Mutation of CERKL, a Novel Human Ceramide Kinase Gene, Causes Autosomal Recessive Retinitis Pigmentosa (RP26)

Abstract

Retinitis pigmentosa (RP), the main cause of adult blindness, is a genetically heterogeneous disorder characterized by progressive loss of photoreceptors through apoptosis. Up to now, 39 genes and loci have been implicated in nonsyndromic RP, yet the genetic bases of >50% of the cases, particularly of the recessive forms, remain unknown. Previous linkage analysis in a Spanish consanguineous family allowed us to define a novel autosomal recessive RP (arRP) locus, RP26, within an 11-cM interval (17.4 Mb) on 2q31.2-q32.3. In the present study, we further refine the RP26 locus down to 2.5 Mb, by microsatellite and single-nucleotide polymorphism (SNP) homozygosity mapping. After unsuccessful mutational analysis of the nine genes initially reported in this region, a detailed gene search based on expressed-sequence-tag data was undertaken. We finally identified a novel gene encoding a ceramide kinase (CERKL), which encompassed 13 exons. All of the patients from the RP26 family bear a homozygous mutation in exon 5, which generates a premature termination codon. The same mutation was also characterized in another, unrelated, Spanish pedigree with arRP. Human CERKL is expressed in the retina, among other adult and fetal tissues. A more detailed analysis by in situ hybridization on adult murine retina sections shows expression of Cerkl in the ganglion cell layer. Ceramide kinases convert the sphingolipid metabolite ceramide into ceramide-1-phosphate, both key mediators of cellular apoptosis and survival. Ceramide metabolism plays an essential role in the viability of neuronal cells, the membranes of which are particularly rich in sphingolipids. Therefore, CERKL deficiency could shift the relative levels of the signaling sphingolipid metabolites and increase sensitivity of photoreceptor and other retinal cells to apoptotic stimuli. This is the first genetic report suggesting a direct link between retinal neurodegeneration in RP and sphingolipid-mediated apoptosis.

Introduction

Inherited retinal degenerative diseases are characterized by the progressive loss of mature photoreceptor cells through apoptosis. Among them, retinitis pigmentosa (RP [MIM 268000]), the most common hereditary cause of blindness, comprises a clinically and genetically heterogeneous group of retinal disorders that affects ∼1.5 million people worldwide (Sullivan and Daiger 1996). Patients with RP typically show night blindness, gradual visual impairment, and bone spicule–like pigment deposits in the retina.

RP is a paradigm of monogenic diseases with extremely high genetic heterogeneity (Hims et al. 2003). This complexity has clearly hindered the identification of causative genes, particularly in the recessive forms, since large nuclear families suitable for linkage studies are scarce. In fact, although 39 genes and loci have been described, many others, still unknown, could account for the remaining unassigned cases. At present, the reported RP causative genes fall into four categories: (a) genes directly involved in the phototransduction cascade, (b) genes encoding proteins responsible for the structure and polarity of the photoreceptors, (c) genes encoding proteins of the visual cycle, and (d) regulatory genes (such as transcription and splicing factors). In the study of autosomal recessive RP (arRP), mutations on 16 genes and linkage to five chromosomal regions have been reported (Wang et al. 2001; RetNet Web site). Therefore, RP gene identification is still a priority in this field, since it not only is needed for diagnosis but also is crucial for giving insights into the molecular basis of the disease.

RP26, one of the remaining uncharacterized loci, was previously mapped to an 11-cM interval on 2q31.2-q32 in a consanguineous family with arRP (Bayés et al. 1998). Later, this 11-cM interval, between markers D2S148 and D2S117, was ascribed to a physical distance of 17.4 Mb that comprised more than 50 genes. As the incoming human genome data were made available in the public databases, new markers—either microsatellites or, mainly, SNPs—were retrieved and analyzed, first for cosegregation and finally for homozygosity mapping. All of this new information allowed us to narrow the candidate region to 2.5 Mb, and mutational screening was subsequently performed to identify the RP26 gene.

Material and Methods

Families and DNA

The two families with arRP we analyzed, P2 and E1, have been described elsewhere (Bayés et al. 1996, 1998). Blood samples were collected from family members, and DNA was extracted using a commercial kit (Wizard Genomic DNA Purification Kit [Promega]). Informed consent was gathered from all family members. This research followed the tenets of the Declaration of Helsinki.

