In vivo model to study the impact of genetic variation on clinical outcome of mastitis in dairy heifers

Title: Background: In dairy herds, mastitis causes detrimental economic losses. Genetic selection offers a sustainable tool to select animals with reduced susceptibility towards postpartum diseases. Studying underlying mechanisms is important to assess the physiological processes that cause differences between selected haplotypes. Therefore, the objective of this study was to establish an in vivo infection model to study the impact of selecting for alternative paternal haplotypes in a particular genomic region on cattle chromosome 18 for mastitis susceptibility under defined conditions in dairy heifers. Results: At the start of pathogen challenge, no significant differences between the favorable (Q) and unfavorable (q) haplotypes were detected. Intramammary infection (IMI) with Staphylococcus aureus 1027 (S. aureus, n = 24, 96 h) or Escherichia coli 1303 (E. coli, n = 12, 24 h) was successfully induced in all heifers. This finding was confirmed by clinical signs of mastitis and repeated recovery of the respective pathogen from milk samples of challenged quarters in each heifer. After S. aureus challenge, Q-heifers showed lower somatic cell counts 24 h and 36 h after challenge (P < 0.05), lower bacterial shedding in milk 12 h after challenge (P < 0.01) and a minor decrease in total milk yield 12 h and 24 h after challenge (P < 0.01) compared to q-heifers. Conclusion: An in vivo infection model to study the impact of genetic selection for mastitis susceptibility under defined conditions in dairy heifers was successfully established and revealed significant differences between the two genetically selected haplotype groups. This result might explain their differences in susceptibility towards IMI. These clinical findings form the basis for further in-depth molecular analysis to clarify the underlying genetic mechanisms for mastitis resistance.


Background
For decades, mastitis has caused large-scale economic losses worldwide in dairy farming due to treatment costs, discarded milk, reduced milk yield and increased culling rates [1][2][3][4][5][6]. A recent study from Canada estimated costs on typical dairy farms to be 662 Canadian Dollars per milking cow per year, in which nearly half of the costs were associated with subclinical mastitis [7]. Additionally, indirect costs arise due to reduced fertility of cows suffering from clinical or subclinical mastitis [8][9][10]. Clinical mastitis (CM) is defined as intramammary infection (IMI) with clinical symptoms, such as altered milk secretion, local (pain, swelling) or systemic signs of inflammation (fever, disturbed general condition). IMI with Escherichia coli ( E. coli) frequently causes CM, which can severely affect the well-being of the animal but often results in transient IMI with a comparably high self-cure rate [11][12][13][14][15]. In comparison, subclinical mastitis (SCM) includes IMI without clinical symptoms but an increased somatic cell count (SCC) in milk, decreased milk yield and reduced milk quality. Staphylococcus aureus ( S. aureus) is one major pathogen causing SCM or mild cases of CM in dairy cows [16]. Due to intermittent shedding, S. aureus is difficult to detect, and treatment of affected animals is often futile, since S. aureus IMI tends to persist within the udder and causes chronic cases of SCM [1,6,17]. In the dairy industry, CM and SCM are the major reasons for antimicrobial usage [18,19]. Additionally, cows with CM or SCM are prone to suffer from other diseases [8,20,21]. Several studies have reported correlations between different reproductive and metabolic disorders and respective management strategies to be the key factor for improvement in this area [15,[22][23][24]. This improvement aims not only to reduce the antimicrobial usage in dairy cows but also to meet the requirements of wellinformed and demanding consumers of dairy products. Irrespective of economic aspects, mastitis and its associated implications have detrimental effects on animal welfare [25]. Genetic selection offers a sustainable tool to select animals with decreased susceptibility towards postpartum diseases. Several groups have reported promising associations between Bos taurus autosome 18 (BTA 18) and performance traits [26][27][28][29]. Our own studies revealed differing immune competence of primary mammary epithelial cells (MEC) originating from two BTA 18 haplotypes: half sib heifers inheriting an alternative haplotype of a confirmed quantitative trait locus (QTL) for somatic cell score (SCS) in the telomeric region of BTA 18 showed different somatic cell scores in vivo [30]. The MECs of these heifers differed in their expression profiles after pathogen challenge in vitro [31,32]. These findings indicate a reduced susceptibility towards intramammary infections in heifers inheriting the favorable QTL allele. Another study recently showed that in addition to selection for disease susceptibility, host infectivity should be considered an important aspect in efficiently reducing diseases in cattle [33]. Studying the underlying mechanisms is important to explore the physiological processes, which cause the reported differences between the haplotypes to carve out and benefit from positive implications and to be aware of negative implications of applied selection strategies. Numerous experimental in vivo mastitis models have been 6 established by various researchers over the last several decades, as recently reviewed by Petzl et al. (2018) [34]. However, to the best of our knowledge, no in vivo mastitis model comparing differing BTA 18 haplotype heifers has been performed to date. Therefore, the objective of this project was to establish an in vivo infection model to study the impact of genetic selection for mastitis resistance under defined conditions in dairy heifers. During the selection process of the BTA 18 haplotypes, SCC served as a target phenotype for mastitis incidence and udder health. The severity and resolution of mastitis is known to be heavily influenced by the species of the infecting pathogen [35], and it was shown that Gram-negative pathogens trigger different immune reactions in the host compared to Gram-positive pathogens [16]. To address the pathogen-specific clinical outcome of mastitis, E. coli served as a surrogate pathogen for acute CM and S. aureus as a surrogate pathogen typically causing SCM or mild CM in dairy cows. The suitability of both strains to serve as typical pathogens has recently been demonstrated [16].

