0 1992 Harwood Academic Publishers GmbH
Growth F I X ~ O Y S1992, , Vol. 7, pp. 1-14 Reprints available directly from the publisher Photocopying permitted by license only
Printed in the United Kingdom
Acidic Fibroblast Growth Factor Accelerates Dermal Wound Healing THEODORE N. MELLIN, ROBERT J. MENNIE, DOREEN E. CASHEN, JOHN J. RONAN, JOANNA CAPPARELLA, MARY LOU JAMES, JERRY DISALVO, JOHN FRANK*, DAVID LINEMEYER, GUILLERMO GIMENEZ-GALLEGO and KENNETH A. THOMAS Department of Biochemist y, Merck Sharp G. Dohme Researrh Laboratories, PO Box 2000, Rahway, N\ 07065; arid *Department of Safety Assessment, Merck Sharp G. Dohme Research Laboratories, West Point, PA 29486
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
(Received September 23 1991, Accepted December 16 1991) Acidic fibroblast growth factor (aFGF) is a potent mitogen in vitro for many cells of ectodermal and mesodermal embryonic origin including skin-derived epidermal keratinocytes, dermal fibroblasts and vascular endothelial cells. Based on the mitogenic activity for these skin-derived cells, we tested the ability of topically applied aFGF to promote healing of full-thickness dermal wounds in healthy rodents. Low doses of aFGF can produce almost a two-fold maximum acceleration in the rate of closure of fullthickness dermal punch biopsy wounds in young healthy mice and rats. The mitogen also produces a 3 to 4 day acceleration in the time to complete closure in rats. Quantitative histomorphometric analysis of wound tissue shows that aFGF induces a marked stimulation of angiogenesis,, granulation tissue formation and the growth of new epithelium, but does not promote dermal contraction. Application of aFGF to linear incisions in rat skin produces a transient increase in wound tensile strength accompanied by enhanced cellularlty and deposition of collagen. Therefore, aFGF functions as a pharmacological agent that can accelerate dermal wound healing in rodents and could act therapeutically to promote dermal tissue repair in humans. KEYWORDS: wound closure, wound strength, granulation tissue, nngiogenesis, endothelial cell, BS-I lectin
INTRODUCTION
identified (Lee et al., 1989; Dionne et al., 1990; Keegan et al., 19911, some of which have been demonstrated to show preferential binding of selected FGFs (Bottaro et al., 1990). The structures and functions of aFGF and bFGF, the first two FGFs to be purified to homogeneity, have been extensively characterized. They are both relatively small (16-18 kDa) proteins that are 55% identical in their amino acid sequence and have very similar tertiary structures (Zhu et al., 1991). Both FGFs complex avidly with heparin which stabilizes them from denaturation and proteolytic degradation. However, only aFGF requires exogenous heparin for full mitogenic activity it7 vitvo on many but not all cells, presumably functioning to stabilize the active conformation of the protein under culture conditions (Ortega et al., 1991). Both aFGF and bFGF but not the other 5 known FGFs lack recognizable secretory leader sequences. Therefore,
Fibroblast growth factors (FGFs) constitute a family of homologous proteins which are potent niitogens for most, if not all, mesodermallyderived cells and many ceIls of ectodermal embryonic origin. Seven homologous family members, acidic FGF (aFGF), basic FGF (bFGF), INT-2, KFGF/HST, FGF-5, FGF-6 and keratinocyte growth factor (KGF), have been identified, purified and structurally characterized at both the protein and DNA level (reviewed in Burgess and Macaig, 1989; Klagsbrun, 1989; Rifkin and Moscatelli, 1989; Ortega and Thomas, 1990). Multiple high affinity receptors have been
Correspondence to: Dr Theodore N. Mellin. Room 80W-243. Merch, Sharp & Dohme Research Laboratories. P.O. Box 2000. Ruhway. Neu Jersey 07065.
I
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
2
MELLIN
their release in vivo could be associated with tissue damage where they might act as endogenous mitogens promoting tissue repair required for efficient wound healing. An important process in tissue repair is the neovascularization required to promote oxygenation, nutrient supply and waste removal. Both aFGF and bFGF are potent mitogens and chemotactic agents for vascular endothelial cells in vitvo and induce new blood vessel growth, or angiogenesis, in vivo (Thomas et al., 1985; Lobb et al., 1985; Herbert et al., 1988; Eppley et al., 1988; Thompson et al., 1989). They are also mitogenic for dermal fibroblasts (Katsuoka et al., 1987; Shipley et al., 1989) and keratinocytes (MillerDavis et al., 1988; O’Keefe et al., 1988; Shipley et al., 1989) in culture. Thus, these FGFs drive proliferation of the major cellular components of skin. Both heparin-stabilized aFGF and bFGF exhibit similar specific mitogenic activities for vascular endothelial cells and dermal fibroblasts, presumably reflecting binding to common plasma membrane receptors. Surprisingly, epidermal keratinocytes respond preferentially to aFGF compared to bFGF in either the presence or absence of heparin (Shipley et al., 1989). This differential mitogenic activity might reflect the abundance of a specific ”KGF” receptor that binds tighter to KGF and aFGF than to bFGF in radioreceptor assays (Bottaro et al., 1990). Similarly, aFGF was reported to be more potent than bFGF in promoting corneal epithelial repair in z7ism (Fredj-Reygrobellet et al., 1987). Partially (Buntrock et al., 1982a, 1982b, 1984) and fully (Davidson et al., 1985; Sprugel et al., 1987; Broadley et al., 1988) purified FGFs have been reported to stimulate formation of granulation tissue iii m‘z70. Furthermore, partially purified mitogens from retina, later identified as aFGF and bFGF (Baird et al., 19851, have been shown to accelerate reepithelialization following removal of the epidermis (Fourtanier et al., 1986). Basic FGF has also been demonstrated to accelerate healing of incisional wounds in rats (McGee et al., 19881, partial-thickness wounds in pigs (Hebda et al., 1990) and full-thickness wounds in healing-impaired diabetic mice (Greenhalgh et al., 1990; Tsuboi and Rifkin, 1990). These activities suggest that FGFs might play an important role in mediating the repair of injured tissue. We report here that aFGF accelerates the rate of dermal wound closure and incisional wound
i’t
al.
strength in normal healthy rodents. A preliminary account of some of these results was previously reported (Thomas et al., 1987).