Cosegregation and Homozygosity Mapping

The RP26 locus was further refined using three flanking markers: D2S2978, D2S2261, and D2S273 (table 1). For subsequent homozygosity mapping, additional markers (13 microsatellites and 101 SNPs) were chosen on the basis of their physical position according to the National Center for Biotechnology Information (NCBI) Human Builds 29–33 at the University of California Santa Cruz Genome Bioinformatics Web site (table 1). The SNP sequences were retrieved from the NCBI dbSNP Home Page and The SNP Consortium database. New SNPs observed during this study were submitted and deposited in the NCBI dbSNP database (see table 1). SNP analyses involved PCR assays with flanking primers and subsequent sequencing to verify the genotype. Three amplification steps (30 s each), with the annealing temperature ranging from 48°C to 61°C, were performed. The reaction mixture (50 μl) contained 10 μM of each primer, 2 μM of dNTPs, 1.5 mM MgCl₂, and 1 U of Taq polymerase (Innogenetics).

Table 1.

Microsatellite and SNP Markers Analyzed on 2q31.2-q32.3

Microsatellite or SNP IDa	Locationb	Locus	Genotype inRP26 Patientsc
D2S148; AFM200WA11	chr2: 176902551–176902745		Recombinant
D2S2978; ATA44H04	chr2: 179788236–179788348		Recombinant
D2S2261; AFMB072WG1	chr2: 180180111–180180442		Recombinant
rs1002207; TSC0036660	chr2: 180237706		Noninformative
rs1949453; TSC1026876	chr2: 180594170	UBE2E3	Homozygous
*rs2077783; TSC0046543	chr2: 180834855		Homozygous
*rs720453; TSC0046544	chr2: 180834909		Homozygous
*D2S2310; AFMB355XD5	chr2: 180849531–180849858		Noninformative
*rs155100; TSC0531809	chr2: 181033165	ITGA4	Homozygous
*rs155101; TSC0531810	chr2: 181033223	ITGA4	Homozygous
*rs155102	chr2: 181033525	ITGA4	Homozygous
*rs155103	chr2: 181033821	ITGA4	Homozygous
rs1801262	chr2: 181226533	NEUROD1	Homozygous
rs2583016	chr2: 181228296	NEUROD1	Homozygous
*rs1157595; TSC0326375	chr2: 181432646		Homozygous
*D2S364; AFM303YA9	chr2: 181717611–181718002	PDE1A	Homozygous
*rs3754929	chr2: 181771662	PDE1A	Homozygous
*rs1438065; TSC0655111	chr2: 181902071	PDE1A	Homozygous
*rs833152; TSC0655110	chr2: 181902179	PDE1A	Noninformative
rs288254; TSC0681111	chr2: 182291705	ERdj5	Noninformative
rs7510	chr2: 182472727	NCKAP1	Heterozygous
ss12676100	chr2: 182476461	NCKAP1	Heterozygous
ss12676101	chr2: 182476476	NCKAP1	Heterozygous
ss12676102	chr2: 182482636	NCKAP1	Heterozygous
rs2271671	chr2: 182510096	NCKAP1	Heterozygous
D2S350; AFM292WD1	chr2: 182532022–182532389	NCKAP1	Heterozygous
ss12676103	chr2: 182678153	LOC129401	Heterozygous
ss12676104	chr2: 182678197	LOC129401	Heterozygous
ss12676105	chr2: 182678220	LOC129401	Heterozygous
D2S2273; AFMB297XC1	chr2: 182825486–182825797		Heterozygous
D2S2281; AFMB310XF5	chr2: 182863973–182864335		Heterozygous
D2S2366; AFMA057VG9	chr2: 183175061–183175459		Noninformative
rs826135; TSC1254035	chr2: 183459257		Noninformative
rs826134; TSC0476562	chr2: 183459267		Noninformative
ss12676106	chr2: 183459509		Heterozygous
rs826133; TSC0658245	chr2: 183459526		Noninformative
ss12676107	chr2: 183459659		Heterozygous
ss12676108	chr2: 183459776		Heterozygous
rs826132; TSC0658244	chr2: 183459802		Heterozygous
D2S1391; GATA65C03	chr2: 183675508–183675835		Homozygous
rs1443021; TSC0664031	chr2: 183910062		Heterozygous
rs1443022; TSC0664032	chr2: 183910149		Heterozygous
rs768352; TSC0076040	chr2: 184012769		Homozygous
rs994653; TSC0317578	chr2: 184193063		Homozygous
rs728534; TSC0064305	chr2: 184484995		Homozygous
rs3046266; TSC1530131	chr2: 184485293		Noninformative
rs1366842; TSC0514415	chr2: 184485321		Noninformative
rs3731834	chr2: 184486442		Homozygous
D2S1361; GATA14E05	chr2: 184899775–184900087		Heterozygous
rs1016410; TSC0098979	chr2: 185334459		Heterozygous
