Animals, husbandry, and sample collection
For the study, pregnant, healthy prepartum heifers from the German Holstein breed were selected, which addressed a combination of two genomic target regions (43–48 Mb and 53–59 Mb) on BTA18 . The margins of the sub-regions were determined from a previous BTA18 mastitis model (see below).
The first step of the selection process comprised the identification of German Holstein sires with extreme differences for the summarized SNP effects for SCS of their haplotypes in the target regions (see also Additional file 6). For this purpose, initially SNP haplotyping was performed for all individuals within the VIT genome data base for German Holsteins . Genotyping data had been obtained with the 50 k Illumina BovineSNP50 BeadChip (Illumina Inc., San Diego, CA, USA) from routine genomic evaluation for German Holsteins (February 2013 ). SNPs were filtered for a minor allele frequency > 1% leaving 43,586 autosomal SNPs for further processing. SNP genotypes were tested for agreement with pedigree information . Only animals with a SNP call rate greater than 98% were further considered. For imputation of missing marker genotypes and for phasing the genotypes, Beagle  was used, which relies on population-wide linkage disequilibrium. The UMD3.1 bovine genome assembly  served as backbone for all genomic SNP coordinates.
The target regions on BTA18 for our selection process were established based on a previous study [38, 39], in which three sires with confirmed alternative haplotype effects on SCS on BTA18 had been investigated. Inspection of their favorable and unfavorable haplotypes revealed regions common to the three unfavorable haplotypes (q) of those sires. Together with data from the literature [8, 40, 41], the following boundaries of the target genomic regions for this project were defined: rs41880634 (BTA18: 43,098,071) - rs109689271 (BTA18: 47,983,685) and rs29021987 (BTA18: 53,013,208) - rs43072554 (BTA18: 58,696,066). SNP allele effects were summarized within each of the two haplotypes for each sire for the two target intervals and in addition for the region rs41880634 (BTA18: 43,098,071) to the telomeric end of the chromosome. Subsequently, for each of the sires the difference of the summarized effect for its alternative haplotypes was calculated. The following criteria were applied for filtering of sires to be eligible for heifer selection: i) difference in summarized haplotype effect difference at least two standard deviations larger than the mean haplotype difference of all sires for the region rs41880634 (BTA18: 43,098,071) to the telomeric end of the chromosome, ii) difference in summarized haplotype effect difference at least two standard deviations larger than the mean haplotype difference of all sires in at least one of the intervals 43–48 or 53–59 Mb, and finally iii) the sires were not allowed to have inverse phasing regarding the direction of the haplotype differences in the intervals 43–48 or 53–59 Mb.
A total of 156 sires fulfilled these criteria, which were further filtered for age of female offspring at the start of the experiments (at least 18 month of age) and anticipated day of calving (from insemination records to select heifers calving within the experimental time window). Finally, heifers were also submitted to further specific selection steps for sires’ and maternal grandsires’ breeding values for milk performance, overall somatic cell count, milking behavior and for maternal grandsires to obtain high variability within half-sib group, but similar performance level between half-sib groups . Further selection steps comprised the heifers’ predicted age at calving (< 36 months) and the number of potentially available daughters within half-sib group (potentially more than three daughters inheriting the favorable (Q) and three daughters inheriting the unfavorable haplotype (q)).
After genotyping with the 50 k Illumina SNP chip and haplotyping (essentially as described above) the finally filtered 282 heifers were assigned to the Q or q group according to the inherited BTA18 paternal haplotype for the target regions. Those heifers meeting all health and veterinary requirements were purchased from conventional private dairy farms across Germany and allocated about 6 weeks prior to first calving either to the animal experimental unit of the FBN Dummerstorf (n = 6, 3 Q, 3 q) for a long-term model or to the Clinic for Cattle at the University of Veterinary Medicine Hanover (TiHo) (n = 36, 18 Q, 18 q) for an infection challenge model.
In total, both groups (Q and q) comprised offspring of the same six sires (see also Additional file 6). For the TiHo animals, in each the Q group and in the q group there were a total of five different paternal haplotypes with respect to identical SNP alleles, respectively, because some sires shared identical haplotypes. In the FBN cohort, two of the three paternal q haplotypes and two of the three paternal Q haplotypes were identical regarding SNP alleles.
For the FBN cohort, the experiment was conducted under the reference number 7221.3–1-055/15 with the approval by the responsible authority (LALLF, Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei Mecklenburg-Vorpommern, Rostock, Germany). For the TiHo cohort, the experiment was performed under the reference number 33.12–42502–04-15/2024 by the Lower Saxony Federal State Office for Consumer Protection and Food Safety. Furthermore, this study was submitted to and approved by the ethics committees of the Leibniz Institute for Farm Animal Biology and the University of Veterinary Medicine Hanover, foundation, respectively. All ethical evaluations were performed as required by the German Animal Care law (Tierschutzgesetz, https://www.gesetze-im-internet.de/tierschg/BJNR012770972.html).
The husbandry and sample collection of the TiHo animals was performed as described by Meyerholz et al., under revision. Briefly, all heifers were housed in individual loose stall pens on straw. The animals were milked twice daily, and milk yield was recorded. Weekly quarter milk samples were collected for analysis of milk components, somatic cell count, and microbiological examination. Moreover, weekly quarter milk samples were collected, conserved by bronopol, and analyzed at the MKV Mittelweser e.V. (Milchwirtschaftlicher Kontrollverband Mittelweser e.V., Rehburg-Loccum, Germany) for determination of SCC using the MilkoScan FT Plus (FOSS, Hilleroed, Denmark).