MATERIALS AND METHODS aFGF Preparation Synthetic genes encoding the 140 amino acid forms of bovine and human aFGF were expressed in E . coli and the recombinant products purified to apparent homogeneity as described (Linemeyer et al., 1987, 1990). The samples of fully active mitogen used in uizw were generated by dialysis from 6 M guanidinium chloride, 5 mM dithiothreitol into Ca+2/Mg+2-freephosphate buffered saline (PBS). Recombinant bovine aFGF was used until human recombinant aFGF became available. Comparative studies demonstrated that the mitogenic activities of heparinstabilized aFGF from these two sources were equivalent.
Mitogenic Tissue Culture Assay Balb/c 3T3 cells (American Type Culture Collection, ATCC #CCL-163) were plated into 96-well dishes (Costar) at various densities produced by serial 3-fold dilutions from a high cell density stock solution. The plating densities used were: 9.4~10‘(confluence), 3.1x10J, l.OxlOJ, 3.5x1O7, 1.2 xl0’ and 0.4~10’cells/cm2. Assays of human aFGF-induced incorporation of [rnethyl-’Hlthymidine into DNA were carried out in quadruplicate in serum-free media containing insulin, transferrin and selenium (Collaborative Research), supplemented with 50 p g l m l heparin (porcine intestinal mucosa, Sigma) as described (Ortega et al., 1991). Dose-response curves were generated as a function of cell density and mitogenic activities were expressed as EDiLls. Skin Organ Culture Six mm diameter full-thickness skin punch biopsies from mice and rats were removed a t wounding as later described and placed i n organ culture in 1 ml of Dulbecco’s Modified Eagle Medium with high glucose (DMEM, Gibco) supplemented with 100 P g l m l glutamine, 100 units/ml penicillin, 100 pg/ml streptomycin,
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
aFGF ACCEI.ERATES WOUND HEALING
and 1% heat-inactivated fetal bovine serum (Gibco). Incubations were carried out in 7.5% CO, (pH 7.4) at 37 "C in 6-well tissue culture dishes (Costar). For 1-4 days plugs were rinsed twice daiIy with DMEM and placed in fresh DMEM/ l'i serum containing 50 pg/ml heparin and various doses of aFGF. Each day a portion of the plugs from every dose group was pulsed for 24 hr with 4 pCi of [niefh?/l-iH]thymidine (20 Ci/mmol, NEN) and 6 p g of unlabeled thymidine (Sigma). FoIlowing incorporation of radioactive thymidine, biopsies were rinsed twice with DMEM and placed in fresh serum-containing medium for two successive 17 and 4 hr washes to allow unincorporated radiolabeled thymidine to diffuse completely out of the plugs as determined by a series of control washout experiments (data not shown). Tissues were dissolved in Beckman BTS-450 and incorporated radiolabel counted using Beckman NA scintillation cocktail. Wound Closure Model Bovine and human recombinant aFGFs were evaluated in male Charles River CD (ICR) BR mice, 45+5 g and Sprague-Dawley rats, 170k10 g, respectively. The back and thoracic areas were shaved and disinfected with 70% ethanol. Animals were anesthetized prior to wounding by intramuscular injection of 40 mg of ketamine and 5 m g xylazine per kg body weight. To prevent disturbance of wounds by licking, all animals were individually housed and fitted with continuously adjustable velcro-fastened plastic neck collars of approximately 4.5 in. outside circumference (supplied by Ejay International, Glendora, CA) designed in this laboratory (James and DeTolla, 1987). These collars readily accommodated changes in neck size and were lined with foam neck guards to minimize abrasion. Single full-thickness 6 mm diameter (0.28 cm') excisional wounds were made on the mid-dorsum with sterile biopsy punches (Baker) and left undressed. Wound area measurements were made immediately after wounding and daily therafter by tracing the perimeter of each wound in triplicate onto a sterile glass slide. Areas were determined from the tracings by digital planimetry (Micro-Plan I1 image analysis system, Donsanto Corp., Natick, Mass.) and expressed as percentages of the area measured immediately after wounding. Animals were randomized with
3
respect to treatment, cage location and tracing order. To avoid potential bias, wound tracings and subsequent planimetry were made in a blinded fashion by individuals who did not know which treatment was received by each animal. Mouse wounds (N=18 mice/group) were treated topically twice daily for 10 consecutive days after injury with 20 pulldose of either a placebo control containing 2 pg each of heparin and mouse serum albumin (Sigma) in PBS or the same placebo solution Containing 0.25 p g of bovine aFGF (0.5 ,ug/day). Wounds were photographed at 1, 4, 7 and 10 days post-injury. Scabs were removed on day 6 to accurately visualize the wound for perimeter tracing and to facilitate dose access to the wound. Rat punch biopsy wounds (N=22/group) received two equal 20 pl daily doses totaling either 0 to 1.0 pg of human aFGF in PBS containing 3 p g of heparin and 10 pg of rat serum a h min (Sigma) for 10 consecutive days after injury. Wound closure was also characterized in rats (N= 18/group) as a function of aFGF daily dose (0, 0.10, 1.0 and l 0 p g ) containing a 3-fold and 10fold mass excess of heparin and rat serum albumin (Sigma), respectively. Heparin and albumin, observed to have no detectable effect on the rate of wound closure (data not shown), were included in the placebo control at a level corresponding to that present in the highest dose of aFGF. The effect of dose frequency was examined by treating the rat biopsy wounds with two equal daily doses totaling either 0 or 1 . 0 p g l d a y of human aFGF administered either (1) immediately after wounding (single dose), (2) immediately, 5 and 8 days after injury (intermittent dose), or ( 3 ) daily for 10 days (daily dose). Both aFGF and placebo control doses contained 3 pg of heparin and 10 p g of albumin. Scabs were removed 5 and 8 days post-injury, immediately prior to wound measurement and dosing. Statistical significances of responses were evaluated using either an unpaired Student's f-test or a 2-test. Since aFGF was observed in over a dozen studies to accelerate the rate of wound closure, but never to inhibit it, the statistical significance of this response was evaluated using a 1-tailed t-test (Daniel, 1991). All studies described were carried out in accord with rules and guidelines of the Merck Sharp & Dohme Research Laboratories Institutional Animal Care
4
MELLIN et al.
