Integrin Regulation by Endogenous Expression of 18-kDa Fibroblast Growth Factor-2*

The three high molecular weight (HMW) forms of fi- broblast growth factor-2 (FGF-2) have a distinct intracellular localization and differentially affect cell mobil- ity and growth compared with the fourth 18-kDa form. To characterize further the effects of the 18-kDa and HMW forms of FGF-2, we have examined their ability to modulate integrin expression. Transfected NIH 3T3 cells expressing only 18-kDa FGF-2 exhibited increased cell surface levels of (cid:97) 5 (cid:98) 1, whereas cells expressing only HMW FGF-2 exhibited cell surface (cid:97) 5 (cid:98) 1 levels similar to parental cells. When cells synthesizing 18-kDa FGF-2 were transfected with a cDNA encoding a dominant negative FGF receptor, (cid:97) 5 (cid:98) 1 cell surface levels decreased. Immunoprecipitation of biosynthetically labeled cells indicated that expression of 18-kDa FGF-2 increased the biosynthesis and rate of maturation of (cid:97) 5. Northern blot analysis showed that 18-kDa FGF-2 increases the level of the (cid:97) 5 subunit mRNA but does not affect (cid:98) 1 subunit transcript levels. Experiments utilizing luciferase re- (cid:98) 1 Integrin Subunit— Of the integrins regulated by 18-kDa FGF-2, immunoprecipitation, SDS-PAGE, and PhosphorImager analysis performed as described under “Materials and Methods.” expressing all FGF-2 forms (clone WTFGFc3), all FGF-2 forms with the dominant negative FGF receptor (WT FGF-2/DN, clone WTDNc2), 18-kDa FGF-2 (clone 43FGFc31), or 18-kDa FGF-2 with the dominant negative FGF receptor (18-kDa FGF-2/DN, clone 43DNc11) were with anti- anti- (cid:98) 1 This pattern is representative of two experiments performed that yielded similar

typic angiogenic factor involved in wound-healing processes and tumor neovascularization (3,4). The responses of cells to FGF-2 are mediated through a dual receptor system consisting of high affinity binding transmembrane receptors and lower affinity cell surface and extracellular matrix heparan sulfate proteoglycan binding sites (5)(6)(7)(8)(9)(10). Four distinct high affinity tyrosine kinase plasma membrane receptors encoded by four different genes have been described for FGF (1,11). Each of these genes encodes multiple variants derived from alternative mRNA splicing (12)(13)(14)(15). Although FGF-2 does not contain a signal sequence, it is released from cells and can act as an autocrine and/or paracrine regulator (16 -17). The mechanism for FGF-2 release remains unknown. However, it has been shown to be released independent of the endoplasmic reticulum/Golgi pathway (18).
FGF-2 is synthesized by a wide variety of cells including primary endothelial cells (5,19). FGF-2 induces angiogenesis as it increases endothelial cell proliferation, migration, and proteolytic activity (19 -22). We previously reported that exogenous FGF-2 modulates integrin expression in microvascualar endothelial cells (23). Integrins are heterodimeric receptors composed of ␣ and ␤ subunits. At present, there are 8 different ␤ and 15 different ␣ subunits that can combine to form 21 receptors with distinct ligand specificities (24). Integrins are involved in the processes of cell proliferation, motility, survival, and mesoderm induction (24 -29). Endothelial cell integrins also function together with other families of adhesion molecules during vasculogenesis, angiogenesis, inflammation, and wound healing (30,31). Treatment of endothelial cells with FGF-2 caused significant changes in the surface expression of nine different integrins (23). Thus, modulation of integrins may be one of the FGF-2-induced effects on endothelial cells during angiogenesis.
Several forms of FGF-2 are produced in vivo resulting from alternative initiation of translation either at an AUG codon or at three in-frame CUG codons 5Ј to the AUG (32,33). This results, respectively, in the synthesis of a form of 18 kDa and three high molecular weight (HMW) forms of 22,000, 22,500, and 24,000 (32)(33)(34)(35). The complete sequences of the smaller forms are contained in the larger forms. The relative amounts of the individual molecular weight forms have been reported to differ substantially among various cell lines and tissues during development, implying that the alternative codon usage is highly regulated (36 -39). It has been suggested that cis-acting elements in the FGF-2 mRNA are involved in regulating the translation of the different forms of FGF-2 at the four initiation sites (40).
