Identification of genes involved in ceramide-dependent neuronal apoptosis using cDNA arrays
Genome Biology volume 3, Article number: research0042.1 (2002)
Ceramide is important in many cell responses, such as proliferation, differentiation, growth arrest and apoptosis. Elevated ceramide levels have been shown to induce apoptosis in primary neuronal cultures and neuronally differentiated PC 12 cells.
To investigate gene expression during ceramide-dependent apoptosis, we carried out a global study of gene expression in neuronally differentiated PC 12 cells treated with C2-ceramide using an array of 9,120 cDNA clones. Although the criteria adopted for differential hybridization were stringent, modulation of expression of 239 genes was identified during the effector phase of C2-ceramide-induced cell death. We have made an attempt at classifying these genes on the basis of their putative functions, first with respect to known effects of ceramide or ceramide-mediated transduction systems, and then with respect to regulation of cell growth and apoptosis.
Our cell-culture model has enabled us to establish a profile of gene expression during the effector phase of ceramide-mediated cell death. Of the 239 genes that met the criteria for differential hybridization, 10 correspond to genes previously involved in C2-ceramide or TNF-α signaling pathways and 20 in neuronal disorders, oncogenesis or more broadly in the regulation of proliferation. The remaining 209 genes, with or without known functions, constitute a pool of genes potentially implicated in the regulation of neuronal cell death.
Ceramide is an intracellular lipid second messenger generated in response to a large number of extracellular signals [1,2]. These include tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β), ionizing and ultraviolet radiation, anti-cancer drugs, growth-factor withdrawal, infection by human immunodeficiency virus (HIV) or bacteria. It is reported to participate in cell differentiation , senescence , growth arrest or programmed cell death [1,2], depending on the cell type.
The role of ceramide in programmed cell death or apoptosis has been described in lymphocytes , macrophages , neurons in primary culture [7,8] and neuronally differentiated PC12 cells [9,10,11]. A number of downstream targets of ceramide have been identified. The best documented are the ceramide-activated protein phosphatases (CAPP) and the ceramide-activated protein kinase (CAPK). The former, represented by the PP1 and PP2A families, mediate the effect of ceramide on the transcription factors c-Myc  and c-Jun . CAPK is involved in the mitogen-activated protein (MAP) kinase (MAPK) cascades that include the extracellular-signal regulated kinases (ERK), the c-Jun N-terminal kinases or stress-activated kinases (JNK/ SNK/SAPK) and the p38 family .
Recently, it has been shown that C2-ceramide rapidly decreases phosphorylation of ERKs, but increases p38 and JNK phosphorylation, activating the transcription factors c-Fos, c-Jun and p53, during the effector phase of apoptosis in primary cortical neurons . It also regulates the protein kinase B (Akt/PKB)-dependent survival pathways, inactivating Akt by dephosphorylation and activating the Bcl-2-related protein BAD by phosphorylation [16,17,18]. Ceramide-induced apoptosis in neurons or in neuronally differentiated PC12 cells has been associated with mitochondrially produced reactive oxygen species (ROS) as well as activation and nuclear translocation of the transcription factor NFκB [10,11,19]. All these molecular events are observed during the effector phase of ceramide-induced apoptosis which also includes gene expression and new protein synthesis required for ceramide-mediated cell death, as it has been shown that neuronal cell death can be inhibited by cycloheximide .
The genes that are transcriptionally regulated during ceramide-mediated cell death are still poorly documented. To study gene expression during neuronal cell death, we carried out a differential screen of an array of 9,120 cDNA clones from a human infant brain library (library 1NIB ) with complex cDNA targets derived from neuronally differentiated rat pheocytochroma PC12 cells treated with C2-ceramide compared to control PC12 cells. This model is particularly suitable for establishing a gene-expression profile during ceramide-mediated neuronal death because first, the neuronal cell population is synchronized and homogeneous, unlike brain tissue or primary neuronal cultures, and second, because the use of exogenous C2-ceramide eliminates the risk of interference by transcripts activated by signal transducers upstream of ceramide in the cell-death pathway or in pathways activated in parallel.
Cell death induced in neuronally differentiated PC12 cells by C2-ceramide
The morphological characteristics of differentiated PC12 cells after 24 hours in the presence of 25 μM C2-ceramide were compatible with cell death by apoptosis. Compared with control cultures, as viewed by phase-contrast microscopy (Figure 1a), C2-ceramide-treated cells lost their neurites and became rounded and shrunken after 24 hours of treatment (Figure 1b). The cells that remained viable in the C2-ceramide-treated cultures were refringent (Figure 1b), like those in the control cultures (Figure 1a), and excluded the vital marker propidium iodide (Figure 1c), whereas the dead cells took up propidium iodide that intercalated into their DNA (Figure 1d), revealing condensed and fragmented nuclei. As previously described, when neuronally differentiated PC12 cells or primary cultures of mesencephalic neurons were treated with cell-permeant C2-ceramide (10-50 μM), they died in a dose-dependent manner [7,10]. At 25 μM no significant cell death was observed until 12 hours after the initiation of treatment (Figure 2a). After 24 hours, 50% of the cells had died. By 48 hours, no viable cells remained. Furthermore, we observed activation of caspase-3/CPP32, a member of the cysteine-activated aspartate family of cell-death proteases , that started 8 hours after the beginning of ceramide treatment and was five times the control value by 18 hours (Figure 2b). No significant cell death and caspase-3/CPP32 activity were observed using the inactive C2 analog of ceramide, C2-dihydroceramide (Figure 2).
Validation of hybridization signals
Hybridization of 9,120 cDNA clones with complex cDNA targets from poly(A)+ RNA extracted from C2-ceramide-treated or control cells produced signals of varying intensities (Figure 3a). In order to eliminate clones for which no reproducible hybridization signals were obtained, the signal-intensity values were validated as described in Materials and methods. Thus, 7% of the clones hybridized with the control cDNA target (634) and 14% of clones hybridized with the C2-ceramide-treated cDNA target (1,297) were excluded from further analysis. The remaining 6,494 clones were analyzed for differential hybridization.
Differential gene expression in neuronally differentiated PC12 cells treated with C2-ceramide compared to controls
Changes in gene expression were analyzed during the effector phase of neuronal death, 7 hours after the beginning of C2-ceramide treatment. This time point was chosen because on the one hand it is preceded by the activation of the transcription factor NFκB and c-Jun observed 4 to 6 hours after C2-ceramide treatment in PC12 cells [10,22], and on the other, the apoptotic process is still not induced by caspase-3 activation, which occurs 8 hours after the beginning of C2-ceramide treatment.
