excellent colostrum (Figure 8).Several of these more abundant genera in bad versus excellent colostrum are associated with contamination of colostrum and milk and could lead to neonatal health problems (Fecteau et al., 2002) or food safety breaches (Hugo et al., 2006;Vacheyrou et al., 2011).Pseudomonas, for example, was more abundant in bad versus excellent HF colostrum and are common contaminants of raw milk (De Jonghe et al., 2011).This genus was identified by Lima et al. (2017) as a dominant taxon in bovine colostrum and as a member of the colostrum core microbiome.In contrast, Pseudomonas was more abundant in colostrum of good quality leading to low serum IgG levels versus high serum IgG levels (Figure 10B).
These results show the possibility that some bacteria are associated with a higher or lower IgG concentration in colostrum.However, a main concern looking at these results is the potentially high individual influence on the microbial community because only few samples were represented in the bad (4.3% in BB and 11.3% in HF) and excellent (15.2% in BB and 1.6% in HF) quality categories (Rosenberg and Zilber-Rosenberg, 2016).Research including more bad quality colostrum samples is needed to further elucidate the correlation between IgG concentration and colostral microbial composition.

Colostral Microbial Composition Affects Serum IgG Absorption in Calves
As done for colostrum quality, serum levels of calves were assigned to 3 categories to perform differential abundance analysis between the different categories.Three categories-low (≤10 g/L IgG), moderate (>10 and <15 g/L IgG), and high (≥15 g/L IgG)-were made based on common thresholds used in literature and practice (Furman-Fratczak et al., 2011).Only in colostrum from HF cows was a compositional difference observed, based on serum IgG level in calves; that is, the presence of some genera in colostrum given to calves that had low serum IgG differed from those in colostrum given to calves that had moderate or high IgG levels.The latter raises the question whether the colostral microbial composition affects the fitness of the calf or whether the presence of some genera is linked to better or worse absorption of IgG.The higher abundance of several genera in colostrum administered to calves with a low serum IgG level can be explained by the fact that pathogenic bacteria can interact with the antibodies, leading to lower availability for absorption (Saldana et al., 2019).In contrast, upregulated genera in the calves with high antibody levels should be further investigated to demonstrate a link with calf fitness.Figure 10A shows the top 10 differentially abundant bacteria within the bad quality category leading to low serum IgG levels (bad-low) versus high serum IgG levels (bad-high).From the lower abundant genera in the bad-low group, Solibacillus was previously identified as a member of the buffalo milk core microbiome and was more abundant in milk from healthy quarters compared with milk from mastitic quarters (Catozzi et al., 2017).Proteus spp.can be found in raw milk and are often related to subclinical mastitis (Ontsouka et al., 2003), but are also associated with aromatizing cheese products (Deetae et al., 2009;Quigley et al., 2013b).Whether udder health during previous lactations, at the start of the dry period, or at the start of the lactation affects colostrum quality is poorly studied.This was outside the scope of the present study; however, for future research, it would be interesting to analyze the effect of udder health on IgG concentration and microbial composition of colostrum.Of the top 10 differentially abundant genera, Rheinheimera, Kocuria, and Flavobacterium are psychrotrophic bacteria and were previously identified in raw milk samples (Hantsis-Zacharov and Halpern, 2007;Yuan et al., 2018).Several psychrotrophic bacteria are known to easily form biofilms, making them notorious contaminants of dairy products (Yuan et al., 2019).Differential abundance analysis was performed between colostrum samples of good quality leading to low versus high serum IgG level, and the top 10 differentially abundant genera are represented in Figure 10B.Of these, Pseudomonas and family Lachnospiraceae were already identified as members of bovine colostrum's core microbiome (Lima et al., 2017).Moreover, in the study of Klein-Jöbstl et al. (2019), Pseudomonas was identified in samples of bovine colostrum, the cow's birth canal, and calf's mouth and feces.
However, as mentioned with the interquality difference in colostral microbial composition, the same consideration about possible high individual effects should be made because only a few calves obtained a low (17.7% in HF) and moderate (22.6% in HF) IgG level (Rosenberg and Zilber-Rosenberg, 2016).

Seasonal Effect on the Microbial Composition of Colostrum
In HF and BB, colostral microbial composition differed between calving seasons.The microflora of colostrum in winter could not be differentiated from that of other seasons only in BB.Typical weather conditions such as temperature and humidity can at least partly explain these differences.The latter has previously been shown in raw milk, where variation in microbial composition over the year was most affected by temperature and humidity (Li et al., 2018a).Another possible effect could be differences in management between seasons.For example, cows are given access to pasture from approximately March until October (depending on weather conditions).This not only exposes them to different microbes from those in animals housed in the barn, but the dietary changes associated with outdoor grazing could also result in changes in the microbial community.The effect of outdoor grazing on the microbial community of the cow's teat skin has been demonstrated previously (Frétin et al., 2018).
As previously mentioned, more extensive research is needed on this topic, because many interesting questions remain unanswered.For example, could the microbial composition of bovine colostrum be altered by dietary changes, as seen in human research?Another important question is what effect the heat treatment of colostrum has on its microflora.
To summarize, the colostral microbial composition of HF cows differed from that of BB cows, with possible genetic, environmental, and management influences.The abundance of several genera differed among the Van Hese et al.: TRANSFER OF PASSIVE IMMUNITY AND COLOSTRAL BACTERIA 3 colostrum quality categories; however, a strong effect of individual on these results cannot be excluded.Microbial composition may also affect IgG absorption in the calf.Whether some genera are associated with improved IgG absorption should be further examined in future research.Another interesting finding was the seasonal effect on the microbial composition of colostrum.The underlying factors associated with this finding and whether the differential abundance between seasons is beneficial is another interesting research topic.