The process of alternative initiation of translation has varying consequences for the ultimate fate of the different FGF-2 forms (41)(42)(43)(44). The three HMW forms of FGF-2 contain a nuclear localization sequence that concentrates the growth factor in that organelle. In contrast, 18-kDa FGF-2, which lacks a nuclear localization sequence, is primarily cytosolic.
The existence of multiple forms of FGF-2 with different subcellular localizations raises the question of whether these different species of FGF-2 have specialized functions. We approached this question by creating stably transfected NIH 3T3 cell lines that express exclusively 18-kDa FGF-2, HMW FGF-2, or all forms of FGF-2 (45,46). Both 18-kDa FGF-2 and HMW FGF-2 alone expressed at high levels transformed NIH 3T3 cells. Cells expressing only 18-kDa FGF-2 had high motility and surface-associated 18-kDa FGF-2, whereas cells expressing exclusively HMW FGF-2 had low motility and virtually no surface-associated FGF-2. FGF receptors were down-regulated in cells expressing 18-kDa FGF-2 but not in cells expressing HMW FGF-2. Cells expressing HMW FGF-2 had a reduced serum requirement for growth, but cells expressing 18-kDa FGF-2 proliferated poorly in low serum. These results showed that 18-kDa and HMW FGF-2 have both unique and shared biological activities. Expression of a dominant negative FGF receptor in cells expressing 18-kDa FGF-2 inhibited their migration and suppressed their growth in soft agar as well as their saturation density. In contrast, expression of the dominant negative receptor in cells expressing HMW FGF-2 had no effect on their growth. Thus, 18-kDa and HMW FGF-2 may mediate certain functions through distinct mechanisms. 18-kDa FGF-2 modulates cell motility and proliferation through the interaction with its cell surface receptors, whereas HMW FGF-2 appears to act as a mitogen and an inducer of anchorage-independent growth through an intracellular mechanism.
Our aim in the present study was to characterize additional functional differences between 18-kDa and HMW FGF-2. To this end, we chose to analyze the regulation of integrin expression by FGF-2 in NIH 3T3 cells expressing the various FGF-2 forms. We have determined which form of FGF-2 modulates integrin expression in these cells as well as the mechanism mediating this effect. We found that endogenous expression of 18-kDa FGF-2 modulates ␤1 integrin expression in NIH 3T3 cells, although endogenous expression of HMW FGF-2 does not. The induced expression of ␣5␤1 on the cell surface mediated by endogenous expression of 18-kDa FGF-2 is a result of the combined effects of an increased level of transcript coding for the ␣5 subunit and an increased rate of processing of the ␣5 and ␤1 subunits.

Cells
Secondary Transfection of Cells with 18-kDa FGF-2 or HMW FGF-2 cDNAs-NIH 3T3 cells transfected with the Zip-neo vector containing HMW FGF-2 cDNA were subsequently retransfected either with 18-kDa FGF-2 or with HMW FGF-2 cDNA inserts in the Zip-neo vector plus the pCEP4 vector containing a hygromycin resistance gene (kindly provided by Dr. C. Basilico, New York University Medical Center, New York) at a molar ratio of 8:1. Hygromycin-resistant clones were selected in DME containing 10% FCS, 200 g/ml hygromycin B, and 250 g/ml Geneticin. Secondary transfectants were characterized by Western blotting of cell extracts with anti-FGF-2 antibodies. The cells used were HMW clone transfected with 18-kDa FGF-2 cDNA, 365/43NC33; HMW clones transfected with HMW FGF-2 cDNA, 365/365FGFc14, 365/ 365FGFc38; control clone transfected with hygromycin-resistant gene alone, 365FGFHc3.
Transfection with a Dominant Negative FGF Receptor cDNA-NIH 3T3 cells transfected with the Zip-neo vector containing either HMW, 18 kDa, or WT FGF-2 cDNAs were cotransfected with pRK5 containing a 1.3-kilobase insert of a human bek (FGF receptor 2) cDNA that lacks the C-terminal tyrosine kinase domain (dominant negative FGF receptor; kindly provided by Dr. J. Schlessinger, New York University Medical Center) and the pCEP4 vector. Hygromycin-resistant clones were selected in DME containing 10% FCS, 200 g/ml hygromycin, and 250 g/ml Geneticin. Resistant clones were tested for high affinity FGF-2 receptors according to Moscatelli (5) and by cross-linking to cell surface receptors with 125 I-FGF-2. The cells used in this study were HMW FGF-2 clones transfected with dominant negative FGF receptor cDNA, 365DNc5, 365DNc7; 18-kDa FGF-2 clone transfected with dominant negative FGF receptor cDNA, 43DNc11; WT FGF-2 clone, WTDNc2.

