wounded skin at various stages during the transition of wound granulation ... blasts,10 a mechanism that extends or amplifies its action in the repair process.
American Journal of Pathology, Vol. 152, No. 2, February 1998 Copyright American Society for Investigative Pathology
Enhanced Expression of Transforming Growth Factor-f3 Type I and Type 11 Receptors in Wound Granulation Tissue and Hypertrophic Scar
Peter Schmid,* Peter Itin,t George Chen Bi,§ and David A. Cox*
Cherry,*
From Dermatology Research,* Novartis Pharma AG, and the Department of Dermatology,t University of Basel, Basel, Switzerland; the Department ofDermatology,* Churchill Hospital, Oxford, United Kingdom; and the Department of Burn Surgery,5 Xijing Hospital, Xi'an, People's Republic of China
In the present study we have analyzed and compared, by immunohistochemistry and in situ hybridization, the expression pattern of the R4/ALK5 transforming growth factor (TGF)-f3 type I receptor (RI) and the TGF-.8 type II receptor (RII) in normal human skin, in wounded skin at various stages during the transition of wound granulation tissue to scar, and in longpersisting post-burn hypertrophic scars. In normal human skin, expression of RI and RII was clearly visible in the epidermis, in epidermal appendages, and in vascular cells, although only a small number of dermal fibroblasts revealed detectable levels of TGF-,f receptor expression. In contrast, granulation tissue fibroblasts showed strong expression of both TGF-j8 receptor types, although in normal-healing excisional wounds their density decreased during granulation tissue remodeling. However, in post-burn hypertrophic scars, RI- and RI-overexpressing fibroblasts were found in high densities up to 20 months after injury. From these findings we suggest that the repair process of deep wounds involves the transformation of a subset of fibroblastic cells toward an increased TGF-j3 responsiveness and a transient accumulation of these cells at the wound site. In addition, our study provides evidence that excessive scarring is associated with a failure to eliminate TGF-13 receptor-overexpressing fibroblasts during granulation tissue remodeling, which leads to a persistent autocrine, positive feedback loop that results in overproduction of matrix proteins and subsequent fibrosis. (Am JPatbol 1998, 152:485-493)
The term transforming growth factor (TGF)-f3 refers to a highly homologous family of peptides that are differentially expressed and exert their multifunctional effects on a wide range of target cells. The three mammalian TGF-,B isoforms, designated TGF-f31, TGF-/32, and TGF-f33,
show sequence homologies between 70 and 80% and display similar biological effects in most experimental systems (reviewed in Ref. 1). All three isoforms act on heteromeric receptor complexes that consist of two different transmembrane serine/threonine kinase proteins (designated as type and type 11 receptors). Both receptor types are required for TGF-f signaling, as binding of TGF-,B to type II receptor (RIl) is necessary for the recruitment and activation of type receptor (RI).2 Moreover, the identification of mutations that generate constitutively active type receptors suggest that it is the type receptor that propagates the signal to downstream targets.3'4 Recently, a family of intracellular proteins (the MADR proteins) have been identified that function as downstream components in the TGF-,B receptor signaling pathway. The existence of several type receptors that differentially activate MADR proteins provide a plausible rationale for the multifunctional nature of TGF-,B (reviewed in Refs. 5 and 6). TGF-,Bs integrate the action of several cell types involved in the tissue repair process and are thus important endogenous regulators of wound healing. TGF-,B1 is released from degranulating platelets and chemotactically attracts inflammatory cells7 and fibroblasts8 into the wound site. Autoinduction of the TGF-,B1 isoform has been reported to occur in macrophages9 and fibroblasts,10 a mechanism that extends or amplifies its action in the repair process. TGF-,f has been shown to activate macrophages to produce angiogenic cytokines and stimulate fibroblasts to produce a wide spectrum of matrix proteins, matrix protease inhibitors, and integrin receptors, thereby increasing matrix formation and modulating cell-cell interactions within the wound site (reviewed in Ref. 1 1). In addition, exogenous application of TGF-f3 has been shown to be a potent stimulator of granulation tissue formation in several wound healing models (reviewed in Ref. 11). It has been suggested that a transient upregulation of TGF-p expression is important for normal wound repair,12 whereas a dysregulated and sustained overproduction of this cytokine contributes to tissue fibrosis (reviewed in Ref. 13). Enhanced expression of TGF-,B1 has been observed in tissues of patients with cutaneous
Accepted for publication November 23, 1997. Address reprint requests to Dr. Peter Schmid, Novartis Pharma Inc., K.681.4.43, 4002 Basel, Switzerland. E-mail: peter-1.schmid@ pharma. novartis.com.
