Heterogeneous vascular endothelial growth factor ... - Clinical Science

Clinical Science (1999) 97, 303–312 (Printed in Great Britain)

Heterogeneous vascular endothelial growth factor (VEGF) isoform mRNA and receptor mRNA expression in human glomeruli, and the identification of VEGF148 mRNA, a novel truncated splice variant Catheryne WHITTLE*, Kathleen GILLESPIE*, Rebecca HARRISON†, Peter W. MATHIESON* and Steve J. HARPER*

*Academic Renal Unit, University of Bristol, Southmead Hospital, Westbury-on-Trym, Bristol BS10 5NB, U.K., and †Department of Pathology, Southmead Hospital, Westbury-on-Trym, Bristol BS10 5NB, U.K.

A

B

S

T

R

A

C

T

Vascular endothelial growth factor (VEGF) mediates increased vascular permeability and endothelial mitogenesis, and may orchestrate normal glomerular permselectivity and proteinuria. Distinct isoforms result from differential gene splicing. VEGF binds to two cell surface tyrosine-kinase receptors, KDR (kinase domain region) and Flt-1 (fms-like tyrosine kinase-1). The latter also exists in a soluble form (sFlt), which is inhibitory. We have studied patterns of VEGF-isoform and VEGF-receptor expression in isolated single normal human glomeruli. mRNA from 190 glomeruli (from 20 individuals) was harvested on to magnetic beads, and nested reverse transcription–PCR was performed using primers for the VEGF isoforms and VEGF receptors. Simultaneous nested reverse transcription–PCR for CD45 was conducted in order to exclude leucocyte contamination. Unexpected products were isolated, cloned and sequenced. Multiple patterns of glomerular VEGF mRNA isoform expression were identified. Most frequently (58 %), all three common forms were expressed. VEGF189 (i.e. 189-amino-acid form of VEGF) was expressed in 63 %, VEGF165 in 85 % and VEGF121 in 84 % of glomeruli. Two unexpected PCR products were also identified : 18 % of glomeruli expressed VEGF145, and 27 % of glomeruli expressed a new truncated VEGF splice variant, VEGF148, lacking exon 6, the terminal part of exon 7 and exon 8. Multiple patterns of VEGF-receptor expression were also identified, the most common being expression of all three isoforms (28 %). Overall, KDR was seen in 59 % of glomeruli, Flt-1 in 45 % and sFlt in 57 %. Thus the expression of VEGF within normal glomeruli is complex and variable, with inter- and intra-individual variation. Furthermore, sFlt appears to be the co-dominant form of VEGF receptor expressed within glomeruli, suggesting that, in healthy individuals, a degree of VEGF autoregulation is the norm. The physiological importance of VEGF148 remains to be confirmed.

Key words : glomeruli, permselectivity, proteinuria, VEGF . "%) Abbreviations : Flt, fms-like tyrosine kinase ; GADPH, glyceraldehyde-3-phosphate dehydrogenase ; KDR, kinase domain region ; RT-PCR, reverse transcription–PCR ; sFlt, soluble form of Flt-1 ; VEGF, vascular endothelial growth factor, Correspondence : Dr S J Harper. The nucleotide sequence data reported have been submitted to the EMBL\GenBank\DDBJ Nucleotide Sequence Databases under accession no. AFO91352.

# 1999 The Biochemical Society and the Medical Research Society

303

304

C. Whittle and others

INTRODUCTION The normal glomerulus demonstrates size and charge permselectivity. Intrinsic renal disease is frequently associated with excess protein (mostly albumin) in the urine (proteinuria). Proteinuria is clinically important, not only because of the nephrotic syndrome but also because it appears to influence the progression of established glomerular lesions [1]. Vascular endothelial growth factor (VEGF), also known as vascular permeability factor, is a 34–42 kDa dimeric glycoprotein which mediates increased vascular permeability and endothelial mitogenesis. Differential exon splicing of the VEGF gene results in five main mRNA species which code for five secreted isoforms (subscripts denote numbers of amino acids) : VEGF , #!' VEGF , VEGF , VEGF and VEGF [2–4] (Figure ")* "'& "%& "#" 1). Isoforms VEGF , VEGF and VEGF are widely ")* "'& "#" expressed, whereas VEGF and VEGF are uncom#!' "%& mon [2–4]. The three common forms are expressed in normal adult and infant kidney, and in situ hybridization and immunohistochemical studies have identified the site of production to be primarily the visceral glomerular epithelial cells [5,6]. However, these studies have used probes and antibodies which do not differentiate between all the physico-chemically and possibly functionally different isoforms. VEGF is acidic and freely secreted. VEGF , the "#" "'& predominant form in most tissues, is more basic, has heparin-binding properties and, although a significant proportion remains cell-associated, most is freely secreted. VEGF is very basic ; it is cell-associated after ")* secretion and is bound avidly by heparin and the extracellular matrix, although it may be released as a soluble form (VEGF ) by heparin, heparinase or ""! plasmin [7,8]. Initial data supported the view that,

