FGF-2 Antagonizes the TGF-β1-Mediated Induction of Pericyte α-Smooth Muscle Actin Expression: A Role for Myf-5 and Smad-Mediated Signaling Pathways

Abstract

purpose. Although the FGF and TGF-β families are known to play an important role in regulating vascular endothelial and smooth muscle cell behavior, the influence of these matrix-binding growth factors on microvascular pericyte morphogenesis is not well understood. The current study was undertaken to examine the molecular mechanisms that mediate the effects of the endothelium-produced growth regulators FGF-2 and TGF-β1 on retinal pericyte proliferation and contractile phenotype.

methods. Using purified retinal pericytes, a series of assays were implemented, including RT-PCR, DNA binding, immunoprecipitation, electrophoretic mobility shift, and indirect immunofluorescence, in an attempt to elucidate the FGF/TGF-β1 signaling cascades that mediate retinal microvascular cell growth and contractile phenotype.

results. Treatment of retinal pericytes with FGF-2 and heparin stimulated nearly a log order increase in proliferation, whereas removal of FGF-2 or addition of TGF-β1 caused withdrawal from the growth cycle, inducing a smooth-muscle-like contractile phenotype, as indicated by upregulation of α-smooth muscle actin (α-SMA). This switch from a growth-potentiated to a growth-arrested state followed induction of the transcriptional regulator myf-5, as well as the nuclear translocation of myf-5 and Smad2.

conclusions. Several critical features of the endothelial cell-extracellular matrix-pericyte molecular signaling axis were elucidated in the study that are likely to be responsible for regulating retinal microvascular morphogenesis during normal development, as well as the pathologic angiogenesis accompanying several ocular disorders, including diabetic retinopathy and age-related macular degeneration.

Biochemical analyses and gene cloning studies have clearly demonstrated the importance of the FGF family in fostering endothelial growth and angiogenesis. ^{1 2 3 4} However, although it has been established that FGF-2 is a component of select subendothelial basement membranes in vivo (as well as a component of the endothelial extracellular matrix in vitro), ⁵ its functional role in the matrix remains uncertain. ⁶ Because matrix-bound FGF envelopes populations of quiescent endothelial cells, it has been suggested that it acts as a survival agent that is released after vascular injury. ^{7 8 9} Nevertheless, if FGF-2 acts only in an autocrine or paracrine role to stimulate endothelial proliferation, then it is difficult to explain the exceedingly low rate of endothelial cell turnover in microvascular beds that possess FGF-2-enriched basement membranes. 

Beyond playing a role in wound healing, it is possible that matrix-bound FGF-2 serves other important functions during vascular morphogenesis, such as signaling the recruitment, proliferation, and/or local differentiation of mural cells (e.g., vascular smooth muscle cells or pericytes). Indeed, little is known regarding the molecular and cellular cues that govern the nonrandom manner in which pericytes come to populate specific microvessels. ¹⁰ Despite the demonstration that pericytes can inhibit endothelial cell growth through a cytokine-dependent pathway, ^{11 12} there is no evidence to indicate that pericyte growth or contractile phenotype can be modulated by endothelium-produced or matrix-associated growth regulators such as FGF-2 or TGF-β1. 

Because of the pericyte’s unique position within the microvascular basement membrane ¹³ and because it has been established that endothelial cells deposit FGF-2 in this matrix, we were eager to ascertain whether FGF-2 and other endothelium-produced growth regulators such as TGF-β specifically influence pericyte proliferation and/or contractile phenotype. Furthermore, we attempted to identify the intracellular downstream effectors that signal to regulate pericyte growth and contractile phenotype. 

Herein, we demonstrate that FGF-2 and TGF-β1 serve antagonistic functions in pericyte growth and myogenic determination. Although FGF-2 markedly stimulates pericyte growth, its removal and/or the addition of TGF-β1 causes the withdrawal of pericytes from the growth cycle and the induction of a contractile phenotype. This switch from a growth-potentiated to a growth-arrested state temporally follows a dramatic increase in the mRNA and protein levels of the basic helix-loop-helix (bHLH) transcriptional regulator of muscle differentiation myf-5, as well as the nuclear translocation of the TGF-β phosphosignaling intermediate Smad2. These results help to fill an important gap in our understanding of the molecular mechanisms regulating smooth muscle commitment and myogenic determination in the pericyte lineage. 

Materials and Methods

Antibodies

For localization, growth neutralization, and immunoprecipitation studies, diethylaminoethyl (DEAE)-fractionated or antigen affinity-purified antibodies were used. Anti-FGF-2 IgG was prepared as described ⁵ ; anti-isoactin IgG (α-smooth muscle and β-nonmuscle) were prepared and used as described ^{14 15} ; monoclonal anti-α-smooth muscle actin (α-SMA) was purchased from BioGenex (San Ramone, CA); and anti-myf-5 was prepared by injecting full-length glutathione S-transferase (GST)-myf-5 into New Zealand White rabbits. ¹⁶ Preimmune and immune sera were isolated and characterized by Western blot analysis, immunoabsorption, and immunofluorescence, as previously described. ¹⁷ For Smad antibodies, peptides generated from the proline-rich linker regions of Smad2, -3, -4, -6, and -7 were synthesized by the Tufts Protein and DNA Sequencing Core Facility (Boston, MA) and cross-linked with 0.1% glutaraldehyde. The cross-linked peptides at 5 mg/mL were dialyzed extensively against PBS and mixed 1:1 with gold adjuvant (Titermax; CytRx Corp., Norcross, GA). Female New Zealand White rabbits were initially injected intramuscularly and intradermally with 500 μg cross-linked peptide. Approximately 1 month after the first injection, the rabbits received a booster injection of 250 μg peptide, and 1 week later the first blood sample was obtained. Further booster injections of 125 μg were administered at approximate 2-week intervals, with respective blood samples collected approximately 1 week afterward. Blood was centrifuged to remove red blood cells, and sera and the resultant immune IgG fractions were tested for reactivity to Smad proteins by Western blot analysis and immunofluorescence. In all rabbits, six to nine boosts were performed, yielding immune sera for collection. The animal protocol adhered to the provisions of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Pericyte Cultures

Microvascular pericytes were isolated from bovine retinas, as previously described. ¹⁸ Briefly, retinas were removed from bovine eyes under sterile conditions, and undesirable retinal pigmented epithelial cells were removed by aspiration under a dissecting microscope. Tissue was minced before collagenase digestion, and filtration (Nitex filter; Tetko, Depew, NY) was used to isolate microvessel fragments. Primary isolates were plated in Dulbecco’s modified Eagle’s medium supplemented with 10% calf serum and antibiotics, as previously published. ^{14 18 19 20} Primary pericyte cell cultures were trypsinized and passaged once for all growth and localization studies. Pericyte cultures were characterized, as in the past, for the simultaneous expression of muscle and nonmuscle contractile proteins. ^{14 18 19 20}  

