Overexpression of CERKL, a gene responsible for retinitis pigmentosa in humans, protects cells from apoptosis induced by oxidative stress

Cloning of CERKL splicing variants in the human retina

Specific primers flanking exons 2 and 13 (Table 1) were used to clone CERKL alternative splicing variants by PCR from Marathon-Ready™ human retina cDNA (BD Biosciences, Franklin Lakes, NJ) under the following conditions: 94 °C for 30 s, 58 °C for 30 s, and 72 °C for 2 min for 35 cycles. The primer CERKL_E2_F annealed to exon 2 while the CERKL_E13_R primer annealed to exon 13 and introduced a BamHI site to facilitate subsequent subcloning. The 50 μl reaction mixture contained 10 μM of each primer, 2 μM of dNTPs, 1.5 mM MgCl₂ and 1 U of Pfu Turbo polymerase (Stratagene, La Jolla, CA). The PCR bands detected after electrophoresis on a 1.5% agarose gel were cloned into the EcoRV site of pBluescriptII KS and verified by sequencing. The full-length cDNAs of these CERKL isoforms were reconstituted by digesting the partial clones with EcoRI and BamHI and subcloning the corresponding fragments into a pBluescriptII KS vector that already contained the 5′ end of CERKL (the complete exon 1 and the stretch of exon 2 before the EcoRI site, obtained from the IMAGE clone 3870103).

RT–PCR analysis of the presence of CERKL splicing variants in human tissues

The splicing pattern of CERKL was analyzed in human retina cDNA and a panel of first-strand cDNAs from several human adult and fetal tissues, using specifically designed pairs of primers to amplify and distinguish each isoform (Table 1), as follows: 1) to detect splicing variants CERKLa and CERKLb (RT_CERKLab_F and RT_CERKLab_R); 2) to amplify splicing variant CERKLc (RT_CERKLc_F and RT_CERKLcd_R); 3) to detect splicing variant CERKLd (RT_CERKLd_F and RT_CERKLcd_R). Two-step PCR conditions were as follows: 35 cycles of 94 °C for 30 s and 58 °C for 30 s. The 25 μl reaction mixture contained 10 μM of each primer, 2 μM of dNTPs, 1.5 mM MgCl₂ and 1 U Taq polymerase (Promega, Madison, WI). The concentration of the Marathon-Ready™ human retina cDNA and first-strand cDNAs from the tissue panel (BD Biosciences) in the reactions were 0.01 ng/μl and 0.1 ng/μl, respectively.

Overexpression of CERKL isoforms in cultured cells and preparation of protein cell lysates

Expression vectors containing the full-length open reading frame of the four CERKL isoforms were assembled by PCR amplification of each variant using the CERKL_E2_F primer, which anneals upstream of the EcoRI site in exon 2 and the CERKL_E13_HA_R primer, which anneals in exon 13 (Table 1). Primer CERKL_E13_HA_R also encoded a C-terminal hemagglutinin (HA) epitope tag and included an EcoRV site for subsequent cloning. The PCR products were digested with EcoRI and EcoRV and cloned into a pcDNA3 (Invitrogen Life Technologies, Carlsbad, CA) that contained exons 1 and 2 of CERKL (previously cloned between HindIII and EcoRI sites). The integrity of all constructs was verified by sequencing. To generate the truncated RP mutant protein (R257X), site-directed mutagenesis was performed on the 2 constructs containing the isoforms CERKLa and CERKLb (encompassing the lipid kinase domain). The 847C>T change was introduced with the mutCERKL_F and mutCERKL_R primers (Table 1). In these constructs, the HA epitope was added to the N-terminus of the protein for immunodetection.

Human embryonic kidney (HEK293T) or African green monkey kidney (COS-7) cells were seeded at 3×10⁶ cells/dish in 10 cm cell culture dishes 1 day before transfection. Cells were tranfected with 24 µg/dish using either the pcDNA3 vector alone or each pcDNA3–CERKL construct and 60 µl/dish of Lipofectamine™2000 (Invitrogen Life Technologies), following the manufacturer’s instructions. After 2 days, the cells were harvested and lysed by sonication in lysis buffer [20 mM MOPS pH 7.2, 2 mM EGTA, 1 mM dithiothreitol, 10% glycerol, and complete™ protease inhibitor (Roche Diagnostics, Indianapolis, IN)], as described elsewhere [9].

