Expression and Splicing of FGF Receptor mRNAs during ARPE-19 Cell Differentiation In Vitro

Abstract

purpose. The expression and alternative splicing of the four FGF receptor (FGFR) mRNAs are regulated in a developmental- and tissue-specific fashion. Capability of differentiation in vitro of the retinal pigment epithelial cell line ARPE-19 has been previously demonstrated. In this study, the hypothesis that FGF receptor gene expression and the alternative splicing of the FGFR1 mRNA is regulated as a function of ARPE-19 differentiation in vitro was tested.

methods. ARPE-19 cells were plated at sparse or confluent densities and maintained in culture up to 14 months. The expression of FGF receptors and the ratio of the FGFR1β to FGFR1α splice variants of the FGFR1 transcript were quantified by a published PCR technique. Two in vivo samples of human RPE served as controls.

results. Sparse cultures of ARPE-19 cells predominantly express FGFR1. When these cultures are allowed to differentiate, FGFR2 is also expressed. Samples of mRNA from RPE cells in vivo exhibit FGFR1 and FGFR2 expression as well as FGFR3 expression, a form that is minimally apparent in vitro. The ratio of the FGFR1β to FGFR1α splice variant decreases as a function of cell differentiation in vitro and approaches the ratio observed in human RPE cells in vivo. Stimulation of cultures in vitro with FGF2 as a prototypical differentiation agent does not regulate the ratio of the FGFR1β to FGFR1α splice variant.

conclusions. Differentiation of the ARPE-19 cell line in vitro recapitulates many but not all the in vivo patterns of FGFR expression and splicing. This in vitro system may be useful for selected studies on how cellular differentiation regulates FGF receptor gene expression and splicing.

The family of fibroblast growth factors (FGFs) consists of at least 15 structurally related polypeptides, several of which have been shown to be expressed in the retinal pigment epithelium (RPE). ^{1 2} FGFs play key roles in a wide variety of crucial biological activities, including trophic support, differentiation, and angiogenesis. ^{3 4 5} All biological effects of FGFs are mediated by binding to high-affinity receptors. The four high-affinity receptors, FGFR1–FGFR4, are composed of a signal peptide, two or three immunoglobulin (Ig)-like loops in the extracellular domain, a single hydrophobic transmembrane domain, and a highly conserved tyrosine kinase domain split by a short kinase insert sequence. ^{6 7} Alternative splicing of RNA transcripts gives rise to variant products of the FGFR genes in the extracellular and intracellular domains. To date, more than 20 single-site variations have been described. Most occur in the FGFR1 and FGFR2 gene products. ⁸ Alternative splicing of a single exon in the FGFR1 and FGFR2 genes results in the isoforms FGFR1α and FGFR2α, both of which exhibit an additional Ig-like loop (loop I), NH₂-terminal to loops II and III of the respective β isoforms. ⁶  

Wang et al. ⁹ have shown that the β isoform of FGFR1 has a 10-fold higher affinity for FGF2 than does the α isoform. This observation suggests that alternative splicing of FGFR1 may lead to functional changes in cellular responses to FGF2 both in vitro and in vivo. Interestingly, dedifferentiation of cells has been correlated with a shift in alternative splicing of FGFR1 and FGFR2 from the α to the β isoforms. This phenomenon has been documented for the malignant progression of human pancreatic cells, prostatic cells, and brain astrocytes. ^{10 11 12} Regulation of FGF receptor splicing has also been demonstrated for nontransformed cells as a part of intimal proliferation during coronary graft rejection. ¹³  

RPE cells also undergo dedifferentiation as a part of the development of proliferative diseases of the posterior pole, such as proliferative vitreoretinopathy (PVR) and age-related macular degeneration (AMD). If this dedifferentiation were accompanied by a relative shift from FGFR1α to FGFR1β, the resulting cells would be more responsive to FGF2, and this transition might therefore play a key role in the development of these pathologies. 

