Two Functional Forms of Vascular Endothelial Growth Factor Receptor-2/Flk-1 mRNA Are Expressed in Normal Rat Retina*

Vascular endothelial growth factor (VEGF) is an important mediator of ocular neovascularization by exerting its endothelial specific mitogenic effects through high affinity tyrosine kinase receptors. By screening a rat retina cDNA library, we have isolated a clone encoding the full-length prototypic form of the rat VEGF receptor-2/Flk-1, as well as a short form of the mRNA that encodes the complete seven N-terminal immunoglobulin-like extracellular ligand-binding domains, transmembrane region, NH2-terminal half of the intracellular kinase domain, and kinase insert domain but does not encode the COOH-terminal half of the intracellular kinase domain and carboxyl-terminal region. Both forms of mRNA are detected in rat retina, although the short form is expressed at a lower level. VEGF induced a biphasic increase of cytoplasmic calcium with both forms in HK 293 transfected cells, indicating that both forms of the VEGF receptor-2/Flk-1 are functional and that the COOH-terminal half of the intracellular kinase domain and carboxyl region of VEGF receptor-2/Flk-1 are not strictly necessary for either ligand binding or this biological activity.

Recent studies have provided evidence for a correlation of VEGF expression with retinal neovascularization in experimental models (14 -17). Elevated levels of VEGF have been found in the vitreous of patients with proliferative diabetic retinopathy (18). It has also been demonstrated that several types of cultured retinal cells secrete VEGF and that hypoxia stimulates VEGF production (19 -21). VEGF levels also increase in a primate model of iris neovascularization (22).
Although several studies have shown that VEGF is expressed and regulated in retinal cells, very little is known about the expression, regulation, and signal transduction mechanisms of the VEGF receptors in neural retina. The cDNAs for the VEGFR-2/Flk-1 have been cloned and characterized from humans (9,11) and mice (8,10,23). However, the full-length cDNA sequence for the VEGFR-2/Flk-1 cloned from rat has not been reported.
In the current study, we have examined the expression pattern of VEGF receptor mRNA in different rat tissues. By isolating and characterizing the full-length cDNA sequences for VEGFR-2/Flk-1 from rat retina, we have found two forms of the receptor: a long form that is similar to the previously reported sequences for mouse and human, and a short form that has part of the intracellular tyrosine kinase domain deleted from the carboxyl terminus. Both forms of the receptor were expressed in mammalian cells and responded to VEGF with a rapid mobilization of calcium.

EXPERIMENTAL PROCEDURES
Probe Preparation by RT-PCR-VEGFR-2/Flk-1 primers were designed from the conserved amino acid sequences ANEGELKT and DSITSSQ, located in the intracellular amino-terminal half of the tyrosine kinase domain and kinase insert region of mouse Flk-1 (10). The primer sequences used were MFlk2575S (5Ј-GCC AAT GAA GGG GAA CTG AAG AC-3Ј) and MFlk3111AS (5Ј-CTG GCT GCT GGT TAT GCT GTC-3Ј). A ϳ500-bp RT-PCR product was amplified from 2 g of rat retina total RNA and cloned into pCRII vector (Invitrogen, San Diego, CA). The cloned PCR product was sequenced on both strands by the chain termination method (24) using T7 and SP6 primers to confirm that the fragment was related to the VEGFR-2/Flk-1. This RT-PCR product was then used for cDNA library screening, Northern and Southern blot analysis, and in situ hybridization.
