- Email: [email protected]
Platelet Derived Growth Factor (PDGF) Responsive Epidermis Formed from Human Keratinocytes Transduced with the PDGFβ Receptor Gene
ORIGINAL ARTICLE
Platelet Derived Growth Factor (PDGF) Responsive Epidermis Formed from Human Keratinocytes Transduced with the PDGFb Receptor Gene Ola Rollman, U¡e B. Jensen,n Arne stman,w Lars Bolund,n Sigru¤n M. Gu¤stafsdo¤ttir,z and Thomas G. Jensenn Department of Medical Sciences, Dermatology and Venereology, Uppsala University Hospital, Uppsala, Sweden; nInstitute of Human Genetics, Aarhus University, Aarhus, Denmark; wLudwig Institute for Cancer Research, Biomedical Center, Uppsala, Sweden; zDepartment of Genetics and Pathology, Rudbeck Laboratory, Uppsala, Sweden
Platelet-derived growth factor is a major proliferative and migratory stimulus for connective tissue cells during the initiation of skin repair processes. In response to injury, locally produced platelet-derived growth factor is secreted by a diversity of cutaneous cell types whereas target activity is con¢ned to cells of mesenchymal origin, e.g. dermal ¢broblasts and smooth muscle cells. Although epidermal cells contribute to cutaneous platelet-derived growth factor activity by their ample capacity to secrete platelet-derived growth factor ligand, normal epidermal keratinocytes are not known to express any member of the platelet-derived growth factor receptor family. In order to study if epidermis may be genetically transformed to a platelet-derived growth factor sensitive compartment we aimed to introduce the gene encoding human platelet-derived growth factor receptor b (PDGFbR) into epidermal keratinocytes using a retrovirus-derived vector. Successful gene transfer to primary cells was con¢rmed by immuno£uorescence staining, southern blotting, and ligand-induced receptor autophosphorylation. By culturing a mixture of PDGFbR-transduced and unmodi¢ed keratinocytes at the air^liquid interface on devitalized dermis, we were able to establish a multilayered epithelium showing histologic similarities to that evolved from native
keratinocytes or keratinocytes transduced with the reporter gene encoding enhanced green £uorescent protein. Receptor-modi¢ed epidermal tissue cultured for 6 days and examined by immuno£uorescence microscopy was shown to contain PDGFbR-expressing keratinocytes distributed in all layers of living epidermis. By continued tissue culture in serum-containing medium, the epidermis became increasingly corni¢ed although receptor-positive cells were still observed within the viable basal compartment. Stimulation of PDGFbR-transduced epidermis with recombinant platelet-derived growth factor BB had a mitogenic e¡ect as re£ected by an increased frequency of Ki-67 positive keratinocytes. The study demonstrates that transgene expression of human PDGFbR can be achieved in epidermal keratinocytes by retroviral transduction, and that ligand activation of such gene-modi¢ed skin equivalent enhances cell proliferation. In perspective, viral PDGFbR gene transfer to keratinocytes may be a useful approach in studies of receptor tyrosine kinase mediated skin repair and epithelialization. Key words: genetic transduction/green-£uorescent protein/PDGF receptor tyrosine kinase/wound healing/proximity ligation. J Invest Dermatol 120:742 ^749, 2003
S
¢broblasts, and smooth muscle cells, contribute to PDGF production in the injured microenvironment (reviewed in Heldin and Westermark, 1999). Mesenchymal cells in the skin express PDGF-speci¢c cell surface receptors, which convey autocrine or paracrine responses via activation of the signal transduction network (Heldin et al, 1981; Reuterdahl et al, 1993). Under physiologic conditions, epidermal cells do not express transmembrane receptors recognizing any classical PDGF isoform (AA, AB, and BB dimers); thus target cells for these ligands in normal skin seem restricted to dermal constituents exclusively. Animal data indicate that delayed skin repair is associated with reduced expression of both PDGF ligands and their corresponding receptors (Beer et al, 1997). Hence, attempts have been made to stimulate cutaneous healing by boosting the PDGF activity in the wound environment (Moulin et al, 1998). Human studies in diabetic lower extremity ulcer (Steed, 1995; Embil et al, 2000; Margolis et al, 2000) and decubital ulcer (Robson et al, 1992) have
kin tissue repair is regulated by intrinsic peptides such as platelet-derived growth factor (PDGF), a mitogenic hormone secreted at sites of trauma. Locally released PDGF attracts monocytes, promotes neovascularization, and stimulates connective tissue cells to proliferate, migrate, and synthesize extracellular matrix components. Together with platelets and macrophages from the circulation, a multitude of cells residing in human skin, e.g. keratinocytes, endothelial cells,
Manuscript received April 7, 2002; revised August 2, 2002; accepted for publication January 9, 2003 Reprint requests to: Ola Rollman, M.D., Department of Medical Sciences, Section of Dermatology and Venereology, Akademiska Hospital, Uppsala University, S-751 85 Uppsala, Sweden. Email: ola.rollman@medsci. uu.se Abbreviations: DED, deepidermized dermis; GFP, green £uorescent protein; NHK, normal human keratinocytes; PAE, porcine aortic endothelial cells; PDGFbR, platelet-derived growth factor receptor b.
0022-202X/03/$15.00 . Copyright r 2003 by The Society for Investigative Dermatology, Inc. 742
VOL. 120, NO. 5 MAY 2003
shown that topical administration of recombinant PDGF-BB promotes granulation tissue and vascular formation. The clinical achievements of exogenous PDGF therapy, however, are rather modest and require that the peptide be applied repeatedly in pharmacologic concentrations during a considerable length of time (Wieman et al, 1998; Rees et al, 1999). In search of a more con¢ned and durable delivery of PDGF at the exact location of tissue repair, a collagen-embedded PDGF-B DNA plasmid (Tyrone et al, 2000) or a viral vector containing the PDGF-B gene (Liechty et al, 1999; Breitbart et al, 2001) were found to improve dermal healing in rabbit ischemic ulcers. Similarly, grafting of PDGF-A-transduced human skin to athymic mice was reported to reduce wound contraction (Eming et al, 1998) and to stimulate blood vessel formation and dermal cellularity (Eming et al, 1995). Supp and colleagues, on the other hand, found no evidence of a dermal response from retroviral PDGF-A gene transfer in mice (Supp et al, 2000). Hitherto, gene-therapeutic approaches using PDGF in wound healing have focused mainly on the ligand rather than the receptor, and primarily on dermis rather than epidermis. To potentiate the e¡ect of PDGF-mediated skin repair ^ including reepithelialization ^ one can hypothesize that it might be therapeutically advantageous to make adjacent epidermis sensitive to PDGF stimuli. With the aim of converting human epidermis to a target tissue for PDGF stimulation, we have introduced the PDGFbR gene to normal epidermal keratinocytes by a retrovirus-mediated technique. Gene-modi¢ed keratinocytes were used as a source to establish a receptor-positive skin equivalent responsive to recombinant PDGF-BB by enhanced cell proliferation similarly to nonepithelial cells expressing endogenous PDGFbR. MATERIALS AND METHODS Retroviral vectors and packaging cell lines A 3.4 kb cDNA encoding human PDGFbR (Claesson-Welsh et al, 1988) was cloned into the retroviral vector GCsam (Chuah et al, 1995), which drives transgene expression from a Moloney murine leukemia virus long-terminal repeat (MoMLV LTR). The vector was packaged in PG13 cells as described previously (Onodera et al, 1997). DNA was transfected into the ecotropic packaging cell line GP þ E 86 (Markowitz et al, 1990) by calcium phosphate coprecipitation (Mammalian Transfection Kit, Stratagene, La Jolla, CA). Supernatant from transfected cells was harvested after 24 h, supplemented with polybrene at 8 mg per ml, passed through a 0.45 mm ¢lter, and used to transduce PG13 cells subsequently cloned by limiting dilution. After 15 d, 10 PG13 clones were analyzed for vector production by a screening technique similar to that described previously (Jensen et al, 1997). The supernatant was transferred onto cultured keratinocytes prior to immuno£uorescence staining using a monoclonal PDGFbR antibody (as below). One PG13 clone out of 10 tested was positive and was used for further experiments. The retroviral vector GCsamGFP was constructed by inserting the enhanced green £uorescent protein (eGFP) gene, isolated from pEGFPN1 (Clontech, Palo Alto, CA), into the GCsam retroviral vector. GCsamGFP was packaged in PG13 cells as described above. Transduced PG13 cells were cloned by limiting dilution, 10 of which were analyzed for vector production after 15 d by measuring the GFP £uorescence of target keratinocytes using £ow cytometry. The PG13 clone yielding the strongest GFP £uorescence of target cells was used for further experiments. Culture and transduction of primary keratinocytes Normal human keratinocytes (NHK) were obtained from neonatal foreskin samples and cocultured with lethally irradiated 3T3 feeder cells (Rheinwald and Green, 1975) in ‘‘classical medium’’, i.e., a 3:1 mixture of Dulbecco’s modi¢ed Eagle’s medium and Ham’s F-12 nutrient medium, supplemented with 10% fetal bovine serum, hydrocortisone (0.5 mg per ml), cholera enterotoxin (10^10 M), epidermal growth factor (EGF, 10 ng per ml), insulin (5 mg per ml), penicillin (100 IU per ml), streptomycin (100 mg per ml), adenine (1.8 10^4 M), and nonessential amino acids. The keratinocytes were subcultivated in ‘‘serum-free medium’’, i.e., Gibco’s keratinocyte-SFM basal medium supplemented with gentamycin (5 mg per ml) plus attached recombinant EGF (0.1^0.2 ng per ml) and bovine pituitary extract (25 mg per ml). Second passage keratinocytes were transduced at 60% con£uency as described by Garlick et al (1991). Polybrene (8 mg per ml) in 20 mM HEPESHCl pH 7.4 was added to
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
743
fresh supernatant from packaging cell lines cultivated at 371C, and passed through a 0.45 mm ¢lter. Keratinocytes, plated in T-75 £asks the day before transduction, were exposed to 2 ml retroviral supernatant for 2 h at 371C with intermittent gentle agitation. The supernatant was then removed and the keratinocytes were maintained in serum-free medium until analyzed or used for raft cultures. For immuno£uorescence and blotting studies, porcine aortic endothelial (PAE) cells were used as negative controls, whereas PAE cells stably transfected with the PDGFbR gene served as positive controls (ClaessonWelsh et al, 1988). The PAE cells were grown in F-10 medium (Gibco) containing 10% fetal bovine serum. Immuno£uorescence and £ow cytometry of transduced cells Keratinocyte cultures grown on chamber slides were examined by immuno£uorescence microscopy as described previously (Jensen et al, 1996) using a monoclonal PDGFbR antibody (RDI-PDGFRBabm, Research Diagnostics, Flanders, NJ) reactive with the extracellular domain of the receptor subunit (Hart et al, 1987). The primary antibody (1:200 dilution) was applied to living cells maintained on ice for 30 min. After ¢xation in neutral bu¡ered 10% formalin, the secondary £uorescent antibody Alexa Fluor 488 goat antimouse IgG (Molecular Probes, Leiden, The Netherlands) was applied at 1:200 dilution for 15 min. Nuclear staining was performed with Hoechst 33258 before £uorescence microscopy. Cultured cells were also analyzed in a Leica TCS confocal laser scanning microscope using propidium iodide for nuclear staining. Flow cytometry was performed with a FACS Calibur £ow £uorimeter (Becton Dickinson). GFP and propidium iodide were excited at 488 nm and analyzed at 514 nm (FL1) and 546 (FL2), respectively. Southern blotting Southern blotting procedures were performed as described previously (Sambrook et al, 1989). Genomic DNA was isolated and digested with the restriction enzyme Nhe I. Five micrograms of DNA was electrophoresed on a 0.7% agarose gel and blotted to a Magnacharge membrane (Frisenette, Ebeltoft, Denmark). The membrane was hybridized with a 32P random labeled 3.4 kb PDGFbR cDNA probe. Phosphotyrosine immunoblotting Cell cultures were placed on ice and exposed to