We hypothesize that the differentiation and dedifferentiation of RPE cells regulate both the relative expression of FGF receptor genes and the alternative splicing of FGF receptor mRNAs. To test this hypothesis, we have examined the regulation of FGF receptor gene expression and the alternative splicing of the FGFR1 mRNA as a function of ARPE-19 differentiation in vitro. 

Materials and Methods

Cell Culture

Routine experiments were performed with the following cell lines. ARPE-19 cells are human diploid nontransformed RPE cells, which display many differentiated properties typical of RPE in vivo. ¹⁴ ARPE-19 cells were plated at low (10,000 cells/cm²) or high (100,000 cells/cm²) density and maintained in culture for 3 days. Differentiated ARPE-19 cells were maintained in culture for 2.5, 7, or 12 months. The T-47D human breast carcinoma cell line was purchased from American Type Culture Collection (ATCC, Manassas, VA) and was used as a positive control for FGF receptor gene expression. ¹⁵ T-47D cells were plated at low density and grown under the conditions suggested by the manufacturer. 

Growth Factors and Matrix Preparation

Recombinant human FGF2, PDGF, and TGF-β were purchased from R&D Systems (Minneapolis, MN). Laminin and Matrigel were purchased from Collaborative Biomedical Research Products (Bedford, MA). Laminin (5 or 10 μg/cm²) and Matrigel (dilutions of 1:20 and 1:40) were prepared as recommended by the manufacturer. The technique for the preparation of in vitro extracellular matrix was a slight modification of a previously published method. ¹⁶ Briefly, differentiated ARPE-19 cells were rinsed with Hanks’ balanced salt solution (HBSS) without calcium and magnesium and incubated with 0.01% Triton X-100 for 1 hour. The flask was agitated to loosen the cells, and the remaining extracellular matrix was then washed three times with HBSS and once with serum-free medium. 

Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR)

Total RNA was isolated from the various cell lines under low-density, high-density, and differentiated conditions using Trizol reagent according to the manufacturer’s instructions (GIBCO BRL, Grand Island, NY), and quantified by spectrophotometry. The RT-PCR quantification of the relative abundance of mRNA species related to the four FGF receptor genes was performed by the method of Xin et al. ¹⁷ The design of the PCR primers (pTKI and pTKII) complementary to two oligonucleotide sequences common to all four of the known forms of FGF receptor genes allowed for the universal amplification of mRNA transcripts for the FGF receptor genes characterized to date (Fig. 1) . Also, receptor-specific primers pR1-Int, pR2-Int, pR3-Int, and pR4-Int, derived from the TK insert regions (Fig. 1) , were used to detect different FGF receptor gene by Southern blot analysis. ¹⁷ First-strand cDNAs were produced using random primers and SuperScript II RNase H⁻ Reverse Transcriptase (RT; GIBCO BRL). Two in vivo samples of human RPE first-strand cDNA were a generous gift from Peter Campochiaro (Department of Ophthalmology, The Johns Hopkins University School of Medicine, Baltimore, MD). The resulting cDNAs for all four FGF receptor mRNA species were amplified using a primer pair corresponding to sequences conserved in all four FGF receptor genes. The PCR reactions were resolved by agarose gel electrophoresis and visualized by ethidium bromide staining, yielding a single band of 480 bp. After Southern blot transfer, the relative abundance of transcripts unique to each gene was detected with end-labeled oligonucleotides specific for each receptor gene. The specific activity of each oligonucleotide probe was adjusted with cold oligonucleotides, and blots were quantified using phosphorimager analysis. 

The relative abundance of the FGFR1α isoform (III-Ig loop) and FGFR1β isoform (II-Ig loop) was quantified using the method of Luqmani et al. ¹⁸ A primer pair unique to FGFR1 (p1 and p2) was used to amplify a sequence in the extracellular domain, including exons I, II, and part of exon III (Fig. 1) . RT-PCR of FGFR1α and FGFR1β results in amplified bands of 464 and 197 bp, respectively. ¹⁸ The relative abundance of these two bands was quantified by phosphorimager analysis after Southern transfer and hybridization with an oligonucleotide probe (pR1αβ; Fig. 1 ) common to both sequences. 