485
486
Schmid et al
AJP February 1998, Vol. 152, No. 2
and noncutaneous fibrosis. In the former category, TGF-,3 overexpression has been linked to scleroderma,14 hypertrophic scars,15-17 and keloids,18 and cultured fibroblasts obtained from hypertrophic scars displayed an increased TGF-f3 responsiveness in comparison with cells from unaffected skin.19 However, the question remains whether excessive scarring also involves an aberrant TGF-13 receptor signaling pathway, possibly as a result of autocrine or paracrine stimulation by excess ligand, or whether it is the consequence of a failure to eliminate TGF-,B overresponsive fibroblasts that may be generated during the repair process. To address this question we have investigated, by in situ hybridization and immunohistochemistry, the distribution of RI and of RIl in normal skin, in cutaneous wounds during the normal transition of granulation tissue to scar formation, and in long-persisting post-burn hypertrophic scars.
Materials and Methods Tissue Samples Wound tissues were obtained at various stages during the transition of granulation tissue to scar (1 to 5 weeks) from patients undergoing secondary control resections after tumor excision (n = 32). Hypertrophic scar tissues (n = 10) and normal skin (n = 10) were obtained from patients undergoing surgical correction of severe postburn hypertrophic scars, which had persisted between 6 months and 8 years. In addition, normal skin biopsies were obtained from healthy volunteers (n = 4). All tissues were fixed in 4% paraformaldehyde and embedded in paraffin. Six-micron paraffin sections were placed on Silane-Prep (Sigma, Poole, UK) slides and dried for 1 hour at 560C.
Immunohistochemistry Rabbit polyclonal IgG, which recognizes residues 158 to 179 of the precursor form of R4/ALK5 of human origin (catalog item sc-398, Santa Cruz Biotechnology, Santa Cruz, CA) was used for immunohistochemical staining of RI. For the immunohistochemical detection of RII receptor we used a combination of two polyclonal rabbit IgGs (Santa Cruz BioTechnology) that recognize amino acids 550 to 565 (catalog item sc-220) and 246 to 266 (catalog item sc-400) of the precursor form of human RIl, respectively. Immunostaining of histiocytes was performed with monoclonal mouse anti-CD68 antibodies (Dako, Glostrup, Denmark). To identify vascular smooth muscle cells and myofibroblasts we used monoclonal mouse anti-a-smooth muscle actin (a-SMA) antibodies (Dako). Sections were dewaxed in xylene, rehydrated in decreasing ethanol solutions and water, and incubated for 1 hour in phosphate-buffered saline (PBS), pH 7.2, containing 1% bovine serum albumin. The primary antibodies were diluted (0.5 ,ug/ml) in common antibody diluent (BioGenex, San Ramon, CA) and applied overnight at 4°C in a humidified chamber equilibrated with PBS. Antibody staining was performed with a Super Sensitive-Alk
Phos kit (BioGenex) containing biotinylated anti-rabbit IgG and streptavidin-alkaline phosphatase complexes. Color reaction was performed with the new Fuchsin substrate system (Dako) containing 1 mmol/L levamisole. Sections were counterstained with hematoxylin and mounted with Crystal Mount (Biomeda). To exclude the possibility that the TGF-3 receptorexpressing cells are histiocytes, double staining for CD68 and TGF-3 type receptor was performed by incubating sections with a mixture of monoclonal anti-CD68 antibodies and polyclonal rabbit anti-type receptor antibodies. To distinguish endothelial cells from pericytes and smooth muscle cells, double staining was performed with monoclonal antibodies against a-SMA and polyclonal rabbit antibodies against RI. CD68 and a-SMA were visualized with horseradish-peroxidase-conjugated antimouse antibodies (Dako) and diaminobenzidine chromogen (brown color deposit) before staining of RI with the Super Sensitive-Alk Phos kit and New Fuchsin chromogen (red color deposit).