human recombinant PDGF-BB (generous gift from Amgen, CA) at 100 ng per ml in phosphate-bu¡ered saline (PBS), supplemented with bovine serum albumin (BSA) at 1mg per ml or PBS alone, for 60 min before preparing a cell lysate. A WGA Sepharose (Pharmacia, Uppsala, Sweden) fraction was analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis separation and immunoblotted using a phosphotyrosine antibody (PY20) and a PDGFbR antibody (P-20) purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies were detected by chemiluminescence using a sheep antimouse secondary antibody conjugated to horseradish peroxidase, NA931 (Amersham Pharmacia Biotech, Stockholm, Sweden). Radioreceptor assay Con£uent keratinocyte cultures in 12-well plates (approximately 2.6 105 cells per well) were incubated on ice for 90 min with a mixture of 5 ng per ml of 125I-labeled PDGF-BB (speci¢c activity 12,250 cpm per m) plus unlabeled recombinant PDGF-BB (Amgen) at concentrations ranging between 0 and 1000 ng per ml. Binding of radiolabeled ligand was analyzed in duplicate using a radioreceptor assay (Heldin et al, 1988). PDGF-BB assay The concentration of PDGF-BB was measured using proximity-ligation assay allowing detection of zeptomole amounts of ligand (Fredriksson et al, 2002). Formation of gene-modi¢ed epidermis at the air^liquid interface Virally transduced keratinocytes were used to generate a multilayered epithelium with deepidermized dermis (DED) as substrate (Freeman et al, 1976; Regnier et al, 1990). DEDs were obtained from normal skin at breast reduction surgery and prepared according to Rikimaru et al (1997). At culture, raft-like supports for the DEDs were prepared from Cell Strainer cups (Becton Dickinson Labware, Franklin Lakes, NJ) by cutting o¡ the lateral linings and dividing the vertical plastic ribs at 5 mm distance from the horizontal nylon mesh. Two 3 3 mm fenestrations were made in the mesh (70 mm pore size) to ensure su⁄cient medium exchange. The modi¢ed Cell Strainers were placed, one each per well, in 6 -well culture plates. The DEDs (one to three per well) were placed onto the mesh and then fresh classical medium (containing PDGF-BB at 0.07 pM or 2 pg per ml as determined by the PDGF-BB assay) was added to the upper level of the DED. Keratinocytes (genetically unmodi¢ed, PDGFbR-transduced, and GFP-transduced, respectively) were collected by centrifugation and a ‘‘pellet’’ containing 105 cells suspended in 15 ml medium was placed on top of the DEDs for air^liquid interface culture. Parallel cultures were grown
744
ROLLMAN ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
in ‘‘EGF-de¢cient medium’’ (i.e., classical medium with no added EGF) supplemented or not with 25 ng per ml of recombinant PDGF-BB. Cells were allowed to attach for 10 min and then incubated in 5% CO2 in humidi¢ed air at 371C. The culture medium (as above) was replaced every 3 d. Analysis of the PDGF-BB content of conditioned classical (PDGF-BB unsupplemented) medium showed values ranging between 0.05 and 0.3 pM (1.5^9.0 pg per ml). Cultured cells were examined for green £uorescence using an Olympus SZX12 stereomicroscope equipped with a 100 W mercury lamp, a coaxial £uorescence illuminator, and a built-in GFP ¢lter cube. At harvest on days 6 or 16, tissues were either snap frozen in OTC compound (Tissue-Tek, Miles,Torrance, CA) for storage at 701C or ¢xed in 4% paraformaldehyde. Morphology and immunohistochemical staining of epidermal reconstructs Para⁄n-embedded tissue sections were stained with hematoxylineosin for routine histology. Immunohistochemical analysis using anti-involucrin antibody CLA130 (BioSite, Stockholm, Sweden) at 1:1000 dilution was performed on 5 mm sections with biotinylated antimouse IgG (1:80 dilution) as secondary antibody. Visualization was with the standard avidinbiotin-coupled immunoperoxidase technique (Vectastain ABC kit, Vector Laboratories, Burlingame, CA). To quantify proliferative cells in para⁄n-embedded specimens, three parallel sections (20 mm apart) were obtained from the tissue center, microwaved 3 5 min at 750 W in 10 mM citrate bu¡er pH 6.0, blocked with 10% normal rabbit serum for 15 min, and incubated with the monoclonal mouse anti-Ki- 67 antibody MIB-1, 1:50 dilution (Immunotech, Marseille, France), at þ 41C for 10 h. As secondary antibody, biotinylated antimouse IgG, 1:80 dilution, was added 30 min prior to incubation with avidinbiotinperoxidase complex. The enzyme reaction was developed with 3 -amino-9-ethylcarbazole. In each coded section, the number of Ki- 67 positive cells was counted in three randomly selected visual ¢elds by the same microscopist. The number of stained cells in the basal or adjacent suprabasal position was related to the area of viable epidermis and to the length of the basement membrane zone, respectively, using a Leica DM RB microscope equipped with a Leica DC 200 digital camera and Q500 image analysis software (Leica Microsystems). The mean number of Ki- 67 positive cells in each sample was obtained by averaging mean values from the three sections. The thickness of epidermis was measured at the horizontal midpoint of each visual ¢eld. Immuno£uorescence microscopy of epidermal reconstructs The expression of GFP in cryosections was examined using a Leica £uorescence microscope equipped with a green £uorescence ¢lter. Immuno£uorescence was performed on acetone-¢xed 6 mm cryosections using monoclonal antikeratin 14 (LL002, 1:50 dilution; Novocastra Laboratories, Newcastle-upon-Tyne, U.K.), antikeratin 1 (clone LHK1, 1:10 dilution, generous gift of I.M. Leigh) and antikeratin 16 (LL0025, 1:50 dilution, generous gift of I.M. Leigh). Alexa Fluor 594 (Molecular Probes, Leiden, The Netherlands) was used as secondary £uorescent antibody at 1:800 dilution. PDGFbR immuno£uorescence on acetone-¢xed sections was performed by blocking with 10% normal horse serum for 15 min, exposure to the primary antibody (RDI-PDGFRBabm as speci¢ed above) at 1:1600 dilution for 10 h at þ 41C, and incubation with biotinylated mouse IgG at 1:200 dilution for 30 min prior to staining with £uorescent Texas Red avidin conjugate (Vector Laboratories, Burlingame, CA) at 1:100 dilution for 30 min. Nuclear staining was with Hoechst 33258 (Molecular Probes) added to the mounting medium. Negative controls were prepared by omitting the primary antibody. The proportion of PDGFbR-positive cells relative to Hoechst-labeled cells in the epidermis was determined by £uorescence microscopy using ¢lters optimized for rhodamine (excitation at 54675 nm; barrier ¢lter at 590 nm) and Hoechst dye. Statistics Statistical analysis was performed using ANOVA random e¡ects applying a mixed model (SAS/STAT software, SAS Institute, Cary, NC).
RESULTS PDGFbR gene integration and expression by cultured keratinocytes The PDGFbR gene was retrovirally transferred to primary keratinocytes as veri¢ed by southern blotting analysis of genomic DNA from transduced cells. The DNA was isolated and digested with Nhe I, which cleaves in each LTR sequence of the vector, but not in the transgene. The southern blot was hybridized with a PDGFbR cDNA probe. As seen in Fig 1(c), the PDGFbR-transduced cells yielded a band with the expected size (5.3 kb) indicating the presence of an integrated
vector in the host cells. On average, keratinocytes transduced with the PDGFbR gene contained approximately one copy of the transgene. The expression of PDGFbR protein was veri¢ed by indirect immuno£uorescence of living cells using a speci¢c monoclonal antibody against the extracellular domain of the human PDGFbR subunit. Flow cytometry analysis showed that 42% of the cells in culture were transduced with the PDGFbR gene. This is less e⁄cient than commonly achieved by retroviral gene transfer and is probably due to the relatively high density (60% con£uency) of keratinocytes at the time of transduction. PDGFbR-immunopositive cells in monolayer culture displayed a lucent £uorescence of the cell surface (Fig 1b) with some additional granular staining suggesting that the antibody also recognizes intracellular epitopes. The membrane pattern of receptor £uorescence was more evident using confocal laser scanning microscopy (results not shown). Control stainings of GFP-transduced and genetically unmodi¢ed (Fig 1a) keratinocytes did not reveal any immunostaining using the PDGFbR antibody. Competitive radioactive PDGF labeling and Scatchard analysis demonstrated that PDGFbR-transduced cells expressed approximately 160,000 receptors per cell with a binding a⁄nity Kd of 0.4 nM (results not shown). Autophosphorylation of the PDGF receptor by ligand stimulation Immunoblot analysis of lysates prepared from PDGFbR-transduced cells yielded a 180 kDa band corresponding to the size of mature PDGFbR (Fig 2). There were no detectable bands corresponding to the precursor form (164 kDa) or smaller proteins, which have been recognized by Western blot analysis of dermal ¢broblasts (Hart et al, 1987). Phosphotyrosine blotting of PDGFbR-transduced cells stimulated with recombinant PDGF-BB showed receptor autophosphorylation con¢rming that the receptor was activated by ligand exposure. Visualization of reporter gene expression by £uorescence microscopy of GFP-transduced cells and tissue reconstructs Fluorescence microscopy of GFP-transduced keratinocytes in culture showed a distinct cytoplasmatic signal in a majority (470%) of cells in monolayer culture. A bright green £uorescence was maintained during subsequent air^liquid interface culture as seen by vital microscopy and also demonstrated in cryosections of GFP-transduced epidermal tissue (Fig 3). The morphology and di¡erentiation marker patterns are similar in normal and PDGFbR-modi¢ed epidermal reconstructs The overall tissue architecture of normal and gene-modi¢ed epidermis grown in classical medium for 6 d was similar in all types of DED-based raft cultures. As shown in Fig 3, a strati¢ed and well-organized epithelium displaying all light microscopy layers typical of normal human epidermis was established at the air^liquid interface. After 16 d in culture, tissue morphology of all samples changed markedly in that a compact stratum corneum covering one to three layers of viable keratinocytes was developed. This feature, which is characteristic of air-exposed classical epidermal cultures (Regnier et al, 1990), was equally prominent in all three types of tissues studied. Immuno£uorescence microscopy showed that the di¡erentiation-speci¢c marker keratin 1 (Ponec et al, 1997) was similarly expressed in normal versus gene-modi¢ed epidermis cultured under classical conditions. The staining pattern of involucrin was also comparable ^ irrespective of keratinocyte source ^ and localized in the upper epidermal layers (results not shown). Cytokeratin 14 protein was expressed not only by basal keratinocytes, as found in normal skin (Stoler et al, 1988), but also in the suprabasal layers of viable epidermis. This is a hallmark of hyperproliferative epidermis such as psoriasis lesions (Castelijns et al, 1999) and skin equivalents generated with the
VOL. 120, NO. 5 MAY 2003
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
745
Figure 2. PDGFbR is autophosphorylated following ligand stimulation. WGA fractions of keratinocyte (NHK) and control cell (PAE) lysates were electrophoresed, transferred to nitrocellulose, and probed with antihuman PDGFbR antibody as described in Materials and Methods. Cell lysates from PDGFbR-transduced keratinocytes (GCsamPDGFbR/NHK) show strong expression of 180 kDa PDGFbR (upper panel, lanes 5, 6), and receptor autophosphorylation following recombinant PDGF-BB stimulation (lower panel, lane 6). Positive control cells transfected with the PDGFbR gene (PDGFbR/PAE) show a weak receptor signal (upper panel, lanes 1, 2) and a strong phosphotyrosine signal (lower panel, lane 2) after ligand exposure. Untransfected control PAE cells (ctr/PAE) and GFP-transduced keratinocytes (GCsamGFP/NHK) show neither receptor nor phosphotyrosine signaling (lanes 3, 4 and 7, 8, respectively).
raft technique (Rikimaru et al, 1997). Furthermore, the hyperproliferative marker, keratin 16, was expressed within the entire viable epidermis and stained similarly regardless of the origin of the keratinocytes (results not shown).