Metabolic Labeling
NIH 3T3 cells plated at subconfluence on 10-or 3.5-cm dishes were preincubated at 37°C for 1 h with Met/Cys-free minimum Eagle's medium and incubated at 37°C for 16 h in Met/Cys-free minimum Eagle's medium supplemented with 2% FCS and 100 Ci/ml 35 S-Translabel. For pulse-chase experiments, cells were pulse-labeled with 500 Ci/ml 35 S-Trans-label for 1 h. Cells were washed immediately with cold PBS or chased in DME plus 5% FCS. After cells were washed, they were extracted for 20 min with cold 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 1% Triton X-100, 150 mM NaCl, and 50 mM Tris-HCl, pH 7.4 (RIPA buffer), containing protease inhibitors, removed from the dish with a Costar cell scraper, and sonicated.

Immunoprecipitation
The 125 I-and 35 S-labeled cell lysates were immunoprecipitated as described previously (23). Protein band intensity was quantitated by PhosphorImager analysis (Molecular Dynamics, Sunnyvale, CA).

Northern Blot Analysis
Northern blot analysis was performed with total cellular RNA isolated by single-step guanidinium thiocyanate/phenol/chloroform extraction using Trizol reagent. Mouse ␤1 and mouse ␣5 cDNA probes were kindly provided by Dr. H. S. Baldwin (Wistar Institute, Philadelphia), and human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe was kindly provided by Dr. A. M. Curatola (New York University Medical Center). Radioactive probes were made by using an oligonucleotide random priming procedure (Boehringer Mannheim) and [ 32 P]dCTP (3000 Ci/mmol). The membranes were washed four times with 2 ϫ SSC, 0.1% SDS for 10 min at room temperature and 1 ϫ SSC, 0.1% SDS for 2 min at 55°C. Blots were exposed to XAR-5 film and bands on the blots were quantitated by PhosphorImager analysis.

Luciferase Plasmid Construct and Luciferase Assay
To create the plasmid p␣5-926LUC, the region of the human ␣5 gene from Ϫ926 to ϩ23 bp was amplified using PCR with the template Pst-1A as described previously (50). The sequence Ϫ926 to Ϫ665 is available from GenBank/EMBL under accession number U48214. The sequence of the 5Ј oligonucleotide corresponded to position Ϫ926 to Ϫ901 of the ␣5 gene containing a PstI restriction enzyme site on its 5Ј end (underlined), d(AGCTCTGCAGGTTTACACCGATTAGGAGCTG-AAGGT)-3Ј. The 3Ј oligonucleotide corresponded to the reverse complement of the ␣5 sequence from ϩ4 to ϩ23 with a SalI restriction enzyme site on its 3Ј end. The resulting PCR product was digested with SalI and PstI and cloned into the plasmid SK Bluescript (Stratagene, La Jolla, CA). The insert was sequenced to ensure that no mutations occurred during PCR. To create the ␣5-luciferase expression vector, this plasmid was digested with the enzymes SmaI (contained immediately 5Ј of the PstI site in the SK plasmid) and SalI, and the fragment was gel-purified and cloned into the SmaI and XhoI sites in the vector pGL2-Basic (Promega, Madison, WI).
NIH 3T3 cells were plated at subconfluence on 10-cm tissue culture dishes in DME containing FCS (10%). 18 h later, cells were transiently transfected using lipofectamine reagent with expression construct p␣5-926LUC or with a control construct, pLUC, containing a promoterless luciferase gene and a ␤-galactosidase (␤-gal) cotransfection control plasmid. 48 h after transfection, cells were washed once with PBS, scraped, and pelleted by centrifugation. Cells were resuspended in 100 l of 0.25 M Tris-Cl, pH 7.8, and lysed by three freeze-thaw cycles. ␤-Gal activity was measured as described previously (51). Results of ␤-gal activity were normalized according to protein concentration in each cell extract determined by the bicinchroninic acid method (Pierce). 35 l of lysates were transferred to a Microlight 1 96-well plate (Dynatech Laboratories Inc., Chantilly, VA). Luciferase activity was assayed using a ML3000 Microtiter Plate Luminometer (Dynatech Laboratories Inc.) that dis-

Analysis of mRNA Stability
The stability of ␣5 mRNA was determined by adding the transcription inhibitor actinomycin D (5 g/ml) to NIH 3T3 cells. Total RNA was isolated after various times of incubation with the inhibitor and subjected to Northern blot hybridization as described above.

