An original SERPINA3 gene cluster: Elucidation of genomic organization and gene expression in the Bos taurus 21q24 region

Serine peptidase inhibitors represent a constantly expanding group of structurally related proteins. Amongst them, the most important family is undoubtedly the serpins, an acronym for ser ine p roteinase in hibitors [1] which describes the functional properties of the superfamily. The name serpin was originally coined in recognition of the fact that most serpins are inhibitors of serine proteinases. However, it is now clearly inappropriate because few members of this superfamily lacked any proteinase inhibitory properties [2, 3]. Approximately 500 serpins have been identified to date and can be found in all superkingdoms including animals, plants, bacteria as well as some viruses [4, 5]. Both extracellular and intracellular serpins have been identified [6]. Most of them are glycoproteins (MW 40,000–60,000) composed of a single polypeptide chain and variable number of oligosaccharide moieties [7]. The protein structure of serpins is characterized by 3 β-sheets and 8 or 9 α-helices [8]. Serpins present a conserved domain, the reactive-centre-loop domain (RCL) which connects β-sheets A and C and often acts as "bait" for the target serine protease [9]. Phylogenetic relationships between orthologous and paralogous serpins have been studied. Irving et al. [4] compared several hundred serpin proteins and proposed an arbitrary classification into eight major and eight minor sub-families from A (antitrypsin-like) to P (plants), based on clade recognitions and supports. In vertebrates, serpins are involved in many extracellular processes dependent on serine proteinases such as blood coagulation, fibrinolysis, cell migration, complement activation and inflammation [10, 9]. Serpin dysfunction can have pathological consequences and contributes to diseases such as thrombosis, cancer [11] and serpinopathies including cirrhosis and emphysema [12]. Serpins are subdivided into the nine first groups A to I. The two largest groups so far, in all of the serpins, are the antitrypsin-like and the ovalbumin-like serpins which are designated "SERPINA" and "SERPINB" respectively and composed of 13 functional members each. In the "SERPINA" clade, some of them are involved in a diversity of biological functions. For example, human SERPINA3 (α₁-antichymotrypsin) is found and identified as a major component of the fibrillary amyloid plaques of brains from patients with Alzheimer's disease, one of the most common forms of dementia [13, 14].

According to the structure of their respective genes (number of introns and exons and relative position at genomic level), the vertebrate serpin superfamily was subdivided into at least six groups [15]. Most known serpin genes, including the α₁-antichymotrypsin which belongs to group 2, contain a non-coding first exon and a partly non-coding last exon.

Clustering of serpin genes occurs in the genome of human and other species. They present similarities in gene structure: the SERPINA genes characteristically consist of four exons with identical positioning and phasing of the intron-exon boundaries. SERPINA1 (α₁-antitrypsin), SERPINA3 (α₁-antichymotrypsin), SERPINA5 (PCI, protein C inhibitor), SERPINA9 (centerin), SERPINA10 (ZPI, protein Z-dependent protease inhibitor) as well as SERPINA11 are mapped together in the same cluster on human chromosome 14q32 [16]. More recently, the gene encoding the SERPINA4 (kallistatin precursor) was also mapped within this cluster. This close proximity in the same cluster suggests that these genes might have arisen by tandem duplications from a common gene ancestor [16, 17]. However, in spite of their close physical proximity, the corresponding proteins have disparate functions, raising interesting questions about the evolution of gene function and expression. For example, the SERPINA9 is likely to function in vivo in the germinal centre B-cells [18]; the SERPINA5, in addition to its activities within the blood clotting and fibrinolytic cascades, seems to participate in several biological processes including reproduction and tumor growth [19] and SERPINA10 produces a rapid inhibition of the coagulation factor Xa in the presence of protein Z (PZ), procoagulant phospholipids and Ca²⁺[20]. An approximate protein similarity of 40–45% between the human SERPINAs [21] could explain these diverse physiological functions.

In mouse, the genomic organization of the serpin family shows significant differences. The cluster of genes corresponding to human chromosome 14q32 is mapped on the chromosome 12F1 [22]. Both α₁-protease inhibitor and contrapsin genes had undergone considerable expansion in this genome. The mouse α₁-protease inhibitor (α₁-antitrypsin – SERPINA1) cluster comprises 5 members. The multigenic cluster corresponding to contrapsin (α₁-antichymotrypsin – SERPINA3) is termed the Spi-2 locus [23] and, until recently, was thought to encompass 14 members [24] with a degree of overall sequence similarity of 65 to 80%, despite markedly divergent RCL domains. Some of these genes present a significant degree of sequence homology with the single-copy human SERPINA3 gene [25]. However, the RCL was substantially different. In mouse genome, these two clusters represent the result of multiple gene duplication events.