and Use Committee and the "Guide for the Care and Use of Laboratory Animals" [DHSS Publication No. (NIH) 85-23, revised 19851.
phenyl phosphate substrate (Sigma) were added. The alkaline phosphate reaction proceeded at 37 "C with colorimetric measurements made at 405 nm every 15 min for 1 hr.
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
Wound Incisional Model Linear incisions of 6 cm length were made paravertebrally on the dorsal aspect of anesthetized male Sprague-Dawley rats (600f50 g) and closed with 3 surgical clips as described by Mustoe et al. (1987). Twenty-four rats each received 2.0pg of human aFGF in 25Opl of PBS containing 25pg each of heparin and rat serum albumin once daily for 14 days (0.4 p g topical on incision margins and 0.4pg s . ~ at . the base of the incision at 1.5cm intervals). A second group of 24 rats received a PBS placebo control containing only heparin and rat serum albumin. Six placebo control and six aFGF-treated rats were sacrificed at each of four time intervals (4,7, 10 and 14 days) post-injury. Four 8 m m wide wound strips were excised from the dorsal skin of each rat between the wound clips, using a double bladed template as described by Mustoe et al. (1987). Breaking strength was measured immediately after excision on two of the four skin strips using a custom-made tensometer. Tension was applied at a constant rate of 1 cm/min from the screw drive of an infusion pump (Harvard Apparatus Model 906, Harvard Apparatus, Natrick, MA), using a 2.0 kg force transducer (Gould, Oxnard, CA). Breaking strength was recorded on an X-Y recorder as the point of maximal stress before wound separation. Statistical significances of responses were evaluated using an unpaired 2-tailed Student's t-test.
Analysis of Immunological Response Sera were collected from the orbital sinus of mice and rats at day 10 for ELISA assays to evaluate the immunogenicity of aFGF. Immulon 196-well plates (Dynatech Laboratories, Inc.) were coated with l o n g of aFGF per well, rinsed and incubated 1 hr at 37 "C with either mouse or rat sera diluted 1:50 and 1:100, respectively. Following rinses to remove unbound primary antibody, a second 1 hr 37 "C incubation was performed with a 1 : l O O O dilution of goat anti-rabbit IgG conjugated to alkaline phosphate (Miles-Yeda, LTDMiles Scientific, Naperville, IL). The wells were again rinsed and 100 p1 of 1 mg/ml p-nitro-
Histomorphometric Analysis of Wound Repair Groups of 10 rats, treated daily with either 1.0 p g of human aFGF or placebo, were sacrificed at 5 and 8 days after injury and standard punchbiopsy wound sites were excised. Wound tissue, with adjacent unwounded skin, was pinned on dental wax to maintain normal morphology and fixed in a Bouins/lO% Zn formalin solution (Anatech Ltd.). Each wound was bisected and the halves embedded in paraffin. Tissue blocks were cut at 4 p m thickness and central sections through the wound were deparaffinized, cleared and rehydrated. Vascular endothelial cells in full-thickness wound sections were visualized with a modified Bandeira simplicifolia agglutinin-I (BS-I lectin) /avidin-biotin-peroxidase complex (ABC) method (Alroy et al., 1987). Selectivity was improved by using a single a-D-galactosespecific BS-I isolectin B4, rather than the standard BS-I mixture of 5 isolectins (Vector Labs). Signal amplification was achieved by a double antibody procedure using an affinity-purified anti-BS-I lectin primary goat antibody followed by an affinity-purified biotinylated rabbit anti-goat secondary antibody. Avidin linked to horseradish peroxidase (Vectastin ABC Peroxidase Kit) was bound to the biotinylated second antibody and visualized by deposition of peroxidase-oxidized diaminobenzidine (Bjornsson et al., 1991). Computer-assisted morphometric analysis of representative mid-wound histological sections was performed with a Joyce-Loebl Magiscan MD image analyzer. Images were captured at 40x magnification with a Sony RGB color video camera mounted on a Zeiss Axioskop compound microscope and displayed on a high resolution color monitor. Granulation tissue zones in the wound bed were defined by the abundant BS-I stained blood vessels between the edges of normal dermal tissue identified by its characteristic highly organized collagen fibril staining. Wound contraction was assessed by measuring the distance between these wound margins at a middermal level. Measurements of both granulation tissue and vascular cross-sectional areas were
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
aFGF ACCELERATES WOUND HEALING
then made automatically by computer digital image processing. Vascular cross-sectional areas were computed including vascular luminal areas. The lengths and cross-sectional areas of new epidermis were also measured. Incisional wounds, contained in two of four dorsal skin strips harvested from each rat, were fixed 10% (v/v> formalin. Five pm thick sagittal sections were cut through the center of each paraffin embedded wound and stained with Masson trichrome for evaluation of wound cellularity and new collagen formation as described by McGee et al., 1988 and Pierce et al., 1988. Briefly, the area of new collagen was visualized by its lighter staining and cellularity and was quantitated by computerized image analysis. Cellularity in the region of newly deposited collagen was measured as the fractional cross-sectional area of cell bodies. Day 4 wounds were not included in this analysis because they were too fragile to provide high quality sections. Statistical significances of histological responses of the punch biopsy and incisional wounds were evaluated using an unpaired 2-tailed Student's t-test, since either promotion or inhibition of the individual parameters could have been observed.
