The Chinese Meishan (ME) breed of pig is unique for many reproductive traits. Compared with Western breeds of swine, ME females reach puberty earlier, ovulate more ova per estrus, and have greater uterine capacity, while intact males (boars) have smaller testes and extremely elevated plasma levels of pituitary-derived glycoprotein hormones. In an effort to identify the genetic mechanisms controlling the elevated plasma levels of pituitary-derived glycoprotein hormones [in particular, follicle-stimulating hormone (FSH)] and to determine whether some of these genetic factors are also responsible for differences in other phenotypes, we scanned the entire genome for regions that affected plasma FSH in boars from a Meishan-White Composite (equal contributions of Chester White, Landrace, Large White, and Yorkshire) resource population. Initially, the entire genome of 121 boars was scanned for regions that potentially influenced plasma FSH. The most significant genomic regions were further studied in a total of 436 boars. Three genomic regions located on chromosomes 3, 10, and X apparently possess genes that significantly affect FSH level, and one region provided suggestive evidence for the presence of FSH-controlling genes located on chromosome 8. The region on the X chromosome also affected testes size. Similar genomic regions to those identified on chromosomes 3, 8, and 10 in this study have been identified to affect ovulation rate in female litter mates, supporting the hypothesis that plasma FSH in pubertal boars and ovulation rate in females is controlled by a similar set of genes.
- quantitative trait loci
- testes size
breeds of swine developed in China were imported into the United States and other countries in the late 1980’s because of the superior reproductive performance of females of these breeds. In general, Meishan (ME) females reach puberty 90 days earlier, ovulate 3–4 more ova per estrus and give birth to 3–4 more live piglets (5, 14) than females of occidental breeds. Despite the economic importance of number born to the swine industry, the use of Chinese germ plasm in commercial swine production has been minimal due to their undesirable growth rate and carcass composition. ME germ plasm has been an invaluable resource for gene mapping studies, as it has been included in nearly every reference and resource population used throughout the world (2, 28, 32, 34).
The ME pig is an interesting resource with which to study the genetics of reproduction. Quantitative trait loci (QTL) have been identified for ovulation rate, uterine capacity and age at puberty in females (34, 37). The ME boar also possesses unique phenotypes for various reproductive characters. Plasma from ME boars contains elevated levels of pituitary-derived hormones [follicle-stimulating hormone (FSH), luteinizing hormone (LH), and thyroid-stimulating hormone (TSH)] and smaller testes than typical western breeds of swine while maintaining normal or above average fertility and libido (21, 22, 38). Of the traits that differ, the only economically important trait is testes size. Unfortunately, acquiring accurate measures of testes size on intact line males is not feasible.
The decision to study FSH as our pituitary-derived hormone was due to many advantages. Levels of plasma FSH within a boar are relatively stable and are less pulsatile than LH (3, 38). Sertoli cells within the seminiferous tubules of testicles are the only cells in males which possess FSH receptors, are essential for spermatogenesis, and their number within a testicle correlates positively with mature testicular size (12). Variation in plasma FSH concentration in boars has been significantly associated with testis size, sperm produced per day per gram of testes, and sperm produced per day (3, 26). In addition, mature boars from an occidental line selected for increased ovulation rate and embryonic survival had more plasma FSH than control line boars (23). Because of these findings, greater understanding of genetic variation of FSH in boars and its relationship to testicular size was the primary impetus for the current research.
In an effort to study the genetic mechanisms affecting the synthesis and secretion of pituitary-derived glycoprotein hormones and testes development, along with its potential relationship with female reproduction, we designed matings from animals in the US Meat Animal Research Center’s (MARC) Swine Resource Population and collected extensive phenotypic data on gilts and boars. A previous study with this population identified eight putative QTL affecting the female reproductive traits of ovulation rate (chromosomes 3, 8, 9, 10, and 16), age at puberty (chromosomes 1 and 10), and uterine capacity (chromosome 8) (34). The present study reports the results of a genomic scan for QTL affecting plasma FSH concentrations in pubertal boars and relates these findings with other QTL previously detected within this population.
