Effects of ciliary muscle plasmid electrotransfer of TNF-α soluble receptor variants in experimental uveitis

A Corrigendum to this article was published on 13 August 2009


Intraocular inflammation has been recognized as a major factor leading to blindness. Because tumor necrosis factor-α (TNF-α) enhances intraocular cytotoxic events, systemic anti-TNF therapies have been introduced in the treatment of severe intraocular inflammation, but frequent re-injections are needed and are associated with severe side effects. We have devised a local intraocular nonviral gene therapy to deliver effective and sustained anti-TNF therapy in inflamed eyes. In this study, we show that transfection of the ciliary muscle by plasmids encoding for three different variants of the p55 TNF-α soluble receptor, using electrotransfer, resulted in sustained intraocular secretion of the encoded proteins, without any detection in the serum. In the eye, even the shorter monomeric variant resulted in efficient neutralization of TNF-α in a rat experimental model of endotoxin-induced uveitis, as long as 3 months after transfection. A subsequent downregulation of interleukin (IL)-6 and iNOS and upregulation of IL-10 expression was observed together with a decreased rolling of inflammatory cells in anterior segment vessels and reduced infiltration within the ocular tissues. Our results indicate that using a nonviral gene therapy strategy, the local self-production of monomeric TNF-α soluble receptors induces a local immunomodulation enabling the control of intraocular inflammation.


Tumor necrosis factor-α (TNF-α) is important in the amplification of inflammation within tissues. In inflammatory sites, it is mostly produced by infiltrating macrophages, T cells and can be produced by some resident cells. TNF-α also activates macrophages and influences the driving of CD4+ T cells to a Th1 phenotype.1 It was demonstrated that TNF-α has an important function in the pathogenesis of experimental noninfectious posterior segment intraocular inflammations.1, 2 In experimental autoimmune uveitis (EAU), an antigen-specific CD4+ T-cell-mediated autoimmune intraocular inflammatory disease, downregulation of TNF-α production reduced the intensity of retinal destruction.1 In endotoxin-induced uveitis (EIU) in rats, TNF-α neutralization not only reduced leukocytes rolling and activation but also retinal cell apoptosis.2 In patients with uveitis, high levels of TNF-α have been found in the ocular media.3, 4 Furthermore, polymorphisms within the TNF-α promoter region and the TNF-α receptors genes have been associated with a significant increased susceptibility for developing intraocular inflammation in HLA-B27+ patients.5, 6 Based on these initial findings, pilot studies have demonstrated the efficacy of anti-TNF-α strategies for the treatment of refractory uveitis using either etanercept (a p75 TNF receptor fusion protein), infliximab (a chimeric immunoglobulin G (IgG) monoclonal antibody (mAb) directed against TNF-α) or adalimumab (a human monoclonal TNF-α antibody).7, 8 In these patients systemic administration of antibodies directed toward TNF-α (infliximab or adalimumab) appeared more beneficial than the administration of the p75 TNF soluble receptor fusion protein (etanercept).9, 10

In EAU, neutralizing TNF-α activity with a systemic p55 TNF receptor fusion protein (TNFR-Ig) delayed the onset of inflammation and suppressed Th1 effector mechanisms protecting against retinal damage in the treated rats.11, 12 In patients, systemic TNF-α inhibition with p55 TNFR-Ig had a satisfactory therapeutic index along with a deviation of the Th2 type of immune response.13, 14 Despite these promising initial therapeutic results, it was evident that to achieve therapeutic efficacy repeated systemic administrations of anti-TNF agents were needed.15 These repeated administrations are associated with severe side effects including an increased risk of opportunistic infections and malignancies.16, 17, 18, 19 It was also observed that a constant level of drug is needed to avoid reactivation of the inflammatory processes.15, 19 Thus, when uveitis is the main (or sole) manifestation of inflammation, sustained levels of local intraocular anti-TNF therapy may allow for the control of the intraocular inflammatory processes with little (or no) unwarranted systemic effects.

