Abstract
purpose. To compare the effects of the three human isoforms of transforming
growth factor (TGF)-β in vivo using a mouse model of conjunctival
scarring, both in normal eyes and after treatment with MMC, with a view
to delineating the role of this growth factor in glaucoma filtration
surgery.
methods. Application of recombinant human TGF-β was assessed in a prospective,
randomized study of mouse conjunctival scarring, in which
subconjunctival TGF-β1, -β2, and -β3 (all 10−9 M)
were compared with control (phosphate-buffered saline [PBS] carrier)
and mitomycin C (MMC; 0.4 mg/ml) treatment at 6 hours, and 1, 3, and 7
days after surgery (six eyes/treatment/time point). Effects of TGF-β2
on eyes previously treated with MMC were also assessed. Histologic
studies of enucleated eyes were performed to analyze development of the
scarring response, extracellular matrix deposition, and the
inflammatory cell profile.
results. All three isoforms of TGF-β behaved in a similar manner in vivo,
being associated with a rapid-onset and exaggerated scarring response
compared with control and MMC treatment. TGF-β–treated eyes showed
evidence of an earlier peak in inflammatory cell activity
(P < 0.05) and increased collagen type III
deposition (P < 0.05). TGF-β2 treatment
significantly stimulated scarring after MMC application
(P < 0.05).
conclusions. TGF-β1, -β2, and -β3 appear to have similar actions in vivo and
stimulate the conjunctival scarring response. Application of TGF-β2
modified the effects of MMC. All TGF-β isoforms may be potent
modulators of the conjunctival scarring response. These studies
indicate that TGF-β2 may naturally modify the antiscarring effects of
antimetabolites such as MMC in glaucoma filtration
surgery.
The conjunctival scarring response is a major determinant of
morbidity and visual prognosis in a wide spectrum of ocular diseases
including cicatricial conditions (e.g., trachoma and pemphigoid) and
those in which the results of treatment depend on the healing response
after surgery, such as glaucoma. The commonest cause of failure of
glaucoma filtration surgery is the occurrence of subconjunctival
scarring at the bleb and sclerostomy sites,
1 2 with
increased scar deposition being associated with poor control of
postoperative intraocular pressure.
Although the role of growth factors in conjunctival scarring is
believed to be important, that of transforming growth factor-β
(TGF-β), the most potent growth factor involved in wound healing
throughout the body,
3 4 5 6 is unclear. There are three
isoforms of TGF-β found in humans. Of these, TGF-β1 and -β2 are
known to stimulate greatly the dermal scarring
response.
4 5 6 7 8 9 10 The actions of the third isoform, TGF-β3,
in wound healing are less well established, with some studies
suggesting that it may actually inhibit in vivo
scarring.
4 11
Although the effects of exogenous TGF-β have been studied in skin,
the actions of all three isoforms after administration in the eye have
not been examined to date. However, it is known that compared with
TGF-β1 and -β3, TGF-β2 is the predominantly expressed ocular
isoform, having been identified in normal and diseased
eyes
12 13 and implicated in the pathogenesis of several
ocular scarring diseases such as proliferative vitreoretinopathy and
cataract formation.
14 15 More recent work, however, has
suggested TGF-β1 and -β3 may be important in the fibrosis occurring
in the cicatrizing disease, ocular pemphigoid.
16
The conjunctival scarring response in glaucoma filtration surgery is
thought to be greatly influenced by the passage of aqueous through the
surgical site, and in particular, growth factors in the aqueous such as
TGF-β.
17 However, compared with the other growth factors
found in aqueous humor, TGF-β2 has been shown to be the most
potent
18 and is significantly raised in glaucomatous
eyes.
19 In the light of these findings, our study
specifically investigates the effects of TGF-β on the conjunctival
scarring response.
Using our recently established model of conjunctival scarring in the
mouse eye,
20 we have compared the effects of all three
TGF-β isoforms. The model is unique because unlike fistulizing
glaucoma surgery, which causes a complex, dynamically changing wound
environment involving the sclera, the conjunctiva, the aqueous humor,
its various cytokines and chemical mediators, it consists of a
wound-healing response that is specifically localized to the
subconjunctival space. We have investigated the conjunctival scarring
response induced by TGF-β in the mouse eye, using histologic
techniques that identify collagen, elastic-related fibers, and
inflammatory cells and have compared this to control and the commonly
used modulating agent mitomycin C (MMC). In addition, we have assessed
the effects of TGF-β2 on MMC-treated eyes.