In vitro and in vivo kinase activity assays

DAG and ceramide kinase activities were assayed in vitro as described [10]. Briefly, the source of the enzymatic activity were protein lysates from either HEK293T or COS-7 cells transfected with each CERKL isoform (250 µg total protein), the empty vector or a CERK-expressing construct (negative and positive control, respectively) obtained by sonication in 20 mM MOPS pH 7.0, 2 mM EGTA, 1 mM DTT, 10% glycerol plus protease inhibitors. The source of lipids were either: a) micelles obtained from commercial sphingolipids, which contained 880 µM lipids (ceramide and sphingosine, both from bovine brain sphingomyelin, or 1,2-dioleoyl-sn-glycerol, Sigma-Aldrich, St. Louis, MO), 1 mM cardiolipin, 1.5% β-octylglucoside, 0.2 mM diethylenetriamine-pentaacetic acid (DETAPAC); b) lipid micelles, obtained by sonication of 50 μg of lipids extracted from cultured 661W murine photoreceptor-derived cells in 1 mM DETAPAC, 7.5% β-octylglucoside; or c) total 661W cell lysates that were heat-inactivated at 65 °C for 20 min to abrogate endogenous enzymatic activities. Reactions were performed at 100 µl final volume containing 20 mM MOPS pH 7.2, 50 mM NaCl, 1 mM dithiothreitol, 2 mM EGTA, and 3 mM CaCl₂, with the corresponding lipid and protein sources. Reactions were started by adding MgCl₂ at 0.5 mM final concentration plus γ-³²P-ATP (2–5 µCi/reaction), incubated at 30 °C for 30 min, 60 min, or overnight, and stopped with 250 µl of chloroform, 250 µl of methanol and 125 µl HCl 2.4 M. After vortexing, the phases were separated by centrifugation, and the organic phase removed and vacuum-dried. Lipids were resuspended in 25 µl chloroform:methanol (95:5, v/v), spotted onto silica gel plates and separated by thin-layer chromatography (TLC). For the enzymatic assays in vivo, we seeded 1.5×10⁶ 661W cells on 10 cm diameter plates and transfected using either the two pEGP-CERKL long isoforms (a and b) or the pEGFP empty-vector (negative control). Transfected cells were enriched by using Fluorescence Activated Cell Sorting (FACS) to select green fluorescent protein (GFP) positive cells. These cells were reseeded on smaller plates, and 4 μl of radioactive ³²P-orthophosphate was added to the medium. Cells were collected after 4 h, and lysed directly in a solution of chloroform:methanol:chlorhydric acid (100:100:1). After vigorous vortexing, cellular debris was sedimented by centrifugation in a microfuge (100x g) and the liquid phase was vacuum-dried on completion. The lipid pellet was resuspended in a solution of 1:1 chloroform:methanol and loaded onto silica gel 60 TLC plates (Merck, Whitehouse Station, NJ). All TLCs (assays in vivo and in vitro) were run in chloroform:acetone:methanol:acetic acid:water (10:4:3:2:1, v/v). Radiolabeled lipids were visualized by autoradiography.

Purification of recombinant proteins

To express the four CERKL isoforms as glutathione-S-transferase (GST) fusion proteins, we subcloned each variant in-frame into pGEX-4T-1 (Amersham Biosciences, Chalfont St. Giles, UK), between the XhoI and NotI sites. The constructs were transformed into BL21 Codon Plus E.coli cells (Stratagene) to avoid premature protein truncation due to the strongly biased codon usage of the CERKL human gene. Recombinant proteins were produced in 500 ml cultures induced by the addition of 0.5 mM Isopropyl-β-D-thiogalactopyranoside (IPTG). Purified GST-recombinant proteins were obtained after elution with reduced glutathione of the protein bound to the glutathione-sepharose bead batch, following the manufacturer’s instructions (Amersham Biosciences).