MATERIALS AND METHODS
Animals in the first four generations of the MARC Swine Resource Population were utilized. Figure 1 presents a schematic diagram of the matings conducted to produce the animals. The population contained germ plasm from purebred ME and a White Composite (WC) population comprising equal contributions from Chester White, Landrace, Large White, and Yorkshire breeds. Males (n = 5) and females (n = 5) of each breed were randomly selected to initiate the population. Reciprocal matings were conducted to produce 41 F1 females. The F1 females were mated back to either ME or WC boars used in the previous generation, yielding progeny that were either ¾ WC ¼ ME or ¼ WC ¾ ME in the backcross generation (BC). The F3 generation was produced by mating ¾ WC ¼ ME to ¼ WC ¾ ME individuals. Thus the final generation was ½ WC ½ ME. Both F1 and BC females were mated to produce two litters. BC litters were produced in two farrowing seasons, and F3 litters were produced in three farrowing seasons.
Collection of genotypic data.
Genetic markers for the initial scan were described in Rohrer and Keele (33). Procedures utilized to genotype microsatellite markers were described in Rohrer et al. (32). Briefly, reaction conditions contained 12–15 ng of porcine genomic DNA, 5 pmol of each primer, and standard Taq polymerase buffer. Thirty cycles were performed with 30 s annealing, 60 s elongation, and 30 s denaturing times. Annealing temperatures were marker dependent and varied between 55°C and 65°C. Products were labeled by directly incorporating 0.1 μCi of [α-32P]dATP and visualized with autoradiography.
A two-tiered data analysis procedure was implemented. In the initial scan, all animals with pubertal plasma FSH levels recorded from the BC generation (n = 33) along with 88 animals with similar measurements from the F3 generation were genotyped for 157 markers spaced throughout the genome. Markers were selected to span the entire genome with a maximum distance between adjacent markers of 20 cM. Approximately one-third of the selected animals in the F3 generation were chosen based on the mean pubertal FSH values for the entire litter, whereas the other boars were selected based on the performance of gilt litter mates for either ovulation rate or age at puberty. Therefore, one-sixth of the F3 boars were from litters with the highest pubertal FSH, and one-sixth were from litters with the lowest mean FSH. This form of selective genotyping should not have significantly affected our initial results, because the majority of animals genotyped were not selected based on FSH levels.
Genotypic data were then analyzed on all boars with phenotypic data for regions of the genome that showed promising results in the initial scan of the genome. A minimum of four markers used previously that flanked the putative QTL were genotyped across the remainder of the F3 boars (n = 315). For the regions on chromosome 8 and X, additional markers were genotyped across the entire population.
Blood collection and FSH radioimmunoassay.
Blood was obtained by venipuncture from all boars at least twice during pubertal development (5–8 mo of age). Individual differences in plasma FSH concentration are consistent from pubertal into postpubertal development of boars and can be described with limited numbers of samples (9, 39). Plasma was frozen and subsequently evaluated for concentrations of FSH as described (9, 19). The radioimmunoassay used antisera AFP-C5288113 in conjunction with iodinated ovine FSH (AFP-6446C) and porcine USDA-B1 as the reference preparation. The interassay coefficients of variation were 13% and 12.9% for pools of plasma from boars and castrated males that assayed 257 and 1,379 ng/ml, respectively. Pubertal plasma FSH was the mean value for duplicate assays from each blood sample collected between 5 and 8 mo of age.
Collection of testes weight.
Postpubertal paired testes weight was recorded for 125 animals (31 BC and 94 F3 animals). Data were collected by either castration of the animal after induction of anesthesia or killing the animal and weighing the removed testes. Weight of the animal at the time of castration was recorded. Daily sperm production per gram of testes was then determined by counting homogenization-resistant spermatid nuclei in testicular homogenates (1, 29).
An interval mapping approach was utilized for all QTL analyses described below. The methodology utilized a least-squares regression approach which fit coefficients for the probability an allele originated from a ME animal in the P generation (13). The procedures were as described in Rohrer et al. (34). The five most significant chromosomal regions were selected to be analyzed in the follow-up study.