rs1019430; TSC0223209	chr2: 185334589		Heterozygous
rs878845; TSC0212635	chr2: 185580725		Heterozygous
ss12676109	chr2: 186138356	ITGAV	Heterozygous
ss12676110	chr2: 186170238	ITGAV	Heterozygous
ss12676111	chr2: 186181185	ITGAV	Heterozygous
ss12676112	chr2: 186186304	ITGAV	Heterozygous
ss12676113	chr2: 186202416	ITGAV	Heterozygous
rs3816386	chr2: 186211713	ITGAV	Heterozygous
rs2018302; TSC0091486	chr2: 186467891		Heterozygous
rs1467990; TSC0384843	chr2: 186468012		Noninformative
rs2018314; TSC0085681	chr2: 186468088		Heterozygous
rs1000623; TSC0033486	chr2: 186468142		Heterozygous
rs840570; TSC1028023	chr2: 186855755		Noninformative
rs2308091	chr2: 186856059		Noninformative
rs84069	chr2: 186856203		Noninformative
rs84068	chr2: 186856285		Noninformative
D2S152; AFM207XG1	chr2: 186914674–186915071	CALCRL	Noninformative
rs1398061; TSC0576737	chr2: 186941063	CALCRL	Noninformative
rs1464338; TSC0378776	chr2: 187177069		Noninformative
rs1464337; TSC0378775	chr2: 187177070		Noninformative
rs2176858; TSC1217605	chr2: 187177146		Noninformative
rs2033838; TSC1047282	chr2: 187551834		Heterozygous
rs2028374; TSC1091521	chr2: 187666577		Heterozygous
rs1354906; TSC0490640	chr2: 188129892	CED-6	Homozygous
rs1007120: TSC0084553	chr2: 188175289		Noninformative
rs925825; TSC0241755	chr2: 188431720		Heterozygous
rs893407; TSC0165376	chr2: 189033193		Heterozygous
rs2289404	chr2: 189318673	ORMDL1	Noninformative
rs6942; TSC0023616	chr2: 189319070	ORMDL1	Homozygous
rs3791767	chr2: 189322574	ORMDL1	Noninformative
rs288817; TSC1078112	chr2: 189529648		Noninformative
ss12676114	chr2: 189529690		Homozygous
rs998173; TSC0011271	chr2: 189694293		Homozygous
rs3791789	chr2: 189787488	HIBCH	Heterozygous
D2S2262; AFMB082YE1	chr2: 189985176–189985493		Heterozygous
rs7721	chr2: 190069246		Heterozygous
rs1128723	chr2: 190069374		Heterozygous
rs8962	chr2: 190074316		Noninformative
rs1468685; TSC0385949	chr2: 190268436		Heterozygous
rs1263136; TSC1488116	chr2: 190298357		Noninformative
rs1263125; TSC0446273	chr2: 190303089		Homozygous
D2S118; AFM066XC1	chr2: 190309134–190309312		Homozygous
rs1263100; TSC0398872	chr2: 190327138		Heterozygous
rs1476896; TSC0398871	chr2: 190327184		Homozygous
D2S389; AFM333WF9	chr2: 190347305–190347628		Homozygous
rs1882395; TSC0896111	chr2: 190387668		Homozygous
D2S1775; GATA71C07	chr2: 190402485–190402614		Homozygous
rs3217036	chr2: 190462343	GLS	Noninformative
rs2355571	chr2: 190477428	GLS	Heterozygous
rs2883713	chr2: 190491082	GLS	Heterozygous
rs3199237	chr2: 190497863	GLS	Noninformative
rs3207595	chr2: 190497884	GLS	Noninformative
rs3207596	chr2: 190497887	GLS	Noninformative
rs3199238	chr2: 190497893	GLS	Noninformative
rs3207597	chr2: 190497917	GLS	Noninformative
rs1801893	chr2: 190498994	GLS	Noninformative
rs3215259	chr2: 190521768	GLS	Noninformative
rs1058589	chr2: 190522002	GLS	Noninformative
rs1058590	chr2: 190522011	GLS	Noninformative
rs1058591	chr2: 190522017	GLS	Noninformative
rs1058592	chr2: 190522020	GLS	Noninformative
rs1547550; TSC0429185	chr2: 190548390	STAT1	Heterozygous
rs925847; TSC0241788	chr2: 190600205	STAT4	Noninformative
rs3024891	chr2: 190601614	STAT4	Heterozygous
D2S2246; AFMB007WC1	chr2: 191085514–191085746		Heterozygous
D2S318; AFM105XC1	chr2: 191278428–191278834		Homozygous
D2S161; AFM224ZF4	chr2: 191431730–191432052		Heterozygous
D2S280; AFM155YE1	chr2: 191557728–191557932	TMEFF2	Homozygous
rs3738882	chr2: 191751560	TMEFF2	Homozygous
rs935367; TSC0352562	chr2: 191842107		Noninformative
D2S315; AFM081YG5	chr2: 192076101–192076267		Homozygous
rs717621; TSC0040380	chr2: 192153621		Heterozygous
D2S273; UT5048	chr2: 192288946–192289538		Recombinant
D2S117; AFM065YF11	chr2: 194331362–194331699		Recombinant