RESULTS
I
1000
-
h
\
Q
W
0
0" LLI
10 loo
1.o
5 I
I
I
I 111'1
1
I
1
1 I l l "
I
I
I
I
I
I
1 1 1 1 1 l
E
L d I1 103
I
11111
1
104 105 Cell Density (cellkm2)
FIGURE 1. Activity of aFGF as a function of cell density. Human aFGF was assayed on Balb/c 3T3 cells plated at a series of densities from very sparse to a tightly packed confluent monolayer. Full dose-response curves of [methyl'Hlthymidine incorporation into DNA were used to calculate ED=,& of aFGF in the presence of 50pg/ml heparin as a function of cell density.
490 pg/ml in tightly packed monolayers. Thus, as the cell density increases, greater concentrations of aFGF are required to stimulate these cells.
Organ Culture
Based on these density-dependent cell culture responses, aFGF was assayed on rodent skin biopsies in organ culture. This was done to deterThe mitogenic activity of aFGF was originally mine if primary skin cells in an organized matrix measured on subconfluent mouse Balb/c 3T3 respond to aFGF and, if so, at what doses. A low fibroblasts in 6-well assay dishes (Thomas et al., concentration of serum was included in the cul1984). In order to increase assay capacity, while ture media not only to extend the viability of the maintaining adequate levels of [met/~yZ-~H]thymi-cultured skin but also to better emulate the dine incorporation into DNA, we initiated assays environment of a wound in vivo. Mouse 6 m m diameter punch biopsy skin with confluent cells using 96-well plates (Linemeyer et al., 1990). These changes unexpec- plugs incorporated [ methyI-3H]thymidine in tedly increased the quantity of aFGF needed to response to 0.5 p g l m l daily doses of bovine aFGF produce half-maximal mitogenic stimulation. (Fig. 2A). Subsequently, the thicker rat punch Since the mitogenic response as a function of cell biopsies were treated daily with 0,0.025,0.25, 2.5 density might be useful in extrapolating to effect- or 25pg/ml human aFGF. After one day of ive doses in three-dimensional tissues, we exposure to mitogenic stimulation, only a very characterized the EDSoas a function of cell den- shallow dose-response curve was evident (Fig. sity in monolayer cultures. A clear relationship 28). On days 2 and 3, however, a clear response was between EDsoand cell density emerged as shown peak of incorporated [ rr~ethyl-~Hlthymidine in Fig. 1. Half-maximal stimulation of DNA syn- apparent at 0.25-2.5 pg/ml with suboptimal thesis ranged from a low aFGF concentration of incorporation of radiolabel at higher concen1.4pg/ml at very sparse cell density to trations of aFGF. Plugs remained viable for 4
Mitogenic Response as a Function of Cell Density
MELLIN et a/.
6
days after which time they deteriorated both in ability to incorporate thymidine and in structural integrity .
A 16
I
I
I
I
T
12 8
aFGF Acceleration of Full-Thickness Excisional Wound Healing Punch biopsies were made in mice and rats that removed both the epidermis and entire dermis. This full-thickness injury was chosen in an attempt to simulate deep dermal decubitus ("bed-sores") and diabetic ulcers in humans. Such wounds frequently penetrate through the dermis and are often debrided to remove necrotic tissue prior to treatment thereby generating a deep wound bed similar to that produced by punch biopsies in the rodents. Measurements of average wound area (Fig. 3A)
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
4
0 I
I
I
1
I
1
2
3
4
1
Treatment Time (days)
A
1.1 $
B
1.4
1.8
1.3
1.4
1.2
(Fold Decrease)
50
30 a-" 20
100
0
0.01
0.1
1.0
10
2
100
aFGF (uglmllday) FlGljRE 2. Response c)f cultured skin punch biopsv plugs to aFCF. Excised 6 nim diameter full-thickness skin plugs were incubated at 37 "C in 1% heat-in'lctiv'~te~1fetal bovine serum and SO pg/inl heparin. ( A ) Duplicate mouse skin plugs \vt'rt\ labelled tor 1 day with [iJi[,~/i!//-~l~ltliyiiiidiii~, after incuhatlon for 1-4 d a y s with either 0 (squares) o r 0.5 (circles) pg/nil ot bovine aFGF. (B) R a t skin plugs were trtvted daily tor I (circles), 2 (squares) o r 3 (dianioiids) day5 w i t h either no o r increasing doses o f h u m a n aFCF. Each d r ~ hv plugs t r o n i L'acli dose group were labelled for 1 day nit11 [ ~ ~ r c t / r ~ ~ / - ' H ~ t l ~ y ~ ~ i i ~ l ~ ~ ~ ~ ~ tollowtd by one d a y of radiolabel washout. Pluss were digested and ~ n c o r p o r a t e dradiol,ibel Lva'; counted r h t a art' plotted as nieans+SEM with error ranges sni,illt~r than tht. svnibol 5iYe of the mean n o t showw.
