Severe alpha-1 antitrypsin deficiency in composite heterozygotes inheriting a new splicing mutation QO_Madrid

The family was composed by four siblings, three males and one female, two of them PI*QO_Porto/QO_Madrid compound heterozygotes with severe AAT deficiency, and the remaining ones PI*M/QO_Porto as well heterozygotes but with moderate AAT deficiency (Figure 1). Their parents died years ago, and none of them had descendants or other blood relatives. Signed informed consent for the study was obtained from patients.

All these patients were previously characterized for AAT serum levels, AAT protein phenotype and genotyping of the common PI*S and PI*Z alleles. The determination of serum AAT was conducted by immunonephelometry with an autoanalyzer ArrayTM Protein System (Beckman-Instruments, Brea, California, USA).

Protease inhibitor (PI) typing was made by means of isoelectric focusing (IEF) technique as previously described [10],[35]. All four siblings were studied for the existence of AAT protein variants by IEF but conclusive patterns were not found.

In addition, all subjects were initially genotyped for the commonest defective variants PI*Z and PI*S by a LightCycler PCR, using primers and hybridization probes as described before [36]. None of these common deficient variants were found in our cases.

Since the low serum AAT concentration of the subjects could not be attributed to PI*Z genotypes, the entire coding sequence of the SERPINA1 gene was analyzed (Reference sequences NG_008290.1, NM_000295.4). Exons 2 to 5 and exon-intron junctions were analyzed by means of a Sanger automated sequencing, using previously described primers [37]. After that, to exclude the existence of other rare variants outside coding regions of the gene, primers for amplification of additional gene fragments to cover the whole gene were designed (Additional file 1: Table S1). The entire gene sequence were analyzed by Sanger automated sequencing (ABI PRISM 377 Applied BioSystems) in the four studied individuals of the family.

RNA extraction from peripheral blood was performed using RNAeasy kit (Quiagen) following manufacturer's recommendations. Then, cDNA synthesis was carried out by reverse transcription PCR (RT-PCR) using the Maxima First Strand cDNA Synthesis kit (Thermo Scientific). To analyze transcripts and splicing (from normal hepatocytes and peripheral blood samples) occurring in patients and controls, the regions between E1A/E2, E1B/E2 and E1C/E2 or E1C/E5 were amplified using several primers located in exon 1A, 1B, 1C, exon 2 and exon 5 as follows: E1A-F: 5'-TCCTGTGCCTGCCAGAAGAG-3'; E1B-F, 5'-ATCAGGCATTTTGGGGTGACT-3' [29]; E1C, 5'CTGTCTCCTCAGCTTCAGGC3'; E2-R, 5'-TTCTTGGCCTCTTCGGTGTC-3' and E5-R, 5'-CCATGAAGAGGGGAGACTTGG-3'. In addition, a primer localized in the initial region of intron 1C INT1C, 5'-GGGGATGGAGAATGTGAGCC-3' was also used to analyze the existence of possible transcripts using intron sequence in mutant cases. PCR amplification of the different products was performed under the following conditions: 35 cycles of 94°C for 45 s, 60°C for 45 s, and 72°C for 45 s. Amplified products were visualized in 1-2% agarose gels, purified by PCR purification Kit (Qiagen) and subsequently cloned into pGEM-T easy vector (Promega, Madison WI, USA) or directly analyzed by sequencing with an ABI PRISM 377 sequencer. Ligation reactions were used to transform DH5' competent cells. Clones containing PCR products were selected by blue/white colony and standard ampicillin selection. Positive transformants were analyzed by PCR and sequenced using AAT primers.

By using different prediction methods integrated in Alamut 1.3 software we evaluated splice signal detection (SpliceSiteFinder-like, MaxEntScan, GeneSplicer, Human Splicing Finder or Known constitutive signals) and Exonic Splicing Enhancers ESE binding site detection (ESEFinder) comparing reference and mutated sequences.