The guidelines proposed by Lander and Kruglyak (20) for determining significance and reporting results were followed. They proposed to report significant values as expected number of false positives per genome scan and categorize results as suggestive when the value was <1.0 and significant when <0.05. Lander and Kruglyak (20) also provided an equation for determining the expected number of false positives (modified for F ratios) as (C + 2·G·ρ·f·dfn) × [1 − probf(f, dfn, dfd)], where C = haploid chromosome number (19 for pigs), G = genome size in morgans (25 for MARC map), ρ = an autocorrelation coefficient that depends on the structure of the population studied (a conservative value of 2.0 was used), f = F ratio, dfn = numerator degrees of freedom, dfd = denominator degrees of freedom, and probf(f,dfn,dfd) is the nominal probability of the F ratio test statistic. Because this calculation uses an F ratio that assumes normality and the data were skewed to the right, these values are approximations. This procedure is the most conservative test for QTL studies, but was designed to be robust to the typical QTL procedures implemented by most scientists and used in this study. Most QTL studies are initiated by analyzing genotypic data from markers evenly spaced across the genome (marker intervals vary from 5 to 30 cM, depending on the study). When interesting intervals are identified, more markers are genotyped to refine the estimated position of the QTL. The formula used is not sensitive to marker density, so the addition of new markers does not change the critical thresholds. Other procedures would require determining new thresholds for the new data set.
Similar statistical procedures were utilized for both the initial scan and the first analysis of the follow-up study. However, the results of the follow-up analysis definitively identified a QTL residing on the X chromosome located near position 80 on the linkage map. Therefore, pubertal FSH concentrations were adjusted for the breed of chromosome inherited at X:80, and the final analyses for autosomal regions were conducted on the adjusted FSH values, to reduce the residual variance and improve the overall fit of the final model. Because the results based on the unadjusted vs. adjusted FSH values varied depending on the genomic location analyzed, the initial scan was reanalyzed using the adjusted phenotypes, and one additional region was selected to be further studied. Only associations that reached the suggestive level of significance (one false positive result per genome scan) in the final data set based on all phenotyped animals are reported.
Three orthogonal linear contrasts that explained all of the genotypic variation for a given locus were fitted. The three contrasts were for additive genetic effects, dominance deviations, and a third contrast comparing the difference between heterozygous individuals that received the ME allele from their sire vs. those heterozygotes that received the ME allele from their dam. This final contrast would be most significant if a genetic imprinting mode of inheritance was present. Genetic effects were eliminated if the reduced model was more significant at the genome-wide level. The three models that were compared were ones that fit all three genetic contrasts, one fitting both additive and dominance terms and the purely additive model.
To determine whether the genomic regions that affected plasma FSH also controlled testes size, similar analyses were conducted on the four regions with putative QTL for FSH using paired testes weight as the dependent variable. A covariate for body weight was included to adjust for differences in the size of the animals at the time of castration. Testes weights were also adjusted for the X chromosomal region, and the adjusted paired testes weights were analyzed.
Considerable variation for pubertal FSH was observed in the data. The mean FSH of all boars in this study was 388 ng/ml, with a range of 48–1,964 and a standard deviation of 327.4. The mean and range of FSH values were similar in BC and F3 pigs. The five highest F ratios from the initial scan were located on chromosomes [Sus scrofa (SSC)] 7, 8, 10, 18 and X. Regional localizations for each peak was the telomeric portion of SSC 8p, 10q, and 18q, the centromeric region of SSC X and at both ends of SSC 7. Both peaks on SSC 7 were similar in magnitude, and the profile of the F ratios spanning chromosome 7 was not as expected a priori unless two QTL actually resided on this chromosome. Therefore, both regions on chromosome 7 were included for additional analyses.