6 8 Days Post-Injury
4
10
aFGF ACCELERATES WOUND HEALING
the average size and gross appearance of fullthickness mouse wounds treated with either bovine aFGF or the corresponding placebo are shown in Fig. 3B. A clear aFGF-mediated increase in wound closure is readily apparent a t 4 and 7 days post-injury. In order to compensate for the thicker skin of the rat, the daily dose of human aFGF used to treat full-thickness wounds was doubled to 1.O
demonstrate a 1.4-fold (P
Growth F I X ~ O Y S1992, , Vol. 7, pp. 1-14 Reprints available directly from the publisher Photocopying permitted by license only
Printed in the United Kingdom
Acidic Fibroblast Growth Factor Accelerates Dermal Wound Healing THEODORE N. MELLIN, ROBERT J. MENNIE, DOREEN E. CASHEN, JOHN J. RONAN, JOANNA CAPPARELLA, MARY LOU JAMES, JERRY DISALVO, JOHN FRANK*, DAVID LINEMEYER, GUILLERMO GIMENEZ-GALLEGO and KENNETH A. THOMAS Department of Biochemist y, Merck Sharp G. Dohme Researrh Laboratories, PO Box 2000, Rahway, N\ 07065; arid *Department of Safety Assessment, Merck Sharp G. Dohme Research Laboratories, West Point, PA 29486
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
(Received September 23 1991, Accepted December 16 1991) Acidic fibroblast growth factor (aFGF) is a potent mitogen in vitro for many cells of ectodermal and mesodermal embryonic origin including skin-derived epidermal keratinocytes, dermal fibroblasts and vascular endothelial cells. Based on the mitogenic activity for these skin-derived cells, we tested the ability of topically applied aFGF to promote healing of full-thickness dermal wounds in healthy rodents. Low doses of aFGF can produce almost a two-fold maximum acceleration in the rate of closure of fullthickness dermal punch biopsy wounds in young healthy mice and rats. The mitogen also produces a 3 to 4 day acceleration in the time to complete closure in rats. Quantitative histomorphometric analysis of wound tissue shows that aFGF induces a marked stimulation of angiogenesis,, granulation tissue formation and the growth of new epithelium, but does not promote dermal contraction. Application of aFGF to linear incisions in rat skin produces a transient increase in wound tensile strength accompanied by enhanced cellularlty and deposition of collagen. Therefore, aFGF functions as a pharmacological agent that can accelerate dermal wound healing in rodents and could act therapeutically to promote dermal tissue repair in humans. KEYWORDS: wound closure, wound strength, granulation tissue, nngiogenesis, endothelial cell, BS-I lectin
INTRODUCTION
identified (Lee et al., 1989; Dionne et al., 1990; Keegan et al., 19911, some of which have been demonstrated to show preferential binding of selected FGFs (Bottaro et al., 1990). The structures and functions of aFGF and bFGF, the first two FGFs to be purified to homogeneity, have been extensively characterized. They are both relatively small (16-18 kDa) proteins that are 55% identical in their amino acid sequence and have very similar tertiary structures (Zhu et al., 1991). Both FGFs complex avidly with heparin which stabilizes them from denaturation and proteolytic degradation. However, only aFGF requires exogenous heparin for full mitogenic activity it7 vitvo on many but not all cells, presumably functioning to stabilize the active conformation of the protein under culture conditions (Ortega et al., 1991). Both aFGF and bFGF but not the other 5 known FGFs lack recognizable secretory leader sequences. Therefore,
Fibroblast growth factors (FGFs) constitute a family of homologous proteins which are potent niitogens for most, if not all, mesodermallyderived cells and many ceIls of ectodermal embryonic origin. Seven homologous family members, acidic FGF (aFGF), basic FGF (bFGF), INT-2, KFGF/HST, FGF-5, FGF-6 and keratinocyte growth factor (KGF), have been identified, purified and structurally characterized at both the protein and DNA level (reviewed in Burgess and Macaig, 1989; Klagsbrun, 1989; Rifkin and Moscatelli, 1989; Ortega and Thomas, 1990). Multiple high affinity receptors have been
Correspondence to: Dr Theodore N. Mellin. Room 80W-243. Merch, Sharp & Dohme Research Laboratories. P.O. Box 2000. Ruhway. Neu Jersey 07065.
I
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
2
MELLIN
their release in vivo could be associated with tissue damage where they might act as endogenous mitogens promoting tissue repair required for efficient wound healing. An important process in tissue repair is the neovascularization required to promote oxygenation, nutrient supply and waste removal. Both aFGF and bFGF are potent mitogens and chemotactic agents for vascular endothelial cells in vitvo and induce new blood vessel growth, or angiogenesis, in vivo (Thomas et al., 1985; Lobb et al., 1985; Herbert et al., 1988; Eppley et al., 1988; Thompson et al., 1989). They are also mitogenic for dermal fibroblasts (Katsuoka et al., 1987; Shipley et al., 1989) and keratinocytes (MillerDavis et al., 1988; O’Keefe et al., 1988; Shipley et al., 1989) in culture. Thus, these FGFs drive proliferation of the major cellular components of skin. Both heparin-stabilized aFGF and bFGF exhibit similar specific mitogenic activities for vascular endothelial cells and dermal fibroblasts, presumably reflecting binding to common plasma membrane receptors. Surprisingly, epidermal keratinocytes respond preferentially to aFGF compared to bFGF in either the presence or absence of heparin (Shipley et al., 1989). This differential mitogenic activity might reflect the abundance of a specific ”KGF” receptor that binds tighter to KGF and aFGF than to bFGF in radioreceptor assays (Bottaro et al., 1990). Similarly, aFGF was reported to be more potent than bFGF in promoting corneal epithelial repair in z7ism (Fredj-Reygrobellet et al., 1987). Partially (Buntrock et al., 1982a, 1982b, 1984) and fully (Davidson et al., 1985; Sprugel et al., 1987; Broadley et al., 1988) purified FGFs have been reported to stimulate formation of granulation tissue iii m‘z70. Furthermore, partially purified mitogens from retina, later identified as aFGF and bFGF (Baird et al., 19851, have been shown to accelerate reepithelialization following removal of the epidermis (Fourtanier et al., 1986). Basic FGF has also been demonstrated to accelerate healing of incisional wounds in rats (McGee et al., 19881, partial-thickness wounds in pigs (Hebda et al., 1990) and full-thickness wounds in healing-impaired diabetic mice (Greenhalgh et al., 1990; Tsuboi and Rifkin, 1990). These activities suggest that FGFs might play an important role in mediating the repair of injured tissue. We report here that aFGF accelerates the rate of dermal wound closure and incisional wound
i’t
al.
strength in normal healthy rodents. A preliminary account of some of these results was previously reported (Thomas et al., 1987).