Lipid-protein overlay assay

Binding of CERKL isoforms to sphingolipids was assessed using SphingoStrips™ and PIP Strips™ (Echelon Biosciences, Salt Lake City, UT), which consist of several nitrocellulose-immobilized sphingolipids and phosphoinositides (100 pmol/spot) respectively. The membranes were equilibrated for 5 min in 10 mM Tris-HCl pH 8.0, 150 mM NaCl, and 0.1% v/v Tween-20 (TBST). They were then blocked with 3% BSA in TBST (blocking solution) for 1 h at room temperature. Subsequently, the membrane was then incubated overnight at 4 °C in blocking solution containing 0.5 µg/ml of the GST-fusion protein on a rocking platform. The membranes were washed twice for 15 min in TBST with gentle agitation, and then incubated for 1 h with 1:1,000 anti-GST monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA). After washing as before, membranes were incubated for 1 h with 1:3,000 horseradish peroxidase-conjugated anti-mouse IgG antibody (Sigma). Finally, the membranes were washed 6 times in an hour in TBST, and developed with the ECL western blotting detection system (Amersham Biosciences).

Transfection, immunolocalization and confocal laser microscopy

COS-7 cells were grown in Dubecco’s Modified Eagle Medium (DMEM) containing 10% fetal bovine serum, 4 mM L-glutamine, 100 U/ml of penicillin, and 100 µg/ml streptomycin (Invitrogen Life Technologies). Cells were grown on coverslips in 12 well plates. Transient transfections were performed using Lipofectamine™2000 (Invitrogen Life Technologies) according to the manufacturer’s protocol. Next, 24 h or 48 h post-transfection cells were rinsed with PBS 1X (137 mM NaCl, 2.7 mM KCl, 1.5 mM KH₂PO₄, 8 mM Na₂HPO₄, pH 7.4) and fixed in 3% paraformaldehyde and 2% sucrose in 0.1 M phosphate buffer at 4 °C for 30 min. Cells were then washed and permeabilized for 10 min with a solution that contained 0.1% Triton X-100, 20 mM glycine, and 10 mM PBS. After permeabilization, cells were rinsed and blocked in a solution composed of 1% BSA, 20 mM glycine, and 10 mM PBS. Then, cells were incubated with 1:250 anti-HA polyclonal antibody and either specific 1:25 anti-calnexin, 1:250 anti-GM130, or 1:1,000 anti-EEA1 monoclonal primary antibodies (all from BD Biosciences) at 37 °C for 1 h. Mitochondria were labeled by adding reduced MitoTracker Orange (Molecular Probes, Invitrogen Life Technologies, Carlsbad, CA) to the cell culture medium at a final concentration of 500 nM for 45 min before the fixation procedure and subsequent immunocytochemistry. Upon washing, cells were incubated with 1:300 AlexaFluor 488-conjugated anti-rabbit and 1:300 AlexaFluor 546-conjugated anti-mouse (Molecular Probes) secondary antibodies. When required, slides were counter-stained with 1:200 DAPI (Sigma) nuclear blue dye in PBS for 15 min. All preparations were mounted in Vectashield medium for fluorescence (Vector Laboratories, Burlingame, CA) and analyzed by confocal laser scanning microscopy with Olympus Fluoview 500 (Olympus, Tokyo, Japan) and Leica TSC NT and TSC SPII (Leica, Wetzlar, Germany) laser scanning microscopes. In silico predictions for CERKL subcellular localization were performed using PSORT, ProSLP v2.0, ESLPred, and SubLoc v1.0.

