1. ^*Klinik für Dermatologie, Charité, Schumannstr., Berlin, Germany
2. ^†Department of Medical Genetics, Academic Hospital, Hanzeplein, Groningen, The Netherlands
3. ^‡Department of Genotyping Facility, Medical Biology, Academic Hospital, Hanzeplein, Groningen, The Netherlands
4. ^§Max Planck Institut für molekulare Genetik, Ihnestr, Berlin, Germany
5. ^¶BARMER Ostseeklinik, Kirchenort, Seebad Prerow, Germany
6. ^#Fachklinik Bad Bentheim, Am Bade Postfach, Bad Bentheim, Germany

Correspondence: Dr med. John Foerster, Klinik fuer Dermatologie und Allergologie Charité, Humboldt Universitaet Berlin, Luisenstr. 5, 10117 Berlin, Germany. Email: john.foerster@charite.de

Received 1 August 2003; Revised 23 June 2004; Accepted 12 July 2004; Published online 21 December 2004.

Abstract

The psoriasis susceptibility locus 1 (PSORS1) mutation is assumed to reside within a region around human leukocyte antigen-C spanning 250 kb, termed risk haplotype (RH) 1/2. By re-analyzing a published data set with a previously developed method, the haplotype sharing statistic, we confirm localization of PSORS1 to the RH1 region and refine its location to marker M6S168. We replicate this result in an independent patient sample. The target region harbors fragments of a human endogenous retrovirus K (HERV-K) endogenous retrovirus. Two single-nucleotide polymorphisms with alleles differing between high- and low-risk haplotypes are located within the HERV-K dUTPase. One of these encodes a predicted non-conserved Glu–Arg exchange. The HERV-K dUTPase is expressed in peripheral blood and in normal as well as lesional psoriatic skin. Our results indicate that an endogenous retroviral dUTPase constitutes a candidate gene for the PSORS1 mutation.

The allele causing strong association of psoriasis with the so-called PSORS1 locus at human leukocyte antigen (HLA)-C is still elusive (reviewed in^{Elder et al, 2001;Capon et al, 2002}).^{Nair et al (2000)} identified an approximately 250 kb spanning interval (designated risk haplotype (RH)1/RH2) based on transmission disequilibrium text (TDT) analysis of founding haplotypes in a large type 1 psoriasis cohort. This interval contains three pseudogenes (HCGII-2, HCGIX-3, and NOB4, all on acc. Nr. AP000508), as well as two genes, corneodesmosin (NM__001264) and HCR (AF216493), with a putative role in the skin. The latter have both been proposed as PSORS1 candidate genes (^{Schmitt-Egenolf et al, 2001;Asumalahti et al, 2002}).

We re-analyzed the haplotype data published by (^{Nair et al, 2000}) using a previously developed haplotype sharing statistic termed HSS. We used the published transmission and non-transmission frequencies to assemble patient and control haplotype sets as detailed in the online supplement (Methods) and compared these sets for haplotype sharing using the HSS method. As shown in Figure 1a, excess haplotype sharing among patient haplotypes peaks strikingly at marker M6S168. The -log10 p-value at this marker of 11.9 corresponds to a corrected p-value of 1.7 10-11; however, excess haplotype sharing as such is detectable over most of the entire 1.1 Mb interval plotted. On the basis of haplotype sharing excess, we also performed a so-called HSS directional test (Figure 1b). This test further defines the interval likely to contain a disease-causing mutation at a marker exhibiting maximal excess haplotype sharing (see online Methods section). A clear decrease in excess haplotype sharing among patients is visible to the right of M6S105 in the telomeric direction and to the left of M6S172 in the centromeric direction. The curves intersect in the 6.4 kb interval bordered by M6S168 and M6S178. Taken together, these results confirm the localization of the PSORS1 mutation within the RH1/RH2 interval (dashed black line in Figure 1a) by an independent statistical method and suggest a more refined localization between markers M6S168 and M6S178. We also verified this mapping result in a novel patient sample (see online supplementary Fig S1). Thus, HSS analysis of the new patient sample and the data set from^{Nair et al (2000)} both suggest that the interval around M6S168 (between M6S105 and M6S178) may harbor a disease-predisposing sequence variant.

