Sam68-Like Mammalian Protein 2, Identified by Digital Differential Display as Expressed by Podocytes, Is Induced in Proteinuria and Involved in Splice Site Selection of Vascular Endothelial Growth Factor
Clemens D. Cohen*,
Peter P. Doran,
Simone M. Blattner*,
Monika Merkle*,
Guo Q. Wang,
Holger Schmid*,
Peter W. Mathieson,
Moin A. Saleem,
Anna Henger*,
Maria P. Rastaldi and
Matthias Kretzler*
* Medizinische Poliklinik, Ludwig-Maximilians-University, Munich, Germany; Department of Medicine and Therapeutics, University College, Dublin, Ireland; Renal Immunopathology Laboratory, San Carlo Borromeo Hospital, Milan, Italy; and Children’s Renal Unit and Academic Renal Unit, University of Bristol, Bristol, United Kingdom
Address correspondence to: Dr. Clemens D. Cohen, Medizinische Poliklinik, Ludwig-Maximilians-University, Pettenkoferstrasse 8A, Munich, 80336, Germany. Phone: +49-89-2180-75841; Fax: +49-89-2180-75860; E-mail: ccohen{at}med.uni-muenchen.de
Received for publication February 22, 2005.
Accepted for publication March 10, 2005.
Podocytes, the glomerular epithelial cells of the kidney, shareimportant features with neuronal cells. In addition to phenotypicaland functional similarities, a number of gene products havebeen found to be expressed exclusively or predominantly by bothcell types. With the hypothesis of a common transcriptome sharedby podocytes and neurons, digital differential display was usedto identify novel podocyte-expressed gene products. Comparisonof brain and kidney cDNA libraries with those of other organsidentified Sam68-like mammalian protein 2 (SLM-2), a memberof the STAR family of RNA processing proteins, as expressedby podocytes. SLM-2 expression was found to be restricted inthe kidney to podocytes. In proteinuric diseases, SLM-2, a knownregulator of neuronal mRNA splice site selection, was foundsignificantly upregulated on mRNA and protein levels. Knockdownof SLM-2 by short interfering RNA in podocytes was performedto evaluate its biologic role. RNA splicing of vascular endothelialgrowth factor (VEGF), a key regulator of the filtration barrierand expressed as functionally distinct splice isoforms, wasevaluated. VEGF165 expression was found to be reduced by 25%after SLM-2 knockdown. In vivo, the glomerular expression ofSLM-2 correlated with the mRNA levels of VEGF165. This studydemonstrates the power of digital differential display to predictcell type–specific gene expression by hypothesis-drivenanalysis of tissue cDNA libraries. SLM-2-dependent VEGF splicingindicates the importance of mRNA splice site selection for glomerularfiltration barrier function.
Podocytes represent a highly specialized cell type with complexcell function. As the final barrier of the glomerular filtrationunit, podocytes build a highly complex network by interdigitatingso-called foot processes (1). Recently, several studies ledto the detection of specific gene products that are crucialfor the physiology of this cell type. These proteins are especiallyinvolved in the organization of the slit diaphragm (the cell–cellcontact between the foot processes), the specialized cytoskeletonof podocytes, and cell adhesion to the glomerular basement membrane.Many of these proteins are reported to be expressed primarilyor exclusively in neurons and podocytes. Besides nephrin, anessential protein of the slit diaphragm, densin; synaptopodin;glomerular epithelial protein 1; the synaptic vesicle moleculerab3A and its effector rabphilin-3a; and the amino acid transportersCAT3 and EAAT3 show preferential or restricted expression inboth of these cell types (2–8). This may reflect physiologicsimilarities, as both cells share important characteristicssuch as arborized cell process formation, highly specializedcell–cell contacts, and a complex cytoskeletal organization(9).
These intriguing similarities of glomerular epithelial and neuronalcells prompted us to take advantage of the hypothesis of a sharedtranscriptome to predict new podocyte-expressed genes. The digitaldifferential display (DDD) method was used to compare cDNA librariesof brain and kidney with those of all other tissues. This technique,part of the Cancer Genome Anatomy Project and available at theNational Center for Biotechnology Information, exploits thelarge number of publicly available cDNA libraries correspondingto different tissues (10). It permits selection of (1) cDNAlibraries to be compared, (2) comparison of the constituentsequences, and (3) output of a list of differentially expressedsequences. DDD experiments have been performed successfullyto compare cancerous and healthy tissue (e.g., reference 11)and to isolate cDNA specific for a given library (12). Thistechnique has not previously been used to predict cell type–specificgene expression by a hypothesis-driven comparison of two organswith other tissues. We searched for mRNA with high expressionin brain and kidney but no or significantly lower expressionin other tissues. This should allow the detection of mRNA templatespreferentially expressed by renal cells with a gene expressionresembling the one of brain cells. According to the shared transcriptomeof podocytes and neurons, the encoded proteins should be foundpreferentially expressed in podocytes.
This approach allowed the prediction of Sam68-like mammalianprotein 2 (SLM-2) to be expressed in podocytes, which was confirmedin vitro and in vivo. SLM-2 (13) is the gene product of thehuman gene KHDRBS3 on chromosome 8. The gene name is an acronymfor the known characteristics of SLM-2: KH domain containing,RNA binding, signal transduction associated. Here, we demonstratethat SLM-2 is regulated in glomerular diseases and is involvedin alternative mRNA splicing of the vascular endothelial growthfactor (VEGF-A).
DDD
DDD, a bioinformatic tool available at the National Center forBiotechnology Information (www.ncbi.nlm.nih.gov/UniGene/info_ddd.html),analyzes the frequencies of cDNA and expressed sequence tag(EST) in expression libraries. Briefly, this technique comparesthe constituent sequences of different libraries to determinethe relative frequency of each transcript in the libraries analyzed.Fisher exact test is used to determine the statistical significanceof the number of times that sequences from the selected librariesare assigned to a specific UniGene cluster. In this manner,the relative abundance of transcripts in each of the librariescan be determined and thus differential expression of genesidentified. A recently developed program, Digital Extractor(14), was used to compare human cDNA libraries of normal adultkidney and brain with other organ cDNA libraries.