In Situ Hybridization In situ hybridization using paraffin sections and 35S-labeled riboprobes was performed according to a previously described method.20 The RI riboprobe template was a cDNA fragment (320 bp) corresponding to the extracellular domain of the human ALK-5 cDNA sequence and cloned into pBluescript (Stratagene, La Jolla, CA). The RII riboprobe template was a 600-bp cDNA fragment corresponding to +260 to +860 bp of the human TGF-p type 11 receptor sequence, cloned into pBluescript (Stratagene). Sense and antisense RNA probes were labeled with [a-355]UTP (>400 Ci/mmol; SJ 263, Amersham, Little Chalfont, UK) using an RNA transcription kit (Boehringer Mannheim, Mannheim, Germany) according to the manufacturer's instructions. Signal detection was performed using emulsion autoradiography (Ilford K5) and bright-field/dark-field
photomicroscopy.
Results RI and Rll Immunostaining in Normal Skin In normal human skin, immunostaining of RI was clearly detectable in the epidermis, hair follicle epithelia, and vascular cells (Figure la). However, only a small subset of stromal cells within the papillary and reticular dermis revealed detectable immunoreactivity with the anti-RI antibodies (Figure la). Anti-RII antibodies revealed a staining pattern that was very similar to that of anti-RI antibodies (Figure lb). Double-staining experiments using monoclonal antiCD68 antibodies and polyclonal anti-RI antibodies provided evidence that the macrophage marker and RI are differentially expressed in the dermis (Figure lc). Some cells with fibroblastic morphology revealed staining with the anti-RI antibodies (indicated by red color deposits) but not with anti-CD68 antibodies (indicated by brown
TGF-B Receptors in Wound Healing and Scarring
487
AJP February 1998, Vol. 152, No. 2
or~
e
:.
~
.*.6.
b f-
:~
i. _q
`
0f .'
I
r x 4w
_0
* ..
:, tF; '''-!
4
p0
.!:
A.a
9,
.,
.f
I. b
v
Figure 1. a and b: Immunohistochemical staining of RI and RII in normal human skin. Immunostaining (red color deposit) of RI (a) and RII (b) is clearly visible in epidermal cells and in vascular cells, although most fibroblastic cells are devoid of specific staining. c: Double staining of CD68 (brown color deposit) and RI (red color deposit) in the papillary dermis. CD68 and RI show differential staining patterns. The red-brown color deposit in a dermal cell (arrow) indicates that histiocytes co-express CD68 and RI. In addition, RI+/CD68- and double-negative stromal cells are visible. Vascular cells show RI staining but no CD68 immunoreactivity. d: Double staining of a-SMA (brown color deposit) and RI (red color deposit) in an arterial blood vessel. a-SMA and RI show differential staining pattems. a-SMA staining (brown color) is visible in cells located in the vascular adventitia, whereas endothelial cells lining the lumen of the vessel demonstrate RI (red color) expression. Red color deposits are also visible in vascular structures positive for a-SMA staining, indicating that vascular smooth muscle cells express TGF-,3 receptors as well.
488
Schmid et al
AJP February 1998, Vol. 152, No. 2
Figure 2. Immunohistochemical staining of RI (a and C) and 1111 (b and d) in normal wound granulation tissue. a and b: Granulation tissue 2 weeks after injury. cadd: Early scar of a normal healing wound 5 weeks after injury. Granulation tissue fibroblasts display a high density and strong immunostaining of RI (a) and RhI (b). In early scar tissue, RI-positive (c) and RII-positive (d) fibroblastic cells are still present but show a reduced cellularity when compared with fully developed granulation tissue.