Growth Studies

For testing the effects of positive growth modulators (FGF-2- or endothelial-cell-conditioned media), as well as negative regulators (TGF-β1, anti-FGF-2, heparin), first- and second-passage pericytes were plated in triplicate into 24- or 48-well clusters. Initial seeding densities ranged from 5,000 to 14,000 cells per well. Cells growing in the presence or absence of the various test growth regulators received 5 to 12 ng/mL FGF-2 or 0 to 1 ng/mL TGF-β1 every other day for up to 2-weeks. For heparin sensitivity studies, doses ranged from 1 ng to 100 μg/mL (Elkins-Sinn, Cherry Hill, NJ). Heparin or chondroitin 4-sulfate (Sigma-Aldrich Co., St. Louis, MO) were added alone or in combination with FGF-2 or neutralizing antibodies, which were used at 10 to 100 μg/mL. When effects of anti-FGF-2 on FGF-2-stimulated or endothelial-cell-conditioned media were analyzed, preimmune or absorbed IgG controls were added, either simultaneously or in sequence, with no demonstrable differences. At the end of a given experiment, which was performed at least six times for each condition, triplicate wells were washed, trypsinized, and directly counted (model ZF cell counter; Coulter Electronics, Inc., Hialeah, FL). Results were plotted on computer (KaleidaGraph; Synergy Software, Reading, PA), depicting mean values together with standard errors. 

PCR Primers

For reverse transcription-polymerase chain reaction (RT-PCR), the following forward and reverse primers were synthesized based on published (BLAST2; www.ncbi.nlm.nih.gov/blast/ provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD) cDNA sequences as follows: myf-5 (742-bp product), forward, 5′-ATGGACATGATGGACGGCTGCCAG-3′, and reverse, 5′-GGTGATCCGATCCACTATGCTGGA-3′; forward (BP 690-1420), 5′-TCCAGCATAGTGGATCGGATCACC-3′, and reverse, 5′-TTTATTTATGGTGGAGTTTCCAAAAAGCTC-3′; β-actin (373-bp product), forward, 5′-TTCTACAATGAGCTGCGTGTGGCT-3′, and reverse, 5′-GCTTCTCCTT-AATGTCACGCACGA-3′; α-SMA (636-bp product), forward, 5′-TGTGAAGAGGAAGACAGCACAGCT-3′, and reverse, 5′-GATGGCTGGAAGAGGGTCTCCGGG-3′; myoD (394-bp product), forward, 5′-AGCTCCAACTGCTCCGACGGCATGAA-3′, and reverse, 5′-AGGAGAGAATCC-AGTTGATGGAAACA-3′; myogenin (702-bp product), forward, 5′-GG-ACTGGGGACCCCTGAGCAT-3′, and reverse, 5′-GTTAAATTCCCTTGCTGGGCT-3′; and herculin or mrf4 (349-bp product), forward, 5′-CCGGCTGGATCAGCAAGAGAA-3′, and reverse, 5′-TAAGCTACTAGCCGTTCCGAT-3′ and 5′-TTCTCCACCACCTCCTCCACG-3′. 

Reverse Transcription-Polymerase Chain Reaction

Ten to 20 μg of total RNA ²¹ that had been purified by CsCl density-gradient centrifugation was treated with 10 U/μL DNase I and RNase inhibitor before phenol-chloroform extraction of RNA. The aqueous phase was collected and precipitated before washing, resuspension in diethyl pyrocarbonate (DEPC) H₂O, and reverse transcription. After denaturation at 70°C for 10 minutes, samples derived from pericytes grown under various test conditions were chilled on ice and brought up in an RT cocktail (10 μL cocktail/reaction = 4 μL 5× first-strand buffer, 2 μL 0.1 M dithiothreitol [DTT], 4 μL 2.5 mM dNTP). Samples were warmed at 37°C for 2 minutes and 1.0 μL (200 U) reverse transcriptase (SuperScript II; Invitrogen-Gibco, Gaithersburg, MD) was added before incubation for 60 minutes at 37°C and heat inactivation for 5 minutes at 95°C. For PCR, 4 μL RT reaction was added to 16 μL PCR cocktail and amplified with Taq polymerase for 38 cycles (94°C for 30 seconds, 62°C for 30 seconds, and 72°C for 1 minute, with a 10-minute 72°C extension) using the forward and reverse primers listed herein. The abundance of PCR products primed under these experimental conditions was quantified by quantitative analysis of ethidium bromide-stained bands electrophoresed in agarose gels (IS-1000 Digital Imaging System; Alpha Innotech, San Leandro, CA). All PCR products were sequenced in the Tufts University DNA/Protein Core Facility, to establish sequence identity and to confirm that the correct PCR products were obtained. 

Generation of α-SMA p125 Promoter Fragments

α-SMA promoter fragments were generated by PCR with DNA polymerase (Pfu turbo; Stratagene, La Jolla, CA). Primers used are as follows: 3′-CCCAAGCTTGGGACTCTGGGTGGGTGGTGTCT and 5′-CGGGGTACCCCGAGGTCCCTATATGGTTGTGT. Base pairs shown in italic denote sites for HindIII (3′ primer) and KpnI (5′ primer; Invitrogen-Gibco) digestion for subcloning into the HindIII-KpnI site of the pA₃ vector. Approximately 50 ng of α-SMA full-length promoter cloned into a vector (pCAT Basic; Promega, Madison, WI) was mixed with 1× buffer (Pfu; Stratagene), 0.2 mM dNTPs, 100 pM each primer, and 2.5 U DNA polymerase (Pfu turbo; Stratagene) in a total volume of 50 μL. Reactions were cycled for 30 rounds of the following conditions in a thermal cycler (PTC-100; MJ Research, Watertown, MA): 94°C for 1 minute, 60°C for 1 minute, and 72°C for 1 minute. Reaction products were run on 2% agarose/Tris-acetate-EDTA (TAE) gels, and for each construct, a parallel control lane was stained with ethidium bromide. The adjacent band corresponding to the correct size of each PCR fragment was then excised from the unstained lanes, electroeluted, phenol-chloroform extracted, and precipitated with ethanol. PCR products and approximately 500 ng pA₃ vector, were then digested with both HindIII and KpnI at 37°C overnight. Digested products were mixed (α-SMA p125 or α-SMA p300+pA₃), phenol-chloroform extracted, precipitated with ethanol, and resuspended in 15 μL H₂O and 4 μL 5× T4 ligase buffer (Invitrogen-Gibco) with 1 U T4 ligase (Invitrogen-Gibco). Ligations were performed at 16°C overnight. Five microliters of ligation product was electroporated into DH5α Escherichia coli hosts (model 1652076 Gene Pulser; Bio-Rad, Richmond, CA), and positive transformants were selected by standard methods. ²²  