Hybridization between the rat PC12 cell-derived targets and the human cDNA macroarray was carried out as described in Materials and methods. Modulation of gene expression was quantitated by calculating the ratio of the intensity of the normalized hybridization signal obtained with the C2-ceramide cDNA target to that obtained with the control target. Clones were considered to be differentially hybridized in C2-ceramide-treated cells compared to control cells if the ratio between the corresponding hybridization intensity values was ≥ 2 (up-hybridized clones) or ≤ 0.5 (down-hybridized clones) which are the limits of confidence for the method. To decrease the risk of false-positive results, clones with hybridization signals that were less than twofold above background were also excluded, resulting in the elimination of 538 clones. In addition, the remaining clones were hybridized with complex cDNA targets from poly(A)+ RNA extracted from C2-dihydroceramide-treated cells used as negative control and compared to untreated cells. No modulation of expression was observed (except for one clone excluded from the analysis) in the presence of this inactive analog of C2-ceramide (data not shown). Among the 239 clones that met the criteria for differential hybridization, 132 were up-hybridized in C2-ceramide-treated cells and 107 were down-hybridized. The distribution of the hybridization-intensity values between the control and the C2-ceramide complex cDNA targets is presented in Figure 3b. Approximately 55% (72/132) of the up-hybridized clones were hybridized 3-6-fold more in C2-ceramide-treated cells than in the control and 40% (41/107) of the down-hybridized clones were hybridized 3-9-fold less.
Partial 5' and 3' sequences of the 239 clones were compared with all the sequences in the database developed in our laboratory (the Genexpress Index ) and in public databases. Of the 239 clones, 179 clones corresponded to already identified human genes, 113 of which have defined functions. The remaining 60 clones corresponded to genes with limited characterization. Under the hypothesis that differential hybridization of the clones reflects linear modulation of expression of the corresponding genes, we assume that we have detected differential gene expression using cDNA array technology that can be interpreted according to the information available.
Ten differentially expressed genes encode proteins with a role in ceramide or TNF-α signaling pathways (Figure 4, Table 1; see  for links to database entries for each gene). Two of these genes, PLA2G4C  and CLN3 [26,27] seem to have a role in ceramide-mediated cell death or survival. Two upregulated genes (ETV5 , NPTX2 [29,30]) and two downregulated genes (COL1SA1 [31,32], TNFAIP1 ) encode proteins that are modulated by TNF-α. Four genes, three upregulated (AXL , BIRC1 , RSU1 ) and one downregulated (MAPK10 ) encode proteins with a role in the TNF-α signaling pathway.
Twenty clones correspond to genes encoding proteins that have been involved in the regulation of apoptosis and/or cell growth (Figure 5, Table 2, see ). Fourteen are up-hybridized and six are down-hybridized by C2-ceramide. Ten of the upregulated and two of the downregulated genes encode proteins stimulating apoptosis and/or growth arrest. The other genes (four upregulated and four downregulated) encode proteins downregulating apoptosis and/or stimulating growth.
The remaining 83 clones corresponding to 82 genes with known or putative functions have no obvious relation to the apoptosis process (Table 3, see ). Of the total number of differentially hybridized clones, 66 correspond to mRNA sequences (Table 4, see ) and 60 to poorly characterized genes (Table 5, see ) that encode proteins without known function.
To confirm the results obtained by macroarray analysis, differentially expressed transcripts representing upregulated or downregulated genes were analyzed for differential expression by reverse transcription PCR (RT-PCR) or northern blots. As shown in Figure 6, the upregulation of ETV5, M6PR and APCL was confirmed by RT-PCR, and the downregulation of two genes with unknown function (mRNA DKFZp586C1723 and GENX 2969) was confirmed by northern blotting.
Extracellular signaling molecules such as cytokines, growth-factor deprivation and DNA damage caused by chemotherapeutic agents or irradiation activate ceramide-mediated signal transduction pathways leading to cell death. These pathways have been investigated in the immune system, where they are known to have an important role, and in neurons, as they are suspected to play a part in neurodegenerative disorders . A number of steps in the signaling cascades have been elucidated. However, although the translation inhibitor cycloheximide inhibits the ceramide-mediated death of mesencephalic neurons , the expression patterns of genes modulated during ceramide-mediated cell death remain unknown. In a global approach to this question, we have used cDNA macroarray technology to determine the profile of gene expression in a neuronal model of cell death, neuronally differentiated and C2-ceramide-treated PC12 cells, in which ceramide-dependent changes in gene expression could be isolated from the effects of other transcription modulators.
Identification of genes closely implicated in the ceramide and/or TNF-α signaling pathway
We were able to detect differential expression of 10 genes known to be involved in the ceramide or TNF-α signaling pathways (see Figure 4, Table 1) thus validating our study. A summary illustration of the putative role of these genes is presented in Figure 7. Briefly, two genes, encoding phospholipase A2 group IVC (PLA2G4C) and ceroid-lipofuscinosis, neuronal 3, juvenile (CLN3) are already known to be involved in ceramide-mediated signal transduction. The first, PLA2G4C, belongs to the cytosolic phospholipase A2 gene family that encodes two different proteins: calcium-independent and calcium-dependent cytosolic phospholipases . TNF-α regulates the expression of PLA2G4A mRNA in HeLa cells  and in human bronchial epithelial cells , which is indirect evidence of modulation by ceramide, but the role of ceramide was not demonstrated directly in these studies. However, ceramide was shown directly to upregulate the expression of the gene encoding cytosolic phospholipase A2 in the fibroblast cell line L929 . Conversely, the activation of this gene was reported to be necessary for ceramide accumulation and cell death in the same cells . We show for the first time that this gene is involved in neuronal apoptosis.
The second gene, CLN3, is expressed in a variety of human tissues including the brain, where the product is necessary for neuronal survival [26,27]. Interestingly, CLN3 does not inhibit C2-ceramide-induced apoptosis but modulates endogenous ceramide synthesis and suppresses apoptosis by preventing generation of ceramide . Thus, C2-ceramide can activate a negative feedback mechanism regulating endogenous ceramide generation as well as activate the downstream targets of the endogenous lipid.