Cell Culture
Three glomerular visceral epithelial cell lines were used andcultured according to previously published protocols: A conditionallyimmortalized human podocyte cell line (15), a stable T-SV40immortalized human glomerular visceral epithelial cell line(16), and a conditionally immortalized murine podocyte cellline (17).
Reverse Transcriptase–PCR and Real-Time Reverse Transcriptase–PCR
Total RNA was isolated using silica-gel columns following therespective protocol (RNeasy Mini; Qiagen, Hilden, Germany).Control mRNA from whole human kidney and microdissected humanglomeruli were isolated according to a reported protocol (18).Reverse transcriptase–PCR (RT-PCR) was performed usingheat-activated TaqDNA polymerase (Amplitaq Gold; Applied Biosystems,Darmstadt, Germany). The amplified cDNA was analyzed on a nondenaturing5% polyacrylamide gel, stained with Vistra Green (Amersham,Braunschweig, Germany), and visualized with ImageQuant Softwareon a Storm Fluorophosphorimager (both Molecular Dynamics, Krefeld,Germany). Real-time RT-PCR was performed on an ABI PRISM 7700Sequence Detection System (Applied Biosystems) as describedpreviously (18). All renal biopsies were performed accordingto the local ethical committees’ directives, and sampleswere processed according to the protocol of the European RenalcDNA bank (18).
The following oligonucleotide primers (300 nmol/L) and probes(100 nmol/L) were used: Human SLM-2 (NM 006558), sense primer5'-CCG TAA AAC AGT TCC CTA AGT TCA AC-3' (exon 2/3, i.e., spanningexons 2 to 3), antisense primer 5'-CTT CCT TGG CCT TGT CTC TCAT-3' (exon 3/4); fluorescence labeled probe (FAM) 5'-CCA CGTGGC AAT TCT CTG AAG CGT TTA-3'; human Ankyrin repeat domain7 (ANKRD7; NM 019644), sense primer 5'-CTG CAT AGA TAC CCA CAATTC ACT G-3' (exon 5/6), antisense primer 5'-TTG TAG AAA TGGGTT CCA TAT CTT CC-3' (exon 6/7); and human hypothetical proteinXP_172449 (XM172449), sense primer 5'-ACT CGC AGG CCC TGT GG-3',antisense primer 5'-CTC GGT GTC AAG GAG CAA GAG-3'. Commerciallyavailable predeveloped TaqMan reagents were used for human glyceraldehyde-3-phosphatedehydrogenase (GAPDH) and 18S rRNA (Applied Biosystems). Theprimers spanning at least one exon-intron boundary did not showany amplification signal tested on genomic DNA. All primersand probes were obtained from Applied Biosystems.
Western Blot
Cultured glomerular epithelial cells were harvested with lysisbuffer (50 mM Tris-HCl [pH 7.4], 1% Nonidet P-40, 0.25% sodiumdeoxycholate, 150 mM NaCl, 1 mM EGTA, 1 mM Na3VO4, and CompleteProtease Inhibitor Cocktail [Roche, Mannheim, Germany]). Extractedproteins were boiled in loading buffer for 10 min, resolvedby 10% SDS-PAGE under reducing conditions, and transferred toan Immobilon-P membrane (Millipore, Eschborn, Germany). Themembrane was blocked with 5% skim milk, incubated in a 1:100dilution of a purified polyclonal goat anti–SLM-2 antibody(M20; Santa Cruz Biotechnology, Heidelberg, Germany) overnight,and rinsed with PBS that contained 0.1% Tween 20. Immune complexeswere visualized using enhanced chemoluminescence (ECL; AmershamBiosciences, Freiburg, Germany). Commercially available murinebrain protein extract served as positive control for SLM-2 detection(ImmunoGlobe, Himmelstadt, Germany).
Immunofluorescence
Glomerular epithelial cells were grown on collagen-coated glasscoverslips for 24 h, fixed with formaldehyde (2% formaldehydeand 4% sucrose in PBS) for 5 min, incubated for 10 min with0.3% Triton in PBS, and washed in PBS. After 30 min of incubationwith blocking solution (2% FCS, 2% BSA, and 0.2% fish gelatinin PBS), the fixed cells were incubated with 1:10 to 1:50 polyclonalgoat anti–SLM-2 antibody (Santa Cruz Biotechnology) for60 min. After washing, the cells were incubated with 1:200 Cy2-labeledanti-goat antibody (Dianova, Hamburg, Germany) for 60 min. Coverslipswere fixed with mowiol solution (Calbiochem-Novabiochem, BadSoden, Germany), and fluorescence microscopy was performed.
Immunohistochemistry
Brain and kidney specimens from 3-mo-old Balb/C mice or humanbiopsies were fixed in 4% buffered paraformaldehyde and embeddedin paraffin. Immunohistochemistry was performed on 8-µm-thickbrain sections and 5-µm-thick kidney sections that weresequentially deparaffinized, rehydrated, and microwave-treatedin 0.01 M citrate buffer. After washing, sections were incubatedwith 0.5% avidin and 0.01% biotin to suppress endogenous avidin-bindingactivity, then postfixed in a methanol-H2O2 solution to blockendogenous peroxidase. The specimens then were incubated sequentiallywith a blocking solution (Zymed, Milan, Italy), the primarypolyclonal goat anti-mouse SLM-2 antibody (Santa Cruz Biotechnology),followed by the secondary rabbit anti-goat biotinylated antibody(Zymed) and the peroxidase-labeled streptavidin (Zymed). Peroxidaseactivity was detected with 3,5-diaminobenzidine (Sigma, St.Louis, MO). Sections then were counterstained by Cresyl Blue(DDK, Milan, Italy), dehydrated, and mounted in Permount (DDK).Nuclear staining in the brain was taken as positive control.Specificity of antibody labeling was demonstrated by the lackof staining after substituting PBS and proper control immunoglobulins(Zymed) for the primary antibody. Positive and negative controlswere run together.