Endogenous 18-kDa FGF-2 but Not HMW FGF-2 Increases
Surface ␤1 Integrin Levels-Previously we demonstrated that integrin expression in cultured endothelial cells could be modulated by exogenous FGF-2 (23). In order to explore the effects of endogenous FGF-2 expression on integrins, we chose to examine integrin levels in NIH 3T3 cells as these cells expressing different forms of FGF-2 are a useful model to study FGF-2 activity (45,46). To establish that FGF-2 was capable of modulating integrin levels in 3T3 cells, control cells were treated for 48 h with and without FGF-2 (15 ng/ml), cell surface proteins were iodinated, and cell extracts were immunoprecipitated with anti-integrin antibodies. In agreement with previous observations demonstrating the induction of ␤1 integrins in capillary endothelial cells, addition of FGF-2 increased ␤1 integrin cell surface levels as visualized using anti-␤1 antibodies. Specifically, immunoprecipitation with anti-␣5 and anti-␣6 antibodies indicated that the two most abundant integrins, ␣5␤1 and ␣6␤1, increased by 3.2-and 3.5-fold, respectively ( Fig. 1A and Table I). Thus, exogenous FGF-2 stimulates integrin expression in NIH 3T3 fibroblasts as well as in capillary endothelial cells.
To test whether endogenous FGF-2 synthesis also affects integrin expression, we examined the cell surface integrins in NIH 3T3 cells stably transfected with the FGF-2 cDNA encoding all forms of FGF-2 (wild type (WT) FGF-2). Immunoprecipitation of surface-labeled cells with anti-␤1 antibody indicated that overall cell surface ␤1 integrins were increased by 3.1-fold compared with control cells (Fig. 1A). Specifically, immunoprecipitation with anti-␣5 and anti-␣6 antibodies revealed that the ␣5␤1 and ␣6␤1 integrins were enhanced by 3.1-and 2.3-fold, respectively. In addition, we checked whether the levels of other ␤1 integrins were also modulated by endogenous FGF-2. The results obtained by immunoprecipitation of cells synthesizing WT FGF-2 using anti-␣3 and anti-␣v antibodies indicated that ␣3␤1 levels were decreased 2-fold, and ␣v␤1 levels were not changed compared with control cells (Fig. 1B). Other ␤1 integrins were not detected in these cells by our analysis. Our results thus show that endogenous FGF-2 differentially modulates cell surface levels of specific ␤1 integrins in NIH 3T3

cells.
To determine which form(s) of FGF-2 expressed by the WT FGF-2 transfected cells modulate(s) integrin levels, we examined the cell surface integrin content in NIH 3T3 cells stably transfected with cDNAs encoding different species of FGF-2. By Western blot analysis, we have demonstrated that comparable, steady-state levels of FGF-2 are produced in the cell clones transfected with either 18-kDa FGF-2 or HMW FGF-2 cDNA (data not shown). A pulse-chase experiment with metabolically labeled WT FGF-2 transfected cells has shown that the rates of synthesis and degradation of the different FGF-2 forms are also similar. 2 Therefore, any differences in biological responses of 18-kDa and HMW FGF-2 cannot be due to differences in levels or rates of synthesis or degradation of FGF-2 forms among the various cell clones used in this study.
To simplify the analysis of integrins regulated by FGF-2, we focused on ␣5␤1 levels. Fig. 2A illustrates that NIH 3T3 cells synthesizing 18-kDa FGF-2 alone have 2-3-fold increased cell surface ␤1 integrin levels. Specifically, immunoprecipitation of cells with antibodies against the ␣5 and ␤1 subunits revealed that the amount of ␣5␤1 integrin was increased compared with control levels. In contrast to the results obtained with cells synthesizing 18-kDa FGF-2, immunoprecipitation of cell extracts from NIH 3T3 cells expressing only HMW FGF-2 with antibodies to ␤1 and ␣5 subunits indicated that no change occurs in the level of ␤1 integrins. Thus, endogenous 18-kDa FGF-2 increases the ␤1 surface integrins on NIH 3T3 cells, but HMW FGF-2 does not. Similar results were obtained with three additional clones of each cell type synthesizing comparable levels of FGF-2 as determined by Western blot analysis (data not shown).