TGF-13 Receptors in Wound Healing and Scarring
489
AJP February 1998, Vol. 152, No. 2
.
>S4
2{h k'*
Uf
.Aw.
Figure 3. Immunohistochemical staining of RI (a and d), RII (b and e), and ca-SMA (c and f) in post-bum hypertrophic scars. a to c: At 15 months after injury. d to f: At 8 years after injury. Strong immunostaining of RI (a) and RII (b) is visible in fibroblastic cells that display a very high density and are arranged in nodular structures. c: Myofibroblasts are also visible structures, although not all fibroblasts within these nodules display the myofibroblast phenotype. In the 8-year-old post-burn hypertrophic scar, RI (d) and RII (e) immunostaining is not detectable or only barely visible in cells with fibroblastic morphology, although blood vessels demonstrate weak but clearly detectable levels of RI and RII immunoreactivity. f: In the 8-year persisting hypertrophic scar, myofibroblasts are not visible, although vascular smooth muscle cells are strongly labeled.
color deposits). A small number of cells showed a redbrown staining, indicating that histiocytes co-express CD68 and RI. However, the majority of dermal fibroblasts revealed staining neither of CD68 nor of RI. Double-staining experiments using monoclonal anti-aSMA antibodies and polyclonal anti-RI rabbit antibodies also showed that the smooth muscle cell marker (indicated by brown color deposits) and RI (indicated by red color deposits) display differential staining patterns in arterial blood vessels (Figure ld). Cells located in the vascular adventitia showed a-SMA staining (brown color), whereas cells lining the lumen of the vessels exhibited RI immunoreactivity (red color), indicating that endothelial cells express TGF-f3 receptors. However, RI immunostaining was not confined to endothelial cells, as red color deposits were also observed within a-SMA-
positive (brown) structures (Figure ld), suggesting that vascular smooth muscle cells express TGF-j3 receptors as well.
RI and Rll Immunostaining in Granulation Tissues and Early Scars In wound granulation tissue, the majority of cells showing a fibroblastic morphology revealed strong staining of RI and RII (Figure 2 a and b). These cells were not uniformly distributed but occurred as clusters in which the highest cellularity was generally observed in the deepest
regions of the granulation tissue. Accumulations of Rland RII-expressing fibroblasts were present in all granulation tissues analyzed, although their cellularity markedly
490 Schmid et al AJP February 1998, Vol. 152, No. 2 Table 1. Details of Hypertrophic Scar Specirnens Analyzed
Age of scar 6 months 6 months 7 months 8 months 12 months 12 months 13 months 15 months 20 months 8 years
Sex Female Male Male Male Female Male Male Male Male Female
Patient's age
Position of biopsy
RI and Rll staining of fibroblasts
4 years Back of left hand 25 years Anterior neck 17 years Right hand 34 years Anterior chest 6 years Anterior neck 25 years Right axillaries 3 years Left upper arm 23 years Right leg 27 years Right leg 11 years Face All biopsies showed strong immunostaining of RI and RII in the epidermis, which served as an internal
++ ++ ++ ++ ++ ++ ++ ++ ++
a-SMA staining of fibroblasts + +/-
+/+
+/+ + + +
control. Vascular staining was also observed in all specimens. All biopsies showed strong a-smooth muscle actin (a-SMA) immunostaining in smooth muscles and on the walls of arterial blood vessels, which served as an internal control. + +, ubiquitous in nodular structures; +, diffuse; +/-, focal; -, not detectable. *Nodular structures were not present, and only few cells with fibroblastic morphology showed weak staining, although vascular staining was detectable.
decreased at stages of the repair process later than 3 weeks after injury (Figure 2, c and d). Blood vessels within granulation tissue also exhibited immunostaining of RI and Rll antigens, although the staining intensities among vascular cells displayed strong heterogeneity.
