C8α-γ deficiency was examined in four unrelated African Americans. Two individuals were compound heterozygotes for a previously reported point mutation in exon 9. mRNA from the remaining six C8A alleles contained a 10 nt insertion between nt 992 and 993 corresponding to the junction between exons 6 and 7. This suggested that C8α-γ deficiency in these individuals was caused by a splicing defect. Genomic sequencing revealed a G→A point mutation in intron 6, upstream of the exon 7 acceptor site. This mutation converts a GG to an AG, generates a consensus 3′ splice site that shifts the reading frame, and creates a premature stop codon downstream. To verify that the point mutation caused a splicing defect, we tested wild-type and mutant mRNA substrates, containing 333 nt of the C8α intron 6/exon 7 boundary, in an in vitro splicing assay. This assay generated spliced RNA containing the 10 bp insertion observed in the C8α mRNA of affected patients. In addition, in mutant RNA substrates, the new 3′ splice site was preferentially recognized compared with wild-type. Preferential selection of the mutant splice site likely reflects its positioning adjacent to a polypyrimidine tract that is stronger than that adjacent to the wild-type site. In summary, we have identified a G→A mutation in intron 6 of C8A as a predominant cause of C8α-γ deficiency in African Americans. This mutation creates a new and preferred 3′ splice site, results in a 10 nt insertion in mRNA, shifts the reading frame, and produces a premature stop codon downstream.
Complement-dependent bactericidal activity, a vital component of innate immunity, develops upon assembly of the membrane attack complex from the terminal C components, C5–C9. A deficiency of any of the C5–C9 terminal C components is associated with loss of serum bactericidal activity and increased risk of meningococcal disease (1, 2). One of the terminal components, C8, is composed of three polypeptide chains (α, β, and γ), each encoded by a separate gene: A, B, and G, respectively (3, 4). Mature, functionally active C8 contains two noncovalently associated subunits: C8β and C8α-γ. In contrast, the two polypeptide chains of C8α-γ subunit are covalently linked by a disulfide bond (5).
Two C8 deficiency states have been described corresponding to the absence of either the C8β or the C8α-γ subunit (6). The two types of deficiency are clinically indistinguishable. Seventy-three patients with C8 deficiency were reported in comprehensive reviews of the literature through 1990 (1, 2). The type of C8 deficiency was delineated in 58 patients: 48 (83%) were due to C8β deficiency, and 10 (17%) were due to C8α-γ deficiency. Four additional C8α-γ–deficient patients have been reported since 1990 (7, 8). Of the 14 C8α-γ–deficient patients, 8 (57%) were African Americans, 4 (28.5%) were Japanese, 1 was Hispanic, and 1 was a Sephardic Jew. The genetic basis for C8α-γ deficiency has been defined for two of the Japanese patients (8) but remains uncharacterized in the African American population. We analyzed the C8A gene in four unrelated African American patients and identified a G→A mutation in intron 6 of C8A as a predominant cause of C8α-γ deficiency in African Americans.
Materials and Methods
Four unrelated, C8α-γ–deficient African Americans from different states within the United States and a healthy individual participated in this investigation after signing University of Iowa Institutional Review Board–approved consent forms. Three of the patients were referred for evaluation of infection, two with recurrent neisserial infection and one with pneumococcal bacteremia (Ref. 2; C8 kindreds 39, 48, and 49). The fourth individual was diagnosed during evaluation for chronic urticaria and progressive spondyloarthritis (9). Serum from all patients had absent or very low total hemolytic complement activity (CH50). C8 deficiency was established by a combination of immunodetection assays and functional assays for C8 in the sera from three individuals. CH50 activity in these sera was restored by the addition of C8. C8 deficiency was established in the fourth patient by immunodetection (9). C8α-γ deficiency was established in the first three patients by restoration of CH50 activity after mixing their sera with C8β-deficient sera, which contains functional C8α-γ, as reported previously (6). C8α-γ deficiency was established in the fourth patient by the genetic analysis reported in this study.
SDS-PAGE and immunoblotting
Immunoblotting was completed for two deficient individuals. Purified human C8 (Quidel, San Diego, CA) and serum samples were heated and resolved by SDS-PAGE, under both nonreducing and reducing conditions. The separated proteins were electrophoretically transferred to nitrocellulose and then Western blotted with specific polyclonal rabbit antiserum raised against C8α-γ (10).
Human fibroblasts were isolated from skin biopsies from the first three patients and a healthy control. Fibroblasts were cultured in MEM α and supplemented with 10% FCS, penicillin, and streptomycin. Total RNA was extracted by guanidinium isothiocyanate lysis, followed by CsCl density-gradient ultracentrifugation (11). Polyadenylated RNA was isolated using Poly (A)TRACT (Promega, Madison, WI).