Short Interfering RNA Experiments and VEGF Isoform Analysis
The conditionally immortalized human podocyte cell line (15)was transiently transfected under nonpermissive culture conditionsafter differentiation with sequence-specific short interferingRNA (siRNA; 200 nM) using Lipofectamine 2000 (Invitrogen, Karlsruhe,Germany). Target sequences were KHDRBS3–2 (SLM-2 siRNA2)sense 5'-GGC AAG GAU GAA GAA AAG Utt-3', antisense 5'-ACU UUUCUU CAU CCU UGC Ctt-3'; and KHDRBS3–3 (SLM-2 siRNA3) sense5'-GGA UGA AGA AAA GUA CAU C-3', antisense 5'-GAU GUA CUU UUCUUC AUC C-3', corresponding to regions in the exon 2 of SLM-2.All siRNA, including negative control of nonspecific nucleotidesequence (negative control siRNA 1), were obtained from Ambion(Silencer siRNA; Ambion, Austin, TX).
Primers for analysis of VEGF isoforms (NM 003376) were as follows:For RT-PCR separated by gel electrophoresis, sense 5'-CAA ATGTGA ATG CAG ACC AAA-3' (exon 4/5), antisense 5'-GGG AGG CTCCTT CCT CCT-3' (3'-UTR), expected product size 92 bp (VEGF121)and 224 bp (VEGF165), respectively. The primers were designedspecifically for the detection of stimulatory VEGF isoforms;recently described inhibitory VEGF isoforms were not amplified(19). Real-time RT-PCR VEGF165-specific assay: Sense primer5'-CAA ATG TGA ATG CAG ACC AAA GA-3' (exon 4/5), antisense primer5'-CAG GAA CAT TTA CAC GTC TGC G-3' (exon 7), and fluorescence-labeledprobe (FAM) 5'-AGC AAG ACA AGA AAA TCC CTG TGG GCC-3' (exon5/7). Assay for all human VEGF-A splice isoforms: Sense 5'-GCCTTG CTG CTC TAC CTC CAC-3' (exon 1), antisense 5'-ATG ATT CTGCCC TCC TCC TTC T-3' (exon 2), internal fluorescence-labeledprobe (FAM) 5'-AAG TGG TCC CAG GCT GCA CCC AT-3' (exon 1/2).
Statistical Analyses
Data are given as mean ± SD. Statistical analysis wasperformed using Kruskall-Wallis, Mann-Whitney tests, and Pearsoncorrelations (SPSS 12.0, SPSS Inc., Chicago, IL). P < 0.05was considered to indicate statistically significant differences.
DDD Predicts Brain/Kidney-Specific Templates
A DDD experiment was conducted to compare normal adult kidneyand brain cDNA libraries (total of 63,268 sequences) with allother available normal adult tissue libraries except for testis(204,515 sequences; Figure 1). Testis libraries were excludedbecause of the significant similarity of its transcriptome tothe one of brain (20). DDD takes advantage of the UniGene databaseby comparing the number of times that sequences from differentlibraries are assigned to a particular UniGene cluster. Thisanalysis produced 55 UniGene clusters that were specific tothe brain and/or kidney pool. Two further DDD experiments wereperformed to isolate brain-specific and kidney-specific clusters.The subtraction of these brain-specific and kidney-specificclusters from the brain and/or kidney results identified 11putative brain- and kidney-specific clusters. Detailed analysisof the chromosomal regions was performed. Thereby, four clustershad to be eliminated as they mapped to gene loci of previouslydeleted templates. From the remaining seven clusters, an automatedliterature search led to removal of an additional four sequencesbecause of their known significant expression in other tissuethan brain and kidney. The remaining three sequences, Ankyrinrepeat domain 7 (ANKRD7), hypothetical protein XP_172449, andSLM-2 (KHDRBS3), were assumed to be potentially restricted tobrain and kidney.
Figure 1. Schematic view of the digital differential display (DDD) experiment. Normal adult kidney and brain cDNA libraries (total of 63,268 sequences) were compared with all other available normal adult tissue libraries (204,515 sequences, testis excluded). Fifty-five UniGene clusters that were potentially specific for brain and/or kidney were retrieved. Further DDD experiments resulted in 11 clusters that were putatively specific for brain and kidney. Chromosomal localization and automated literature search led to removal of eight sequences. The expression of the remaining three sequences was predicted to be restricted to brain and kidney and therefore potentially specific for podocytes in the kidney.
SLM-2 Shows a Restricted Expression in Podocytes
RT-PCR experiments were carried out with specific primers oncDNA from two human glomerular epithelial cell lines. No ANKRD7or hypothetical protein XP_172449 mRNA templates could be amplified.RT-PCR for SLM-2 was positive in both glomerular epithelialcell lines and could also be detected on isolated human glomeruliand whole human kidney cDNA (Figure 2A).
Figure 2. Sam68-like mammalian protein 2 (SLM-2) expression in cultured podocytes. (A) Reverse transcriptase–PCR (RT-PCR) for SLM-2 on renal tissue. RT-PCR for SLM-2 resulted in a product of the corresponding size (146 bp) in whole human kidney, isolated human glomeruli, and two human podocyte cell lines (shown Podoc. II [15]); human brain cDNA served as positive control. Non–reverse transcribed samples (RT–) and no-template controls (ddH2O) remained negative in all experiments. (B) Western blot analysis of SLM-2. Protein extracts of two human (Podoc. I [16] and II [15]) and one murine (Podoc. III [17]) podocyte cell lines showed a single band of the expected size for SLM-2 (approximately 50 kD). Control murine brain protein extracts gave a band of identical size. Negative control experiments without primary antibody resulted in no positive signal. (C) Immunofluorescence for SLM-2 on cultured human podocytes. A nuclear signal was demonstrated on all glomerular epithelial cell lines, the human conditionally immortalized cell line shown here (15). Control experiments resulted in no nuclear positivity (insert).
The protein expression of SLM-2 in glomerular epithelial celllines was first demonstrated by Western blot. A single bandof the expected size was found in two human and one murine podocytecell line with an anti–SLM-2 antiserum; murine brain proteinextract served as positive control (Figure 2B). Immunofluorescenceexperiments were performed to demonstrate the intracellularlocalization of SLM-2 in human podocytes. In both cell lines,a predominantly nuclear signal was evident as reported in previousstudies for SLM-2 (21) (Figure 2C).