In order to analyze the effect of the splicing variants, a minigene containing exons 1B and 1C with the flanking intronic sequences were constructed. Exons 1B and 1C and part of the flanking intronic sequences (1,252 bp; Additional file 2: Figure S2) were amplified with Phusion High Fidelity polymerase (Fisher Scientific, Madrid, Spain) and forward, 5' GCTCTAGAACTAGTGGATCCCCCGGGGAGCAAAAACAGAAACAGG 3' and reverse, 5' ATAAGCTTGATATCGAATTCCTGCACTTTGTTGCTGTTGCTGTATC 3' primers (cloning tails are in italics). This fragment was cloned into the pSAD® splicing vector (patent# P201231427, Consejo Superior de Investigaciones Científicas, Spain) by overlap extension PCR [38], transformed into the DH5' strain of Escherichia coli (Life Technologies, Carslbad, CA, USA), and plated on LB-agar with ampicillin (Fisher Scientific) at 100 μg/μL, X-Gal (5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside, Fisher Scientific) at 40 μg/μL and IPTG 0.1 mM (isopropyl beta-D-thiogalactopyranoside, Fisher Scientific), where recombinant white colonies were selected. This construct constituted the minigene MGserp1a_ex1b-1c.

To include in the minigene the studied genetic variants, mutagenesis was carried out according to PCR mutagenesis protocol over the wild type minigene by using Pfu UltraHF polymerase (Agilent, Santa Clara, CA) and primers for variant c.-5 + 1G > A (QO_Porto) (F:5'ACTGACCTGGGACAGTGAATCATAAGTATGCCTTTCACTGCGA3', R: 5'TCGCAGTGAAAGGCATACTTATGATTCACTGTCCCAGGTCAGT3') and for variant QO_Madrid, c.-5 + 2dupT (F:5'CTGACCTGGGACAGTGAATCGTTAAGTATGCCTTTCACTGCGAGA3', and R:5'TCTCGCAGTGAAAGGCATACTTAACGATTCACTGTCCCAGGTCAG3').

For transfection, approximately 10⁵ HeLa (human cervical carcinoma) cells were grown to 90% confluency in 0.5 mL of medium (DMEM, 10% Fetal Bovine Serum, 1% glucose and 1% Penicillin/Streptomycin; Fisher Scientific) in 4-well plates (Nunc, Roskilde, Denmark). Cells were transfected with 1 μg of each minigene and 2 μL of Lipofectamine 2000 (Life Technologies). To inhibit nonsense mediated decay (NMD), cells were incubated with cycloheximide (Sigma-Aldrich, St. Louis, MO) 300 μg/mL (4 hours).

RNA from transfected cells was purified with the GeneMATRIX Universal RNA purification kit (EURx, Gdansk, Poland) with DNAse I treatment. Then, retrotranscription was carried out with 400 ng of RNA and the Transcriptor first strand cDNA synthesis kit (Roche Applied Science, Penzberg, Germany) and sequence specific primer SA2-PSPL3_RTREV (5'TGAGGAGTGAATTGGTCGAA3') that retrotranscribes only RNA produced by the minigene. Two to five μL of cDNA were amplified with Platinum Taq polymerase (Life Technologies) and specific RT-PCR primers of the vector exons, SD6-PSPL3_RTFW (5'-TCACCTGGACAACCTCAAAG-3') and SA2-PSPL3_RTREV. Samples were denatured at 94°C for 5 min, followed by 35 cycles consisting of 94°C for 30 sec, 59°C for 15 sec, and 72°C (1 min/kb), and a final extension step at 72°C for 5 min.

Semiquantitative fluorescent RT-PCRs were done with FAM-labelled SD6-PSPL3_RTFW primer as previously reported [39]. One μL of a dilution 1/10-1/20 of the RT-PCR products was mixed with 18.5 μL of Hi-Di Formamide (Life Technologies) and 0.5 μL of Genescan 500 Rox Size Standard (Life Technologies). To visualize the expression products, samples were run on an ABI3130 sequencer and analyzed with Peak Scanner (Life Technologies).

The PI*QO_Porto/QO_Madrid index case (II-4) was a male former smoker of 27 pack-years, diagnosed with severe COPD at age of 44, and currently receiving home oxygen therapy. He had dyspnea on moderate and small exertion, and radiological findings of diffuse panlobular emphysema and isolated bronchiectasis. Liver function tests were normal. He presented no detectable IEF bands and his serum AAT levels ranged from 9 to 17 mg/dL (Figure 1).

The other PI*QO_Porto/QO_Madrid sibling (case II-3) was a never smoker asymptomatic female that maintained a nearly normal respiratory function, but early functional signs of lung damage ((Transfer coefficient (diffusion constant) of CO diffusion (KCO) slightly decreased), and mild signs of lung emphysema, detected by computed tomography (CT) scan at the age of 41. She also showed absence of IEF bands, and AAT serum levels between 8 and 15 mg/dL. Her liver function tests were within normal parameters.