The TiHo heifers were fed one of three component diets (dry period: < 270 days post insemination (p.i.), prepartum period: > 270 days p.i., and lactation period: after calving). The diets comprised in the dry period hay and minerals, in the prepartum period hay, grass silage, corn silage, concentrates, and minerals and in the lactation period grass silage, corn silage, rapeseed extraction meal, soy extraction meal, concentrates, and minerals. The animals left the observation period at day 39 ± 4 after calving.
Six (3 Q, 3 q) heifers were kept in a free-stall barn at the Leibniz Institute for Farm Animal Biology in Dummerstorf (FBN). Husbandry and sample collection at the FBN were performed as follows: Until parturition, cows were housed in calving boxes. After calving, the cows were moved to a dairy cattle loose stall barn and were kept in the same group during the entire observation period. The FBN cows were feed ad libitum with their daily feed intake measured via weighing troughs controlled by the Roughage Intake Control (RIC) system (Insentec, Marknesse, The Netherlands) . The cows were fed different total mixed ratios (TMR) depending on their lactation status (dry: starting at arrival, transit: starting 14 days ante partum (a.p.), lactating cows: starting post partum (p.p.)) with adjusted energy content. One representative ratio each for dry, transit, and lactating cows can be found in the supplements of this publication (see Additional file 7). The animals had free access to water.
The FBN cows were milked twice a day in an auto-tandem milking parlor (DeLaval, Tumba, Sweden) with daily recording of milk yield. Moreover, once a week the milk of one afternoon and the following morning milking was pooled and analyzed for content of fat, protein, lactose, urea, and somatic cells at a milk lab (LKV, Landeskontrollverband für Leistungs- und Qualitätsprüfung Mecklenburg-Vorpommern e.V., Güstrow, Germany) using infrared spectroscopy (MilkoScan FT and Fossomatic FC, FOSS, Hilleroed, Denmark). Furthermore, the LKV determined SCC on udder quarter level at specific time points during the lactation (day 2, 7, 14, 21, 28, 35, 42, 70, 150, and 240 p.p.). The SCS was calculated by the following formula: SCS = log2 (SCC / 100,000) + 3 . Energy-corrected milk (ECM) was calculated according to Kirchgessner (1997): ECM = average daily milk yield x (0.37 x milk fat percentage + 0.21 x milk protein percentage + 0.95) / 3.1 [9, 44].
The body weight (BW) was recorded daily after the animals were leaving the milking parlor at the FBN. Furthermore, weekly backfat thickness (BFT) was measured by ultrasonic measurement (SonoSite Titan, SonoSite GmbH, Erlangen, Germany) in the sacral region following an established method , and simultaneously body condition score (BCS) was assigned according to a standardized scheme . The energy balance (EB) p.p. was calculated with the following formula: EB (MJ NEL) = NEL-intake − (kg ECM × 3.14 + 0.293 × kg BW 0.75) .
In the observation interval until week 35 p.p. at the FBN, a veterinary clinical examination was performed weekly (daily the first 5 days after calving) to monitor the animals’ health. All veterinary diagnoses and treatments as well as zootechnical interventions (claw care, inseminations etc.) were electronically documented. Rectal temperature was measured daily after the morning milking. In case of infections or diseases, the animals were treated according to good veterinary practice. After first calving, the cows at the FBN were inseminated starting at day 64 ± 23 after parturition, and potential pregnancies were recorded.
Blood was collected by licensed veterinarians from the Vena jugularis starting 10 days before the calculated calving date, then 2 days after parturition, followed by weekly sampling until day 42 p.p.. The last sampling days in the lactation were day 70, 150, and 240. The samples from day 10 a.p., 2 p.p., 14 p.p., 70 p.p., 150 p.p., and 240 p.p. were sent to an accredited laboratory (synlab.vet, Berlin, Germany) for differential blood count using flow cytometry and microscope. Serum concentrations of NEFA (non- esterified fatty acids) and BHB (beta-hydroxybutyric acid) were determined in samples of day 10 a.p., 2 p.p., 7 p.p., 14 p.p., 21 p.p., and 42 p.p. using the ABX Pentra 400 (HORIBA, Ltd., Kyoto, Japan). Furthermore, plasma samples from the same days were examined for insulin-like growth factor-I (IGF-I) and growth hormone (GH) using validated immunoassays .
In addition to samples from whole milkings, quarter milk samples from the FBN cows were taken at the same time points as blood samples and were sent to the lab MQD (Qualitätsprüfungs- und Dienstleistungsgesellschaft Mecklenburg-Vorpommern GmbH, Güstrow, Germany), where in addition to analysis of the SCC a bacteriological status for each udder quarter was determined by qualitative macroscopic evaluation of colonies grown on blood agar.
The cows were killed by immediate exsanguination after stunning with a captive bolt gun (FBN cohort: approximately 6 weeks into their second lactation; TiHo cohort: at day 39 ± 4 after calving).
The data analysis was performed by scripts and packages within the R platform (version 3.4.3) . For the graphical representation of the data, the package ggplot2 was used . To evaluate differences between the Q and q animals, we fitted a linear model to the data using the lm function [51, 52] with fixed effects of group (either Q or q) and week of lactation. For those traits measured daily / weekly across the lactation (feed intake, body weight, ECM, BFT, BCS, and SCS) we fitted orthogonal polynomials or a natural spline to the data to account for missing data points and outliers due to technical problems. For statistical evaluation of a potential significance in different proportions of udder quarters with extremely low somatic cell count and bacterial colonization, a Pearson’s Chi-squared test implemented in the MASS package in R  was applied.