Figure 1. PDGFbR transgene integration is expressed by transduced keratinocytes in culture. Indirect immuno£uorescence of (a) un¢xed normal and (b) PDGFbR-transduced keratinocytes cultured on chamber slides and stained with antihuman PDGFbR primary antibody and Alexa Fluor 488 secondary antibody. Nuclear staining was performed using propidium iodide. (c) Southern blotting analysis of genomic DNA isolated from PDGFbR-transduced (left lane), untransduced (middle lane), and GFPtransduced (right lane) keratinocytes digested with Nhe I, electrophoresed, and hybridized with radiolabeled PDGFbR cDNA probe. A 5.3 kb fragment with the expected size of the transgene is seen in the left lane.
PDGFbR expression in gene-modi¢ed epidermis The expression of PDGFbR by transduced keratinocytes was studied both in monolayer cultures and in reconstituted epidermis using immunostaining techniques. The proportion of PDGFbRexpressing cells did not change over time from the initial phase of monolayer culture (42% positive cells by £uorescenceactivated cell sorter analysis) to the subsequent stage of air-exposed multilayered cultures when examined by immuno£uorescence microscopy. In cryosections of PDGFbRtransduced epidermis, receptor-positive keratinocytes ^ both single and clustered ^ were displayed between non£uorescent epidermal areas. In 6 -d-old cultures, there was a bright membrane pattern of receptor staining by cells representing all epidermal levels (Fig 3). By 16 d in culture, the fraction of PDGFbR-expressing cells was di⁄cult to assess due to the pronounced accumulation of epidermal squames in all samples examined. The density of receptor-positive cells in the more basal, undi¡erentiated layers, however, indicated that the percentage of PDGFbR-expressing cells was largely unchanged in late-stage cultures grown under classical conditions.
746
ROLLMAN ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 3. Epidermal morphology and immunostaining patterns in PDGFbR-transduced tissues are similar to the situation in genetically unmodi¢ed reconstructs. Epidermis was regenerated from untransduced (left), GFP-transduced (middle), and PDGFbR-transduced (right) keratinocytes (NHK) cultivated in classical keratinocyte medium (with EGF, without recombinant PDGF-BB) for 6 d. Haematoxylineosin stained sections of all reconstructs displayed similar morphologic features (¢rst row). Cryosections of GFP-transduced tissue showed an intense epidermal signal by £uorescence microscopy (second row). By immuno£uorescence, PDGFbR-transduced tissue showed positive receptor staining using a PDGFbR antibody visualized with Texas Red avidin secondary antibody (third row). Immunostainings of keratin 1 (fourth row), keratin 14 (¢fth row), and Ki- 67 antigen (not shown) displayed comparable patterns between all tissues. Magni¢cation: 250 .
Epidermal reconstructs expressing the PDGFbR transgene respond to ligand stimulation by increased frequency of cycling cells Normal and gene-modi¢ed epidermal cultures were established in classical medium (with no added recombinant PDGF-BB) containing low concentrations of natural PDGF-BB (o9 pg per ml conditioned medium). After 6 and 16 d, respectively, in tissue culture the samples were sectioned and immunohistochemically stained with the Ki- 67 antibody. As shown in Fig 4, the proportion of proliferative cells in the three
types of epidermis was almost identical. Taken together, the mean (7SEM) numbers of Ki- 67 positive cells per millimeter basement membrane length were 3874 (day 6) and 1172 (day 16, not shown), respectively. Moreover, no di¡erence in Ki- 67 frequency was observed between normal and gene-modi¢ed epidermis cultured in EGF-de¢cient medium devoid of recombinant PDGF-BB. When stimulated with recombinant PDGFBB (25 ng per ml medium), however, the frequency of Ki- 67 positive cells increased signi¢cantly in PDGFbR-transduced
VOL. 120, NO. 5 MAY 2003
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
Figure 4. PDGFbR-positive epidermis responds to recombinant PDGF-BB stimulation by increased proliferation. Keratinocyte cultures were established at the air^liquid interface on DEDs and stained immunohistochemically using the proliferation marker Ki- 67. Upper panel: the frequency of Ki 67 positive cells per millimeter basement membrane length in untransduced (left bar), GFP-transduced (middle), and PDGFbRtransduced (right) epidermis cultured in classical medium (containing less than 9 pg per ml of PDGF-BB) for 6 d. Lower panel: the frequency of Ki67 positive cells normalized to basement membrane length (left histogram) or viable epidermal area (right histogram) in 6 -d-old epidermal reconstructs grown in classical medium (minus EGF) supplemented with recombinant PDGF-BB at 25 ng per ml. Means 7 SEM (n ¼ 24 ¢elds per cell type). n po0.05; nnpo0.01.
epidermis (5172.8 cells per mm) compared to GFP-transduced (3373.1, po0.05) or unmodi¢ed (2272.0, po0.001) tissue. Comparable results were obtained if the number of Ki- 67 positive cells was normalized to the area of viable epidermis. The proportion of cycling cells in PDGFbR-transduced epidermis was even higher when cultured with recombinant PDGF (without added EGF) than in classical medium supplemented with EGF. Also, when receptor-positive reconstructs were stimulated with recombinant PDGF, a slightly thicker epithelium (mean 94.8 mm) developed compared to unmodi¢ed controls (mean 84.7 mm); the di¡erence, however, was not statistically signi¢cant. It should be noted that the density of cycling cells was somewhat higher in the GFPtransduced epidermis compared to untransduced tissue if related to length of basement membrane (po0.05), but nonsigni¢cant if related to the viable epidermal area. DISCUSSION
PDGF is a key regulator in cutaneous tissue repair and synthesized from ligand-producing cells at the site of injury. Although secreted rather ubiquitously from skin cells and thrombocytes, the main target cells for receptor-mediated PDGF activity are dermal constituents recognizing various peptide isoforms. PDGF-BB interacts with both a- and gb-receptor subtypes, and represents a potent ligand in terms of connective tissue response. As a result, PDGF-BB binding induces dermal cells to proliferate and migrate, thus promoting granulation tissue formation. Analogously, wound healing stimulation by gene-modi¢ed keratinocytes overexpressing the PDGF-B gene has been shown to induce
747
cell-rich connective tissue repair in athymic mice (Eming et al, 1999). Human keratinocytes are major contributors to endogenous PDGF-AA secretion but these cells do not seem to secrete other proteins belonging to the classical PDGF receptor family (Ansel et al, 1993). Despite some production of PDGFbR transcripts by highly di¡erentiated keratinocytes in culture (Teumer et al, 1994) and injured pig skin (Antoniades et al, 1991), the existence of PDGF receptor type b on NHK has not yet been demonstrated (Rubin et al, 1988; Ansel et al, 1993). In this study, epidermal keratinocytes were transformed into PDGF-responsive cells using a retroviral vector. Radioreceptor assay and Scatchard analysis indicated that recombinant PDGFBB was bound to cells expressing the transmembrane receptor with an approximate Kd of 0.4 nM and an estimated receptor density of 1.6 105 per keratinocyte. These results correspond to values obtained with human ¢broblasts expressing the endogenous PDGFbR (stman et al, 1989). As £ow cytometry demonstrated that PDGFbR gene transfer e⁄cacy was 42%, the epidermal reconstructs were initiated from a mixture of keratinocytes, a majority of which did not express the receptor. Accordingly, cryosections of gene-modi¢ed epidermis showed a patchy appearance representing 30%^50% PDGFbR-immunopositive cells by day 6. This indicates that the normal and gene-modi¢ed cells expanded similarly in classical medium containing less than 0.3 pM of natural PDGF-BB. In many sections, there was a tendency to a clustered distribution of receptor-positive cells in successive layers suggesting that transduced keratinocytes and neighboring daughter cells expressed the transgene. During short-term culture there was no obvious survival advantage ^ or disadvantage ^ of PDGFbR-transduced cells when cultured under classical (recombinant PDGF-BB unsupplemented) conditions. After 16 d in culture the proportion of PDGFbR-expressing cells was di⁄cult to assess due to the predominance of stratum corneum (showing no receptor staining) at this stage of air-exposed culture. The basal cell compartment, however, showed persistent £uorescence from receptor-expressing cells. The raft culture model is a valuable tool in biologic studies of multilayered epidermis (Regnier et al, 1990). As opposed to animal models based on gene-modi¢ed cells or tissue grafts, the in vitro situation facilitates controlled application of bioactive compounds without ample contribution from other sources. As a fundamental point of this study was to assess proliferative activity within the gene-modi¢ed epithelium itself rather than the surrounding tissues. Thus, we maintained reconstituted epidermis in vitro (with a de¢ned PDGF-BB environment) instead of performing subsequent transplantation (in vivo) procedures. We used devitalized dermis as substrate, epidermal keratinocytes as the single source of cells, and classical growth medium with a low background level ranging between 1.5 and 9.0 pg per ml of PDGF-BB. Using the raft model, we found no detrimental e¡ects from retroviral exposure on the architecture, short-term survival, or di¡erentiation-speci¢c marker expression in transduced epidermal tissues. The various reconstructs cultured in classical medium did not show any apparent di¡erences in their keratin or involucrin immuno£uorescence patterns. It should be emphasized that reconstituted epidermis grown on rafts is more proliferative than native skin, particularly during the ¢rst week in culture (Gibbs et al, 2000). This is in accordance with the hyperproliferative state of our equivalents when grown in classical medium containing EGF. Accordingly, the hyperproliferation-associated keratin 16 and the basal keratin 14 were both expressed fairly homogeneously across the viable compartment as judged by immuno£uorescence microscopy. A reason why basal keratin 14 is expressed suprabasally may be that the transcript is not degraded rapidly enough to abolish the protein signal from these regions of hyperproliferative epidermis (Castelijns et al, 1999). Ligand binding to endogenous PDGFbR is known to induce autophosphorylation of the tyrosine kinase receptor and subsequent activation of the signal transduction network leading to increased proliferation and mobility of target cells. In this study, receptor phosphorylation and growth promotion was induced by
748
ROLLMAN ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
ligand exposure to cells expressing the PDGFbR transgene. In native epidermis, the proliferative activity is usually restricted to the basal cell layer and recognizable, e.g., by using antibodies against the Ki- 67 antigen expressed by cycling cells. In our three di¡erent skin equivalents cultured under classical conditions (with no recombinant PDGF in the medium) there were similar frequencies of Ki- 67 positive cells although some cycling cells were also displayed at the adjacent suprabasal level. When stimulated with recombinant PDGF-BB at 25 ng per ml (a concentration 42500 times higher than that of endogenous PDGF-BB in classical medium), however, the density of proliferative cells was signi¢cantly increased in PDGFbR-transduced tissue compared to GFP-transduced and untransduced equivalents. Hypothetically, the PDGFbR-expressing epidermal cells may have responded to PDGF protein from diverse sources: autocrine stimulation by transduced cells, paracrine stimulation by neighboring keratinocytes, and medium-derived stimulation by natural or exogenous ligand. Recombinant ligand was probably the predominant source as the concentrations of PDGF-BB, the natural isoform to which PDGFbR is preferentially receptive, are much lower or even undetectable in skin tissue (Grayson et al, 1993; Cooper et al, 1994), cultured keratinocytes (Ansel et al, 1993), and culture medium (this study). Application of recombinant PDGF at 25 ng per ml resulted in a 2.3 -fold increase in the density of Ki- 67 positive cells compared to untransduced controls. The results are analogous to those in a previous gene transfer study where human EGFR, another transmembrane receptor with tyrosine kinase activity, induced hyperproliferation and enhanced reepithelialization of porcine skin wounds (Nanney et al, 2000). The relevance of the ¢nding that the GFP construct seems to give some backgound activity is uncertain as the di¡erence in Ki- 67 density between untransduced and GFP-transduced tissues is very close to the 0.05 signi¢cance level, and even nonsigni¢cant when related to epidermal area. There are a few reports, however, claiming that GFP may a¡ect mammalian cells (Hanazono et al, 1997; Liu et al, 1999). More speci¢cally, GFP gene transfer has been associated with rapidly growing transfectants (Gubin et al, 1999; Migliaccio et al, 2000). In summary, we show that NHK in culture are amenable to viral PDGFbR gene delivery and that these cells can be utilized to produce a gene-modi¢ed epidermal reconstruct. We also demonstrate that ligand stimulation of such engineered tissue stimulates the mitogenic activity implying that a functional tyrosine kinase receptor and downstream signal transduction pathways are operative. If future studies con¢rm that the proliferative response is maintained also under in vivo conditions, then this novel gene expression by keratinocytes might be exploited in slow healing ulcer states where PDGF protein levels are downregulated (Shukla et al, 1998). Although a link between overexpressed PDGF receptor stimulation and tumor development deserves further consideration, retrovirus-mediated introduction of PDGFbR to epidermal cells may be a useful tool in forthcoming strategies to modify keratinocyte proliferation and epithelialization via a tyrosine kinase receptor. We thank Dr. S. Bittmann for supplying us with skin samples.The technical skill of Inger Pihl-Lundin, Anette Thomsen, Anne Keblovszki, and Bodil Schmidt is acknowledged. Thanks are due to Simon Fredriksson for advice on the PDGF assay. The study was supported by the Edvard Welander and Finsen Foundation, NorFA, and the Swedish Medical Research Foundation, project no K98-13F-12406-01.