Northern Blot Analysis-To examine the expression of VEGFR-2 from various tissues, total RNA was isolated from adult rat brain, heart, kidney, liver, lung, skeletal muscle, retina, and spleen using the guanidinium thiocyanate acid-phenol method (25). Poly(A ϩ ) RNA was prepared using Poly(A)Ttract mRNA Isolation System (Promega, Madison, WI). Five g of poly(A ϩ ) RNA from each tissue was fractionated on 1% agarose-formaldehyde gels (26) and transferred to NYTRAN membranes (Schleicher & Schuell). The DNA probes were as follows: for rat VEGFR-2/Flk-1, a 537-bp cDNA fragment generated from rat retina total RNA by RT-PCR with primers MFLK2575S and MFLK3111AS as described above; for rat ␤-actin, a ϳ700-bp fragment generated by RT-PCR with rat ␤-actin amplimer set (CLONTECH, Palo Alto, CA). Probes were radiolabeled with [␣-32 P]dCTP using a random primer * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Screening of Rat Retina cDNA Library-The 537-bp rat VEGFR-2/ Flk-1 cDNA fragment generated as described above was radiolabeled with [␣-32 P]dCTP using a random primer DNA labeling kit (Stratagene, La Jolla, CA) and used as a probe to screen an oligo(dT)-primed rat retina cDNA library in the vector ZAP (Stratagene, La Jolla, CA) using standard procedures (26). Secondary and tertiary screening were performed until well isolated positive plaques were obtained. The pBluescript phagemids with target insert were rescued from ZAP vector by in vivo excision using ExAssist/SOLR system (Stratagene, La Jolla, CA). The phagemid DNAs were isolated, and the size of the cloned inserts was determined by restriction digestion followed by agarose gel electrophoresis. DNA was sequenced from both directions with successive primers. Sequence analysis was performed with Hitachi MacDNAsis software (National Biosciences, Plymouth, MN).
5Ј-Rapid Amplification of cDNA Ends (5Ј-RACE)-Since the VEGFR-2/Flk-1 cDNA inserts that were isolated from rat retina cDNA library lacked 5Ј-ends, we used the 5Ј-RACE (28) to determine the 5Ј-end sequence by using the Marathon cDNA amplification kit (CLONTECH). First-strand cDNA was synthesized from 1 g of rat retina poly(A ϩ ) RNA with an oligo(dT) primer and Moloney murine leukemia virus reverse transcriptase. Second strand cDNA synthesis was performed with a mixture of E. coli DNA polymerase I, RNase H, and E. coli DNA ligase. Following the creation of blunt ends with T4 DNA polymerase, the double-stranded cDNA was ligated to the Marathon cDNA adaptor by T4 DNA ligase. The 5Ј-RACE reaction was performed with the Marathon adaptor primer and gene-specific antisense primer RRFlk1996AS, 5Ј-ATG TGG ACC GAT GTT GCC TGT GAG C-3Ј that was designed from the 5Ј-end sequence of the isolated cDNA insert. Klen Taq DNA polymerase mix (CLONTECH), which provides proofreading activity, was used for the PCR reactions. The 5Ј-RACE PCR product was cloned into pCRII (Invitrogen) and characterized by DNA sequencing from both directions with successive primers.
Southern Blot Analysis-To confirm the existence of mRNA encoding the short form of VEGFR-2/Flk-1 in the rat retina, Marathon cDNA from rat retina (described above) was amplified with primers RRFlk2610S (5Ј-GCC AAT GAA GGG GAA CTG AAG ACA-3Ј) and RRFLK5155AS (5Ј-CCA GGG CAG ACA CAA GTG GGT AT-3Ј). The PCR product was electrophoresed in an 1% agarose gel and transferred onto a NYTRAN membrane. The blot was hybridized with a 537-bp rat VEGFR-2/Flk-1 cDNA probe as described for the Northern blot analysis.