Successful establishment of an in vivo infection model
No major pathogens were detected in the last bacteriological examinations of milk samples obtained from each heifer before the start of the challenge experiment. At the start of the experimental challenge, the animals were free of withdrawal periods, and none of the animals showed signs of systemic diseases.
Intramammary infection with S. aureus (n = 24, 96 h) or E. coli (n = 12, 24 h) was induced in all heifers, and samples were obtained every 12 h after IMI, as illustrated in Figure 1. The success of intramammary infection was confirmed by clinical signs 7 of mastitis: changes in milk secretion and udder firmness were observed after challenge with both pathogens (Table 1 and 2). Repeated recovery of the respective pathogen from milk samples in each cow also served to confirm the success of the intramammary infection. Quantification of bacteria was performed via plate count of colony forming units (CFU) per ml ( Figure 2). A significant increase in SCC and a decrease in total milk yield were observed after challenge with both pathogens (Figure 3 and 4). The first signs of mastitis were detected 24 h after S. aureus challenge. As expected, the onset of local changes after intramammary challenge with E. coli was earlier (12 h) and higher in severity compared to animals challenged with S. aureus (Table 1  Comparable systemic effects after pathogen challenge in Q and q To evaluate the severity of the induced mastitis and to detect differences between the divergent haplotypes, the general health condition of the heifers was monitored via such parameters as heart rate, respiratory frequency, inner body temperature, filling and activity of the rumen and feed intake. In the S. aureus group, as well as in the E. coli group, no significant differences concerning general health condition between the divergent haplotypes during the experimental setup (data not shown) were found. Intravaginal temperature during the challenge did not differ between Q-8 and q-heifers, regardless of maximum body temperature (S.

No differences in local clinical signs of mastitis between haplotypes
Local signs of CM were examined every 12 h using a milk secretion and an udder palpation scoring system not only to prove the success and evaluate the extent of the experimentally induced mastitis as described above but also to compare these local effects between the divergent haplotype groups. No differences between the Q-and the q-haplotype were detected in this regard, either within the S. aureus or within the E. coli group (Table 1 and  Lower SCC in Q compared to q after intramammary S. aureus challenge The SCC applies as the main parameter to evaluate udder health because it indicates the inflammatory response during an IMI. At the time point before the intramammary challenge (0 h), all heifers included in this study showed mean low SCC, and no significant differences were detected between the divergent haplotypes (Q: 42.2 * 10 3 /ml ± 10.1 vs. q: 58.5 * 10 3 /ml ± 12.5; P > 0.1; Figure 3). A significant increase in SCC was detected in the infected quarters of all 24 heifers from the S. aureus group 24 h after intramammary challenge. Comparison of the haplotypes revealed significant differences concerning SCC during the course of the experiment: 24 and 36 h after challenge, Q-heifers showed lower SCC levels in milk samples from infected quarters compared to q-heifers ( Figure 3). The SCC of milk samples from noninfected udder quarters did not differ between Q-and q-heifers (data not shown). In the E. coli group, a significant increase of SCC in the milk of the infected quarter was detected earlier compared to the S. aureus group already at 12 h after challenge, but no differences between the divergent haplotypes have been found ( Figure 3).

S. aureus challenge
Total milk yield declined in all heifers after intramammary challenge with either S. aureus or E. coli (maximal decline ~ 35 % and ~ 50 %, respectively; Figure 4). In the S. aureus group, the decline of total milk yield was ~10 % less pronounced in Qheifers compared to q-heifers: the total milk yield in percent 12 h and 24 h after challenge relative to that at the beginning of the challenge was higher, and 12 h and 24 h after challenge, total milk yield in percent relative to that at the start of the challenge was higher in Q-compared to q-heifers (12 h after challenge Q: 92.7 % ± 2.8 vs. q: 82.0 % ± 2.2; P < 0.05; 24 h after challenge Q: 99.8 ± 2.0 vs. q: 90.1 ± 2.8; P < 0.01; Figure 4). In contrast, Q-and q-heifers did not differ concerning the reduction in total milk yield after intramammary challenge with E. coli.