MATERIALS AND METHODS aFGF Preparation Synthetic genes encoding the 140 amino acid forms of bovine and human aFGF were expressed in E . coli and the recombinant products purified to apparent homogeneity as described (Linemeyer et al., 1987, 1990). The samples of fully active mitogen used in uizw were generated by dialysis from 6 M guanidinium chloride, 5 mM dithiothreitol into Ca+2/Mg+2-freephosphate buffered saline (PBS). Recombinant bovine aFGF was used until human recombinant aFGF became available. Comparative studies demonstrated that the mitogenic activities of heparinstabilized aFGF from these two sources were equivalent.
Mitogenic Tissue Culture Assay Balb/c 3T3 cells (American Type Culture Collection, ATCC #CCL-163) were plated into 96-well dishes (Costar) at various densities produced by serial 3-fold dilutions from a high cell density stock solution. The plating densities used were: 9.4~10‘(confluence), 3.1x10J, l.OxlOJ, 3.5x1O7, 1.2 xl0’ and 0.4~10’cells/cm2. Assays of human aFGF-induced incorporation of [rnethyl-’Hlthymidine into DNA were carried out in quadruplicate in serum-free media containing insulin, transferrin and selenium (Collaborative Research), supplemented with 50 p g l m l heparin (porcine intestinal mucosa, Sigma) as described (Ortega et al., 1991). Dose-response curves were generated as a function of cell density and mitogenic activities were expressed as EDiLls. Skin Organ Culture Six mm diameter full-thickness skin punch biopsies from mice and rats were removed a t wounding as later described and placed i n organ culture in 1 ml of Dulbecco’s Modified Eagle Medium with high glucose (DMEM, Gibco) supplemented with 100 P g l m l glutamine, 100 units/ml penicillin, 100 pg/ml streptomycin,
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
aFGF ACCEI.ERATES WOUND HEALING
and 1% heat-inactivated fetal bovine serum (Gibco). Incubations were carried out in 7.5% CO, (pH 7.4) at 37 "C in 6-well tissue culture dishes (Costar). For 1-4 days plugs were rinsed twice daiIy with DMEM and placed in fresh DMEM/ l'i serum containing 50 pg/ml heparin and various doses of aFGF. Each day a portion of the plugs from every dose group was pulsed for 24 hr with 4 pCi of [niefh?/l-iH]thymidine (20 Ci/mmol, NEN) and 6 p g of unlabeled thymidine (Sigma). FoIlowing incorporation of radioactive thymidine, biopsies were rinsed twice with DMEM and placed in fresh serum-containing medium for two successive 17 and 4 hr washes to allow unincorporated radiolabeled thymidine to diffuse completely out of the plugs as determined by a series of control washout experiments (data not shown). Tissues were dissolved in Beckman BTS-450 and incorporated radiolabel counted using Beckman NA scintillation cocktail. Wound Closure Model Bovine and human recombinant aFGFs were evaluated in male Charles River CD (ICR) BR mice, 45+5 g and Sprague-Dawley rats, 170k10 g, respectively. The back and thoracic areas were shaved and disinfected with 70% ethanol. Animals were anesthetized prior to wounding by intramuscular injection of 40 mg of ketamine and 5 m g xylazine per kg body weight. To prevent disturbance of wounds by licking, all animals were individually housed and fitted with continuously adjustable velcro-fastened plastic neck collars of approximately 4.5 in. outside circumference (supplied by Ejay International, Glendora, CA) designed in this laboratory (James and DeTolla, 1987). These collars readily accommodated changes in neck size and were lined with foam neck guards to minimize abrasion. Single full-thickness 6 mm diameter (0.28 cm') excisional wounds were made on the mid-dorsum with sterile biopsy punches (Baker) and left undressed. Wound area measurements were made immediately after wounding and daily therafter by tracing the perimeter of each wound in triplicate onto a sterile glass slide. Areas were determined from the tracings by digital planimetry (Micro-Plan I1 image analysis system, Donsanto Corp., Natick, Mass.) and expressed as percentages of the area measured immediately after wounding. Animals were randomized with
3
respect to treatment, cage location and tracing order. To avoid potential bias, wound tracings and subsequent planimetry were made in a blinded fashion by individuals who did not know which treatment was received by each animal. Mouse wounds (N=18 mice/group) were treated topically twice daily for 10 consecutive days after injury with 20 pulldose of either a placebo control containing 2 pg each of heparin and mouse serum albumin (Sigma) in PBS or the same placebo solution Containing 0.25 p g of bovine aFGF (0.5 ,ug/day). Wounds were photographed at 1, 4, 7 and 10 days post-injury. Scabs were removed on day 6 to accurately visualize the wound for perimeter tracing and to facilitate dose access to the wound. Rat punch biopsy wounds (N=22/group) received two equal 20 pl daily doses totaling either 0 to 1.0 pg of human aFGF in PBS containing 3 p g of heparin and 10 pg of rat serum a h min (Sigma) for 10 consecutive days after injury. Wound closure was also characterized in rats (N= 18/group) as a function of aFGF daily dose (0, 0.10, 1.0 and l 0 p g ) containing a 3-fold and 10fold mass excess of heparin and rat serum albumin (Sigma), respectively. Heparin and albumin, observed to have no detectable effect on the rate of wound closure (data not shown), were included in the placebo control at a level corresponding to that present in the highest dose of aFGF. The effect of dose frequency was examined by treating the rat biopsy wounds with two equal daily doses totaling either 0 or 1 . 0 p g l d a y of human aFGF administered either (1) immediately after wounding (single dose), (2) immediately, 5 and 8 days after injury (intermittent dose), or ( 3 ) daily for 10 days (daily dose). Both aFGF and placebo control doses contained 3 pg of heparin and 10 p g of albumin. Scabs were removed 5 and 8 days post-injury, immediately prior to wound measurement and dosing. Statistical significances of responses were evaluated using either an unpaired Student's f-test or a 2-test. Since aFGF was observed in over a dozen studies to accelerate the rate of wound closure, but never to inhibit it, the statistical significance of this response was evaluated using a 1-tailed t-test (Daniel, 1991). All studies described were carried out in accord with rules and guidelines of the Merck Sharp & Dohme Research Laboratories Institutional Animal Care
4
MELLIN et al.