The exogenous TGF-βs used in these studies were all recombinant
human proteins. TGF-β2 and -β3 were generous gifts of Ciba Geigy,
Switzerland, and TGF-β1 was a gift from Professor Gregory Schultz,
University of Florida, Gainesville. All TGF-β injections were
reconstituted from stock solutions of the active form (1 mg/ml) in
phosphate buffered saline (PBS) containing 1% bovine serum albumin
(Sigma) at a concentration of 10−9 M. PBS was
used as the carrier and the control. Mitomycin-C (MMC, Kyowa, Essex,
United Kingdom) was used at the clinically used concentration of 0.4
mg/ml.
All experiments were performed on 6-week-old BALB/c mice,
and conformed to the ARVO Statement for the Use of Animals in
Ophthalmic and Vision Research. Mice eyes were randomly allocated to
one of seven treatments (six eyes/treatment/time point):
subconjunctival injections of a single of 25 μl of TGF-β1, -β2,
or -β3 (10−9 M), or control (PBS carrier); MMC
(0.4 mg/ml); or dual therapy consisting of subconjunctival injections
of MMC (0.4 mg/ml) followed 24 hours later by either TGF-β2
(10−9 M) (MMC+TGF-β2), or PBS control
(MMC+control).
Surgery was performed using the method previously
described.
20 Briefly, all animals underwent general
anesthesia with intraperitoneal injections (0.2 ml Hypnorm; Janssen
Pharmaceutical, Oxford, UK; and Hypnovel; Roche, Welwyn Garden City;
UK) mixed 1:1:6 with sterile distilled water. A subconjunctival
injection of 25 μl of each test substance with a 27-gauge needle was
given at the same location in each animal, under visualization with a
microscope (OPMI MDU; Carl Zeiss, Oberkochen, Germany). This position
was marked with a corneal 10-0 nylon suture at postmortem but before
enucleation. Mice were assessed clinically and killed by cervical
dislocation at 6 hours and 1, 3, and 7 days after surgery.
All enucleated eyes were prepared for histologic analysis after
fixation for 24 hours with 10% buffered formaldehyde and embedded
whole in paraffin wax. Development of scar tissue was studied in
sequential 5-μm thick sections using the following special stains:
hematoxylin and eosin to assess cellularity, van Gieson and picrocirius
red (viewed under polarized light) to demonstrate collagen deposition,
and aldehyde fuchsin for elastic and elaunin fibers together with
oxidation-aldehyde fuchsin for additionally demonstrating oxytalan
fibers. In addition, immunofluorescence was used to demonstrate
collagen types I and III. Primary antibodies were goat anti-mouse
collagen I (1:25 dilution, Europath, Southern Biotechnology) and chick
anti-mouse collagen III antibodies (1:25 dilution, Europath, Southern
Biotechnology) with a secondary fluorescein isothiocyanate
(FITC)-labeled chick anti-goat and rabbit anti-chick IgG antibodies
(1:100 dilution, Serotec). Photographs were taken at an
excitation–emission wavelength of 490 and 525nm.
Lymphocyte and macrophage identification was performed using the
lymphocyte CD3 rat anti-human antibody (1:50 dilution, Serotec) and
macrophage F4/80 rat anti-mouse antibody (1:20 dilution, Serotec) as
primary antibodies incubated for 2 hours, followed by a further 2-hour
incubation with a FITC-labeled goat anti-rat IgG secondary antibody
(1:100 dilution, Serotec). Fibroblasts were identified using
phalloidin, which stains the fibroblast cytoskeleton.
21 Briefly, sections were soaked in Tris-buffered saline (TBS; 25 mM Tris,
140 mM NaC1, 3 mM KCl [pH 7.4]) before a 2-hour incubation in
FITC-phalloidin (Sigma) at 2.5 μg/ml in TBS. Sections were then
rinsed and mounted in PBS and viewed by fluorescence microscopy.