Figure 1

although VEGF , VEGF and VEGF may each ")* "'& "#" influence capillary permeability, only the two smaller forms are strong mitogens, VEGF being the most "'& potent [3,8]. However, more recent evidence suggests that VEGF may also have full mitogenic potential after ")* extracellular cleavage by urokinase [9]. VEGF receptors are of two types : stimulatory, membrane-bound Flt-1 (fms-like tyrosine kinase-1 ; also known as VEGFR-1) and KDR (kinase domain region ; VEGFR-2) ; and inhibitory, soluble Flt (sFlt). sFlt is a splice variant of Flt-1 which renders VEGF inactive when bound. These receptors are expressed by the glomerular endothelium [5,10]. The membrane-bound receptors also appear to be functionally distinct. In vitro, at least, KDR mediates mitogenesis, whereas Flt-1 lacks this ability [11] The intimate anatomical juxtaposition of the source of VEGF and its receptors [5,10], its extreme potency as a mediator of vascular permeability [12], the ability of VEGF to bind heparan sulphate (an important con")* tributor to the anionic nature of the glomerular basement ultrafiltration membrane), and the ability of VEGF to induce endothelial fenestrations [13] have resulted in VEGF being implicated in the control of normal glomerular permselectivity in health and of proteinuria in glomerular disease [14]. Despite its important potential role in normal glomerular physiology and pathology, and the structural and functional heterogeneity demonstrated by both VEGF and its receptors, current knowledge of constitutive glomerular VEGF isoform and receptor expression is limited. We describe a technique to analyse gene expression for the various forms of these mediators in individual glomeruli. In the course of these studies, we identified a novel truncated splice variant of VEGF and also observed the expression of a splice variant (VEGF )

VEGF splice variants resulting from differential exon splicing


"%&

Vascular endothelial growth factor in normal human kidney

previously only described in tissue derived from the female reproductive tract.

METHODS Our glomerular harvesting and single-glomerulus reverse transcription–PCR (RT-PCR) techniques were adapted from well established previously published methods [15]. Sequences of all relevant primers are given in Table 1, and the positions of VEGF primers are shown in Figure 2(A).

Nephrectomy tissue Nephrectomy tissue was supplied by the Department of Urology, Southmead Hospital, from patients undergoing

Table 1

nephrectomy for polar renal tumour (age range 47–78 years). All patients were non-diabetic and normotensive, with normal excretory renal function and no urinary sediment. A uniform collection system was used. Immediately after nephrectomy, excess fat was removed from the normal pole by blunt dissection, and several ‘ full-thickness ’ (capsule to deep medulla) needle biopsies were taken. These were placed in a Petri dish containing PBS (pretreated with diethyl pyrocarbonate as an inhibitor of RNases) and glomeruli were harvested as described below. The time lag from nephrectomy to obtaining isolated glomeruli in lysis buffer was in the order of 30 min. In addition, sections of kidney were taken for histological analysis.

Nested PCR primer sequences

Gene

Primer name

Forward primer

Reverse primer

Amplicon length

GAPDH

External GAPDH Internal GAPDH External CD45 Internal CD45 External VEGF Internal VEGF isoforms Internal VEGF common exon External Flt-1 Internal Flt-1 External KDR Internal KDR External sFlt Internal sFlt

CACCCATGGCAAATTCCATG GCCAAAAGGGTCATCATCTC AGCTCGAAAGCCCTTTAACC AGAATACTGGCCGTCAATGG GTGCATTGGAGCCTTGCCTT (exon 1) GTGAATGCAGACCAAAGAAAG AAGGAGGAGGGCAGAATCAT CAGGAATGTATACACAGGGG ATCAGAGATCAGGAAGCACC TGATGTGGTTCTGAGTCCGT GACTTCAACTGGGAATACCC CAGGAATGTATACACAGGGG ACAATCAGAGGTGAGCACTG

GCAGGTTTTTCTAGACGGCA GTAGAGGCAGGGATGATGTTC ACATCCACTTTGTTCTCGGC GCTGAAGGCATTCACTCTCC GCAGCGTGGTTTCTGTATCG (3h untranslated region) AAACCCTGAGGGAGGCTC GCTGTAGGAAGCTCATCTCT CGGCACGTAGGTGATTTCTT GGAACTTCATCTGGGTCCAT GGTAACCAAGGTACTTCGCA CATGGACCCTGACAAATGTG AAACACAGAGAAGGCAGTGC CTGCTATCATCTCCGAACTC

524 287 629 237 774, 723, 651, 519 96, 228, 300, 351 281 654 441 399 208 387 219

CD45 VEGF

Flt-1 KDR sFlt

A

B

Figure 2

Positions of external and internal primers

(A) Simplified VEGF gene exon splicing, showing sites of primers for VEGF isoforms. E, external primers ; T, primers for VEGF conserved exons ; I, primers spanning exon deletions to identify isoform mRNAs. (B) Sites of primer pairs (a, b and c) used to identify the mRNA of the new splice variant VEGF148 (see Table 2). # 1999 The Biochemical Society and the Medical Research Society

305

306


Since the tissue studied was derived from surgical specimens and would otherwise have been discarded, no Ethical Committee approval or individual patient consent was required.