and Use Committee and the "Guide for the Care and Use of Laboratory Animals" [DHSS Publication No. (NIH) 85-23, revised 19851.
phenyl phosphate substrate (Sigma) were added. The alkaline phosphate reaction proceeded at 37 "C with colorimetric measurements made at 405 nm every 15 min for 1 hr.
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
Wound Incisional Model Linear incisions of 6 cm length were made paravertebrally on the dorsal aspect of anesthetized male Sprague-Dawley rats (600f50 g) and closed with 3 surgical clips as described by Mustoe et al. (1987). Twenty-four rats each received 2.0pg of human aFGF in 25Opl of PBS containing 25pg each of heparin and rat serum albumin once daily for 14 days (0.4 p g topical on incision margins and 0.4pg s . ~ at . the base of the incision at 1.5cm intervals). A second group of 24 rats received a PBS placebo control containing only heparin and rat serum albumin. Six placebo control and six aFGF-treated rats were sacrificed at each of four time intervals (4,7, 10 and 14 days) post-injury. Four 8 m m wide wound strips were excised from the dorsal skin of each rat between the wound clips, using a double bladed template as described by Mustoe et al. (1987). Breaking strength was measured immediately after excision on two of the four skin strips using a custom-made tensometer. Tension was applied at a constant rate of 1 cm/min from the screw drive of an infusion pump (Harvard Apparatus Model 906, Harvard Apparatus, Natrick, MA), using a 2.0 kg force transducer (Gould, Oxnard, CA). Breaking strength was recorded on an X-Y recorder as the point of maximal stress before wound separation. Statistical significances of responses were evaluated using an unpaired 2-tailed Student's t-test.
Analysis of Immunological Response Sera were collected from the orbital sinus of mice and rats at day 10 for ELISA assays to evaluate the immunogenicity of aFGF. Immulon 196-well plates (Dynatech Laboratories, Inc.) were coated with l o n g of aFGF per well, rinsed and incubated 1 hr at 37 "C with either mouse or rat sera diluted 1:50 and 1:100, respectively. Following rinses to remove unbound primary antibody, a second 1 hr 37 "C incubation was performed with a 1 : l O O O dilution of goat anti-rabbit IgG conjugated to alkaline phosphate (Miles-Yeda, LTDMiles Scientific, Naperville, IL). The wells were again rinsed and 100 p1 of 1 mg/ml p-nitro-
Histomorphometric Analysis of Wound Repair Groups of 10 rats, treated daily with either 1.0 p g of human aFGF or placebo, were sacrificed at 5 and 8 days after injury and standard punchbiopsy wound sites were excised. Wound tissue, with adjacent unwounded skin, was pinned on dental wax to maintain normal morphology and fixed in a Bouins/lO% Zn formalin solution (Anatech Ltd.). Each wound was bisected and the halves embedded in paraffin. Tissue blocks were cut at 4 p m thickness and central sections through the wound were deparaffinized, cleared and rehydrated. Vascular endothelial cells in full-thickness wound sections were visualized with a modified Bandeira simplicifolia agglutinin-I (BS-I lectin) /avidin-biotin-peroxidase complex (ABC) method (Alroy et al., 1987). Selectivity was improved by using a single a-D-galactosespecific BS-I isolectin B4, rather than the standard BS-I mixture of 5 isolectins (Vector Labs). Signal amplification was achieved by a double antibody procedure using an affinity-purified anti-BS-I lectin primary goat antibody followed by an affinity-purified biotinylated rabbit anti-goat secondary antibody. Avidin linked to horseradish peroxidase (Vectastin ABC Peroxidase Kit) was bound to the biotinylated second antibody and visualized by deposition of peroxidase-oxidized diaminobenzidine (Bjornsson et al., 1991). Computer-assisted morphometric analysis of representative mid-wound histological sections was performed with a Joyce-Loebl Magiscan MD image analyzer. Images were captured at 40x magnification with a Sony RGB color video camera mounted on a Zeiss Axioskop compound microscope and displayed on a high resolution color monitor. Granulation tissue zones in the wound bed were defined by the abundant BS-I stained blood vessels between the edges of normal dermal tissue identified by its characteristic highly organized collagen fibril staining. Wound contraction was assessed by measuring the distance between these wound margins at a middermal level. Measurements of both granulation tissue and vascular cross-sectional areas were
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
aFGF ACCELERATES WOUND HEALING
then made automatically by computer digital image processing. Vascular cross-sectional areas were computed including vascular luminal areas. The lengths and cross-sectional areas of new epidermis were also measured. Incisional wounds, contained in two of four dorsal skin strips harvested from each rat, were fixed 10% (v/v> formalin. Five pm thick sagittal sections were cut through the center of each paraffin embedded wound and stained with Masson trichrome for evaluation of wound cellularity and new collagen formation as described by McGee et al., 1988 and Pierce et al., 1988. Briefly, the area of new collagen was visualized by its lighter staining and cellularity and was quantitated by computerized image analysis. Cellularity in the region of newly deposited collagen was measured as the fractional cross-sectional area of cell bodies. Day 4 wounds were not included in this analysis because they were too fragile to provide high quality sections. Statistical significances of histological responses of the punch biopsy and incisional wounds were evaluated using an unpaired 2-tailed Student's t-test, since either promotion or inhibition of the individual parameters could have been observed.