Sequential sections were assessed for cellularity profile and
extracellular matrix deposition by three independent and masked
observers (MFC, RA, JG), using the grading system modified from Shah et
al.
22 (scale: 0–4; where 0 is normal conjunctiva; 1 is
1–25% of normal; 2 is 2–50% of normal; 3 is 51–75% of normal; and
4 is >75%; no prefix, more than; − prefix, less than). Parameters
assessed included total cellularity; collagen types I and III and
elastic fiber deposition; and fibroblast, macrophage, and lymphocyte
profiles. For each treatment group a mean grade per parameter at each
time point (with 95% confidence intervals [CI]) was calculated.
Analysis was performed using computer software (SPSS for Windows; SPSS;
Chicago, IL) at individual time points using a one-way analysis of
variance. All treatments were compared with control (PBS). The observed
significance levels from multiple comparisons were adjusted using the
Bonferroni test, with P < 0.05 indicating
significance.
Mouse eyes showed evidence of a visible subconjunctival bleb after
all subconjunctival injection, which was macroscopically apparent until
24 hours
(Fig. 1) . Little difference between macroscopic appearances of injected eyes
was noted thereafter, and no clinical evidence of periocular
inflammation was seen at any time after surgery.
Histologic evaluation showed a characteristic response to
subconjunctival administration of TGF-β
(Fig. 2) . Semiquantitative analysis revealed all TGF-β isoforms produced a
similar scarring response. There was no significant difference in any
of the parameters studied between TGF-β isoforms at any time point
including total cellularity (
Fig. 3 A).
Compared with control, TGF-β treatment was associated with an
earlier peak in the appearance of macrophages and lymphocytes (
Figs. 3B 3C ;
P < 0.05). Maximal numbers of macrophages and
lymphocytes for TGF-β treatment occurred at 6 hours, compared with
control in which maximal counts occurred at day 1. No significant
difference between treatment groups was seen in fibroblast activity,
although TGF-β was associated with increased fibroblast activity with
time.
TGF-β2 significantly stimulated collagen III deposition compared with
control at day 7 (
Fig. 3E ;
P < 0.05). Although both
TGF-β1 and -β3 also showed increased collagen III deposition
compared with control, this did not reach statistical significance. No
statistically significant difference in collagen I, picrocirius and
aldehyde fuchsin staining was associated with TGF-β treatment,
although there was a trend toward more picrocirius staining with
TGF-β compared with control after day 1.
As expected, there was a significant reduction in total
cellularity associated with MMC treatment compared with control and all
TGF-β-treated eyes (days 3 and 7; P < 0.05). In
comparison with control, MMC treatment significantly inhibited
macrophage and lymphocyte levels on day 1 and fibroblast activity on
days 1, 3, and 7 (P < 0.05). Macrophage activity was
significantly reduced by MMC treatment compared with TGF-β1, -β2,
and -β3 at 6 hours (P < 0.05), and there was a
significant difference in lymphocyte and fibroblast levels in MMC and
TGF-β2-treated eyes at 6 hours and 7 days, respectively
(P < 0.05).
A reduction in extracellular matrix components was also associated with
MMC treatment. Compared with control, this was significant on days 3
and 7 for collagen III (P < 0.05), day 7 for collagen
I (P < 0.05), day 3 for elastic fiber, and day 7 for
picrocirius red staining (P < 0.05). Differences among
MMC and TGF-β1, -β2, and -β3 treatments were present with a
reduction in collagen III on days 3 and 7 (P < 0.05),
elastic fibers at 6 hours and on day 3 (P < 0.05), and
picrocirius red staining on day 7.
Figure 4 shows histologic sections characteristics of the conjunctival scarring
response after treatment with MMC.
Figures 4C and 4F demonstrate the
modulation of the antiscarring effects associated with MMC by a
subsequent injection of TGF-β2 (10
−9 M), 7
days after treatment. Also of note is the appearance of abnormal
conjunctival epithelial cells in all MMC-treated eyes, as shown in
Figures 4A 4B and 4C .