^aMicrosatellite IDs are those that begin with “D,” and SNP IDs are those that begin with “rs” or “ss.” SNPs IDs that begin with “ss” are those identified in the present study and are deposited in the NCBI dbSNP database. Microsatellite and SNP IDs preceded by an asterisk (*) are the flanking CERKL markers used for haplotype analysis and comparison in families P2 and E1.

^bLocation in NCBI Build 32.

^cBoldface italic type indicates candidate homozygous region.

Mutation Screening

Primers located at the flanking intron regions were designed to amplify each coding exon for all the analyzed genes in all samples. Three amplification steps (30 s each), with the annealing temperature ranging from 45°C to 60°C, were performed. Sequencing was performed using the BigDye Terminator Cycle Sequencing v3.1 kit and the automatic sequencers ABI 3700 and ABI 3730 (Applied Biosystems).

Mutation Restriction Analysis

A forward primer (5′-GAAGAATGCTGGGATGGAAACGGAC-3′) containing a mismatch nucleotide (underlined) that created an AvaII restriction site in the exon 5 wild-type sequence was designed. A three-step PCR (annealing temperature 50°C) was performed on the genomic DNA of the P2 and E1 family members, through use of the mismatch forward primer and exon 5 reverse primer (5′-GTTGTGCTGTCTAGATTAGC-3′), under the conditions mentioned above. The 157-bp PCR fragments were digested with AvaII and were resolved in a 4% low-melting-temperature agarose gel.

Human and Murine CERKL cDNA Characterization

Database searches from the available human genomic sequence allowed the identification of a truncated EST (GenBank accession number BE797822) and a partial cDNA (GenBank accession number BC020465), which overlapped 18 bp. To characterize the full-length cDNA, a human retina cDNA library (BD Biosciences) was screened (1.2×10⁶ plaque-forming units) with a mixture of all the CERKL coding exons as a probe. The hybridization was performed overnight under highly stringent conditions (42°C in 50% formamide, 5× Denhardt’s, 5× SSPE, 0.1% sodium dodecyl sulfate, and 100 μg/ml salmon sperm DNA), and several washes were performed at 55°C in 2× SSC and 1× SSC before autoradiography. Only a partial CERKL cDNA was obtained. For the full-length coding sequence, we devised a PCR strategy using two specific primers flanking the conceptual ATG and STOP codons on the Marathon-Ready retina cDNA (BD Biosciences) under the following conditions: 94°C for 30 s, 60°C for 30 s, and 72°C for 2 min for 40 cycles. The reaction mixture (50 μl) contained 10 μM of each primer, 2 μM of dNTPs, 1.5 mM MgCl₂, 1 U of Taq polymerase (Innogenetics), and 5% DMSO.

TBLASTN database searches at the NCBI BLAST server revealed two murine sequences homologous to human CERKL, one EST (GenBank accession number BY742285) and one partial cDNA (GenBank accession number BC046474), which had 377 nt of overlap with one another. Protein alignment was performed with CLUSTALW 1.8 at the BCM Search Launcher Web site and was depicted with BOXSHADE 3.21 at the EMBnet BoxShade Server.