In Situ Hybridization-For in situ hybridization studies, eyes from normal adult rats were enucleated, fixed in 4% paraformaldehyde at 4°C overnight, and embedded in paraffin. Serial 5-m sections of the whole eyes were placed on Superfrost microscope slides (Fisher), and the slides were stored at 4°C. Tissue sections were prehybridized and hybridized as described previously (29). The 537-bp cDNA fragment (as described above) that is specific to both long and short forms of the VEGFR-2/Flk-1 and a long form-specific 324-bp VEGFR-2/Flk-1 cDNA fragment amplified with primers RRFlk3608S (5Ј-AAT GCG GGC TCC TGA CTA CAC CA-3Ј) and RRFlk3903AS (5Ј-CGG CTT TTT CGC TTG CTG TTC TG-3Ј) were used as the templates to synthesize the RNA probes. Single-stranded antisense and sense 35 S-labeled RNA probes for both cDNA fragments were prepared by in vitro transcription with [ 35 S]UTP in the presence of either T7 or SP6 RNA polymerase corresponding to the promoters flanking the insert sequences in pCRII vector. Following hybridization, sections were immersed in LM-1 Hypercoat nuclear emulsion (Amersham Corp.) and exposed for 15 days at 4°C. Hybridization with labeled sense RNA probes served as controls for nonspecific binding. Slides were developed, counterstained with hematotoxylin and eosin, and photographed with both light and dark field microscopy.
Cloning of Full-length Coding Sequences and in Vitro Translation-Two pairs of primers were designed to clone the full-length coding sequences for both long and short forms of rat retina VEGFR-2/Flk-1. For the long form, the primers are as follows: RRFlk237S/XhoI, 5Ј-GCC GCT CGA GGC GAG GTG CAG GAT GGA GAG CAG GG-3Ј and RRFlkC4291AS/XbaI, 5Ј-GCC CTC TAG AAG TTG GGG GCG GGG ACG GGA CAG GG-3Ј. For the short form, the primers are as follows: RRFlk237S/XhoI and RRFlk3168AS/XbaI, 5Ј-GCC CTC TAG ACG TCA CGG AGG GAT TTC TCT TCC AC-3Ј. The XhoI or XbaI restriction site (underlined sequence) was included in the 5Ј-end of each primer to facilitate the subcloning of the amplified cDNA into plasmid vector. The Marathon cDNA from rat retina (described above) was amplified with each pair of primers using Klen Taq DNA polymerase mix (CLON-TECH). Following the digestion with XhoI and XbaI, the PCR products were cloned into the mammalian expression vector pcDNA3 (Invitrogen). The constructs (RRFlk-L/pcDNA3 and RRFlk-S/pcDNA3) were sequenced to confirm the authenticity of the reading frame. One g of plasmid DNA from each construct was used for the in vitro transcription and translation with the TNT T7 polymerase coupled reticulocyte lysate system (Promega) in the presence of [ 35 S]methionine. The translation products were analyzed by 7.5% SDS-polyacrylamide gel electrophoresis and visualized by autoradiography using a PhosphorImager (Molecular Dynamics).
Mammalian Cell Expression-HK 293 cells (human embryonic kidney cells) (30) were transfected with 20 g of plasmid DNA from each construct and pcDNA3 using a calcium phosphate transfection kit (Invitrogen). The cells were grown for 48 h in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal serum and 1% penicillin and streptomycin. The VEGFR-2/Flk-1 mRNA production was monitored by Northern blot analysis. The RNA samples isolated from transfected HK 293 cells were treated with RNase-free DNase I to eliminate the contamination of plasmid DNA.
Image Analysis of Intracellular Calcium-The transfected HK 293 cells grown for 48 h were incubated with 3 M Fura-2/AM at 37°C for 30 min and then placed on a coverslip precoated with Cell-Tak (3.5 g/cm 2 ), which served as an observation port in a perfusion chamber of 1-ml volume. The chamber was continuously perfused with Ringer's solution (140 mM NaCl, 1.2 mM MgSO 4 , 1.0 mM CaCl 2 , 15 nM HEPES at pH 7.4, 11 mM somatostatin-glucose, and 0.1% bovine serum albumin) at a rate of 2 ml/min. Human recombinant VEGF 165 (1 nM) (Sigma) or VEGF 121 (1 nM) (R & D Systems, Minneapolis, MN) was added into the chamber while the cells were observed by a Nikon Fluo 40 objective with a Zeiss Axiovert 135 microscope in the epifluorescence mode using a long pass filter cut-off at 510 nm and a dichroic at 385 nm. Alternating wavelength excitation of 340 and 380 nm was provided by a motorized filter wheel. The image pairs were monitored by a COHU ISIT camera, captured (less than 1 s apart) under the control of Image-1 F1 software (Universal Imaging, West Chester, PA) and stored as digitized 256-gray level images. Data are expressed as the ratio (340/380 nm) of brightness levels in chosen fields. The calcium concentrations were determined as described previously (31,32).

