Anal atresia is a rare and severe disorder in swine occurring with an incidence of 0.1–1.0%. A whole-genome scan based on affected half-sibs was performed to identify susceptibility loci for anal atresia. The analysis included 27 families with a total of 95 animals and 65 affected piglets among them. Animals were genotyped for 126 microsatellite markers distributed across the 18 autosomal porcine chromosomes and the X chromosome, covering an estimated 2080 cM. Single-point and multipoint nonparametric linkage scores were calculated using the computer package ALLEGRO 1.0. Significant linkage results were obtained for chromosomes 1, 3, and 12. Markers on these chromosomes and additionally on chromosomes for which candidate genes have been postulated in previous studies were subjected to the transmission disequilibrium test (TDT). The test statistic exceeded the genomewide significance level for adjacent markers SW1621 (P = 7 × 10−7) and SW1902 (P = 3 × 10−3) on chromosome 1, supporting the results of the linkage analysis. A specific haplotype associated with anal atresia that could prove useful for selection against the disorder was revealed. Suggestive linkage and association were also found for markers S0081 on chromosome 9 and SW957 on chromosome 12.
ANAL atresia (AA), a naturally occurring nonsyndromal disorder in pigs, has been described by several authors (Henricson 1963; Norrish and Rennie 1968). The prevalence of the disorder varies between 0.1 and 1.0% in different breeds (Triebler 1984; Stigler et al. 1991; Thaller 1992; Thaller et al. 1996; Hori et al. 2001). A similar anomaly is reported in humans with a prevalence of 0.048% (Stoll et al. 1997). The spectrum of the different outcome varies from ectopic anus to atresia ani and recti with urogenitale fistula to complex deformations of the cloaca. In humans, anorectal malformations are often associated with syndromes: for example, Hirschsprung disease (Mahboubi and Templeton 1984), VACTERL (Khoury et al. 1983), and Townes-Brocks (Townes and Brocks 1972) or the Pallister-Hall (Hall et al. 1980) syndrome.
The etiology and causal genetic and environmental factors of anorectal malformations remain poorly understood. A closer examination of affected piglets showed that the main developmental defect occurs during embryogenesis at the cloacal plate, where the dorsal part of variable size was missing (van der Putte 1986; Kluth et al. 1995). As a result, the migration of the dorsal cloacal plate and neighboring structures to the body surface of the tail groove does not take place. Different types of anorectal malformations have been reported, depending on the extent of the cloacal plate defect, and vary from minor abnormalities such as anal stenosis to more serious defects such as imperforate anus of low (atresia ani) and high types (atresia recti).
Although the exact genetic background of nonsyndromal anorectal malformations in swine is unknown, it appears to be complex. Several studies have been conducted to investigate the mode of inheritance and different genetic models were postulated for the disorder. A monogenic recessive model with incomplete penetrance was presumed by Kinzelbach (1932), Norrish and Rennie (1968), and Triebler (1984). By contrast, two-locus recessive models were hypothesized, assuming either complete (Hamori 1962) or incomplete penetrance (Henricson 1963). The comprehensive comparison of various genetic models by means of segregation analysis and based on >30,000 litters favored a mixed genetic model, where a single locus and polygenic effects are involved in the development of the congenital defect (Thaller et al. 1996). A recessive locus acting together with a dominant one was assumed by Hori et al. (2001). Using a resource population with an artificially increased incidence for AA, Hori et al. (2001) mapped a susceptibility locus for the disorder to Sus scrofa chromosome (SSC) 15 at marker position SW 2072 with a LOD score of 2.7.
Several chromosome deletions in humans are known to cause syndromal anorectal malformations: on HSA 1q (Takano et al. 1997), HSA 7q36 (Lynch et al. 1995; Seri et al. 1999), and HSA 13q (Bartsch et al. 1996; Kuhnle et al. 2000). In addition, a tetrasomy on HSA 12p (Steinbach and Rehder 1987) is also known to be associated with anorectal malformations and syndromes. Danforth's short tail (SD) mouse is the best-known animal model, which explains the etiology and pathogenesis of anorectal malformations in mammalian species (Danforth 1930). The SD locus, the molecular nature of which is still unknown, is likely to be responsible for the phenotype imperforate anus with fistula to the urinary tract (Gluecksohn-Schoenheimer 1943). In mice, the SD locus has been mapped close to the genes PAX8 and SPNA1 on chromosome 2 by linkage analysis (Koseki et al. 1993; Lane and Birkenmeier 1993). In swine, PAX8 (paired box gene 8) has been mapped to SSC 12q11–q15 by Marklund et al. (1998) and SPNA1 (spectrin α-chain) to SSC 1q21 by Lee et al. (2001), respectively.