Immunostaining of RI, Rll, and a-SMA in Hypertrophic Scars All hypertrophic scars persisting between 6 and 20 months after thermal injury demonstrated the dense nodular organization and fine, random organization of collagen fibers, characteristic of hypertrophic scarring.21'22 Most fibroblasts localized within these nodular structures revealed abundant immunostaining of RI and RII (Figure 3, a and b). However, Rl- and RIl-expressing fibroblasts were very rare in a hypertrophic scar persisting 8 years after injury. This lesion showed thick collagen bundles, although the fibroblastic cellularity was low and nodular structures were no longer visible (Figure 3, d-f). All hypertrophic scars persisting between 6 and 20 months exhibited a-SMA staining of myofibroblasts located within nodular structures (Figure 3c). The abundance and distribution of myofibroblasts within individual lesions exhibited a high degree of variability. In most hypertrophic scars, myofibroblasts were distributed throughout the fibrotic tissue, whereas some hypertrophic scars exhibited a more focal distribution of cells that displayed the myofibroblast phenotype (see Table 1). A comparison of the immunostaining patterns of RI, RII, and a-SMA in serial sections of hypertrophic scars clearly showed that all myofibroblasts expressed abundant TGF-3 receptors. However, not all TGF-f3 receptor-overexpressing fibroblasts were a-SMA positive (Figure 3, a-c). Myofibroblasts were not detected in the 8-year persisting hypertrophic scar, although in this lesion vascular smooth muscle cells displayed a strong immunoreactivity with antibodies against SMA (Figure 3f), which served as a positive internal control. Blood vessels showed immunoreactivity with anti-RI and anti-RII antibodies, although the staining intensity of vascular cells was generally weaker than in cells with
fibroblastic morphology (Figure 3, a and b). As in wound granulation tissue, the staining intensities among individual blood vessels displayed strong heterogeneity (not shown).
RI and Rll mRNA Expression in Normal Skin, Granulation Tissues, and Hypertrophic Scars In normal human skin, RI and RII mRNA expression was clearly visible in the epidermis and in epidermal appendages, although mRNA expression of both receptor types was only barely detectable in the normal dermis (Figures 4, a/b and elf). The weak hybridization signals within the normal dermis were not uniformly distributed but appeared to accumulate in areas containing vascular structures (Figure 4, a/b and elf). However, due to strewing of silver grains and nonspecific background signals, a distinct cellular localization was not feasible. A slight accumulation of silver grains was also observed in the proximity of some cells with fibroblastic morphology, indicating that a minority of stromal fibroblasts express TGF-p receptor mRNA. However, a clear differentiation of positive and negative stromal cells was not possible due to nonspecific background signals. In contrast, hypertrophic scar fibroblasts, when compared with stromal cells in normal skin, revealed significantly elevated levels of RI and RII mRNA expression signals (Figure 4, c/d and g/h).
Discussion There is an increasing body of evidence that suggests that fibroblasts are a morphologically and functionally heterogeneous cell population, not only across different histological sites but also within a single tissue layer (reviewed in Ref. 23). Individual fibroblasts may therefore display distinct biological functions in dermal tissue repair and homeostasis. This hypothesis is supported by our observation that in the nonwounded dermis only a small subset of fibroblastic cells show detectable TGF-P receptor immunoreactivity, which may well account for
..
'_*.gb>b:*''s*",bf..;'S br
491 AJP February 1998, Vol. 152, No. 2
w:
Y,
+ *e ,
tt _
TGF-,B Receptors in Wound Healing and Scarring
'r
i
w
V
,
4W
_||f q_f
|'!,
e
§1s,,Fw' *
..
;sE
¢' $
e
C
t
:4
}
'
w
f
.1w-
I
A;. ewe It .
I
,1e t
C.,
2.
.> 1
W
_,
Ig...::,: s, w.
-.4.
4 *,:
,Is!
"
_
.,
_
it.,
#,#
r'
' ,.'4
1".
'4
., -
, .V4
ft ' 'a
# ... ,,,,,..s(F,s,,;%. .........t
ra
r
.Xb. w
*
i
:.
# M_
C
4
!:,8:
I
* ' '@e,
7
i ..
ir
t
/
t
b'
^ I. A.
0.
tf t
^