Expression of VEGFR-2/Flk-1 in Adult Rat Tissues-RT-
PCR was used to amplify a DNA fragment corresponding to the kinase insert domain of VEGFR-2/Flk-1 from rat retina total RNA with a pair of primers designed from the mouse Flk-1 sequence (10). The 537-bp PCR product was subcloned into a plasmid vector pCRII and sequenced. Computer-assisted comparison with mouse Flk-1 DNA sequence (10) revealed a 93% identity, which suggests that this DNA fragment represents the rat version of the VEGFR-2/Flk-1 gene. This DNA fragment was used to examine the expression pattern of VEGFR-2/Flk-1 in normal adult rat tissues by Northern blot analysis (Fig. 1). The VEGFR-2/Flk-1 mRNA was detected in several tissues including brain, heart, kidney, liver, lung, skeletal muscle, retina, and spleen with a major band at about 6 kb (Fig. 1A). The expression level for most tissues was low. However, the mRNA expression of VEGFR-2/Flk-1 in retina was relatively abundant, at a level as high as that in lung, which has been reported as a tissue with high expression of this gene (10,23). A smaller band was also detected in retina and lung at about 4 kb (Fig. 1A).
Cloning and Characterization of Rat VEGFR-2/Flk-1 cDNA-Three independent positive clones were found using a 537-bp rat VEGFR-2 cDNA fragment as probe to screen 5.4 ϫ 10 5 recombinants from a rat retina cDNA library. They are designated as RRFlk1429, RRFlk3600, and RRFlk2053, respectively, based on their size (Fig. 2). Partial DNA sequencing indicated that the three clones shared the same 3Ј-end of about 800 bp containing poly(A ϩ ), but none of the clones reached the complete 5Ј-end of the Flk-1 cDNA sequence when compared with the cDNA sequence data from other species (8 -11). To determine the 5Ј-end sequence of rat retina VEGFR-2/Flk-1 cDNA, 5Ј-RACE was performed with the antisense primer RRFlk1996AS. From this, a 2020-bp cDNA fragment was generated and cloned. As a result of sequencing and comparing the overlapping clones, the complete cDNA sequence for rat retina VEGFR-2/Flk-1 was found to contain 5892 nucleotides (Gen-Bank TM accession number U93306 ). An open reading frame of 4029 nucleotides is flanked by 248 nucleotides of 5Ј-untranslated sequence and 1615 nucleotides of 3Ј-untranslated sequence. The presence of a consensus polyadenylation signal at nucleotide 5858 suggests that the 3Ј-noncoding region is complete. The first ATG codon found in this open reading frame is in good agreement with the consensus sequence for translation initiation (33). The deduced 1343-amino acid polypeptide is 24 and 11 residues shorter than mouse and human VEGFR-2/ Flk-1 (9 -11) (Fig. 3A). Overall, it shows 69.2 and 51.2% iden-tity in nucleotide sequence and 95 and 87% identity in deduced amino acid sequence to its mouse and human homologues. Fig.  3B shows that there is a high degree of sequence conservation in the structure of VEGFR-2/Flk-1 at the amino acid level among rats, humans, and mice.