Subcellular fractionation by sucrose gradient

HeLa cells (human cervical cancer derived cell line ATCC CCL-2) were seeded (1.5×10⁶ cells/dish) in 10 cm plates and transfected with the CERKLa isoform tagged with the HA epitope. At 48 h post-transfection, cells were washed twice with 1X PBS, and then incubated 20 min at 4 °C in a solution that contained 10 mM Tris HCl, pH 7.4, 3% sucrose w/v, and 1% protease inhibitors (Roche). After this incubation, the concentration of sucrose was increased up to 0.25 M, and cells were broken by 15 strokes in an iced-chilled Dounce homogenizer. Cellular debris, nuclei, and unbroken cells were separated by centrifugation at 1,300x g for 5 min at 4 °C, and the supernatant was further centrifuged overnight at 100,000x g at 4 °C. The pellet containing cell organelles was resuspended in 1.5 M sucrose and laid at the bottom of a 1.8 ml swing-out (Beckman, Fullerton, CA) centrifuge tube. Subsequent layers of equal volumes of 1.15, 0.9, 0.6, and 0.25 M sucrose were added, and the final sucrose gradient was centrifuged for 2 h at 100,000x g at 4 °C. Next, 11 fractions of approximately 150 μl were collected from bottom to top. Samples were loaded onto 10% polyacrylamide SDS–PAGE gels, transferred to 0.45 µm polyvinylidene fluoride PDVP membranes (Immobilon-P, Millipore, Billerica, MA) and underwent immunodetection with the corresponding antibodies.

Immunodetection

After blocking membranes for 1 h in 10% nonfat milk in phosphate buffered saline, 0.1% Tween-20 (MTP), membranes were incubated overnight at 4 °C or 1 h at room temperature, separately, with any of the following primary antibodies, all diluted at 1:1000, anti-GM130 (BD Biosciences), anti-PDI (BD Biosciences), anti human TGN38 (Santa Cruz Biotechnology, Inc.) or anti-HA (Covance, Princetown, NJ). After extensive washes in MTP, membranes were incubated for 1 h with 1:3,000 of the corresponding peroxidase-conjugated secondary antibody (Sigma), and the bands were visualized using the ECL western blotting detection system (Amersham Biosciences).

Propidium iodide staining assay

COS-7 cells were grown on coverslips in 24 well plates. Transient transfections were performed on the following: the constructs encoding each of the four isoforms (pcDNA3-CERKL), the truncated variant (pcDNA3-R257X), empty pcDNA3.1 (negative control), or pcDNA6.2-DEST-CERK (provided by F. Bornancin, Novartis). At 44 h post-transfection, cells were treated with 200 µM H₂O₂ (Fluka, Sigma-Aldrich, St. Louis, MO) and 3 h later propidium iodide was added at a final concentration of 7 µg/ml. After 1 h, cells were rinsed twice with PBS and fixed in 3% paraformaldehyde and 2% sucrose in 0.1 M phosphate buffer at 4 °C for 30 min. They were then washed 2 times in 0.1 M PBS and mounted in Vectashield medium (Vector Laboratories). Several fields were analyzed under the visible and green light channels in a fluorescence microscope (Leica DM IL) coupled to a Leica DFC camera (Leica, Wetzlar, Germmany). A minimum of 500 cells were counted and compared to detect pignotic nuclei in the total cell count. Three independent replicates were performed for each construct.

Analysis of PARP cleavage-dependent apoptosis

HeLa cells were seeded in 24 well plates and transfected with either empty pcDNA3.1 (negative control) or pcDNA3-CERKLa (containing the lipid kinase domain). At 40 h post-transfection, the medium was replaced by fresh solution containing one of the following: 300 µM H₂O₂, 400 µM H₂O₂, or 0.3 M sodium nitroprusside (Fluka), or serum-deprived medium. After 24 h, cells were rinsed with PBS, harvested, and lysed directly in protein electrophoresis loading buffer. Samples were then boiled for 5 min and centrifuged to sediment cellular debris. Protein preparations were loaded onto 7.5% SDS–PAGE gels. Either 1:1,000 monoclonal anti-PARP (Invitrogen Life Technologies) or 1:8,000 monoclonal anti-tubulin (Sigma) were used as primary antibodies. Anti-mouse secondary antibody was conjugated to 1:3,000 horseradish peroxidase. Immunodetection was performed using chemiluminescent ECL reagents (GE Healthcare, Waukesha, WI), and the intensity of each band was quantified using the Quantity-One 1-D Analysis Software (Bio-Rad, Hercules, CA).