Comparisons of histologic effects of MMC followed by PBS (MMC+control)
and MMC followed by TGF-β2 (MMC+TGF-β2) treatment are shown in
Figure 5 . These results should also be compared with single applications of
TGF-β2 injections, displayed in
Figure 3 .
TGF-β2 treatment (MMC+TGF-β2) was associated with significantly
more cellular activity on day 3 (
Fig. 5A ;
P < 0.05),
compared with either MMC+control or MMC treatment alone. TGF-β2
significantly stimulated levels of macrophages on days 1 and 7 (
Fig. 5B ;
P < 0.05), and lymphocytes and fibroblasts on days
3 and 7 (
Fig. 5C 5D ;
P < 0.05). Although MMC+TGF-β2
was associated with greatest cellular activity, both groups showed peak
macrophage, lymphocyte, and fibroblast activity at 6 hours.
TGF-β2 treatment was associated with stimulation of extracellular
matrix components. Compared with MMC+control, this was significant on
days 3 and 7 for collagen III and collagen I (
Figs. 5E 5F ;
P < 0.05). Differences in picrocirius red staining was
also present on days 1, 3, and 7(
Fig. 5D ;
P < 0.05).
Although not significant, there was a trend for TGF-β2 treatment to
be associated with greater levels of all studied extracellular matrix
components than in control
(Fig. 5D) .
Our studies have shown that all three human TGF-β isoforms, when
applied exogenously and at the same concentration, produced a similar
conjunctival scarring response. This response was characterized by an
earlier and more pronounced peak in inflammatory cell activity with
evidence of enhanced fibroblast activity in TGF-β treatment groups
compared with control. In addition, TGF-β2 was associated with
increased collagen III deposition, although no significant difference
between TGF-β and control groups was demonstrated in extracellular
matrix architecture.
Little is known about the effect of exogenous TGF-β in conjunctival
scarring. However, elsewhere in the eye, TGF-β has been advocated as
a biologic chorioretinal “glue” for use in repairing retinal
tears
23 and macular holes.
24 Glaser et al.
24 suggested that visual acuity after TGF-β2 treatment
in macular hole surgery significantly improved in a dose-related manner
(range, 0.28–5.32 × 10
−7 M TGF-β2).
Early work by Roberts et al.
10 showed that a
subcutaneous injection of TGF-β1 and -β2 (0–16 ×
10
−7 M) in newborn mice, stimulated the formation of
granulation tissue, associated with induction of angiogenesis,
increased fibroblast number, and collagen deposition. TGF-β
administration into the peritoneum has also been shown to induce
fibrosis.
25 Williams et al.
25 demonstrated
that application of 16 × 10
−8 M TGF-β1 after
surgical injury to the uterine horns in rats, significantly increased
the number of adhesions formed after surgery, with evidence of an
increased number of inflammatory cells and fibroblasts,
histologically.
25
Application of TGF-β2 (4 and 40 × 10
−8 M) on
healing fractures in the rabbit, demonstrated that at the higher dose,
TGF-β2 promoted callus formation in stable conditions.
26 However, in unstable conditions, TGF-β2 was found to be opposite in
that they retarded and reduced bone and cartilage formation in the
callus. TGF-β2 was not found to accelerate fracture healing. This is
suggestive that the actions of TGF-β2 are determined by its
extracellular environment.
The effects of exogenous TGF-β application have also been studied in
several models of dermal wound healing. Shah et al.
22 have
demonstrated that exogenous application of TGF-β1 to a linear
incisional wound in rat skin affects the response in a dose-dependent
manner (range, 0.8–20 × 10
−9 M). At 4 and 20 × 10
−9 M concentrations of TGF-β1, wounds showed
increased vascularity and extracellular matrix deposition compared with
controls, with evidence of scarring. However, no significant difference
in cellularity was noted between TGF-β1 and control groups.
In one of the few studies comparing the in vivo effects of all three
TGF-β isoforms, Shah et al.