The remaining two siblings (II-1 and II-2) aged 51 and 55, with PI*M/QO_Porto genotypes had a moderate AAT deficiency, with AAT serum values around 70 mg/dL, and IEF revealing a normal electrophoretic pattern, compatible with an M phenotype. This partial reduction in the AAT levels but a normal protein phenotype corresponds to their heterozygous genotype, constituted by one normal (M) and one null allele (QO_Porto). These subjects suffered from a moderate congenital mental retardation of unknown etiology. None of them showed symptoms or signs of respiratory disease, and their lung function tests remain normal at present, despite both are former smokers (smoking exposure of 10 pack-years each one). The CT scan detected mild basal emphysema in II-1 and no pathological findings in II-2.

The entire sequence of the SERPINA1 gene, including coding and non-coding regions, was analyzed. DNA sequencing of the exon 1C and intron 1C revealed in the four siblings a heterozygous base substitution G to A at position +1 of the splice donor site of intron 1C (NM_000295.4:c.-5 + 1G > A), previously described as QO_Porto splicing mutation [29]. Analysis of the complete sequence of the SERPINA1 gene showed that this was the only pathogenic mutation found in individuals II-1 and II-2 (Figure 2). The heterozygous status for QO_Porto allele in these patients was compatible with their moderate AAT deficiency.

Interestingly, the other two family members, II-3 and II-4, who showed a severe AAT deficiency, in addition to the QO_Porto mutation were carriers of a duplication of the thymine (T) in position +2 of the same splice donor site of intron 1C (NM_000295.4: c.-5 + 2dupT) (Figure 2). Since splicing sites are highly conserved sequences, this variation presumably disrupts the splicing donor site of this intron. This effect would be compatible with the strong AAT deficiency that both subjects presented. Since the whole sequence of the gene, coding and noncoding regions, was analyzed in these two patients, and there was no other pathogenic change, we assumed reasonably that this change is a deficient allele, not previously described, that was termed QO_Madrid following the classical nomenclature agreed by experts in the field, with QO representing a non-expressing gene at the protein level, and Madrid the name of the place of origin of the first carrier of the allele [40],[41].

In addition, all four patients were heterozygous for the M2 (G/A, Arg101His, rs709932) and M3 (A/C, Glu376Asp, rs1303) common polymorphisms of the SERPINA1 gene. Sequencing of the expression products generated by using primers located in exon 1C and exon 5 demonstrated that patients II-1 and II-2, carriers of the QO_Porto allele, only expressed transcripts with the M3 (C, Asp376) variant. Since QO_Porto mutation affects the normal expression of AAT mRNA, QO_Porto mutation in these patients occur in the other not expressed allele carrying the M2 (A, His101) but not the M3 (C, Asp376) variant (Figure 1). Consequently, the QO_Madrid mutation found in patients II-3 and II-4 segregated with the M3 (C, Asp376) variant.

Thymine duplication at the site of splicing of intron 1C (i.e., the QO_Madrid mutation) is located in a highly conserved splicing sequence of vertebrate splice donor sites and may lead to deregulation of gene expression. Computational predictions of splicing signals demonstrated disappearance of the constitutive donor splicing signals in the case of the QO_Madrid mutation, similarly to what happened for the QO_Porto splicing mutation. In addition several putative binding sites for Exonic Splicing Enhancers (ESE elements), such as SC35, SRp55 or SRp40 splicing factors, were predicted to be also affected by the duplication of T in position +2 of the QO_Madrid allele. Thus, SC35 site would be disrupted by QO_Madrid mutation and the predicted scores for SRp55 or SRp40 were reduced (Figure 3).

To further investigate the molecular effect on splicing and expression of the SERPINA1 gene with the QO_Madrid mutation which co-occurred with QO_Porto mutation in patients II-3 and II-4, expression products generated by different primer pairs in exons 1A, 1B, 1C and exons 2 and 5 were analyzed (Figure 4A). RT-PCR expression analysis of fragment E1C-E2 showed normally spliced products in control liver tissue and peripheral blood, as well as in the two moderate AAT deficiency patients II-1 and II-2 with PI*M/QO_Porto genotype (Figure 4B). In these patients sequence analysis demonstrated only one expressed allele based on the Arg101His polymorphism, corresponding to the normal M3 allele. However, no expression products were detected in patients carrying both splicing variants QO_Porto and QO_Madrid, clearly indicating that QO_Madrid mutation also impairs normal splicing of intron 1C and represents a new null allele of the SERPINA1 gene.