Knock-out experiments for the GLI-Kruppel family member 2 (GLI2) and GLI-Kruppel family member 3 (GLI3) transcription factors of the sonic Hedgehog (SHH) pathway show that lack of these genes results in several forms of anorectal malformations, including imperforate anus and recto-urethral fistula and cloacal abnormalities (Kimmel et al. 2000). Genes of the SHH pathway are therefore strong candidates underlying the susceptibility to these disorders. In addition, members of the retinoid-mediated signaling pathway are also promising functional candidate genes. In mice and rats, it is known that a dose of 100 mg/kg all-trans retinoic-acid (ATRA) (Padmanabhan 1998; Bitoh et al. 2001) or 60 mg/kg etretinate (Kubota et al. 2000) during gestation can cause anorectal malformations with urogenital fistula in the offspring. An impaired distribution of retinoid acid receptor-α (RARA) could also be observed in the hindgut of affected embryos (Bitoh et al. 2001).
In this study, we present a whole-genome screen designed to identify genomic regions that might affect the development of anorectal malformation in pigs. The results provide a basis for further positional and functional candidate-based association studies. We carried out a two-stage genome scan as proposed by Hauser et al. (1996) and Guo and Elston (2001). To identify susceptibility loci for AA in the Bavarian swine population, we used a nonparametric approach for linkage analysis implemented in the software package ALLEGRO 1.0 (Gudbjartsson et al. 2000). Linkage and association studies based on the transmission disequilibrium test (TDT) statistic (Spielman and Ewens 1996) were performed for chromosome regions either with indicative linkage results (nonparametric linkage, NPL, > 1) or that have been proposed in other studies (Hori et al. 2001).
MATERIALS AND METHODS
Collection of affected piglets with verified diagnosis:
Due to the generally low incidence of AA and early postpartum mortality of affected piglets, a populationwide collection system was established to get access to the animals. In cooperation with Bavarian farmers and breeding organizations, a total of 439 putatively affected piglets were collected. All were sent to the veterinary research institute “Landesuntersuchungsamt für das Gesundheitswesen Südbayern e. V.” to be diagnosed for different forms of AA. Of these animals, 12 piglets were reported as having findings other than AA such as peritonitis or rectal prolapse, leaving 427 affected animals. An excess of affected males (89.1%) was observed. About 70.0% of the affected piglets were Piétrain × German Landrace crossbred, which is the most common commercial cross in Bavaria. The predominant sire breed was Piétrain (91.0%) followed by German Landrace, Large White, and some hybrid boars.
In a first step, extensive paternity testing was conducted to select paternally related affected piglets. Twenty microsatellite markers yielded an exclusion probability for a wrong paternity of 99.98%. A total of 35 families consisting of 88 affected piglets (78 atresia ani, 10 atresia recti), 34 boars, and four sows remained for genotyping and were included in the analyses. The explicit family structures used for linkage analysis are listed in Table 1. Most of the families were paternal half-sib families with up to five affected half-sibs. Seven full-sib families could be established due to the fact that 2 affected piglets occurred within the same litter. Some fathers of full-sibs had additional affected offspring from further matings; hence there is a mixture of full- and half-sibs among the offspring of these boars. Finally, a small four-generation pedigree showed up with a father, two sons, and one grandson having together three affected offspring. A subset comprising 27 families with 65 affected piglets, 27 boars, and 3 sows was used for the initial genome scan (Table 1). During the first scan, 8 additional families emerged from the ongoing sampling procedure and provided an independent data set for confirmation studies on SSC 1. To investigate sex-linked inheritance, 9 families with two affected full-sibs were genotyped together with the sows at markers on the X chromosome (SSC X).