Glomerular harvesting Single subcapsular, mid-cortical and juxta-medullary glomeruli were isolated from needle biopsies. A total of 190 glomeruli were studied from 20 individuals. Glomeruli were collected under an Olympus SZ11 dissection microscope using a pair of fine tweezers ground to a point and a 12-gauge needle. Only glomeruli visibly isolated from surrounding tissue were used. It became evident with time that glomeruli could be encouraged to spontaneously ‘ pop ’ out of their Bowman’s capsule by gently squeezing the biopsy core with fine tweezers.

Harvesting of glomerular mRNA and reverse transcription mRNA was extracted from isolated glomeruli using a poly(T) Dynabead kit (Dynal RNA Direct), using the manufacturer’s instructions [15]. Briefly, each glomerulus was lysed and then incubated with poly(T) beads. The beads were washed several times to remove cell debris, prior to a reverse transcription step, which was carried out while the RNA was still attached to the Dynabeads, again following the manufacturer’s instructions (Reverse Transcription kit ; Promega). Following reverse transcription, the cDNA-linked Dynabeads were incubated in 50 µl of TE buffer (10 mM Tris\1 mM EDTA, pH 8) and heated to 95 mC for 1 min. This TE buffer was then discarded, removing the RNA that had become detached from the beads. The cDNA-linked Dynabeads were then resuspended in 50 µl of TE buffer for storage or PCR.

Initial experiments and standardization of nested RT-PCR In preliminary experiments, standard RT-PCR was performed on isolated glomeruli. However, we found that the 40–45 cycles required to detect VEGF also resulted in poor-quality smears. Nested RT-PCR was therefore used. This was standardized by varying the cycle number and template quantity in each phase. We found that a cycle number of 18 in the first PCR and of 25 in the second produced specific bands. Further initial experiments from a number of cases demonstrated that the pattern of VEGF mRNA isoforms within a single glomerulus was always reproducible with repeated experiments. # 1999 The Biochemical Society and the Medical Research Society

Nested PCR for VEGF isoforms and total VEGF cDNA-linked Dynabeads were washed in 10i PCR buffer (Hybaid Proof 2 buffer) prior to addition to a 25 µl PCR reaction mixture containing external forward and reverse primers (Figure 2 ; Table 1) (400 nM each), 1i Proof 2 buffer (including 2.5 mM MgCl ) and 200 µM # dNTPs (Hybaid). The reaction was ‘ hot started ’ by addition of 0.75 unit of Taq polymerase (Hybaid ; PWO proof mix) in a first-round synthesis carried out as follows : 95 mC for 7 min (‘ hot start ’ after 3 min), 95 mC for 1 min, 55 mC for 1.5 min and 72 mC for 2 min. The cDNA-linked Dynabeads were then removed by pelleting after 2 min at 95 mC. The supernatant containing the first strand of cDNA was then amplified using 18 cycles of 95 mC for 30 s, 55 mC for 50 s and 72 mC for 50 s, with a final extension for 5 min at 72 mC. A 1 µl portion of the first-round PCR was used as a template in the second round of amplification. The reaction mixture was as above, except that either internal forward and reverse primers for VEGF isoforms (spanning exon deletions) or a VEGF common exon were used (Figure 2 ; Table 1). The final MgCl concentration was # 1.5 mM, and 0.75 unit of Taq polymerase was added per reaction. Conditions for the second round were as follows : 25 cycles of 95 mC for 30 s, 55 mC for 40 s and 72 mC for 40 s, followed by a 5 min extension at 72 mC. All PCR reactions were carried out in a Hybaid thermal cycler.

Nested PCR for glyceraldehyde-3-phosphate dehydrogenase (GAPDH), CD45 and the VEGF receptors KDR, Flt-1 and sFlt cDNA-linked Dynabeads were washed in 10i PCR buffer containing 1.5 mM MgCl (Promega) prior to # addition to a 25 µl PCR reaction mixture containing external forward and reverse primers (Figure 2 ; Table 1) (400 nM each), 1i Promega PCR buffer and 200 µM dNTPs (Hybaid). The reaction was ‘ hot started ’ by the addition of 1 unit of Taq polymerase (Promega) in a firstround synthesis carried out as follows : 95 mC for 7 min (‘ hot start ’ after 3 min), 95 mC for 1 min, 55 mC for 1.5 min and 72 mC for 2 min. The cDNA-linked Dynabeads were then removed by pelleting after 2 min at 95 mC. The supernatant containing the first strand of cDNA was then amplified using 18 cycles of 95 mC for 30 s, 55 mC for 40 s and 72 mC for 40 s, with a final extension of 5 min at 72 mC. A 1 µl portion of the first-round PCR was used as template in the second round of amplification. The reaction mixture was as above, except that either internal forward and reverse primers for the appropriate molecule were used and 1 unit of Taq polymerase was added.


