Editor—In the February 1998 issue of the Journal, Verlingue et al 1reported an absence of mutations in the promoter region of the CFTR (cystic fibrosis transmembrane conductance regulator) gene. They analysed a region that spans over 3.9 kb of sequences upstream of the first CFTR exon, including the CFTR promoter, down to 1.3 kb within the first intron. These sequences, shown previously to contain potential regulatory elements,2-4 had been selected on the basis of conservation throughout evolution (phylogenetic footprints) from rodents to primates.2 3 Verlingue et al 1 analysed a cohort of 205 subjects including patients with classical cystic fibrosis (CF), disseminated bronchiectasis, or congenital bilateral absence of the vas deferens (CBAVD), carrying either one or no mutation after scanning all 27 CFTR exons by DGGE (denaturing gradient gel electrophoresis). They further screened 5.2 kb of targeted sequences spanning the CFTR promoter region, but were unable to detect any putative disease related mutation in their sample.

We report the first three nucleotide alterations in the CFTR minimal promoter, defined as a 250 bp fragment upstream of the ATG translation start codon.4 These sequence changes were identified by DGGE followed by direct DNA sequencing on an ABI 377 automated sequencer by using primers specially designed for optimal amplification of this GC rich region (forward: 5′-GGCTCGAGGCTGGGAGTC-3′, reverse: 5′-TTCCATGGTCTCTCGGGCGCTGGGGT-3′). From a total of 450 patients referred to our laboratory for molecular analysis of the CFTR gene since 1989, we detected 118 different mutations responsible for CF or CBAVD (M Claustres, unpublished data). The high allelic heterogeneity of the populations studied might have raised the chances of discovering putative mutations in regulatory sequences.

Mutation −33G→A was identified in a male adult with CBAVD from southern France with an as yet undefined second mutation. The patient was homozygous for allele 7T at the splice acceptor site in intron 8 and heterozygous for M470V in exon 10.

Mutation −94G→T was characterised in a South African Black CF female. She had been diagnosed with CF on the basis of two positive sweat tests and severe chronic lung disease. The other allele remains unknown.

Mutation −102T→A was detected in two unrelated CF patients from southern France previously found to be compound heterozygotes for two CF mutations in the coding portion of CFTR. Careful familial segregation studies showed that −102T→A was associated in cis (on the same allele) with the mutation S549R (T→G). One female patient, born in early 1992 and diagnosed when she was 5 years old (sweat chloride value 118 mmol/l), had a genotype −102T→A+S549R (T→G)/ΔF508. The male patient, born in 1980 and diagnosed at the age of 9 (sweat test 74 and 96 mmol/l), had a genotype −102T→A+ S549R (T→G)/S945L. Both patients had mild pulmonary disease and were classified as pancreatic sufficient by the clinicians. As the S549R mutation has been previously described as a “severe allele” associated with pancreatic insufficient status,5 we wondered whether −102T→A could modulate the clinical phenotype in these patients. A small number of complex alleles containing more than one mutation have previously been observed in CF,6 7 and at least two of them, R553Q/ΔF5086 and 7T-R117H8 have been shown to contribute to a milder phenotype by inducing changes in the conformation of mutant protein or in the splicing of mRNA mutant transcripts or both. We have undertaken a collaborative study of phenotype/genotype correlations in patients carrying S549R (T→G), with or without −102T→A, and functional analyses in transient transfections in order to support our preliminary findings.

None of these sequence alterations was detected in a further 238 normal, 376 CF, and 158 CBAVD chromosomes from our sample, which suggests that they are either very rare polymorphisms or potential regulatory mutations. It is noteworthy that each of these variations is located in evolutionarily conserved sequences (S Vuillaumier, unpublished data). The −102T→A alteration was detected within a 28 bp sequence that shares 89% homology with a segment of the human α1(I) collagen promoter involved in the regulation of the gene.9 A computer aided search for the presence of transcription factor binding sites, using the Patsearch program,10 indicated that −33G→A is located in a PEA3-like motif, −94G→T in a GC box, and −102T→A in potential binding sites CCAAT-like or CarG-like among other possible cis acting elements. Interestingly, computer analysis of the antisense strand, determined that −102T→A creates a YY1 repressor site, which matches completely with the consensus sequence.11

There is a relative paucity of information regarding naturally occurring variation in the 5′ upstream regions of human genes, with available data indicating low nucleotide diversity at these sites. So far, only four other sequence changes have been described in the 5′ region of CFTR, three polymorphisms, −966T→G,1−895T→G, and −816C→T,12 and a putative deleterious mutation −741T→G.12 Moreover, in spite of several studies on the patterns of CFTR expression in men and rodents, the molecular mecanisms involved in CFTR gene regulation remain unknown. Further functional analyses will be required in order to determine whether the base pair substitutions reported in this study could lead to subtle variations in levels of CFTR expression.