Four other genes or families of genes known to be transcriptionally regulated by TNF-α were also modulated by C2-ceramide in our model (Table 1). Of these, ETS variant 5 (ETV5) belongs to the family of ETS transcription factor genes. Increased expression of both ETS1 mRNA and the protein has been observed in human fibroblasts after TNF-α or IL-1β stimulation . PEA3 (a mouse protein corresponding to ETV5) inhibits tumorigenesis in vivo . Moreover, ETV5 and ETS1 can cooperate with c-Jun/c-Fos [44,45], potential regulators of apoptosis in many cell types and specially in the mammalian nervous system . The second gene regulated by TNF-α is NPTX2, encoding neuronal pentraxin II. Pentraxins are a family of proteins that include C-reactive protein and serum amyloid P. They have been found in the brain plaques characteristic of Alzheimer's disease and are toxic to neuronal cell cultures [47,48]. Furthermore, the expression of NPTX3 is increased in response to TNF-α or IL-1β stimulation via activation of NFκB [29,30]. The regulation of the pentraxin gene family by C2-ceramide treatment is consistent with our previous studies showing NFκB activation by C2-ceramide in PC12 cells and in primary cultures of neurons [10,19]. The last two genes known to be regulated by TNF-α and identified in our model are COL18A1, encoding type XVIII collagen alpha 1, and TNFAIP1, encoding TNF-α-induced protein 1. These proteins, downregulated by C2-ceramide, are modulated by TNF-α in various cell types [31,32,33].
We also identified four genes encoding proteins known to participate in TNF-α-activated signal transduction pathways. Thus AXL, upregulated by a factor of 3.65 (Table 1), encodes a tyrosine kinase receptor. Signaling through this receptor is reported to protect against TNF-α-induced apoptosis in fibroblasts and its absence increases apoptosis after serum deprivation . Interestingly, ARK, the mouse protein corresponding to AXL, activates the survival pathway mediated by the serine-threonine kinase Akt , which is negatively regulated by ceramide [16,17,50], and is also reported to modulate ceramide synthesis . The second gene we identified is BIRC1, encoding baculoviral IAP repeat-containing 1 protein. This protein, putatively involved in spinal muscular atrophy , is an inhibitor of cell death induced by various apoptotic stimuli, including TNF-α . The third identified gene, RSU1, encodes Ras suppressor protein 1, which is involved in TNF-α signaling by blocking the Ras-dependent response. Levels of both RSU1 mRNA and protein have been correlated with a decrease in growth rate and tumorigenic potential in U251 glioblastoma cells  and it induces growth arrest in PC12 cells . This is consistent with the report that ceramide regulates apoptosis via modulation of the Ras signaling pathway . In addition, RSU1 has been identified as an inhibitor of Jun kinase activation . This point is interesting, as the fourth gene presenting in this group, MAPK10/J.NK3, encoding the JNK family member mitogen-activated protein kinase 10, is downregulated by C2-ceramide in our model.
The identification of these eight genes, which are involved in the TNF-α signaling pathway, in C2-ceramide treated PC12 cells, suggests that their modulation of expression by TNF-α could be the result of a ceramide-dependent mechanism.
Commitment to apoptosis: upregulation of pro-apoptotic genes and downregulation of anti-apoptotic genes by the ceramide pathway
Twenty genes regulated by C2-ceramide correspond to genes known to be involved in regulation of apoptosis and/or cell growth (Figure 5, Table 2). Twelve of these genes are known to be associated with oncogenesis and four with neuronal disorders. Of the upregulated genes, 10 out of 14 are known to be associated with a pro-apoptotic or anti-proliferation process and 3 out of 14 are mainly implicated in protection of the cell against cytotoxicity or damage. Of the downregulated genes, 4 out of 6 are associated with an anti-apoptotic or a proliferation process. This highlights the fact that the cells are engaged in programmed cell death. The putative roles of these genes are illustrated in Figure 7, which focuses on the pro-apoptotic or anti-proliferation process versus anti-apoptotic or proliferation processes.
Briefly, of the known pro-apoptotic or anti-proliferative genes that are upregulated in our model, RPL4 encodes the ribosomal protein L4 that has been shown to be transcriptionally stimulated prior to apoptosis induced by the 5-azacytidine in the PC12 cells . PDCD6IP, upregulated by C2-ceramide in our model, encodes a protein that interacts with ALG2, a Ca2+-binding protein that is required for apoptosis induced by diverse stimuli, including ceramide treatment [55,56,57]. M6PR encodes the cation-dependent mannose-6-phosphate receptor, which has been implicated in retinoid-induced apoptosis . NMA, encoding a putative transmembrane protein, is expressed at low levels in metastatic human melanoma cell lines and xenografts, and is completely absent in highly metastatic human melanoma cell lines . APCL, encoding adenomatous polyposis coli like protein, is a tumor-suppressor gene . METAP2 encodes methionine aminopeptidase eIF-2-associated p67, which interacts with eukaryotic translation initiation factor eIF-2  and could regulate p53 signaling . BAT3, downregulated in some transformed cells, encodes HLA-B associated transcript-3, which interacts with the tumor-suppressor protein DAN that contains growth or tumor suppressive activity in vitro . PHB encodes the protein prohibitin, a potential tumor-suppressor protein that binds to the retinoblastoma (Rb) protein and represses E2F transcriptional activity [64,65]. BNIP3L, encoding BCL2/adenovirus E1B 19 kD-interacting protein 3-like, is a pro-apoptotic gene which has a growth-inhibitory effect on cancer cells . CDH13, encoding cadherin 13, is significantly downregulated in human breast carcinoma cell lines and breast cancer, whereas its overexpression decreases tumor-cell growth [67,68].
Of the known anti-apoptotic or proliferative genes that are downregulated in our model, LOC51582 encodes an antizyme inhibitor, which regulates the antizyme activity proposed to be involved in the polyamine biosynthesis pathway [69,70]. Interestingly, overexpression of antizyme inhibits cell growth [71,72], whereas LOC51582 is downregulated in our model. This observation is consistent with a role of antizyme in the apoptotic process and suggests that ceramide can regulate its activity. LOC51283, a regulator of the activity of the Bcl-2 family proteins, encodes a novel apoptosis regulator, which has been identified as an inhibitor of Bax-induced cell death . Its downregulation by C2-ceramide confirms its involvement in the ceramide-dependent regulation of cell death. The last gene presented here, MET, encodes the MET proto-oncogene, known to be a receptor of the hepatocyte growth factor that has been described to protect neuronal cells from apoptosis via the phosphatidylinositol-3 kinase/Akt pathway .