Phylogenetic Analysis

Protein sequences from several members of the diacylglycerol, sphingosine, and ceramide kinase families were retrieved from public databases and aligned with human and murine CERKL through use of CLUSTALX version 1.64 (Thompson et al. 1997). The resulting phylogenetic tree was constructed using the neighbor-joining algorithm and considering the diacylglycerol kinase sequences as outgroup.

RT-PCRs

Two specific primers on exons 8 (5′-CCATTTAACAGCTCTGATGATGTGCAAG-3′) and 12 (5′-GACCTCTGATGCAACTTCCATTAAGTCAC-3′), which amplified 429 bp, were used to detect CERKL expression on a panel of first-strand cDNAs of several human adult and fetal tissues (BD Biosciences). Specific CERKL expression was confirmed by sequencing the amplified products. Amplification of GAPD was used to compare and normalize the samples, as suggested by the manufacturer. PCR conditions were as follows: 35 cycles of 94°C for 30 s and 65°C for 60 s. The reaction mixture (50 μl) contained 10 μM of each primer, 2 μM of dNTPs, 1.5 mM MgCl₂, and 1 U Taq pol (Innogenetics).

In Situ Hybridizations

For in situ hybridization, adult mouse eye slides from the C57BL/6J strain were obtained from Novagen. Eyes were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at 7 μm. Sections were washed for 7 min three times with xylene to remove paraffin, for 3 min twice with ethanol 100%, for 3 min with ethanol 90%, for 3 min with ethanol 70%, and for 3 min with diethyl pyrocarbonate–treated water, and then were fixed with 4% paraformaldehyde for 20 min. Sections were treated with 2 μg/ml proteinase K for 30 min, washed for 5 min twice with phosphate-buffered saline, and postfixed with 4% paraformaldehyde. Acetylation with 0.1 M triethanolamine-HCl (pH 8.0) containing 0.25% acetic anhydride was performed for 10 min. Slides were subsequently washed, ethanol dehydrated, and air dried. Hybridization was performed overnight at 55°C with 2 μg/ml digoxigenin-labeled riboprobes in 50% formamide, 1 × Denhardt’s solution, 10% dextran sulfate, 0.3 M NaCl, 10 mM Tris-HCl (pH 8.0), 5 mM EDTA (pH 8.0), 10 mM NaH₂PO₄, and 200 μg/ml salmon sperm DNA.

The riboprobes were generated from the T7 and T3 promoters of a pBluescript II KS vector containing either a 201-bp HindIII-PstI fragment, a 350-bp HindIII fragment, a 513-bp PstI-SphI fragment of the murine Cerkl cDNA IMAGE 4504238, or a 198-bp fragment from coding exon 4 of murine rhodopsin.

After hybridization, the slides were washed in 2× SSC for 20 min; washed for 5 min twice in a solution of 50% formamide and 2× SSC (washes at 50°C); equilibrated in NTE (0.5 M NaCl, 10 mM Tris-HCl pH 8.0, 5 mM EDTA) at 37°C; and then treated with 10 μg/ml RNase A in NTE at 37°C for 30 min. Subsequently, the sections were washed in NTE, 2× SSC, and 0.1× SSC (15 min each), equilibrated in buffer 1 (100 mM Tris-HCl pH 7.5, 150 mM NaCl), and blocked in blocking buffer (1% BSA and 0.1% Triton X-100 in buffer 1) for 30 min. An anti-digoxigenin-AP conjugate antibody (1:500; Roche Applied Science) in blocking buffer was incubated overnight at 4°C. The sections were washed twice in buffer 1 for 15 min and in buffer 2 (100 mM Tris-HCl pH 9.5, 150 mM NaCl, and 50 mM MgCl₂) for 10 min prior to exposure to the alkaline phosphatase substrate, nitroblue tetrazolium–5-bromo-4-chloro-3-indoyl phosphate (NBT-BCIP; Roche Applied Science). The reaction was stopped by several washes in distilled water. The sections were cover-slipped with 70% glycerol in PBS and were photographed using a Nomarski optics microscope (Axioplan, Carl Zeiss) equipped with a Nikon Coolpix digital camera.

Results

Refinement of the Cosegregation Region and Candidate Gene Analysis

The analysis of two proximal (D2S2978 and D2S2261) and one distal (D2S273) flanking markers (table 1) allowed further refinement of the RP26 locus to a 12.5-Mb interval (fig. 1A). As a first attempt to identify the causative gene, a mutational screening of candidate genes selected on the basis of their known retinal function or the phenotype of Drosophila mutants or knockout mice was performed. The analysis of five candidates (ORMDL1, ITGAV, GLS, NEUROD1, and FRZB) within this interval did not reveal any pathogenic mutation.