Northern Analysis

RNA (15 μg) was electrophoresed in formaldehyde-agarose gels and transferred to 0.45-μm Hybond-N membrane (Amersham, Arlington Heights, IL) according to standard procedures. ¹⁶ After UV crosslinking, the blots were probed with ³²P-labeled cDNAs for FGFR1 (item 1042862; ATCC,) and FGFR2 (item 1035480; ATCC), washed, exposed, and developed according to standard procedures. Blots were quantified using phosphorimager analysis and normalized against 28S rRNA. 

mRNA Half-Life Determination

The mRNA half-life of FGFR1 transcripts was measured using the RNA polymerase inhibitor 5,6-dichloro-1-β-d-ribofuranosyl benzimidazole riboside (DRB; Sigma, St. Louis, MO) at a final concentration of 25 μg/ml. ¹⁹ After addition of DRB to cells, RNA was extracted after 0, 3, 6, 9,12, 15, 18, and 24 hours. Half-life was measured by plotting log of normalized phosphorimager counts versus time after DRB exposure. 

Results

Relative Abundance of FGFR1–FGFR4 Transcripts in ARPE-19 Cells as a Function of Cell Differentiation

To assess the relative expression of the four FGFR genes in the ARPE-19 cell line, we used a previously published RT-PCR method. ¹⁷ A schematic of an FGF receptor and the locations of primers used for analyzing the expression levels of different FGF receptors are shown in Figure 1 . Table 1 shows the percent total abundance of all FGF receptors in proliferating and differentiated ARPE-19 cells in vitro, as well as in human RPE cells in vivo. Criteria for demonstrating the differentiation of ARPE-19 cells in vitro were established in previous publications and included cuboidal morphology, functional polarity, and the expression of RPE-specific genes. FGFR1 is the predominant form under all conditions tested in vitro and in vivo. FGFR2, FGFR3, and FGFR4 are not substantially expressed in proliferating ARPE-19 cells. When these cells are allowed to differentiate for 7 months in vitro, FGFR2 is expressed, albeit at lower levels than FGFR1. In addition, the results demonstrate that human RPE cells abundantly express FGFR3 in vivo, compared to its minimal expression in vitro. The control T-47D cells expressed all four FGF receptor genes. ¹⁵  

Northern Blot Analysis of FGFR1 and FGFR2 Expression in ARPE-19 Cells as a Function of Cell Differentiation

Expression of FGFR1 and FGFR2 in ARPE-19 cells was examined by Northern blot analysis to provide a comparison for the relative expression data presented in Table 1 . As shown in Figure 2A , FGFR1 steady state mRNA levels increase approximately 2.5-fold as a function of cell differentiation. Figure 2B shows that FGFR2 gene expression is observed in differentiated ARPE-19 cells but not in proliferating cells. 

Analysis of the Ratio of the FGFR1β to FGFR1α Splice Variants as a Function of Differentiation

We examined how FGFR1 splicing varies as a function of proliferation and differentiation. We elected to study the FGFR1α and FGFR1β splice variants because these variants have different affinities for FGF2, and splicing might therefore have a functional consequence with respect to receptor function. The analysis was conducted by RT-PCR, and the position of each PCR primer (p1 and p2) is shown relative to FGFR1 (Fig. 1) . Figure 3 shows that the ratio of β to α isoforms of FGFR1 decreases from a value of 1.94 ± 0.12 in proliferating cells to 0.75 ± 0.05 in differentiated ARPE-19 cells (2.5 months). This ratio approaches a value of 0.27 in differentiated RPE cells in vivo. 