Cloning of Short Form of the VEGF Receptor-2/Flk-1-On the basis of sequence analysis, one of the three clones isolated from a retina cDNA library, RRFlk3600, appears to be derived from the prototypic form of VEGFR-2/Flk-1, whereas the other two clones encode a short form of the VEGFR-2/Flk-1 with a truncated COOH terminus. The two truncated VEGFR-2/Flk-1 clones are also independent of each other because they terminate at different 5Ј-sites. The shorter 1.4-kb clone (RRFlk1429) begins at nucleotide 2588, whereas the longer 2.0-kb clone (RRFlk2053) begins at nucleotide 1964 of the full-length prototypic form of the rat VEGFR-2/Flk-1 sequence (GenBank TM accession number U93306). The clone RRFlk2053 spans the coding sequence from the sixth immunoglobulin-like domain to the 3Ј-poly(A ϩ ) end. The cDNA coding sequence of both of the truncated VEGFR-2/Flk-1 clones are identical to the full-length prototypic form up to nucleotide 3222, at which the open reading frame is terminated by TGA (Fig. 4A). The complete cDNA sequence for the short form of rat VEGFR-2/Flk-1 was obtained by lining up the overlapping 2020-bp 5Ј-RACE product (described above) and clone RRFlk2053, which contains 4016 nucleotides (GenBank TM accession number U93307 ). An open reading frame of 2973 nucleotides encoding 991 amino acids is initiated by an ATG codon at nucleotide 249. A deletion of 1876 nucleotides (1082 nucleotides for the coding sequence and 794 nucleotides for the noncoding sequence) from the rat prototypic form of VEGFR-2/Flk-1 mRNA results in the deletion of 352 amino acids from the COOH-terminal half of the intracellular kinase domain and carboxyl-terminal region in the short form of rat VEGFR-2/Flk-1 (Fig. 4B).
The expression of both long and short forms of VEGFR-2/ Flk-1 in rat retina was confirmed by RT-PCR with primers (RRFlk2610S and RRFlk5155AS) spanning the deletion site and common to both forms of the receptor, followed by Southern blot analysis with a 537-kb DNA probe corresponding to the kinase insert domain, which is present in both forms of the receptor. Two bands (2545 bp for the long form and 669 bp for the short form of VEGFR-2/Flk-1) were produced by RT-PCR using rat retina poly(A ϩ ) RNA (Fig. 4C). The Southern blot in Fig. 4D shows that both bands were specific to VEGFR-2/Flk-1. The 669-bp band was very weak in the ethidium bromidestained gel due to the low level of expression of the short form VEGFR-2/Flk-1 in rat retina, but it was easily visualized after Southern hybridization. No PCR product was detected in control reactions without RT (data not shown), eliminating the possibility of contamination by genomic DNA. the location of VEGFR-2/Flk-1 mRNA expression in rat retina, in situ hybridization was performed. Fig. 5 shows the results of a representative in situ hybridization performed in normal adult rat retina. Hybridization with an antisense probe specific to both the long and short forms of VEGFR-2/Flk-1 shows an intense signal for VEGFR-2/Flk-1 mRNA in the inner nuclear layer of retina and also some labeling over choroid (Fig. 5, A  and B). A similar distribution pattern was obtained with the antisense probe specific to only long form of VEGFR-2/Flk-1 except that less labeling was found in the boundary of inner nuclear layer and inner plexiform layer (Fig. 5, E and F), compared with the same area in Fig. 5, A and B. Hybridization with sense probes shows low background in most areas of retina (Fig. 5, C, D, G, and H), with the exception of the photoreceptor outer segments, which showed homogenous nonspecific labeling with both antisense and sense probes (Fig. 5,  B, D, F, and H), and thus the signal over this area does not imply the presence of VEGFR-2/Flk-1 transcripts. Closer examination with high power magnification in light field microscopy (Fig. 6) clearly shows that the dark grains denoting VEGFR-2/Flk-1 mRNA/probe hybrids are located in the inner nuclear layer of the retina with high density for both antisense probes (Fig. 6, A and C, down arrows). Significant labeling can be found in the choroid for both antisense probes (Fig. 6, A and   FIG. 3. Comparison of the predicted amino acid sequences of rat, mouse, and human VEGFR-2/Flk-1. A, amino acid sequence deduced from the nucleotide sequence of rat VEGFR-2/Flk-1 cDNA. The rat Flk-1 sequence of 1343 amino acid residues is compared with that of human Flk-1 (1354 residues) and rat Flk-1 (1367 residues). The Hitachi MacDNAsis software (National Biosciences, Plymouth, MN) was used to align the sequences. The program introduced gaps to achieve maximum alignment. Identical amino acids are indicated by asterisks. C, solid black arrows). Hybridization with antisense probe specific to both the long and short forms of VEGFR-2/Flk-1 also shows some grains in the border of the inner nuclear layer and inner plexiform layer (Fig. 6A, upward arrows), where very few grains were found when the section was hybridized with antisense probe specific to only the long form of VEGFR-2/Flk-1 (Fig. 6C). This confirms the observation from dark field microscopy in Fig. 5. Hybridization with either sense probe (Fig. 6, B and D) shows low background in most areas in the retina except the photoreceptor outer segments, where the homogenous nonspecific labeling was found for both antisense and sense probes (Fig. 6, A, B, C, and D).