The first analysis conducted on the dataset containing genotypic data from all animals with phenotypes presented convincing evidence (F1,429 = 88.7; probability of a false positive ∼2.5 × 10−15) of a QTL located near SSC X:80 (locations presented as chromosome:location; Table 1). Figure 2 presents the mean FSH values for boars that inherited the X:80 segment from the ME (probability of inheriting the ME allele of >0.90) vs. the WC (probability of inheriting the ME allele of <0.10) breed. The probabilities were determined by the genotypic marker data and were the same values used as regression coefficients for the QTL analysis. The variance within the animals inheriting ME alleles was four times greater than in the sample of animals with the WC allele. The different variances could be due to incorrect assignment of breed of origin for more boars coded as ME than WC (expectation for misclassified animals was 10%); however, a more likely explanation is that the variance in plasma FSH is proportional to the mean. The mean FSH for the boars that were not assigned a breed of origin (probability of ME less than 0.90 but greater than 0.10) was intermediate at 405.2 with an intermediate variance.
The analyses for chromosomes 7, 8, 10, and 18 with the adjusted phenotypes detected one significant QTL located at 10:101 and one suggestive QTL at 8:16 (Table 1). The region on chromosome 18 fell just below our threshold (P genome = 1.1) but may still possess a gene affecting FSH. Both regions on chromosome 7 were not significant in analyses using the adjusted phenotypes. Analysis of the adjusted data reduced the mean squared error by more than 15% and should have increased the power of subsequent analyses.
Because of the perceived improvement of power achieved by analyzing the adjusted phenotypic data, the initial scan was analyzed using the FSH measurements adjusted for the QTL located on SSC X. The four most significant genomic regions of this scan were located at SSC 3p, 8p, 10q, and 18q. The phenotypic variation associated with the X chromosome was effectively removed by the adjustment. Genotypic data were collected for the remaining F3 animals for markers surrounding the SSC 3p region. Analysis of the complete data for chromosome 3 identified a significant QTL at 3:49 with an expected number of false positives of 0.048.
The mode of inheritance of these QTL were quite different. Since males only possess one copy of genes located around X:80, there is no opportunity for intralocus interactions. For the QTL located at 3:49, the inheritance appeared to be completely additive as the addition of the other genetic effects decreased the significance of these regions. However, the predominant genetic contrast for the regions at 8:16 and 10:101 was the contrast between the alternate heterozygotes. Least square means for each genotypic class at both loci are presented in Fig. 3. In both cases the FSH levels of heterozygous animals were considerably higher than either of the homozygotes. The heterozygote that received a ME allele from its dam and a WC allele from its sire had greater FSH than the heterozygote that received alleles in the opposite alignment. There is no established genetic mechanism that would account for these genotypic means. Despite the fact that the average of the heterozygous animals exceeded the average of the homozygotes for both SSC 8:16 and 10:101, individual means comparisons using linear contrasts did not detect a significant difference between the predicted FSH value for heterozygous animals receiving the WC allele from their dam and the ME allele from their sire than either homozygous class (Fig. 3). This mode of inheritance resembles polar overdominance, which has been shown to affect muscle hypertrophy in sheep (6).
The analysis of testes size revealed a highly significant association of the centromeric region of the X chromosome (position 80 cM) with testes size (F1,113 = 59.0; P genome 3.7 × 10−8). Inheritance of a ME allele at SSC X:80 resulted in a reduction of 193 g in paired testes weight relative to boars that inherited a WC allele at SSC X:80. After adjusting the measurements for the SSC X QTL, no other associations were detected that reached nominal significance at the position of the FSH QTL. No associations significant at either the genome-wide or nominal level were detected between the four chromosomal regions affecting plasma FSH and daily sperm production per gram of testes.
Germ plasm from Chinese pigs has been studied to determine the basis for their early onset of puberty and increased litter size with genetic and physiological methods. This study represents the first step to determine the genetic mechanisms and genes responsible for the altered function/expression of the pituitary-derived glycoprotein hormones observed in ME boars. Four genomic regions were detected which presented suggestive or significant evidence of the presence of a QTL affecting plasma levels of FSH in pubertal boars.
For each region detected, it is nearly impossible to statistically determine whether the effects observed are due to a single gene or a complex of genes acting in concert. Similarly, for associations of different traits to similar regions of the genome it is impossible to differentiate between pleiotropic effects (one gene affecting multiple traits) of a single gene vs. separate genes having an effect on each trait independently. Factors to be considered when assessing the potential for pleiotropic gene effects are whether the mode of gene action is similar, the distance between the estimated positions of the QTL, and whether the direction of the effects is similar to what should be expected based on genetic and/or phenotypic correlations.