Table 2

Primer sequences used to identify the mRNA of the new splice variant VEGF 148

Primer name

Forward primer

Reverse primer

Amplicon length

External VEGF (a) Internal VEGFi (b) Deletion VEGF (c) Exon 7 VEGF

gtgcattggagccttgcctt gtgaatgcagaccaaagaaag agactcgcgttgcaagatgt ctcagagcggagaaagcatt

gcagcgtggtttctgtatcg aaaccctgagggaggctc gcagcgtggtttctgtatcg gcagcgtggtttctgtatcg

774, 723, 651, 519 96, 228, 300, 351 135 199, 234

Conditions for the second round were as follows : 25 cycles of 95 mC for 30 s, 55 mC for 40 s and 72 mC for 40 s, followed by a 5 min extension at 72 mC.

Detection of PCR products PCR products were mixed with 3 µl of loading dye (0.25 % Bromophenol Blue\50 % glycerol) and electrophoresed in a 3 % (w\v) agarose (Sigma) gel. Bands were visualized by ethidium bromide staining on a UVP dualintensity transilluminator.

Nested PCR controls and experimental design Each glomerulus was examined for : (a) ‘ total ’ VEGF, using primers located in exons 2–4 (which are conserved in all VEGF isoforms), to act as an internal control, (b) VEGF isoforms, (c) VEGF receptors, (d) GAPDH, a housekeeping gene, and (e) CD45, the leucocyte common antigen, to detect any contamination by RNA from trafficking leucocytes. Each reaction included negative controls in which either template or the reverse transcription enzyme was omitted. Any glomerulus that was positive for CD45, and all glomeruli from kidneys subsequently found to be histologically abnormal, were excluded from analysis. In addition, all VEGF-isoform and VEGF-receptor specific bands were cloned and sequenced to confirm their identity. Multiple clones were sequenced of each of the expected VEGF-isoform and VEGF-receptor PCR products, and each confirmed the published sequences of VEGF , VEGF , VEGF , Flt-1, KDR and sFlt, as "#" "'& ")* anticipated.

Sequencing of unexpected PCR products The unexpected PCR products were ligated into pGEM T vector (Promega) using the manufacturer’s instructions. The products of ligation were transformed into competent TOP 10Fh Escherichia coli (Invitrogen, Groningen, The Netherlands) according to the recommended protocol, and positive colonies were selected on ampicillin (100 µg\ml)\agar plates. Clones shown to contain VEGF inserts by PCR with M13 primers were purified using a Qiagen QIA Quick PCR Purification kit

(Qiagen, Crawley, W. Sussex, U.K.) and sequenced on an ABI 373 automatic sequencer using fluorescent dye terminators. The sequence of one unexpected product was novel. To confirm this novel sequence, two additional primer sets were designed (primer sets b and c ; Table 2, Figure 2B). In the case of primer set c, the band of interest was excised and used as the template in a further PCR reaction. The products were separated on Spredex 600 gels (VH Bio, Newcastle-Upon-Tyne, U.K.), ligated using an Invitrogen TOPO TA cloning kit and transformed and sequenced as described above. A total of nine clones from the three primer sets designed to identify the new isoform of VEGF were sequenced, all of which gave the same results.

RESULTS A total of 190 single glomeruli were isolated. Of these, 26 (from two nephrectomy specimens) were subsequently excluded because of minor histological abnormalities in the nephrectomy specimens (tubulo-interstitial disease). In addition, six glomeruli (2.6 %) demonstrated a faint CD45 band and were also excluded from analysis. Thus 158 glomeruli remained, from 18 individuals. Individual glomerular numbers ranged from 5 to 16. All glomeruli demonstrated GAPDH expression ; 11 (7 %) were GAPDH-positive but negative for VEGF mRNA.

VEGF isoform mRNA expression Marked heterogeneity was seen in VEGF isoform and receptor expression in different glomeruli from the same nephrectomy specimen ; indeed from the same needle biopsy. Figure 3 shows an electrophoretic gel of the VEGF mRNA isoforms expressed by 11 glomeruli from the same needle biopsy. These glomeruli were harvested sequentially from the subcortical area to the juxtamedullary zone. This clearly demonstrates that individual glomeruli show different patterns of VEGF mRNA isoform expression. The first glomerulus expressed the three common forms of VEGF mRNA (Figure 3) ; however, the VEGF isoform mRNA expression of the glomerulus in lane 2 (which was anatomically approx. 50 µm away from the first) was totally different. # 1999 The Biochemical Society and the Medical Research Society

307

308


Figure 3 Heterogeneity of VEGF mRNA isoform expression in 11 glomeruli from a single needle biopsy

PCR products were separated on a 3 % (w/v) agarose gel. Each lane contains material derived from a single glomerulus. Size (bp) is indicated on the right.

Classification of each glomerulus according to the three common VEGF isoforms detected revealed multiple patterns of expression. However, two frequent patterns were seen (Table 3). The commonest pattern was seen in 58 % of glomeruli, which expressed all three common forms previously identified in renal tissue : VEGF , VEGF and VEGF . However, there was ")* "'& "#" much variability between individuals, with the frequency of this pattern ranging from 17 to 86 %. In addition, 20 % (individual range 0–60 %) of glomeruli expressed only VEGF and VEGF , and 11 % (range 0–50 %) "'& "#" expressed one isoform only. No VEGF mRNA isoforms were detected in 7 % of the glomeruli. Overall, VEGF ")* was expressed in 63 % of glomeruli, VEGF in 85 % and "'& VEGF in 84 %. In addition, unexpected PCR products "#" were consistently seen in a minority of glomeruli (Figure 4). Subsequent identification of these products revealed the expression of VEGF (Figure 4, Product D) in 18 % "%& of glomeruli, and 27 % of glomeruli (range 0–67 %) expressed a new splice variant, VEGF (Figure 4, "%) Product C). This new splice variant was seen in glomeruli from eight individuals only.