Generation of Single-Stranded α-SMA 3′ Untranslated Region Probe for Northern Analysis

The 1361 rat α-SMA actin cDNA ²³ cloned into a plasmid (pBluescript II KS; Stratagene) was digested with DdeI (Promega) and used as a template in an asymmetric PCR reaction that contained approximately 50 ng of DdeI-digested α-SMA/plasmid DNA; 1.5 mM MgCl₂; 1× PCR buffer (Invitrogen-Gibco); 0.2 mM dATP, dTTP, and dGTP; 8 nM dCTP; 50 pM T3 primer; 1 μL ³²P dCTP (3000 Ci/mM; ICN, Costa Mesa, CA); and 2.5 U Taq DNA polymerase (Invitrogen-Gibco) in a total volume of 20 μL. Reactions were cycled for 40 rounds in the following conditions: 94°C for 1 minute; 50°C for 2 minutes, and 72°C for 2 minutes. PCR products were purified through probe purification columns (NucTrap; Stratagene), as described by the manufacturer. The final product was a single-stranded ∼191-bp fragment from the 3′ end of the α-SMA clone. 

Generation of Double-Stranded DNA Probes for EMSA and DNA-Binding Assays

For electrophoretic mobility shift assay (EMSA) and DNA-binding assay, double-stranded constructs were generated as follows. A probe comprising the first 125 bp of the rat α-SMA promoter was generated by PCR, as described earlier, using DNA polymerase (Pfu; Stratagene) and the primers listed herein, without the overhangs for the HindIII and KpnI sites. The 30-bp promoter fragments were synthesized as complementary single-stranded fragments by the Tufts Protein and DNA Sequencing Core Facility and were annealed by mixing equimolar amounts of each complementary strand, boiling for 5 minutes, and cooling slowly to room temperature. 

Northern Blot Analysis

Northern blot analysis was performed as described previously, ²⁴ with minor modifications. Pericytes grown in 150-mm plastic dishes in the presence of 10% serum-DMEM were downshifted to 0.01% serum-DMEM for 48 hours. The cells were then treated with 1 ng/mL TGF-β (Sigma-Aldrich) or 5 ng/mL FGF-2+1 μg/mL heparin for various periods. Total cellular RNA was extracted by density gradient centrifugation of guanidinium isothiocyanate lysates through 5.7 M CsCl cushions ²² in a rotor (model SW41; Beckman Instruments, Fullerton, CA) at 24,000 rpm in an ultracentrifuge (model L8-60M; Beckman Instruments). RNA pellets were resuspended in 400 μL DEPC-treated H₂O, extracted with phenol-chloroform and ethanol precipitated. For each condition, approximately 1 μg total RNA was run on a 1.2% agarose gel containing 3% formaldehyde. RNA was transferred to nylon membranes (Nytran; Schleicher and Schuell, Keene, NH) overnight in 10× SSC (1.5 M NaCl, 0.15 M Na citrate) by capillary phoresis. RNA was cross-linked to the membranes at optimal setting using a UV cross-linker (model FB-UVXL-1000; Fisher Scientific, Pittsburgh, PA), and the blot was probed for 24 to 48 hours with approximately 10⁶ cpm/mL of a single-stranded ³²P-labeled α-SMA 3′ untranslated region (UTR) fragment generated by asymmetric PCR, as described for the rat α-SMA-specific cDNA. After incubation with probe, the blot was washed at 65°C as follows: two times for 5 minutes each in 2× SSC and 0.1% SDS and one time for 5 minutes in 0.1× SSC. The blot was then air dried at room temperature and exposed to film at −80°C. Cells untreated with cytokine were used as the negative control. 

Isolation of Pericyte Nuclei

Nuclei were isolated from bovine retinal pericytes by a modification of the procedure of Workman. ²⁵ Nuclei were isolated intact by hypotonic lysis and NP-40 detergent extraction rather than by Dounce (Bellco Glass, Vineland, NJ) homogenization, and nuclear proteins were not extracted with high salt. Briefly, bovine retinal pericytes in 150-mm tissue culture plates (Corning Glass Co., Corning, NY) were cultured in 5% (for EMSA with a 125-bp fragment) or 10% (for EMSA with 30-bp fragments) bovine calf serum-DMEM for 48 hours and then downshifted to 1% serum, with or without 1 ng/mL TGF-β, for various periods. The cells were iced and washed twice with PBS. Cells were scraped into 5 mL PBS per plate and were pelleted at 4°C and 1500 rpm for 10 minutes in a bench-top centrifuge (RT6000B; Sorvall, Newtown, CT). The cell pellet was gently vortexed 5 seconds, 2 mL per plate hypotonic buffer (10 mM Tris [pH 7.5], 10 mM NaCl, 3 mM MgCl₂, 1 mM phenylmethylsulfonyl fluoride, 5 mM E-64 cysteine protease inhibitor, and 2.5 mM Na orthovanadate) was added drop-wise, and the cells were gently vortexed for 10 seconds. Resuspended cells were swollen for 5 minutes on ice, pelleted, and resuspended as before in hypotonic buffer with 1% NP-40. After 10 minutes on ice, the nuclei were pelleted and resuspended as before in hypotonic buffer with 1% NP-40 and were gently vortexed three times for 5 seconds each. Nuclei were pelleted and resuspended at approximately 10⁶/mL in glycerol storage buffer (50 mM Tris [pH 8.4], 5 mM MgCl₂, 0.1 mM EDTA, 20% glycerol, 1 mM phenylmethylsulfonyl fluoride, 5 mM E-64, 2.5 mM Na orthovanadate), snap frozen in liquid nitrogen, and stored at −80°C. 

Electrophoretic Mobility Shift Assays

EMSAs were performed according to the procedure described by Hautmann et al. ²⁶ A double-stranded α-SMA promoter fragment probe (entire 125-bp fragment of the rat α-SMA gene promoter or segments spanning −1 to −29, −30 to −59, −60 to −89, −90 to −125, and −202 to −235 [E-box2], or −236 to −269 [E-box1]) was end labeled with 1 μL γ[³²P]ATP (3000 Ci/mM) and 1 μL T4 polynucleotide kinase (Invitrogen-Gibco) as described by the manufacturer. End-labeled ³²P (∼7 × 10⁴ cpm) was incubated with approximately 5000 nuclei from bovine retinal pericytes that had been treated for various lengths of time with 1 ng/mL TGF-β (Sigma-Aldrich) in EMSA binding buffer (20 mM HEPES [pH 7.9], 50 mM KCl, 0.2 mM EDTA, 0.5 mM DTT, and 2% NP-40) in a total volume of 20 μL 0.7 μg poly dA dT (Amersham Pharmacia Biotech, Piscataway, NJ) was added to eliminate nonspecific protein-DNA interactions. After 20 minutes at room temperature, complexes were separated on 4.5% nondenaturing polyacrylamide gel and visualized by autoradiography. For competition reactions, nuclei were incubated with an approximate 100× molar excess (over labeled DNA fragment) of an unlabeled promoter fragment for 10 minutes before incubation with the labeled one. 