We investigated whether exogenously added FGF-2 increases integrin levels in NIH 3T3 cells transfected with HMW FGF-2 cDNA. As in control cells, addition of FGF-2 increased cell surface levels of ␤1 integrins including the ␣5␤1 and ␣6␤1 receptors in cells expressing only HMW FGF-2 (Table I). Thus, the low level of integrins in NIH 3T3 cells synthesizing only HMW FGF-2 does not result from an inability to respond to extracellular FGF-2. In contrast to control cells and cells expressing HMW FGF-2, FGF-2 addition to cells synthesizing WT FGF-2 did not further increase integrin levels. In cells synthesizing 18-kDa FGF-2, integrin levels were further induced only slightly by exogenous addition of FGF-2 ( Table I).
The previous results support the hypothesis that endogenous HMW FGF-2 does not increase ␤1 integrins above levels observed in control cells, although exogenous FGF-2 can mediate this effect on these cells. To test further whether the synthesis of 18-kDa FGF-2 is responsible for changes in integrin levels, cells synthesizing only HMW FGF-2 were transfected with cDNAs encoding 18-kDa FGF-2 and a hygromycin-resistant gene. As shown in Fig. 2B, cells expressing both HMW FGF-2 and 18-kDa FGF-2 displayed a 3-fold increase in ␤1 integrins compared with cells expressing HMW FGF-2 alone. This result confirms the hypothesis that 18-kDa FGF-2 modulates integrin cell surface levels in NIH 3T3 cells.
Induction of Surface ␤1 Integrins by Endogenous 18-kDa FGF-2 Is FGF Receptor-mediated-Although the FGF-2 translation product lacks a signal peptide sequence normally required for secretion, the protein is released from cells. Bikfalvi et al. (46) demonstrated that in NIH 3T3 cells, extracellular interaction of 18-kDa FGF-2 with its cell surface receptor is required for the growth factor's biological activities including increases in cell migration, cell proliferation, and FGF receptor down-regulation. In contrast, interaction with cell surface high affinity receptors is not necessary for the biological activities stimulated by HMW FGF-2 such as increases in cell proliferation and growth in low serum. Therefore, we next questioned whether the regulation of integrin expression by 18-kDa FGF-2 was dependent on the presence of high affinity FGF receptors. In order to test this hypothesis, cells were transfected with a cDNA encoding a dominant negative mutant type-2 FGF receptor lacking the tyrosine kinase domain. Previously, it was demonstrated that the effect of this mutant receptor is transdominant (i.e. it inhibits signaling by all FGF receptor types) (52). ␤1 Integrin Subunit-Of the integrins regulated by 18-kDa FGF-2, we have focused our studies on ␣5␤1, which has been shown to control migration of neural crest-like cells and Chinese hamster ovary cells (53,54). To characterize the mechanism by which 18-kDa FGF-2 increases cell surface ␣5␤1 levels, we first examined the biosynthesis of the ␤1 subunit. Cells were metabolically labeled under steady-state conditions, and cell extracts were immunoprecipitated with ␤1-specific antibody. Fig. 4A illustrates that the combined amount of precursor and mature ␤1 forms was not affected by expression of 18-kDa FGF-2 compared with control cells. However, endogenous expression of 18-kDa FGF-2 increased the amount of mature ␤1 subunit by 4-fold that was paralleled by a decrease of 4-fold in the precursor form. In control cells, 20% of the total of precursor and mature ␤1 forms was present as mature ␤1, whereas, in the two clones synthesizing 18-kDa FGF-2, 75% and 40% was present as mature ␤1. The effect appeared to be dependent on the expression level of 18-kDa FGF-2 because of the two clones examined (lanes 2 and 3); the clone synthesizing a higher level of 18-kDa FGF-2 (lane 3) had the higher percent mature ␤1. Both clones of cells that expressed only HMW FGF-2 did not significantly differ in the level of either premature or mature ␤1 with respect to control cells even though the cells analyzed produced levels of HMW FGF-2 comparable with the levels of FGF-2 in the two clones transfected with 18-kDa FGF-2 cDNA. This experiment demonstrates that 18-kDa FGF-2 expression increases the amount of precursor ␤1 converted to the mature form. To confirm this result, Western blot analysis was performed because it is another method to detect steady-state ␤1 integrin levels. The results were in agreement with the previous findings as more mature ␤1 was present in cell extracts from cells expressing 18-kDa FGF-2 compared with control cells. In addition, there was a parallel decrease in precursor ␤1 (data not shown).