To determine whether differential stabilities of the two FGFR1 splice variants could account for differences in steady state mRNA levels, the half-lives of these two mRNAs were measured at confluence. Figure 4A demonstrates that the FGFR1 steady state mRNA has a half-life of 10 hours. RT-PCR was performed using the same RNA samples for the determination of the ratio of β to α isoforms (Fig. 4B) . The ratio does not change over a 24-hour period after DRB treatment, indicating that both isoforms have the same half-life. 

Effects of Growth Factors and Extracellular Matrix on the Ratio ofβ to α Isoforms of FGFR1

Regulation of expression of alternatively spliced variants by cytokines or growth factors has been previously demonstrated. ^{20 21 22 23 24} To investigate whether the splicing pattern of α and β isoforms is induced by growth factors, we treated proliferating ARPE-19 cells with FGF2. The results in Figure 5 reveal that treatment of ARPE-19 cells with FGF2 (20 ng/ml) did not affect the ratio of FGFR1β to FGFR1α. The same experiment performed with PDGF (20 ng/ml) or TGF-β (2 ng/ml) also did not result in a ratio change of FGFR1β to FGFR1α (data not shown). 

As RPE cells become differentiated, matrix proteins are deposited and might play a role in the observed ratio change of β to α isoforms demonstrated in Figure 3 . We therefore studied the role of plastic, laminin, Matrigel, and in vitro deposited matrix from differentiated ARPE-19 cells on the regulation of the ratio of FGFR1 splice variants. The results demonstrated that growth on these different matrices did not regulate the ratio of FGFR1β to FGFR1α in ARPE-19 cells seeded at low or high density (data not shown). 

Discussion

The data presented in this study support the hypothesis that expression and alternative splicing of FGF receptors is regulated by cellular differentiation in vitro. We have demonstrated that the relative expression levels of the four FGF receptor genes change as a function of differentiation in ARPE-19 cells. FGFR1 is the predominant form in proliferating cultures of ARPE-19 cells, and FGFR2 expression is upregulated by differentiation in vitro. ARPE-19 cells do not exhibit significant FGFR3 or FGFR4 expression under any of the conditions we tested in vitro. However, our data demonstrate that in vivo human RPE express significant levels of FGFR3. The ratio of β toα isoforms of the FGFR1 transcript also decreases as a function of cell differentiation in vitro, approaching the ratio observed in human RPE cells in vivo. We could not demonstrate that either growth factor treatment or growth on matrices changes the ratio of FGFR1β to FGFR1α in ARPE-19 cells. 

The ARPE-19 cell line was chosen for these studies because our laboratory has previously demonstrated that this cell line is capable of differentiation in vitro in a time-dependent fashion and displays differentiated properties similar to those observed in human RPE cells in vivo. ^{14 25} These properties include cuboidal morphology, the expression of RPE-specific genes, and functional polarity. ¹⁴ Our laboratory has also shown that differentiation of ARPE-19 cells in vitro uncovers silencer activity in the FGF-5 gene promoter. ²⁶  

In a previous study, Northern blot analysis of normal and dystrophic rat RPE cells demonstrated expression of FGFR1 and FGFR2, but FGFR3 and FGFR4 were not evaluated. ^{27 28} Three different FGF receptor genes were also found to be expressed in 7.5-day chick embryonic RPE. ²⁹ These included FGFR2, FGFR3, and an RPE-specific form, which was shown to be 70% identical with FGFR1. We could only find significant FGFR3 expression in vivo. FGFR2 clearly showed a pattern of increasing expression with cellular differentiation. These results on the relative expression of FGFR1 and FGFR2 parallel the findings of Ali et al. ³⁰ The expression of FGFR1 was unaffected by the differentiation of the F-9 embryonal carcinoma cell line. Differentiation of F-9 cells, however, led to a dramatic upregulation of FGFR2 expression. 