Four genes out of the other genes presented in Table 2 have already been implicated in neuronal disorders, suggesting that ceramide may be a key second messenger in these pathologies. The upregulation of the glutamate receptor gene (GRIA2) seems to be an indicator of tolerance to ischemia . The absence of somatostatin, encoded by SST (downregulated in our model), is associated with apoptotic neurons in patients with Alzheimer's disease . SMN, encoding Survival of motor neuron 2, downregulated by C2-ceramide, strongly contributes to the severity of the spinal muscular atrophy . MUT mRNA is upregulated in ischemia, in relation to a decrease in the accumulation of its neurotoxic metabolite .
In conclusion, our cell culture model has enabled us to establish a profile of gene expression during the effector phase of ceramide-mediated cell death. In spite of the stringency of the criteria adopted for differential hybridization, a large number of cDNA clones, 239 of the 9,120 in our cDNA array derived from a normalized infant brain library, correspond to genes up- or downregulated by C2-ceramide treatment. Already-known genes account for 179 of the transcripts, 113 of which have a putative function.
On the basis of their putative functions, we have made an attempt at classifying these transcripts, first with respect to known effects of ceramide or ceramide-mediated transduction systems, then with respect to regulation of cell growth and apoptosis. The 30 genes in Tables 1 and 2 met these criteria, validating the approach and suggesting that the other modulated genes may also be relevant with regard to the progression of the cell-death mechanisms. These genes were classified as having no obvious relation to cell death or survival (Table 3), no known function (Table 4) or as poorly characterized (Table 5). As a result of our study, these genes now have tentative functions. The full list can be consulted with the relevant data on the dedicated website .
Interestingly, given the large number of genes known to be modulated by NFκB in the immune system , it was surprising that only pentraxin was detected in our model. This suggests either that NFκB is less important in neurons than in lymphocytes, or that its targets are different. Conversely, the transcriptional regulators responsible for the differential expression of the genes detected in our study remain to be discovered. In any case, our results show that transcriptional regulation plays an important role in ceramide-mediated cell death and that some of the modulated transcripts, in agreement with published studies, are involved in other cell-death mechanisms as well.
Materials and methods
Rat PC12 cells , which acquire a neuronal phenotype in the presence of nerve growth factor (NGF), were plated at a density of 2,000-3,000 cells/cm2 in 75 cm2 culture flasks coated with polyethylenimine (1 mg/ml) in Leibovitz modified L15 medium (Gibco BRL) supplemented with 2% horse serum and 150 ng/ml NGF (grade II; Alomone Labs, Jerusalem, Israel) as previously described . Apoptosis was induced, after 6 days in the presence of NGF, with the cell-permeant C2 analog of ceramide (C2-ceramide), N-acetylsphingosine (Biomol Research Laboratories, Plymouth Meeting, PA), at a concentration of 25 μM. As negative control, an inactive C2 analog of ceramide (C2-dihydroceramide), N-acetylsphinganine (Biomol Research Laboratories), was used in the same condition as C2-ceramide.
Morphological characterization of apoptosis and cell counts
Neurite retraction and cell shrinkage were visualized by phase-contrast microscopy. Condensed and fragmented nuclei were made visible in situ as described in , by intercalation into nuclear DNA of the fluorescent probe propidium iodide. Propidium iodide, which only enters dead cells that have become permeable, was visualized by epifluorescence with a rhodamine filter (excitation, 548-580 nm; emission, 580-610 nm). Viability was quantified by counting cells in at least 10 randomly chosen fields with a 20x objective. The percentage of cells excluding the vital dye propidium iodide was calculated at each time point after the beginning of C2-ceramide or C2-dihydroceramide treatment with respect to the corresponding control.
Measurement of caspase-3-like activity
Caspase-3-like activity was measured using the CaspACE Assay system (Promega, Madison, WI). Cell extracts containing equivalent amounts of protein were used to measure DEVDase (caspase-3-like) activity: the chromophore p-nitroaniline (pNA), released from the colorimetric substrate (Ac-DEVD-pNA) upon cleavage by DEVDase produces a yellow color that is monitored by a photometer at 405 nm.
Preparation of the cDNA macroarray
cDNA clones from a normalized infant brain library (library 1NIB; ) were randomly selected to provide a set of 9,120 cDNA clones. The 3' and/or 5' ends of these clones had been previously sequenced . The sequences, registered in GenBank , were compared to those in public data bases, permitting tentative identification of the corresponding gene transcripts. The cDNA clones were used to prepare PCR products using oligonucleotide primers complementary to sequences in the vector. They were spotted by robot (Flexis; Perkin Elmer, Shelton, CT) at medium density (25 PCR products/cm2) on nylon membranes (Hybond-N+; Amersham Biosciences, Uppsala, Sweden) as previously described . The entire collection of 9,120 cDNA clones was spotted on a set of four filters.
Purification of poly(A)+mRNA
Total RNA was extracted from control PC12 cultures and from PC12 cultures treated with C2-ceramide or C2-dihydroceramide (approximately 106 cells) with the RNeasy midi kit (Qiagen, Courtaboeuf, France), according to the manufacturer's instructions. The integrity of the RNA was confirmed by agarose gel electrophoresis. Poly(A)+ mRNA was extracted from total RNA with oligo(dT)-conjugated magnetic beads (Dynabeads; Dynal, Oslo, Norway), as described in the manufacturer's protocols.
Complex cDNA target synthesis
Complex cDNA targets were synthesised by reverse transcription of 500 ng poly(A)+ mRNA extracted from control, C2-dihydroceramide- or C2-ceramide-treated PC12 cells. The reaction was performed with the Superscript™ Preamplification System (Invitrogen) as previously described . The reaction mixture contained random-oligonucleotide primers (500 ng), 50 μCi [α-33P]dATP, 3,000 Ci/mmol (Amersham), 500 μM d(T, C, G)TP (Amersham) and 50 μM dideoxyGTP (Invitrogen).
The filters were prehybridized at 68°C for 30 min in ExpressHyb hybridization solution (Clontech, Palo Alto, CA), hybridized for 2 h in the same solution to which the radiolabeled complex cDNA target was added, then washed twice for 30 min at 25°C in standard saline citrate (SSC) 1×/0.1% sodium dodecyl sulfate (SDS) and twice for 30 min at 25°C in SSC 0.1×/0.1% SDS. The washed filters were exposed to phosphorus screens (Molecular Dynamics, Sunnyvlae, CA) for 16 h.