VEGF 148 mRNA The sequencing data identified a 35 bp deletion at the end of exon 7. This dictated a frame-shift and a premature stop codon at the start of the short (19-base) exon 8. The

Table 3 Frequency of VEGF mRNA isoform expression in normal glomeruli

Values in parentheses represent ranges for individual tissue samples. mRNA species present

Frequency (%)

VEGF121, VEGF165 and VEGF189 VEGF121 and VEGF165 One form only (VEGF121, VEGF165 or VEGF189) Minor patterns (VEGF121 and VEGF189, or VEGF165 and VEGF189) None

58 (17–86) 20 (0–60) 11 (0–50) 4 (0–20) 7 (0–30)


Figure 4 Unexpected nested RT-PCR products seen in some glomeruli using internal primer set ‘ a ’

Lanes 1–3, individual glomeruli from the same needle biopsy ; lane 4 and 5, products from five combined glomeruli ; lane 6, control (no reverse transcriptase) ; lane 7, control (beads only) ; lane 8, single glomerulus ; lane 9, peripheral blood mononuclear cells. Product A, 300 bp (VEGF189) ; Product B, 228 bp (VEGF165) ; Product C, 194 bp (VEGF148) ; Product D, 169 bp (VEGF145) ; Product E, 96 bp (VEGF121). Size (bp) is indicated on the left. protein product of the new mRNA splice variant would be predicted to contain 148 amino acids after signal peptide cleavage, i.e. VEGF (a truncated form of "%) VEGF ). A comparison of the nucleotide and predicted "'& amino acid sequences of VEGF with those of the "%) predominant VEGF is shown in Figure 5. "'& The novel sequence was identical in multiple experiments with nine different clones and three sets of primer pairs, in five different glomeruli from four different individuals. The sequences and positions of the three primer pairs are detailed in Table 2 and Figure 2(B) respectively. These were : (a) one pair placed beyond exon deletions to enable all isoforms to be identified, (b) a second pair, the forward primer of which spanned the novel exon boundary of VEGF and (c) a third pair in "%) which the forward primer was positioned in exon 7. We have also detected the expression of VEGF mRNA in "%) a number of stored and fresh tissue samples (Figure 6).

VEGF-receptor mRNA expression The expression of mRNAs for the three VEGF receptors was also heterogeneous (Table 4). Although expression of all three was the commonest pattern (28 % of glomeruli ; individual range 0–64 %), 14 % of glomeruli (range 0–33 %) expressed KDR and sFlt, and 21 % (range 0–33 %) expressed only one of the three receptors. In 24 % of glomeruli, no mRNA was detected for any of the receptors. Overall, KDR was expressed in 59 % of glomeruli, Flt-1 in 45 % and sFlt in 57 %. There was substantial variation in both VEGF-isoform and VEGF-receptor mRNA expression in isolated single glomeruli within and between individuals. The scatter of frequencies of particular isoform and receptor mRNA patterns was always normally distributed. No specific patterns of VEGF-isoform or VEGFreceptor expression were apparent when comparing subcortical with juxta-medullary glomeruli. No specific pattern of VEGF-isoform expression was convincingly


Figure 5

Comparison of the nucleotide and predicted amino acid sequences of VEGF 148 and VEGF 165 Table 4 Frequency of VEGF mRNA receptor expression in normal glomeruli

Values in parentheses represent ranges for individual tissue samples.

Figure 6

Expression of VEGF 148 in other tissues

Lanes 1 and 2, single glomeruli ; lane 3, brain ; lane 4, tonsil ; lane 5, liver ; lane 6, placenta ; lane 7, peripheral blood mononuclear cells. Size (bp) is indicated on the left. associated with any particular pattern of VEGF-receptor expression. However, VEGF and VEGF products "%& "%) were only identified in the presence of VEGF .

"'&

mRNA species present

Frequency (%)

Flt-1, KDR and sFlt KDR and sFlt One only (Flt-1, KDR or sFlt) None Minor patterns (Flt-1 and KDR, or Flt-1 and sFlt)

28 (0–64) 14 (0–33) 20 (0–33) 24 (0–50) 14 (0–80)

Enumeration of glomerular cross-sections on haematoxylin and eosin staining revealed significant glomerular sclerosis in 5.3 % of glomerular cross-sections, a value comparable with the proportion of glomeruli in which no detectable VEGF mRNA was identified. # 1999 The Biochemical Society and the Medical Research Society