Expression of Both Forms of VEGFR-2-The full-length coding sequences for both long and short forms of VEGFR-2/ Flk-1 generated from RT-PCR were cloned into the pcDNA3 vector, which contains a CMV early promoter region upstream of the multiple cloning site. To confirm that the open reading frame sequences present in the RRFlk-L/pcDNA3 and RRFlk-S/pcDNA3 clones code for translatable proteins, the cDNAs from each construct were in vitro transcribed using T7 polymerase and translated in rabbit reticulocyte lysate. Fig. 7 shows the translation products of the two cDNAs. The translated product from the RRFlk-L/pcDNA3 clone was a protein of ϳ150 kDa, whereas the translation product from the RRFlk-S/pcDNA3 clone was a protein of ϳ110 kDa. In the presence of microsome membrane to allow glycosylation, the molecular mass shifted for both constructs to ϳ190 and ϳ145 kDa, respectively. Multiple protein translation products with smaller molecular mass were also found for both constructs, most probably because of the in frame ATGs in the nucleotide sequences.
The RRFlk-L/pcDNA3 and RRFlk-S/pcDNA3 were transfected transiently into HK 293 cells using a calcium precipitation method. VEGFR-2/Flk-1 mRNA production was assessed by Northern blot analysis using the 537-bp VEGFR-2/Flk-1 probe (described above) after 48 h of transfection. Northern blot analysis shows that a ϳ4-kb mRNA band was found after transfection with RRFlk-L/pcDNA3, whereas a ϳ3-kb mRNA band was found after transfection with RRFlk-S/pcDNA3 (Fig.  8). These bands correspond to the expected size of open reading frames for the long and short forms of VEGFR-2/Flk-1. No signal was found from control HK 293 cells transfected with pcDNA3 alone. As a positive control, 2 g of poly(A ϩ ) RNA from rat retina was loaded on the gel, and two hybridized bands were found with sizes of ϳ6 and ϳ4 kb.
