Spectrin St Claude, a Splicing Mutation of the Human α-Spectrin Gene Associated With Severe Poikilocytic Anemia

THE SHAPE of the red blood cell, as well as its unique mechanical properties, such as its remarkable deformability and prodigious stability under shear stress, depends on a multiprotein network, the membrane skeleton. The membrane skeleton is composed primarily of spectrin tetramers (αβ)₂ . These tetramers are composed of two head-to-head heterodimers consisting of long α and β-spectrin chains, laterally associated in an antiparallel arrangement.

The structure of spectrin has been analyzed by mild tryptic digestion. This technique has defined nine different domains resistant to trypsin: five in α-spectrin (αI through αV) and four in β-spectrin (βI through βIV).1 The primary structure of α and β-spectrin has been determined by peptide2 and cDNA sequence analysis.3,4 Both spectrin chains are primarily composed of repeating units 106 residues in length, which are folded in a coiled-coil triple helical arrangement.5 6 

The contributions of spectrin to the specific mechanical properties of the erythrocyte membrane skeleton are highlighted by its defects in disorders of red blood cell shape such as hereditary elliptocytosis (HE), hereditary pyropoikilocytosis (HPP), and hereditary spherocytosis (HS).7 8 Most of the defects identified in spectrin from individuals with HE and HPP are located either in the αI domain (the NH₂ -terminal end of a chain) or in the βI domain (the COOH-terminal end of β chain), the regions of spectrin that participate in heterodimer self-association, and lead to membrane instability.

We previously reported α-spectrin–related, severe poikilocytic anemia in a young patient.9 Tryptic digestion of spectrin from this individual showed an abnormal sensitivity to digestion. These studies localized the underlying molecular defect to the middle of the α-spectrin chain within the αII domain. In this report, we identify the underlying genetic defect in this individual, a polypyrimidine tract T → G mutation 13 nucleotides upstream of the 3′ acceptor splice site of exon 20 of the α-spectrin gene. This mutation abolishes the normal splice site and creates a new acceptor splice site. This results in the production of two α-spectrin mRNAs: one mRNA with the skipping of exon 20, the other one with the insertion of 12 intronic nucleotides upstream of exon 20, including a termination codon. We have designated this variant as spectrin St Claude. Spectrin St Claude constitutes a new, low expression allele of α-spectrin. These results highlight the importance of the regions of spectrin outside the self-association site in maintaining the stability of the red blood cell membrane and provide additional information to our understanding of mRNA processing and mechanisms of human genetic disease.

Case report.The case has been reported in detail by Lecomte et al.9 Briefly, the propositus (P) was born in 1979 as the second child of apparently unrelated parents, both from Guadeloupe in the French West Indies. The baby had a severe hemolytic anemia requiring frequent blood transfusions. After partial splenectomy in 1982, the patient had to undergo complete splenectomy 2 years later. Since that time, the propositus has experienced a compensated hemolysis. Blood smears show marked poikilocytosis with spherocytes, microspherocytes, and a few elliptocytes as observed in HPP patients. Both parents and a brother are completely asymptomatic with normal hematological parameters and red blood cell morphology. Ghosts from both parents have normal resistance to mechanical stress and normal deformability as demonstrated by ektacytometry.9 

Analysis of cDNA.Total reticulocyte RNA was prepared from freshly collected blood as described10 and reverse transcribed into single-strand cDNA using random hexamers.11 cDNA fragments coding for the αII domain of spectrin were amplified by the polymerase chain reaction (PCR) with Taq polymerase (Appligene, Illkirch, France) using the gene-specific primers listed in Table 1.4 After agarose gel electrophoresis, the PCR products (primers D and E) were excised and purified using QIAquick purification kit (QIAGEN-COGER, Courtaboeuf, France). Direct sequencing was performed by PCR amplification (Genome Express SA [Grenoble, France]) in a final volume of 20 μL using 100 ng of PCR products, 5 pmol/L of primer and 9.5 μL of dye terminators premix (Applied Biosystems, Foster City, CA) according to Applied Biosystems protocol. After heating to 94°C for 2 minutes, the reaction was cycled as follows: 25 cycles of 30 seconds at 94°C, 30 seconds at 55°C, and 4 minutes at 60°C (9600 Thermal Cycler; Perkin-Elmer, Roissy, France). Removal of excess of dye terminators was performed using Quick Spin Columns (Boehringer-Mannheim, Meylan, France). The samples were dried in a vacuum centrifuge and dissolved with 4 μL of deionized formamide EDTA pH 8.0 (5/1). The samples were loaded on an Applied Biosystems 373A Automated DNA sequencer and run for 12 hours on a 6% denaturing acrylamide gel.