DNA isolation and genotyping:
Total genomic DNA was isolated from muscle tissue of the piglets and from semen of the boars following standard protocols. The DNA was quantified using a Hoefer DNA Quant Fluorometer. PCR reactions were performed in a 20-μl total volume with 50 ng of template DNA, 5 pmol each of forward and reverse primer, and a final concentration of 50 mm KCl, 10 mm Tris-HCl (pH 8.3), 200 μm of each dNTP, 1.5 mm MgCl2, and 0.5 units of Perkin-Elmer (Norwalk, CT) AmpliTaq Polymerase. In each reaction, the forward primer was 5′-labeled with one of three fluorochromes (6-FAM, TET, or HEX). Altogether, 117 autosomal and 9 X-chromosomal microsatellite markers were used for the genome scan as well as for confirmation studies on SSC 1. Each microsatellite was amplified separately and then multiplexed into panels of 8–12 PCR reactions per animal on the basis of pooling ratios designed to normalize signal intensity across markers before being run on ABI 377 sequencers along with the size-standard ladder GENESIZE TAMRA 500. The gels were tracked and reviewed using GENESCAN 3.1 software and the alleles were called using GENOTYPER 2.5 software (Applied Biosystems, Foster City, CA). Allele calling by the latter program was accomplished by standardized analysis settings and followed by manual inspection and correction. All genotypes and alleles called were checked for Mendelian inconsistencies.
The order of the 126 preselected microsatellite markers and the genetic distances between them were taken from the published MARC U.S. Department of Agriculture maps (Rohrer et al. 1996). They cover all 18 autosomal chromosomes, the X chromosome, and the pseudoautosomal region of the Y chromosome; their total length summed to 2080 cM. The average marker interval was 19.62 cM and ∼95.0% of the intervals were <33 cM.
The software used for nonparametric linkage analysis, ALLEGRO 1.0 (Gudbjartsson et al. 2000), requires allele frequencies for markers if one parent is unknown and marker genotypes of the other parent and offspring do not allow the transmission of parental alleles to be traced unambiguously. As our design was based on affected paternal half-sibs with missing genotypes of sows, the maternal allele frequencies were estimated directly on the basis of the entire data set of genotyped animals. By this procedure, only alleles for which the maternal origin could be unequivocally determined were considered.
We used a nonparametric approach based on identical-by-descent (IBD) allele sharing among affected individuals. Two-point and multipoint nonparametric linkage analyses of the pedigrees were carried out with ALLEGRO, which is a follow-up of GENEHUNTER (Kruglyak et al. 1996). Briefly, inheritance vectors are set up that describe all possible outcomes of meioses within a specific pedigree. A priori, all inheritance vectors have equal probability and marker information is used to assign a posteriori probabilities to each of the inheritance vectors. Next, as inheritance vectors provide complete knowledge of inheritance, it can easily be determined for each of the vectors whether affected animals share alleles IBD, i.e., that come from a common founder. ALLEGRO allows for two different scoring functions, Spairs (Weeks and Lange 1988) or Sall (Whittemore and Halpern 1994), for allele sharing. Whereas Spairs considers only pairwise allele sharing among affected animals, Sall puts special emphasis on pedigrees where more than two affected animals share the same founder allele withwhere n is the number of affected individuals, 2f is the number of founder alleles in the pedigree, h is a selection that picks one of the two alleles of each affected one, and bi(h) is the number of times founder allele i appears in h for a given inheritance vector v. The sum is taken over 2n possible ways of choosing h. Thereafter, a so-called Z-score is calculated by summing the values of the scoring functions, weighted with the respective posterior probability, over all inheritance vectors. Finally, the Z-scores of each pedigree are standardized and their weighted sum over the independent pedigrees provides the NPLall test statistic that approximately follows a standard normal distribution under the null hypothesis.
The information content supplied by ALLEGRO provides a measure with which to judge the extent to which the inheritance information present in the pedigrees could be extracted by the given markers compared to a fully informative situation. It therefore allows an overview of which chromosomes were sufficiently covered by markers and in which chromosomal regions additional markers might be helpful when conducting a genome scan. The information content at a specific chromosomal position, x, is defined according to Kruglyak et al. (1996) aswith E being the entropy measured aswhere Pi is the probability of the inheritance vector i, and E0 is the entropy if no marker information is available. The information content ranges from 0 to 1, with the latter value reflecting complete knowledge of inheritance.