Figure 1.

The RP26 locus and the CERKL gene and protein. A, RP26 physical map of the 12.5-Mb cosegregation interval between markers D2S2261 and D2S273. B, Localization of CERKL within the candidate region (“CR”) defined by homozygosity mapping. Genes (rectangles) and the transcription direction (arrowheads) are indicated. C, CERKL exon/intron structure. The location of the R275X null mutation in exon 5 is shown.

Homozygosity Mapping and Mutation Screening

The reported consanguinity of the RP26 family and the availability of new markers after the sequencing of the human genome made homozygosity mapping feasible as a means to reduce the candidate region. Sixteen microsatellite markers and 101 SNPs located within the 12.5-Mb interval (table 1) were analyzed in all P2 family members. Eventually, only one homozygous segment spanning >1 Mb, between markers D2S2261 and D2S350 (fig. 1A), was identified. This candidate region, spanning 2.5 Mb, harbored five annotated genes and four partially characterized mRNAs (fig. 1B), which were prioritized for mutational analysis on the basis of (a) retinal expression, as assessed by PCR on a retina cDNA library; and (b) availability of functional data. All of them were eventually sequenced (in fact, two genes, NEUROD1 and FRZB, were analyzed in the previous section), but none showed any pathogenic variant.

Identification of a Novel Gene, CERKL

A more exhaustive database search for uncharacterized cDNAs and spliced ESTs within the candidate 2.5-Mb region was undertaken and revealed two close sequences, one EST (GenBank accession number BE797822) and one eye/retinoblastoma incomplete cDNA (GenBank accession number BC020465), located between the ITGA4 and the NEUROD1 genes. The conceptual protein from the in silico assembly of both sequences showed homology with eukaryotic ceramide, sphingosine, and diacylglycerol kinases and strongly suggested that the sequences analyzed corresponded to a novel, unreported gene. After screening a human retina cDNA library and performing serial PCR assays on retina cDNA, we obtained the full-length 1,596-nt coding sequence, which spanned 13 exons (fig. 1C). The predicted 532-aa protein showed the highest similarity (29% identity; 50% similarity) with the human ceramide kinase (CERK [Sugiura et al. 2002]) (fig. 2) and, therefore, the gene was named “human ceramide kinase–like” (CERKL [HUGO-approved nomenclature]). In addition, database searches revealed one murine EST (GenBank accession number BY742285) and one partial cDNA (GenBank accession number BC046474), which, upon assembly and conceptual translation, produced a protein that was highly homologous to human CERKL (75% identity; 85.6% similarity) (fig. 2). To verify that CERKL belonged to the ceramide kinase subfamily of lipid kinases, a phylogenetic tree was constructed using 24 protein sequences from different eukaryotic phyla. Human and murine CERKL sequences clearly clustered within the branch of ceramide kinases (fig. 3).

Figure 2.

Protein alignment of human (“h”) and murine (“m”) CERKL and CERK. Identical residues are highlighted in black, and conservative positions are in gray. The solid black line underlines the conserved COG1597 (LCB5) domain detected after comparison with the NCBI Conserved Domain Database. Murine CERKL sequence was assembled from GenBank sequences BY742285 and BC046474, human CERK sequence from AB079066, and murine CERK sequence from AB79067.

Figure 3.