Endogenous 18-kDa FGF-2 Increases the Rate of Processing of the ␤1 Subunit-A pulse-chase experiment was performed to test whether 18-kDa FGF-2 affects the rate of conversion of premature ␤1 to mature ␤1 subunit. Control cells and cells expressing 18-kDa FGF-2 were metabolically labeled for 1 h and chased for various lengths of time. The cell extracts were immunoprecipitated with ␤1 antibody and the antigens separated by SDS-PAGE. Fig. 5A shows that the ␤1 subunit was synthesized as an immature form that, after 3 h of chase, was partially converted to the mature form. Between 12 and 24 h of chase, the protein was completely converted to the mature form. The kinetics of processing were faster in cells expressing 18-kDa FGF-2. After 3 h of chase the amount of mature ␤1 was twice the level found in control cells. After 6 h of chase, the premature form was almost completely converted to the mature form. As observed above, the total amount of ␤1 subunit synthesized was not affected. In addition, control cells and cells expressing HMW FGF-2 were pulse-labeled and immunoprecipitated with anti-␤1 antibody. The kinetics of processing of ␤1 was unaffected by HMW FGF-2 expression (data not shown).
Biosynthesis of the ␣5 Subunit Is Increased by 18-kDa FGF-2 Expression-To investigate the synthesis of the ␣5 subunit, five clones of cells (one control clone, two clones expressing only 18-kDa FGF-2, and two clones expressing only HMW FGF-2) were metabolically labeled, and cell extracts were immunoprecipitated with ␣5-specific antibody. Fig. 4B shows that the synthesis of the ␣5 subunit was increased 5-fold. The effect on ␣5 subunit synthesis was dependent on the expression level of 18-kDa FGF-2 because the clone expressing higher 18-kDa FGF-2 (lane 2) synthesized more ␣5 subunit. In one clone expressing only HMW FGF-2 (lane 4), the synthesis of the ␣5 subunit was not affected. In the second clone (lane 5), the ␣5 level was slightly higher than control levels. However, additional experiments indicate that this increase was not reproducible (data not shown). Western blot analysis using anti-␣5 antibody confirmed the previous result demonstrating that 18-kDa FGF-2 but not HMW FGF-2 increases the biosynthesis of the ␣5 subunit (data not shown). To study whether 18-kDa FGF-2 affects the rate of processing of the ␣5 subunit, a pulsechase experiment using cell extracts from control cells and cells synthesizing 18-kDa FGF-2 was performed. The results obtained using control cells showed that ␣5 was synthesized as a precursor form, which after 30 min of chase was partially converted to the mature form (Fig. 5B). After 3 h of chase the protein was completely converted to the mature form and was associated with the ␤1 subunit. In contrast, 18-kDa FGF-2 synthesis increased the rate of processing of the ␣5 subunit as mature ␣5 was already visible at the start of chase and the ␣5␤1 complex was observed by 1 h of chase. Thus, 18-kDa FGF-2 increases the biosynthesis as well as the rate of maturation of the ␣5 subunit. Whereas the rate of maturation of the ␤1 subunit is similarly increased by 18-kDa FGF-2, the biosynthesis of ␤1 is unaffected.
Pre-translational Regulation of the ␣5 and ␤1 Subunits by FGF-2 Expression-To analyze further the mechanism of increased ␣5 synthesis, Northern blot analysis was performed. Total RNA extracted from control cells and cells synthesizing 18-kDa FGF-2 was hybridized with ␣5-specific, ␤1-specific, and GAPDH-specific cDNA probes. The results showed that, after normalization to GAPDH mRNA levels by PhosphorImager scanning analysis, the level of mRNA coding for the ␣5 subunit was increased by 3-fold by 18-kDa FGF-2 synthesis (Fig. 6A). However, the level of ␤1 mRNA was unchanged by 18-kDa FGF-2 (Fig. 6B). The level of ␤1 or ␣5 mRNAs did not change with synthesis of HMW FGF-2 (data not shown). Thus, endogenous 18-kDa FGF-2 specifically increases the message level of the ␣5 subunit.
Increase in ␣5 mRNA Levels by FGF-2 Expression Is Due to Enhanced Transcriptional Activity and Not mRNA Stabilization-The increase in ␣5 mRNA levels could reflect increased transcription of the gene or an increase in message stability. The stability of ␣5 mRNA was determined in NIH 3T3 cells by treatment with actinomycin D (5 g/ml) to inhibit transcription. RNA was isolated after various time intervals, and equal amounts of total cellular RNA were analyzed by Northern blotting (Fig. 7). By this method, the half-life of ␣5 mRNA was determined to be approximately 1 h in control cells. In cells synthesizing 18-kDa FGF-2, the half-life was also approximately 1 h. These data suggest that 18-kDa FGF-2 expression does not result in an increase in ␣5 mRNA stability.
