IL-10 and Antibodies to TGF-β₂ and PDGF Inhibit RPE-Mediated Retinal Contraction

Abstract

purpose. Retinal pigment epithelial (RPE) cells are believed to play a pivotal role in the formation and contraction of epiretinal membranes in proliferative vitreoretinopathy (PVR). In the present study, an organ culture method was used that mimics the contractile stage of PVR, to investigate the contribution of a variety of growth factors in human RPE cell–mediated contraction of the retina.

methods. Cultured human RPE cells were seeded onto bovine retinal explants. After attachment, cultures received one of the following exogenous growth factors: platelet-derived growth factor (PDGF)-AB, PDGF-BB, basic fibroblast growth factor (bFGF), transforming growth factor (TGF)-β₁, TGF-β₂, or interleukin (IL)-10; or a neutralizing antibody to PDGF and/or TGF-β₂. Control explants were either untreated or received a null antibody. Contraction was assessed by image analysis and expressed as percentage reduction in retinal area.

results. RPE cells produced a more than 50% contraction of the retina after 7 days in untreated samples. PDGF and TGF-β₂ stimulated RPE-mediated contraction by a further 20% at 100 ng/ml. IL-10 decreased contraction by 63%, whereas the other growth factors gave rise to similar contraction to untreated controls. Neutralizing antibodies against PDGF and TGF-β₂ reduced RPE-mediated contraction by up to 70% in comparison with untreated controls. The neutralizing antibodies also inhibited the effects of exogenous PDGF and TGF-β₂ on RPE-mediated contraction of the retina (P < 0.01).

conclusions. These findings confirm a role for both PDGF and TGF-β₂ in RPE cell–mediated contraction of the retina. Such contraction can be inhibited by neutralizing antibodies against PDGF and TGF-β₂, which, together with IL-10, are putative candidates for therapeutic intervention in PVR.

Proliferative vitreoretinopathy (PVR) is a common sequel to rhegmatogenous retinal detachment and is the leading cause of failure of surgery to correct it. Characterized by the formation of contractile membranes in the vitreal cavity and on both surfaces of the retina, ¹ the disease has been compared with an inappropriate wound-healing response. This comparison, coupled with immunohistochemical analysis of excised membranes, has focused interest on the contribution of specific growth factors/cytokines to the progression of the disease. Retinal pigment epithelial (RPE) cells are thought to play a pivotal role in both the formation and the contraction of these membranes and are known to secrete, and to be modulated by, a wide range of growth factors, including transforming growth factor (TGF)-β, platelet-derived growth factor (PDGF), and fibroblast growth factor (FGF). ²  

TGF-β isoforms 1, 2, and 3 have been identified in the posterior segment of the eye by both molecular and immunohistochemical techniques. ^{3 4 5} RPE cells, fibroblasts, platelets, and macrophages (all epiretinal membrane components) are known to secrete TGF-β and are considered to be the major source of exogenous TGF-β in the eye. ^{4 6} Although not definitively proven in man, TGF-β₂ is considered to be the predominant TGF-β isoform in the posterior segment; studies in primates have shown that the β₂:β₁ ratios are 6:1 for the neural retina and 425:1 for vitreous. ⁵ TGF-β is hypothesized to play a major role in PVR, because levels in the vitreous of patients with PVR are increased fourfold over vitreous extracted from patients without such disease, ⁷ and TGF-β has been localized to excised epiretinal membranes (Mike Boulton, unpublished data, 1997). Furthermore, when RPE cells in culture are exposed to TGF-β, they are stimulated to increase fibronectin synthesis and secretion ⁸ and collagen gel contraction, ^{9 10 11} both features associated with the pathobiology of PVR. ² TGF-β can be modulated by a variety of growth factors-cytokines of which interleukin (IL)-10 is a potent antagonist of TGF-β production. ^{12 13}  

The presence of PDGF in membranes excised from patients with PVR has been confirmed immunohistochemically ¹⁴ ;PDGF concentration is also elevated in the vitreous of patients with the disease. ¹⁴ Fibroblasts, macrophages, platelets, and RPE cells have all been shown to secrete PDGF isoforms. ^{6 15 16} In response to PDGF, RPE cells proliferate and upregulate secretion of PDGF in an autocrine feedback loop. ¹⁶ PDGF has been shown to be chemotactic for both fibroblasts and RPE cells ^{6 17 18} and enhances the contraction of RPE cells and fibroblasts in collagen gels. ^{9 19}  