Reverse transcription and PCR
mRNA was transcribed, PCR amplified, and C8α and C8γ cDNA sequencing completed for two deficient individuals. Reverse transcription was carried out, as described previously (10), using Moloney murine leukemia virus reverse transcriptase (Invitrogen, Carlsbad, CA), mRNA (0.1–0.5 μg/ml), and Oligo-dt (Pharmacia, Piscataway, NJ) or oligonucleotide primers based on the reported cDNA sequences for C8α and C8γ. Specific regions spanning the coding sequences for C8α and γ were amplified using primer pairs designed from published sequences (12, 13). An informative primer pair amplified the 558 bp fragment of C8α cDNA between bp 918 and 1451 (5′ primer, 5′-TTATTGGTGGGTGTAGGTGTATCC-3′; 3′ primer, 5′-CAGGAACGGTATGTAATGGTGCTC-3′). PCR products were sequenced directly (14).
Genomic DNA isolation, amplification, and sequencing
Genomic DNA was harvested from lysed fibroblasts or from peripheral WBCs from patient four, RNased (Boehringer Mannheim Biologics, Indianapolis, IN), and proteinase K digested (Boehringer Mannheim Biologics) (15). Genomic sequencing of C8A was completed for all four patients. Genomic DNA was amplified using primers based on both the published C8A genomic sequence (16) and additional intron sequence from GenBank. The junction between intron 6 and exon 7 was amplified using primers: 5′-ATTACAGGCATGAGCCACTGC-3′ and 5′-TTTGCTTTGTCAATCACCAGG-3′. Amplified DNA was sequenced directly in our laboratory (S35 Sanger Sequencing) or at the Genomics Division of the Iowa Institute of Human Genetics (BigDye Terminator chemistry – v3.1).
In vitro splicing reactions
Genomic DNA, from one healthy and one C8α-γ–deficient individual, was amplified using the primer pair (5′: [5′-GGGGAATTCATTACAGGCATGAGCCACTGC-3′]; 3′: [5′-CCGCTCGAGTTTGCTTTGTCAATCACCAGG-3′]) containing EcoRI and XhoI sites, respectively. These primers define a 333 bp fragment extending from 106 bp upstream to 227 downstream of the 3′ splice site (SS) of exon 7. The amplified PCR product was sequenced, digested with XhoI and EcoRI (New England BioLabs, Beverly, MA), and ligated into pHS1-X, which contains an HIV-1 cassette that can be transcribed in vitro (17). Plasmids were purified by CsCl and linearized by BamHI, 177 bp downstream of the intron 6/exon 7 junction. Linearized DNA was transcribed in vitro by T3 RNA polymerase (Ambion, Austin, TX) using radiolabeled 32P-UTP and added to splicing reactions incubated at 30°C for 2 h (17). Splicing reactions containing equal amounts of radioactivity were loaded onto denaturing polyacrylamide gels, run at 600 V, dried, and exposed to Hyperfilm MP (Amersham Life Sciences).
Sera from C8-deficient African American patients lack the C8α-γ subunit
A previous study showed that sera from C8α-γ–deficient patients contained low but detectable levels of a nonfunctional C8α-γ subunit (18). We looked for C8α-γ in sera from two deficient patients (Fig. 1). When resolved on nonreducing gels, the positive controls (purified C8 and healthy human sera) showed a reactive species migrating at 86 kDa, consistent with the presence of C8α-γ. In contrast, sera from the deficient patients, when probed with C8α-γ antiserum, contained no detectable immunoreactive subunit of this m.w. When resolved under reducing conditions, the pooled healthy human sera contained proteins migrating at 64 and 22 kDa (the molecular mass of C8α and γ, respectively), but patient sera showed no similarly reactive species. Identical results were obtained when the sera were selectively precipitated with polyethylene glycol or ammonium sulfate (data not shown). We conclude that the sera from these African American patients do not contain detectable levels of the C8A or C8G gene products or do so at levels below the assay sensitivity.
Patients’ C8α cDNA contains a 10-nt insertion
To determine whether affected patients express mutant C8α or C8γ, the corresponding cDNAs were analyzed by RT-PCR. Sequencing confirmed that the C8γ coding region in the two patients examined was normal. In contrast, the sequence of C8α cDNA contained a 10 bp insertion (5′-TTGCTGGCAG-3′) between nt 992 and 993 (Fig. 2). Comparing the sequence of the PCR product to the C8A genomic sequence revealed that nt 992 and 993 mark the splice junction between exons 6 and 7 and that the 10 bp sequence was identical to the intronic sequence immediately upstream of the junction between intron 6 and exon 7 (16), suggesting the existence of an insertional splicing defect.