Localization Studies

In Vitro.

For localization of FGF-2, FGFR, myf-5, or the isoactins, first- or second-passage pericytes were plated onto glass coverslips in the bottom of 24-well plates or directly onto culture chamber slides (Tissue Tek; Sakura Finetek, Torrance, CA) in DMEM supplemented with serum, FGF-2, or TGF-β1, as previously published. ^{19 20} Duplicate or triplicate samples from several different retinal isolations (n > 6) were fixed after treatment with various growth factors for different lengths of time (0 minutes to several hours; doses and control ligands are identical with those evaluated for pericyte growth studies). Pericytes were fixed for 5 minutes at room temperature in a freshly prepared solution of 4% formaldehyde in DMEM, permeabilized in 0.1% Triton X-100 buffered with 50 mM HEPES (pH 7.1), 50 mM PIPES (piperazine-N-N′-bis(2-ethanesulfonic acid; pH 6.9), 0.1 mM EGTA, 75 mM KCl, and 1.0 mM MgCl₂ for 90 seconds at room temperature before equilibration in PBS. For antibody localization studies, primary antibodies were used at 10 to 50 μg/mL in PBS and applied to cells for 1 hour at room temperature. Secondary antibodies were also applied for 1 hour at room temperature. Texas red, fluorescein, or rhodamine-labeled goat anti-rabbit or mouse IgG (100 μg/mL; Cooper Biomedical, Gaithersburg, MD) were used as previously described. ^{19 20} Quantitative analyses of myf-5 and α-SMA positive pericyte cultures were performed in a blind fashion on triplicate cultures for each time point over the time course of the experiment. Each set of experiments was repeated at least five times. For myf-5 colocalization, positively stained nuclei were digitally counted and expressed as a percentage of the total nuclear pool, identified by DNA labeling with 4′,6′-diamino-2-phenylindole (DAPI). For digital image analysis of immunofluorescence from cultured cells, nuclear fluorescence (for anti-myf-5 localization) and stress-fiber-associated fluorescence (α-actin IgG localization) were quantified directly on the stage of an inverted fluorescence microscope (Axiovert; Carl Zeiss Meditec, Thornwood, NY) equipped with high-resolution optics (Planapochromat, numeric aperture 1.4; Carl Zeiss Meditec) interfaced with a computer-based microscope imaging workstation (MetaMorph; Universal Imaging Co., Downingtown, PA) and an intensified target video camera (ISIT 66; Dage-MTI, Michigan City, IN). Digitized fluorescence from experimental and control groups was plotted along with standard error on computer (KaleidaGraph; Synergy Software). 

Western Blot Analysis.

For Western blot analysis, cell lysates were derived from pericytes treated as described ^{14 27 28} before electrophoresis and transfer to nitrocellulose. After blocking for 1 hour at room temperature in Tris-buffered saline-Tween 20 (TBST) containing 5% fat-free milk, membranes were incubated in antibodies, as previously described. ²⁸ Blots were washed extensively in TBST before incubation in secondary antibodies (1:3000 dilution, horseradish-peroxidase [HRP]-conjugated). After another hour of room temperature incubation, blots were washed and developed. Signals were detected using enhanced chemiluminescence, as described by the manufacturer (Super Signal; Pierce Chemical Co, Rockford, IL). 

Results

Endothelium-Derived Growth Regulators, Pericyte Growth, and Contractile Phenotype

Earlier work has demonstrated an abundance of pericytes within specific microvascular beds. ²⁹ Because there is a 1:1 correspondence between the presence of pericytes and the localization of FGF-2 within the cerebrovascular basement membrane, ⁵ we hypothesized that endothelium-synthesized and matrix-bound FGF-2 plays a pivotal role in either signaling the recruitment of pericyte precursors or fostering pericyte proliferation. Because pericytes occupy the basal lamina along with the microvascular endothelium, presentation of either free FGF-2 or FGF-2-bound heparin sulfate proteoglycan would place the mitogen advantageously for signaling pericyte growth. When purified populations of retinal pericytes were plated in tissue culture medium in the presence of nanogram/milliliter doses of FGF-2, there was a rapid and marked induction of pericyte growth that was more than two times higher than in control cells (Fig. 1) . FGF-2 stimulation of pericyte growth was markedly enhanced when FGF-2 was added in combination with heparin (Fig. 1) , causing a four- to sixfold potentiation of pericyte growth over control cells. This potentiation of FGF-2-induced growth was not observed when other sulfated glycosaminoglycans were added simultaneously with FGF-2. For example, pericytes cultured with identical concentrations of chondroitin 4-sulfate together with FGF-2 showed no significant difference from cells grown with FGF-2 alone. In addition, FGF-2 growth stimulation was completely obliterated when a molar excess of anti-FGF-2 was added simultaneously to pericyte cultures, thus neutralizing the FGF-2-mediated growth response. ⁵ Conversely, addition of nanogram/milliliter doses of TGF-β1 caused pericyte growth arrest within several hours after its addition to serum-containing media, regardless of whether low (0.5%) or high (5%–10%) concentrations of serum were used (Fig. 1) . 

Having demonstrated that soluble and matrix-bound growth regulators play an important role in stimulating pericyte proliferation, we recognized that removal of serum and/or FGF-2 from the culture medium might cause pericytes to withdraw from the cell cycle and differentiate, akin to what occurs in the cell culture conditions that have been established for launching skeletal myogenesis. ^{30 31} If similar inductive events were regulating commitment and differentiation in pericytes, we would predict that vascular smooth muscle contractile protein gene expression would be switched on as the encoded contractile protein isoforms accumulated coincident with growth arrest. Indeed, isoelectric focusing and Western blot analysis, as well as immunofluorescence microscopy with isoactin-specific antibodies, revealed that whole-cell extracts and fixed pericyte cultures were devoid of α-smooth muscle contractile protein isoforms when pericytes were grown in the presence of FGF-2. In contrast, when cells were withdrawn from media containing FGF-2 (not shown) or when pericytes are cultured in the absence of FGF-2, but in the presence of TGF-β1, a known inducer of α-SMA expression, smooth muscle contractile protein isoforms accumulated (Fig. 2) . Thus, when pericytes were forced to differentiate by withdrawal of FGF-2 or addition of TGF-β1, they expressed α-SMA protein. 