Basic FGF (bFGF) has been localized to PVR membranes ²⁰ and is known to be secreted by a number of cell types, including RPE cells, involved in the pathologic course of PVR. ²¹  

The purpose of this study was to investigate the contribution of a variety of growth factors to human RPE cell–mediated contraction of the retina, using a novel organ culture model in which the formation of contractile cellular epiretinal membranes represents the early pathobiology of PVR ²² and to assess the efficacy of a neutralizing antibody against those growth factors that affect contraction. 

Materials and Methods

Materials

Growth factors were purchased as follows: PDGF-AB, PDGF-BB from Sigma (Poole, UK); bFGF, TGF-β₁, TGF-β₂ from R&D (Oxon, UK); IL-10 from Genzyme (Kent, UK). TGF-β was prepared in phosphate-buffered saline (PBS) supplemented with 0.1% bovine serum albumin and 4 mM HCl, and the other growth factors were prepared in PBS according to the manufacturer’s instructions. A neutralizing polyclonal antibody to PDGF was obtained from R&D. Three to 5 μg/ml of this antibody neutralizes 50% of the bioactivity due to 10 ng/ml of human PDGF; the antibody is effective against all isoforms of PDGF. A neutralizing monoclonal antibody to active human TGF-β₂ (IgG4 6B1) and a null isotype matching monoclonal antibody (2G6) with no cross reactivity with TGF-β were kindly donated by Cambridge Antibody Technology, Melbourn, UK. The anti TGF-β₂ antibody had a 50% inhibitory concentration (IC₅₀) of 1.2 nM when assayed in the TF1 proliferation assay. 

RPE Cell–Mediated Retinal Contraction

Retinal contraction studies were undertaken as previously described. ²² In brief, retinas isolated from bovine eyes obtained from a local abattoir within 4 hours of death were mounted on cellulose ester membranes (Supor; Gelman Sciences, Northampton, UK). Explants approximately 5 × 5 mm² were cut from the equatorial retina, producing a preparation with an identical area of retina and support and allowing nontraumatic preparation of the tissue. The explants were transferred to organ culture dishes where they rested on a stainless steel mesh (16 explants per dish). Medium (Trowels T8; Gibco, Paisley, UK) supplemented with 10% fetal calf serum, 20 mM HEPES buffer, 200 mg/l streptomycin, 200 mg/l kanamycin, 120 mg/l benzyl penicillin, and 70 mg/l glutamine was added to the level of the mesh—that is, in contact with the support membrane beneath the explant. The dishes were then incubated at 37°C in a humidified atmosphere of air and 5% CO₂. 

Retinal explants and their underlying membrane supports were cultured for 2 days before transfer to a 96-well dish. Human RPE cells (passages 5–9), isolated and cultured as previously described, ²³ were seeded into each attachment well at a cell density of 4 × 10⁴ (i.e., the number of cells added to each attachment well suspended in 100 μl T8 medium plus 10% fetal calf serum). The cells were allowed to attach to the explant surface for 2 hours before the explants were transferred back to the organ culture dish; this was referred to as time 0, after which 10 μl of test substance was pipetted on top of each explant. The underlying medium was changed every 2 days. For individual experiments, RPE cells of the same passage and from the same donor were always used. 

Addition of Growth Factors

Growth factors (PDGF-AB, PDGF-BB, bFGF, TGF-β₁, TGF-β₂, and IL-10) were prepared in serum-free T8 medium at concentrations of 1, 10, and 100 ng/ml. At time 0, retinal explants were placed in separate organ culture dishes (six explants per dish), and 10 μl of growth factor solution was pipetted on top of each explant, once only or three times daily for 7 days. Controls were replacement of growth factor solution with growth factor diluent, explants plus RPE cells with no additions, and explants without RPE cells but addition of growth factors or diluent. Contraction was assessed by image analysis (Seescan Imaging, Cambridge, UK) of photographs taken at 0, 3, 5, and 7 days by measuring the area of the explant viewed from above, expressed as a percentage of the area of the support membrane. ²²  