REFERENCES Ansel JC, Tiesman JP, Olerud JE, et al: Human keratinocytes are a major source of cutaneous platelet derived growth factor. J Clin Invest 92:671^678, 1993 Antoniades HN, Galanopoulos T, Neville-Golden J, Kiritsy CP, Lynch SE: Injury induces in vivo expression of platelet-derived growth factor (PDGF) and PDGF
receptor mRNAs in skin epithelial cells and PDGF mRNA in connective tissue ¢broblasts. Proc Natl Acad Sci 88:565^569, 1991 Beer HD, Longaker MT, Werner S: Reduced expression of PDGF and PDGF receptors during impaired wound healing. J Invest Dermatol 109:132^138, 1997 Breitbart AS, Grande DA, Laser J, et al: Treatment of ischemic wounds using cultured dermal ¢broblasts transduced retrovirally with PDGF-B and VEGF121 genes. Ann Plast Surg 46:555^561, 2001 Castelijns FA, Gerritsen MJ, van Vlijmen-Willems IM, van Erp PJ, van de Kerkhof PC: Proliferation is the main epidermal target in the treatment of psoriatic plaques with once daily application of tacalcitol ointment. Acta Derm Venereol 79:111^114, 1999 Chuah MK,Vandendriessche T, Morgan RA: Development and analysis of retroviral vectors expressing human factor VIII as a potential gene therapy for haemophilia A. Hum GeneTher 6:1363^1377, 1995 Claesson-Welsh L, Eriksson A, Moren A, et al: cDNA cloning and expression of a human platelet-derived growth factor (PDGF) receptor speci¢c for B-chaincontaining PDGF molecules. Mol Cell Biol 8:3476^3486, 1988 Cooper DM,Yu EZ, Hennessey P, Ko F, Robson MC: Determination of endogenous cytokines in chronic wounds. Ann Surg 219:688^691, 1994 Embil JM, Papp K, Sibbald G, Tousignant J, Smiell JM, Wong B, Lau CY: Recombinant human platelet-derived growth factor-BB (becaplermin) for healing chronic lower extremity diabetic ulcers: An open-label clinical evaluation of e⁄cacy. Wound Repair Regen 8:162^168, 2000 Eming SA, Lee J, Snow RG, Tompkins RG, Yarmush ML, Morgan JR: Genetically modi¢ed human epidermis over-expressing PDGF-A directs the development of a cellular and vascular connective tissue stroma when transplanted to athymic mice ^ implications for the use of genetically modi¢ed keratinocytes to modulate dermal regeneration. J Invest Dermatol 105:756^763, 1995 Eming SA, Medalie DA,Tompkins RG,Yarmush ML, Morgan JR: Genetically modi¢ed human keratinocytes overexpressing PDGF-A enhance the performance of a composite skin graft. Hum GeneTher 1:529^539, 1998 Eming SA, Yarmush ML, Krueger GG, Morgan JR: Regulation of the spatial organization of mesenchymal connective tissue: e¡ects of cell-associated versus released isoforms of platelet-derived growth factor. Am J Pathol 154:281^289, 1999 Fredriksson S, Gullberg M, Jarvius J, et al: Protein detection using proximity-dependent DNA ligation assays. Nat Biotechnol 20:473^477, 2002 Freeman AE, Igel HJ, Herrman BJ, Kleinfeld KL: Growth and characterization of human skin epithelial cell cultures. In Vitro 12:352^362, 1976 Garlick JA, Katz AB, Fenjves ES, Taichman LB: Retrovirus-mediated transduction of cultured epidermal keratinocytes. J Invest Dermatol 97:824^829, 1991 Gibbs S, Silva Pinto AN, Murli S, Huber M, Hohl D, Ponec M: Epidermal growth factor and keratinocyte growth factor di¡erentially regulate epidermal migration, growth, and di¡erentiation. Wound Repair Regen 8:192^203, 2000 Grayson LS, Hansbrough JF, Zapata-Sirvent RL, Dore CA, Morgan JL, Nicolson MA: Quantitation of cytokine levels in skin graft donor site wound £uid. Burns 19:401^405, 1993 Gubin AN, Koduru S, Njoroge JM, Bhatnagar R, Miller JL: Stable expression of green £uorescent protein after liposomal transfection of K562 cells without selective growth conditions. Biotechniques 27:1162^1170, 1999 Hanazono Y, Yu JM, Dunbar CE, Emmons RV: Green £uorescent protein retroviral vectors: Low titer and high recombination frequency suggest a selective disadvantage. Hum GeneTher 8:1313^1319, 1997 Hart CE, Seifert RA, Ross R, Bowen-Pope DF: Synthesis, phosphorylation, and degradation of multiple forms of the platelet-derived growth factor receptor studied using a monoclonal antibody. J Biol Chem 262:10780^10785, 1987 Heldin CH, Backstrom G, stman A, et al: Binding of di¡erent dimeric forms of PDGF to human ¢broblasts: Evidence for two separate receptor types. EMBO J 7:1387^1393, 1988 Heldin CH, Westermark B: Mechanism of action and in vivo role of platelet-derived growth factor. Physiol Rev 79:1283^1316, 1999 Heldin CH, Westermark B, Wasteson A: Speci¢c receptors for platelet-derived growth factor on cells derived from connective tissue and glia. Proc Natl Acad Sci 78:3664^3668, 1981 Jensen TG, Andresen BS, Jensen HK, et al: Rapid characterization of disease-causing mutations in the low density lipoprotein receptor gene (LDL-R) by overexpression in COS cells. Z Gastroenterol Suppl 3:9^12, 1996 Jensen TG, Sullivan DM, Morgan RA, Taichman LB, Nussenblatt RB, Blaese RM, Csaky KG: Retrovirus-mediated gene transfer of ornithine-delta-aminotransferase into keratinocytes from gyrate atrophy patients. Hum GeneTher 8:2125^ 2132, 1997 Liechty KW, Nesbit M, Herlyn M, Radu A, Adzick NS, Crombleholme TM: Adenoviral-mediated over-expression of platelet-derived growth factor-B corrects ischemic impaired wound healing. J Invest Dermatol 113:375^383, 1999 Liu HS, January MS, Chou CK, Chen PH, Ke NJ: Is green £uorescent protein toxic to the living cells? Biochem Biophys Res Commun 260:712^717, 1999 Margolis DJ, Crombleholme T, Herlyn M: Clinical protocol. Phase I trial to evaluate the safety of H5.020CMV. PDGF-B for the treatment of a diabetic insensate foot ulcer. Wound Repair Regen 8:480^493, 2000 Markowitz D, Hesdor¡er C,Ward M, Go¡ S, Bank A: Retroviral gene transfer using safe and e⁄cient packaging cell lines. Ann N Y Acad Sci 612:407^414, 1990 Migliaccio AR, Bengra C, Ling J, et al: Stable and unstable transgene integration sites in the human genome: Extinction of the green £uorescent protein transgene in K562 cells. Gene 256:197^214, 2000
VOL. 120, NO. 5 MAY 2003
Moulin V, Lawny F, Barritault D, Caruelle JP: Platelet releasate treatment improves skin healing in diabetic rats through endogenous growth factor secretion. Cell Mol Biol 44:961^971, 1998 Nanney LB, Paulsen S, Davidson MK, Cardwell NL, Whitsitt JS, Davidson JM: Boosting epidermal growth factor receptor expression by gene gun transfection stimulates epidermal growth in vivo. Wound Repair Regen 8: 117^127, 2000 Onodera M, Yachie A, Nelson DM, Welchlin H, Morgan RA, Blaese RM: A simple and reliable method for screening retroviral producer clones without selectable markers. Hum GeneTher 8:1189^1194, 1997 stman A, Bckstr˛m G, Fong N, et al: Expression of three recombinant homodimeric isoforms of PDGF in Saccharomyces cerevisiae: evidence for di¡erence in receptor binding and functional activities. Growth Factors 1:271^281, 1989 Ponec M, Gibbs S,Weerheim A, Kempenaar J, Mulder A, Mommaas AM: Epidermal growth factor and temperature regulate keratinocyte di¡erentiation. Arch Dermatol Res 289:317^326, 1997 Rees RS, Robson MC, Smiell JM, Perry BH: Becaplermin gel in the treatment of pressure ulcers: A phase II randomized, double-blind, placebo-controlled study. Wound Repair Regen 7:141^147, 1999 Regnier M, Asselineau D, Lenoir MC: Human epidermis reconstructed on dermal substrates in vitro: An alternative to animals in skin pharmacology. Skin Pharmacol 3:70^85, 1990 Reuterdahl C, Sundberg C, Rubin K, Funa K, Gerdin B: Tissue localization of b receptors for platelet-derived growth factor and platelet-derived growth factor B chain during wound repair in humans. J Clin Invest 91:2065^2075, 1993 Rheinwald JG, Green H: Serial cultivation of strains of human epidermal keratinocytes: the formation of keratinizing colonies from single cells. Cell 6:331^343, 1975 Rikimaru K, Moles JP, Watt FM: Correlation between hyperproliferation and suprabasal integrin expression in human epidermis reconstituted in culture. Exp Dermatol 6:214^221, 1997
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
749
Robson MC, Phillips LG, Thomason A, Robson LE, Pierce GF: Platelet-derived growth factor BB for the treatment of chronic pressure ulcers. Lancet 339: 23^25, 1992 Rubin K,Terracio L, R˛nnstrand L, Heldin CH, Klareskog L: Expression of plateletderived growth factor receptors is induced on connective tissue cells during chronic synovial in£ammation. Scand J Immunol 27:285^294, 1988 Sambrook J, Fritsch EF, Maniatis T In: Molecular Cloning: A Laboratory Manual. Cold Spring Harbor: Cold Spring Harbor Laboratory Press, 1989 Shukla A, Dubey MP, Srivastava R, Srivastava BS: Di¡erential expression of proteins during healing of cutaneous wounds in experimental normal and chronic models. Biochem Biophys Res Commun 244:434^439, 1998 Steed DL: Clinical evaluation of recombinant human platelet-derived growth factor for the treatment of lower extremity diabetic ulcers. Diabetic Ulcer Study Group. J Vasc Surg 21:71^78, 1995 Stoler A, Kopan R, Duvic M, Fuchs E: Use of monospeci¢c antisera and cRNA probes to localize the major changes in keratin expression during normal and abnormal epidermal di¡erentiation. J Cell Biol 107:427^446, 1988 Supp DM, Bell SM, Morgan JR, Boyce ST: Genetic modi¢cation of cultured skin substitutes by transduction of human keratinocytes and ¢broblasts with platelet-derived growth factor-A. Wound Repair Regen 8:26^35, 2000 Teumer J, Zezulak K, Green H: Measurement of speci¢c mRNA content of keratinocytes of di¡erent sizes in relation to growth and di¡erentiation. In: Leigh IM, Lane EB, Watt FM eds. The Keratinocyte Handbook, Cambridge: Cambridge University Press, 1994 Tyrone JW, Mogford JE, Chandler LA, Ma C, Xia Y, Pierce GF, Mustoe TA: Collagen-embedded platelet-derived growth factor DNA plasmid promotes wound healing in a dermal ulcer model. J Surg Res 93:230^236, 2000 Wieman TJ, Smiell JM, Su Y: E⁄cacy and safety of a topical gel formulation of recombinant human platelet-derived growth factor-BB (becaplermin) in patients with chronic neuropathic diabetic ulcers. A phase III randomized placebo-controlled double-blind study. Diabetes Care 21:822^827, 1998
Platelet Derived Growth Factor (PDGF) Responsive Epidermis Formed from Human Keratinocytes Transduced with the PDGFb Receptor Gene Ola Rollman, U¡e B. Jensen,n Arne stman,w Lars Bolund,n Sigru¤n M. Gu¤stafsdo¤ttir,z and Thomas G. Jensenn Department of Medical Sciences, Dermatology and Venereology, Uppsala University Hospital, Uppsala, Sweden; nInstitute of Human Genetics, Aarhus University, Aarhus, Denmark; wLudwig Institute for Cancer Research, Biomedical Center, Uppsala, Sweden; zDepartment of Genetics and Pathology, Rudbeck Laboratory, Uppsala, Sweden
Platelet-derived growth factor is a major proliferative and migratory stimulus for connective tissue cells during the initiation of skin repair processes. In response to injury, locally produced platelet-derived growth factor is secreted by a diversity of cutaneous cell types whereas target activity is con¢ned to cells of mesenchymal origin, e.g. dermal ¢broblasts and smooth muscle cells. Although epidermal cells contribute to cutaneous platelet-derived growth factor activity by their ample capacity to secrete platelet-derived growth factor ligand, normal epidermal keratinocytes are not known to express any member of the platelet-derived growth factor receptor family. In order to study if epidermis may be genetically transformed to a platelet-derived growth factor sensitive compartment we aimed to introduce the gene encoding human platelet-derived growth factor receptor b (PDGFbR) into epidermal keratinocytes using a retrovirus-derived vector. Successful gene transfer to primary cells was con¢rmed by immuno£uorescence staining, southern blotting, and ligand-induced receptor autophosphorylation. By culturing a mixture of PDGFbR-transduced and unmodi¢ed keratinocytes at the air^liquid interface on devitalized dermis, we were able to establish a multilayered epithelium showing histologic similarities to that evolved from native