Hybridization signal quantitation
Image acquisition was carried out with the Phosphorlmager (Molecular Dynamics). The hybridization signal corresponding to each cDNA clone was quantitated with a specifically designed software (XdotsReader; Cose, Dugny, France) and the local background signal was subtracted. The intensity of the hybridization signal for each clone was then divided by the average intensity of all the clones on each filter to obtain normalized values. Hybridization was done in quadruplicate so that, for each clone/target combination, four values were obtained, compared and validated if at least three out of the four values were similar (SD ± 25%). The final value assigned to each clone was the average of the validated values.
Total RNA (20 μg) were fractionated under denaturing conditions in a 1.2% agarose gel and transferred onto a Hybond-N+ membrane (Amersham). Specific probes were generated from cDNA clones of interest by PCR using vector-specific primers. The PCR products were purified using the microcon kit (Amicon, Wageningen, The Netherlands) and radiolabeled by random priming (Gibco BRL). Oligonucleotides corresponding to 18S rRNA (control probe) were 32P-labeled using [γ-33P]ATP and T4 RNA kinase. For northern blot analysis, the blots were prehybridized 2 h in ULTRAhyb hybridization buffer (Ambion, Austin, TX), hybridized with the labelled probe (1-2 × 106 cpm/ml) for 16 h at 42°C in the same solution, and washed as for the high-density filters. The washed filters were exposed to phosphorus screens (Molecular Dynamics) for 48 h. The hybridization signal of the specific probes was analyzed with the ImageQuant software (Molecular Dynamics) and compared to the signal obtained with the control probe.
Total RNA of PC12 cells cultured with or without C2-ceramide was purified according to the protocol described above. Total RNA (2 μg) were reverse transcribed using the Superscript™ Preamplification System (Invitrogen) according to the manufacturer's protocol. An aliquot of the reaction was then used for PCR amplification with the Advantage PCR kit (Clontech) and primers specific to the gene of interest. The amplification products were visualized after electrophoresis in a 1.5% agarose gel with ethidium bromide. The signals were analyzed with ImageQuant software and compared to HPRT (hypoxanthine phosphoribosyl transferase) as control gene.
Hannun YA: Functions of ceramide in coordinating cellular responses to stress. Science. 1996, 274: 1855-1859. 10.1126/science.274.5294.1855.
Mathias S, Pena LA, Kolesnick RN: Signal transduction of stress via ceramide. Biochem J. 1998, 335: 465-480.
Furuya S, Mitoma J, Makino A, Hirabayashi Y: Ceramide and its interconvertible metabolite sphingosine function as indispensable lipid factors involved in survival and dendritic differentiation of cerebellar Purkinje cells. J Neurochem. 1998, 71: 366-377.
Obeid LM, Venable ME: Signal transduction in cellular senescence. J Am Geriatr Soc. 1997, 45: 361-366.
Kolesnick R, Golde DW: The sphingomyelin pathway in tumor necrosis factor and interleukin-1 signaling. Cell. 1994, 77: 325-328.
Kyprianou N, Alexander RB, Isaacs JT: Activation of programmed cell death by recombinant human tumor necrosis factor plus topoisomerase II-targeted drugs in L929 tumor cells. J Natl Cancer Inst. 1991, 83: 346-350.
Brugg B, Michel PP, Agid Y, Ruberg M: Ceramide induces apoptosis in cultured mesencephalic neurons. J Neurochem. 1996, 66: 733-739.
Mitoma J, Ito M, Furuya S, Hirabayashi Y: Bipotential roles of ceramide in the growth of hippocampal neurons: promotion of cell survival and dendritic outgrowth in dose- and developmental stage-dependent manners. J Neurosci Res. 1998, 51: 712-722. 10.1002/(SICI)1097-4547(19980315)51:6<712::AID-JNR5>3.3.CO;2-Q.
Hartfield PJ, Mayne GC, Murray AW: Ceramide induces apoptosis in PC12 cells. FEBS Lett. 1997, 401: 148-152. 10.1016/S0014-5793(96)01460-3.
France-Lanord V, Brugg B, Michel PP, Agid Y, Ruberg M: Mitochondrial free radical signal in ceramide-dependent apoptosis: a putative mechanism for neuronal death in Parkinson's disease. J Neurochem. 1997, 69: 1612-1621.
Lambeng N, Michel PP, Brugg B, Agid Y, Ruberg M: Mechanisms of apoptosis in PC12 cells irreversibly differentiated with nerve growth factor and cyclic AMP. Brain Res. 1999, 821: 60-68. 10.1016/S0006-8993(99)01061-6.
Wolff RA, Dobrowsky RT, Bielawska A, Obeid LM, Hannun YA: Role of ceramide-activated protein phosphatase in ceramide-mediated signal transduction. J Biol Chem. 1994, 269: 19605-19609.
Reyes JG, Robayna IG, Delgado PS, Gonzalez IH, Aguiar JQ, Rosas FE, Fanjul LF, Galarreta CM: c-Jun is a downstream target for ceramide-activated protein phosphatase in A431 cells. J Biol Chem. 1996, 271: 21375-21380. 10.1074/jbc.271.15.8719.
Su B, Karin M: Mitogen-activated protein kinase cascades and regulation of gene expression. Curr Opin Immunol. 1996, 8: 402-411. 10.1016/S0952-7915(96)80131-2.
Willaime S, Vanhoutte P, Caboche J, Lemaigre-Dubreuil Y, Mariani J, Brugg B: Ceramide-induced apoptosis in cortical neurons is mediated by an increase in p38 phosphorylation and not by the decrease in ERK phosphorylation. Eur J Neurosci. 2001, 13: 2037-2046. 10.1046/j.0953-816x.2001.01581.x.
Schubert KM, Scheid MP, Duronio V: Ceramide inhibits protein kinase B/Akt by promoting dephosphorylation of serine 473. J Biol Chem. 2000, 275: 13330-13335. 10.1074/jbc.275.18.13330.
Salinas M, Lopez-Valdaliso R, Martin D, Alvarez A, Cuadrado A: Inhibition of PKB/Akt1 by C2-ceramide involves activation of ceramide-activated protein phosphatase in PC12 cells. Mol Cell Neurosci. 2000, 15: 156-169. 10.1006/mcne.1999.0813.