Discussion
The objective of this study was to establish an in vivo infection model to study the impact of genetic selection for mastitis susceptibility under defined conditions in dairy heifers. Thirty-six Holstein Friesian heifers selected for favorable (Q) and unfavorable (q) paternal BTA 18-haplotypes for SCC were included in this study. SCC served as a surrogate trait for mastitis susceptibility, indicating low (Q) or high (q) mastitis susceptibility. At the start of the experiment, no significant differences regarding udder-specific parameters were found between the two haplotypes.
Although Q-heifers had a lower incidence of metritis, lower blood concentrations of betahydroxbutyrate compared to q-heifers, numerical lower incidence of CM and SCM in the postpartum period and significant differences in SCC [36]  Accurate surveillance before and after calving was essential to prepare and synchronize the two haplotype groups, despite differing periparturient performance.
The experiment was conducted as planned in all heifers, and none of the animals had to be excluded from the study based on defined exclusion criteria. After intramammary challenge with E. coli or S. aureus, all animals developed IMI and displayed clinical signs of mastitis in a pathogen-specific manner, and re-isolation of the respective pathogens was successful in all cases. A significant increase in the SCC and decrease in milk yield was assessed after S. aureus challenge, as well as after E. coli challenge. This finding complies with results of previous studies, which compared pathogen-and time-dependent variability of the innate immune response in dairy cows challenged with S. aureus or E. coli [37,38]. It can be generalized from the results that via thorough standardization of the animals and their environment, establishment of an intramammary infection model to study the influence of the respective haplotype was achieved. Synchronization of Q-and qheifers was achieved to such an extent that genetically determined differences were not blurred by environmental effects.
In the present study, the two BTA 18 haplotype groups showed initial differences during their clinical response towards experimental IMI, but these differences were limited to S. aureus IMI. Significantly lower SCC in Q-compared to q-heifers 24 h and 36 h after challenge with S. aureus and significantly lower bacterial load in milk samples 12 h after challenge may suggest differing capacities of antimicrobial reaction patterns between the two haplotype groups. The less prominent decrease in total milk yield 12 h and 24 h after challenge with S. aureus in Q-heifers compared to q-heifers completes this picture of less intense reaction towards intramammary challenge in Q-heifers. This result, in turn, indicates that Qhaplotype heifers are more resilient against IMI than those featuring the qhaplotype. These findings prove that the genetic selection for chromosome BTA 18 haplotypes performed in this study has an impact on experimentally induced mastitis.
It was unexpected that differences in the clinical response between haplotypes during S. aureus IMI could clinically be discriminated only in the initial phase. One decade ago, Rupp et al. (2009) published their study concerning an animal model with two divergent groups of ewes that had been selected for reduced susceptibility towards IMI based on SCS [39]. The selection criteria included extreme breeding values of the respective rams, but no genotyping was applied. The results indicated that ewes from the 'high SCS line' revealed sustained better capacities to eliminate IMI after parturition and during lactation. However, these ewes were only confronted with naturally occurring IMI, and no controlled experimental challenge model was established to carefully scrutinize the genetically determined differential resilience against IMI. To the best of our knowledge, no comparable studies have been published to date reporting on experimentally induced IMI to compare the impact of BTA 18 haplotypes on the resolution and outcome of mastitis in dairy cows.
After IMI with E. coli, no differences were found between Q-heifers and q-heifers based on our diagnostic parameters. This lack of differences might be caused by different reasons. First, the virulence of the Gram-negative E. coli pathogen was higher than that of the S. aureus pathogen, as evidenced by the finding that bacterial counts in milk 12 h after IMI increased by more than orders of magnitude.
This finding caused a stronger assault than S. aureus infection and elicited a strong host reaction that might have overridden the effectiveness of the defense mechanisms influenced by the genetic selection, as applied in this study. Second, the host immune defense against mammary infection with Gram-negative (e.g.,