Furthermore, expression of SLM-2 in vivo was studied by immunohistochemistry.First, murine and human brain sections were studied to establishthe antiserum for SLM-2. A nuclear signal could be demonstratedin both species as reported previously for SLM-2 (22) (Figure 3A).In murine kidney sections, the nuclear staining was restrictedto glomerular epithelial cells (Figure 3B). No other glomerularcells were positive for SLM-2. Only in a few individual tubularepithelial cells could discrete nuclear signals also be detected.In human renal sections, the staining was less intense but againrestricted to glomerular epithelial cells (Figure 4B). Summarized,these findings confirm the podocyte-specific expression of SLM-2predicted by the computational approach.
Figure 3. Immunohistochemistry of SLM-2 on brain and kidney. (A) SLM-2 staining on murine brain. A positive nuclear signal was found in nearly all neurons of murine brain sections. Similar patterns were found on human brain sections (data not shown). Negative controls that used only the secondary antibody gave no positive signal (insert). (B) SLM-2 staining on murine kidney. In murine kidney, podocytes demonstrated a positive nuclear staining for SLM-2 (arrows). No other glomerular cell type was stained positive for SLM-2. In the tubulointerstitial compartment, a few tubular epithelial cells and some intravascular cells gave slight positive signals (data not shown). Negative controls gave no positive signal (insert).
Figure 4. SLM-2 expression in microdissected human glomeruli. (A) SLM-2 mRNA expression. SLM-2 mRNA was quantified in microdissected renal biopsies from control tissue and biopsies from various proteinuric diseases. SLM-2 was significantly induced in all proteinuric groups compared with control samples, independent of the housekeeper gene used for normalization (18S rRNA [shown], glyceraldehyde-3-phosphate dehydrogenase [GAPDH[; *P < 0.05, **P < 0.01 versus controls; CON, controls; MCD, minimal-change disease; BNS, benign nephrosclerosis; MGN, membranous glomerulonephritis; FSGS, focal segmental glomerulosclerosis). (B) Immunohistochemistry for SLM-2. Immunohistochemistry for SLM-2 on human tissue showed low positivity in healthy control subjects but pronounced staining for SLM-2 in glomerular epithelial cells. Shown here are a healthy control subject (A) and a biopsy of a patient with FSGS (B; nonsclerosed part of the glomerular tuft). (C) Quantification of SLM-2 immunohistochemistry. Quantification of the immunohistochemical staining for SLM-2 demonstrated a significantly higher number of positive nuclei in both human proteinuric diseases (MGN and FSGS) compared with CON (n = 5 for each condition; **P < 0.01 versus CON).
SLM-2 Is Induced in Human Proteinuric Diseases
To study a potential regulation of SLM-2 in podocytes in vivo,we investigated the expression of SLM-2 mRNA in microdissectedglomeruli from patients with proteinuria and nonproteinuriccontrol subjects. Glomeruli from 77 patients with proteinuriaand nonproteinuric control subjects were investigated (minimal-changedisease [MCD], n = 13; benign nephrosclerosis [BNS], n = 16;membranous glomerulonephritis [MGN], n = 31; focal segmentalglomerulosclerosis [FSGS], n = 9; controls, n = 8 [four tumornephrectomies and four pretransplant biopsies of living-relateddonors]). SLM-2 mRNA was significantly induced in the diseasegroups, independent from the housekeeper used for normalization(18S rRNA, P = 0.01; GAPDH, P = 0.02; SLM-2 expression normalizedfor 18S rRNA: control 1.45 ± 0.62, MCD 3.23 ±1.38, BNS 2.92 ± 1.17, MGN 2.79 ± 2.26, FSGS 3.08± 1.82; normalized to GAPDH: control 0.06 ± 0.04,MCD 0.23 ± 0.15, BNS 0.14 ± 0.10, MGN 0.10 ±0.07, FSGS 0.14 ± 0.12; Figure 4A).
To confirm a corresponding induction on protein level, we performedSLM-2 staining of fixed sections from routine renal biopsies.In control tissue (n = 5), only a minority of glomeruli showedpositive nuclear staining. In contrast, sections from two differentproteinuric diseases (MGN, n = 5; FSGS, n = 5) showed a significantlyincreased number of SLM-2–positive nuclei in glomerularepithelial cells (control 0.2 ± 0.45, MGN 2.0 ±0.7, FSGS 2.6 ± 1.14; P < 0.01 for both disease groupsversus control subjects; Figure 4, B and C).
Knockdown of SLM-2 in Podocytes Leads to Alternative Splicing of VEGF
These data may reflect a role of SLM-2 in the biology of podocytesduring proteinuria. To study the role of SLM-2 in podocytesin vitro, we established a knockdown model. Conditionally immortalizedhuman podocytes were transfected with two different siRNA directedagainst human SLM-2. Both siRNA led to a significant reductionof SLM-2 mRNA below 20% of control conditions (normalized toGAPDH: time control set as 1.0, negative control 0.99 ±0.35, SLM-2 siRNA2 0.16 ± 0.04, SLM-2 siRNA3 0.13 ±0.04; normalized to 18S rRNA: 1.0, 1.19 ± 0.59, 0.14± 0.04, 0.11 ± 0.06; n = 4; P < 0.05 for eachSLM-2 siRNA compared with control conditions; Figure 5A).
Figure 5. Alternative splicing of vascular endothelial growth factor (VEGF-A) after SLM-2 knockdown. (A) Knockdown efficiency by RNA interference. Human conditionally immortalized podocytes were transfected with negative control siRNA1, SLM-2 siRNA2, and SLM-2 siRNA3, respectively. SLM-2 expression was determined 48 h after transfection by real-time RT-PCR. Transfection of SLM-2 siRNA led to a significant decrease of SLM-2 templates below 20% of control conditions. Data shown are normalized to GAPDH; n = 4; *P < 0.05 for SLM-2 siRNA versus control conditions. (B) VEGF RT-PCR with gel electrophoresis. RT-PCR with primers amplifying all VEGF-A isoforms showed a reduction of the VEGF165 isoform (top band) compared with the VEGF121 isoform (bottom band). The order of control and siRNA groups corresponds to C. (C) Real-time RT-PCR for VEGF isoforms. Real-time RT-PCR with isoform-specific primers was performed to re-confirm the above results and showed a 25% reduction of the VEGF165 isoform normalized to all VEGF isoforms in the SLM-2 knockdowns (*P < 0.05 versus control). (D) Sequences of potential SLM-2 binding sites. Exon 7 of the human VEGF gene, included in the VEGF165 isoform, shows an element with high similarity to the potential exonic splicing enhancers binding SLM-2 in other known genes: The sequences in the human transformer-2 (tra) exon 2, human tau exon 2, and murine CD44 exon v5 were described by Stoss et al. (22). The human CD44 exon v5 and VEGF exon 7 were found by alignment. Identical nucleotides to the human VEGF exon 7 sequence are shown in bold.