Recently, we have developed a novel nonviral gene therapy strategy to express and produce locally a human TNF-α soluble receptor type 1 (p55) chimeric protein (hTNFR-Is) by the ciliary muscle. Furthermore, we have shown that high levels of this protein were found in the ocular media of rat eyes after ciliary muscle electrotransfer (ET) transfection.20

In the present study, we are reporting the results demonstrating that ciliary muscle ET of a plasmid encoding for different variants of TNF-α soluble p55 receptor markedly inhibited the development of EIU in the treated rat eyes, through mechanisms similar to those observed after systemic administration of anti-TNF-α, without any circulating TNF soluble receptors.


Efficient transfection of the ciliary muscle by electrotransfer

Eight days after ciliary muscle targeted ET of a plasmid encoding for the lacZ reporter gene (Figures 1a and b; see Materials and methods for details), β-galactosidase activity is macroscopically localized within the tissues around the corneal limbus of the treated eyes, restricted to the site of ET application (Figure 1c). Observation of the flatmounted tissues allows identification of the blue staining in the ciliary muscle (Figure 1d, white arrows) at the roots of the ciliary body, without transfection of ciliary processes (black arrows) demonstrating a targeted and localized transfection and gene expression.

Figure 1

Ciliary muscle electrotransfer (ET) of lacZ reporter gene in the rat eye. (a) Injection of plasmid in the ciliary muscle is followed by (b) ET, using one electrode inserted in the ciliary muscle, the other electrode forming an annular ring covering the ciliary muscle. (c) LacZ reporter gene expression (blue coloration) visualized macroscopically from outside the eye, 7 days after pVAX1-LacZ ET. (d) Flatmount of iris/ciliary body 7 days after pVAX1-LacZ ET: β-galactosidase activity is restricted to the ciliary muscle (white arrows) whereas the ciliary epithelium (black arrow) is not transfected and does not show any blue color.

Dose optimization of the plasmid encoding for TNF-α receptor and dose–response effect

Figure 2a represents the clinical scores of EIU using different plasmid doses in three different experiments. Electrotransfer of the empty plasmid (pVAX2 backbone) has no effect on clinical scores of EIU at all tested doses (P>0.05). No significant clinical effect on disease severity was observed between the group of rats receiving a ciliary muscle injection of 3 μg of the therapeutic plasmid pVAX2 hTNFR-Is/mIgG1 (3.7±0.2) when compared to control groups of untreated rats (3.8±0.2) or of rats receiving ET of the empty plasmid (3.8±0.2) (P>0.05 for both; n=8 for each group). Although injection of 15 μg of plasmid also did not significantly influence the uveitis score (3.1±0.5, P>0.05; n=12), injection of the highest dose of pVAX2 hTNFR-Is/mIgG1 (30 μg) significantly reduced the clinical score of EIU (1.6±0.5, P=0.007; n=14) compared to untreated rats (3.9±0.3) or rats treated by pVAX2 backbone ET (4.1±0.1). These observations show that injection of high plasmid dose in the ciliary muscle without current has significant clinical efficacy.

Figure 2

Clinical and biological effects as a function of plasmid dose combined to electrotransfer (ET). (a) Clinical scores (expressed as arbitrary units, AU) evaluated 24 h after lipopolysaccharide (LPS) challenge in eyes pretreated with injection (−ET) or electrotransfer (+ET) of 3 μg (n=8 eyes), 15 μg (n=12) or 30 μg (n=14) of therapeutic plasmid (pVAX2 hTNFR-Is/mIgG1encoding the chimera) in the ciliary muscle. Untreated endotoxin-induced uveitis (EIU) eyes or eyes electrotransferred with the pVAX2 plasmid backbone were used as controls (n=8, 12 or 14 for each group). ***P<0.001 and **P<0.01 vs control; ###P<0.001 and ##P<0.01 vs pVAX2+ET; $$$P<0.001 vs pVAX2 hTNFR-Is/mIgG1−ET. (b) Levels of tumor necrosis factor-α (TNF-α) (enzyme-linked immunosorbent assay, ELISA) in the ocular media of EIU eyes treated by injection±ET of 3 or 30 μg (n=4 samples of two eyes) of plasmid encoding the chimera. Untreated EIU eyes or pVAX2 +ET treated eyes served as controls. ***P<0.005 and *P<0.03 vs control; ###P<0.005 and #P<0.03 vs pVAX2+ET. ND, not detectable. (c) Levels of total proteins (Bradford assay) in the aqueous humors of EIU eyes treated by injection±ET of 15 μg (n=6 samples of two eyes) of the chimera encoding plasmid compared to untreated EIU eyes (control) and eyes treated with pVAX2 backbone. **P<0.005 vs control; ##P<0.005 vs pVAX2; $P<0.02 vs pVAX2 hTNFR-Is/mIgG1−ET. (d) Number of infiltrating polymorphonuclear leukocytes (PMNs) per eye section in the ocular media of EIU eyes untreated (control) or treated by injection±ET of 3 μg (n=12 sections of four eyes for each group) of plasmid encoding the chimera. ***P<0.0001 vs control; ###P<0.0001 vs pVAX2+ET. All the results are expressed as mean ± s.e.m. Statistical analyses: Mann–Whitney U-test.