FGF1 and FGF2 have been shown to directly regulate exon IIIb/IIIc splicing of FGFR2 and FGFR3 in a human keratinocyte cell line and a rat bladder carcinoma cell line. ³¹ Despite the ability of some FGFs to directly regulate FGF receptor splicing in exon III, we could not detect a direct effect of FGF2 on the ratio of FGFR1β to FGFR1α in ARPE-19 cells. The literature contains several other reports demonstrating growth factor or cytokine regulation of alternative splicing. For example, modification of tenascin isoform ratios induced by TGF-β, FGF2, and PDGF has been previously reported. ^{20 21 22 23} TGF-β has been demonstrated to alter fibronectin pre-mRNA splicing as well. ²⁴ In our studies, TGF-β and PDGF did not regulate the ratio of β to α isoforms of FGFR1 transcripts (data not shown). 

Wang et al. showed a 10-fold increase in the affinity of FGFR1β for FGF2 when compared with the affinity of FGFR1α for FGF2. Together with our findings, this observation suggests that undifferentiated RPE cells in vivo may be more responsive to FGF2 than differentiated RPE cells. This difference in the potential response of undifferentiated versus differentiated RPE cells to FGF2 may have implications both in development and pathology. We previously documented time-dependent changes in FGF2 expression during the development of the bovine and murine retinas. ^{32 33} Altered levels of FGF2 expression when combined with the events of RPE differentiation in vivo might lead to the discrete timing of biological responses to FGF2 in differentiating RPE cells. 

Pathology presents a more clear-cut case where the relative response of differentiated and undifferentiated RPE cells to FGF2 may be important. In both PVR and AMD, normal RPE cells with an epithelial phenotype undergo a transition to RPE cells with a mesenchymal or proliferative phenotype. Such cells are commonly found to be a constituent of fibrotic or fibrovascular cellular membranes, where they contribute substantially to the observed pathology. Several other studies in the literature have examined gene expression in RPE cells as a function of dedifferentiation. ^{34 35 36}  

Several studies have documented FGF2 expression in these pathologies. ^{37 38 39} One natural consequence of these findings might be that undifferentiated cells are selectively acted on by FGF2 with respect to differentiated cells. This may in turn lead to altered proliferation, viability, or as yet undetected biological responses. It remains a task for future work to identify the relative levels of FGFR1α and FGFR1β in vivo and any relevance of these observed ratios to pathology. 

 Supported in part by National Institutes of Health grants EY06473 (LMH) and F32 EY06650 (CMG) and by an unrestricted grant from Research to Prevent Blindness (RPB). LMH is a recipient of an RPB Senior Scientist Award.

 Submitted for publication November 22, 1999; revised February 7, 2000; accepted February 15, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Schematic of FGF receptor and the location of PCR and oligonucleotide probes. The position of each PCR primer is shown relative to the FGF receptor. The structure of the FGF receptor, including the amino terminus (H₂N), initiation start site (I), immunoglobulin-like disulfide loops I–III, acidic box domain (A), transmembrane domain (TM), tyrosine kinase consensus sequences I and II (TKI, TKII), kinase insert region (KI), and carboxyl-terminus (COOH). The primers pTKI and pTKII were selected based on the consensus cDNA sequence of highly conserved tyrosine kinase domains for all four FGF receptors. ¹⁷ The specific primers pR1-Int, pR2-Int, pR3-Int, and pR4-Int derived from the highly divergent TK insert regions were used to detect the different FGF receptor genes by Southern blot analysis (Table 1) . PCR using primers p1 and p2 resulted in the amplification of two structural variants of FGFR1, the α isoform (464 bp), and the β isoform (197 bp). An oligonucleotide probe (PR1αβ) was used to detect both isoforms (Fig. 3) .

Table 1.

