Duplication in Primates

C4 and CYPZl are two adjacent, but functionally unrelated genes residing in the middle of the mammalian major histocompatibility complex (Mhc). The C4 gene codes for the fourth component of the complement cascade, whereas the CYP2l gene specifies an enzyme (cytochrome P450c21) of the glucocorticoid and mineralocorticoid pathways. The genes occur frequently in multiple copies on a single chromosome arranged in the order C4 . . . CYPPl . . . C 4 . . . CYP21. The unit of duplication (a module) is the C4-CYP21 gene pair. We sequenced the flanking regions of the C4-CYP21 modules and the intermodular regions of the chimpanzee, gorilla, and orangutan, as well as the intermodular region of an Old World monkey, the pigtail macaque. By aligning the sequences, we could identify the duplication breakpoints in these species. The breakpoint turned out to be at exactly the same position as that found previously in humans. The sequences flanking paralogous genes in the same species were found to be more similar to one another than sequences flanking orthologous genes in different species. We interpret these results as indicating that the original (primigenial) duplication occurred before the separation of apes from Old World monkeys more than 23 million years ago. The nature of the sequence at the breakpoint suggests that the duplication occurred by nonhomologous recombination. Since then, the C4-CYP21 haplotypes have been expanding and contracting by homologous crossing over which has homogenized the sequences in each species. We speculate that the reason for the concerted evolution of the primate C4-CYP21 region may be a requirement for the coevolution of certain components of the complement pathway, including the C4 component. We contrast the evolution of the C4-CYP21 region with that of other Mhc regions. C OMPLEMENT component 4, C4, is one of more than 20 proteins constituting the complement cascade, which is involved in defense of vertebrate bodies against pathogens. Activation of the cascade by antigen-antibody complexes or by other means leads ultimately to the assembly of lytic complexes on the cell surface, perforation of the plasma membrane, and killing of the cell (Ross 1986). Some of the activated components of the cascade are also involved in a variety of other biological functions. Cytochrome P450c2 1 (2 1-hydroxylase), CYP2 1, is an enzyme participating in the conversion of cholesterol to aldosterone or cortisol in the cortex of the vertebrate adrenal gland (MILLER 1988). In mammals, complement component 4 and cytochrome P450c21 are encoded in adjacent genes, C4 and CYP21, respectively, residing in the middle of the major histocompatibility complex, Mhc (KAWAGUCHI, O'HUIGIN and KLEIN 1991). The per haplotype number of C4 and CYP21 copies varies from species to species and also among individuals of certain species. Single C4-CYP21 haplotypes have been found in humans (MCLEAN et al. 1988; and others), Syrian hamster (L~vI-STRAUSS et al. 1985), dog (KAY and DAWKINS 1984; DOXIADIS et al. 1985), cat (DE ' Permanent address: Department of Dermatology, Yokohama City University School of Medicine, Yokohama, Japan. Genetics 134: 331-339 (May, 1993) KROON et al. 1986), guinea pig (BITTER-SUERMANN et al. 1977), and several whale species (SPILLIAERT, PALSDOTTIR and ARNASON 1990). Haplotypes with two C4 and two CYP21 genes are common in humans (McLEAN et al. 1988; and others), chimpanzees (KAWAGUCHI et al. 1990; BONTROP et al. 1990; CHRISTIANSEN et al. 1991), gorillas (KAWAGUCHI and KLEIN 1992), macaques (MEVAG et al. 1983), mice (ROOS, ATKINSON and SHREFFLER 1978), rats (TOM et al. 1985), cattle (YOSHIOKA et al. 1986; CHUNC, MATTESON and MILLER 1985, 1986), pigs (KIRSZENBAUM et al. 1985), and horses (KAY et al. 1987). Haplotypes with three C4 and three CYP21 genes have been reported in humans (MCLEAN et al. 1988) and orangutans (KAWAGUCHI and KLEIN 1992). Finally, haplotypes with four C4 and four CYP21 genes occur occasionally in humans (MCLEAN et al. 1988) and orangutans (ZHANG et al. 1993). In haplotypes carrying multiple copies of C 4 and CYP2l genes, the two types of gene always alternate on the linkage map ( i e . , C 4 . . . CYP2l . . . C 4 . . . CYP21, etc.), suggesting that the basic unit is a C4-CYP21 module and that multimodular haplotypes arise by multiplication of this,unit (KAWAGUCHI, O'HUIGIN and KLEIN 1991). The human C4-CYP21 module is about 35 kilobases (kb) long and the distance between the two genes, transcription332 Y. Horiuchi et al. ally oriented in the same direction, is approximately 3 kb (CARROLL, CAMPBELL and PORTER 1985; DUNHAM et al. 1987). Recent studies indicate, however, that in reality the module contains three genes, the third one being transcribed from the DNA strand complementary to that from which the C4 and CYP21 genes are transcribed (MOREL et al. 1989). The third gene codes for a protein related to the extracellular matrix protein tenascin (MATSUMOTO et al. 1992; GITELMAN, BRISTOW and MILLER 1992) and is either referred to as MHC-F3 because of its location in the Mhc region and the presence of fibronectin type I11 domains (MATSUMOTO et al. 1992), or simply as “X” (MOREL et al. 1989). The last exon coding for the 3”untranslated (3’UT) region of the CYP21 gene overlaps with the last exon of the MHC-F3(X) gene, but the two genes are oriented in opposite directions. In haplotypes with multiple copies of the CYP21 gene, all but one of the MHC-F?