Basu S, Bayoumy S, Zhang Y, Lozano J, Kolesnick R: BAD enables ceramide to signal apoptosis via Ras and Raf-1. J Biol Chem. 1998, 273: 30419-30426. 10.1074/jbc.273.46.30419.
Hunot S, Brugg B, Ricard D, Michel PP, Muriel MP, Ruberg M, Faucheux BA, Agid Y, Hirsch EC: Nuclear translocation of NF-kappaB is increased in dopaminergic neurons of patients with Parkinson disease. Proc Natl Acad Sci USA. 1997, 94: 7531-7536. 10.1073/pnas.94.14.7531.
Soares MB, Bonaldo MF, Jelene P, Su L, Lawton L, Efstratiadis A: Construction and characterization of a normalized cDNA library. Proc Natl Acad Sci USA. 1994, 91: 9228-9232.
Nicholson DW: Caspase structure, proteolytic substrates, and function during apoptotic cell death. Cell Death Differ. 1999, 6: 1028-1042. 10.1038/sj/cdd/4400598.
Hartfield PJ, Bilney AJ, Murray AW: Neurotrophic factors prevent ceramide-induced apoptosis downstream of c-Jun N-terminal kinase activation in PC12 cells. J Neurochem. 1998, 71: 161-169.
Houlgatte R, Mariage-Samson R, Duprat S, Tessier A, Bentolila S, Lamy B, Auffray C: The Genexpress Index: a resource for gene discovery and the genic map of the human genome. Genome Res. 1995, 5: 272-304.
Identification of genes involved in ceramide-dependent neuronal apoptosis using cDNA array. [http://idefix.ers1984.vjf.cnrs.fr/APOPTOSE/]
Jayadev S, Hayter HL, Andrieu N, Gamard CJ, Liu B, Balu R, Hayakawa M, Ito F, Hannun YA: Phospholipase A2 is necessary for tumor necrosis factor alpha-induced ceramide generation in L929 cells. J Biol Chem. 1997, 272: 17196-17203. 10.1074/jbc.272.27.17196.
Janes RW, Munroe PB, Mitchison HM, Gardiner RM, Mole SE, Wallace BA: A model for Batten disease protein CLN3: functional implications from homology and mutations. FEBS Lett. 1996, 399: 75-77. 10.1016/S0014-5793(96)01290-2.
The International Batten Disease Consortium: Isolation of a novel gene underlying Batten disease, CLN3. Cell. 1995, 82: 949-957.
Gilles F, Raes MB, Stehelin D, Vandenbunder B, Fafeur V: The c-ets-1 proto-oncogene is a new early-response gene differentially regulated by cytokines and growth factors in human fibroblasts. Exp Cell Res. 1996, 222: 370-378. 10.1006/excr.1996.0046.
Basile A, Sica A, d'Aniello E, Breviario F, Garrido G, Castellano M, Mantovani A, Introna M: Characterization of the promoter for the human long pentraxin PTX3. Role of NF-kappaB in tumor necrosis factor-alpha and interleukin-1beta regulation. J Biol Chem. 1997, 272: 8172-8178. 10.1074/jbc.272.13.8172.
Mantovani A, Muzio M, Ghezzi P, Colotta C, Introna M: Regulation of inhibitory pathways of the interleukin-1 system. Ann NY Acad Sci. 1998, 840: 338-351.
Solis-Herruzo JA, Brenner DA, Chojkier M: Tumor necrosis factor alpha inhibits collagen gene transcription and collagen synthesis in cultured human fibroblasts. J Biol Chem. 1988, 263: 5841-5845.
Hernandez-Munoz I, de la Torre P, Sanchez-Alcazar JA, Garcia I, Santiago E, Munoz-Yague MT, Solis-Herruzo JA: Tumor necrosis factor alpha inhibits collagen alpha 1(I) gene expression in rat hepatic stellate cells through a G protein. Gastroenterology. 1997, 113: 625-640.
Wolf FW, Marks RM, Sarma V, Byers MG, Katz RW, Shows TB, Dixit VM: Characterization of a novel tumor necrosis factor-alpha-induced endothelial primary response gene. J Biol Chem. 1992, 267: 1317-1326.
Bellosta P, Zhang Q, Goff SP, Basilico C: Signaling through the ARK tyrosine kinase receptor protects from apoptosis in the absence of growth stimulation. Oncogene. 1997, 15: 2387-2397. 10.1038/sj/onc/1201419.
Liston P, Roy N, Tamai K, Lefebvre C, Baird S, Cherton-Horvat G, Farahani R, McLean M, Ikeda JE, MacKenzie A, Korneluk RG: Suppression of apoptosis in mammalian cells by NAIP and a related family of IAP genes. Nature. 1996, 379: 349-353. 10.1038/379349a0.
Masuelli L, Ettenberg S, Vasaturo F, Vestergaard-Sykes K, Cutler ML: The ras suppressor, RSU-1, enhances nerve growth factor-induced differentiation of PC12 cells and induces p21CIP expression. Cell Growth Differ. 1999, 10: 555-564.
Masuelli L, Cutler ML: Increased expression of the Ras suppressor Rsu-1 enhances Erk-2 activation and inhibits Jun kinase activation. Mol Cell Biol. 1996, 16: 5466-5476.
Underwood KW, Song C, Kriz RW, Chang XJ, Knopf JL, Lin LL: A novel calcium-independent phospholipase A2, cPLA2-gamma, that is prenylated and contains homology to cPLA2. J Biol Chem. 1998, 273: 21926-21932. 10.1074/jbc.273.34.21926.
Hoeck WG, Ramesha CS, Chang DJ, Fan N, Heller RA: Cytoplasmic phospholipase A2 activity and gene expression are stimulated by tumor necrosis factor: dexamethasone blocks the induced synthesis. Proc Natl Acad Sci USA. 1993, 90: 4475-4479.
Wu T, Ikezono T, Angus CW, Shelhamer JH: Tumor necrosis factor-alpha induces the 85-kDa cytosolic phospholipase A2 gene expression in human bronchial epithelial cells. Biochim Biophys Acta. 1996, 1310: 175-184. 10.1016/0167-4889(95)00143-3.