The TDT of Spielman and Ewens (1996) was applied to all markers located either on chromosomes showing a NPL statistic >1.0 or lying within chromosomal regions harboring suspected candidate genes. The TDT tests for deviation from Mendelian expectations with regard to transmission of alternative marker alleles from parents to affected offspring. A significant result simultaneously identifies linkage and association between marker alleles or haplotypes consisting of multiple tightly linked markers with the putative disease locus. The test statistic TmHet (Spielman and Ewens 1996) used is defined aswith t being the number of alleles per marker. Throughout the analysis, a nominal significance level for linkage and association of 0.05 was applied first. Additionally, a Bonferroni correction (Snedecor and Cochran 1980) was used to account for multiple testing m markers on the same chromosome, yielding an approximate chromosome-wise significance level of α = 0.05/m.
Figure 1 shows the profiles of the multipoint NPL scores and the respective information content for each chromosome obtained from the initial genomewide scan based on 27 families. The information content was 0.52 on average and, with the exception of SSC 17, <0.20 for only a few chromosomal segments. These values are comparable to a first scan in an asthma study conducted in humans (Laitinen et al. 2001). Linkage signals exceeded the nominal significance level of 5.0% for three chromosomal regions with maximum multipoint NPL scores of 2.57 (P = 0.006) at marker SW1621 on SSC 1, 1.94 (P = 0.028) at marker S0229 on SSC 12, and 1.90 (P = 0.031) at marker S0002 on SSC 3. Following our multistage mapping approach, we considered the results for SSC 1 as the most promising and concentrated our further efforts on that chromosome. First, we genotyped eight additional microsatellite markers to increase the information content to 0.74 compared to 0.47 in the initial scan. Again, the most likely position for a disease locus was at marker SW1621, but with a slightly lower NPL score of 2.39 (P = 0.01) (Table 2). Next, we included eight new families and further half-sibs to existing families that were consecutively sampled during genotyping and data analysis of the first stage. The separate analysis of the eight new families genotyped at 14 markers did not confirm linkage results on SW1621 but showed a maximum NPL score of 2.61 (P = 0.006) at marker SW2185 12 cM proximal to SW1621 (data not shown). Reasons for this finding could be the much lower information content of 0.54 at SW1621 compared to 0.91 at SW2185 in the new families. However, analyzing all 35 families together with a total of 14 markers yielded significant multipoint NPL scores >2.0, spanning a chromosomal region of ∼20 cM in length and including markers SW2185 and SW1621 (Figure 2).
Our multipoint NPL scores rely on marker distances extracted from publicly available marker maps that might differ from those in our own material. We investigated possible consequences of this by calculating single-point NPL scores across the genome (data not shown) and for the respective data sets on chromosome 1 (Table 2). These linkage analyses do not depend on the order of markers or on the relative marker distances. In general, similar profiles of test statistics emerged across chromosomes, but with clearly lower values for single-compared to multipoint NPL statistics throughout the analyses. Thus, we concluded that the marker maps and marker distances used were adequate and that the applied multipoint approach was justified. The results further demonstrate the greatly improved power achieved by the simultaneous use of multiple markers, especially for the coarse scale mapping at the first stage.
Next, we addressed the question of whether or not atresia ani and atresia recti have the same genetic foundation and differ only in the degree of anal malformation, e.g., due possibly to unknown environmental factors, or whether genetic heterogeneity plays a role. As can be seen in the materials and methods section, few affected piglets were diagnosed with the more severe atresia recti phenotype. It was therefore not possible to analyze these animals separately. Instead, we reanalyzed the data sets excluding atresia recti animals and compared the resulting test statistics. All NPL scores were lower when atresia recti animals were omitted, with the exception of marker S0229 on SSC 12 with a slightly higher NPL of 1.98 (P = 0.025) compared to 1.94 (P = 0.028) in the original analysis. In theory, higher values for the test statistic would be expected if different genes determine the outcome of distinguishable disease phenotypes. By contrast, the power will decrease with a reduced number of affected pairs if the phenotypes are virtually identical with respect to their genetic determinism. Our results therefore support the hypothesis of the same genetic background for atresia ani and atresia recti.