We also checked using primers in the E1C-Int1C whether transcripts including intron 1C sequence were expressed because of disruption of the splicing site due to the presence of both intron 1C splicing mutations, QO_Porto and QO_Madrid. Similar to what was described for the QO_Porto mutation [29], no expression products retaining intron 1C sequence were detected indicating that if these transcripts are synthesized they might be probably rapidly degraded.

Transcripts that include E1A and E1B were also analyzed. Amplification of the E1A/E2 fragment in all patients, normal peripheral blood samples, and in a normal liver sample analyzed, produced one single band visible in agarose gel in all cases. Sequencing analysis revealed that the only mRNA product resulted from the direct splice of exon 1A to exon 2 (Figure 4C). Alternative spliced forms, previously described by others in monocytes/macrophages [8],[42], were not detected in our samples. Moreover, in all patients transcripts which include E1A showed expression of both alleles, since both variants Arg101 and His101 (G/A) were found. This indicates that QO_Porto and QO_Madrid mutations do not impede transcription not requiring the splice donor site of intron 1C.

Regarding transcripts which included exon 1B, multiple expression products were detected after amplification with primer pair E1B-E2. After cloning, five different products were identified in patients II-1 and II-2 (Figure 4D). The most frequent species, as shown by the number of colonies, were those including exons 1B joined to 1C and to exon 2 (named as product [1] in Figure 4D). Alternative products ([2] and [3] in Figure 4D) generated using a cryptic splice donor site in exon 1B previously described, producing a 18 bp shorter fragment at the 3' end of the exon 1B, were also found [43]. In addition, alternatively spliced products formed by direct splice of exon 1B to E2 ([4] in Figure 4D), as well as transcripts retaining the intron 1B ([5] in Figure 4D), were also found but in less amount.

In patients with QO_Porto and QO_Madrid mutations (II-3 and II-4) only two different transcripts species were detected by using E1B and E2 primer pair. In these cases both fragments corresponded to RNA products containing only E1B joined to E2 (Figure 4D), one of the products previously described using the cryptic splice donor site in exon 1B [44]. However, these patients do not include E1C in transcription products since splice donor site of intron 1C is completely disrupted by the presence of QO_Porto and QO_Madrid splicing mutations.

The wild type (wt) and the mutant minigenes were functionally assayed in HeLa cells. The wt construct produced at least 7 different transcript as well as a wide range of rare splicing isoforms (Figure 5), where the skipping of both exons 1B and 1C (only with vector exons v1 + v2) was the most frequent event (51.3%). This isoform resembles the most abundant transcript of AAT in macrophages without exons 1B and 1C [43]. We also identified four distinct isoforms with exons 1B and 1C that resulted from the combination of four splice sites: the canonical donor and acceptor sites, and one donor (18 nucleotides downstream in intron 1B) and one acceptor (3 nucleotides downstream in exon 1C) cryptic sites (transcripts Tr5, Tr6, Tr7 and Tr8, together accounting for 22.1%; Figure 5). Finally, Tr1 and Tr2 would represent transcripts lacking exon 1B. These results are in accordance to the transcripts described by Rollini and Fournier (2000) [43]. Therefore, the splicing pattern of wild type MGserp1a_ex1b-1c mimicked that of macrophages cell lines.

Interestingly, mutations c.-5 + 1G > A and c.-5 + 2dupT triggered the elimination of six transcripts of the wild type minigene (Tr1, 2, 5, 6, 7, 8) and generated six new ones, Tr10, Tr11 and Tr12 (c.-5 + 1G > A), and Tr3, Tr4 and Tr9 (c.-5 + 2dupT). The V1 + V2 transcript was significantly enriched in c.-5 + 1G > A (83.0% vs. 51.3% in wt) although other three aberrant transcripts (Tr10-12) would correspond to partial retentions of intron 1C. The new mutation QO_Madrid (c.-5 + 2dupT) induced isoforms Tr3 and 4 that were relatively abundant (12.5 and 34.6%, respectively) and corresponded to exon 1C skipping and alternative usage of the two donor sites of intron 1B. This result reproduced the splicing profile of patient RNA.