In the pig, four serine protease inhibitors were detected at the protein level: PI1 (SERPINA1), PI2 (SERPINA3-1), PI3 and PI4 (SERPINA3 paralogues) [26–28]. PI1 and PI2 corresponding genes were mapped by radioactive in situ hybridization (RISH) to the distal end of chromosome 7q23-26 [29, 30]. A complete cDNA, named SERPINA3-1, corresponding to PI2 and a gene, named SERPINA3-2, were sequenced [31] and clearly affiliated to the porcine α₁-antichymotrypsin (SERPINA3) family. Porcine PI2 and SERPINA3-2 only showed 76% amino acid identity, with the largest difference near the C-terminus containing the deduced reactive site P1-P'1.

The knowledge on bovine serpin genes is much more restricted. Hwang et al. [32] identified a serpin, previously named Endopin1, at the cDNA level, in bovine endocrine chromaffin cells. The same serpin was then purified from bovine skeletal muscle [33]. This serpin (mEndopin1A) was found to be essentially intracellular and widely distributed in bovine tissues suggesting that this protein might have central biological role in a large set of tissues [34]. More recently, a second endopin, named mEndopin1B, was purified from bovine diaphragm and the genes encoding mEndopin1A and mEndopin1B were characterized [35].

In order to clarify the complexity surrounding the bovine SERPINA3 genes, we describe here the isolation, the complete sequence and the genomic organization of the bovine α₁-antichymotrypsin (SERPINA3) multigenic cluster. This cluster was assigned to chromosome 21 using a radiation hybrid panel. In addition, we provide preliminary expression data on the bovine SERPINA3 genes. Characterization of these new genes and phylogenetic analysis of the corresponding proteins provide new insights on the evolution of SERPINA3 gene cluster. We also propose a nomenclature to designate all the members of this clade.

Two bovine serpins named mEndopin1A (renamed here bovSERPINA3-1) and mEndopin1B (renamed here bovSERPINA3-3) were recently purified from diaphragma and characterized [33, 35]. Although the sequence identity between both muscle endopins is close to 93%, the major differences were found within the RCL. These two proteins are homologous to the unique human SERPINA3 and share with it a high identity (approximately 67%). The genes encoding these two proteins were shown to be expressed notably in muscle. In this study, we show that bovSERPINA3-1 and bovSERPINA3-3 are members of a family comprising at least six other deduced proteins. The genes encoding these proteins are mapped to the presently named SERPINA3 cluster localized on chromosome 21q24. Our results indicate that the locus spans over ~235 Kb and contains at least eight SERPINA3 genes with a local order of SERPINA3-1/SERPINA3-2, SERPINA3-6, SERPINA3-3/SERPINA3-4, SERPINA3-7, SERPINA3-5 and SERPINA3-8, and the pseudogene SERPINA3P localized at the 3'end of the cluster. We also mapped another SERPINA gene, SERPINA5 at the 5' end of SERPINA3-1/SERPINA3-2. This mapping is not in agreement with the publicly available Btau_2.0 genomic sequence assembly. To correlate with our proposed genomic organization based on the physical map [39], it was necessary to invert the relative positions of the two contigs (NW_930022 and NW_930023) on chromosome 21. However, all the genes identified in this work had not been listed in Btau 2.0 assembly. To date, the SERPINA3-2, SERPINA3-4 and SERPINA3-6 genes are not identified in the whole genome shotgun sequence data. This absence is probably caused by the high degree of similarity between bovine SERPINA3 genes. The current Btau 3.1 [41] doesn't clarify the situation. Indeed, the allocated genes are still the same but are now localized on two contigs distant of 2 Mb from each other (NW_001494061 and NW_001494063). The new Btau 4 version, not yet publicly available, might help us to resolve this discrepancy.