309

310


DISCUSSION VEGF is believed to play a key role in the glomerulus in health and disease [14]. In view of the complexity of this molecule’s expression, differential splicing and receptor subtypes, we felt it was essential to study the normal glomerulus before attempting to analyse diseased tissue. The available antibodies to VEGF do not differentiate between VEGF isoforms, and nor do in situ hybridization probes. In addition, ligand-binding studies do not differentiate between receptor types. We devised a sensitive technique capable of differentiating between the various VEGF isoforms and receptors, and our study is the first to report the complex patterns of expression of these molecules in isolated normal human glomeruli. We analysed a large number of glomeruli from 20 individuals. We also tested whether VEGF-isoform and VEGF-receptor expression differ between subcortical and juxta-medullary glomeruli as a result of differences in the glomerular micro-environment between these two sites (e.g. the relative hypoxia of juxta-medullary glomeruli). Single isolated glomeruli harvested as described therefore allowed us to study glomeruli from known anatomical sites within the renal cortex ; this would not have been possible with sieved glomeruli or cultured podocytes, both of which are known to express the three common VEGF isoforms. Our results demonstrate a heterogeneity in glomerular VEGF-isoform and -receptor expression, and therefore suggest an added level of complexity to normal glomerular and VEGF biology that was not previously appreciated. Since access to truly ‘ normal ’ human tissue is ethically and practically difficult, we took tissue from the uninvolved poles of kidneys removed because of tumours. Most donors were therefore elderly, and all had malignant disease. However, we did our utmost to exclude any possible confounding factors. After nephrectomy, perinephric fat was removed from the kidney to enable the tumour to be clearly visualized. Biopsies were taken from the opposite pole. Samples of kidney adjacent to the biopsy sites, but nearer the tumour, were taken for histological assessment. Glomeruli from two nephrectomy samples were subsequently excluded from analysis because of minor changes apparent on histological examination. These were changes associated with chronic interstitial inflammation ; no glomerular pathology was identified. Although these could be regarded as changes associated with normal aging [16], we decided to exclude these kidneys from analysis (although their inclusion would not have materially altered our conclusions). Furthermore, any glomeruli that were positive for CD45 (the leucocyte common antigen) were also excluded from analysis. The only alternative source of ‘ normal ’ glomeruli would be unused transplant organs. However, these are uncommon, are frequently derived from a similar-aged cohort and are often preserved ex vivo for # 1999 The Biochemical Society and the Medical Research Society

several hours. In addition, animal studies have demonstrated significant immunological abnormalities in organs derived from ‘ brain-dead ’ animals [17]. Thus our results must be interpreted with the knowledge that the donor tissue may not be truly ‘ normal ’, but we submit that our sources were the best available and our exclusion criteria were very rigorous. Relative VEGF-isoform and VEGF-receptor expression is heterogeneous in normal glomeruli. The reason for this is unclear ; however, our data suggest that glomeruli exhibit a degree of functional independence. Perhaps this should not surprise us, since in primary renal disease it is not uncommon to observe adjacent glomeruli demonstrating a variety of lesions (e.g. global sclerosis, crescent formation, mesangial proliferation). Some micro-environmental factors must govern this variation. Although it would be preferable to confirm this heterogeneity in the relative expression of VEGF isoforms and receptors between different glomeruli using a technique such as in situ hybridization, this has not been possible. Despite considerable experience in developing new strategies with non-isotopic in situ hybridization [18–24], we have not been able to develop a probe cocktail that will distinguish between the VEGF isoform mRNAs. This is primarily because of the striking similarity between isoforms, with the only sequence specific for each being the spliced exon–exon boundaries. The predominant glomerular VEGF isoforms were VEGF and VEGF . These are generally freely "'& "#" secreted and are potent mitogens [3,8], although they have been shown to have permeability properties also. The dominant stimulatory VEGF receptor in normal glomeruli is KDR, which is known to mediate mitogenesis [11], although the glomeruli are not sites of new vessel formation. However, the shear stresses on glomerular endothelial cells are considerable, and it is generally accepted that 80 % of cell turnover in normal glomeruli is due to endothelial cell proliferation. It has therefore been proposed that the integrity of the endothelium is maintained by this stimulus. VEGF is the least abundant common form in total ")* renal tissue and sieved glomeruli [5]. It was present in 63 % of single isolated glomeruli. This form of VEGF is strongly basic, and therefore cationic. It is known to bind strongly to heparan sulphate, an important anionic component of the glomerular basement membrane. It can be hypothesized that this form of VEGF may have a greater chance of affecting the control of normal permselectivity and proteinuria, since, after secretion, VEGF is generally associated with cells and the ")* extracellular matrix, and could move up the filtration pressure gradient by moving down an electrostatic one. The present study provides additional new information about the expression of some uncommon VEGF isoforms. The mRNA splice variant VEGF was initially "%& described in tumour cell lines derived from the female