keratinocytes or keratinocytes transduced with the reporter gene encoding enhanced green £uorescent protein. Receptor-modi¢ed epidermal tissue cultured for 6 days and examined by immuno£uorescence microscopy was shown to contain PDGFbR-expressing keratinocytes distributed in all layers of living epidermis. By continued tissue culture in serum-containing medium, the epidermis became increasingly corni¢ed although receptor-positive cells were still observed within the viable basal compartment. Stimulation of PDGFbR-transduced epidermis with recombinant platelet-derived growth factor BB had a mitogenic e¡ect as re£ected by an increased frequency of Ki-67 positive keratinocytes. The study demonstrates that transgene expression of human PDGFbR can be achieved in epidermal keratinocytes by retroviral transduction, and that ligand activation of such gene-modi¢ed skin equivalent enhances cell proliferation. In perspective, viral PDGFbR gene transfer to keratinocytes may be a useful approach in studies of receptor tyrosine kinase mediated skin repair and epithelialization. Key words: genetic transduction/green-£uorescent protein/PDGF receptor tyrosine kinase/wound healing/proximity ligation. J Invest Dermatol 120:742 ^749, 2003
S
¢broblasts, and smooth muscle cells, contribute to PDGF production in the injured microenvironment (reviewed in Heldin and Westermark, 1999). Mesenchymal cells in the skin express PDGF-speci¢c cell surface receptors, which convey autocrine or paracrine responses via activation of the signal transduction network (Heldin et al, 1981; Reuterdahl et al, 1993). Under physiologic conditions, epidermal cells do not express transmembrane receptors recognizing any classical PDGF isoform (AA, AB, and BB dimers); thus target cells for these ligands in normal skin seem restricted to dermal constituents exclusively. Animal data indicate that delayed skin repair is associated with reduced expression of both PDGF ligands and their corresponding receptors (Beer et al, 1997). Hence, attempts have been made to stimulate cutaneous healing by boosting the PDGF activity in the wound environment (Moulin et al, 1998). Human studies in diabetic lower extremity ulcer (Steed, 1995; Embil et al, 2000; Margolis et al, 2000) and decubital ulcer (Robson et al, 1992) have
kin tissue repair is regulated by intrinsic peptides such as platelet-derived growth factor (PDGF), a mitogenic hormone secreted at sites of trauma. Locally released PDGF attracts monocytes, promotes neovascularization, and stimulates connective tissue cells to proliferate, migrate, and synthesize extracellular matrix components. Together with platelets and macrophages from the circulation, a multitude of cells residing in human skin, e.g. keratinocytes, endothelial cells,
Manuscript received April 7, 2002; revised August 2, 2002; accepted for publication January 9, 2003 Reprint requests to: Ola Rollman, M.D., Department of Medical Sciences, Section of Dermatology and Venereology, Akademiska Hospital, Uppsala University, S-751 85 Uppsala, Sweden. Email: ola.rollman@medsci. uu.se Abbreviations: DED, deepidermized dermis; GFP, green £uorescent protein; NHK, normal human keratinocytes; PAE, porcine aortic endothelial cells; PDGFbR, platelet-derived growth factor receptor b.
0022-202X/03/$15.00 . Copyright r 2003 by The Society for Investigative Dermatology, Inc. 742
VOL. 120, NO. 5 MAY 2003
shown that topical administration of recombinant PDGF-BB promotes granulation tissue and vascular formation. The clinical achievements of exogenous PDGF therapy, however, are rather modest and require that the peptide be applied repeatedly in pharmacologic concentrations during a considerable length of time (Wieman et al, 1998; Rees et al, 1999). In search of a more con¢ned and durable delivery of PDGF at the exact location of tissue repair, a collagen-embedded PDGF-B DNA plasmid (Tyrone et al, 2000) or a viral vector containing the PDGF-B gene (Liechty et al, 1999; Breitbart et al, 2001) were found to improve dermal healing in rabbit ischemic ulcers. Similarly, grafting of PDGF-A-transduced human skin to athymic mice was reported to reduce wound contraction (Eming et al, 1998) and to stimulate blood vessel formation and dermal cellularity (Eming et al, 1995). Supp and colleagues, on the other hand, found no evidence of a dermal response from retroviral PDGF-A gene transfer in mice (Supp et al, 2000). Hitherto, gene-therapeutic approaches using PDGF in wound healing have focused mainly on the ligand rather than the receptor, and primarily on dermis rather than epidermis. To potentiate the e¡ect of PDGF-mediated skin repair ^ including reepithelialization ^ one can hypothesize that it might be therapeutically advantageous to make adjacent epidermis sensitive to PDGF stimuli. With the aim of converting human epidermis to a target tissue for PDGF stimulation, we have introduced the PDGFbR gene to normal epidermal keratinocytes by a retrovirus-mediated technique. Gene-modi¢ed keratinocytes were used as a source to establish a receptor-positive skin equivalent responsive to recombinant PDGF-BB by enhanced cell proliferation similarly to nonepithelial cells expressing endogenous PDGFbR. MATERIALS AND METHODS Retroviral vectors and packaging cell lines A 3.4 kb cDNA encoding human PDGFbR (Claesson-Welsh et al, 1988) was cloned into the retroviral vector GCsam (Chuah et al, 1995), which drives transgene expression from a Moloney murine leukemia virus long-terminal repeat (MoMLV LTR). The vector was packaged in PG13 cells as described previously (Onodera et al, 1997). DNA was transfected into the ecotropic packaging cell line GP þ E 86 (Markowitz et al, 1990) by calcium phosphate coprecipitation (Mammalian Transfection Kit, Stratagene, La Jolla, CA). Supernatant from transfected cells was harvested after 24 h, supplemented with polybrene at 8 mg per ml, passed through a 0.45 mm ¢lter, and used to transduce PG13 cells subsequently cloned by limiting dilution. After 15 d, 10 PG13 clones were analyzed for vector production by a screening technique similar to that described previously (Jensen et al, 1997). The supernatant was transferred onto cultured keratinocytes prior to immuno£uorescence staining using a monoclonal PDGFbR antibody (as below). One PG13 clone out of 10 tested was positive and was used for further experiments. The retroviral vector GCsamGFP was constructed by inserting the enhanced green £uorescent protein (eGFP) gene, isolated from pEGFPN1 (Clontech, Palo Alto, CA), into the GCsam retroviral vector. GCsamGFP was packaged in PG13 cells as described above. Transduced PG13 cells were cloned by limiting dilution, 10 of which were analyzed for vector production after 15 d by measuring the GFP £uorescence of target keratinocytes using £ow cytometry. The PG13 clone yielding the strongest GFP £uorescence of target cells was used for further experiments. Culture and transduction of primary keratinocytes Normal human keratinocytes (NHK) were obtained from neonatal foreskin samples and cocultured with lethally irradiated 3T3 feeder cells (Rheinwald and Green, 1975) in ‘‘classical medium’’, i.e., a 3:1 mixture of Dulbecco’s modi¢ed Eagle’s medium and Ham’s F-12 nutrient medium, supplemented with 10% fetal bovine serum, hydrocortisone (0.5 mg per ml), cholera enterotoxin (10^10 M), epidermal growth factor (EGF, 10 ng per ml), insulin (5 mg per ml), penicillin (100 IU per ml), streptomycin (100 mg per ml), adenine (1.8 10^4 M), and nonessential amino acids. The keratinocytes were subcultivated in ‘‘serum-free medium’’, i.e., Gibco’s keratinocyte-SFM basal medium supplemented with gentamycin (5 mg per ml) plus attached recombinant EGF (0.1^0.2 ng per ml) and bovine pituitary extract (25 mg per ml). Second passage keratinocytes were transduced at 60% con£uency as described by Garlick et al (1991). Polybrene (8 mg per ml) in 20 mM HEPESHCl pH 7.4 was added to
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
743
fresh supernatant from packaging cell lines cultivated at 371C, and passed through a 0.45 mm ¢lter. Keratinocytes, plated in T-75 £asks the day before transduction, were exposed to 2 ml retroviral supernatant for 2 h at 371C with intermittent gentle agitation. The supernatant was then removed and the keratinocytes were maintained in serum-free medium until analyzed or used for raft cultures. For immuno£uorescence and blotting studies, porcine aortic endothelial (PAE) cells were used as negative controls, whereas PAE cells stably transfected with the PDGFbR gene served as positive controls (ClaessonWelsh et al, 1988). The PAE cells were grown in F-10 medium (Gibco) containing 10% fetal bovine serum. Immuno£uorescence and £ow cytometry of transduced cells Keratinocyte cultures grown on chamber slides were examined by immuno£uorescence microscopy as described previously (Jensen et al, 1996) using a monoclonal PDGFbR antibody (RDI-PDGFRBabm, Research Diagnostics, Flanders, NJ) reactive with the extracellular domain of the receptor subunit (Hart et al, 1987). The primary antibody (1:200 dilution) was applied to living cells maintained on ice for 30 min. After ¢xation in neutral bu¡ered 10% formalin, the secondary £uorescent antibody Alexa Fluor 488 goat antimouse IgG (Molecular Probes, Leiden, The Netherlands) was applied at 1:200 dilution for 15 min. Nuclear staining was performed with Hoechst 33258 before £uorescence microscopy. Cultured cells were also analyzed in a Leica TCS confocal laser scanning microscope using propidium iodide for nuclear staining. Flow cytometry was performed with a FACS Calibur £ow £uorimeter (Becton Dickinson). GFP and propidium iodide were excited at 488 nm and analyzed at 514 nm (FL1) and 546 (FL2), respectively. Southern blotting Southern blotting procedures were performed as described previously (Sambrook et al, 1989). Genomic DNA was isolated and digested with the restriction enzyme Nhe I. Five micrograms of DNA was electrophoresed on a 0.7% agarose gel and blotted to a Magnacharge membrane (Frisenette, Ebeltoft, Denmark). The membrane was hybridized with a 32P random labeled 3.4 kb PDGFbR cDNA probe. Phosphotyrosine immunoblotting Cell cultures were placed on ice and exposed to human recombinant PDGF-BB (generous gift from Amgen, CA) at 100 ng per ml in phosphate-bu¡ered saline (PBS), supplemented with bovine serum albumin (BSA) at 1mg per ml or PBS alone, for 60 min before preparing a cell lysate. A WGA Sepharose (Pharmacia, Uppsala, Sweden) fraction was analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis separation and immunoblotted using a phosphotyrosine antibody (PY20) and a PDGFbR antibody (P-20) purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies were detected by chemiluminescence using a sheep antimouse secondary antibody conjugated to horseradish peroxidase, NA931 (Amersham Pharmacia Biotech, Stockholm, Sweden). Radioreceptor assay Con£uent keratinocyte cultures in 12-well plates (approximately 2.6 105 cells per well) were incubated on ice for 90 min with a mixture of 5 ng per ml of 125I-labeled PDGF-BB (speci¢c activity 12,250 cpm per m) plus unlabeled recombinant PDGF-BB (Amgen) at concentrations ranging between 0 and 1000 ng per ml. Binding of radiolabeled ligand was analyzed in duplicate using a radioreceptor assay (Heldin et al, 1988). PDGF-BB assay The concentration of PDGF-BB was measured using proximity-ligation assay allowing detection of zeptomole amounts of ligand (Fredriksson et al, 2002). Formation of gene-modi¢ed epidermis at the air^liquid interface Virally transduced keratinocytes were used to generate a multilayered epithelium with deepidermized dermis (DED) as substrate (Freeman et al, 1976; Regnier et al, 1990). DEDs were obtained from normal skin at breast reduction surgery and prepared according to Rikimaru et al (1997). At culture, raft-like supports for the DEDs were prepared from Cell Strainer cups (Becton Dickinson Labware, Franklin Lakes, NJ) by cutting o¡ the lateral linings and dividing the vertical plastic ribs at 5 mm distance from the horizontal nylon mesh. Two 3 3 mm fenestrations were made in the mesh (70 mm pore size) to ensure su⁄cient medium exchange. The modi¢ed Cell Strainers were placed, one each per well, in 6 -well culture plates. The DEDs (one to three per well) were placed onto the mesh and then fresh classical medium (containing PDGF-BB at 0.07 pM or 2 pg per ml as determined by the PDGF-BB assay) was added to the upper level of the DED. Keratinocytes (genetically unmodi¢ed, PDGFbR-transduced, and GFP-transduced, respectively) were collected by centrifugation and a ‘‘pellet’’ containing 105 cells suspended in 15 ml medium was placed on top of the DEDs for air^liquid interface culture. Parallel cultures were grown