SLM-2 has been shown to influence alternative splice site selectionof different genes. Two of these genes are CD44 and tau; SLM-2promotes inclusion of the variable exon v5 in CD44 and exons2 and 3 in tau (22–24). Although mRNA templates for CD44and tau could be detected in cultured human podocytes and microdissectedglomeruli, only SLM-2–independent isoforms were expressedin vitro and in vivo (CD44-v5– and tau-2–3–;data not shown). Next, VEGF-A, which is known to be expressedas alternatively spliced isoforms in podocytes, was studiedand endogenous VEGF was found to be alternatively spliced afterknockdown of SLM-2. SLM-2 knockdown led to a 25% reduction ofthe isoform VEGF165, which contains exon 7, compared with controlconditions (VEGF165/all VEGF isoforms: time control set as 1.0,negative control siRNA 0.96 ± 0.08, SLM-2 siRNA2 0.74± 0.22, SLM-2 siRNA3 0.76 ± 0.16; n = 4; P <0.05 for both SLM-2 siRNA; Figure 5, B and C). This substantiatesa role of SLM-2 for the inclusion of VEGF exon 7 and thereforepromoting expression of VEGF165 over VEGF121. As Stoss et al.(22) were able to characterize the potential binding elementof SLM-2 in exonic splicing enhancers of different genes, wesearched for homologies in the exon 7 of human VEGF. As shownin Figure 5D, a corresponding purin-rich binding site can alsobe found in exon 7 of the human VEGF gene. Nephrin, anotherpodocyte-expressed gene with reported splice isoforms, did notshow a potential SLM-2 binding element (25,26). As SLM-2 issignificantly upregulated in proteinuric diseases, the expressionof VEGF165 was analyzed in the 77 human renal biopsies specifiedabove. The expression of SLM-2 showed a highly significant correlationwith the expression of VEGF165 (r = 0.63, P < 0.01; Figure 6),potentially indicating a role of SLM-2 in mRNA splice siteselection in podocytes in vivo.
Figure 6. Correlation of VEGF165 isoform and SLM-2 in human biopsies. Expression of SLM-2 mRNA correlated positively with the expression of VEGF165 in microdissected human glomeruli (r = 0.63, P < 0.01). The expression levels of both genes were determined on isolated glomeruli of the 77 patients specified above; each patient’s sample is represented by one dot.
The computational approach of DDD allowed the prediction ofa novel podocyte expressed gene, SLM-2. Its expression is limitedto glomerular epithelial cells in the kidney, it is upregulatedin proteinuric diseases, and it is involved in mRNA splice siteselection. SLM-2 represents a mammalian member of the STAR family;STAR is an acronym for signal transduction and activation ofRNA. Its members share an extended hnRNP K homology (KH) RNAbinding domain and are involved in mitosis and cell-cycle regulation,signal transduction, and alternative splice site selection (reviewedin reference 27). The best characterized member of these RNA-bindingproteins is Sam68 (Src-associated substrate during mitosis of68 kD), first described as a target of the tyrosin kinase Srcduring mitosis (28,29). SLM-2 that is closely related to Sam68was first described by three independent groups in 1999. DiFruscio et al. (13) detected SLM-2 by screening a ZAP II mousebrain cDNA library and demonstrated its RNA-binding potential.This RNA binding is inhibited after phosphorylation by the epithelial-specifictyrosine kinase BRK/sik (30), whereas other kinases, such asSrc and Fyn, do not phosphorylate SLM-2 (13,31). Venables etal. first showed SLM-2’s nuclear localization (21); later,this group could demonstrate its regulation by proteasomic degradation(32). A significant expression of SLM-2 in the kidney was firstdetected by Lee’s group (33), who was also showing thatexpression of SLM-2 can lead to cell growth arrest.
As presented in our study, SLM-2 is expressed in glomerularepithelial cells and influences the splice site selection ofan important glomerular signal transductor, VEGF. The role ofVEGF in development and function of the renal glomerulus hasbeen studied comprehensively and was reviewed recently (34).It appears in three major isoforms, named for the number ofamino acids VEGF121, VEGF165, and VEGF189, corresponding onmRNA levels to exons 1 to 5 and 8 in all isoforms and additionallyexon 7 (VEGF165) or 6 and 7 (VEGF189) (35). In the glomerulus,all major VEGF isoforms are produced by the podocyte (36). TheVEGF receptors are found mainly on endothelial cells, but aweaker expression on podocytes suggests an additional paracrinesignaling pattern. Individual VEGF isoforms play specific rolesin glomerular function. Podocyte-specific deletion of VEGF inmice leads to death at birth and absence of a glomerular filtrationbarrier (37). Mice that express only VEGF122 have a significantreduction in the number of vascularized glomeruli compared withcontrols (38). Thus, VEGF122 provides partial but incompleterescue compared with the complete podocyte-specific knockout.In contrast, podocyte-specific overexpression of VEGF165 leadsto a striking collapsing glomerulopathy (37). The systemic administrationof VEGF165 can accelerate glomerular repair in necrotizing glomerulonephritis(39). Differences between VEGF isoforms are potentially dueto differences in binding to and release from the glomerularbasement membrane (40,41). However, little is known about themolecular regulation of alternative splice site selection ofVEGF mRNA. The data presented in this study suggest a significantrole of SLM-2 on isoform-specific expression of VEGF, herebypotentially influencing the functional regulation of the glomerularfilter.