Semiquantitation of PCR products was performed as following. The PCR reactions were performed in a volume of 25 μL containing 0.12 μg cDNA, 10 pmol of each primer F and G, 0.2 mmol/L of each dNTP (Life Technologies SARL, Cergy-Pontoise, France), 2.5 mmol/L MgCl₂ , 2 U Taq DNA polymerase (Appligene oncor), Tris-HCl 10 mmol/L pH 9, KCl 50 mmol/L, Triton X100 0.1%, bovine serum albumin (BSA) 0.2 mg/mL, 0.5 μCi ³²Pα-dCTP (Amersham). Samples were amplified using a Hybaid thermocycler for 20, 25, 30, or 35 cycles (each cycle consisted of 20 seconds at 95°C, 20 seconds at 53°C, and 20 seconds at 70°C), with an initial denaturation step at 95°C for 4 minutes. A total of 2 μL of the PCR product was mixed with 5 μL of loading buffer (10 mmol/L EDTA pH 8, 0.1% xylene cyanol blue, 0.1% bromophenol blue in formamide) and loaded on a denaturating 6% acrylamide gel (Long Ranger BioProbe, 7 mol/L urea and 1.2× Tris borate EDTA [TBE]). Electrophoresis was performed at 80 W for 3 hours. The PCR products were then analyzed using Instant Imager analysis.

Analysis of genomic DNA.Exons 19, 20, and 21, with their flanking intronic sequences, were amplified by PCR using the primers shown in Fig 1. Amplified products were purified using the Geneclean II kit (Bio 101, La Jolla, CA) according to the manufacturer's instructions. Forty microliters of purified product (corresponding to about 1 μg of double-stranded template) were added to 40 μL of streptavidin-coupled magnetic beads (Dynabeads M280 streptavidin, Dynal, Oslo, Norway) and single-stranded DNA was purified as described by the manufacturer. Sequencing reactions were performed with the antisense primers used for PCR amplification, and T⁷ sequencing kit (Pharmacia, St Quentin-en-Yvelines, France).

Restriction endonuclease digestion. MwoI (Biolabs-Ozyme, Montigny-le-Bretonneux, France) digestion of PCR-amplified genomic DNA was performed for 2 hours at 60°C according to the manufacturer's instructions.

Analysis of α-spectrin cDNA.Previous biochemical studies have shown that the defect in the α-spectrin chain observed in the propositus was located in the COOH-terminal region of the αII domain, adjoining the αIII domain. cDNA encoding the entire αII domain and the begining of the αIII domain was amplified by PCR with two sets of primers: A + C, which amplify a fragment of 748 bp corresponding to nucleotides 2368 to 3115 (exons 16 to 21) and B + E, which amplify a fragment of 566 bp corresponding to nucleotides 2866 to 3431 (exons 19 to 23).

In the normal control subject, agarose gel electrophoresis showed that both PCR reactions produced a single amplification product of the expected size. In contrast, in the propositus and his parents, analysis of both PCR reactions showed the presence of an additional amplification product about 100 bp shorter than the expected band (not shown). To further analyze this smaller amplification product, a third PCR amplification, encompassing exons 19 to 21, was performed using primers B + D, which amplify a fragment of 429 bp corresponding to nucleotides 2866 to 3294. Analysis of these PCR reactions showed that two discrete amplification products were present in reactions from the propositus and his parents, one migrating at the expected size of 429 bp and the other migrating around 329 bp, about 100 bp shorter than normal (Fig 2A). Nucleotide sequencing of the lower amplification product showed that the sequences encoding exon 20 were absent and a junction between exons 19 and 21 was present (Fig 2B). Surprisingly, the nucleotide sequence of the upper species of amplification product, thought to be normal in size, contained an additional 12 nucleotides, 5′-CAA TAA ATG CAG-3′, corresponding to the intronic sequence immediately upstream of exon 20¹² (Fig 2B). This insertion leads to the introduction of a stop codon (TAA) in the corresponding deduced amino acid sequence, predicting premature chain termination. No normal α-spectrin cDNA from this region was detected in the proband.

To examine whether the same splicing events occurred in the α-spectrin gene of the parents, PCR amplification of reticulocyte cDNA from both parents was performed using a new set of primers: F + G, which are expected to amplify a fragment of 217 bp corresponding to nucleotides 2971 to 3187. PCR products were separated on a 6% acrylamide gel, followed by detection with ethidium bromide. After a long-running electrophoresis, analysis of the PCR products showed: (1) a single band of the expected size in a normal control (217 bp); (2) two bands in the propositus, one corresponding to the insertion of 12 additional nucleotides (229 bp), the other to the skipping of exon 20 (124 bp); (3) three bands in both parents, one (229 bp) corresponding to the insertion of 12 additional nucleotides, another (217 bp) to the wild-type product, and the third (124 bp) to the skipping of exon 20 (Fig 3). These results demonstrated that α-spectrin cDNA from both parents contained wild-type product from the normal allele, as well as both products from the abnormal allele.

