METHODS

Participants

We recruited 13 families with 1–2 members affected by IPS and at least one affected and one healthy or affected sibling. We also included three families with one affected child but no siblings for haplotype analysis. Altogether, 20 affected members and 13 healthy siblings were available for DNA studies (fig 1). Overall, 14 families originated from a defined region of middle Norway and two from the adjacent region of middle Sweden. None of the families were known to be related, and we ascertained no consanguinity. TG-D clinically examined the Norwegian families, who were all verified ultrastructurally by skin biopsies in Heidelberg. AV clinically examined the Swedish patients. Affected members of all families displayed the typical IPS phenotype, including premature birth with variable perinatal complications. Each patient’s skin was affected severely at birth, but this turned gradually into a milder phenotype, with characteristic flat follicular hyperkeratosis. Hypereosinophilia was verified in all patients at varying ages. In two of the informative families, an affected sibling died from neonatal complications (not shown in fig 1). The ultrastructural findings were consistent with ichthyosis congenita type IV in all patients and showed characteristic aggregations of bending membranes of the horn cells of epidermis. (^1,20 and Gedde-Dahl Jr T personal communication, 2003).

Genotyping of families

We extracted genomic DNA from whole blood and genotyped it with radioactively or fluorescently labelled microsatellites. The initial genome scan consisted of 364 microsatellite markers selected with an average distance of 10–15 cM throughout the genome. For radioactive labelling, we used forward primers labelled with ³²P-adenosine triphosphate. We amplified the markers by polymerase chain reaction (PCR) in a 10 μl reaction volume that contained 10 ng genomic DNA, 1 μl 10X PCR buffer (200 mM Tris (pH 8.4), 500 mM potassium chloride, and 15 mM magnesium chloride; Invitrogen, San Diego, CA, USA), 0.3 μl nucleotides (dNTPs) (25 mM), 0.2 μl of each primer (10 μM), and 0.03 μl Taq DNA Polymerase (5 U/μl) (Invitrogen, San Diego, CA, USA).

For fluorescent labelling, we modified the forward primers with the fluorescent labels carboxyfluorescein (FAM), hexachloro-6-carboxy-fluorescein (HEX), or tetrachloro-6-carboxy-fluorescein (TET). We amplified the markers by PCR in a 10 μl reaction volume that contained 20 ng of genomic DNA, 1 μl 10X Platinum PCR buffer without magnesium chloride (200 mM Tris (pH 8.4) and 500 mM potassium chloride) (Invitrogen, San Diego, CA, USA), 0.3 μl magnesium chloride (50 mM), 0.3 μl dNTPs (25 mM), 0.05 μl of each primer (100 μM), 0.03 μl Platinum Taq DNA Polymerase (5 U/μl) (Invitrogen, San Diego, CA, USA). We subjected the reactions to an initial denaturation for 5 minutes at 96°C followed by 35 cycles at 96°C for 30 seconds, 55°C for 45 seconds, and 72°C for 45 seconds, before a final elongation at 72°C for 7 minutes. We separated the radioactively labelled PCR products on a 4.5% polyacrylamide gel with 7 M urea in 0.5 × TEB buffer. We developed radioactively labelled fragments on Kodak BioMax MS films (Amersham Biosciences, Piscataway, NJ, USA). We separated the pooled fluorescently labelled PCR products on a 4% polyacrylamide gel on an ABI 377 DNA Sequencer (Applied Biosystems, Foster City, CA, USA). We analysed fragments with Genescan software (version 3.1.3) and Genotyper software (version 3.7) (Applied Biosystems, Foster City, CA, USA).

Linkage and haplotype studies

We drew haplotypes with the Cyrillic family drawing program (version 2.1.3) (Cherwell Scientific Publishing, Oxford, UK). Two point LOD score calculations were made with the MLINK program of the LINKAGE package (version 5.1).²³ We performed linkage analysis assuming an autosomal recessive mode of inheritance, an equal male and female recombination rate, and full penetrance. We set the disease allele frequency to one per 150 000, as some obligate carriers were born outside the high risk area, and we gave the marker alleles equal frequencies in the population. As all parents were genotypically known but phase unknown, however, this material concerns “certain” families in which gene frequencies are irrelevant to the linkage analysis outcome.²⁴

Association studies

For the association study, we used the allele sizes of genetically linked markers in the affected children and their parents to construct the parental “affected” and “unaffected” (control) haplotypes. The regional aggregation of families made the parental control haplotypes safer than non-regionally weighted control families. This corresponds to the transmission disequilibrium test (TDT).

We calculated the association of allele sizes to the disease loci with the ψ² distribution.

RESULTS

Linkage and haplotype analysis

Genomewide scan

Genotype analysis excluded linkage to the known ARCI loci with markers that covered the six loci at chromosomes 2q34, 3p21, 14q11, 17p13, 19p12–q12, and 19p13.1–p13.2. An initial genomewide linkage analysis with 364 radioactively labelled markers was performed with four families affected by IPS. The markers were selected with an average distance of 10–15 cM throughout the genome. The results gave a first indication for linkage on five different chromosomal loci. The family material was expanded to include a total of nine families to investigate these loci. This gave a maximal two-point LOD score >1.0 (θ_max = 0) on chromosome 8 (LOD 1.70 with marker D8S277), chromosome 9 (LOD 1.70 with marker D9S60), chromosome 10 (LOD 1.17 with marker D10S215), chromosome 15 (LOD 1.24 with marker D15S125) and chromosome 20 (LOD 1.70 with marker D20S120). Investigation of flanking markers and haplotypes excluded linkage for the loci except for chromosome 9.