It also demonstrates the importance of RNA splicing for celltype–specific regulation of gene expression. Besides transcriptionalregulation and RNA interference, this biologic mechanism enablesthe cell to expand the limited amount of encoded genes towardthe plethora of cell-specific proteins (42,43). RNA splice siteselection has not previously been studied in podocytes but couldadd a novel level to understanding the complex biology of thishighly specialized cell type.
The computational approach of DDD used in this study allowedfor the first time the prediction of a cell type–specificgene expression by comparison of two organs with other cDNAlibraries. This was possible because of the shared transcriptomeof neuronal cells and podocytes. Although this successful proofof concept led to the identification of SLM-2, three limitationshave to be discussed. First, SLM-2 has its most prominent expressionin testis and therefore was even named testis-STAR (21). Thereason that our DDD results revealed this gene is the exclusionof testis in the list of organs included in the analysis. Thiswas necessary because of the exhausting number of transcriptsfound in human testis, making a testis-included DDD screen almostimpossible. Furthermore, for our approach, brain cDNA were especiallyimportant, and it was shown recently by Guo et al. (20) thatthe transcriptomes of brain and testis are the most similarones. This made the exclusion of testis central for searchingcDNA predominantly expressed in brain and kidney. The secondlimitation of our approach is the lack of suborgan-specificcDNA libraries. The availability of glomerular cDNA libraries(instead of whole kidney) possibly would have increased thepredictive impact of the analysis. More specific, the reasonthat no known podocyte-specific gene was found by the DDD experimentmay be the low quantitative representation of podocytes in whole-kidneylibraries. As third weakness of the DDD approach, one may callthe low predictive value of 33% (one of three predicted genesconfirmed). This predictive power is comparable with the onesof previous studies that used DDD, in which 20 to 30% of predictedtemplates could be confirmed (11,44,45). It underlines the importanceof confirmation experiments after computational analysis, butas with all novel techniques, the number of false-positive callsshould decrease when DDD becomes a more frequently used toolfor gene expression prediction.
This study stresses the potential that DDD has to predict celltype–specific gene expression. It allowed prediction ofthe podocyte-restricted expression of SLM-2 in the glomerulus.Along with the confirmation of this finding, SLM-2 was shownto be upregulated in proteinuric diseases and involved in splicesite selection of VEGF. Studying the regulation of SLM-2 activityin vitro and in vivo may further elucidate alternative splicingas a central mechanism of podocyte biology.
Acknowledgments
This study was supported by the State of Bavaria "BayerischerHabilitationsförderpreis" and Friedrich-Baur-Stiftung toC.D.C., and EU-FPV: Chronic Kidney Disease (QLRT-2001-01215)to P.P.D., M.P.R., and M.K.
We thank S. Irrgang, K. Frach, and I. Edenhofer for excellenttechnical assistance; B. Luckow for support; and D. Schlondorfffor suggestions, support, and critical revision of the manuscript.
We thank all members of the European Renal cDNA Bank (ERCB)and their patients for support and cooperation. The membersof the ERCB at the time of this study were as follows: J.P.Rougier, P. Ronco, Paris, France; M.P. Rastaldi, G. D'Amico,Milan, Italy; F. Mampaso, Madrid, Spain; P. Doran, H.R. Brady,Dublin, Ireland; D. Monks, C. Wanner, Würzburg, Germany;A.J. Rees, Aberdeen, UK; F. Strutz, G. Muller, Gottingen, Germany;P. Mertens, J. Floege, Aachen, Germany; T. Risler, Tubingen,Germany; L. Gesualdo, F.P. Schena, Bari, Italy; J. Gerth, U.Ott, G. Wolf, Jena, Germany; R. Oberbauer, D. Kerjaschki, Vienna,Austria; B. Banas, B. Kramer, Regensburg, Germany; W. Samtleben,Munich, Germany; H. Peters, H.H. Neumayer, Berlin, Germany;K. Ivens, B. Grabensee, Dusseldorf, Germany; M. Zeier, H.J.Gröne, Heidelberg, Germany; M. Merta, V. Tesar, Prague,Czech Republic; K. Blouch, K. Lemley, Stanford, California;R. Nelson, Phoenix, Arizona; C.D. Cohen, M. Kretzler, D. Schlondorff,Munich, Germany.
Footnotes
Published online ahead of print. Publication date availableat www.jasn.org.
Pavenstadt H, Kriz W, Kretzler M: Cell biology of the glomerular podocyte.
Physiol Rev 83
: 253
–307, 2003[Abstract/Free Full Text]
Beltcheva O, Kontusaari S, Fetissov S, Putaala H, Kilpelainen P, Hokfelt T, Tryggvason K: Alternatively used promoters and distinct elements direct tissue-specific expression of nephrin.
J Am Soc Nephrol 14
: 352
–358, 2003[Abstract/Free Full Text]
Ahola H, Heikkila E, Astrom E, Inagaki M, Izawa I, Pavenstadt H, Kerjaschki D, Holthofer H: A novel protein, densin, expressed by glomerular podocytes.
J Am Soc Nephrol 14
: 1731
–1737, 2003[Abstract/Free Full Text]
Mundel P, Heid HW, Mundel TM, Kruger M, Reiser J, Kriz W: Synaptopodin: An actin-associated protein in telencephalic dendrites and renal podocytes.
J Cell Biol 139
: 193
–204, 1997[Abstract/Free Full Text]
Thomas PE, Wharram BL, Goyal M, Wiggins JE, Holzman LB, Wiggins RC: GLEPP1, a renal glomerular epithelial cell (podocyte) membrane protein-tyrosine phosphatase. Identification, molecular cloning, and characterization in rabbit.
J Biol Chem 269
: 19953
–19962, 1994[Abstract/Free Full Text]
Rastaldi MP, Armelloni S, Berra S, Li M, Pesaresi M, Poczewski H, Langer B, Kerjaschki D, Henger A, Blattner SM, Kretzler M, Wanke R, D’Amico G: Glomerular podocytes possess the synaptic vesicle molecule Rab3A and its specific effector rabphilin-3a.
Am J Pathol 163
: 889
–899, 2003[Abstract/Free Full Text]
Tagawa M, Shirasawa T, Yahagi Y, Tomoda T, Kuroyanagi H, Fujimura S, Sakiyama S, Maruyama N: Identification of a receptor-type protein tyrosine phosphatase expressed in postmitotic maturing neurons: Its structure and expression in the central nervous system.