E. coli) infection is governed and determined by mammary epithelial cells (MEC),
while this dominant cell type of the lactating udder does only play a minor role in defending against Gram-positive mammary pathogens, such as S. aureus or Streptococcus uberis [16,40]. Hence, those immune mechanisms determined by the BTA 18 haplotype having been selected for in this study might not reside in MEC but 13 rather in other immune-relevant cell types. This conclusion agrees notably well with those of Bonnefont et al. 2012 [41], who analyzed MEC from genetically selected ewes of different resilience against mastitis.
IMI models with intramammary application of E. coli that were previously performed within our work group were limited to 24 h [37,42]. In these studies, pathogenspecific reaction patterns were demonstrated, and maximum inner body temperature and changes in milk secretion were detected approximately 12 h to 14 h after challenge. Due to ethical reasons and to preserve comparability of the results with previous studies, E. coli IMI was limited to 24 h in the present study, as well. However, it was unexpected that heifers in this challenge experiment showed macroscopic changes in milk secretion and udder firmness not before 24 h after challenge. One explanation for this delayed reaction of the mammary tissue might be that the animals were early-lactating animals compared to mid-lactating animals, which had been used in previous studies. Vangroenweghe et al. (2004) demonstrated that early lactating primiparous cows showed moderate clinical symptoms towards IMI with E. coli [13], and Van Werven et al. (1997) showed a significant effect of parity on the severity of clinical mastitis induced via E. coli [43]. It is further known that the clinical course of IMI induced via E. coli might be quite severe, but the infection is self-limited, and a high self-cure rate can be observed. Hence, it can only be speculated whether further sampling for a longer period would have revealed more striking differences between the two haplotypes concerning bacteriological and clinical cure of E. coli IMI.
Another unexpected result was the frequent observation of fever during S. aureus IMI. The reason for this finding might be the close monitoring of the inner body temperature via the intravaginal device, recording data every three minutes. In 14 previous studies, rectal temperature was only measured every 6-12 h [37], meaning that potential peaks in between might have been missed.
In their review, Schukken et al. (2011) summarized that long-term self-cure in S. aureus IMI is possible and that persistence of the bacteria within the udder varies from individual to individual [35]. The course of infection/inflammation in this study could not be monitored longer than 96 h due to limitations within the experimental setup; thus, the resolution or persistence of the S. aureus infection in the long term has not been determined. Because ewes from the 'high SCS line' were more susceptible to natural IMI with clinical symptoms [39], it would have been revealing to compare the efficiency of the two haplotypes in eliminating the bacteria from the infected mammary quarter, but this aspect was not within the scope of the present study.

Conclusions
An in vivo infection model to study the impact of specific genetic selection for mastitis susceptibility under defined conditions in dairy heifers was successfully established in this study. Significant differences between the two genetically selected haplotypes focused on SCC and bacterial shedding, which might explain the differing susceptibility towards mastitis. These findings must be supplemented with further data from studies with regard to haplotype-dependent susceptibility towards natural infections and monitoring of subsequent lactations to clarify both the economic feasibility of that genetic selection scheme and the underlying immune mechanisms. The present challenge model is applicable for studying differences between groups of cows embedded in holistic approaches.  At the end of the experiment, the heifers were killed with a captive bolt gun and exsanguination immediately followed by necropsy and tissue sampling for further investigations (Figure 1).

Pathogens for intramammary challenge
The applied strains of S. aureus1027 and E. coli1303 are field isolates from cases of subclinical and clinical mastitis, respectively. Genomic and proteomic characteristics of S. aureus1027, including common virulence markers and virulence gene expression, have been examined [44], and the genome sequence of E. coli1303 has been published [45]. The strains were stocked in a cryobank system at -80°C. To create a stock solution for comparable intramammary challenge doses,

E. coli was cultured on Violet-Red-Bile-Agar (VRB), and S. aureus was cultured on
Columbia-Sheep-Blood-Agar (CSB) and incubated (24 h, 37°C). Afterwards, one colony per bacterial strain was applied into a tube with 10 ml brain heart infusion broth (BHI) and subsequently incubated (6 h, 37°C). Of this solution, 100 µl was applied into 9.9 ml tryptic soy broth (TSB). After 18 h of incubation, the inoculum was prepared to perform serial dilutions. These serial dilutions were plated on VRB For assessing udder health before and during the challenge, all udder quarters were examined for signs of inflammation such as swelling, redness, pain or increased udder surface temperature as well as for the evaluation of milk secretion according to All ethical evaluations were performed as required by the German Animal Care law and associated legislative regulations ("Tierschutzgesetz", https://www.gesetze-iminternet.de/tierschg/BJNR012770972.html).

Consent for publication
Not applicable.

Availability of data and material
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.        Graphical illustration of somatic cell count from Q-/q-heifers after intramammary challenge S Figure 4 Graphical illustration of total milk yield from Q-/q-heifers after intramammary challenge Tota