reproductive tract [25]. A recent functional study has shown this translated product to have a unique combination of properties distinct from those of the other isoforms [4]. We found it expressed in 18 % of normal glomeruli, and subsequently in peripheral blood mononuclear cells. This is the first description of its occurrence outside the female reproductive tract or tissue derived from it. Furthermore, we have identified a new truncated splice variant, VEGF . This form results from a 35 bp deletion "%) at the end of exon 7. This suggests that exon 7 may be in two parts that are differentially spliced, although there is no intervening intron. There is a clear precedence for this in exon 6 of VEGF. The deletion changes the reading frame and a premature stop codon results, producing the additional deletion of exon 8. The predicted product of VEGF mRNA would be a "%) truncated form of VEGF , perhaps without its modest "'& ability to bind heparin but retaining a mitogenic capacity. Certainly it would contain amino acids 1–110, which seem to be needed for this function [9]. Furthermore the VEGF product would not be dissimilar to the enzymic "%) product VEGF , which results from cleavage of matrix""! associated VEGF . Conversely, since VEGF is ")* "%) unique among the isoforms (in lacking exon 8), it may have no biological activity, but may affect the function of other well characterized VEGF species by the formation of heterodimers. Alternatively, it may have a more complex role. There are clear precedents for such splice variants with potential function-modulatory capacity. The oestrogen receptor is produced in a number of splice variants [26]. In addition, a heterogeneity of the expression of oestrogen receptor levels, size of protein produced and response to oestrogen have been described in various cell types within one tissue, akin to the heterogeneity we describe in VEGF isoform and receptor expression. Furthermore, a truncated splice variant, TERP-1 (‘ truncated estogen receptor product ’), has been shown to be biologically active. It acts as a transcriptional stimulant to full-length products, thereby increasing the sensitivity of some cells to oestrogen [26]. p73 (a structural and functional homologue of the transcription factor p53) also has multiple splice variants (known as p73α, p73β, p73γ and p73δ), which act as homo- or hetero-dimers. Similar to VEGF , the splicing "%) of p73δ demonstrates a shift from the original reading frame and a premature stop codon. This form of p73 is less efficient at activating transcription, and may modulate the other isoforms through the formation of heterodimers [27]. A number of glomeruli expressed one form of VEGF receptor only. Of the seven patterns of VEGF-receptor expression identified, the most frequent was the expression of all three receptors. Akin to the situation in the placenta, the co-dominant VEGF receptor in normal

glomeruli is sFlt, which inactivates VEGF. These data suggest, therefore, that in health a degree of glomerular VEGF autoregulation is the norm. Nested RT-PCR is a very sensitive technique, more so than in situ hybridization, RNase protection or Northern analysis. We demonstrated the presence of GAPDH bands in all glomeruli, but in 24 % of glomeruli we were unable to visualize nested RT-PCR products for any of the VEGF receptors. However, binding studies with labelled VEGF have demonstrated VEGF binding in all glomeruli [10]. We presume, therefore, that despite good evidence of VEGF receptor protein in all glomeruli, our results simply reflect a level of VEGF-receptor message below the detection sensitivity of the assay. There were no convincing differences in the patterns of VEGF-isoform or VEGF-receptor expression between subcapsular and juxta-medullary glomeruli. Furthermore, there did not appear to be any fixed relationship between the patterns of VEGF-isoform and VEGFreceptor expression in individual glomeruli. In summary, the relative expression of VEGF isoforms and VEGF receptors by individual glomeruli is complex and variable. This should be borne in mind in the future study of glomeruli from diseased kidneys. Our data are of necessity descriptive. The functional significance of the variable patterns of expression of VEGF isoforms and VEGF receptors remains to be determined, as do the factors responsible for their regulation. Furthermore, additional studies will be required in order to assess the potential functional capacity of VEGF , which may be "%) biologically active directly or via its influence on other more potent isoforms, adding a further layer to the complex biology of this molecule.

ACKNOWLEDGMENTS This work was supported by the National Kidney Research Fund (grants R32\2\96 and R35\1\98) and the Southmead Hospital Research Foundation (RF 70). We are grateful to the surgeons of the Department of Urology, Southmead Hospital, for making nephrectomy tissue available. S.J.H. is supported by Wellcome Trust Grant 057936\Z\99\Z.

REFERENCES 1 Williams, J. D. and Coles, G. A. (1994) Proteinuria : a direct cause of renal morbidity. Kidney Int. 45, 443–450 2 Ferrara, N. and Davis-Smith, T. (1997) The biology of vascular endothelial growth factor. Endocr. Rev. 18, 4–25 3 Houck, K. A., Ferrara, N., Winer, J., Cachianes, G., Li, B. and Leung, D. W. (1991) The vascular endothelial growth factor family : identification of a fourth molecular species and characterization of alternative splicing of RNA. Mol. Endocrinol. 5, 1806–1814 4 Poltorak, Z., Cohen, T., Sivan, R. et al. (1997) VEGF , a "%& secreted vascular endothelial growth factor isoform that binds to extracellular matrix. J. Biol. Chem. 272, 7151–7158