CERKL transcripts are alternatively spliced in the human retina

The initially characterized CERKL gene spanned 13 exons in chromosome 2q31.2–31.3 (Figure 1A). While attempting to clone the full-length CERKL on retina cDNA using primers spanning the ATG (forward) and the STOP (reverse) codons, we observed several faint bands, which were indicative of alternatively spliced variants. Therefore, we devised a two-step strategy of RT–PCR and cloning (see Methods) to characterize the full-length isoforms based on the structure of the gene. The RT–PCR analysis of human retina mRNA using primers from exons 2 and 13 allowed us to identify several spliced transcripts (Figure 1A) that would result in several protein isoforms (Figure 1 B): 1) CERKLa, which corresponded to the already reported variant [2], spanning 13 exons and encoding a 532 amino acid protein (GenBank accession number AY357073); 2) CERKLb, a 14 exon isoform, which includes an additional exon (4b) between exons 4 and 5 and encodes a 558 amino acid protein (accession number AY690329); 3) CERKLc, a splicing variant devoid of exons 3-4-4b-5, which encodes a 419 amino acid protein (accession number AY690330); and 4) CERKLd, a splicing variant that skips exons 4-4b-5 and encodes a 463 amino acid protein (accession number AY690331). Two very faint bands corresponded to not in-frame splicing variants (accession numbers AY690332 and AY690333) that would generate prematurely truncated proteins. These were considered aberrant transcripts, and given the mRNA quality control regulation, if they were produced in cells, they would be most probably cleared by the nonsense mediated decay mechanism.

The longest isoforms (CERKLa and CERKLb) contained the complete DAG domain and a putative N-terminal pleckstrin homology (PH) region. In contrast, the short forms (CERKLc and CERKLd) skipped most of the DAG kinase domain, although the putative PH domain (encoded in exons 1 and 2) was preserved (Figure 1B).

Noticeably, exons 4b and 5, which are alternatively spliced, show the rare GC donor splice site instead of the more common GT (marked by asterisks in Figure 1A). Both exons and the GC donor site sequences are conserved in the chimpanzee genome. The mouse Cerkl gene, which shows overall strong conservation in exon and intron structure, has no equivalent for exon 4b, thus suggesting that there are only 3 CERKL isoforms in the mouse. Again, the donor site of exon 5 in rodents is GC.

CERKL shows a complex alternative splicing pattern in human tissues

As we had previously found that CERKL was expressed in a variety of human adult and fetal tissues [2], we searched for CERKL isoforms in tissues other than retina. To this end, we used a panel of first strand cDNAs from several human tissues. In accordance with our previous data, no CERKL isoform was found in adult skeletal or heart muscle. However, some of the four in-frame splicing variants were detected in other adult and fetal tissues (Figure 2). cDNAs encoding protein isoforms CERKLa and CERKLb (with an intact DAG kinase domain) were present in adult liver and pancreas, as well as in fetal brain, lung, and kidney. One or both short isoforms (c and d) were detected in adult lung, adult kidney, adult pancreas, fetal lung, and fetal liver. Notably, retina was the only tissue in which all 4 isoforms were expressed under the conditions tested. However, basal levels of most isoforms were observed in most tissues after a saturating number of PCR cycles (data not shown).

CERKL does not phosphorylate ceramide under standard conditions

Given the structural similarities of CERKL to the reported ceramide kinases, a series of experiments were conducted to assess ceramide kinase activity. The protocol from [10] with minor modifications was used for in vitro ceramide kinase assays. We changed either of the following: 1) the source of the ceramides, namely a commercial mixture of brain ceramides (Sigma) or lipid micelles prepared from 661W cells (mouse photoreceptor precursor cell line, kindly provided by M.R. Al-Ubaidi) [11]; or 2) the source of the enzymatic activity to be tested, namely protein lysates from transiently transfected HEK293T and COS-7 cells overexpressing each CERKL isoform, or affinity-purified recombinant GST-CERKL variants (after expression in E. coli cells and only used in assays in vitro). No kinase activity was detected for any isoform at any combination of the lipid and protein lysate sources (Figure 3A). Additionally, lysates from co-transfections with constructs expressing 2 different isoforms, or the 4 of them together, were assayed for kinase activity. However, no significant difference with respect cells transfected with the empty vector or single CERKL isoforms could be detected (data not shown). We then attempted to detect CERKL kinase activity in vivo using photoreceptor-derived 661W cells. As 661W are neuronal, they transfect poorly; therefore, we used CERKL-GFP fusion constructs to enrich by FACS the population of transfected cells. Unfortunately, no differences were detected between cells transfected with CERKL and the empty vector (Figure 3B).