744
ROLLMAN ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
in ‘‘EGF-de¢cient medium’’ (i.e., classical medium with no added EGF) supplemented or not with 25 ng per ml of recombinant PDGF-BB. Cells were allowed to attach for 10 min and then incubated in 5% CO2 in humidi¢ed air at 371C. The culture medium (as above) was replaced every 3 d. Analysis of the PDGF-BB content of conditioned classical (PDGF-BB unsupplemented) medium showed values ranging between 0.05 and 0.3 pM (1.5^9.0 pg per ml). Cultured cells were examined for green £uorescence using an Olympus SZX12 stereomicroscope equipped with a 100 W mercury lamp, a coaxial £uorescence illuminator, and a built-in GFP ¢lter cube. At harvest on days 6 or 16, tissues were either snap frozen in OTC compound (Tissue-Tek, Miles,Torrance, CA) for storage at 701C or ¢xed in 4% paraformaldehyde. Morphology and immunohistochemical staining of epidermal reconstructs Para⁄n-embedded tissue sections were stained with hematoxylineosin for routine histology. Immunohistochemical analysis using anti-involucrin antibody CLA130 (BioSite, Stockholm, Sweden) at 1:1000 dilution was performed on 5 mm sections with biotinylated antimouse IgG (1:80 dilution) as secondary antibody. Visualization was with the standard avidinbiotin-coupled immunoperoxidase technique (Vectastain ABC kit, Vector Laboratories, Burlingame, CA). To quantify proliferative cells in para⁄n-embedded specimens, three parallel sections (20 mm apart) were obtained from the tissue center, microwaved 3 5 min at 750 W in 10 mM citrate bu¡er pH 6.0, blocked with 10% normal rabbit serum for 15 min, and incubated with the monoclonal mouse anti-Ki- 67 antibody MIB-1, 1:50 dilution (Immunotech, Marseille, France), at þ 41C for 10 h. As secondary antibody, biotinylated antimouse IgG, 1:80 dilution, was added 30 min prior to incubation with avidinbiotinperoxidase complex. The enzyme reaction was developed with 3 -amino-9-ethylcarbazole. In each coded section, the number of Ki- 67 positive cells was counted in three randomly selected visual ¢elds by the same microscopist. The number of stained cells in the basal or adjacent suprabasal position was related to the area of viable epidermis and to the length of the basement membrane zone, respectively, using a Leica DM RB microscope equipped with a Leica DC 200 digital camera and Q500 image analysis software (Leica Microsystems). The mean number of Ki- 67 positive cells in each sample was obtained by averaging mean values from the three sections. The thickness of epidermis was measured at the horizontal midpoint of each visual ¢eld. Immuno£uorescence microscopy of epidermal reconstructs The expression of GFP in cryosections was examined using a Leica £uorescence microscope equipped with a green £uorescence ¢lter. Immuno£uorescence was performed on acetone-¢xed 6 mm cryosections using monoclonal antikeratin 14 (LL002, 1:50 dilution; Novocastra Laboratories, Newcastle-upon-Tyne, U.K.), antikeratin 1 (clone LHK1, 1:10 dilution, generous gift of I.M. Leigh) and antikeratin 16 (LL0025, 1:50 dilution, generous gift of I.M. Leigh). Alexa Fluor 594 (Molecular Probes, Leiden, The Netherlands) was used as secondary £uorescent antibody at 1:800 dilution. PDGFbR immuno£uorescence on acetone-¢xed sections was performed by blocking with 10% normal horse serum for 15 min, exposure to the primary antibody (RDI-PDGFRBabm as speci¢ed above) at 1:1600 dilution for 10 h at þ 41C, and incubation with biotinylated mouse IgG at 1:200 dilution for 30 min prior to staining with £uorescent Texas Red avidin conjugate (Vector Laboratories, Burlingame, CA) at 1:100 dilution for 30 min. Nuclear staining was with Hoechst 33258 (Molecular Probes) added to the mounting medium. Negative controls were prepared by omitting the primary antibody. The proportion of PDGFbR-positive cells relative to Hoechst-labeled cells in the epidermis was determined by £uorescence microscopy using ¢lters optimized for rhodamine (excitation at 54675 nm; barrier ¢lter at 590 nm) and Hoechst dye. Statistics Statistical analysis was performed using ANOVA random e¡ects applying a mixed model (SAS/STAT software, SAS Institute, Cary, NC).
RESULTS PDGFbR gene integration and expression by cultured keratinocytes The PDGFbR gene was retrovirally transferred to primary keratinocytes as veri¢ed by southern blotting analysis of genomic DNA from transduced cells. The DNA was isolated and digested with Nhe I, which cleaves in each LTR sequence of the vector, but not in the transgene. The southern blot was hybridized with a PDGFbR cDNA probe. As seen in Fig 1(c), the PDGFbR-transduced cells yielded a band with the expected size (5.3 kb) indicating the presence of an integrated
vector in the host cells. On average, keratinocytes transduced with the PDGFbR gene contained approximately one copy of the transgene. The expression of PDGFbR protein was veri¢ed by indirect immuno£uorescence of living cells using a speci¢c monoclonal antibody against the extracellular domain of the human PDGFbR subunit. Flow cytometry analysis showed that 42% of the cells in culture were transduced with the PDGFbR gene. This is less e⁄cient than commonly achieved by retroviral gene transfer and is probably due to the relatively high density (60% con£uency) of keratinocytes at the time of transduction. PDGFbR-immunopositive cells in monolayer culture displayed a lucent £uorescence of the cell surface (Fig 1b) with some additional granular staining suggesting that the antibody also recognizes intracellular epitopes. The membrane pattern of receptor £uorescence was more evident using confocal laser scanning microscopy (results not shown). Control stainings of GFP-transduced and genetically unmodi¢ed (Fig 1a) keratinocytes did not reveal any immunostaining using the PDGFbR antibody. Competitive radioactive PDGF labeling and Scatchard analysis demonstrated that PDGFbR-transduced cells expressed approximately 160,000 receptors per cell with a binding a⁄nity Kd of 0.4 nM (results not shown). Autophosphorylation of the PDGF receptor by ligand stimulation Immunoblot analysis of lysates prepared from PDGFbR-transduced cells yielded a 180 kDa band corresponding to the size of mature PDGFbR (Fig 2). There were no detectable bands corresponding to the precursor form (164 kDa) or smaller proteins, which have been recognized by Western blot analysis of dermal ¢broblasts (Hart et al, 1987). Phosphotyrosine blotting of PDGFbR-transduced cells stimulated with recombinant PDGF-BB showed receptor autophosphorylation con¢rming that the receptor was activated by ligand exposure. Visualization of reporter gene expression by £uorescence microscopy of GFP-transduced cells and tissue reconstructs Fluorescence microscopy of GFP-transduced keratinocytes in culture showed a distinct cytoplasmatic signal in a majority (470%) of cells in monolayer culture. A bright green £uorescence was maintained during subsequent air^liquid interface culture as seen by vital microscopy and also demonstrated in cryosections of GFP-transduced epidermal tissue (Fig 3). The morphology and di¡erentiation marker patterns are similar in normal and PDGFbR-modi¢ed epidermal reconstructs The overall tissue architecture of normal and gene-modi¢ed epidermis grown in classical medium for 6 d was similar in all types of DED-based raft cultures. As shown in Fig 3, a strati¢ed and well-organized epithelium displaying all light microscopy layers typical of normal human epidermis was established at the air^liquid interface. After 16 d in culture, tissue morphology of all samples changed markedly in that a compact stratum corneum covering one to three layers of viable keratinocytes was developed. This feature, which is characteristic of air-exposed classical epidermal cultures (Regnier et al, 1990), was equally prominent in all three types of tissues studied. Immuno£uorescence microscopy showed that the di¡erentiation-speci¢c marker keratin 1 (Ponec et al, 1997) was similarly expressed in normal versus gene-modi¢ed epidermis cultured under classical conditions. The staining pattern of involucrin was also comparable ^ irrespective of keratinocyte source ^ and localized in the upper epidermal layers (results not shown). Cytokeratin 14 protein was expressed not only by basal keratinocytes, as found in normal skin (Stoler et al, 1988), but also in the suprabasal layers of viable epidermis. This is a hallmark of hyperproliferative epidermis such as psoriasis lesions (Castelijns et al, 1999) and skin equivalents generated with the
VOL. 120, NO. 5 MAY 2003
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
745
Figure 2. PDGFbR is autophosphorylated following ligand stimulation. WGA fractions of keratinocyte (NHK) and control cell (PAE) lysates were electrophoresed, transferred to nitrocellulose, and probed with antihuman PDGFbR antibody as described in Materials and Methods. Cell lysates from PDGFbR-transduced keratinocytes (GCsamPDGFbR/NHK) show strong expression of 180 kDa PDGFbR (upper panel, lanes 5, 6), and receptor autophosphorylation following recombinant PDGF-BB stimulation (lower panel, lane 6). Positive control cells transfected with the PDGFbR gene (PDGFbR/PAE) show a weak receptor signal (upper panel, lanes 1, 2) and a strong phosphotyrosine signal (lower panel, lane 2) after ligand exposure. Untransfected control PAE cells (ctr/PAE) and GFP-transduced keratinocytes (GCsamGFP/NHK) show neither receptor nor phosphotyrosine signaling (lanes 3, 4 and 7, 8, respectively).
raft technique (Rikimaru et al, 1997). Furthermore, the hyperproliferative marker, keratin 16, was expressed within the entire viable epidermis and stained similarly regardless of the origin of the keratinocytes (results not shown).
Figure 1. PDGFbR transgene integration is expressed by transduced keratinocytes in culture. Indirect immuno£uorescence of (a) un¢xed normal and (b) PDGFbR-transduced keratinocytes cultured on chamber slides and stained with antihuman PDGFbR primary antibody and Alexa Fluor 488 secondary antibody. Nuclear staining was performed using propidium iodide. (c) Southern blotting analysis of genomic DNA isolated from PDGFbR-transduced (left lane), untransduced (middle lane), and GFPtransduced (right lane) keratinocytes digested with Nhe I, electrophoresed, and hybridized with radiolabeled PDGFbR cDNA probe. A 5.3 kb fragment with the expected size of the transgene is seen in the left lane.