It is clear that a gene or gene complex that affects FSH levels in boars is present on the X chromosome located near the centromere. This particular region has previously been identified as affecting weight at 8 wk of age (31), back-fat thickness (33), and possibly ovulation rate and uterine capacity (34). Strong evidence for a QTL located in this region of the X chromosome was also detected for weight of testes, but no effect was detected on daily sperm production per gram of testes. Therefore, this chromosomal region must have an effect on total daily sperm production due to its effect on testicular size. The evidence for small testicular size in boars with unusually elevated plasma FSH is strong in populations containing ME germ plasm as well as populations of western breed types (9, 39).
Comparative mapping has shown that the centromeric region of the porcine X chromosome most likely corresponds to the centromeric region of human X chromosome (15). Unfortunately, this region of SSC X displays a low level of genetic recombination, so the gene(s) that causes the observed effect on FSH, testes size, and body composition could lie anywhere between SSC Xp1.2 to Xq2.1 (∼28% of the entire chromosome). One gene that maps to this location and may be a suitable positional candidate gene would be the androgen receptor (30). Mutations in the human androgen receptor gene have been identified as the primary cause for various forms of androgen insensitivity (11, 24). Androgen insensitivity often affects testes size, testes function, and occasionally plasma FSH concentrations in men (7, 25, 36).
In mice, the authors of a study involving various crosses between BALB/c and CBA mice concluded that the origin of the X chromosome significantly affected testes weight, especially at ages less than 8 wk (16). However, they found the line of Y chromosomal origin had a greater impact on mature testes size than the X chromosome. Evidence of other autosomal genes influencing testes size was also detected (16). Breed of origin for the Y chromosome could not be evaluated in the present study as too few boars had a Y chromosome from the ME breed. Paigen et al. (27) identified a QTL causing cholesterol gallstones in mice located on the region of the mouse X chromosome orthologous to the present QTL for FSH. Their QTL overlapped a QTL for obesity in mice, and the authors speculated that the causative gene affected cholesterol metabolism. Paigen et al. (27) speculated that the androgen receptor could be the causative gene. However, correlations between plasma FSH and cholesterol metabolism are not documented.
Regions that were detected to affect plasma FSH were similar to locations associated with QTL for ovulation rate on chromosomes 3, 8, and 10. It remains to be determined whether the same genes affect both traits for any of these regions. The modes of inheritance for the ovulation rate and FSH peaks located on chromosome 3 were similar, and the WC contributed the allele that increased ovulation rate and plasma FSH, supporting the hypothesis that these results are due to pleiotropic effects of a single gene. If both QTL are caused by one gene, then the mechanism behind this QTL’s effect on ovulation rate is possibly related to regulation of FSH in the female with a similar effect on FSH expression in boars.
The location of the peak for plasma FSH in this study on chromosome 10 (position 101 cM) is located between QTL for ovulation rate (position 89 cM) and age at puberty (position 125 cM) in females (34). The results from Rohrer et al. (34) indicated that ME alleles decreased ovulation rate and age at puberty. Further studies are needed to determine whether there are three QTL segregating in this population, or if the effect detected for FSH is caused by the same gene(s) as one or both of the QTL detected for female reproduction traits. Since the FSH measurements were recorded during the pubertal phase for these boars, it is likely that the gene affecting age at puberty in females also controls age at puberty in males and would affect the plasma FSH levels recorded in this study. Earlier age at puberty would be expected to produce higher plasma FSH levels in boars.
If both QTL for female reproduction also affect male plasma FSH concentrations, then the ME breed would have contributed one allele that decreases FSH levels (located at the ovulation rate QTL) and one that increases FSH levels (at the QTL for age at puberty). The opposite configuration of alleles would be present in the WC breed. When two QTL are linked with each breed contributing one positive and one negative allele (repulsion phase), then heterozygotes at both loci will exhibit pseudo-overdominance (8) or the performance of the heterozygote will surpass the performance of both homozygous genotypes. However, the difference in performance between the two heterozygous genotypes must be caused by an unknown genetic phenomenon or estimation error. When SSC 10 was analyzed with only the additive genetic effect in the statistical model, the estimated effect of the ME allele was negative at the location of the ovulation rate QTL (approximately −46 at position 89 cM) and positive at the age at puberty QTL (∼105 at 125 cM). Although these effects were not statistically significant, they show a trend that may be biologically important and provides evidence that this QTL result is actually caused by two different genes linked in repulsion phase.