VanFigure 1 .
Figure1.Absorption of colostral antibodies in calves expressed in terms of serum IgG levels in relation to the total (Tot) amount of antibodies administered within 24 h after birth in Belgian Blue (BB) and Holstein Friesian (HF) calves.The dashed line represents the threshold for failure of passive transfer (FPT; 10 g/L).All calves below this threshold suffer from FPT.A positive correlation was found between serum IgG (g/L) and total amount of antibodies (g) fed (R 2 = 0.15 for HF; R 2 = 0.26 for BB), although large variation in absorption between calves was observed.

VanFigure 2 .
Figure 2. Bacterial composition of colostrum from Belgian Blue (BB) and Holstein Friesian (HF) cows, showing (A) top 10 phyla and (B) top 10 genera that were most abundant in BB and HF colostrum.

Figure 3 .
Figure 3. Bacterial richness (Chao1), evenness (Shannon), and diversity (inverse Simpson) between Belgian Blue (BB) and Holstein Friesian (HF) colostrum.P-values were calculated with the Wilcoxon ranked sum test and are displayed above the boxplots.Individual points represent the richness estimate per sample, brackets per individual point in the Chao1 measure represent the theoretical standard error range associated with that measure.Boxplots are overlayered on top of the points for the 2 breeds with the median represented by the line in the box; the lower and upper hinge correspond to the first and third quartiles (the 25th and 75th percentiles).A P-value < 0.05 is statistically significant.Evenness and diversity were significantly different between BB and HF colostrum.

Van
Figure 4. Bacterial richness (Chao1), evenness (Shannon), and diversity (inverse Simpson) between calving season in Holstein Friesian (HF) colostrum.Individual points represent the richness estimate per sample, brackets per individual point in the Chao1 measure represent the theoretical standard error range associated with that measure.Boxplots are overlayered on top of the points for the 2 breeds with the median represented by the line in the box; the lower and upper hinge correspond to the first and third quartiles (the 25th and 75th percentiles).P-values were calculated with the Wilcoxon ranked sum test and are displayed above the boxplots.

VanFigure 6 .
Figure 6.Several taxa were differentially abundant between colostrum of Belgian Blue (BB) and Holstein Friesian (HF) cows.This barplot shows the 10 most common differential abundant genera.Base mean was log-transformed (logbaseMean) for better visualization.Log2 fold changes (log2FC) indicate the differential abundance of the particular genera, where a positive (negative) log2FC indicates a higher (lower) abundance in BB compared with HF.If taxonomy could not be assigned at the genus level, the most specific taxonomy level available was assigned (F = family).

Figure 8 .
Figure8.In Holstein Friesian (HF) colostrum, the top 10 differential abundant genera between (A) bad versus good colostrum quality were all less abundant in bad quality colostrum; (B) the top 10 between bad versus excellent colostrum quality were all more abundant in bad quality colostrum.Base mean was log-transformed (logbaseMean) for better visualization.Log2 fold changes (log2FC) indicate the differential abundance of the particular genera, where a positive (negative) log2FC indicates a higher (lower) abundance in bad versus good or excellent quality.If taxonomy could not be assigned on the genus level, the most specific taxonomy level available was assigned (O = order, F = family).

VanFigure 9 .
Figure 9. Top 10 most common differential abundant genera between colostrum leading to low versus high serum IgG levels in Holstein Friesian (HF) calves.Base mean was log-transformed (logbaseMean) for better visualization.Log2 fold changes (log2FC) indicate the differential abundance of the particular genera; a positive (negative) log2FC indicates a higher (lower) abundance in colostrum leading to low versus high serum IgG level.If taxonomy could not be assigned on the genus level, the most specific taxonomy level available was assigned (O = order, F = family).

Van
Figure 10.(A) In Holstein Friesian (HF) colostrum of bad quality, of the top 10 differential abundant genera between the bad-low versus bad-high group, 6 and 4 taxa were less [negative log2 fold change (log2FC)] and more (positive log2FC) abundant, respectively, in the bad-low versus the bad-high group.(B)In HF colostrum of good quality, of the top 10 differential abundant genera between the good-low versus goodhigh group, 9 and 1 taxa were less and more abundant, respectively, in the good-low versus the good-high group (where bad and good refer to colostrum quality, and low and high refer to serum IgG level of the calf).If taxonomy could not be assigned on the genus level, the most specific taxonomy level available was assigned (O = order, F = family).
Van Hese et al.: TRANSFER OF PASSIVE IMMUNITY AND COLOSTRAL BACTERIA

Table 1 .
Van Hese et al.: TRANSFER OF PASSIVE IMMUNITY AND COLOSTRAL BACTERIA Diet composition (g/kg of DM, unless otherwise noted) during the far-off and close-up dry periods for Holstein Friesian (HF) cows and diet composition for Belgian Blue (BB) cows

Table 2 .
Van Hese et al.: TRANSFER OF PASSIVE IMMUNITY AND COLOSTRAL BACTERIA Mean ± SD of serum IgG by category, total IgG administered, colostrum IgG from the first milking, apparent efficiency of absorption (AEA), and prevalence of failure of passive transfer (FPT) between Belgian Blue and Holstein Friesian cows ³Colostrum IgG thresholds: bad (≤50 g IgG/L), good (>50 and <100 g IgG/L), excellent (≥100 g IgG/L).