Intronic mutations affecting splicing of MBTPS2 cause ichthyosis follicularis, alopecia and photophobia (IFAP) syndrome


Dr Frank Oeffner, Centre of Human Genetics, Bahnhofstrasse 7a, 35037 Marburg, Germany, Tel.: 49(0)731/98490 40, Fax: 49(0)731/98490 20, e-mail:


Abstract:  Ichthyosis follicularis, alopecia and photophobia (IFAP) syndrome is an X-linked genodermatosis with congenital atrichia being the most prominent feature. Recently, we have shown that functional deficiency of MBTPS2 (membrane-bound transcription factor protease site 2) – a zinc metalloprotease essential for cholesterol homeostasis and endoplasmic reticulum stress response – causes the disease. Here, we present results obtained by analysing two intronic MBTPS2 mutations, c.671-9T>G and c.225-6T>A, using in silico and cell-based splicing assays. Accordingly, the c.225-6T>A transversion generated a new splice acceptor site, which caused extension of exon 3 by four bases and subsequently introduced a premature stop codon. Both, minigene experiments and RT-PCR analysis with patient-derived mRNA, demonstrated that the c.671-9T>G mutation resulted in skipping of exon 6, most likely because of disruption of the polypyrimidin tract or a putative intronic splicing enhancer (ISE). Our combined biocomputational and experimental analysis strongly suggested that both intronic alterations are disease-causing mutations.


maximum entropy model


polymerase chain reaction


Ichthyosis follicularis, alopecia and photophobia (IFAP) syndrome implies generalized ichthyosis, congenital alopecia, and photophobia (1–3). Non-consistent features include neurological anomalies (4) and further symptoms like inguinal herniation (5–7). In line with the X-linked mode of inheritance, only men suffer from the full-blown clinical picture, whereas female carriers display a mosaic pattern of minor symptoms (8).

Questions addressed

Recently, we have demonstrated that IFAP is caused by missense mutations in MBTPS2. Here, we address whether the disease could be provoked by alterations in the non-coding region of the gene. To this end, we combine biocomputational approaches and experimental validation to study two intronic alterations: c.225-6T>A and c.671-9T>G.

Experimental design

Our study includes three unrelated cases of IFAP syndrome. Patient 1 is a North-African boy, individual 6-II:1 as described previously (9). In addition to the IFAP triad, he displayed abnormal external genitalia and delayed psychomotor development. Patient 2, a Caucasian boy, also featured the typical IFAP triad. His psychomotor development was normal but he had inguinal herniation and short stature. Patient 3, a 26-year-old Ashkenazi man, had similar clinical features. A more detailed description of the patients’ medical history is provided in the supplementary material (Data S1).

MBTPS2 exons and flanking intronic sequences were PCR-amplified from patient DNA and sequenced as previously reported (9). Mutations were independently confirmed by methods like single-strand conformation analysis or amplification refractory mutation system technology. To exclude that nucleotide changes represented polymorphisms, we analysed DNA samples from >100 unrelated healthy controls and scrutinized the database NCBI_dbSNP for the presence of the mutated allele.

Splice scores were calculated using SpliceView and two different resources from the Human Splicing Finder (HSF) software package (10) – HSF matrices (11) and the MaxEnt program (12). To identify auxiliary splicing elements such as intronic splicing enhancers (ISEs), we employed ACESCAN2 and additional resources also supplied by HSF, like search for splicing silencer motifs (13) and identification of intronic identity elements (14).

To investigate experimentally the impact of intronic mutations, we adopted the pET01 minigene system (MoBiTec, Göttingen, Germany) endowed with an intrinsic splicing function (Fig. 1). This strategy – depicted more thoroughly in the supplementary section (Data S1, Table S1) – involved the cloning of wild-type and mutant MBTPS2 genomic DNA together with the analysis of transcripts generated by COS7 cells and additional RT-PCR analyses with cDNA from Patient 2 (Data S1).

Figure 1.

 Functional splicing assay for intronic MBTPS2 mutations. (a) Schematic of the minigenes constructed to analyse intronic MBTPS2 mutations. 5′- and 3′-exon are from the vector pET01. MBTPS2 exons and parts of flanking introns (IVS) are inserted at XhoI and BamHI restriction sites. PCR primers 2–5′ and 3–3′ (half arrows) annealing within exons 5′ and 3′ are employed for RT-PCR. The promoter regulating minigene expression is indicated by an arrowhead upstream of the 5′-exon. (b) Representation of inserted wild-type MBTPS2 fragments (WT1/2) and the intronic mutations MUT1 and MUT2. Exonic sequences in upper case, intronic in lower case. The mutated nucleotides are in bold and underlined. (c) Electropherograms of RT-PCR products from minigene transcripts. M = 100-bp ladder.