Different results are observed when ET current (200 V cm−1) is applied after injection of the plasmid. Under these conditions, the lowest dose of pVAX2 hTNFR-Is/mIgG1 (3 μg) already significantly reduced EIU clinical scores (1.2±0.2) compared to the untreated control (3.8±0.2) (P<0.0001) or to the group of rats treated with the same dose of plasmid without current (3.7±0.2, P<0.0001). No additional clinical effects could be observed in these experiments when increasing the plasmid dose to 15 or 30 μg.

In the ocular media, levels of hTNFR-Is/mIgG1 were 262±70 pg ml−1 (n=6) in eyes that received injection of 30 μg of plasmid (not shown). The levels were strikingly higher (8675±4960 pg ml−1; n=6) in eyes that received ET after injection of 30 μg of plasmid (P<0.0001). In the serum, no hTNFR-Is receptors could be measured. Also, no hTNFR-Is was detected in eyes submitted to ET of the pVAX2 backbone.

As shown in Figures 2a and b, injection of 3 μg of plasmid without ET did not induce clinical inhibition of EIU despite the fact that TNF-α levels in the ocular media were reduced to 250±45 pg ml−1 compared to 510±44 pg ml−1 in control uveitic eyes and 478±33 in naked plasmid electrotransferred eyes (P<0.005). Injection of 3 μg of plasmid combined with ET reduced the TNF-α levels to 126±16 pg ml−1. In eyes receiving 30 μg of plasmid (with or without ET) the rat TNF-α production was markedly decreased and remained below detectable levels in both types of treatment.

Total proteins in the aqueous humor were used as a marker of blood–ocular barriers breakdown, as already used by other authors.21 Protein levels in the ocular media were not reduced in eyes injected with 15 μg of plasmid, correlating well with the lack of clinical efficacy of the injection without ET (Figures 2a and c). In eyes treated with 15 μg of plasmid combined with ET, EIU score was significantly decreased and the protein levels in aqueous humors were 33.1±4.3 mg ml−1. P<0.02, <0.02 and <0.005 values are reached when these levels are compared to those detected in eyes treated with anti-TNF plasmid injection only (50.8±2.8 mg ml−1), eyes electrotransferred with the pVAX2 plasmid backbone (65.8±4.2 mg ml−1) or in control untreated eyes (68.0±9.2 mg ml−1), respectively (Figure 2c).

Production of the different hTNFR-Is variants in ocular media after ET

As described above, the electrotransferred ciliary muscle was able to secrete the hTNFR-Is/mIgG1 protein. This protein is a fusion form of the human TNF-α soluble receptor p55 (hTNFR-Is) linked to the Fc fragment of mouse IgG1. The ability of the ciliary muscle to produce two other forms (monomeric and dimeric) of the human TNF-α soluble receptor p55 was assessed. As already described,22 the monomeric form is the physiological form of the human soluble p55 TNF-α receptor (hTNFR-Is) whereas the dimeric form (hTNFR-Is)2 is an original form made by association of two hTNFR-Is linked by a polyglycine polylinker (see Figure 3 for the schematic representation of the three variants).