 

Relative Abundance of FGFR1–FGFR4 Transcripts in ARPE-19 Cells as a Function of Differentiation

	FGFR1*	FGFR2	FGFR3	FGFR4
Low density (proliferating), †	95.3 ± 0.9	0.2 ± 0.1	3.1 ± 1.6	1.4 ± 0.7
High density (proliferating), †	96.5 ± 1.2	0.2 ± 0.1	2.7 ± 1.5	0.6 ± 0.5
2 months (differentiated), †	90.6 ± 1.5	8.5 ± 3.9	0.8 ± 0.5	0.1 ± 0.1
7 months (differentiated), †	73.2 ± 1.3	25.9 ± 1.5	0.4 ± 0.2	0.5 ± 0.4
In vivo RPE cells, ‡	50.9	24.3	20.9	3.9
T-47D (positive control), §	24.0	18.0	10.3	47.7

*  Results for each receptor are expressed as the percent of the total abundance of all FGF receptor genes.

†  Results are expressed as the average of three independent experiments ± SEM.

‡  Two in vivo samples of human RPE cells served as controls, and the results are expressed as the average of both.

§  Data for positive control (T-47D) are from a single experiment.

Figure 2.

 

Northern blot analysis of absolute FGFR1 and FGFR2 mRNA levels in ARPE-19 cells as a function of cell differentiation. Total RNA was isolated from proliferating and differentiated ARPE-19 culture using conditions described in Materials and Methods. Northern blot analysis was performed, and the blots were hybridized with ³²P-labeled FGFR1 (A) and FGFR2 (B) cDNA. Northern blot analysis were normalized against 28S rRNA and quantified by phosphorimager analysis. The results are expressed as the average of two independent experiments.

Figure 3.

 

Analysis of the ratio of FGFR1β to FGFR1α splice variants as a function of differentiation. ARPE-19 cells were plated at low (10,000 cells/cm²) or high (100,000 cells/cm²) density and maintained in culture for 3 days. Differentiated ARPE-19 cells were maintained in culture for 2.5, 7, or 12 months. RT-PCR was performed for the determination of the relative abundance of the FGFR-1α and FGFR-1β isoforms. Quantification of bands was performed by phosphorimager analysis after Southern blot transfer. Results are expressed as the ratio of the β to α isoform of FGFR1. Three independent experiments were performed in triplicate, and the ratio values are reported as means ± SEM. Ratios from two in vivo samples of human RPE cells are presented as the average of both.

Figure 4.

 

Measurement of FGFR1 mRNA half-life. ARPE-19 cells were plated at 100,000 cells/cm² in complete medium and maintained in culture for 3 days. Fresh medium containing 25 μg/ml DRB was added, and total RNA was isolated at 0, 3, 6, 9, 12, 15, 18, and 24 hours after DRB addition. Northern blot analysis was performed and FGFR1 transcripts were quantified by phosphorimager analysis (A). Results are expressed as log₁₀ of the normalized phosphorimager counts. RT-PCR was performed using the same RNA samples for the determination of the ratio of β to α isoforms (B). The bands were quantified by phosphorimager analysis after Southern blot transfer. Values for the ratios from three independent experiments were averaged and are plotted ± SEM.

Figure 5.

 

The effect of FGF2 on the relative ratio of β to α isoforms of FGFR1 as a function of time. ARPE-19 cells were plated at 15,000 cells/cm² in complete medium and maintained in culture for 3 days and serum starved for 48 hours. Fresh medium containing FGF2 (20 ng/ml) was added to each culture. Total RNA was isolated 5 minutes, 30 minutes, 1, 2, 4, 12, 24, or 48 hours after addition of FGF2. RT-PCR was performed for the determination of the relative abundance of α and β isoforms. Quantification of bands was performed by phosphorimager analysis after Southern blot transfer. The results of the ratio of β to α isoforms are presented relative to the untreated control at each indicated time point. Data from time points ranging from 5 minutes to 2 hours are expressed as the average of three experiments (± SEM). Data from 4-, 12-, 24-, and 48-hour time points are the result of one experiment.

We thank James T. Handa for his critical reading of this manuscript.