Hayakawa M, Jayadev S, Tsujimoto M, Hannun YA, Ito F: Role of ceramide in stimulation of the transcription of cytosolic phospholipase A2 and cyclooxygenase 2. Biochem Biophys Res Commun. 1996, 220: 681-686. 10.1006/bbrc.1996.0464.
Puranam KL, Guo WX, Qian WH, Nikbakht K, Boustany RM: CLN3 defines a novel antiapoptotic pathway operative in neurode-generation and mediated by ceramide. Mol Genet Metab. 1999, 66: 294-308. 10.1006/mgme.1999.2834.
Xing X, Wang SC, Xia W, Zou Y, Shao R, Kwong KY, Yu Z, Zhang S, Miller S, Huang L, Hung MC: The ets protein PEA3 suppresses HER-2/neu overexpression and inhibits tumorigenesis. Nat Med. 2000, 6: 189-195. 10.1038/72294.
Nakae K, Nakajima K, Inazawa J, Kitaoka T, Hirano T: ERM, a PEA3 subfamily of Ets transcription factors, can cooperate with c-Jun. J Biol Chem. 1995, 270: 23795-23800. 10.1074/jbc.270.40.23795.
Basuyaux JP, Ferreira E, Stehelin D, Buttice G: The Ets transcription factors interact with each other and with the c-Fos/ c-Jun complex via distinct protein domains in a DNA-dependent and -independent manner. J Biol Chem. 1997, 272: 26188-26195. 10.1074/jbc.272.42.26188.
Herdegen T, Leah JD: Inducible and constitutive transcription factors in the mammalian nervous system: control of gene expression by Jun, Fos and Krox, and CREB/ATF proteins. Brain Res Brain Res Rev. 1998, 28: 370-490. 10.1016/S0165-0173(98)00018-6.
Duong T, Nikolaeva M, Acton PJ: C-reactive protein-like immunoreactivity in the neurofibrillary tangles of Alzheimer's disease. Brain Res. 1997, 749: 152-156. 10.1016/S0006-8993(96)01359-5.
Duong T, Acton PJ, Johnson RA: The in vitro neuronal toxicity of pentraxins associated with Alzheimer's disease brain lesions. Brain Res. 1998, 813: 303-312. 10.1016/S0006-8993(98)00966-4.
Allen MP, Zeng C, Schneider K, Xiong X, Meintzer MK, Bellosta P, Basilico C, Varnum B, Heidenreich KA, Wierman ME: Growth arrest-specific gene 6 (Gas6)/adhesion related kinase (Ark) signaling promotes gonadotropin-releasing hormone neuronal survival via extracellular signal-regulated kinase (ERK) and Akt. Mol Endocrinol. 1999, 13: 191-201.
Zhou H, Summers SA, Birnbaum MJ, Pittman RN: Inhibition of Akt kinase by cell-permeable ceramide and its implications for ceramide-induced apoptosis. J Biol Chem. 1998, 273: 16568-16575. 10.1074/jbc.273.26.16568.
Goswami R, Kilkus J, Dawson SA, Dawson G: Overexpression of Akt (protein kinase B) confers protection against apoptosis and prevents formation of ceramide in response to pro-apoptotic stimuli. J Neurosci Res. 1999, 57: 884-893. 10.1002/(SICI)1097-4547(19990915)57:6<884::AID-JNR14>3.3.CO;2-1.
Roy N, Mahadevan MS, McLean M, Shutler G, Yaraghi Z, Farahani R, Baird S, Besner-Johnston A, Lefebvre C, Kang X, et al: The gene for neuronal apoptosis inhibitory protein is partially deleted in individuals with spinal muscular atrophy. Cell. 1995, 80: 167-178.
Tsuda T, Marinetti MR, Masuelli L, Cutler ML: The Ras suppressor RSU-1 localizes to 10p13 and its expression in the U251 glioblastoma cell line correlates with a decrease in growth rate and tumorigenic potential. Oncogene. 1995, 11: 397-403.
Kajikawa S, Nakayama H, Suzuki M, Takashima A, Murayama O, Nishihara M, Takahashi M, Doi K: Increased expression of rat ribosomal protein L4 mRNA in 5-azacytidine-treated PC12 cells prior to apoptosis. Biochem Biophys Res Commun. 1998, 252: 220-224. 10.1006/bbrc.1998.9633.
Vito P, Pellegrini L, Guiet C, D'Adamio L: Cloning of AIP1, a novel protein that associates with the apoptosis-linked gene ALG-2 in a Ca2+-dependent reaction. J Biol Chem. 1999, 274: 1533-1540. 10.1074/jbc.274.3.1533.
Vito P, Lacana E, D'Adamio L: Interfering with apoptosis: Ca(2+)-binding protein ALG-2 and Alzheimer's disease gene ALG-3. Science. 1996, 271: 521-525.
Lacana E, Ganjei JK, Vito P, D'Adamio L: Dissociation of apoptosis and activation of IL-1beta-converting enzyme/Ced-3 proteases by ALG-2 and the truncated Alzheimer's gene ALG-3. J Immunol. 1997, 158: 5129-5135.
Kang JX, Bell J, Beard RL, Chandraratna RA: Mannose 6-phosphate/insulin-like growth factor II receptor mediates the growth-inhibitory effects of retinoids. Cell Growth Differ. 1999, 10: 591-600.
Degen WG, Weterman MA, van Groningen JJ, Cornelissen IM, Lemmers JP, Agterbos MA, Geurts van Kessel A, Swart GW, Bloemers HP: Expression of nma, a novel gene, inversely correlates with the metastatic potential of human melanoma cell lines and xenografts. Int J Cancer. 1996, 65: 460-465. 10.1002/(SICI)1097-0215(19960208)65:4<460::AID-IJC12>3.0.CO;2-E.
Nakagawa H, Koyama K, Murata Y, Morito M, Akiyama T, Nakamura Y: APCL, a central nervous system-specific homologue of adenomatous polyposis coli tumor suppressor, binds to p53-binding protein 2 and translocates it to the perinucleus. Cancer Res. 2000, 60: 101-105.
Ray MK, Datta B, Chakraborty A, Chattopadhyay A, Meza-Keuthen S, Gupta NK: The eukaryotic initiation factor 2-associated 67-kDa polypeptide (p67) plays a critical role in regulation of protein synthesis initiation in animal cells. Proc Natl Acad Sci USA. 1992, 89: 539-543.