Unrooted phylogenetic tree of several members of the diacylglycerol, sphingosine, and ceramide kinases, including the human and murine CERKL sequences described in the present study (boxed). The tree was obtained with the neighbor-joining algorithm of the program ClustalX version 1.64b. Numbers show the values for the bootstrap analysis. SPHKs = sphingosine kinases; CERKs = ceramide kinases; and DGKs = diacylglycerol kinases. Abbreviations and GenBank accession numbers are as follows: human sphingosine kinase 1 (hSPHK1), AAF73423; murine sphingosine kinase 1 (mSPHK1), AAC61697; human sphingosine kinase 2 (hSPHK2), AAF74124; murine sphingosine kinase 2 (mSPHK2), AAF74125; Drosophila sphingosine kinase (dSPHK), AAF48045; Ciona intestinalis sphingosine kinase (ciSPHK), AK112588; yeast sphingosine kinase Lcb4 (Lcb4p), NP_014814; yeast sphingosine kinase Lcb5 (Lcb5p), NP_013361; Arabidopsis thaliana sphingosine kinase (AtSPHK), AY128394; human ceramide kinase (hCERK), AB079066; murine ceramide kinase (mCERK), AB079067; Ciona intestinalis ceramide kinase (ciCERK), AK112750; Drosophila ceramide kinase (dCERK), AAF52040; human ceramide kinase-like (hCERKL), AY357073; murine ceramide kinase-like (mCERKL), BY742285 and BC046474; Caenorhabditis elegans ceramide kinase (cCERK), AAC67466; human diacylglycerol kinase β subunit (hDGKb), Q9Y6T7; murine diacylglycerol kinase β subunit (mDGKb), XP_147651; human diacylglycerol kinase epsilon subunit (hDGKe), NP_003638; murine diacylglycerol kinase epsilon subunit (mDGKe), NP_062378); Caenorhabditis elegans diacylglycerol kinase 1 (cDGK1), NP_508190; Arabidopsis thaliana diacylglycerol kinase 1 (AtDGK1), Q39017; Drosophila diacylglycerol kinase 2 (dDGK2), Q09103; and Drosophila diacylglycerol kinase 1 (dDGK1), Q01583.

Premature Truncation of CERKL Causes arRP

The CERKL coding exons were sequenced in all members of the RP26 family. All patients were homozygous for a nonsense mutation (R257X; CGA→TGA) in exon 5, which prematurely truncates the protein within the predicted catalytic domain (fig. 4A and 4B). This mutation was not observed in 170 unrelated, ethnically matched control individuals. We subsequently evaluated cosegregation of RP with two CERKL flanking markers (D2S2310 and D2S364) in a panel of small nuclear Spanish families with RP whose causative gene was still unknown. An additional cosegregating pedigree, unrelated to the RP26 family, was identified. The CERKL sequence analysis showed that the affected members were also homozygous for the R257X mutation (fig. 4B), thus reinforcing the pathogenicity of this CERKL variant. Although both families share the same mutation, the haplotype analysis of the 12 nearest flanking markers (see table 1), stretching 600 kb around CERKL, does not support a common ancestry (data not shown).

Figure 4.

A, DNA sequence of exon 5 in an unaffected individual and patient VI.15 of the P2 family, showing the homozygous nonsense R257X mutation. B, Pedigrees of P2 and E1 families, showing the ARMS analysis of R257X mutation. A specific primer was designed to generate an AvaII restriction site in the amplified wild-type allele. All RP-affected individuals are homozygous for the mutant, nonrestricted allele. Individuals whose DNA samples were not available are indicated by a number sign (#).

CERKL Shows a Tissue-Specific Pattern of Expression

Northern expression analysis of CERKL did not reveal any transcript on two panels of several human tissues, even after long exposure times (data not shown). This suggested either specific expression in nontested tissues or very low transcriptional levels, in contrast to the higher and more ubiquitous expression of the CERK gene (Sugiura et al. 2002). The more sensitive RT-PCR assays revealed moderate expression in adult human retina, kidney, lung, and pancreas, and low levels in brain, placenta, and liver (fig. 5). The RT-PCRs on human fetal tissues also showed expression in lung, kidney, and brain (fig. 5).

Figure 5.

CERKL and GAPD (control) RT-PCR expression analysis on several human tissues, including retina

To assess CERKL expression in the retina more accurately, we performed in situ hybridization on mouse eye sections. The mouse CERKL ortholog is predominantly expressed in the retina ganglion cell layer, although a faint signal is also detected in the inner nuclear and photoreceptor cell layers (fig. 6A and 6B). The axons from the ganglion cells merge in the optic nerve and, therefore, constitute the last cellular component of phototransduction in the retina.

Figure 6.

A, Cerkl in situ hybridization on murine eye sections with the antisense riboprobe. B, Cerkl in situ hybridization on murine eye sections with the sense riboprobe (negative control). C, Rhodopsin in situ hybridization on murine eye sections (positive control). RPE = retinal pigment epithelium; PhR = photoreceptor cell layer; ONL = outer nuclear layer; OPL = outer plexiform layer; INL = inner nuclear layer; IPL = inner plexiform layer; GCL = ganglion cell layer.

As a positive control, we assayed rhodopsin, which is highly expressed in the photoreceptor cells (fig. 6C). It is remarkable that, at a comparable specific labeling and concentration probe, rhodopsin hybridization signal was much more intense and was detected much earlier than Cerkl expression, again supporting the low transcriptional levels of the latter.