To determine whether transcription of the ␣5 gene was increased by 18-kDa FGF-2, the ␣5 promoter activity was measured in control cells and cells synthesizing 18-kDa FGF-2. Cells were transiently transfected with either p␣5-926LUC or pLUC and a transfection control ␤-gal plasmid. Luciferase activity in extracts prepared from cells transfected with either luciferase construct were normalized to ␤-gal activity. In cells synthesizing 18-kDa FGF-2, there was a 6-fold increase in luciferase activity with expression of p␣5-926LUC compared with control cells (Fig. 8). Thus, 18-kDa FGF-2 acts on the ␣5 promoter to drive expression of the luciferase reporter gene. As a further control, cells synthesizing HMW FGF were transfected with the luciferase constructs. HMW FGF-2 expression did not increase ␣5 promoter activity significantly above that in control cells (data not shown). Thus, 18-kDa FGF-2 enhances ␣5 mRNA levels by increasing the transcription of the ␣5 gene. DISCUSSION The data reported in this paper show that the pattern of expression of ␤1 integrins at the cell surface of NIH 3T3 cells is influenced by endogenous 18-kDa FGF-2 but not by HMW FGF-2. This conclusion is based on the following observations. surface mediated by 18-kDa FGF-2. These results add support to the model that 18-kDa FGF-2 is released from cells and interacts with cell surface FGF receptors, which induces receptor phosphorylation and signal propagation, and ultimately triggers various biological responses. These include down-regulation of FGF receptors, increases in motility, stimulation of growth, and modulation of integrin expression. The responses we have observed are dependent upon the absolute amount of 18-kDa FGF-2 because clones of cells synthesizing low levels of 18-kDa FGF-2 mediate these effects to a lesser extent than cells synthesizing high levels, probably because less growth factor is released and available to interact with receptors. Clones of NIH 3T3 cells expressing high levels of HMW FGF-2 do not regulate integrin levels probably because the growth factor is not released from cells in sufficient quantity (46). This may be due to the nuclear localization sequence that efficiently targets HMW FGF-2 into the nucleus. Whereas certain biological activities such as growth in low serum may be mediated by HMW forms of FGF-2, integrin modulation is not.
The most abundant integrin in the parental NIH 3T3 cells was ␣5␤1, which displayed a striking up-regulation at the cell surface in cells synthesizing 18-kDa FGF-2. Metabolic labeling of NIH 3T3 cells expressing FGF-2 followed by immunoprecipitation with anti-␣5 antibody showed that the modulation of ␣5␤1 appearance at the cell surface reflects a concomitant modification of the biosynthesis of the ␣5 subunit. Northern blot analysis demonstrated that the change in the rate of biosynthesis is a result of an increase in the transcript level of ␣5, and luciferase assays indicated that this increase is a consequence of modulation of the rate of transcription of the ␣5 gene. In contrast to ␣5, the biosynthesis of ␤1 is not enhanced by endogenous 18-kDa FGF-2, probably because an excess pool of precursor ␤1 already exists in control cells. However, 18-kDa FGF-2 dramatically increases the rate of processing of ␤1 as measured by pulse-chase experiments, and this enhanced rate of processing increases the level of mature ␤1. Similarly, the rate of processing of ␣5 is stimulated by 18-kDa FGF-2.
We previously observed that NIH 3T3 cells expressing 18-kDa FGF-2 are more migratory than control cells, whereas cells expressing HMW FGF-2 migrated to the same degree as control cells. These differences in migration between the cells expressing 18-kDa and HMW FGF-2 may be due to differences in the levels of ␣5␤1. In neural crest-like cells, the repertoire of integrins and the extent of integrin expression determined the rate of cell migration and the particular pathway of cell migration (53). Expression of ␣5␤1 or ␣4␤1 in mouse sarcoma S180 cells, which behave similarly to neural crest cells and normally synthesize low levels of these integrins, promoted an increase in cell motility in vitro. When these cells were grafted into an embryo, they migrated in distinct pathways compared with parental cells. The cells expressing ␣5 migrated simultaneously in both ventral and dorsolateral pathways in contrast to the parent cells that migrated only in the ventral path. Similarly, cells expressing low levels of ␣4 migrated in both ventral and dorsolateral pathways. However, the cells expressing high levels of ␣4 remained nonmigratory. Thus, the repertoire and levels of integrins enabled the cells to utilize different pathways of migration and regulate their speed of migration in vivo. Based on these observations, it is likely that increased ␣5␤1 levels in NIH 3T3 cells synthesizing 18-kDa FGF-2 play a role in the enhanced migration compared with control cells, but this hypothesis still remains to be proven.
