MATERIALS AND METHODS

Patients affected by HHT were diagnosed by EB or EM using the published criteria.³ All patients are of Italian ancestry and details on the city of origin of each family were also obtained. Detailed clinical data will be reported elsewhere.

We collected blood samples, after informed consent, from 54 index cases and from available relatives at risk of being affected on the basis of the family pedigree; two of them belonged to large families in which linkage analysis showed the involvement of ENG and were excluded from ACVRL1 analysis.

DNA from the remaining 52 index cases was extracted by routine techniques; exons 3, 7, and 8 were amplified with the primers reported by Berg et al¹¹ and analysed by SSCP with two different running conditions¹⁶ in order to reduce false negative results (exon 3, which contains a common polymorphism) or dHPLC plus SSCP (exons 7 and 8).

In each experiment, a control for which the presence of a wild type sequence had previously been shown by direct sequencing and, for exon 3, a carrier of the common polymorphism were also included.

All subjects displaying an SSCP (or dHPLC) pattern different from controls or carriers of the common polymorphism of exon 3 were sequenced, using the Big Dye terminator method associated with Taq FS enzyme (Cycle Sequencing reaction) and an ABI-PRISM 3700 DNA analyser (Applied Biosystems).

The sequences obtained were compared to the reported gene sequence (working draft NT 009609) using the BLASTN program.

RESULTS AND DISCUSSION

The 13 different mutations in exons 3, 7, and 8 found in 16 unrelated probands after screening the 52 HHT patients are reported in table 2.

Only the mutation found in case LPV has been already reported,¹¹ while the others are all new. Four mutations (cases LL, PS, SA, VC, table 2) cause a frameshift and introduce a stop codon and two mutations (cases Dit201 and MoA, table 2) introduce a stop codon; thus these six mutations are certainly responsible for the production of a truncated and defective protein.

Two mutations occurred twice (delH97N98 and R67W), but no relationship could be found up to the great grandparents for each couple of families; however, in both cases the family names originate from closely related geographical areas in north Italy.

Del H97N98 deletes two amino acids of the extracellular domain, where N98 is a potential glycosylation site; the two amino acids are conserved in Mus musculus and Rattus norvegicus and N98 is also conserved in Xenopus laevis and Gallus gallus. R67 is conserved in mouse and rat while a homologous amino acid is present in Xenopus laevis and Gallus gallus. The presence of the same change at a conserved amino acid in two unrelated (or distantly related) families is a point in favour of its causal relation to the disease. The same changes were not observed in any other allele examined (n=50).

P301, carrying the R67W mutation (table 2) was found in a family previously reported as unlinked to either chromosome 9 or 12.¹⁷ After re-evaluation of all family members according to the updated diagnostic criteria³ and a new linkage analysis, the evidence for exclusion of chromosome 9 remained strong, while the evidence for the exclusion of chromosome 12 was not significant; thus, the proband was included in the study. The extensive liver involvement in this family is confirmed and it is noteworthy that intrahepatic AV shunts are also present in the apparently unrelated family sharing the same mutation (case MC, table 2), which comes from the same geographical region. A different missense mutation at the same position, R67Q, was reported by Berg et al,⁷ confirming that the substitution of this arginine causes HHT2. In addition, Lux et al¹⁸ have shown the loss of signalling activity in vitro for mutations C51Y (case 3 in table 2) and R67Q.

The amino acid changes D330Y, C344F, A352P, P378L, and A400D (table 2, fig 1) all occur at highly conserved residues (Rattus norvegicus, Xenopus laevis, Gallus gallus, Drosophila melanogaster), so they are very likely to produce protein changes causing the disease. None of them has been found in any other examined allele (n=50).

The cosegregation of the mutations with the disease was analysed and shown in families of cases LPV, Dit201, MC, P301, VG, and Iac302, for which other affected members were available.

The genotypes of the common polymorphism of exon 3 c(intron 3 5`IVS+11)t (cp3), reported in the HHT mutation database, but not formally published, were found equally distributed in the whole group of HHT patients and in the control population (CC/CT/TT: 33.9/49.0/17.1 in patients and 30.8/ 53.8 /15.4 in controls); the overall C allele frequency is 0.58 and Hardy Weinberg equilibrium is conserved. In the families of cases Dit201, P301, and MC, cp3 did not segregate with the disease.

We also analysed the short tandem repeat (STR) D12S1677 which is located between exons 9 and 10; the size of the alleles was randomly distributed among the 52 probands and the probands of the two families carrying the same mutations shared one (cases PP/VG in table 2) or both D12S1677 alleles (cases MC/P301 in table 2).

The 29 mutations in the ACVRL1 gene reported so far in patients affected by HHT2 (table 1) are distributed in eight of the nine coding exons, but exons 3, 7, and 8 contain 59% of them. No mutations have been reported more than twice and no data on geographical distribution are available. Owing to the small number of mutations identified and to the extremely wide phenotypic variability, no detailed phenotype-genotype correlation is known.

A much lower incidence of lung AV malformation in ACVRL1 linked families was suggested by Berg et al,⁷ but recently Kjeldsen et al¹⁴ reported a T1193A (exon 8) mutation in a family with a high prevalence of PAVM and Trembath et al¹⁰ identified ACVRL1 mutations in HHT patients with clinical manifestations typical of primary pulmonary hypertension.

McDonald et al⁸ reported in a single large family that the mutation G998T (exon 7) was associated with a significantly higher liver involvement, with 17% of screened patients showing liver AV fistulas. They proposed that these clinical findings may be specific for the mutation, and stressed the extreme variability present even in subjects carrying the same mutation.

Intrahepatic AV shunts were present in six of the 10 patients carrying mutations in the ACVRL1 gene who agreed to undergo liver ultrasound examination. A positive family history for similar liver involvement is present in the families of cases P301, MC, and FP. In the same set of patients, lung x rays and pulse oximetry consistently yielded normal results. Our findings suggest the hypothesis that mutations in the ACVRL1 gene may be associated with higher risk of liver AV malformations.

Exon 3 contains information for the extracellular domain, while exons 7 and 8 contain information for a portion of the intracellular tyrosine kinase domain; however, at present, it is not possible to correlate precisely changes in the different segments of the protein with specific clinical symptoms.

Testing by SSCP and/or dHPLC exons 3, 7, and 8 of ACVRL1 and sequencing only cases showing abnormal patterns allowed us to show that 30.7% of a group of 52 unselected HHT Italian patients may be classified as HHT2.

The prevalence of the disorder in the population is not well determined, probably because of the wide phenotypic variability; the best available estimates are those for a well defined region of Haut-Jura, Lyon, France⁵ and for the county of Fyn, Denmark,¹⁴ in which a prevalence of 15.6 per 100 000 was calculated. In our sample, the geographical origin of the families was evenly distributed over the whole of Italy so there is no evidence for a specific geographical aggregation of the disease in Italy as in France⁵; preliminary unpublished data by EB indicate a prevalence of 18/100 000 in the region of Piacenza, in northern Italy.

The probands from two pairs of families (MC and P301, and PP and VG, respectively, table 2) shared the same mutations, one or two alleles at the intronic D12S167 polymorphism and at the common polymorphism, and came from the same geographical area. Thus, they are very like to have a common ancestor.

In conclusion, we report a consistent number of new mutations in the ACVRL1 gene identified in a large series of unselected Italian HHT patients; for six mutations an association with intrahepatic AV shunts was also documented. Our results contribute to the understanding of the spectrum of ACVRL1 mutations leading to HHT2 and the high rate of intrahepatic AV shunts in HHT2, if confirmed, will have relevant implications for genetic counselling and clinical management of patients.