The most likely position for the QTL detected on chromosome 8 affecting plasma FSH is located between a QTL for ovulation rate (position 5 cM) and one for uterine capacity (position 71 cM). Recent data have shown that selection for increased uterine capacity produces a correlated increase in plasma FSH in prepubertal gilts, similar to selection for increased ovulation rate (Ford JJ and Wise TH, unpublished observations). Similar to the SSC 10 results, these QTL are also in repulsion phase, as the ME breed contributes lower number of ova ovulated but greater uterine capacity. The estimated location for the plasma FSH QTL is much closer to the ovulation rate QTL than the uterine capacity QTL and may indicate that the QTL detected is the same gene that affects ovulation rate due to the coarse resolution of QTL mapping. Unlike the SSC 10 example, the additive model does not identify two different peaks in repulsion phase. Therefore, this observed effect may actually be due to a single gene that affects plasma FSH in a polar overdominant genetic mechanism (6). Despite the different types of gene action, this same locus may or may not affect ovulation rate. For the callipyge locus in sheep the mode of gene action is different for meat tenderness traits than for muscle mass (10).
Initial efforts to correlate plasma FSH concentrations and ovulation rate focused on cycling females, and no direct conclusions could be drawn, possibly due to maturation of negative feedback regulation of FSH secretion. In prepubertal female pigs, lines with greater ovulation rate have greater plasma FSH concentrations during their prepubertal development (4; Ford JJ, unpublished data). However, after puberty the association of plasma FSH with ovulation rate does not exist or is restricted to very specific stages of the estrous cycle (17, 18, 23). In contrast, the positive association of plasma FSH concentrations and ovulation rate of their respective line becomes apparent during pubertal development in boars (4) and continues after puberty.
These results build a plausible story relating FSH levels in the plasma of pubertal boars with genetic mechanisms controlling ovulation rate in females. The occurrence of QTL affecting FSH in boars at the same position as QTL for ovulation rate implies that these two traits should have a positive genetic correlation. This was not detected in an analysis of animals in this population (35); however, the population structure was not sufficient to estimate genetic parameters reliably for correlated traits which cannot be measured on the same animals. The QTL results do substantiate the findings of Mariscal et al. (23) and Cassady et al. (4) who detected a significant correlated response in FSH level of boars in lines selected for increased female reproduction.
It is now possible to select positional candidate genes from the human gene map which may contain the genetic variation causing the observed effects. Selection of candidate genes for a given QTL will be greatly facilitated by prior knowledge of other traits affected. As previously mentioned, the androgen receptor lies in the chromosomal region detected on the X chromosome and potentially affects all the traits linked to that chromosomal region (FSH, testes size, and body composition). It will need to be determined whether the quantity and/or quality (amino acid composition) of the protein produced by the candidate gene differs between ME and WC pigs. The quality of the protein produced can be assessed by sequencing the coding region for the gene in both breeds. Quantitation of a specific protein is more complicated as it requires a precise method to measure the quantity of protein and knowledge of the important tissue(s) and critical developmental times for protein expression. If a difference is detected, then determination of the biological nature of how this altered protein expression produces the observed phenotype is required to verify that the genetic variation identified actually causes the phenotypic variation studied.
We acknowledge the technical assistance of S. Hassler, A. Kruger, and K. Simmerman; the MARC swine personnel for care of animals; S. Kluver for manuscript preparation; and USDA/NIDDK for FSH antisera, ovine FSH for iodination, and porcine FSH reference preparation.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: G. A. Rohrer, US Meat Animal Research Center, Spur 18D, PO Box 166, Clay Center, NE 68933-0166 (E-mail:).
- Copyright © 2001 the American Physiological Society