Functional Response of the Expressed Receptor-The functional expression of the receptors was examined by measurement of a calcium signal in transfected HK 293 cells after the addition of recombinant human VEGF 165 . The imaging system allowed calcium responses to be measured from several cells simultaneously. Approximately 20% of the HK 293 cells transfected with RRFlk-L/pcDNA3 responded to 1 nM VEGF 165 (Fig. 9A), whereas those HK 293 cells transfected with RRFlk-S/pcDNA3 showed slightly more cells (25%) responding (Fig. 9B). The response consisted of a rapid increase in intracellular calcium (ϳ500 nM) with a return to near base line in ϳ1 min. No response of intracellular calcium was observed in HK 293 cells transfected with pcDNA3 alone (Fig. 9C). Similar calcium responses were obtained when 1 nM VEGF 121 was used to stimulate HK 293 cells transfected with either RRFlk-L/pcDNA3 or RRFlk-S/pcDNA3 (data not shown). To further demonstrate unequivocally that the VEGF-induced re-sponses of intracellular calcium in the transfected HK 293 cells were specifically due to VEGF binding, an anti-VEGF antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was used to neutralize VEGF activity. Incubation of VEGF with excess soluble anti-VEGF antibody (10 g/ml) completely inhibited the VEGF activity upon inducing response of intracellular calcium in HK 293 cells transfected with either the long or short form of VEGFR-2/Flk-1 cDNAs (Fig. 9, A and B). DISCUSSION In the present study, we have cloned, sequenced, and expressed two variant cDNA clones for VEGFR-2/Flk-1 from a rat retina cDNA library. They code for two distinct forms of VEGFR-2/Flk-1, which represents the first report of the variant for the VEGFR-2/Flk-1. The long form, RRFlk-L, encodes the prototypic form of VEGFR-2/Flk-1 that includes a hydrophobic leader sequence following the initiator methionine; the extracellular domain containing seven Ig-like domains (the fourth of which does not contain cysteine residues); a transmembrane region; an intracellular kinase domain that is interrupted into an NH 2 -terminal half and a COOH-terminal half by a kinase insert domain; and a carboxyl-terminal region (Fig.  3A). The short form, RRFlk-S, encodes a truncated form of VEGFR-2/Flk-1, which is identical to the long form, except for a deletion of the COOH-terminal half of the kinase domain and carboxyl region in the intracellular portion.
The cDNA encoding RRFlk-S was found in two of three independent RRFlk clones, so the corresponding mRNA is neither an a artifact of cloning nor likely to be rare. The expression of the short form of VEGFR-2/Flk-1 (RRFlk-S) in rat retina was confirmed by RT-PCR with a pair of primers spanning the deletion site, followed by Southern blot with the probe corresponding to the kinase insert domain (Fig. 4). Additional evidence also came from Northern blot analysis when we increased the amount of poly(A ϩ ) RNA up to 5 g loaded on the gel, which shows two bands at ϳ6 and ϳ4 kb corresponding to mRNAs encoding the long and short forms of VEGFR-2/Flk-1, respectively, in rat lung and retina (Fig. 1A). The short form of VEGFR-2/Flk-1 is expressed at a very low level, because it was difficult to detect the ϳ4-kb band from any tissue examined in the same condition when 1 g of poly(A ϩ ) RNA was loaded on the gel (data not shown). The RRFlk-L and RRFlk-S are probably alternatively spliced variants of a single gene, because the 3Ј-terminal tail of RRFlk-S (nucleotides 3223-4016) is found to be exactly identical to the extreme 3Ј-untranslated region of RRFlk-L (nucleotides 5099 -5898). However, the splice site cannot be determined without the genomic structure data of VEGFR-2/Flk-1.
The identification of splice variants among the tyrosine kinase receptors has been increasing at a phenomenal rate. Splice variants have been identified for the Eph-related receptor tyrosine kinase Cek9 (34); fibroblast growth factor receptor, SpFGFR1 (35); neurotrophin receptor TrkB (36); the receptor for colony-stimulating factor 1, FLT3 (37); hepatocyte growth factor receptor, c-met (38); and nerve growth factor receptor, TrkA (39,40). Furthermore, Kendall and Thomas (41) reported the existence of a cDNA encoding a truncated form of VEGF receptor-1/Flt-1 lacking the seventh immunoglobulin-like domain, the transmembrane sequences, and the entire cytoplasmic domain in human umbilical vein endothelial cells. It was hypothesized that this variant might encode a soluble form of VEGF receptor that acts as a specific antagonist of VEGF. However, it is interesting to note that in the short form of VEGFR-2/Flk-1 the cytoplasmic domain of the receptor is truncated, indicating that alternative RNA splicing may play an important role in the generation of functionally divergent receptor activities.