The absence of kinase activity prompted us to test the ability of CERKL isoforms to bind sphingolipids by protein-lipid overlay. As PH domains have been reported to have an affinity for phosphoinositides, these substrates were also included. We used the CERKLa isoform (532 aa, containing the complete DAG kinase domain) either as a pure preparation of recombinant GST-CERKL fusion protein (in E.coli) or as total fresh protein lysate from HEK293T cultured cells transiently transfected with the CERKLa HA-tagged construct. Again, no positive binding signals were produced (data not shown).

Subcellular CERKL localization is highly dynamic

Ceramide and other sphingoid lipids are highly dynamic membrane components. Proteins involved in sphingolipid metabolism appear to be specifically localized in subcellular compartments of the secretory pathway, from the ER-Golgi network to the plasma membrane. In this context, we aimed to determine CERKL subcellular localization by transient expression of each alternative splicing variant in cultured COS-7 cells, as the 4 isoforms display a specific domain architecture, which might account for slight differences in function or compartment localization. We used C-terminal HA-tagged CERKL isoforms (CERKL-HA), as well as the fusion CERKL-GFP constructs, in these transient transfections.

Interestingly, neither a unique localization pattern nor a restricted subcellular localization could be ascribed to any CERKL isoform, as all variants showed similar patterns that appeared to be cell-specific. Two main protein localizations emerged consistently in all preparations, as seen for the isoform CERKLa (532 amino acids; Figure 4A–D). Most cells showed CERKL in a strong perinuclear distribution, while the nucleus was practically devoid of signal (Figure 4C,D). Additionally, few cells in each field presented CERKL within the nucleus, where it was either localized to, or clearly excluded from, the nucleoli (Figure 4A,B,D). No exclusive colocalization with reference markers of subcellular compartments was observed, as CERKL colocalized both with the ER (calnexin) and Golgi matrix (GM130) markers (Figure 4E). To further assess association of CERKL with other membranous compartments, we assayed colocalization to endosomes (EEA1) or mitochondria (labeled by MitoTracker staining). However, no precise colocalization was observed (Figure 4E). Instead, the mutant RP truncated protein CERKLa R257X (the localization of CERKLb 257X was identical; data not shown) showed a more stable localization pattern in the ER and nuclei (Figure 4F), in agreement with other reports [12,13]. Similar CERKL subcellular localization results were obtained at 24 h or 48 h post-transfection, although a shift toward perinuclear accumulation was clearly observed. The same pattern was obtained when analyzing the localization of all 4 CERKL-GFP isoform constructs (data not shown). Overall, these results point to a highly dynamic localization of CERKL, which may relocate according to the stimuli acting upon a cell.

To further confirm that CERKL was localized at the membranous organelles, we performed subcellular organelle fractionation in a sucrose gradient. Cells transfected with the wild-type CERKLa isoform were homogenized, and nuclei and cellular debris were discarded. Membranes were pelleted by centrifugation, resuspended in 1.5 M sucrose, and further centrifuged in a 5-step sucrose gradient, ranging from 1.5 M to 0.25 M. Fractions were collected from bottom to top, run in an SDS–PAGE and immunodetected against HA (for CERKL detection) or distinctive organelle markers. The results (Figure 5) clearly showed that CERKL distributed to the same fractions where the Trans-Golgi (TGN38), Golgi (GM130), and ER (PDI) markers localized, although CERKL was more abundant in the Trans-Golgi and Golgi fractions.