Biochem J 321
: 865
–871, 1997
Gloy J, Reitinger S, Fischer KG, Schreiber R, Boucherot A, Kunzelmann K, Mundel P, Pavenstadt H: Amino acid transport in podocytes.
Am J Physiol Renal Physiol 278
: F999
–F1005, 2000[Abstract/Free Full Text]
Kobayashi N: Mechanism of the process formation; podocytes vs. neurons.
Microsc Res Tech 57
: 217
–223, 2002[CrossRef][Medline]
Strausberg RL, Buetow KH, Emmert-Buck MR, Klausner RD: The cancer genome anatomy project: Building an annotated gene index.
Trends Genet 16
: 103
–106, 2000[CrossRef][Medline]
Scheurle D, DeYoung MP, Binninger DM, Page H, Jahanzeb M, Narayanan R: Cancer gene discovery using digital differential display.
Cancer Res 60
: 4037
–4043, 2000[Abstract/Free Full Text]
Lord-Grignon J, Tetreault N, Mears AJ, Swaroop A, Bernier G: Characterization of new transcripts enriched in the mouse retina and identification of candidate retinal disease genes.
Invest Ophthalmol Vis Sci 45
: 3313
–3319, 2004[Abstract/Free Full Text]
Di Fruscio M, Chen T, Richard S: Characterization of Sam68-like mammalian proteins SLM-1 and SLM-2: SLM-1 is a Src substrate during mitosis.
Proc Natl Acad Sci U S A 96
: 2710
–2715, 1999[Abstract/Free Full Text]
Madden SF, O’Donovan B, Furney SJ, Brady HR, Silvestre G, Doran PP: Digital extractor: Analysis of digital differential display output.
Bioinformatics 19
: 1594
–1595, 2003[Abstract/Free Full Text]
Saleem MA, O’Hare MJ, Reiser J, Coward RJ, Inward CD, Farren T, Xing CY, Ni L, Mathieson PW, Mundel P: A conditionally immortalized human podocyte cell line demonstrating nephrin and podocin expression.
J Am Soc Nephrol 13
: 630
–638, 2002[Abstract/Free Full Text]
Delarue F, Virone A, Hagege J, Lacave R, Peraldi MN, Adida C, Rondeau E, Feunteun J, Sraer JD: Stable cell line of T-SV40 immortalized human glomerular visceral epithelial cells.
Kidney Int 40
: 906
–912, 1991[Medline]
Mundel P, Reiser J, Zuniga Mejia Borja A, Pavenstadt H, Davidson GR, Kriz W, Zeller R: Rearrangements of the cytoskeleton and cell contacts induce process formation during differentiation of conditionally immortalized mouse podocyte cell lines.
Exp Cell Res 236
: 248
–258, 1997[CrossRef][Medline]
Cohen CD, Frach K, Schlondorff D, Kretzler M: Quantitative gene expression analysis in renal biopsies: A novel protocol for a high-throughput multicenter application.
Kidney Int 61
: 133
–140, 2002[CrossRef][Medline]
Cui TG, Foster RR, Saleem M, Mathieson PW, Gillatt DA, Bates DO, Harper SJ: Differentiated human podocytes endogenously express an inhibitory isoform of vascular endothelial growth factor (VEGF165b) mRNA and protein.
Am J Physiol Renal Physiol 286
: F767
–F773, 2004[Abstract/Free Full Text]
Guo J, Zhu P, Wu C, Yu L, Zhao S, Gu X: In silico analysis indicates a similar gene expression pattern between human brain and testis.
Cytogenet Genome Res 103
: 58
–62, 2003[CrossRef][Medline]
Venables JP, Vernet C, Chew SL, Elliott DJ, Cowmeadow RB, Wu J, Cooke HJ, Artzt K, Eperon IC: T-STAR/ETOILE: A novel relative of SAM68 that interacts with an RNA-binding protein implicated in spermatogenesis.
Hum Mol Genet 8
: 959
–969, 1999[Abstract/Free Full Text]
Stoss O, Olbrich M, Hartmann AM, Konig H, Memmott J, Andreadis A, Stamm S: The STAR/GSG family protein rSLM-2 regulates the selection of alternative splice sites.
J Biol Chem 276
: 8665
–8673, 2001[Abstract/Free Full Text]
Arikan MC, Memmott J, Broderick JA, Lafyatis R, Screaton G, Stamm S, Andreadis A: Modulation of the membrane-binding projection domain of tau protein: Splicing regulation of exon 3.
Brain Res Mol Brain Res 101
: 109
–121, 2002[Medline]
Li K, Arikan MC, Andreadis A: Modulation of the membrane-binding domain of tau protein: Splicing regulation of exon 2.
Brain Res Mol Brain Res 116
: 94
–105, 2003[Medline]
Ahola H, Wang SX, Luimula P, Solin ML, Holzman LB, Holthofer H: Cloning and expression of the rat nephrin homolog.
Am J Pathol 155
: 907
–913, 1999[Abstract/Free Full Text]
Holthofer H, Ahola H, Solin ML, Wang S, Palmen T, Luimula P, Miettinen A, Kerjaschki D: Nephrin localizes at the podocyte filtration slit area and is characteristically spliced in the human kidney.
Am J Pathol 155
: 1681
–1687, 1999[Abstract/Free Full Text]
Lukong KE, Richard S: Sam68, the KH domain-containing superSTAR.
Biochim Biophys Acta 1653
: 73
–86, 2003[Medline]
Taylor SJ, Shalloway D: An RNA-binding protein associated with Src through its SH2 and SH3 domains in mitosis.
Nature 368
: 867
–871, 1994[CrossRef][Medline]
Fumagalli S, Totty NF, Hsuan JJ, Courtneidge SA: A target for Src in mitosis.
Nature 368
: 871
–874, 1994[CrossRef][Medline]
Haegebarth A, Heap D, Bie W, Derry JJ, Richard S, Tyner AL: The nuclear tyrosine kinase BRK/Sik phosphorylates and inhibits the RNA-binding activities of the Sam68-like mammalian proteins SLM-1 and SLM-2.