In situ hybridization studies show that VEGFR-2/Flk-1 mRNA is highly expressed by cells in the inner nuclear layer of the retina (Figs. 5 and 6). This is consistent with a prior report that Flk-1 transcripts were expressed in a row of cells within the inner nuclear layer of retina in adult mice (42). Our in situ hybridization studies also show that VEGFR-2/Flk-1 transcripts are expressed in choroid of the retina (Figs. 5 and 6). Since the RRFlk-L shares all of the sequence information that RRFlk-S has, it is difficult to precisely distinguish the difference in location between these two forms of the VEGFR-2/Flk-1 transcript in the retina. To localize mRNA expression of the two forms of VEGFR-2/Flk-1 in the retina, we used two different antisense RNA probes for in situ hybridization; one is specific to both RRFlk-L and RRFlk-S, and the other is specific only to RRFlk-L. Comparison of the in situ hybridization results (Figs. 5 and 6) with two different probes suggests that RRFlk-S appears to be expressed by the cells in the border between the inner nuclear layer and inner plexiform layer of retina, where RRFlk-L is less expressed. However, this does not exclude the possibility that RRFlk-S is also expressed in the inner nuclear layer and choroid where RRFlk-L is expressed.
To functionally characterize the cDNA encoding for VEGFR-2/Flk-1 isolated from rat retina, we subcloned the full-length coding sequences for both RRFlk-L and RRFlk-S into the expression vector pcDNA3 and transiently expressed them in HK 293 cells. The in vitro transcription and translation (Fig. 7) clearly indicated the presence of translatable, functional coding sequences. The Northern blot analysis after 48 h of transfection (Fig. 8) indicated that both of RRFlk-L and RRFlk-S mRNA were well expressed. It has been reported that VEGF induces a rapid Ca 2ϩ entry into cultured endothelial cells from bovine aorta, human umbilical vein, and bovine adrenal cortex (43). The measurement of cytoplasmic calcium has been considered the most sensitive assay currently available for detecting VEGF activity (44), and it has been used in several studies including the biological activation of VEGF receptor-1/Flt-1 (7) and hypoxia-induced VEGF gene expression (45). We used a similar technique to examine the functional response of the expressed VEGF receptors. The addition of VEGF to the HK 293 cells induced a 6 -7-fold increase in intracellular calcium in HK 293 cells transfected with either RRFlk-L/pcDNA3 or RRFlk-S/pcDNA3 (Fig. 9, A and B). Mock transfection of HK 293 cells with pcDNA3 did not respond to VEGF (Fig. 9C). These findings show that VEGF can induce a specific functional response of both long and short forms of VEGFR-2/Flk-1 encoded by cDNAs isolated from rat retina. They also indicate that VEGF is a ligand for both receptors and that the COOHterminal half of the kinase domain and carboxyl region of the VEGF receptor-2/Flk-1 (amino acids 992-1343) are not required for this biological response. Further studies will therefore be required to determine if there is a functional difference between the two receptor variants.
The Northern blot data indicates that the VEGFR-2/Flk-1 is expressed in the retina of normal adult rat at a relatively high level compared with other tissues examined (Fig. 1). In situ hybridization in adult rat retina shows that VEGFR-2/Flk-1 is expressed in the cells of inner nuclear layer and choroid (Figs. 5 and 6). In situ hybridization in mouse retina (42) has detected the VEGFR-2/Flk-1 expression in the neural retina during mouse embryogenesis and in the retinal Muller glial cell of adult mouse. The Muller cells are radially aligned and span most of the thickness of the retina and play an important role in maintaining the extracellular homeostasis of ions, metabolites, and signaling molecules. The ligand for VEGFR-2/Flk-1, VEGF, is likewise expressed in Muller glial cells (17) at a relatively high level in the retina and has been proposed to play a role in vascular permeability and angiogenesis (44). Since VEGFR-2/Flk-1 expression appears not to be restricted to retinal endothelial cells, it is tempting to speculate that the high level of expression for VEGF and VEGFR-2/Flk-1 in the retina might reflect an alternative function for this ligand and receptor. Since VEGF is secreted by the retina under normal conditions, VEGFR-2/Flk-1 on Muller cells may function as a specific uptake mechanism and regulator of the extracellular concentration of this potent signaling protein.