In this study, we have used the latest bovine SNP chip technology and association analysis to map the osteopetrosis disease locus to a ~3.4-Mb interval at the telomeric end of bovine chromosome 4. Both a case/control association analysis and a homozygosity analysis implicated essentially the same chromosomal segment on BTA4, centered at the 117-118 Mb location. Molecular examination of the SLC4A2 functional candidate gene at this position not only resulted in a new annotation of this gene, but also revealed a large deletion mutation within this genomic sequence. By analyzing SLC4A2 transcripts present in a known osteopetrosis carrier, we have confirmed the presence of a previously unannotated first exon identified in dbEST, and defined the biological relevance of the SLC4A2 deletion mutation. We have further demonstrated that this mutation is concordant with the osteopetrosis phenotype. Thus, we believe we have identified the causative mutation for osteopetrosis in Red Angus cattle.
At the level of genomic DNA, the osteopetrosis mutation results in the deletion of 2781 bp of the SLC4A2 gene sequence, and thus the removal of roughly one-third of intron 1, the entire sequences of exon 2 and intron 2, and nearly half of exon 3. This has been confirmed by the successful amplification and sequencing of the genomic fragment spanning the deletion breakpoint junction in affected individuals, followed by comparison of the mutant sequence with the normal SLC4A2 sequence. Inspection of the genomic sequence encompassing the deletion mutation did not reveal an obvious mechanism for spontaneous chromosomal breakage at this location. Deletions of this type can often be attributed to genomic instability; typically this instability is associated with the formation of alternative, non-B DNA secondary structures . As a general rule, DNA sequences that form such secondary structures are repetitive ; yet, analysis of the SLC4A2 sequence with RepeatMasker http://www.repeatmasker.org did not yield masked sequences within the immediate vicinity of either chromosomal break (Additional file 3 Figure S1). The nearest repetitive sequences are at least 550 bp from either side of each breakage point, and only a few short (30-70 bp) repeats were detected within the deleted sequence. Thus, the cause of the deletion mutation is unclear.
At the level of transcription, the deletion mutation results in the production of mutant transcripts lacking exons 2 and 3. This was expected, given the deletion of exon 2 as well as the splice acceptor for exon 3, and has been confirmed by analysis of 5' RACE products from a known osteopetrosis carrier cow. Despite evidence of multiple SLC4A2 transcripts (referred to as types "a", "b" and "c") in other species, only normal and mutant type "a" transcripts were detected by 5' RACE. The ~4.4-kb type "a" transcript is the longest known SLC4A2 isoform, and has been detected in the human, mouse and rat [29–32]. Unlike the other transcript types, this isoform is ubiquitously expressed, and has been shown to be upregulated in osteoclasts . It consists of exons 1 through 23, and the start codon is located within exon 2. Slightly shorter, ~4.2-kb "b" isoforms have also been described in the human, mouse and rat [29–32]. Both "b1" (human, mouse and rat) and "b2" (human and mouse) isoforms initiate transcription from alternative first exons, containing alternative start codons, that are located in intron 2. Expression of these transcripts appears to be predominant in the stomach, but also at reduced levels in other tissues; in humans, the relative abundance of these transcripts appears to be approximately 10% of that of the "a" isoform . Therefore it is possible that no type "b" transcripts were detected in this study as a result of either a lack of expression in osteoclasts or expression at a level insufficient to be observed among the clones picked. If type "b" isoforms are present in bovine osteoclasts, it is expected that the deletion mutation would result in the removal of both alternative first exons as well as exon 3. Finally, a third, ~3.8-kb "c" type isoform has been detected in both mice and rats [29, 32]. Similar to the "b" isoforms, "c1" (mouse and rat) and "c2" (mouse) isoforms initiate from alternative first exons; these exons are located in intron 5, and splice to exon 6, which contains the start codon. The "c1" isoform has been described as stomach-specific, whereas the "c2" isoform can be detected at low levels in other tissues . In this case, the lack of observation of a "c" transcript type in cattle, and likewise in human, is likely due to the fact that the start codon utilized in mouse and rat sequences is not conserved. Both the human and bovine sequences contain a methionine codon further downstream within exon 6, yet, based on a lack of detection in either species, this does not appear to be a translational start. Although the nucleotides surrounding this codon may represent an adequate Kozak sequence, they do not comprise a strong initiator sequence .