CERKL protects cells from apoptosis in oxidative stress conditions

One of the working hypotheses to explain why mutations in CERKL cause RP is based on its cellular protective role through the regulation of the ceramide/ceramide-1-phosphate ratio [14]. Photoreceptors are under constant oxidative challenge. Environmental stress as well as genetic mutations (e.g., on CERKL) would make them more susceptible to damage, tilting the survival/death balance and triggering apoptosis. To test whether CERKL has a protective role, we assayed the effect of CERKL overexpression in transiently transfected cells under normal or oxidative stress conditions. As most neuronal cell lines, retina derived cells are very poorly transfected, so we had to resort to transfections on COS-7 and HeLa cell lines to obtain statistically significant results.

COS-7 cells were transfected with each of the CERKL isoforms as well as the mutant RP protein, and cell death was assayed by propidium iodide staining after treatment with 200 μM H₂O₂ for 4 h. The results were compared to those obtained with cells transfected with either empty vector or a construct expressing the known ceramide kinase, CERK (this construct was a gift from F. Bornancin). Dead cells were stained with propidium iodide and quantified (as a percentage) against total cell number (Figure 6A). Expression of the CERKL or CERK constructs caused a slight increase in mortality in basal conditions. Under oxidative stress, there was a 3.4-fold increase in mortality in the empty-vector transfected cells, whereas cells overexpressing any of the 4 CERKL isoforms maintained a similar mortality rate (no increase in mortality) with a high statistical significance (Mann–Whitney, p<0.05; Figure 6B). Remarkably, the CERKL R257X (the RP mutant) failed to protect cells; the mortality rate increased under oxidative stress (Figure 6B), although to a lesser level than the empty-vector cells (1.8 fold increase in mortality; statistical significance, p<0.05 in comparison to either CERKL isoforms or the empty vector). As propididum iodide staining was performed on coverslips, the transfection efficiency of each construct was assessed by western detection on protein lysates of replicate transfections.

Propidium iodide labels both apoptotic and necrotic cell nuclei. We attempted then to determine whether this protective effect of CERKL overexpression was on preventing cells from entering apoptosis. Activation of caspases is a well established event in apoptosis, and detection of caspase-cleaved substrates, such as PARP-1 (poly-ADP-ribose-polymerase)-1, is one of the most widespread hallmarks to quantify cell apoptosis [15,16]. PARP caspase-dependent cleavage was immunodetected in HeLa cells that were transiently transfected with either the empty vector or the CERKLa isoform (532 amino acids, encompassing the DAG kinase domain) in normal or under oxidative stress conditions, and using several reagents and concentrations: 300 μM and 400 μM H₂O₂, fetal bovine serum (FBS) deprivation and 0.3 mM sodium nitroprusside (SNP). The analysis was performed at different window times to determine the best range of conditions for quantification of any potential protective effect. CERKL did protect against apoptosis induced by oxidative stress (Figure 7A) in cells treated with 300 μM H₂O₂. This effect was dependent on the degree of oxidative insult, as it was much more effective at 300 μM (no apoptosis increase), than at 400 μM H₂O₂ (no significant protection). It was also dependent on the type of inducer, as cells overexpressing CERKLa did not behave differently from empty-vector transfected cells after FBS deprivation or SNP addition.

This protective effect was also dependent on time, and presented an effective window frame. It was already detectable at 12 h (Figure 7B) with an average of 20% protection (statistical significance p<0.05, Mann–Whitney test). In addition, this effect appeared to be much more prominent at 24 h (Figure 7C), where the protection appeared to be maximal, around 50% protection (p<0.05, Mann–Whitney test). However, the effect was no longer observed at 48 h (data not shown), where PARP cleavage was high and comparable between empty-vector and CERKLa transfected cells. The expression of CERKLa was comparable in all transfections (65%–68%), as assayed by western blot (data not shown), thus the observed differences were solely attributable to the overexpression of this isoform. Interestingly, this protective effect at 24 h-treatment was only found for the full-length CERKLa but not the truncated R257X mutant (neither in the a nor the b isoform constructs; Figure 7D), in accordance to the pathogenicity of this mutation.