In the proband, semiquantitative reverse transcriptase (RT)-PCR with Instant Imager (Packard, Rungis, France) analysis shows the predominance of the mRNA species with the skipping of exon 20 compared with the mRNA with the 12 nt insert (75%/25%). In the parents, the product from the mRNA species with the skipping of exon 20 is also predominant compared with the mRNA with the 12 nt insert.

Analysis of α-spectrin genomic DNA.To determine the precise genetic basis of these splicing abnormalities, fragments corresponding to exons 19, 20, and 21 and their flanking intronic sequences were PCR-amplified using genomic DNA from the propositus and a normal control. These amplification products were subjected to direct nucleotide sequence analysis. The sequences of exons 19 and 21 and their flanking introns from the propositus were identical to control sequences. In the propositus, a T → G transversion was detected at position -13 upstream of exon 20 in the polypyrimidine tract (data not shown); the normal sequence was not found. This mutation creates a restriction site for the enzyme Mwo I. Mwo I digestion of PCR-amplified genomic DNA followed by agarose gel electrophoresis identified fragments corresponding only to the wild-type allele in a normal control; to the wild-type and mutant alleles in both parents and a brother, confirming heterozygosity for this mutation; and only to the mutant allele in the propositus, confirming homozygosity for this mutation (Fig 4).

Analysis of genomic DNA from individuals of European and African extraction.The presence of the intronic T → G mutation was investigated in 60 French white adults and 64 black adults from Benin, Africa, using Mwo I digestion of PCR-amplified genomic DNA. The Africans studied did not carry any of the widespread α^HE mutations reported so far in the Benin population, such as L154dup, L207P, or L260P.13 None of the amplified DNAs from whites were sensitive to Mwo I digestion. Among the Benin, African individuals, 4 individuals were heterozygous for the mutation as studied by Mwo I digestion. Thus, 4 of 128 alleles of individuals from Benin carried this mutation (3%).

In this report, we have identified the underlying molecular defect of the previously reported spectrin α^II/47 variant, renamed spectrin St Claude, observed in a young patient with severe recessive poikilocytosis. This patient is homozygous for a mutation of the α-spectrin gene, a T → G transversion at nucleotide position −13 upstream of the 3′ acceptor splice site of exon 20, which leads to abnormal splicing of the α-spectrin mRNA.

Mutations at nucleotide position −13 upstream of the 3′ acceptor splice site leading to abnormal splice variants have previously been reported. In the α-glucosidase gene, a T → G transversion leads to exon skipping, as well as utilization of a cryptic splice site in the exon.14 In the steroid 21-hydroxylase gene, a C → G transversion results in the activation of several cryptic sites.15 Other mutations upstream of the acceptor site have been reported such as −8T → G and −7C → G mutations in the β-globin gene, both leading to a decreased efficiency in the splicing process.16-18 In reported cases, these mutations occur in the pyrimidine-rich region and dramatically alter the splicing process. The pyrimidine tract plays an essential role in the splicing process,19 acting at an early stage of spliceosome assembly and playing a role in the first step of the catalytic process. Numerous studies using site-directed mutagenesis have shown that the length of the pyrimidine tract and its composition are important in the acceptor site recognition and that the inhibition of the splicing process depends on both the position and the nature of a substitution in the pyrimidine tract. At a given position, adenosine and guanosine residues are not equivalent; substitutions of pyrimidines by guanosine residues have a more dramatic effect on pre-mRNA splicing than substitutions by adenosine residues. In spectrin St Claude, the mutation abolishes the normal 3′ acceptor splice site of exon 20 and creates a new acceptor splice site (AT → AG) in intron 19. Similar effects on splicing have been described for mutations at nt −15²⁰ and −110²¹ in the β-globin gene.

In spectrin St Claude, the new splice site introduces 12 additional intronic nucleotides upstream of the coding sequence of exon 20 into one species of α-spectrin mRNA. The deduced amino acid sequence of this new mRNA species includes a stop codon (Fig 5) leading to a protein with a predicted molecular weight of 108 kD, roughly one third of the normal size (281 kD) (Fig 6). This short α-spectrin chain was detected neither in whole membrane proteins by denaturing electrophoresis nor in spectrin extracts by immunoblotting using antibodies directed against the αII domain of spectrin.9 This mutant protein is predicted to lack the entire COOH-terminus of α-spectrin including the nucleation site, an essential domain for α/β dimer formation and membrane assembly. As defined by Speicher et al,22 23 the lateral assembly between α and β chains begins at a specific nucleation site, located at the tail of the molecule and involves the COOH-terminus of the α chain and the NH₂ -terminus of the β chain (Fig 6); then, the two chains zip together to the head. As this truncated α chain cannot associate with β chain, it cannot be recruited to the membrane by this process and is probably degraded in the cytosol.