Other studies support the above hypothesis. Variants of Chinese hamster ovary cells were selected that expressed reduced levels of ␣5␤1 (54). These cells exhibited slower migration than the parental cell line. This result taken together with the studies described above, strongly suggests a direct correlation between the concentration of ␣5␤1 and speed of migration. However, if the level of ␣5␤1 is increased significantly, cell migration is decreased (47). One explanation for this effect is that very high levels of ␣5 may produce an affinity to the substratum that reduces rather than increases motility. As increased expression of the ␣5 subunit enhances fibronectin assembly at the cell surface, this could immobilize the cells (47,53).
In addition to regulating migration of the NIH 3T3 cells expressing 18-kDa FGF-2, ␣5␤1 may contribute to the proliferation of these cells. ␣5␤1 expression by HT29 colon carcinoma cells decreases cell proliferation by inducing the transcription of growth arrest gene 1, a gene product that induces growth arrest and blocks transcription of several immediate early genes (55). These changes occur in the absence of cell attachment to fibronectin. However, ligation of ␣5␤1 to fibronectin down-regulates growth arrest-specific gene 1 expression, activates immediate early gene transcription, and induces cell proliferation. Thus, ␣5␤1 can generate both positive and negative signals depending on whether it is bound to its substrate fibronectin. Therefore, ␣5␤1 expression in NIH 3T3 cells transformed by 18-kDa and HMW FGF-2 may contribute to enhanced cell proliferation.
Integrins have been shown to be required during angiogenesis. Several studies have demonstrated that blocking the activity of integrins affects angiogenesis. In particular, antibodies against ␣v␤3 or ␣v␤5 severely perturbed angiogenesis induced in the chorioallantoic membrane by FGF-2 or VEGF, respectively (56,57). In fact, antibodies against ␣v␤3 induced apoptosis in proliferative angiogenic vascular cells suggesting that ligation of ␣v␤3 may be required for the survival and maturation of newly forming blood vessels (56). Antibodies against ␤1 or ␣v␤3 integrins injected into quail embryos arrest or severely disrupt vasculogenesis indicating an important role for both ␤1 and ␣v␤3 integrins during vasculogenesis (58,59). However, it is most likely that the specific integrins playing a role during angiogenesis depend on the tissue type. Although it is clear that many cytokines and growth factors can induce angiogenesis, little is known about the molecular mechanisms underlying this activity. Changes in the level of expression or function of integrins may be necessary during angiogenesis. We have previously demonstrated that FGF-2 can regulate inte- FIG. 8. Integrin ␣5 promoter activity in control NIH 3T3 cells or cells synthesizing 18-kDa FGF-2. Cells expressing FGF-2 (clone 43FGFc31) or control cells (Zipneoc2) were transfected with p␣5-926LUC or pLUC and a transfection control ␤-gal plasmid. Luciferase and ␤-gal activities were measured as described under "Materials and Methods." Luciferase activity was normalized by dividing by ␤-gal activity (A 420 ). Each column represents a mean of three experiments with a bar indicating the standard error of the mean. RLU, relative light units. grin levels in microvascular endothelial cells. We show here that the 18-kDa form of FGF-2 is the only endogenous form that mediates this effect. HMW FGF-2 does not modulate integrin production and, therefore, blocking the extracellular activity of the 18-kDa form may be sufficient to block integrin modulation by FGF-2 in vivo. It is possible that blocking the FGF-2-induced modulation of integrin levels is sufficient to inhibit angiogenesis.
In summary, we have shown that endogenous expression of 18-kDa FGF-2, but not HMW FGF-2, modifies surface integrin levels. This involves 18-kDa FGF-2 interaction with FGF receptors and signaling of changes in integrin biosynthesis and processing. Enhanced ␣5␤1 levels caused by endogenous 18-kDa FGF-2 may play a role in the increased migration and proliferation of the cells. Furthermore, modification of integrin expression in vivo by 18-kDa FGF-2 may be important during several FGF-2-mediated processes including mesoderm formation, wound healing, and angiogenesis.