At the level of translation, then, it is expected that the deletion mutation would result in a complete lack of bovine SLC4A2 protein expression. Although this has not been confirmed in this study, much can be inferred by comparison with recent studies of SLC4A2 knockout (Ae2-/-) mice. Mice deficient in all SLC4A2 isoforms, as a result of targeted replacement of exons 14-17 with A the neomycin resistance gene, fail to resorb bone and develop severe osteopetrosis [25, 26]. The gross and histological features observed in Ae2-/- mice are similar to those observed in osteopetrosis-affected calves; Ae2-/- mice exhibit improper bone marrow cavity formation [24–26], growth retardation[24–26, 34], early lethality (typically by age of weaning) [26, 34], abnormal mandibles  and impaired tooth development [25, 34]. In both SLC4A2-deficient mice and cattle, abnormal osteoclast morphology is apparent; however, histological differences are observed. In these mice, osteoclasts are enlarged with abundant cytoplasm [25, 26], whereas the cattle osteoclasts examined here appeared small with a reduced volume of cytoplasm. Additionally, Ae2-/- knockout mice exhibit similar numbers of osteoclasts in bone when compared to wild-type mice, whereas we noted a marked reduction in the number of osteoclasts in the affected calf samples. This finding is consistent with previous reports of osteopetrosis in cattle [6, 9, 10].
Recent studies in mice have shown that the anion exchanger SLC4A2 is necessary for proper osteoclast function [24–26]. This finding suggests that, during normal bone development and remodeling, SLC4A2 is the osteoclast anion exchanger required to exchange bicarbonate ions produced by carbonic anhydrase II for chloride ions that are then transported across the ruffled border membrane, along with protons, to acidify the resorption lacuna and demineralize bone. It is expected, then, that a lack of SLC4A2 protein expression would prevent bone resorption due to the resulting lack of acidification. Indeed, studies in mice have shown that SLC4A2 -deficient osteoclasts are unable to form an acidified vacuole , and in this study, the observed resorption lacunae are particularly shallow. However, unlike human osteoclasts harboring mutations in either carbonic anhydrase II  or the v-ATPase proton pump , which also mediate the acidification activity of osteoclasts, SLC4A2 -deficient mouse osteoclasts appear morphologically abnormal and lack a ruffled border membrane [24–26]; thus, a bone resorption defect due to a lack of SLC4A2 may be more severe. Additional functional complications may result from the improper alkalinization of the osteoclast cytoplasm. It has been suggested that the expression of other proteins involved in bone resorption may be affected by a loss of SLC4A2, and indeed, abnormal expression of β3 integrin has been observed in SLC4A2 -deficient mouse osteoclasts . As β3 integrin is involved in the cytoskeletal organization step of the bone resorption process, this may explain the defect in ruffled border membrane formation. It has also been proposed that SLC4A2 may have multiple functions in the osteoclast, as another member of this protein family, SLC4A1, can function as a cytoskeletal anchor protein . Thus, loss of SLC4A2 may affect the bone-resorbing function of osteoclasts through a number of different mechanisms; however, the effect of the cattle SLC4A2 deletion mutation on acidification activity is currently unclear.