Refinement of the interval

The family material was extended to 13 families affected by IPS from Norway and Sweden. We tested the power of our study by simulation analysis with all 13 families; this provided a maximum expected LOD score of 3.73. To refine the region on chromosome 9, adjacent markers were selected from public databases (National Center for Biotechnology Information, Ensembl, and deCODE). We determined the markers’ order and positions by comparing information from all three databases. Further analysis resulted in a maximum combined LOD score of 3.73 with marker D9S778 at θ_max = 0. All families affected by IPS that we investigated supported linkage to chromosome 9q, yet with individually insignificant scores. Haplotype analysis showed key recombination events in two families. The IPS locus was restricted towards the centromere by a recombination event between markers D9S250 and D9S778. The telomeric border was defined by a recombination event between markers D9S61 and D9S63 (fig 2). This localised the IPS locus between the markers D9S250 and D9S63: a region of about 11 cM. A region consisting of 18 marker loci in the interval D9S250–D9S63 segregated completely with the phenotype in all families (table 1). This interval corresponded to a physical distance of approximately 9 Mb at 9q33.3–34.13 (table 2).

We performed association studies with the markers to identify a commonly shared haplotype in the families. This gave a positive ψ² score (p<0.05) for three adjacent markers: D9S1798 (p = 0.0012), D9S1821 (p = 0.0286), and D9S112 (p = 1.1×10⁻⁷) for allele sizes 240 bp (allele 3), 171 bp (allele 2), and 138 bp (allele 3), respectively. The results showed a possible common haplotype that consisted of these three markers. This region, however, was not shared by all individuals (loss of homozygosity). Investigation of the haplotypes that surrounded the three markers showed two possible core haplotypes, A and B, in the families. The central marker D9S904, which did not show association (p = 0.2142 for allele 1 and p = 0.4126 for allele 2), separated the two possible core haplotypes. The D9S904 marker allele was the only one shared by all individuals, which was illustrated in families IR61 and IR90 (fig 3). The two haplotypes were expanded with the centromeric markers D9S1789 and D9S1825 and the telomeric markers D9S918, D9S61, and D9S1795.

DISCUSSION

Autosomal recessive congenital ichthyosis is a heterogeneous group of disorders clinically and genetically. Ichthyosis prematurity syndrome seems to be a homogeneous subgroup clinically and from pathognomic ultrastructural features. A genomewide screen was initially performed on four families affected by IPS, and linkage to several chromosome regions was established. The material was extended to a total of 13 families, and linkage to all other chromosomes but chromosome 9 could be excluded. Linkage to the six previously known ARCI loci was excluded with markers from each region. All families analysed in our study supported linkage to a single chromosome 9 region, and they proved linkage with a maximum cumulative LOD score of 3.73. The relatively high incidence of IPS at birth in the three counties of middle Norway with a population of 600 000 suggests a strong and ancient founder effect. Haplotype analysis of the 11 cM region on chromosome 9 showed two possible core haplotypes, A and B, in the 13 families. The central region of the two haplotypes, represented by marker D9S904, is shared in all affected individuals. We expected a single common ancestor haplotype for the mutation in all families, but the results showed two haplotypes that imply the presence of two founder mutations. One possible explanation is that historical recombination events have changed the original haplotype at a resolution beyond the markers used. Alternatively, a mutant allele may have shifted haplotype by a historic gene conversion that has resulted in two separate haplotypes for a single mutation. With the hypothesis of one or two common ancestral chromosomes, the region could be restricted by allelic association to a 1 Mb region between marker D9S112 and D9S918 (fig 3). This 9q region needs to be characterised with additional microsatellite and biallelic single nucleotide polymorphism markers in affected individuals to define and expand the possible core haplotypes.

The 11 cM interval identified on chromosome 9q by linkage analysis contains about 80 known or predicted genes and transcripts. The 1 cM region predicted by the haplotypes and allelic association contains only 16 known or predicted genes and transcripts (GenBank).

Of these 16 genes, only eight has a known function, including the LIM homeobox transcription factor 1 β (LMX1B (MIM 602575)), zinc finger protein 297B (ZNF297B), Ral guanine nucleotide exchange factor RalGPS1A (RALGPS1A), angiopoietin like 2 (ANGPTL2 (MIM 605001)), solute carrier family 2 (facilitated glucose transporter) member 8 (SLC2A8 (MIM 605245)), zinc finger protein 79 (pT7) (ZNF79 (MIM 194552)), ribosomal protein L12 (RPL12 (MIM 180475)), and syntaxin binding protein 1 (STXBP1 (MIM 602926)).

To further restrict the candidate region, samples from additional families affected by IPS are important for the identification of critical recombination events and common haplotypes. A future characterisation of the gene responsible for IPS may lead to a better understanding of the mechanisms behind normal skin formation and possibly to a clarification of factors behind premature birth and genetic predisposition to atopy.