4 applied exogenous TGF-β1,
-β2, and -β3 to the same incisional rat model of dermal scarring
described above. All isoforms were assessed at different concentrations
of 0.04, 0.4, 4, and 20 × 10
−9 M. Differences
between the isoforms were noted at 4 and 20 × 10
−9 M
in increased fibronectin in association with TGF-β1 and -β2 only,
increased collagen fibril organization with TGF-β3 only, decreased
monocyte and macrophage profile of wounds treated with TGF-β3
(compared with TGF-β1 and -β2 which were similar to control), and
increased collagen I and III in TGF-β1 and -β2 treatments only. All
isoforms were associated with increased vascularity and angiogenesis.
The authors suggest that TGF-β3 inhibited scarring and promoted
better collagen organization, compared with TGF-β1 and -β2, which
stimulated dermal scarring.
However, Cox
11 has shown that TGF-β3 application, both
in thermal wounds in mice and incisional and second-intention wounds in
rats, stimulates the cutaneous scarring response. This stimulation by
TGF-β3, similar to that associated with TGF-β1 and -β2, resulted
in accelerated re-epithelialization, increased fibroblast and phagocyte
activity, and increased protein, DNA, and collagen production.
Our results, similar to those of Cox,
11 suggest that in
conjunctival scarring all three TGF-β isoforms have similar
stimulatory effects. This is supported by our data from in vitro work
in our laboratory investigating effects of TGF-β on conjunctival
fibroblasts.
27 The TGF-β used in this study was 100%
active. This is different from cellular TGF-β, which, when produced,
is secreted in a latent form that has to be activated into a mature,
active form. It is known that only between 22% and 61% of TGF-β
found in aqueous is in its active rather than its latent form
17 28 and the average concentration of active TGF-β2 in normal
aqueous is between 0.73 and 10.98 × 10
−11 M compared
with 10
−9 M, a much higher concentration, used in our
study.
Using the mouse model of conjunctival scarring, we have shown that
TGF-β2 applied to mice conjunctiva after MMC treatment modulated and
significantly reversed the antiscarring effects of MMC. Specifically,
this was demonstrated by TGF-β2 treatment stimulating cellular
activity including macrophages, lymphocytes, and fibroblasts compared
with either control or MMC treatment alone. In addition, TGF-β2
treatment was associated with increased collagen type I and III
deposition.
The use of TGF-β2 as a modulating agent to counteract effects of MMC
treatment, has been shown by other investigators.
29 30 Doyle et al.
30 demonstrated that a single peribleb
injection of 4 × 10
−7 M TGF-β2 effectively treated
50% of bleb leaks deliberately induced in a rabbit model of
MMC-assisted filtration surgery. Histologic examination revealed
increased peribleb cellularity and denser collagen deposition in
TGF-β eyes compared with control eyes.
The present mouse study also shows the possible use of TGF-β2 as a
modulator of the potent antiscarring effects of conjunctival MMC. This
is important because although MMC has been shown to be highly effective
as an adjuvant treatment for preventing postoperative scarring in
glaucoma filtration surgery, its use is associated with several
complications. These include the production of thin, avascular cystic
blebs with the attendant risks of persistent hypotony, bleb leaks, and
endophthalmitis. TGF-β2 may offer another therapeutic strategy for
enhancing healing after MMC use, although the method of its
instillation would be very important. Our studies on MMC bleb
sizes
31 suggest that a peribleb injection (as administered
by Doyle et al.
30 ) would not be suitable, because leaking
blebs are often thin-walled cystic blebs, with an increase in
cellularity around the bleb itself. A peribleb injection of TGF-β2
would further increase the degree of peribleb cellular activity
contributing to an increase in surrounding scar tissue fibrosis and
contraction, which in turn would produce even smaller blebs with
thinner walls. It would seem more appropriate that TGF-β2 be applied
either transconjunctivally or with an intrableb injection. Obviously,
these methods need further investigation.
In summary, our results show that all three TGF-β human
isoforms have similar actions in subconjunctival scarring in vivo. They
each stimulate the scarring response in an exaggerated and rapid manner
compared with control and MMC. Application of TGF-β2 modulated the
antiscarring effects of MMC. This may be an important clinical finding
and suggests that TGF-β is a potent modulator of the conjunctival
scarring response and may naturally modify the antiscarring effects of
antimetabolites such as MMC in glaucoma filtration surgery.