Addition of Neutralizing Antibodies to TGF-β and PDGF-AB

Neutralizing antibodies to PDGF and TGF-β₂, and control null antibodies, were prepared in serum-free T8 medium at concentrations of 0.01, 0.1, and 1 mg/ml. In initial experiments, 10 μl of antibody solution was pipetted on top of each explant three times daily for 7 days but in subsequent experiments, antibodies were added every 4 hours for the first 24 hours and then three times daily for the remaining 6 days. Controls were replacement of antibody solution with diluent, explants plus RPE cells with no additions, explants without RPE cells but addition of antibodies or diluent, and 100 ng/ml growth factor and 1 mg/ml neutralizing antibody to confirm the neutralizing effect of the antibodies in the organ culture model. Contraction was assessed at days 0, 3, 5, and 7, as described. 

Statistics

Replicates of six explants were created for each experiment and each experiment was repeated at least twice. Data were analyzed by Student’s t-test and analysis of variance, using commercial software (Simfit, Manchester, UK). ²⁴  

Results

Human RPE cells were consistently able to contract bovine retinal explants at all passages used (passages 5–9). However, the degree of contraction was passage dependent with contraction increasing with higher passage number (Fig. 1) and varying between different cell lines (this was independent of donor age). 

Effect of TGF-β and Neutralizing Antibodies to TGF-β on Retinal Contraction

Repeated addition of TGF-β₁ did not statistically affect the RPE cell-induced contraction of retinal explants even at the highest concentration of 100 ng/ml (data not shown). By contrast, repeated addition of TGF-β₂ modulated the RPE-mediated contraction of retinal explants in a dose dependent manner (Fig. 2) . Although there was variation between experiments, the same trend was always observed. The highest concentration of TGF-β₂ (100 ng/ml) significantly stimulated RPE-mediated retinal contraction by 40% more than that observed in the absence of added growth factor at day 7 (P < 0.01). Addition of 10 ng/ml TGF-β₂ did not produce a statistically significant effect on contraction, whereas at 1 ng/ml TGF-β₂ the degree of retinal contraction was significantly decreased by greater than 30% at day 7 compared with contraction in the absence of growth factor (P < 0.01). A single addition of 100 ng/ml TGF-β₂ at time 0 produced a contraction pattern comparable to that seen with repeated additions, whereas a single addition of 1 or 10 ng/ml TGF-β₂ produced no additional RPE-mediated retinal contraction compared with controls in the absence of growth factor (data not shown). In all experiments growth factor diluent had no effect on baseline RPE-mediated contraction, and no retinal contraction was observed in retinas that had not been seeded with RPE cells (either in the presence or absence of growth factor). 

Addition of neutralizing antibody at 1 mg/ml every 4 hours for 24 hours followed by every 8 hours for 6 days significantly reduced RPE-mediated retinal contraction compared with controls (P < 0.001; Fig. 3 ); RPE cell–mediated contraction was reduced by up to 50% in the presence of neutralizing antibody. Potency of the antibody was confirmed by demonstrating that 1 mg/ml of anti-TGF-β₂ antibody was sufficient to neutralize the stimulatory action of repeated additions of 100 ng/ml of human recombinant TGF-β₂ on RPE-mediated contraction of the retina (Fig. 3) . No effect was observed with lower concentrations (10 and 100 μg/ml) of antibody or lower frequency regimen. Addition of the neutralizing monoclonal antibody directed against active human recombinant TGF-β₂, even at a concentration of 1 mg/ml, every 8 hours for 7 days had no significant effect on the RPE-mediated contraction compared with untreated controls (data not shown). The contraction of explants was not significantly affected by addition of either null antibody (1 mg/ml human IgG4) or the antibody diluent. 

Effect of PDGF and a Neutralizing Antibody to PDGF on Retinal Contraction

RPE-mediated contraction of retinal explants was significantly enhanced in a dose-dependent manner by the addition of PDGF-AB (P < 0.01; Fig. 4 ); stimulation was greatest at 100 ng/ml and least at 1 ng/ml being, respectively, 40% and 18% greater than in the untreated controls. PDGF-AB had no effect on retinal explants that had not been seeded with RPE cells. By contrast, none of the concentrations of PDGF-BB tested in this experiment produced any statistically significant effect on retinal contraction by day 7 in culture (data not shown). 