The TDT was applied to markers in genome regions determined either by an indicative result in the linkage analysis or by comparative mapping of human candidate regions to the porcine chromosomes. We included chromosomes 1, 3, 8, 9, 12, 13, 15, 18, and X in our association studies. Table 3 shows markers with chromosome-wise significant TDT results. After accounting for multiple testing, two markers, SW1621 and SW1902 on SSC 1, showed significant TmHet statistics at the 1.0% chromosome-wise significance level. Linkage and association were also found for markers S0081 on SSC 9 (pnominal = 9 × 10−4) and for SW957 on SSC 12 (pnominal = 0.003) at the 1.0 and 5.0% chromosome-wise significance thresholds, respectively. The TDT statistics for markers on SSC 3 were not significant, although marker S0002 showed a significant nonparametric linkage result.
The results on SSC 1 were also supported by a haplotype-based TDT. Haplotypes for affected piglets and their boars consisting of markers SW1621 (alleles 1, 2, and 3) and SW1902 (alleles 1, 2, 3, 4, 5, 6, and 7) were derived using the haplotype option of ALLEGRO. The first position of the haplotype represents the allele at the marker SW1621 and the second position the allele at SW1902. A total of 60.4% of the affected piglets and 50.7% of the boars carry the haplotype 1-2. This haplotype is preferentially transmitted to affected piglets, with the haplotype TDT showing a significant result (P = 1 × 10−4).
The main objective of our study was to identify chromosomal regions possibly containing putative disease loci involved in the etiology of anal atresia. Similar to complex traits in humans, AA is characterized by a low prevalence, variation in appearance of the phenotypes, and an unknown mode of inheritance. Nevertheless, many studies demonstrate a genetic determination of the disorder (Knap 1986), with some of them postulating the presence of one or two major genes from the results of segregation analyses (Stigler et al. 1991; Thaller et al. 1996). To clarify these findings and to map possibly causal genes, we conducted a linkage study based on affected paternal half-sib families. Because the parameters that are prerequisite for classical linkage analysis such as the disease allele frequency, penetrance, or rate of phenocopies are unknown, we chose a nonparametric approach. Robust nonparametric methods were shown to perform only slightly worse compared to parametric approaches even if parameters are known (Lander and Schork 1994; Whittemore and Halpern 1994), but the latter are much more sensitive with respect to input parameters. ALLEGRO offers the possibility of calculating a LOD score with variable prevalences and disease allele frequencies. When analyzing our data we realized that results were difficult to interpret and we could demonstrate that LOD score values on SSC 1 depended heavily on the assumed parameters. Obviously, there is a severe risk when investigating the genetic background using model-based methods if the correct mode of inheritance is unknown.
To increase efficiency, a two-stage screening strategy was employed as proposed by Hauser et al. (1996) and Guo and Elston (2001). This procedure not only had the advantage of reducing genotyping effort, but also allowed us to gain an initial overview in a reasonable time span on the basis of a first subset of half-sib families. Affected piglets are rare and, despite the populationwide collection system established, some effort is necessary to achieve a sufficient number of confirmed paternal half-sibs from field data. Nevertheless, we rigorously tested for correct parentage to avoid any loss in power by falsely including unrelated pairs. The first scan involved 27 families, and most of them were paternal half-sib families, with a total of 65 affected piglets and 118 markers, yielding an average information content of 0.52 across the major part of the genome; this value is in the range of comparable studies conducted in humans (Laitinen et al. 2001). On the basis of results achieved, an additional eight markers were added in the previously determined region of interest on SSC 1 and a total of 35 families were included in the final linkage analysis. As can be seen in Table 2, the information content was markedly increased and the NPL statistic supported the initial findings. The fact that the NPL score of marker SW1621 decreased slightly could be a consequence of its comparatively low information content in the additional families or an overestimation in the first analysis. Interestingly, however, the new marker SW2185 showed only a small linkage signal in the original family panel but a rather high NPL value in the separate analysis of the eight new families. Overall linkage results indicated a disease locus located on chromosome 1q21 in swine.