Effect of TGF-β1 on the Level of α-SMA RNA in Retinal Pericytes

To determine the mode(s) of TGF-β1 regulation of α-SMA under the aforementioned experimental conditions, we first performed Northern blot analysis on total cellular RNA isolated from bovine retinal pericytes treated with TGF-β1. A radiolabeled DNA fragment specific for the 3′ UTR of α-SMA ³² was used as a probe. TGF-β1 induced abundant α-SMA mRNA expression, whereas FGF-2/heparin stimulated little if any accumulation of α-SMA mRNA (Fig. 3A) . The TGF-β1-stimulated increase in α-SMA mRNA plateaued after 4 hours to approximately 2.5-fold over the level observed in unstimulated cells and began to decline to baseline levels after 14 hours (Fig. 3B) . 

Effect of TGF-β1 on Protein Binding to the α-SMA Promoter

Because TGF-β1 stimulated an increase in the steady state levels of α-SMA mRNA, we hypothesized that TGF-β1 does this through transcriptional control mechanisms, perhaps by inducing transcription factors to bind to the α-SMA promoter. Because the first 125 bp of the promoter support TGF-β1-driven transcription in rat smooth muscle cells, ²⁶ this construct was prepared and end labeled with ³²P for use as a probe in an EMSA, using nuclei isolated from retinal pericytes grown under control conditions or after stimulation with TGF-β1. Using this construct, a band shift was evident as early as 0.5 hour after stimulation with TGF-β1 (Fig. 4A) . This shifted band increased in intensity by 1 hour, then decreased between 4.5 and 14 hours of TGF-β1 treatment. The shifted complex was specific for the 125-bp promoter fragment, as an approximate 100-fold molar excess of unlabeled probe completely blocked binding (Fig. 4A) . 

To further resolve where proteins are binding to the 125-bp α-SMA promoter fragment, EMSA was performed with 29- to 30-bp oligomers spanning the first 125 bp of the α-SMA promoter. Because a band shift was seen in an EMSA after 0.5 hour of TGF-β1 stimulation with the 125-bp promoter construct (Fig. 4A) , end-labeled oligomers were incubated with nuclei from retinal pericytes treated with TGF-β1 for 30 minutes after downshifting from 10% serum+FGF-2/heparin. Band shifts were evident with each probe used, and each was competitively blocked by an approximate 100-fold molar excess of unlabeled probe (Fig. 4B) . To further test the specificity of the band shifts, the same EMSA was performed, except that the approximate 100-fold molar excess of an unlabeled nonspecific probe was used as the competitor. Most of the band shifts generated by labeled probes spanning -1 to −29 and −60 to −89 were competitively blocked by unlabeled E-box1, indicating that they are probably representative of nonspecific DNA binding (Fig. 4C) . Furthermore, band shifts produced by labeled probes spanning each E-box were competitively blocked by the unlabeled segment spanning −1 to −29. Nevertheless, a few band shifts were not competitively blocked. These include the band shifts using the −90 to –125 probe (Figs. 4B 4C , arrowhead) and the −30 to −59 probe (Figs. 4B 4C , arrow). The −30 to −59 probe encompassed the TGF-β control element that is necessary for TGF-β-induced α-SMA promoter activation of transcription in rat smooth muscle. ²⁶ These results suggest that TGF-β1 induces the binding of transcription factors to segments −30 to −59 and −90 to −125 of the retinal pericyte α-SMA promoter. 

Effect of TGF-β1 on the Level and Nuclear Translocation of Myf-5 RNA

TGF-β1 induction of specific α-SMA promoter band shifts (Fig. 4) suggests that this cytokine stimulates nuclear accumulation of factors necessary for binding and transactivating the α-SMA promoter. In an effort to learn whether the myoD family of bHLH transcriptional regulators might be the molecular regulators controlling downstream smooth muscle contractile protein gene expression induced by TGF-β1, we performed RT-PCR with primers sets specific for several of the well-studied skeletal muscle myogenic bHLH transcriptional regulators. RT mixes were prepared from mRNAs isolated from pericytes grown under a variety of conditions, including different concentrations of serum and serum containing FGF-2 (with and without heparin) or TGF-β1. Results of these RT-PCR experiments revealed conclusively that myf-5 mRNA levels (as seen by RT-PCR) were increased nearly fourfold when pericytes were removed from media containing FGF-2 and cultured in 0.2% serum alone or with TGF-β1 (Fig. 5) . Further, there was an 18-fold increase in the predicted myf-5 PCR product when pericytes were first stimulated to proliferate under the control of FGF-2 and heparin and then placed in low-serum medium (Fig. 5) . Similarly, TGF-β1 alone induced myf-5 expression (described later). Neither myf-5 mRNA or protein nor any of the other mRNAs encoding other myoD family members was detected in pericytes growing in the presence of FGF-2 or FGF-2/heparin. Only myf-5, but none of the other myoD family members, was induced after the withdrawal of FGF-2 or the addition of TGF-β1. Neither vascular endothelial cells nor smooth muscle cells expressed detectable myf-5 (or any myoD family member), regardless of whether mRNAs were isolated from growing or growth-arrested cells. Furthermore, β-actin mRNA levels were not significantly altered in this set of experimental conditions (Fig. 5) . 

Using antibodies specific for myf-5 and α-SMA, we were interested in learning whether the accumulation of these respective proteins occurs concomitant with growth arrest. Indeed, immunofluorescence microscopy and quantitative image analysis of the anti-myf-5-treated cultures revealed that myf-5 accumulated in pericyte nuclei after the withdrawal of cells from heparin and FGF-2 or the addition of TGF-β1 (Fig. 6) . Double-immunofluorescence microscopy using the myf-5- and α-SMA-specific antibodies revealed that the appearance of nuclear myf-5 preceded the emergence of α-SMA-containing stress fibers and persisted during their emergence (Fig. 7) . 

Smad2 Translocation into the Nucleus

The TGF-β receptor is a dimeric complex composed of type I and II serine/threonine kinase subunits. Receptor type II binds TGF-β and phosphorylates receptor type I, and phosphorylated type I receptor in turn phosphorylates and activates the Smad proteins. ³³ The type III receptor may facilitate the binding of TGF-β to the type II receptor. ³⁴ The Smad proteins are effectors of the activated TGF-β receptor. On phosphorylation by activated type I receptor, Smads translocate into the nucleus and may directly bind to target gene promoters to activate gene transcription. ³⁵ Therefore, the Smads are likely candidates to bind and activate the α-SMA gene promoter in response to TGF-β. To determine whether Smads are involved in TGF-β1-mediated transactivation of the α-SMA promoter, we first generated polyclonal antibodies to Smad2, -3, and -4. Specific antibodies were obtained from serum of rabbits that were injected with synthetic polypeptides corresponding to the most nonconserved regions in the Smad hinge region (Fig. 8) . The specificity of these antibodies was assessed by using them to probe Western blot analysis of Smad-overexpressing COS cell lysates. Rabbit antisera to Smad2 recognized a specific band that corresponded to the correct molecular weight of this polypeptide (Fig. 9A) . 