The splicing out of exon 20 leaves the open reading frame intact in this mRNA species, producing a protein that lacks 31 amino acids with an expected molecular weight of 277 kD. According to the triple helical structure defined from crystallography5 and nuclear magnetic resonance (NMR)6 data, the splicing out of exon 20 leads to the suppression of a B helix, from the residue G 926 up to the residue Q 958. This deletion occurs at the end of the αII domain of spectrin in repeating segment α9, immediately upstream of the SH3 domain (Fig 6). This deletion is consistent with the deletion of a protease hypersensitivity site previously located between αII and αIII domains by biochemical studies.9 

In the propositus, the splicing out of exon 20 seems to occur predominantly over the use of the new splice site as shown by the relative proportions of each product after PCR amplification (75%/25%). Thus, only the product of the skipping event, ie, 75% of the total synthesized α-spectrin, can be recruited to the membrane. Despite this feature, no spectrin deficiency was observed in the erythrocyte membrane of the proband. In mammals, α-spectrin is synthesized in excess.24,25 It has been shown that a marked decrease (40% to 50%) in the synthesis of α-spectrin related to a low expressed allele was not associated with an αβ-spectrin deficiency in the membrane.26 In a similar way, there is compensation for the loss of the severely truncated spectrin α chain, as also observed in homozygous carriers for the low expressed α allele, α^LELY.27 28 

The parents have no demonstrable truncated α-spectrin in their membranes. Because of the excess of normal α-spectrin produced from the wild-type allele, the abnormal spectrin is not or is very poorly recruited to the membrane in heterozygotes. Such features have already been observed in two spectrin variants, such as spectrin Dayton29 and spectrin Oran,30 both cases resulting from exon skipping, with a deletion of a B helix in repeat α2 and α8, respectively. In these two variants, only ≈10% of abnormal spectrin could be observed in the membrane of heterozygous patients, indicating a major disadvantage of the mutated α-chain dimer or tetramer compared with normal. In spectrin Dayton, pulse-chase analysis showed an increased degradation of this newly synthesized spectrin variant before its assembly to the membrane. The poor incorporation into the membrane of spectrin St Claude could also be explained by an increased sensitivity of the mutated α chain to proteolysis, as observed in tryptic peptides maps and during spectrin dimer storage.

The relationship between the molecular defect and the severity of the clinical phenotype in the homozygous patient is yet to be determined. The red blood cell morphology observed in the propositus is similar to that observed in patients with HPP and severe hemolytic HE. In many of these cases, the severity of hemolysis is directly related to the severity of the defect in spectrin self-association,31 which, in turn, is related to the specific causative mutation. Mutations located within the tetramerization site are associated with a severe spectrin self-association defect. In spectrin St Claude, as well as in spectrin Oran, the causal mutation, occurs far from the tetramerization site (repeat 9 and 8, respectively) and leads to only a minor defect in spectrin self-association. Thus, this does not account for the erythrocyte mechanical instability and hemolysis observed in the homozygous propositus.

The suppression of a B helix might disturb the spectrin structure in its central part. These results suggest that the role of spectrin dimers in membrane stability not only involves the interactions with itself to form tetramer or with other partners to build up the skeleton network, but also depends on the structural integrity of its central rod-shaped region.

The spectrin St Claude proband and both of his parents originated from Guadeloupe, an island in the French West Indies. Recent studies on the haplotypes of the β^S globin gene in Guadeloupian patients have revealed the predominance of the Benin haplotype (73% of all β^S chromosomes),32 suggesting a strong Benin genetic background in the Guadeloupian population. Because both spectrin St Claude parents were unrelated, asymptomatic heterozygous carriers of the −13T → G transversion, we looked for the −13T → G mutation in individuals originating from Benin. Interestingly, the spectrin St Claude allele was found in 3% of Benin individuals, suggesting that this allele may be particularly widespread in certain African populations. Additional epidemiological studies of the frequency of the spectrin St Claude allele in Guadeloupians and different African populations are currently in progress.

We are grateful to family O. for their kind cooperation. We thank M. Romana for making arrangements for patient and parents' blood collecting, B.G. Forget and L. Kotula for providing the flanking intronic sequences of the exons corresponding to spectrin αII domain, C. Glele-Kakaı̈ and I. Zohoun for providing Benin DNAs, C. Galand and G. Hetet for providing Caucasian DNAs.