Association studies confirm linkage results on SSC 1 and SSC 12:
In the context of a multistep procedure and to avoid a loss in power by unnecessary multiple testing, we applied the TDT test for association of markers only within a priori interesting chromosomes containing putative candidate genes or within chromosomes with significant linkage signals. The TDT allows us to confirm linkage results and improves the resolution as it relies on linkage disequilibria that are expected to stretch over smaller chromosome segments. The closely linked markers SW1621 and SW1902 located in the middle of the region with the highest NPL scores on SSC 1 (Figure 2) showed large values for the test statistics that were significant at the 1.0% level applying a Bonferroni correction. We consider this independent result on SSC 1 as a strong confirmation of the linkage signals and indicating the presence of a locus lying a relative short distance from the tested markers with some impact on AA. The same is true for SSC 12, but statistics testing for linkage and association applying the TDT were less pronounced. None of the markers on SSC 3 showed association, which could be due to linkage equilibrium resulting from comparatively large distances between the markers and the putative disease locus. To clarify this finding, further investigations using a denser marker map would be necessary. On the other hand, results for a single marker on SSC 9 indicate linkage and association due to the TDT but NPL statistics were not significant at this chromosomal position. Such discrepancies between results of linkage and association analyses have also been reported by Zhang et al. (2004), who indicated methodological differences as a possible explanation of the disparity found.
We also investigated haplotypes consisting of alleles of the two significant markers SW1621 and SW1902 on chromosome SSC 1 and found an excess in the frequency of a specific haplotype in affected piglets compared to the frequency in boars. It could be concluded that this haplotype might be associated with the occurrence of AA. The fact that this haplotype is rather frequent even in unaffected boars will be the subject of further studies. On the one hand, there is the possibility of the existence of subhaplotypes that represent the causative variant more precisely and would allow us to distinguish between disease-carrying and normal haplotypes among the associated haplotypes. Additional markers in the region of interest will help to answer this question. On the other hand, it has been also reported in human genetics (Tabor et al. 2002) that variants with a comparatively modest effect are much more frequent in populations than alleles of severe monogenic diseases. Haplotype frequencies representing causative variants in the range found in this study are not unexpected in view of a possibly low selection coefficient acting in combination with genetic drift.
Although our study did not aim to determine the mode of inheritance, some indirect conclusions could be made. Our results support the possible impact of single genes as proposed by Thaller et al. (1996). These authors also suggested the influence of polygenic and environmental factors in addition. A monogenic inheritance is unlikely when considering the low prevalence and high frequency of the suspected disease locus. When assuming the involvement of a limited number of genes the penetrance is probably low, because only a few full-sibs with atresia ani were found. Despite being highly speculative, a basic two-locus model with genes located on SSC 1 and SSC 12 might underlie the expression of AA. A digenic hypothesis was suggested by Hori et al. (2001) as well as by Stigler et al. (1991). This theory is also supported by our own comparative mapping of the SD region in mice to homologous genome regions in swine. It seems that the area near the marker genes SPNA1 and PAX 8 maps to SSC 1q21 and SSC 12, respectively.
Influence of sex on atresia ani:
We observed a pronounced excess of males among the affected piglets collected in the field. This finding is in agreement with that of Hamori (1962), although several other authors (Henricson 1963; Hori et al. 2001) did not find a difference in the occurrence of AA between sexes. One explanation of the major bias in the sex ratio of affected piglets might be that females with AA can survive for a couple of weeks because they often develop a fistula between rectum and vagina and are able to defecate through the vagina (Lambrecht and Lierse 1987). Thus, farmers sell afflicted females as sucking pigs rather than sending them to veterinary institutes and the sex bias found could be expected as a consequence. Interestingly, we also observed three cases where sows with AA had litters.
Because a sex-linked genetic contribution cannot be ruled out, we included microsatellite markers on SSC X and on the pseudoautosomal region in our analyses. There was no indication of the involvement of the sex chromosomes, but the amount of data specific for appropriate analysis of the pseudoautosomal region was small. In this case, only sows with at least two affected piglets are informative and nine full-sib families are insufficient to draw a final conclusion. It can be questioned whether a more elaborate diagnosis (concomitance of fistulas) might allow for more precisely defined traits in the future. In view of the major deformation, the defect of cloacal plate in the early stage of embryogenesis that is responsible for an imperforate anus, we consider our trait definition as justified. This is supported by our results from analyzing the subtrait atresia ani separately.