To determine the localization of Smad proteins in pericytes, we performed immunofluorescence on formaldehyde-fixed, permeabilized retinal pericytes, using the Smad-specific rabbit IgG. In retinal pericytes asynchronously growing in 10% serum, the antibody to Smad2 showed bright nuclear fluorescence (Fig. 9B) , as well as several cells that exhibited diffuse cytoplasmic staining without nuclear fluorescence (Fig. 9B , arrowheads). This heterogeneity in anti-Smad2 IgG localization suggests that in asynchronously growing cells in high-serum medium, some cells actively cycle, whereas others are committed to the smooth muscle lineage. To test this hypothesis, retinal pericytes were synchronized by downshifting to 0.01% serum for 48 hours, and Smad2 was localized after stimulation with TGF-β1 for various times. After 5, 10, and 15 minutes of TGF-β1 treatment, Smad2 appeared as puncta surrounding the nucleus (Fig. 9C , arrowhead) or was diffusely distributed in the cytoplasm (Fig. 9C , arrow, 15 min). After 30 minutes of TGF-β1 treatment, however, there was an exclusive and robust nuclear fluorescence of Smad2. Therefore, we conclude that endogenous pericyte Smad2 translocates into the nucleus in response to TGF-β1 treatment of retinal pericytes. 

Discussion

The results of these experiments revealed that pericyte growth and differentiation are differentially regulated by antagonistic matrix- and cell-associated signaling cascades. In vivo, these signals are likely to originate from the microvascular endothelium, its associated matrix, or bound growth regulators. The data described herein also help to establish the pivotal role that endothelial cell differentiation plays in modulating microvascular morphogenesis, especially at the level of perivascular cell recruitment and commitment to its smooth muscle phenotype. 

Matrix-Bound Growth Regulators, Receptor Kinases, and Pericyte Proliferation

Earlier work has revealed that endothelium-derived extracellular matrix can alter pericyte growth and contractile phenotype. ^{20 36} It is also well established that specific subpopulations of microvascular endothelial cells possess unique arrays of matrix-bound growth regulators (e.g., FGF-2). ^{5 37 38} When these specific populations of vascular endothelial cells are grown in tissue culture, accumulation of matrix-bound FGF-2 occurs, though only after endothelial cell-cell contact, as the cells are withdrawing from the growth cycle. ⁵ This suggests a role for FGF-2 beyond signaling the proliferation of endothelium. One possibility is that FGF-2 also influences growth and differentiation of the closely associated pericytes. These observations are in good agreement with the reported abundance of perivascular cells in the brain, ^{17 19} retina, ^{18 39 40} and kidney ^{5 41} (e.g., in the kidney, mesangial cells associate with glomerular capillary basement membranes enriched in heparin sulfate-associated FGF-2 ^{5 41} ). Together, these findings suggest that FGF-2 serves as a pericyte mitogen and/or motogen, signaling growth through its receptor tyrosine kinase (RTK). Clearly, FGF-2 and its receptor are colocalized and are seemingly essential for activation of cell cycle progression and pericyte growth. Furthermore, this and other pivotal vascular signaling cascades (e.g., PDGF-bb and TGF-β1) may be cooperative or antagonistic in mediating pericyte growth and microvascular remodeling during development or in association with vasoproliferative disorders. 

Collectively, our findings help to explain the manner in which pericytes are selectively recruited to specific microvascular beds and further identify the parallel growth-promoting RTK signaling pathways responsible for stimulating pericyte proliferation. In addition, because pericyte proliferation occurs when vascular smooth muscle contractile protein gene and protein expression are seemingly repressed, ²⁰ it calls into question the usefulness of smooth muscle-specific probes as markers for pericytes when the cells are actively proliferating. ⁴²  

Myf-5 and Pericyte Myogenic Determination

During the past decade, considerable insights into the molecular mechanisms controlling skeletal myogenesis have been gained. ^{43 44 45 46} Yet, despite this level of molecular understanding with regard to skeletal myogenesis, much less is known about the lineage, recruitment, and molecular switching events that regulate vascular smooth muscle commitment and differentiation. Our results point to a new role for myf-5, the earliest of the bHLH transcriptional regulators to be expressed, in helping to establish the smooth muscle phenotype in retinal pericytes. That myf-5 mRNA and protein levels are markedly induced in pericytes, but not in vascular smooth muscle cells, suggests that the myogenic precursor pools or the molecular signals responsible for seeding the vasculature with its mural components are distinct. This notion of site-specific myf-5 expression in specific perivascular precursor pools is consistent with recent observations regarding the temporal and spatial myf-5 expression patterns observed during murine embryogenesis, when visceral arch and somitic myf-5 expression patterns differ. ⁴⁷ Together with our results, these findings suggest that there are multiple and spatially segregated regulatory domains and pathways that may actively influence myf-5 expression in committed skeletal precursors and now, in pericyte precursors. Whereas a role for myf-5 in regulating perivascular cell commitment has not been considered, our results reflect a need for the reexamination of these transgenic animals, especially at the single-cell level and, more important, within the various embryonic and postnatal microvascular beds. This is especially important when considering the retinal microvasculature, which undergoes extensive postnatal remodeling, both normally and in association with pathologic angiogenesis. Clearly, these studies have helped to divulge important details in the molecular signaling pathways presumed to be responsible for regulating perivascular cell commitment and differentiation, but future work should yield new insights into the embryologic origins and developmental connections that presumably exist among pericyte, smooth muscle, and skeletal muscle precursors. 

TGF-β1 as an Inducer of Pericyte Commitment and Differentiation

TGF-β1, a member of the TGF gene family, represents the prototype of this multimember family, coordinating a variety of activities, including the regulation of growth, differentiation, and extracellular matrix production. ^{48 49 50 51 52 53 54 55 56 57} Its widespread expression during embryonic development points to its important role in orchestrating epithelium-mesenchyme interactions, angiogenesis, and chondro- and osteogenesis. Only recently, through the targeted disruption of TGF-β1, have these essential role(s) been more clearly delineated. ^{55 56 58 59} TGF-β1 signals through two transmembrane serine/threonine kinases (STKI and STKII), which function together and share approximately 40% homology. ^{60 61 62} Each STK possesses an extracellular cysteine-rich domain responsible for fostering ligand binding, a single transmembrane domain, and an intracellular serine/threonine kinase domain responsible for intracellular signaling through cytoplasmic phosphomediators (Smads). Our recent discovery that TGF-β1 signaling may orchestrate the onset of pericyte differentiation not only indicates the putative importance of STK signaling during retinal microvascular development, but it seems evident that aberrations in TGF-β1 signal transduction may also be critically involved in regulating pericyte dedifferentiation during the onset of disease, in consideration of the known alterations in the retinal vasculature that occur during diabetes. 