Figure 1.

Haplotype-sharing statistic (HSS) re-analysis of haplotype clusters from^{Nair et al (2000)}. The x-axis shows markers from (^{Nair et al, 2000}) telomere to centromere, beginning at marker D6S273. (A) HSS analysis, quantifying excess in mean haplotype length shared among patient haplotypes around each marker as compared with that among control haplotypes. The horizontal dashed lines represent PSORS1, the risk haplotype (RH)1/RH2 risk interval, respectively. (B) Directional analysis quantifying excess in mean haplotype length shared among patients as compared with controls only telomeric (solid line) or only centromeric (dashed line) from each marker. The arrowhead marking the intersection of both curves indicates the most likely interval harboring the PSORS1 mutation.

Full figure and legend (29K)

A BLAST search of this region revealed residual fragments of a human endogenous retrovirus K (HERV-K) family endogenous retrovirus, exhibiting the highest homology to the tandem-repeat virus HML-2.Hom on chromosome 7 (see online Fig S2). Two open reading frames (ORF), designated ORF22576 and ORF23048 based on their position on reference clone AC006048, represent segments of the C-terminal HERV-K Gag gene and the N-terminal Pro gene of the HERV-K genome, respectively. No HERV-K LTR segments are retained in the PSORS1 region. Focussing on the target region between 21 and 25 kb on clone AC006048 containing the two ORF, we identified sequence variations related to psoriasis risk status by comparing high- and low-risk haplotype DNA pools (outlined in online supplementary Table S2). Of 17 nucleotide variations identified within a 4.2 kb segment, four are located within ORF23048 (data not shown). Sequencing a 400 bp segment containing these variations in an extended patient sample revealed that only two of the four variations between the high risk (HR) and low risk (LR) pools correspond to single-nucleotide polymorphism (SNP) (data not shown). One additional SNP is located in the potential 3' untranslated region of ORF23048. These SNP were designated SNP143, SNP200, and SNP315 based on their position in the sequenced 400 bp segment. Their frequencies are detailed in online supplementary Table S3. All 10 individuals in the HR pool were homozygous for the psoriasis-associated haplotype C-A-G at SNP143/200/315, whereas all LR pool individuals were homozygous for the opposite haplotype (not shown).

ORF23048, which contains two of the three identified SNP, represents a fragment of the N-terminal HERV-K protease gene. In the intact retrovirus, this fragment is proteolytically processed, forming the retroviral dUTPase (online supplementary Fig S2). HERV-K dUTPases contain five evolutionary conserved motifs (^{Harris et al, 1997}), of which the most C-terminal motif is least conserved. The amino acid sequence homology between ORF23048 and the HERV-K dUTPase breaks off, after Y182 before motif 5, as a result of a frameshift, whereas homology extends through nt24274 on the nucleotide level (data not shown). The predicted amino acid variations between the HR and LR pools (G184R; R187K) immediately follow the break in sequence homology. The R187K substitution corresponds to SNP143 (GA), the A allele being associated with psoriasis. Based on the primary sequence, it is not possible to predict whether one or both of the PSORS1 dUTPase variants could be enzymatically active.