Comparison to other linkage studies:
Despite the relevance of birth defects like AA for animal welfare and the currently available tools for molecular genetics, only a few studies were undertaken to unravel possible disease loci. A major reason might be the logistic difficulties in obtaining material from diseased and related piglets. Hori et al. (2001) therefore used a backcross of a commercial line with a swine population selected for AA and exhibiting an extremely high incidence of 62.0%. This strain could be maintained only by surgery of affected animals. They report suggestive linkage at the marker SW 2072 on SSC 15, which could not be confirmed in our analysis. By contrast, Hori et al. (2001) found no indication for a predisposing gene on SSC 1 or SSC 12, although their coverage of the genome and marker density were generally low compared to our study. Specifically, the authors used only three markers on SSC 1 in their single-point parametric linkage analysis, none of which was located in the region of interest flanked by markers SW2185 and SW1902. The same is true for SSC 12, where only one marker was included in the analysis by Hori et al. (2001).
The discrepancy on SSC 15 might possibly be explained by false positive results when applying a parametric approach if parameters are unknown. Nevertheless, it is the rule rather than the exception for complex diseases that different chromosomal regions show effects in different breeds or populations. Similar discrepancies were obtained when genetically dissecting diseases in humans such as asthma (Laitinen et al. 2001) and schizophrenia (Blouin et al. 1998).
Putative candidate genes:
A great deal of work identifying candidate genes that are possibly causative for AA and that are located in the QTL regions still remains. Most porcine genes are not yet physically mapped and lists of the gene content of specific porcine genome regions are far from complete. However, comparative mapping of human and porcine chromosome regions might allow us to postulate some candidate genes that might eventually be responsible for predisposition to AA (see Table 4). The comparative maps established by Goureau et al. (1996, 2000), Shi et al. (2001), and Robic et al. (2003) suggest that the region SSC 1q18–q21 flanked by markers SW1621 and SW1902 maps to three different human chromosomal regions: HSA 15q26, HSA 14q22, and HSA 9q21–q34. Our own BLAST searches comparing the porcine flanking sequence of SW1621 and SW1902 to the human genome sequence suggest that it is homologous to HSA 9q34.4. This region harbors several potential candidate genes, members either of the retinoid acid pathway such as retinoic X receptor-α (RXRA) or of the sonic hedgehog signaling pathway such as, for example, patched homolog 1 (PTCH1), a receptor for sonic hedgehog. Some members of the sonic hedgehog pathway are also located in the regions of HSA 15 and HSA 14, such as cysteine knot superfamily 1 (CKTSF1B1), formin (FMN), and bone morphogenetic protein 4 (BMP4), respectively.
The region on SSC 12 near marker SW957 maps to HSA 17q11–q21. Several candidates like the HOXB-cluster genes HOXB9 and HOXB5, which play an important role during embryonic development and in determining tissue patterning and body plan organization, are already mapped on chromosome SSC 12p11–p12. Another member of the retinoid acid pathway is RARA, which is not physically mapped in swine so far and is possibly located in the area on SSC 12 (see Table 4). However, before useful hypotheses about positional candidate genes can be postulated, more validated information about the gene content of this specific chromosome region in swine is needed. The same is true for functional candidates. Very little is known about the etiology of AA, especially about the factors causing a malformation of the cloacal plate, which is likely the underlying defect in embryogenesis. The follow-up work envisaged will apply TDT tests and case/control studies to a carefully selected set of candidate genes on the basis of sufficiently high numbers of affected and unaffected piglets to detect causal polymorphisms and to enlighten physiological pathways.
We thank the Bavarian insemination centers Besamungsstation Bergheim e. V., Besamungsstation Landshut- Pocking e. V., and Besamungsverein Neustadt a. d. Aisch for providing support in collecting animals and semen samples. We are grateful to the veterinarians at the Landesuntersuchsamt Südbayern e. V., especially to Thomas Held and Franz Schrott for the pathological diagnosis of the piglets. We are indebted to Ingolf Russ from Genecontrol e.V. for genotyping. Olaf Bininda-Emonds, funded by Bundesministerium für Bildung und Forschung (Federal Ministery of Education and Research), provided valuable comments on the manuscript. This work was supported by grants from the Förderverein für Biotechnologieforschung e.V., which is an organization of the German pig breeding industry.
Communicating editor: C. Haley
- Received June 25, 2004.
- Accepted July 13, 2005.
- Copyright © 2005 by the Genetics Society of America