In addition to c.225-6T>A in Patient 1 (9), Patients 2 and 3 featured another change, c.671-9T>G. Analysis of control samples and dbSNP check-up excluded that either of these intronic mutations were polymorphisms. The presence of an MBTPS2-intronic SNP c.222G in Patient 2 in contrast to c.222A in Patient 3 supports an independent origin of the mutations.

For the c.225-6T>A change, all three algorithms predicted introduction of a new acceptor causing extension of MBTPS2 exon 3 (Table 1). Two programs suggested nearly identical strengths for the physiological acceptor and the novel splice signal (SpliceView: 88 vs 87, HSF: 87.69 vs 87.56). The MaxEnt algorithm, however, predicted a remarkably higher score for the introduced site (9.14 vs 5.05). Contrary to SpliceView and HSF matrices, MaxEnt considers interactions between nucleotides.

Table 1.   Prediction of the impact of intronic MBTPS2 mutations on splicing
AlleleSplice acceptor siteScore
Splice ViewHSF (S&S)MaxEnt
  1. Splice site alleles are reported in the first column, while the second column shows the corresponding nucleotide sequences (mutations bold, used splice sites underlined). Novel splice acceptor highlighted in grey. Third column: Splice site strength calculated by three different prediction algorithms.

  2. The bold values highlight the different scores for the c.225-6T > A mutation provided by MaxEnt.

WT c.225-6TcttttgacagGTTCAATT9190.3310.63
MUT c.225-6ActttagacagGTTCAATT8787.565.05
MUT c.225-6ActttagACAGGTTCAATT8887.699.14
WT c.671-9TctctttccagGTATCTGG9896.9713.97
MUT c.671-9GcgctttccagGTATCTGG9493.8712.56

With an S&S score difference of 3.1 (Table 1), the c.671-9T>G change has only marginal effects being reflected by minimal changes calculated by SpliceView and MaxEnt. On the other hand, ACESCAN2 suggested disruption of an ISE site (Fig. S1) and foretells – together with HSF – exonic splice silencer motifs in exon 6 (Fig. S2).

Because patient mRNA was not amenable, we relied on a minigene assay to examine the impact of c.225-6T>A on splicing. (Fig. 1 and Fig. S3).The corresponding construct produced a single RT-PCR product (Fig. 1c) featuring a transcript that was extended for the tetranucleotide acag (Fig. S3). Cloning of this fragment did not support an alternative use of both acceptor sites, as each of 30 clones yielded the extended transcript (data not shown).

The c.671-9T>G mutant minigene yielded a prominent band of 246 nucleotides demonstrating exon exclusion (p.Ile225LeufsX25) and an additional faint fragment of 365 nucleotides indicating inclusion of exon 6 (Fig. 1c). RT-PCR with the wild-type insert showed reversed electrophoretic band intensities, with the 365-bp fragment being much more abundant (Fig. 1c). Partial skipping of exon 6 was likewise confirmed by RT-PCR with human control RNA. Contrary to the minigene assay, the cDNA amplification of exons 5–7 from patient cells showed total exclusion of exon 6 (Fig. S4).


Only the MaxEnt algorithm suggested the preferred employment of the novel splice site, which had been created by the c.225-6T mutation. This called for the employment of integrated software tools instead of relying merely on stand-alone applications. We expected that in the relevant tissues of affected individuals, at least some pre-mRNA should be spliced correctly, as functional depletion of MBTPS2 in all cells is considered to be lethal.

The effect of the c.671-9T>G transversion is explicable by the disruption of a putative ISE (Fig. S2). Furthermore, c.671-9T>G lies in the polypyrimidin tract, and several mutations within this splicing element have been already discovered, which either promote exclusion or inclusion of exons (15,16). The preference of the short splice variant seen in patient blood cells might be specific for lymphocytes, because the minigene experiments provide evidence for a small quantity of mRNA including exon 6 (Fig. 1). Exon 6 exclusion was also indicated by wild-type minigene constructs and RT-PCR experiments with RNA from a healthy control, respectively (Fig. 1 and Fig. S4). We hypothesize that the splicing enhancer around position MBTPS2:c.671–9 is indispensable for counteracting the function of this exonic suppressor (ESS), concomitantly predicted by ACESCAN 2 and the HSF software. Our findings mirror the results reported by Nozu and collaborators (17), who also describe an ESS causing exon skipping of wild-type sequences. Similarly, we cannot fully exclude the possibility that our assay has limited splicing capability.

However, our study demonstrates the following issue: For detected intronic mutations, a combination of different biocomputational tools and functional splicing assays can estimate the impact on the phenotype with sufficient reliability, even if the relevant biological tissues are not available.

Online resources


We are deeply indebted to the patients and their families for their cooperation. We thank Dr. RC Betz (Bonn, Germany) for kindly providing the Exontrap Cloning Vector. This research was supported by BMBF (Netzwerk für Ichthyosen und verwandte Verhornungsstörungen, NIRK).