Studies in mice have also shown that loss of SLC4A2 may affect the viability of osteoclasts. Although comparison of normal and Ae2-/- mice reveals similar numbers of osteoclasts, nearly four times as many SLC4A2 -deficient osteoclasts show signs of apoptosis ; this has been attributed to cytoplasmic alkalinization. This finding is consistent with results reported here, and likely explains the marked reduction in osteoclasts observed in affected calves at the time of abortion. Histological examination suggests that, in cattle homozygous for the SLC4A2 deletion mutation, osteoclasts form but die prematurely by apoptosis. Several Howship's lacunae lacking associated osteoclasts were detected in the affected calf samples, suggesting that osteoclasts were once present, yet few osteoclasts were observed and cytologic features of apoptosis were evident among them. Unexpectedly, histological examination also revealed reduced numbers of osteoblasts in calf samples. Based on the increased bone density observed in these animals, osteoblasts were clearly present during fetal development. Thus, loss of SLC4A2 may also affect this cell type. Indeed, studies have reported intercellular communication among bone cells which regulates the balance between bone formation and resorption [reviewed in 36]. Interestingly, a pH-dependent model of osteoclast-osteoblast communication has been proposed in which SLC4A2 plays a key role [36, 37]. In this model, extracellular alkalinization resulting from SLC4A2-mediated bicarbonate ion secretion activates osteoblast enzymes that degrade inhibitors of bone formation. Perhaps, then, a lack of osteoblast activation, due to the loss of SLC4A2 and apoptosis of osteoclasts, leads to subsequent death of osteoblast cells. Temporal differences in the death of these cell types could account for the overgrowth of bones observed in the affected calves.
It is interesting that no mutations in human SLC4A2 have been described to date in the context of osteopetrosis or any other hereditary disease. Approximately 30% of human patients with osteopetrosis have unrecognized molecular defects [15, 38]; thus, it may be worthwhile to examine these orphan human osteopetrotic syndromes for mutations in SLC4A2. However, it has also been suggested that such human SLC4A2 mutations may result in embryonic lethality in humans due to ubiquitous expression of this protein, and as a result, will likely remain undetected . For reasons unknown, mice and cattle lacking SLC4A2 exhibit differences in age at lethality; SLC4A2 -deficient mice tend to die as weanlings, whereas cattle are aborted in late gestation. The abortion of affected calves is likely due to severe changes in brain, specifically compression of brainstem, herniation of cerebellum through foramen magnum, degeneration of caudal cranial nerve neurons, axonal spheroid formation, and mineralization of neuropil at the floor of the lateral ventricles. All of these features appear to be secondary to compression by the thickened osteopetrotic cranial vault. They have not been described, to our knowledge, as a congenital or antenatal defect in severe forms of inherited osteopetrosis in human patients.
We have demonstrated that the deletion mutation genotype is fully associated with the osteopetrosis phenotype in Red Angus cattle. Through the development and use of simple, PCR-based assay to detect the deletion mutation, we accurately identified 100% of affected calves and their carrier parents as such based on genotype. Due to this accuracy, the genotyping assay can now be used as a molecular diagnostic tool for the identification of other osteopetrosis carriers within the Red Angus population. In this study, population analysis identified 25 carriers among 265 AI sires; this allele frequency suggests that the frequency of the mutation in the Red Angus population may be as high as 4.7%. As a caveat, however, this estimation includes a group of animals submitted by one breeder in which nearly 30% were identified as carriers. Taking this into consideration, a more realistic estimate is likely to be on the order of 1.5-2%. Results of all tested AI sires have been reported to the Red Angus Association of America for dissemination to breeders. The accessibility of this information, as well as any future carrier information generated through the use of the diagnostic test, will allow cattle producers to make educated decisions for herd management that not only facilitate disease prevention, but also enable retention of valuable germplasm.
While assessing the Red Angus population, we identified a carrier bull that was born in 1964. This finding is consistent with the emergence of the disorder; however, as no ancestors of this bull were available for testing, the identity of the proband remains unknown. It is interesting to note that this oldest known carrier bull is a descendant of Black Angus grandparents, yet the deletion mutation could not be detected among 578 Black Angus individuals tested. This suggests that, if the same deletion mutation causes osteopetrosis in both Black and Red Angus cattle, the frequency of the mutated allele among Black Angus cattle is less than 0.001. Alternatively, it is possible that at least one other bovine mutation causing osteopetrosis exists. Any additional mutations that may cause this disorder in other breeds of cattle are currently unknown.