PDGFbR expression in gene-modi¢ed epidermis The expression of PDGFbR by transduced keratinocytes was studied both in monolayer cultures and in reconstituted epidermis using immunostaining techniques. The proportion of PDGFbRexpressing cells did not change over time from the initial phase of monolayer culture (42% positive cells by £uorescenceactivated cell sorter analysis) to the subsequent stage of air-exposed multilayered cultures when examined by immuno£uorescence microscopy. In cryosections of PDGFbRtransduced epidermis, receptor-positive keratinocytes ^ both single and clustered ^ were displayed between non£uorescent epidermal areas. In 6 -d-old cultures, there was a bright membrane pattern of receptor staining by cells representing all epidermal levels (Fig 3). By 16 d in culture, the fraction of PDGFbR-expressing cells was di⁄cult to assess due to the pronounced accumulation of epidermal squames in all samples examined. The density of receptor-positive cells in the more basal, undi¡erentiated layers, however, indicated that the percentage of PDGFbR-expressing cells was largely unchanged in late-stage cultures grown under classical conditions.
746
ROLLMAN ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 3. Epidermal morphology and immunostaining patterns in PDGFbR-transduced tissues are similar to the situation in genetically unmodi¢ed reconstructs. Epidermis was regenerated from untransduced (left), GFP-transduced (middle), and PDGFbR-transduced (right) keratinocytes (NHK) cultivated in classical keratinocyte medium (with EGF, without recombinant PDGF-BB) for 6 d. Haematoxylineosin stained sections of all reconstructs displayed similar morphologic features (¢rst row). Cryosections of GFP-transduced tissue showed an intense epidermal signal by £uorescence microscopy (second row). By immuno£uorescence, PDGFbR-transduced tissue showed positive receptor staining using a PDGFbR antibody visualized with Texas Red avidin secondary antibody (third row). Immunostainings of keratin 1 (fourth row), keratin 14 (¢fth row), and Ki- 67 antigen (not shown) displayed comparable patterns between all tissues. Magni¢cation: 250 .
Epidermal reconstructs expressing the PDGFbR transgene respond to ligand stimulation by increased frequency of cycling cells Normal and gene-modi¢ed epidermal cultures were established in classical medium (with no added recombinant PDGF-BB) containing low concentrations of natural PDGF-BB (o9 pg per ml conditioned medium). After 6 and 16 d, respectively, in tissue culture the samples were sectioned and immunohistochemically stained with the Ki- 67 antibody. As shown in Fig 4, the proportion of proliferative cells in the three
types of epidermis was almost identical. Taken together, the mean (7SEM) numbers of Ki- 67 positive cells per millimeter basement membrane length were 3874 (day 6) and 1172 (day 16, not shown), respectively. Moreover, no di¡erence in Ki- 67 frequency was observed between normal and gene-modi¢ed epidermis cultured in EGF-de¢cient medium devoid of recombinant PDGF-BB. When stimulated with recombinant PDGFBB (25 ng per ml medium), however, the frequency of Ki- 67 positive cells increased signi¢cantly in PDGFbR-transduced
VOL. 120, NO. 5 MAY 2003
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
Figure 4. PDGFbR-positive epidermis responds to recombinant PDGF-BB stimulation by increased proliferation. Keratinocyte cultures were established at the air^liquid interface on DEDs and stained immunohistochemically using the proliferation marker Ki- 67. Upper panel: the frequency of Ki 67 positive cells per millimeter basement membrane length in untransduced (left bar), GFP-transduced (middle), and PDGFbRtransduced (right) epidermis cultured in classical medium (containing less than 9 pg per ml of PDGF-BB) for 6 d. Lower panel: the frequency of Ki67 positive cells normalized to basement membrane length (left histogram) or viable epidermal area (right histogram) in 6 -d-old epidermal reconstructs grown in classical medium (minus EGF) supplemented with recombinant PDGF-BB at 25 ng per ml. Means 7 SEM (n ¼ 24 ¢elds per cell type). n po0.05; nnpo0.01.
epidermis (5172.8 cells per mm) compared to GFP-transduced (3373.1, po0.05) or unmodi¢ed (2272.0, po0.001) tissue. Comparable results were obtained if the number of Ki- 67 positive cells was normalized to the area of viable epidermis. The proportion of cycling cells in PDGFbR-transduced epidermis was even higher when cultured with recombinant PDGF (without added EGF) than in classical medium supplemented with EGF. Also, when receptor-positive reconstructs were stimulated with recombinant PDGF, a slightly thicker epithelium (mean 94.8 mm) developed compared to unmodi¢ed controls (mean 84.7 mm); the di¡erence, however, was not statistically signi¢cant. It should be noted that the density of cycling cells was somewhat higher in the GFPtransduced epidermis compared to untransduced tissue if related to length of basement membrane (po0.05), but nonsigni¢cant if related to the viable epidermal area. DISCUSSION
PDGF is a key regulator in cutaneous tissue repair and synthesized from ligand-producing cells at the site of injury. Although secreted rather ubiquitously from skin cells and thrombocytes, the main target cells for receptor-mediated PDGF activity are dermal constituents recognizing various peptide isoforms. PDGF-BB interacts with both a- and gb-receptor subtypes, and represents a potent ligand in terms of connective tissue response. As a result, PDGF-BB binding induces dermal cells to proliferate and migrate, thus promoting granulation tissue formation. Analogously, wound healing stimulation by gene-modi¢ed keratinocytes overexpressing the PDGF-B gene has been shown to induce
747
cell-rich connective tissue repair in athymic mice (Eming et al, 1999). Human keratinocytes are major contributors to endogenous PDGF-AA secretion but these cells do not seem to secrete other proteins belonging to the classical PDGF receptor family (Ansel et al, 1993). Despite some production of PDGFbR transcripts by highly di¡erentiated keratinocytes in culture (Teumer et al, 1994) and injured pig skin (Antoniades et al, 1991), the existence of PDGF receptor type b on NHK has not yet been demonstrated (Rubin et al, 1988; Ansel et al, 1993). In this study, epidermal keratinocytes were transformed into PDGF-responsive cells using a retroviral vector. Radioreceptor assay and Scatchard analysis indicated that recombinant PDGFBB was bound to cells expressing the transmembrane receptor with an approximate Kd of 0.4 nM and an estimated receptor density of 1.6 105 per keratinocyte. These results correspond to values obtained with human ¢broblasts expressing the endogenous PDGFbR (stman et al, 1989). As £ow cytometry demonstrated that PDGFbR gene transfer e⁄cacy was 42%, the epidermal reconstructs were initiated from a mixture of keratinocytes, a majority of which did not express the receptor. Accordingly, cryosections of gene-modi¢ed epidermis showed a patchy appearance representing 30%^50% PDGFbR-immunopositive cells by day 6. This indicates that the normal and gene-modi¢ed cells expanded similarly in classical medium containing less than 0.3 pM of natural PDGF-BB. In many sections, there was a tendency to a clustered distribution of receptor-positive cells in successive layers suggesting that transduced keratinocytes and neighboring daughter cells expressed the transgene. During short-term culture there was no obvious survival advantage ^ or disadvantage ^ of PDGFbR-transduced cells when cultured under classical (recombinant PDGF-BB unsupplemented) conditions. After 16 d in culture the proportion of PDGFbR-expressing cells was di⁄cult to assess due to the predominance of stratum corneum (showing no receptor staining) at this stage of air-exposed culture. The basal cell compartment, however, showed persistent £uorescence from receptor-expressing cells. The raft culture model is a valuable tool in biologic studies of multilayered epidermis (Regnier et al, 1990). As opposed to animal models based on gene-modi¢ed cells or tissue grafts, the in vitro situation facilitates controlled application of bioactive compounds without ample contribution from other sources. As a fundamental point of this study was to assess proliferative activity within the gene-modi¢ed epithelium itself rather than the surrounding tissues. Thus, we maintained reconstituted epidermis in vitro (with a de¢ned PDGF-BB environment) instead of performing subsequent transplantation (in vivo) procedures. We used devitalized dermis as substrate, epidermal keratinocytes as the single source of cells, and classical growth medium with a low background level ranging between 1.5 and 9.0 pg per ml of PDGF-BB. Using the raft model, we found no detrimental e¡ects from retroviral exposure on the architecture, short-term survival, or di¡erentiation-speci¢c marker expression in transduced epidermal tissues. The various reconstructs cultured in classical medium did not show any apparent di¡erences in their keratin or involucrin immuno£uorescence patterns. It should be emphasized that reconstituted epidermis grown on rafts is more proliferative than native skin, particularly during the ¢rst week in culture (Gibbs et al, 2000). This is in accordance with the hyperproliferative state of our equivalents when grown in classical medium containing EGF. Accordingly, the hyperproliferation-associated keratin 16 and the basal keratin 14 were both expressed fairly homogeneously across the viable compartment as judged by immuno£uorescence microscopy. A reason why basal keratin 14 is expressed suprabasally may be that the transcript is not degraded rapidly enough to abolish the protein signal from these regions of hyperproliferative epidermis (Castelijns et al, 1999). Ligand binding to endogenous PDGFbR is known to induce autophosphorylation of the tyrosine kinase receptor and subsequent activation of the signal transduction network leading to increased proliferation and mobility of target cells. In this study, receptor phosphorylation and growth promotion was induced by
748
ROLLMAN ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
ligand exposure to cells expressing the PDGFbR transgene. In native epidermis, the proliferative activity is usually restricted to the basal cell layer and recognizable, e.g., by using antibodies against the Ki- 67 antigen expressed by cycling cells. In our three di¡erent skin equivalents cultured under classical conditions (with no recombinant PDGF in the medium) there were similar frequencies of Ki- 67 positive cells although some cycling cells were also displayed at the adjacent suprabasal level. When stimulated with recombinant PDGF-BB at 25 ng per ml (a concentration 42500 times higher than that of endogenous PDGF-BB in classical medium), however, the density of proliferative cells was signi¢cantly increased in PDGFbR-transduced tissue compared to GFP-transduced and untransduced equivalents. Hypothetically, the PDGFbR-expressing epidermal cells may have responded to PDGF protein from diverse sources: autocrine stimulation by transduced cells, paracrine stimulation by neighboring keratinocytes, and medium-derived stimulation by natural or exogenous ligand. Recombinant ligand was probably the predominant source as the concentrations of PDGF-BB, the natural isoform to which PDGFbR is preferentially receptive, are much lower or even undetectable in skin tissue (Grayson et al, 1993; Cooper et al, 1994), cultured keratinocytes (Ansel et al, 1993), and culture medium (this study). Application of recombinant PDGF at 25 ng per ml resulted in a 2.3 -fold increase in the density of Ki- 67 positive cells compared to untransduced controls. The results are analogous to those in a previous gene transfer study where human EGFR, another transmembrane receptor with tyrosine kinase activity, induced hyperproliferation and enhanced reepithelialization of porcine skin wounds (Nanney et al, 2000). The relevance of the ¢nding that the GFP construct seems to give some backgound activity is uncertain as the di¡erence in Ki- 67 density between untransduced and GFP-transduced tissues is very close to the 0.05 signi¢cance level, and even nonsigni¢cant when related to epidermal area. There are a few reports, however, claiming that GFP may a¡ect mammalian cells (Hanazono et al, 1997; Liu et al, 1999). More speci¢cally, GFP gene transfer has been associated with rapidly growing transfectants (Gubin et al, 1999; Migliaccio et al, 2000). In summary, we show that NHK in culture are amenable to viral PDGFbR gene delivery and that these cells can be utilized to produce a gene-modi¢ed epidermal reconstruct. We also demonstrate that ligand stimulation of such engineered tissue stimulates the mitogenic activity implying that a functional tyrosine kinase receptor and downstream signal transduction pathways are operative. If future studies con¢rm that the proliferative response is maintained also under in vivo conditions, then this novel gene expression by keratinocytes might be exploited in slow healing ulcer states where PDGF protein levels are downregulated (Shukla et al, 1998). Although a link between overexpressed PDGF receptor stimulation and tumor development deserves further consideration, retrovirus-mediated introduction of PDGFbR to epidermal cells may be a useful tool in forthcoming strategies to modify keratinocyte proliferation and epithelialization via a tyrosine kinase receptor. We thank Dr. S. Bittmann for supplying us with skin samples.The technical skill of Inger Pihl-Lundin, Anette Thomsen, Anne Keblovszki, and Bodil Schmidt is acknowledged. Thanks are due to Simon Fredriksson for advice on the PDGF assay. The study was supported by the Edvard Welander and Finsen Foundation, NorFA, and the Swedish Medical Research Foundation, project no K98-13F-12406-01.