RESULTS
I
1000
-
h
\
Q
W
0
0" LLI
10 loo
1.o
5 I
I
I
I 111'1
1
I
1
1 I l l "
I
I
I
I
I
I
1 1 1 1 1 l
E
L d I1 103
I
11111
1
104 105 Cell Density (cellkm2)
FIGURE 1. Activity of aFGF as a function of cell density. Human aFGF was assayed on Balb/c 3T3 cells plated at a series of densities from very sparse to a tightly packed confluent monolayer. Full dose-response curves of [methyl'Hlthymidine incorporation into DNA were used to calculate ED=,& of aFGF in the presence of 50pg/ml heparin as a function of cell density.
490 pg/ml in tightly packed monolayers. Thus, as the cell density increases, greater concentrations of aFGF are required to stimulate these cells.
Organ Culture
Based on these density-dependent cell culture responses, aFGF was assayed on rodent skin biopsies in organ culture. This was done to deterThe mitogenic activity of aFGF was originally mine if primary skin cells in an organized matrix measured on subconfluent mouse Balb/c 3T3 respond to aFGF and, if so, at what doses. A low fibroblasts in 6-well assay dishes (Thomas et al., concentration of serum was included in the cul1984). In order to increase assay capacity, while ture media not only to extend the viability of the maintaining adequate levels of [met/~yZ-~H]thymi-cultured skin but also to better emulate the dine incorporation into DNA, we initiated assays environment of a wound in vivo. Mouse 6 m m diameter punch biopsy skin with confluent cells using 96-well plates (Linemeyer et al., 1990). These changes unexpec- plugs incorporated [ methyI-3H]thymidine in tedly increased the quantity of aFGF needed to response to 0.5 p g l m l daily doses of bovine aFGF produce half-maximal mitogenic stimulation. (Fig. 2A). Subsequently, the thicker rat punch Since the mitogenic response as a function of cell biopsies were treated daily with 0,0.025,0.25, 2.5 density might be useful in extrapolating to effect- or 25pg/ml human aFGF. After one day of ive doses in three-dimensional tissues, we exposure to mitogenic stimulation, only a very characterized the EDSoas a function of cell den- shallow dose-response curve was evident (Fig. sity in monolayer cultures. A clear relationship 28). On days 2 and 3, however, a clear response was between EDsoand cell density emerged as shown peak of incorporated [ rr~ethyl-~Hlthymidine in Fig. 1. Half-maximal stimulation of DNA syn- apparent at 0.25-2.5 pg/ml with suboptimal thesis ranged from a low aFGF concentration of incorporation of radiolabel at higher concen1.4pg/ml at very sparse cell density to trations of aFGF. Plugs remained viable for 4
Mitogenic Response as a Function of Cell Density
MELLIN et a/.
6
days after which time they deteriorated both in ability to incorporate thymidine and in structural integrity .
A 16
I
I
I
I
T
12 8
aFGF Acceleration of Full-Thickness Excisional Wound Healing Punch biopsies were made in mice and rats that removed both the epidermis and entire dermis. This full-thickness injury was chosen in an attempt to simulate deep dermal decubitus ("bed-sores") and diabetic ulcers in humans. Such wounds frequently penetrate through the dermis and are often debrided to remove necrotic tissue prior to treatment thereby generating a deep wound bed similar to that produced by punch biopsies in the rodents. Measurements of average wound area (Fig. 3A)
Growth Factors Downloaded from informahealthcare.com by University of Auckland on 11/13/14 For personal use only.
4
0 I
I
I
1
I
1
2
3
4
1
Treatment Time (days)
A
1.1 $
B
1.4
1.8
1.3
1.4
1.2
(Fold Decrease)
50
30 a-" 20
100
0
0.01
0.1
1.0
10
2
100
aFGF (uglmllday) FlGljRE 2. Response c)f cultured skin punch biopsv plugs to aFCF. Excised 6 nim diameter full-thickness skin plugs were incubated at 37 "C in 1% heat-in'lctiv'~te~1fetal bovine serum and SO pg/inl heparin. ( A ) Duplicate mouse skin plugs \vt'rt\ labelled tor 1 day with [iJi[,~/i!//-~l~ltliyiiiidiii~, after incuhatlon for 1-4 d a y s with either 0 (squares) o r 0.5 (circles) pg/nil ot bovine aFGF. (B) R a t skin plugs were trtvted daily tor I (circles), 2 (squares) o r 3 (dianioiids) day5 w i t h either no o r increasing doses o f h u m a n aFCF. Each d r ~ hv plugs t r o n i L'acli dose group were labelled for 1 day nit11 [ ~ ~ r c t / r ~ ~ / - ' H ~ t l ~ y ~ ~ i i ~ l ~ ~ ~ ~ ~ tollowtd by one d a y of radiolabel washout. Pluss were digested and ~ n c o r p o r a t e dradiol,ibel Lva'; counted r h t a art' plotted as nieans+SEM with error ranges sni,illt~r than tht. svnibol 5iYe of the mean n o t showw.
6 8 Days Post-Injury
4
10
aFGF ACCELERATES WOUND HEALING
the average size and gross appearance of fullthickness mouse wounds treated with either bovine aFGF or the corresponding placebo are shown in Fig. 3B. A clear aFGF-mediated increase in wound closure is readily apparent a t 4 and 7 days post-injury. In order to compensate for the thicker skin of the rat, the daily dose of human aFGF used to treat full-thickness wounds was doubled to 1.O
demonstrate a 1.4-fold (P