(X) copies are truncated (GITELMAN, BRISTOW and MILLER 1992) and most likely pseudogenes. GITELMAN, BRISTOW and MILLER (1 992) sequenced the DNA in the region between the two modules of a bimodular human haplotype and found that the sequence in the 3’ part of the region matches that found upstream from the first module, whereas the sequence in the 5’ part aligns with that found downstream from the second module. The two parts of the intermodular sequences overlap by four nucleotides in the center of the region. This result suggests that the two modules arose by duplication from a single module and that this event occurred by nonhomologous recombination. In the present study we attempt to answer the following questions: When did the duplication occur? Did nonhomologous recombination occur repeatedly in different primate species and different haplotypes, or did it occur only once, and if so, by what mechanism did the subsequent variation in module number arise? T o this end, we sequenced the relevant segments of the C4-CYP2I region in the chimpanzee, gorilla, orangutan and pigtail macaque. MATERIALS AND METHODS Source of cosmid clones and D N A The relevant DNA segments were obtained from cosmid clones. Three cosmid libraries were used. The first library was constructed using DNA isolated from the Epstein-Barr virus (EBV)-transformed B cell line, Hugo, established from the common chimpanzee (Pan troglodytes) at the TNO Institute of Applied Radiobiology and Immunology, Rijswijk, The Netherlands (MAYER et al. 1988). The second library was based on DNA of a western lowland gorilla (Gorilla gorilla) isolated from the fibroblast line Sylvia, which was established from a skin biopsy sample by KIRBY D. SMITH, The Johns Hopkins University School of Medicine, Baltimore, Maryland. The DNA for the third cosmid library was isolated from the cell line CP8 1, which was established from monocytic leukemia cells of a 13-year-old female orangutan at the Los Angeles Zoological Garden (RASHEED et al. 1977). The pigtail macaque (Macaca nemestrina) DNA was derived from the EBVtransformed B cell line 86081 kindly provided to us by LAKSHMI GAUR, HLA Laboratory, Puget Sound Blood Center, Seattle, Washington. Genomic DNA was isolated from the indicated cell lines according to the method described by MANIATIS, FRITSCH and SAMBROOK (1982) and the libraries were constructed and screened according to STEINMETZ et al. (1 985); for a full description of the libraries, see KAWAGUCHI et al. (1990) and KAWAGUCHI and KLEIN (1 992). Analysis of cosmid clones: DNA isolated from cosmid clones following the protocol given in MANIATIS, FRITSCH and SAMBROOK (1982) was digested with restriction endonucleases, and the resulting fragments were separated by agarose gel electrophoresis and transferred to nitrocellulose membranes (Sartorius, Gottingen, Germany). The relevant fragments were subcloned into pBluescript I1 SK+ phagemid vector (Stratagene, Heidelberg, Germany) according to the standard method (DAVIS, DIBNER and BATTEY 1986). Sequencing: Subclones were ligated to the pBluescript I1 SK+ phagemid vectors and the recombinant clones were picked up by the colony hybridization method (DAVIS, DIBNER and BATTEY 1986). Double-stranded DNA was prepared by the plasmid mini-boiling method (HOLMES and QUIGLEY 1981). Five micrograms of DNA were denatured in 0.2 M NaOH, 0.2 mM EDTA for 30 min at 37” and sequenced by the dideoxy chain-termination method (SANGER, NICKLEN and COULSON 1977), using the Sequenase version 2.0 kit (US. Biochemicals, Cleveland, Ohio). Fragments were sequenced on both strands two to three times to eliminate or resolve sequencing errors and ambiguities. Polymerase chain reaction (PCR): Genomic DNA was amplified in the GeneAmp PCR system 9600 (Perkin-Elmer, Uberlingen, Germany). Two hundred nanograms of DNA were initially denatured by heating to 94“ followed by a 30cycle profile of 20 sec at 94”, 20 sec annealing at 5 0 ° , and 30 sec extension at 72“ (using the GeneAmp PCR reagent kit (Perkin-Elmer). The primers used for amplification were 5’-GACTCCTTGATGGATGTTGA-3’ (Tu336) and 5’AAGGACAGCCTGGCGCCCT-3’ (Tu337), which were specific for sequences located 105 base pairs (bp) upstream and 1 16 bp downstream of the human recombination breakpoint in the intermodular region, respectively. The amplified products were isolated by electrophoresis on low melting point agarose gel, cloned in Bluescript I1 SK+ vector, and sequenced. Construction of dendrograms: The neighbor-joining genetic distance method of SAITOU and NEI (1987) was used for evaluating evolutionary relationships among nucleotide sequences. Genetic distances were calculated by the twoparameter method (KIMURA 1980).