311

312


5 Simon, E., Gro$ ne, H.-J., Johren, O. et al. (1995) Expression of vascular endothelial growth factor and its receptors in human renal ontogenesis and in adult kidney. Am. J. Physiol. 268, 240–250 6 Brown, L. F., Berse, B., Tognazzi, K. et al. (1992) Vascular endothelial growth factor mRNA and protein expression in human kidney. Kidney Int. 42, 1457–1461 7 Park, J. E., Keller, G.-A. and Ferrara, N. (1993) The vascular endothelial growth factor isoforms : differential deposition into the subepithelial extracellular matrix and bioactivity of ECM-bound VEGF. Mol. Biol. Cell 4, 1317–1326 8 Keyt, B., Berleau, L., Nguyen, H. et al. (1996) The carboxy-terminal domain (111–165) of VEGF is critical for mitogenic potency. J. Biol. Chem. 271, 7788–7795 9 Plouet, J., Moro, F., Bertagnolli, S. et al. (1997) Extracellular cleavage of the vascular endothelial growth factor 189-amino-acid form by urokinase is required for its mitogenic effect. J. Biol. Chem. 272, 13390–13396 10 Simon, M., Wolfgang, R., Hornig, C. et al. (1998) Receptors of vascular endothelial growth factor\vascular permeability factor in fetal and adult human kidney : localization and ["#&I]VEGF binding sites. J. Am. Soc. Nephrol. 9, 1032–1044 11 Waltenberger, J., Claesson-Welsh, L., Siegbahn, A., Shibuya, M. and Heldin, C.-H. (1994) Differential signal transduction properties of KDR and Flt-1, two receptors for vascular endothelial growth factor. J. Biol. Chem. 269, 26988–26995 12 Senger, D. R., Connelly, D. T., Van de Water, L., Feder, J. and Dvorak, H. F. (1990) Purification and N-terminal amino acid sequence of guinea-pig tumor-secreted vascular endothelial growth factor. Cancer Res. 50, 1774–1778 13 Roberts, W. G. and Palade, G. E. (1995) Increased microvascular permeability and endothelial fenestration induced by vascular endothelial growth factor. J. Cell Sci. 108, 2369–2379 14 Gro$ ne, H.-J. (1995) Angiogenesis and vascular endothelial growth factor (VEGF) : is it relevant in renal patients ? Nephrol. Dial. Transplant. 10, 761–763 15 Bicknell, G. R., Shaw, J. A., Pringle, J. H. and Furness, P. N. (1996) Amplification of specific mRNA from a single human renal glomerulus, with an approach to the separation of epithelial cell mRNA. J. Pathol. 180, 188–193 16 Thomas, S. E., Anderson, S., Gordon, K. L., Oyama, T. T., Shankland, S. T. and Johnson, R. J. (1998)

17 18

19 20

21

22 23

24 25

26

27

Tubulointerstitial disease in aging : evidence for underlying peritubular capillary damage, a potential role for renal ischaemia. J. Am. Soc. Nephrol. 9, 231–242 Pratschke, J., Kusaka, M., Wilhelm, M. J. and Tilney, N. L. (1998) Chronic rejection : effect of brain death and ischaemia\reperfusion. Graft 1 (Suppl. II), 34–36 Harper, S. J., Pringle, J. H., Gillies, A. et al. (1992) Simultaneous in situ hybridization of native mRNA and immunoglobulin detection by conventional immunofluorescence in paraffin embedded sections. J. Clin. Pathol. 45, 114–119 Harper, S. J., Pringle, J. H., Wicks, A. C. B. et al. (1994) Expression of J chain mRNA in duodenal IgA plasma cells in IgA nephropathy. Kidney Int. 45, 836–844 Harper, S. J., Allen, A. C., Be! ne! , M.-C. et al. (1995) Increased dimeric IgA producing B cells in the tonsils in IgA nephropathy determined by in situ hybridisation for J chain mRNA. Clin. Exp. Immunol. 101, 442–448 Harper, S. J., Allen A. C., Pringle, J. H. and Feehally, J. (1996) Increased dimeric producing B cells in the bone marrow in IgA nephropathy determined by in situ hybridisation for J chain mRNA. J. Clin. Pathol. 49, 38–42 Jones, P. H., Harper, S. J. and Watts, F. (1995) Stem cell patterning and fate in human epidermis. Cell 80, 83–93 Harper, S. J., Bailey, E., McKeen, C. et al. (1997) A comparative study of digoxigenin, 2,4-dinitrophenyl and alkaline phosphatase as deoxyoligonucleotide labels in non-isotopic in situ hybridisation. J. Clin. Pathol. 50, 686–690 Bailey, E., Harper, S. J., Pringle, J. H. et al. (1998) Visceral glomerular epithelial cell DNA synthesis in experimental and human membranous disease. Exp. Nephrol. 6, 352–358 Charnock-Jones, D. S., Sharkey, A. M., Raiput-Williams, J. et al. (1993) Identification and localisation of alternatively spliced mRNAs for vascular endothelial growth factor in human uterus and estrogen regulation in endometrial carcimona cell lines. Biol. Reprod. 48, 1120–1128 Friend, K. E., Resnick, E. M., Ang, L. W. and Shupnik, M. A. (1997) Specific modulation of estrogen receptor mRNA isoforms in rat pituitary throughout estrous cycle and in response to steroid hormones. Mol. Cell. Endocrinol. 131, 147–155 De Laurenzi, V., Costannzo, A., Barcaroli, D. et al. (1998) Two new p73 splice variants, γ and δ, with differential transcription activity. J. Exp. Med. 188, 1763–1768 Received 5 January 1999/13 April 1999; accepted 25 May 1999