Discussion

The identification of CERKL and its characterization as an RP causative gene opens new avenues for evaluating its contribution—as well as that of other sphingolipid (SL) metabolism enzymes—to retinal function and disease. The pathogenicity of the R257X mutation is supported not only by the absence of this variant in the control population (the test on 340 chromosomes allows the detection of a polymorphism of 1% with 95% statistical confidence [Collins and Schwartz 2002]) but also, and with more relevance to function, by the premature truncation of the protein (position 257 out of 532 aa), well within the strictly conserved catalytic domain of ceramide, sphingosine, and diacylglycerol kinases (fig. 2).

During the last decade and after exhaustive cosegregation analyses using internal and flanking markers of all the reported arRP genes and loci (including RP26), we have identified the causative gene in only 8 (∼15%) families out of 52 Spanish pedigrees with arRP. When we consider all our data, CERKL (responsible for the disease in two families [present study]) appears to contribute to arRP to a similar extent as other well-known arRP genes, such as PDE6B (three families [Bayés et al. 1996]), TULP1 (one family [Paloma et al. 2000]), ABCA4 (one family [Martínez-Mir et al. 1998]), and CNGA1 (one family [Paloma et al. 2002]). Although it is too early to assess the real contribution of CERKL to retinal disorders, we surmise that, after this first study, additional families and new mutations will be readily described.

When we consider CERKL function, we should remember that SLs are structural membrane components, particularly abundant in neurons. SL accumulation is associated with severe neurodegenerative disorders (Gaucher, Tay-Sachs, and Niemann-Pick A and B diseases [Buccoliero and Futerman 2003]). Moreover, increasing evidence points to both nonphosphorylated and phosphorylated sphingosine and ceramide as essential second messengers in cellular stress, senescence, growth, and apoptosis (Luberto et al. 2002). SLs and their metabolic products act as biosensors of the cellular state, and the enzymes that regulate SL metabolism appear to be fine-tuned cellular switches, connecting several pathways with antagonistic properties (Hannun and Obeid 2002). Thus, ceramide and sphingosine lead to cell growth arrest and apoptosis, whereas sphingosine-1-phosphate and ceramide-1-phosphate have antiapoptotic and neuroprotective effects (Frago et al. 1998, 2003; Hannun and Obeid 2002; Spiegel and Milstien 2002, 2003). In this context, sensory neuropathies (hereditary sensory neuropathy type 1 [Dawkins et al. 2001]) and other neurodegenerative diseases (Batten disease [Puranam et al. 1999]) have recently been associated with ceramide metabolism enzyme deficiencies. It is puzzling that some key enzymes regulating intracellular ceramide levels have remained elusive (Hannun and Obeid 2002)—for example, the first ceramide kinase to be characterized was reported only recently (Sugiura et al. 2002). However, despite the short lapse of time, data on the function of ceramide-1-phosphate and ceramide kinase in several scenarios have increased, stressing the roles of the relative levels of ceramide and ceramide-1-phosphate as intracellular indicators for cell death/survival decisions (Frago et al. 2003) and adding a prominent role in the regulation/modulation of inflammatory responses (Pettus et al. 2003).

Photoreceptor neurodegeneration in RP has been reported to proceed by apoptosis triggered by cellular stress. However, none of the described genetic defects underlying inherited retinal dystrophies has provided a direct link to apoptosis. It is interesting that a very recent report showed that a decrease of the intracellular ceramide pools rescued photoreceptor degeneration in Drosophila mutants (Acharya et al. 2003). Ceramide may be produced in the photoreceptors in response to stress. In this case, deficiency of CERKL, a specific retinal ceramide kinase, would increase the ceramide pool or decrease the ceramide-1-phosphate levels, rendering the photoreceptor cells more susceptible to stress and, thus, more sensitive to apoptotic signals. The cumulative effects of this enhanced susceptibility could lead to progressive depletion of photoreceptors and, eventually, to blindness. Failure to process the ceramide to ceramide-1-phosphate may trigger cell death not only by increasing the endogenous ceramide pool but also by failing to activate the corresponding downstream signalling pathway. Whether CERKL is transcriptionally modulated in response to cellular stress stimuli remains an open question.

This study—to our knowledge, the first genetic report linking RP to ceramide-induced apoptosis—highlights the SL metabolism enzymes as new functional candidates and establishes novel targets for therapeutic intervention of retinal diseases.