We have implicated the Smads as mediators of TGF-β1-driven α-SMA promoter transactivation. TGF-β1 treatment of retinal pericytes drives Smad2 translocation into the nucleus after 30 minutes (Fig. 9) , after which α-SMA mRNA accumulates (Fig. 3) . Mounting evidence suggests that Smads can bind to DNA and activate transcription from target promoters. Drosophila Mad ⁶³ and mammalian Smad3 and Smad4 bind DNA and activate transcription in vitro, ^{64 65} and a Smad3/4 complex binds the consensus sequence GTCTAGAC, which is found in promoters of many TGF-β-responsive genes. ^{35 66} Specificity in Smad activation of target genes, however, is due to the recruitment of accessory transcription factors by Smads to promoters. Smad2/4 and Smad3/4 form complexes with the winged-helix protein FAST-1 on promoters of Xenopus ⁶⁷ and human ⁶⁸ homeobox genes activated by TGF-β and activin-like factors. ^{69 70} Smads have also been shown to form complexes with activator protein (AP)-1 to activate transcription of artificial promoters in vitro. ⁷¹ In addition, Smads can recruit coactivators such as p300 and CREB-binding protein, ^{72 73} as well as corepressors, ⁷⁴ in response to TGF-β. Therefore, the various effects of TGF-β are regulated by the cellular context of Smad as it interacts with different proteins to form distinct transcriptional complexes. Perhaps Smads act in concert with myf-5 to induce pericyte α-SMA gene promoter transactivation and protein expression. In this sense, then, myf-5 would confer specificity in TGF-β1 signaling to the myogenic lineage-derived pericyte. 

The consensus Smad binding element AGAC, defined by Hua et al., ⁷⁵ is found at −42 of the α-SMA promoter used in these experiments. Because region −30 to −59 was bound by a nuclear protein(s) from TGF-β1-treated pericytes (Fig. 4B) and Smad2 translocated from nucleus to cytoplasm in response to TGF-β1 treatment, it is possible that Smad 2 binds to this element and recruits the accessory proteins necessary for a complete transcriptional unit. We were unsuccessful in supershifting the specific band shift in the EMSA (Fig. 4) with Smad antisera. However, preliminary experiments show that mutation of 2 bp within a Smad consensus DNA binding element ^{35 66} within the first 125 bp of the α-SMA promoter results in a drastic reduction in the α-SMA promoter-driven reporter gene expression driven by TGF-β1 (data not shown). Flanking this element on the 5′ end is the 10-bp “TGF-β-control element” defined by Hautmann et al. ²⁶ Characterization of the proteins binding to these regions of the α-SMA promoter may elucidate the mechanism of TGF-β1-stimulated α-SMA gene transcription. 

Nevertheless, more work is needed to establish the exact signaling hierarchy between TGF-β1 signaling and pericyte α-SMA expression. For example, it is possible that TGF-β1 signaling intermediates activate both myf-5 and α-SMA gene transcription. Detailed molecular analysis of these intermediates is likely to reveal the manner in which TGF-β1 acts as a molecular inducer of the pericyte smooth muscle contractile phenotype. 

Antagonistic Signaling Cascades in Modulation of Pericyte Growth and Contractile Phenotype

Our observations suggest that pericyte growth and differentiation are inversely related. A simplistic explanation is that molecular control of either pericyte growth or differentiation is the sum of the positive and negative signals from the vascular basement membrane, with each signaling cascade targeting the regulators of cell cycle progression on the one hand or withdrawal from the cell cycle and differentiation on the other. In fact, there are reports that stimulation of growth by FGF-2 is mediated by phosphorylation of the retinoblastoma susceptibility gene product RB, which is thought to restrict cell cycle progression through late G₁. ^{76 77 78 79} Furthermore, based on our finding that myf-5 mediated pericyte α-smooth muscle contractile protein gene expression (Figs. 5 6 7) , our results suggest that commitment and differentiation occur simultaneously as pericytes withdraw from the growth cycle. Other work in skeletal myogenesis indicates that growth arrest, induction of the CDK inhibitor p21, and skeletal myogenesis ensue after expression of myoD and myf-5. ⁸⁰ This has been proposed as a protective mechanism guarding the differentiated skeletal myocyte against apoptosis. This phenomenon may be evolutionarily conserved, as it is seemingly in place during pericyte myogenesis. 

Our findings demonstrate that FGF-2 and heparin promote pericyte growth (Fig. 1) and suppress α-smooth muscle contractile protein gene expression (Fig. 2) , whereas TGF-β1 caused growth arrest (Fig. 1) and the induction of the contractile phenotype (Fig. 2) in vitro. These observations are consistent with the robust α-SMA mRNA and protein expression occurring “downstream” of Smad 2 accumulation in the nucleus (Fig. 9) . Taken together, these findings lend strong credence to the notion that positive and negative endothelium-derived growth regulators act to control the patterned placement of growing versus contractile pericytes within the retinal microcirculatory bed in vivo (Fig. 10) . Furthermore, in consideration of the role that TGF-β1 signal transduction plays in fostering retinal pericyte differentiation (presumably through induction of smooth muscle gene transcription), it seems likely that other downstream targets are also affected. Because this matrix-driven and kinase-mediated phenotypic modulation could result in local pericyte dedifferentiation, pericytes would be neither contractile nor capable of restraining endothelial cells from reentering the cell cycle in a dedifferentiated state. As such, endothelial cell proliferation and angiogenesis could ensue. Our results predict that perturbations in these antagonistic, matrix-mediated signaling cascades, rather than pericyte death, per se, may serve as the early signals that elicit the recurrent rounds of endothelial proliferation observed during pathologic angiogenesis. 

 Supported in part by National Eye Institute Grant EY09033 (IMH).

 Submitted for publication March 20, 2003; revised May 27 and June 27, 2003; accepted August 7, 2003.

 Disclosure: M. Papetti, None; J. Shujath, None; K.N. Riley, None; I.M. Herman, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

FGF-2 and TGF-β1 modulated pericyte growth and contractile phenotype. Proliferation assays were performed in triplicate 24-well plates over a 4-day time course. FGF-2, FGF-2+heparin, TGF-β1, or serum alone was added, and the number of cells was counted directly. The mean counts ± SE from triplicate wells and several experiments (n > 6) are shown.