Cuddihy AR, Li S, Tam NW, Wong AH, Taya Y, Abraham N, Bell JC, Koromilas AE: Double-stranded-RNA-activated protein kinase PKR enhances transcriptional activation by tumor suppressor p53. Mol Cell Biol. 1999, 19: 2475-2484.
Ozaki T, Hanaoka E, Naka M, Nakagawara A, Sakiyama S: Cloning and characterization of rat BAT3 cDNA. DNA Cell Biol. 1999, 18: 503-512. 10.1089/104454999315222.
Wang S, Nath N, Fusaro G, Chellappan S: Rb and prohibitin target distinct regions of E2F1 for repression and respond to different upstream signals. Mol Cell Biol. 1999, 19: 7447-7460.
Wang S, Nath N, Adlam M, Chellappan S: Prohibitin, a potential tumor suppressor, interacts with RB and regulates E2F function. Oncogene. 1999, 18: 3501-3510. 10.1038/sj/onc/1202684.
Matsushima M, Fujiwara T, Takahashi E, Minaguchi T, Eguchi Y, Tsujimoto Y, Suzumori K, Nakamura Y: Isolation, mapping, and functional analysis of a novel human cDNA (BNIP3L) encoding a protein homologous to human NIP3. Genes Chromosomes Cancer. 1998, 21: 230-235. 10.1002/(SICI)1098-2264(199803)21:3<230::AID-GCC7>3.3.CO;2-G.
Lee SW: H-cadherin, a novel cadherin with growth inhibitory functions and diminished expression in human breast cancer. Nat Med. 1996, 2: 776-782.
Lee SW, Reimer CL, Campbell DB, Cheresh P, Duda RB, Kocher O: H-cadherin expression inhibits in vitro invasiveness and tumor formation in vivo. Carcinogenesis. 1998, 19: 1157-1159. 10.1093/carcin/19.6.1157.
Pegg AE, Wechter RS, Clark RS, Wiest L, Erwin BG: Acetylation of decarboxylated S-adenosylmethionine by mammalian cells. Biochemistry. 1986, 25: 379-384.
Tabor CW, Tabor H: Polyamines. Annu Rev Biochem. 1984, 53: 749-790. 10.1146/annurev.bi.53.070184.003533.
Murakami Y, Ichiba T, Matsufuji S, Hayashi S: Cloning of antizyme inhibitor, a highly homologous protein to ornithine decarboxylase. J Biol Chem. 1996, 271: 3340-3342. 10.1074/jbc.271.7.3340.
Iwata S, Sato Y, Asada M, Takagi M, Tsujimoto A, Inaba T, Yamada T, Sakamoto S, Yata J, Shimogori T, et al: Anti-tumor activity of antizyme which targets the ornithine decarboxylase (ODC) required for cell growth and transformation. Oncogene. 1999, 18: 165-172. 10.1038/sj/onc/1202275.
Zhang H, Xu Q, Krajewski S, Krajewska M, Xie Z, Fuess S, Kitada S, Godzik A, Reed JC: BAR: An apoptosis regulator at the intersection of caspases and Bcl-2 family proteins. Proc Natl Acad Sci USA. 2000, 97: 2597-2602. 10.1073/pnas.97.6.2597.
Zhang L, Himi T, Morita I, Murota S: Hepatocyte growth factor protects cultured rat cerebellar granule neurons from apop-tosis via the phosphatidylinositol-3 kinase/Akt pathway. J Neurosci Res. 2000, 59: 489-496. 10.1002/(SICI)1097-4547(20000215)59:4<489::AID-JNR3>3.3.CO;2-0.
Alsbo CW, Wrang ML, Moller F, Diemer NH: Is the AMPA receptor subunit GluR2 mRNA an early indicator of cell fate after ischemia? A quantitative single cell RT-PCR study. Brain Res. 2001, 894: 101-108. 10.1016/S0006-8993(01)01985-0.
Li WP, Lai HW, Cheng SY, Yew DT: Somatostatin-positive neurons in the different parts of the brain in normal aging and Alzheimer's disease. Biol Signals. 1996, 5: 343-348.
Lorson CL, Strasswimmer J, Yao JM, Baleja JD, Hahnen E, Wirth B, Le T, Burghes AH, Androphy EJ: SMN oligomerization defect correlates with spinal muscular atrophy severity. Nat Genet. 1998, 19: 63-66.
Narasimhan P, Sklar R, Murrell M, Swanson RA, Sharp FR: Methyl-malonyl-CoA mutase induction by cerebral ischemia and neurotoxicity of mitochondrial toxin methylmalonic acid. J Neurosci. 1996, 22: 7336-7346.
Baeuerle PA, Henkel T: Function and activation of NF-kappa B in the immune system. Annu Rev Immunol. 1994, 12: 141-179.
Greene LA, Tischler AS: Establishment of a noradrenergic clonal line of rat adrenal pheochromocytoma cells which respond to nerve growth factor. Proc Natl Acad Sci USA. 1976, 73: 2424-2428.
Michel PP, Vyas S, Agid Y: Synergistic differentiation by chronic exposure to cyclic AMP and nerve growth factor renders rat phaeochromocytoma PC12 cells totally dependent upon trophic support for survival. Eur J Neurosci. 1995, 7: 251-260.
Pietu G, Alibert O, Guichard V, Lamy B, Bois F, Leroy E, Mariage-Sampson R, Houlgatte R, Soularue P, Auffray C: Novel gene transcripts preferentially expressed in human muscles revealed by quantitative hybridization of a high density cDNA array. Genome Res. 1996, 6: 492-503.
Decraene C, Reguigne-Arnould I, Auffray C, Pietu G: Reverse transcription in the presence of dideoxynucleotides to increase the sensitivity of expression monitoring with cDNA arrays. Biotechniques. 1999, 27: 962-966.
This work was supported by the CNRS and grants from European Union to J.M. (TMR projet Neuril) and BIOMED2 programs (EURO-IMAGE Consortium, BMH4-CT-97-2284) to C.A. C.M. was supported by Genome Express and C.D. acknowledges fellowships from the Fédération Française des Groupements Parkinsoniens (FFGP) and the Association pour la Recherche sur le Cancer (ARC).
About this article
Cite this article
Decraene, C., Brugg, B., Ruberg, M. et al. Identification of genes involved in ceramide-dependent neuronal apoptosis using cDNA arrays. Genome Biol 3, research0042.1 (2002). https://doi.org/10.1186/gb-2002-3-8-research0042