We next performed RT-PCR expression analysis dUTPase ORF from whole blood employing primer pair 7 (online methods, Appendix B), previously shown to be specific in the PSORS1 sequencing experiment. As shown in Figure 2a, we observed the expression of the dUTPase ORF in peripheral blood from three of four patients tested and all five healthy volunteers. The PSORS1 dUTPase transcript was also detectable in polyA+ RNA prepared from peripheral blood mononuclear cell (PBMC) of a healthy volunteer (Figure 2b). The identity of all PCR products was verified by sequencing (online Fig S4 and data not shown). The BLAST alignment of the sequenced RT-PCR product supplied in online supplementary Fig S4 demonstrates that the amplified transcript unequivocally derives from the PSORS1 locus based on the perfect match only to the PSORS1-containing clones. The SNP143 T allele was present in the three psoriatic blood samples and three of the five healthy controls expressed the SNP143 C allele. One sample expressed both alleles. Finally, we examined expression of the PSORS1 dUTPase in skin. As shown in Figure 2c, the dUTPase expression was detectable in healthy skin, as well as in one non-lesional and two lesional psoriasis skin samples, albeit only by nested PCR. Again, the derivation of all PCR products was verified by sequencing and BLAST analysis. The psoriasis-associated SNP143 C allele was found in one of the two lesional skin samples and one of the two PBMC samples. Taken together, these data indicate that the HERV-K dUTPase from the PSORS1 locus is commonly expressed in peripheral blood and is also detectable in skin.

Figure 2.

Expression of the human endogenous retrovirus K (HERVK) dUTPase at the PSORS1locus. (A) Peripheal blood from psoriasis patients (upper panel) or healthy donors (lower panel). RT-PCR was performed from total RNA with primers specific for the dUTPase open reading frame (ORF) or -actin. "+" and "-" indicate cDNA synthesis performed in the presence or absence of reverse transcriptase, respectively. The expected size of the PSORS1 dUTPase band is 430 bp and that of -actin is 466 bp. (B) Detection of the PSORS1 dUTPase in polyA+ RNA from peripheral blood mononuclear cells from a healthy donor, performed as in (A). (C) Detection of the dUTPase ORF at PSORS1 in skin by nested RT-PCR as detailed in online methods from two independent psoriatic plaque samples (pso L 1.2), non-lesional skin, and skin from a healthy proband (con skin). The expected size of the nested RT-PCR PSORS1 dUTPase product is 260 bp.

Full figure and legend (39K)

These data confirm localization of PSORS1 to the RH1/RH2 region defined by^{Nair et al (2000)} and suggest its fine mapping in the intervals around marker M6S168. A recent study on genetic association of psoriatic arthritis in a Spanish association also defines an interval including HLA-C and M6S168, but excluding markers in the HCR gene (^{Martinez-Borra et al, 2003}). That study independently confirms the PSORS1 interval suggested by our data. We suggest that HSS analysis be applied to existing SNP data sets in an effort to further confirm the current fine mapping of the PSORS1 location.

We identified a dUTPase derived from a defunct HERV-K endogenous retrovirus in the PSORS1 target interval. Alignment of the predicted protein sequence with other HERV-K dUTPases (online Fig S2) shows that the last of five conserved motifs has been lost from the PSORS1 dUTPase. Therefore, it is uncertain whether the predicted protein is enzymatically active; however, because viral, as opposed to host, dUTPases are active in the cytosol, even low additional PSORS1-derived dUTPase activity may contribute to viral persistence or replication in cell types not otherwise susceptible (^{Fleischmann et al, 2002}). Alternatively, an enzymatically non-functional PSORS1 dUTPase may contribute to psoriasis pathogenesis as a dominant-negative protein or auto-antigen. Independent of the dUTPase, a PSORS1 mutation located at M6S168 may exert positional effects on nearby genes.

Finally, our results suggest that expression of the PSORS1 retroviral dUTPase in PBMC is a common phenomenon. It will be most informative to test whether the recently described highly prevalent systemic IgG response against endogenous retroviruses in psoriasis patients (^{Moles et al, 2003}) includes antibodies against the HERV-K dUTPase fragment described in this report.

All studies involving human samples were approved by the Charité ethics review board. This study was conducted according to Declaration of Helsinki principles. All participants gave written informed consent prior to enrollment.