Figure 2.

 

Addition of TGF-β1 induced pericyte expression of α-SMA. Shown are bovine retinal pericytes cultured as described and stained for α-SMA, after being grown in DMEM with 10% serum (A) or in low serum in the presence of 1 ng/mL TGF-β (B), which caused an increase in the formation of α-SMA-containing stress fibers.

Figure 3.

 

TGF-β1 induced an increase in α-SMA mRNA in retinal pericytes. Bovine retinal pericytes were cultured in 10% bovine calf serum/DMEM and were downshifted to 0.01% serum/DMEM for 48 hours. (A) Downshifted pericytes were cultured in the presence of either 1 ng/mL TGF-β1 or 5 ng/mL FGF-2+1 mg/mL heparin for 3 hours, and Northern blot analysis was performed on approximately 1 mg of RNA from each condition, using a ³²P-labeled probe specific for α-SMA. (B) The cells were then cultured in the presence of 1 ng/mL TGF-β1 for the indicated lengths of time, and Northern blot analysis was performed, as in (A).

Figure 4.

 

Proteins from TGF-β1-treated retinal pericytes bind the α-SMA promoter. (A) Bovine retinal pericytes were cultured in 5% serum+5 ng/mL FGF-2+1 μg/mL heparin for 48 hours. The cells were then washed and treated with 1 ng/mL TGF-β1 for the times indicated. Nuclear lysates were incubated with the ³²P-labeled α-SMA promoter fragment p125, and DNA complexes were separated on 4.5% polyacrylamide gels by native-PAGE. Where indicated, reactions included a 100× molar excess of unlabeled promoter fragment p125. Arrowhead: free probe; arrow: shifted complex. (B) EMSA was performed as in (A), but pericytes were first incubated in 10% serum+5 ng/mL FGF-2+1 μg/mL heparin, and nuclei from pericytes treated with 1 ng/mL TGF-β1 for 30 minutes were incubated with the indicated ²³P-labeled α-SMA promoter fragments. Where indicated, reactions included a 100× molar excess of unlabeled specific promoter fragment. Arrowhead: free probe. (C) EMSA was performed as in (B), but reactions included a 100× molar excess of the unlabeled nonspecific promoter fragment indicated. Arrow and arrowhead in (B) and (C) show band shifts created by probes 30 to 59 and 90 to 125, respectively, that were not completely competed by excess E-box1.

Figure 5.

 

Myf-5 induction coincides with pericyte growth-arrest. Primers were designed, RNA purified and RT-PCR. Top: myf-5; bottom: β-actin. Lane 1: 100 bp ladder; lane 2: 10% calf serum; lane 3: FGF-2; lane 4: FGF-2 treatment before growth factor and serum withdrawal (0.2%); lane 5: FGF-2+heparin; lane 6: FGF-2+heparin treatment before growth factor and serum withdrawal; lane 7: endothelial cells; lane 8: vascular smooth muscle cells. Note the induction of the myf-5 PCR product (732 bp) after the withdrawal of FGF-2 or FGF-2 and heparin. Neither endothelial cells nor smooth muscle cells expressed myf-5. β-Actin control shows 373-bp PCR product. Ethidium bromide staining.

Figure 6.

 

Quantitative analysis of myf-5 expression. Anti-myf-5 immunofluorescence was performed (Figs. 8 9) . After fixation and viewing, cells expressing nuclear myf-5 were digitized. Cells that possessed nuclear myf-5 were then expressed as a percentage of the total population counted. Results are from several experiments (n > 5) in which more than 100 cells were counted for each condition, plated in triplicate. Error bars denote SE. Note that FGF-2 and heparin treatment of pericytes blocked nuclear myf-5 accumulation, whereas TGF-β1 treatment induced myf-5 nuclear localization.

Figure 7.

 

Myf-5 accumulated in the nucleus concomitant with growth arrest. Pericytes were grown in the presence of 10% serum or FGF-2/heparin. Cells were either maintained under these conditions or washed and changed to TGF-β1-containing medium. Cells were then fixed and prepared for anti-myf-5 immunofluorescence. When pericytes were cultured in the presence of serum (A) or FGF-2/heparin (B), anti-myf-5 fluorescence was cytoplasmic. However, after the switch from 10% serum to TGF-β1-containing medium (C), anti-myf-5 fluorescence was nuclear. Confocal imaging confirmed the nuclear localization of myf-5 in pericytes treated with TGF-β1 (D), and stained simultaneously for myf-5 (green) and α-SMA (red).

Figure 8.

 

Schematic of peptides used for rabbit injections to generate polyclonal Smad antibodies. Generalized domain structures of Smad2, -3, -4, -6, and -7 are shown. LVKKLKK represents the conserved nuclear localization signal present in Smad2 and -3, and SS(V/M)/S is the conserved C-terminal site of phosphorylation by TGF-β receptor type I. MH1 and MH2 refer to the highly conserved N- and C-terminal domains of Smad2, -3, and -4. The divergent linker regions between MH1 and MH2 regions are also shown. Sequences of synthesized peptides used to inject rabbits for antibodies are indicated below each diagram.

Figure 9.

 

TGF-β1 induced Smad2 nuclear accumulation in retinal pericytes. (A) Western blot, performed on whole-cell lysates from transfected COS cells, demonstrating the specificity of rabbit anti-Smad2 antibody. (B) Bovine retinal pericytes were seeded onto glass coverslips and continuously grown in the presence of 10% serum. Arrowheads: diffuse cytoplasmic localization of Smad2. (C) Cells were grown in 10% serum and downshifted to 0.01% serum for 48 hours. The downshifted cells were then treated with 1 ng/mL TGF-β1 for the periods indicated. Arrowheads: punctate perinuclear staining of anti-Smad2; arrow: diffuse cytoplasmic Smad2 localization.

Figure 10.

 

Diagram depicting a suggested pathway of FGF-2 and TGF-β1 induction of pericyte growth and differentiation during microvascular morphogenesis. Pericyte recruitment occurs after endothelial deposition of FGF-2 in the extracellular basement membrane. As pericytes begin contacting endothelial cells, TGF-β1 becomes activated, therein fostering endothelial growth arrest as well as causing the induction of myf-5 and pericyte smooth muscle contractile protein gene expression. This smooth muscle contractile phenotype maintains the endothelium in its growth-arrested state until aberrations in the extracellular compartment prevent endothelial cell-pericyte contacts. In turn, pericyte and endothelial cell proliferation ensues, launching recurrent rounds of pathologic angiogenesis.

The authors thank Nicholas Divaris, Justin Glotfelty, and Benson Chen for technical assistance; Amy Yee and Craig Dionne for the gift of antibody reagents, and Pat D’Amore and Anita Roberts for collaborative support and critical review of the manuscript.