Addition of all three concentrations of neutralizing antibody (0.01, 0.1, and 1 mg/ml) to PDGF every 4 hours for 24 hours, followed by every 8 hours for 6 days, significantly reduced RPE-mediated retinal contraction in the absence of antibody or added growth factor (P < 0.001; Fig. 5 ). RPE-mediated contraction was reduced by up to 50% in the presence of neutralizing antibody. Potency of the antibody was confirmed by demonstrating that 1 mg/ml of anti-PDGF antibody was sufficient to neutralize the stimulatory action of 100 ng/ml of PDGF-AB on RPE-mediated contraction of the retina (Fig. 5) . The contraction of explants was not significantly affected by addition of either null antibody (1 mg/ml) or the antibody diluent. 

Effect of a Combination of Neutralizing Antibodies to PDGF and TGF-β₂ on Retinal Contraction

A combination of neutralizing antibodies to PDGF and TGF-β tended to produced a greater inhibition of RPE cell–mediated retinal contraction compared with either antibody on its own; however, this effect was not significant (Table 1) . 

Effect of bFGF and IL-10 on Retinal Contraction

None of the concentrations of bFGF produced any statistically significant change in RPE-mediated retinal contraction compared with that seen in the absence of growth factor by day 7 in culture (data not shown). However, at the highest concentration of IL-10 tested (100 ng/ml) repeated addition of IL-10 resulted in an inhibition of RPE-mediated contraction of the retina; contraction was decreased by 63% compared with untreated controls (Fig. 6) . Addition of IL-10 had no significant effect on contraction at the lower doses tested (1 and 10 ng/ml). 

Discussion

In this study we have demonstrated the importance of both TGF-β₂ and PDGF in RPE-mediated contraction of the retina. Furthermore, we have demonstrated that RPE-mediated contraction of the retina can be inhibited by IL-10 and by neutralizing antibodies to TGF-β and PDGF. 

Both TGF-β and PDGF have been implicated in promoting contraction using a variety of cell types including RPE cells. ^{9 25 26 27} However, the potency of this effect appears to be isoform specific. Exogenous TGF-β₂ significantly stimulated RPE-mediated retinal contraction in this study, whereas TGF-β₁ had no effect. The timing of application may well be critical, because TGF-β₂ is reported to have most effect on contraction in the first 24 hours after application. ²⁸ Furthermore, our results show that a single addition of TGF-β₂ at the beginning of an experiment is enough to enhance contraction significantly. The potency of TGF-β₂ is in agreement with that reported in other studies. ^{26 29} However, the role of TGF-β₁ is equivocal. In contrast to our findings, it has previously been shown to enhance contraction by fibroblasts ^{25 30} and RPE cells in collagen gels. ^{6 10} This apparent discrepancy may reflect the difference in substrate between these studies and our model; the nature of the extracellular matrix is also known to regulate the cellular response to growth factors. ³¹  

PDGF also demonstrated isoform-specific RPE mediated contraction; PDGF-AB stimulated contraction, whereas PBGF-BB had no effect at the concentrations used. This suggests that the A chain of PDGF is essential for RPE-mediated contraction. Choudary et al. ⁹ have demonstrated that PDGF stimulates RPE-mediated collagen gel contraction but did not assess isoform differences. Studies using fibroblasts have reported that both PDGF-AB and PDGF-BB, but not PDGF-AA, stimulate the contraction of type 1 collagen and fibrin gels. ^{19 27 32} This may reflect differences in extracellular matrix or cell type specificity. Interestingly, bFGF had no effect on RPE-mediated contraction of the retina, despite its ability to promote contraction in other cell types. ^{33 34}  

The results from this study suggest that RPE cell–mediated contraction is induced by at least two growth factors and is almost certainly dependent on the nature of the extracellular matrix. However, the induction of the contractile response may be more complicated than this. Choudary et al. ⁹ have demonstrated that RPE-induced contraction of collagen gels by TGF-β and IL-1β was due, not to the direct effect of these growth factors, but to the production of another growth factor or factors by the target cells. Analysis of the conditioned medium revealed a peptide similar to PDGF which, when inhibited, prevented contraction. The observation from our study that RPE cells did not require repeated application of exogenous TGF-β₂ to increase their contraction and that neutralizing antibodies to TGF-β₂ did not result in total inhibition of contraction supports a role for a second growth factor, probably PDGF. The mechanism by which TGF-β and PDGF promote RPE-mediated contraction is unclear but is likely to involve adhesion receptors. TGF-β is known to increase α2β1 integrin expression, and neutralizing antibodies to the α2β1 integrin dimer can inhibit collagen matrix contraction. ³⁵ The β1 integrin is also known to be regulated by PDGF. ^{9 36}  