J Biol Chem 279
: 54398
–54404, 2004[Abstract/Free Full Text]
Stoss O, Novoyatleva T, Gencheva M, Olbrich M, Benderska N, Stamm S: p59(fyn)-mediated phosphorylation regulates the activity of the tissue-specific splicing factor rSLM-1.
Mol Cell Neurosci 27
: 8
–21, 2004[CrossRef][Medline]
Venables JP, Dalgliesh C, Paronetto MP, Skitt L, Thornton JK, Saunders PT, Sette C, Jones KT, Elliott DJ: SIAH1 targets the alternative splicing factor T-STAR for degradation by the proteasome.
Hum Mol Genet 13
: 1525
–1534, 2004[Abstract/Free Full Text]
Lee J, Burr JG: Salpalpha and Salpbeta, growth-arresting homologs of Sam68.
Gene 240
: 133
–147, 1999[CrossRef][Medline]
Eremina V, Quaggin SE: The role of VEGF-A in glomerular development and function.
Curr Opin Nephrol Hypertens 13
: 9
–15, 2004[Medline]
Tischer E, Mitchell R, Hartman T, Silva M, Gospodarowicz D, Fiddes JC, Abraham JA: The human gene for vascular endothelial growth factor. Multiple protein forms are encoded through alternative exon splicing.
J Biol Chem 266
: 11947
–11954, 1991[Abstract/Free Full Text]
Kretzler M, Schroppel B, Merkle M, Huber S, Mundel P, Horster M, Schlondorff D: Detection of multiple vascular endothelial growth factor splice isoforms in single glomerular podocytes.
Kidney Int Suppl 67
: S159
–S161, 1998[Medline]
Eremina V, Sood M, Haigh J, Nagy A, Lajoie G, Ferrara N, Gerber HP, Kikkawa Y, Miner JH, Quaggin SE: Glomerular-specific alterations of VEGF-A expression lead to distinct congenital and acquired renal diseases.
J Clin Invest 111
: 707
–716, 2003[CrossRef][Medline]
Mattot V, Moons L, Lupu F, Chernavvsky D, Gomez RA, Collen D, Carmeliet P: Loss of the VEGF(164) and VEGF(188) isoforms impairs postnatal glomerular angiogenesis and renal arteriogenesis in mice.
J Am Soc Nephrol 13
: 1548
–1560, 2002[Abstract/Free Full Text]
Shimizu A, Masuda Y, Mori T, Kitamura H, Ishizaki M, Sugisaki Y, Fukuda Y: Vascular endothelial growth factor165 resolves glomerular inflammation and accelerates glomerular capillary repair in rat anti-glomerular basement membrane glomerulonephritis.
J Am Soc Nephrol 15
: 2655
–2665, 2004[Abstract/Free Full Text]
Gengrinovitch S, Berman B, David G, Witte L, Neufeld G, Ron D: Glypican-1 is a VEGF165 binding proteoglycan that acts as an extracellular chaperone for VEGF165.
J Biol Chem 274
: 10816
–10822, 1999[Abstract/Free Full Text]
Iozzo RV, San Antonio JD: Heparan sulfate proteoglycans: Heavy hitters in the angiogenesis arena.
J Clin Invest 108
: 349
–355, 2001[CrossRef][Medline]
Black DL: Mechanisms of alternative pre-messenger RNA splicing.
Annu Rev Biochem 72
: 291
–336, 2003[CrossRef][Medline]
Stamm S, Ben-Ari S, Rafalska I, Tang Y, Zhang Z, Toiber D, Thanaraj TA, Soreq H: Function of alternative splicing.
Gene 344C
: 1
–20, 2005
Ghadersohi A, Odunsi K, Lele S, Collins Y, Greco WR, Winston J, Liang P, Sood AK: Prostate derived Ets transcription factor shows better tumor-association than other cancer-associated molecules.
Oncol Rep 11
: 453
–458, 2004[Medline]
Dennis JL, Vass JK, Wit EC, Keith WN, Oien KA: Identification from public data of molecular markers of adenocarcinoma characteristic of the site of origin.
Cancer Res 62
: 5999
–6005, 2002[Abstract/Free Full Text]
Received for publication February 22, 2005.
Accepted for publication March 10, 2005.
This article has been cited by other articles:
D. G. Nowak, J. Woolard, E. M. Amin, O. Konopatskaya, M. A. Saleem, A. J. Churchill, M. R. Ladomery, S. J. Harper, and D. O. Bates Expression of pro- and anti-angiogenic isoforms of VEGF is differentially regulated by splicing and growth factors
J. Cell Sci.,
October 15, 2008;
121(20):
3487 - 3495.
[Abstract][Full Text][PDF]
C. M. Costello, K. Howell, E. Cahill, J. McBryan, M. Konigshoff, O. Eickelberg, S. Gaine, F. Martin, and P. McLoughlin Lung-selective gene responses to alveolar hypoxia: potential role for the bone morphogenetic antagonist gremlin in pulmonary hypertension
Am J Physiol Lung Cell Mol Physiol,
August 1, 2008;
295(2):
L272 - L284.
[Abstract][Full Text][PDF]
M. P. Rastaldi, S. Armelloni, S. Berra, N. Calvaresi, A. Corbelli, L. A. Giardino, M. Li, G. Q. Wang, A. Fornasieri, A. Villa, et al. Glomerular podocytes contain neuron-like functional synaptic vesicles
FASEB J,
May 1, 2006;
20(7):
976 - 978.
[Abstract][Full Text][PDF]
Top
Abstract
Introduction
ZAP II mouse brain cDNA library and demonstrated its RNA-binding potential. This RNA binding is inhibited after phosphorylation by the epithelial-specific tyrosine kinase BRK/sik (30), whereas other kinases, such as Src and Fyn, do not phosphorylate SLM-2 (13,31). Venables et al. first showed SLM-2’s nuclear localization (21); later, this group could demonstrate its regulation by proteasomic degradation (32). A significant expression of SLM-2 in the kidney was first detected by Lee’s group (33), who was also showing that expression of SLM-2 can lead to cell growth arrest. 
(tra) exon 2, human tau exon 2, and murine CD44 exon v5 were described by Stoss et al. (