Affected patients share a mutation that creates a new 3′ SS for exon 7
To determine why 10 bp are inserted at the boundary between intron 6/exon 7, we sequenced genomic DNA from healthy and deficient patients. Six of the eight C8A alleles contained a G→A transition, 12 residues upstream of the intron 6/exon 7 junction. This analysis confirmed an exact match between the 10 bp insertion in the cDNA from deficient individuals and the 10 bp downstream of the mutation in their genomic DNA (Figs. 2, 3A). Two patients were found to be compound heterozygotes. The two C8A alleles in these patients were determined to be due to a previously described C→T transition at bp 1407 in exon 9 (Supplemental Fig. 1) that creates a premature stop codon (8).
The shared G→A transition reported in this study converts a CGG in intron 6 to a CAG 12 nt upstream of the wild-type SS (Fig. 3B), thereby creating a new 3′ SS. Both the original and the newly created CAG 3′ SS for C8A intron 6/exon 7 exactly match the consensus 3′ SS sequence (19). The data also suggest that the mutant 3′ SS is highly preferred in vivo because no normal transcripts were detected in the RT-PCR sequences obtained from the C8α-γ–deficient patients.
The upstream mutant 3′ SS is used exclusively in in vitro splicing reactions
To confirm that the mutation created a new and preferred 3′ consensus acceptor site for exon 7, we tested its function in an in vitro splicing system. Genomic DNA containing the C8 intron 6/exon 7 boundary was purified from healthy and affected individuals (these sequences differed only in the G→A mutation). Amplified fragments were ligated into an in vitro transcription vector that contained an HIV-1 reporter, the resulting plasmids linearized and transcribed. The resulting RNA, composed of the heterologous HIV donor SS followed by 283 nt encompassing the C8A intron 6/exon 7 boundary (Fig. 4A), was tested in in vitro splicing reactions.
The results confirmed the point mutation creates a preferred 3′ SS (Fig. 4B). All unspliced transcripts (568 nt) comigrated, as did the products from the first splicing step, which contain the lariat plus the C8 exon 7. In contrast, both the free intron (released during the second step of splicing) and the spliced exons differed depending on whether the RNA sequence was from healthy or deficient subjects. In the latter case, the splicing products migrated as predicted if the G→A mutation at the −12 position had indeed created a new 3′ SS; the mutant transcript produced an excised intron that was slightly smaller (migrating at ∼396 versus ∼406 nt) and a spliced exon that was slightly larger (migrating at ∼308 versus ∼298 nt) compared with the wild-type samples. Moreover, although the mutant transcript still contained both 3′ SS, only the upstream mutant site was selected in the in vitro splicing reactions. Longer gel exposure did not reveal any mutant RNA transcripts spliced at the wild-type 3′ SS, suggesting the original 3′ SS is never used or is used at a frequency below the level of detection.
We explored the basis for C8α-γ deficiency in four unrelated African American individuals at the level of the secreted protein, mRNA, and genomic DNA. These patients harbored the same mutation in at least one of their C8A alleles: a G→A point mutation 12 residues upstream of the intron 6/exon 7 boundary. This mutation converts a CGG to a CAG and in the process creates a preferred 3′ SS upstream of the normal intron 6/exon 7 junction, causing exon 6 to be extended by 10 nt at the 5′ end. This 10-nt insertion shifts the reading frame and creates a stop codon at bases 1050–1052 (57 nt downstream from the mutant 3′ SS). This finding is consistent with the failure to synthesize C8α or the synthesis of a dysfunctional protein. Either mechanism would account for the absent hemolytic activity in the sera from these individuals and contribute to their risk of infection or autoimmune disease. If translated, the mutant transcript would be predicted to give rise to a 304 aa protein of 34 kDa (∼55% the size of the normal C8α). If this truncated form were synthesized, it might be able to form a covalent bridge with C8γ through cysteines 164 and 40 in the respective proteins (5). We were unable to detect either a smaller C8α-γ subunit or a truncated C8α protein product in the deficient sera, suggesting either that the aberrantly spliced transcript is subject to nonsense-mediated decay (20) or that a truncated protein might be unstable and quickly degraded.
Although disease-associated mutations that cause splicing defects are common, most either destroy the original site or activate cryptic sites; mutations creating a new and preferred SS are uncommon. A review of point mutations in the vicinity of mRNA splice junctions reported that 13 of 101 mutations generated a new SS. Of these, only six created a new consensus 3′ SS, and just three showed evidence that the novel SS was used (21). Of these, the aberrant splicing of phenylalanine hydroxylase mRNA is most similar to the one described in this study. In the case of phenylalanine hydroxylase deficiency, a G→A exchange at the −11 position in intron 10 causes a nine-base insertion in the mRNA that, although in frame, nevertheless destroys enzyme function (22). Thus, the molecular basis for C8α-γ deficiency in African Americans defined in this study is unique.