The observation that the contractile ability of cultured human RPE cells increases with passage number is not unexpected. Grisanti and Guidry ³⁷ have previously demonstrated that RPE cells take on a fibroblastic morphology with increased passage and that this is associated with an increased ability to contract silicone membranes. RPE cells have been proposed as one of the possible sources of the fibroblasts seen in epiretinal membranes, and this enhanced contractile ability could come about either through expression of α-smooth muscle actin, differences in the ability to manipulate surrounding extracellular matrix, or changes in the effect of soluble factors on these cells. The phenotypic change did not affect their response to either TGF-β isoforms, bFGF, PDGF-BB, or IL-10, in that a similar effect was observed at both high and low passage. 

A role for PDGF and TGF-β in PVR is supported by the observations that RPE cells have receptors for both TGF-β and PDGF, ¹⁶ that TGF-β and PDGF promote RPE-induced contraction, that both TGF-β (Mike Boulton, unpublished data, 1997) and PDGF have been localized to epiretinal membranes, ¹⁴ and that intravitreal levels of TGF-β and PDGF are increased in PVR. ^{38 39 40} The source of these growth factors is unclear, but there is evidence for paracrine-autocrine actions, in that a variety of retinal cells, including RPE cells, are capable of synthesizing both TGF-β and PDGF. ^{2 3 4} Thus, a feasible approach to the modulation of PVR is the inhibition of these growth factors, either by neutralizing their activity or through the action of known antagonists. We have clearly demonstrated in this study that neutralizing antibodies to both TGF-β and PDGF inhibit RPE-mediated contraction of the retina. Neutralizing antibodies to TGF-β have previously been shown to inhibit contraction in fibroblast-seeded collagen matrices, ¹¹ but to our knowledge similar studies have not been undertaken on human RPE cells. However, application of neutralizing antibodies against PDGF have previously been shown to inhibit RPE contraction in the collagen gel assay. ^{9 41} It is difficult to determine whether the neutralizing antibodies were acting against endogenous growth factor levels in the fetal calf serum used in these experiments (it is not possible to maintain retinas in the absence of serum) or whether they were neutralizing growth factor produced by the RPE or retina. Given that serum is reported to contain negligible TGF-β₂ ⁴² and approximately 100 ng/ml PDGF, ⁴³ we estimate that our cultures were exposed to no and 10 ng/ml of TGF-β and PDGF, respectively. These levels are sufficiently low to suggest that the neutralizing antibodies are also acting against locally produced growth factors. 

Exogenous IL-10 produced a dose-dependent inhibition of RPE-mediated contraction in our organ culture model. This is perhaps not surprising, because IL-10 is reported to reduce collagen synthesis, induce matrix metalloproteinase production at the posttranslational level, and suppress TGF-β production in bone marrow cells. ^{13 44} Because the addition of IL-10 produced the same effect as neutralizing antibodies to TGF-β, it is tempting to speculate that, in our model, IL-10 not only suppressed the synthesis of TGF-β₂ by RPE cells but also acted as an antagonist. The reported ability of IL-10 to inhibit collagen production and upregulate matrix metalloproteinase secretion is directly antagonistic to some TGF-β₂-induced actions. Further studies might investigate the interrelationship between TGF-β₂ and IL-10 on modulating RPE behavior. 

It is clear from these studies that both TGF-β and PDGF play an important role in RPE-mediated contraction of the retina. As previously discussed, our combined cell–organ culture system provides a model with reproducibility of histologic features and contractile responses representing the early events of PVR. ²² Therefore, our observations that neutralizing antibodies and certain peptides can inhibit this contractile response suggest that these agents should be considered for use in clinical trials. The efficacy of neutralizing antibodies to TGF-β in preventing fibrosis in the eye has recently been confirmed by Cordeiro et al. ⁴⁵ who demonstrated that application of neutralizing monoclonal antibodies to active TGF-β can prevent bleb failure in a recent model of glaucoma filtration surgery. However, our study suggests that it may be necessary to inhibit more than one factor simultaneously to obtain a maximal effect. Further study is necessary to identify new, and more successful, means of therapeutic intervention in PVR. 

 Presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May, 1998.

 Supported by the Wellcome Trust; The Guide Dogs for the Blind Association, Reading, UK; and the Manchester Royal Eye Hospital Endowments, UK.

 Submitted for publication July 2, 1999; revised November 10, 1999; accepted December 1, 1999.

 Commercial relationships policy: C5(LC, MB).

 

Figure 1.

 

Effect of increasing passage number (passages 5–9) on the RPE-mediated contraction of bovine retinal explants seeded with 4 × 10⁴ human RPE cells and maintained for 7 days. Explants were cultured in T8 with 10% fetal calf serum. Each point represents mean of nine explants ± SEM.

Figure 2.

 

The contraction of bovine retinal explants seeded with 4 × 10⁴ RPE cells (passage 6) over a time course of 7 days, comparing the effects of adding 10 μl of 1 ng/ml, 10 ng/ml, or 100 ng/ml human recombinant TGF-β₂ every 8 hours. Controls are either with no additions or the addition of growth factor diluent. Each data point represents the mean value of six replicates ± SEM. *Significant difference from control levels at the 1% level (Student’s t-test).

Figure 3.

 

The contraction of bovine retinal explants seeded with 4 × 10⁴ RPE cells (passage 9) over a time course of 7 days, comparing the effects of adding 10 μl of 100 ng/ml human recombinant TGF-β₂, 0.1 or 1 mg/ml 6B1 (human monoclonal antibody against active human TGF-β₂), or a combination of 1 mg/ml 6B1 and 100 ng/ml TGF-β₂, every 4 hours for 24 hours, then every 8 hours. Controls were untreated or treated with antibody diluent or 1 mg/ml of a null antibody. Each data point represents the mean value of six replicates ± SEM. *Significant difference from control levels at the 1% level (Student’s t-test).

Figure 4.

 

The contraction of bovine retinal explants seeded with 4 × 10⁴ RPE cells (passage 7) over a time-course of 7 days, comparing the effects of adding 10 μl of 1 ng/ml, 10 ng/ml, or 100 ng/ml human recombinant PDGF-AB every 8 hours. Controls were untreated. Each data point represents the mean value of three replicates + SEM.* Significant difference from control levels at the 1% level (Student’s t-test).

Figure 5.

 

The contraction of bovine retinal explants seeded with 4 × 10⁴ RPE cells (passage 6) over a time course of 7 days, comparing the effects of adding 10 μl of 100 ng/ml of PDGF-AB, 0.1 or 1 mg/ml of a neutralizing antibody against PDGF, or a combination of 1 mg/ml of neutralizing antibody and 100 ng/ml PDGF-AB every 4 hours for 24 hours, then every 8 hours. Controls were untreated or were treated with antibody diluent or 1 mg/ml of a null antibody. Each data point represents the mean value of six replicates ± SEM. *Significant difference from control levels at the 1% level (Student’s t-test).

Table 1.

 

Comparison of the Effects of Neutralizing Antibody Alone or in Combination on RPE-Induced Contraction of the Retina after 7 Days in Organ Culture

Treatment	Percentage Contraction
Control	47 ± 12
100 ng/ml TGF-β2	73 ± 10*
100 ng/ml PDGF-AB	77 ± 14*
1 mg/ml anti-TGF-β2	26 ± 9, †
1 mg/ml anti-PDGF	20 ± 10, †
1 mg/ml anti-TGF-β2+ 1 mg/ml anti-PDGF	16 ± 9, †
2 mg/ml null antibody	50 ± 10

 Data are means ± SEM.

*  Significantly greater than control (P < 0.01).

†  Significantly less than null antibody (P < 0.01).

Figure 6.

 

Graph showing the contraction of bovine retinal explants seeded with 4 × 10⁴ RPE cells (passage 7) over a time course of 5 days, comparing the effects of adding 10 μl of either 1 ng/ml, 10 ng/ml, or 100 ng/ml of human recombinant IL-10 every 8 hours. Controls were untreated. Each data point represents the mean value of five replicates ± SEM. *Significant difference from control levels at the 1% level (Student’s t-test).