REFERENCES Ansel JC, Tiesman JP, Olerud JE, et al: Human keratinocytes are a major source of cutaneous platelet derived growth factor. J Clin Invest 92:671^678, 1993 Antoniades HN, Galanopoulos T, Neville-Golden J, Kiritsy CP, Lynch SE: Injury induces in vivo expression of platelet-derived growth factor (PDGF) and PDGF
receptor mRNAs in skin epithelial cells and PDGF mRNA in connective tissue ¢broblasts. Proc Natl Acad Sci 88:565^569, 1991 Beer HD, Longaker MT, Werner S: Reduced expression of PDGF and PDGF receptors during impaired wound healing. J Invest Dermatol 109:132^138, 1997 Breitbart AS, Grande DA, Laser J, et al: Treatment of ischemic wounds using cultured dermal ¢broblasts transduced retrovirally with PDGF-B and VEGF121 genes. Ann Plast Surg 46:555^561, 2001 Castelijns FA, Gerritsen MJ, van Vlijmen-Willems IM, van Erp PJ, van de Kerkhof PC: Proliferation is the main epidermal target in the treatment of psoriatic plaques with once daily application of tacalcitol ointment. Acta Derm Venereol 79:111^114, 1999 Chuah MK,Vandendriessche T, Morgan RA: Development and analysis of retroviral vectors expressing human factor VIII as a potential gene therapy for haemophilia A. Hum GeneTher 6:1363^1377, 1995 Claesson-Welsh L, Eriksson A, Moren A, et al: cDNA cloning and expression of a human platelet-derived growth factor (PDGF) receptor speci¢c for B-chaincontaining PDGF molecules. Mol Cell Biol 8:3476^3486, 1988 Cooper DM,Yu EZ, Hennessey P, Ko F, Robson MC: Determination of endogenous cytokines in chronic wounds. Ann Surg 219:688^691, 1994 Embil JM, Papp K, Sibbald G, Tousignant J, Smiell JM, Wong B, Lau CY: Recombinant human platelet-derived growth factor-BB (becaplermin) for healing chronic lower extremity diabetic ulcers: An open-label clinical evaluation of e⁄cacy. Wound Repair Regen 8:162^168, 2000 Eming SA, Lee J, Snow RG, Tompkins RG, Yarmush ML, Morgan JR: Genetically modi¢ed human epidermis over-expressing PDGF-A directs the development of a cellular and vascular connective tissue stroma when transplanted to athymic mice ^ implications for the use of genetically modi¢ed keratinocytes to modulate dermal regeneration. J Invest Dermatol 105:756^763, 1995 Eming SA, Medalie DA,Tompkins RG,Yarmush ML, Morgan JR: Genetically modi¢ed human keratinocytes overexpressing PDGF-A enhance the performance of a composite skin graft. Hum GeneTher 1:529^539, 1998 Eming SA, Yarmush ML, Krueger GG, Morgan JR: Regulation of the spatial organization of mesenchymal connective tissue: e¡ects of cell-associated versus released isoforms of platelet-derived growth factor. Am J Pathol 154:281^289, 1999 Fredriksson S, Gullberg M, Jarvius J, et al: Protein detection using proximity-dependent DNA ligation assays. Nat Biotechnol 20:473^477, 2002 Freeman AE, Igel HJ, Herrman BJ, Kleinfeld KL: Growth and characterization of human skin epithelial cell cultures. In Vitro 12:352^362, 1976 Garlick JA, Katz AB, Fenjves ES, Taichman LB: Retrovirus-mediated transduction of cultured epidermal keratinocytes. J Invest Dermatol 97:824^829, 1991 Gibbs S, Silva Pinto AN, Murli S, Huber M, Hohl D, Ponec M: Epidermal growth factor and keratinocyte growth factor di¡erentially regulate epidermal migration, growth, and di¡erentiation. Wound Repair Regen 8:192^203, 2000 Grayson LS, Hansbrough JF, Zapata-Sirvent RL, Dore CA, Morgan JL, Nicolson MA: Quantitation of cytokine levels in skin graft donor site wound £uid. Burns 19:401^405, 1993 Gubin AN, Koduru S, Njoroge JM, Bhatnagar R, Miller JL: Stable expression of green £uorescent protein after liposomal transfection of K562 cells without selective growth conditions. Biotechniques 27:1162^1170, 1999 Hanazono Y, Yu JM, Dunbar CE, Emmons RV: Green £uorescent protein retroviral vectors: Low titer and high recombination frequency suggest a selective disadvantage. Hum GeneTher 8:1313^1319, 1997 Hart CE, Seifert RA, Ross R, Bowen-Pope DF: Synthesis, phosphorylation, and degradation of multiple forms of the platelet-derived growth factor receptor studied using a monoclonal antibody. J Biol Chem 262:10780^10785, 1987 Heldin CH, Backstrom G, stman A, et al: Binding of di¡erent dimeric forms of PDGF to human ¢broblasts: Evidence for two separate receptor types. EMBO J 7:1387^1393, 1988 Heldin CH, Westermark B: Mechanism of action and in vivo role of platelet-derived growth factor. Physiol Rev 79:1283^1316, 1999 Heldin CH, Westermark B, Wasteson A: Speci¢c receptors for platelet-derived growth factor on cells derived from connective tissue and glia. Proc Natl Acad Sci 78:3664^3668, 1981 Jensen TG, Andresen BS, Jensen HK, et al: Rapid characterization of disease-causing mutations in the low density lipoprotein receptor gene (LDL-R) by overexpression in COS cells. Z Gastroenterol Suppl 3:9^12, 1996 Jensen TG, Sullivan DM, Morgan RA, Taichman LB, Nussenblatt RB, Blaese RM, Csaky KG: Retrovirus-mediated gene transfer of ornithine-delta-aminotransferase into keratinocytes from gyrate atrophy patients. Hum GeneTher 8:2125^ 2132, 1997 Liechty KW, Nesbit M, Herlyn M, Radu A, Adzick NS, Crombleholme TM: Adenoviral-mediated over-expression of platelet-derived growth factor-B corrects ischemic impaired wound healing. J Invest Dermatol 113:375^383, 1999 Liu HS, January MS, Chou CK, Chen PH, Ke NJ: Is green £uorescent protein toxic to the living cells? Biochem Biophys Res Commun 260:712^717, 1999 Margolis DJ, Crombleholme T, Herlyn M: Clinical protocol. Phase I trial to evaluate the safety of H5.020CMV. PDGF-B for the treatment of a diabetic insensate foot ulcer. Wound Repair Regen 8:480^493, 2000 Markowitz D, Hesdor¡er C,Ward M, Go¡ S, Bank A: Retroviral gene transfer using safe and e⁄cient packaging cell lines. Ann N Y Acad Sci 612:407^414, 1990 Migliaccio AR, Bengra C, Ling J, et al: Stable and unstable transgene integration sites in the human genome: Extinction of the green £uorescent protein transgene in K562 cells. Gene 256:197^214, 2000
VOL. 120, NO. 5 MAY 2003
Moulin V, Lawny F, Barritault D, Caruelle JP: Platelet releasate treatment improves skin healing in diabetic rats through endogenous growth factor secretion. Cell Mol Biol 44:961^971, 1998 Nanney LB, Paulsen S, Davidson MK, Cardwell NL, Whitsitt JS, Davidson JM: Boosting epidermal growth factor receptor expression by gene gun transfection stimulates epidermal growth in vivo. Wound Repair Regen 8: 117^127, 2000 Onodera M, Yachie A, Nelson DM, Welchlin H, Morgan RA, Blaese RM: A simple and reliable method for screening retroviral producer clones without selectable markers. Hum GeneTher 8:1189^1194, 1997 stman A, Bckstr˛m G, Fong N, et al: Expression of three recombinant homodimeric isoforms of PDGF in Saccharomyces cerevisiae: evidence for di¡erence in receptor binding and functional activities. Growth Factors 1:271^281, 1989 Ponec M, Gibbs S,Weerheim A, Kempenaar J, Mulder A, Mommaas AM: Epidermal growth factor and temperature regulate keratinocyte di¡erentiation. Arch Dermatol Res 289:317^326, 1997 Rees RS, Robson MC, Smiell JM, Perry BH: Becaplermin gel in the treatment of pressure ulcers: A phase II randomized, double-blind, placebo-controlled study. Wound Repair Regen 7:141^147, 1999 Regnier M, Asselineau D, Lenoir MC: Human epidermis reconstructed on dermal substrates in vitro: An alternative to animals in skin pharmacology. Skin Pharmacol 3:70^85, 1990 Reuterdahl C, Sundberg C, Rubin K, Funa K, Gerdin B: Tissue localization of b receptors for platelet-derived growth factor and platelet-derived growth factor B chain during wound repair in humans. J Clin Invest 91:2065^2075, 1993 Rheinwald JG, Green H: Serial cultivation of strains of human epidermal keratinocytes: the formation of keratinizing colonies from single cells. Cell 6:331^343, 1975 Rikimaru K, Moles JP, Watt FM: Correlation between hyperproliferation and suprabasal integrin expression in human epidermis reconstituted in culture. Exp Dermatol 6:214^221, 1997
PDGF RECEPTOR GENE TRANSFER TO KERATINOCYTES
749
Robson MC, Phillips LG, Thomason A, Robson LE, Pierce GF: Platelet-derived growth factor BB for the treatment of chronic pressure ulcers. Lancet 339: 23^25, 1992 Rubin K,Terracio L, R˛nnstrand L, Heldin CH, Klareskog L: Expression of plateletderived growth factor receptors is induced on connective tissue cells during chronic synovial in£ammation. Scand J Immunol 27:285^294, 1988 Sambrook J, Fritsch EF, Maniatis T In: Molecular Cloning: A Laboratory Manual. Cold Spring Harbor: Cold Spring Harbor Laboratory Press, 1989 Shukla A, Dubey MP, Srivastava R, Srivastava BS: Di¡erential expression of proteins during healing of cutaneous wounds in experimental normal and chronic models. Biochem Biophys Res Commun 244:434^439, 1998 Steed DL: Clinical evaluation of recombinant human platelet-derived growth factor for the treatment of lower extremity diabetic ulcers. Diabetic Ulcer Study Group. J Vasc Surg 21:71^78, 1995 Stoler A, Kopan R, Duvic M, Fuchs E: Use of monospeci¢c antisera and cRNA probes to localize the major changes in keratin expression during normal and abnormal epidermal di¡erentiation. J Cell Biol 107:427^446, 1988 Supp DM, Bell SM, Morgan JR, Boyce ST: Genetic modi¢cation of cultured skin substitutes by transduction of human keratinocytes and ¢broblasts with platelet-derived growth factor-A. Wound Repair Regen 8:26^35, 2000 Teumer J, Zezulak K, Green H: Measurement of speci¢c mRNA content of keratinocytes of di¡erent sizes in relation to growth and di¡erentiation. In: Leigh IM, Lane EB, Watt FM eds. The Keratinocyte Handbook, Cambridge: Cambridge University Press, 1994 Tyrone JW, Mogford JE, Chandler LA, Ma C, Xia Y, Pierce GF, Mustoe TA: Collagen-embedded platelet-derived growth factor DNA plasmid promotes wound healing in a dermal ulcer model. J Surg Res 93:230^236, 2000 Wieman TJ, Smiell JM, Su Y: E⁄cacy and safety of a topical gel formulation of recombinant human platelet-derived growth factor-BB (becaplermin) in patients with chronic neuropathic diabetic ulcers. A phase III randomized placebo-controlled double-blind study. Diabetes Care 21:822^827, 1998