Structural features in pre-mRNA that affect recognition of a 3′ SS include the nucleotide immediately preceding the consensus AG, the proximity to the branch point, and the strength of the upstream polypyrimidine tract (23–25). For both the wild-type and mutant acceptor sites reported in this study, the nucleotide immediately preceding the consensus AG is a C, the most favorable from the standpoint of a 3′ SS consensus sequence (19). The in vitro splicing experiments demonstrate that mRNA substrates derived from healthy or deficient persons are functional, implying both SS are favorably situated with respect to any cis-acting elements that influence recognition of the wild-type site. Despite this equivalence and the presence of both site SS in the mRNA from the C8α-γ–deficient individuals, the new 3′ SS created by the G→A mutation in these individuals is by far preferred.
One possible factor driving selection of the mutant SS is the relative strength of the polypyrimidine tract. The most important determinants of polypyrimidine tract strength are the proportion of uridine (thymidines in sequencing reactions) residues in the 12-nt stretch preceding the CAG and the number of those residues that are consecutive (23–25). In the normal gene, those 12 nt contain five thymidines, of which just two are consecutive. In contrast, the 12 nt preceding the consensus 3′ SS in the mutated gene contain six thymidines, all but one of which are consecutive (Fig. 3C). This difference is even more pronounced if the comparison is extended further upstream. Thus, in C8α-γ–deficient individuals, the point mutation in C8A intron 6 creates a new 3′ SS with a substantially stronger polypyrimidine tract, which likely accounts for its highly preferred use.
The molecular basis for C8α-γ deficiency reported in this study differs from that reported in two unrelated Japanese persons (8). In those patients, three of the four null alleles involved a point mutation at the second exon–intron boundary, which inactivated the conserved 5′ SS and presumably leads to exon skipping. The fourth allele involved a C→T transition that generated a premature stop codon. Thus, two of the three distinct C8A null alleles described to date involve splicing defects. In contrast, in four of the seven described null alleles in the highly homologous C8B gene, unique C→T transitions produced premature stop codons; none of the remaining three null alleles was caused by a splicing defect (26, 27).
Our findings should be reconciled with those of Tedesco et al. (18), showing sera from deficient patients contain minute quantities (≤0.5%) of a normal-sized C8α-γ subunit, which lacked functional activity in hemolytic overlay agarose gels (18). One explanation for this discrepancy is that the molecular basis for the deficiencies may differ in persons from different ethnic backgrounds, as described above. Alternatively, splicing reactions in vitro and in vivo are distinctly different; in vitro, the substrate is a presynthesized transcript, whereas in vivo splicing is cotranscriptional. Therefore, splicing in vivo might allow recognition of the wild-type site, albeit at a very low frequency. However, if this were the case, one would expect the synthesized C8α-γ to possess normal hemolytic activity. The inability of Tedesco et al. (18) to detect hemolytic activity associated with the minute amount of C8α-γ subunit might reflect the different sensitivities of the functional and immunoassay systems. Last, neither the molecular basis for C8α-γ deficiency described here nor that described by Kojima et al. (8) sheds light on the mechanism underlying the reduced amount of C8β in these sera, other than to indicate it is not due to a direct effect of these genetic alterations on C8β.
In summary, we identified a G→A mutation in intron 6 of C8A as a predominant cause of C8α-γ deficiency in African Americans. This mutation creates a new and highly preferred 3′ SS, results in a 10-bp insertion, shifts the reading frame, and produces a downstream premature stop codon.
The authors gratefully acknowledge the excellent editorial assistance provided by Drs. Diana Colgan and John P. Atkinson and the supportive contribution of referring physicians Drs. Jerry A. Winkelstein, Mary Beth Fasano, James E. Peacock, David S. Stephens, Edward F. Hendershot, Carlos Del Rio, and Dai Park. BigDye Terminator sequencing data presented in this study were obtained at the Genomics Division of the Iowa Institute of Human Genetics, which is supported, in part, by the University of Iowa Carver College of Medicine.
This work was supported by a Merit Review award from the U.S. Department of Veterans Affairs (to P.D.) and by Public Health Service grants from the Division of Cancer Prevention of the National Cancer Institute (CA 28051) and from the National Institute of Allergy and Infectious Diseases (AI 36073) (both to C.M.S.).
The online version of this article contains supplemental material.
The authors have no financial conflicts of interest.