All these genes share an exon-intron organization identical to that of class 2 serpin genes, as established by Ragg and collaborators [15] with four coding exons and one untranslated exon. Moreover, they share a high degree of identity both in exonic and intronic sequences. Therefore, all these genes established a compact bovine SERPINA3 family with at least eight functional members and one pseudogene. SERPINA3 gene clusters exist in other mammal genomes. Mus musculus genome (build 37.1 [45]) contains nine expressed SERPINA3 genes and two pseudogenes located on chromosome 12. The deduced proteins display only 55% to 80% of identity. Porcine SERPINA3 sequence analysis defined at least three expressed genes [UniGene:55446, UniGene:50917, UniGene:3917] [31] with comparable degree of overall identity as observed in mice. On the other hand, primate genomes only possess a single representative SERPINA3 gene, as seen in rhesus macaque, chimpanzee and human, the latter being located on chromosome 14q32 [16].

We described, for the first time in mammals, a serpin sub-family of eight genes with one sub-group composed of six very conserved members: SERPINA3-1,SERPINA3-2,SERPINA3-3,SERPINA3-4,SERPINA3-5 and SERPINA3-6, with high degree of identity over the full sequences (from 88 to 99%). Members of the bovine SERPINA3 sub-family arose by recent duplications after divergence from Sus scrofa and from an ancestral gene, located on the region 21q24, homologous to the unique primate SERPINA3 gene. They show a considerable gene clustering as the mouse ones, but with notable differences. Indeed, rodent SERPINA3 homologues are grouped into 4 sub-clades due to three rounds of duplication prior to speciation between mouse and rat. Following duplications, mainly in mouse, were then lineage-specific [46]. All these SERPINA3 cluster organizations are analogous to the one observed at the family level, that is for SERPINA. In case of human, where 13 members are described, only SERPINA7 and SERPINA8 are located on different chromosomes, Xq22 and 1q42-43 respectively, whereas all others are on chromosome 14q32. They probably also arose by several rounds of duplication within as well as between chromosomes as already demonstrated for human ovalbumine-like serpins (SERPINB) [47]. The single clade supported by 100% BP on ML tree clustered SERPINA7 and SERPINA9, although these most related members are on distinct chromosomes and present different activities, non-inhibitory and inhibitory respectively. This suggests that through evolution, the loss of inhibitory function in some SERPINA occurred several times independently, as seen for SERPINA7 but also for SERPINA6. Finally, in bovine, although the genome is completely sequenced, only four sub-families were identified to date, with SERPINA1, SERPINA3 and SERPINA5 assigned to chromosome 21 and SERPINA7 probably on chromosome X as all other mammal SERPINA7 s.

SERPINA3-7 corresponds to the previously identified endopins 2A/2B [49, 50]. However, despite using specific set of primers, we could not amplify the third bovine endopin 2, named 2C [36] on our four different BAC DNAs.

In mice and porcine, no specific expression analyses have been conducted on proteins encoded by SERPINA3 genes of clusters. However, although bovSERPINA3s are very close, they share differential expression patterns. For example, SERPINA3-1 is expressed in all tested tissues, whereas SERPINA3-4 is amplified only in liver. Moreover, consensus glycosylation sites and the 2D-gel analysis predicted that bovSERPINA3 may be glycoproteins with different degrees of N-glycosylation. In a previous study [35], we have already shown that four potential N-glycosylation sites (Asn^{100,180,230,264}) are common to bovSERPINA3-1 and bovSERPINA3-3. However, an additional site (Asn³¹⁸) is present in bovSERPINA3-3 and could account for the slightly higher molecular mass (75 KDa for bovSERPINA3-3 versus 70 KDa for bovSERPINA3-1) as assessed by SDS-PAGE. So, as indicated in the 2D-gel Western blot analysis of a partially purified muscle bovSERPINA3 fraction (Figure 5), several bovSERPINA3s present various states of phophorylation. Protein phosphorylation is a very important signalling pathway since this post-translational modification ensured by a large set of specific Ser/Thr/Tyr kinases, might be essential for the biological activity of the protein concerned and/or for a modulation of this activity [51, 52]. Those observations lead us to propose a differential functionality for these protein isoforms.

Comprehensive analysis of the bovine genome SERPINA3 content combined with phylogenetic clade recognition has allowed us to characterize one original SERPINA3 gene cluster not identified yet in other mammals. Some of these eight corresponding proteins were shown to be expressed in various tissues and differently glycosylated and/or phosphorylated. This underlines a complex mechanism of regulation for the biological function of bovine SERPINA3s.

This coherent sub-family of serpins could constitute a model to study the duplication events and their underlying molecular proceedings at the origin of multigenic families.

Actually, it is interesting to notice that the numerous SERPINA3s of Bos taurus contrast with the paucity of other SERPINA sub-families compared to Homo sapiens (5 assigned and 3 predicted in bovine [41] versus 13 in human).