Figure 3

Intraocular levels of the three variants of the p55 tumor necrosis factor-α (TNF-α) soluble receptors: differential clinical efficacy and TNF-α neutralization in endotoxin-induced uveitis (EIU). (a) Level of hTNFR-Is (enzyme-linked immunosorbent assay, ELISA) in ocular media and sera 7 days after electrotransfer (ET) of 10 μg (n=6 eyes per group) of plasmid encoding the monomer (hTNFR-Is), the chimera (hTNFR-Is/mIgG1) or the dimer (hTNFR-Is)2 in non-uveitic rats. The monomer (**P<0.003) and the chimera (*P<0.03) were detected at higher level in the intraocular fluids compared to the dimer. For each, no variant was detected in the serum. (b) Clinical scores (slit lamp examination, n=8 eyes per group) and corresponding TNF-α levels in ocular media (ELISA, n=4 samples of 2 eyes per group) evaluated 24 h after uveitis induction in eyes pretreated with ET of 30 μg of plasmid encoding the monomer, the chimera or the dimer. Untreated uveitic eyes were used as control. *P<0.03 vs control; #P<0.03 vs pVAX2 (hTNFR-Is)2; $P<0.03 vs pVAX2 hTNFR-Is. Results are expressed as mean±s.e.m. Statistical analyses: Mann–Whitney U-test. Symbols in black and gray are for clinical scores and TNF-α levels, respectively.

Concentration of the different constructions of hTNFR-Is was measured by enzyme-linked immunosorbent assay (ELISA) in ocular media of normal, non-uveitic eyes, following the injection of 10 μg of plasmid in combination with ET. Seven days after treatment, the monomer hTNFR-Is was found in the ocular fluids of the treated eyes at levels of 4782±299 pg ml−1 whereas the chimeric protein hTNFR-Is/mIgG1 levels were 3576±574 pg ml−1 and those of the dimer (hTNFR-Is)2 were 2151±530 pg ml−1 (Figure 3a).

None of the variants was detected in the serum, demonstrating the local intraocular production of the proteins.

Clinical and biological efficacy of the different hTNFR-Is variants of EIU

Because the three variants of hTNFR-Is were secreted in the ocular media after ciliary muscle ET, their efficacy in EIU was compared. Clinical score and TNF-α intraocular concentration were measured 24 h after EIU induction. As shown in Figure 3b, a significant reduction in EIU clinical score was observed in the group of eyes treated with the monomeric variant (1.75±0.25) compared to the untreated uveitic control eyes (4.25±0.14) (P<0.03). The dimer was produced in low amounts (Figure 3a) and does not significantly reduce EIU (4.5±0.5) (P>0.05; Figure 3b). The chimeric hTNFR-Is, on the other hand, induced the most potent inhibitory effect on clinical EIU (0.75±0.25). The fact that low molecular weight and simply designed monomeric forms of TNF-α soluble receptors can be produced by the ciliary muscle and have a clinical efficacy is of particular interest for further clinical transposition of this therapeutic technology.

In our model, the endogenous TNF-α levels correlated well with the observed clinical scores (Figure 3b). The TNF-α levels were 1039.8±35.4 pg ml−1 in the control group, 825.7±92.6 pg ml−1 in the group treated with the dimer and 341.8±63 pg ml−1 in the group treated with the monomer, with undetectable levels in the group of rats treated with the chimeric variant. Overall, the correlation between the clinical score of disease and TNF-α levels in the ocular media yields a linear relationship between the two (R2=0.89).

In vivo confocal microscopy showing effect of the treatment on rolling of inflammatory cells

As shown in videos (Supplementary Material), intense rolling of the inflammatory cells (video A) with diapedesis was observed in the limbal vessels from control EIU rat eyes, treated by ET of pVAX2 empty plasmid. In rat eyes treated by ET of 30 μg of pVAX2 hTNFR-Is/mIgG1, no (or random occasional) rolling and no diapedesis were observed during the period of examination (video B).

Characterization and quantification of cellular infiltration by immunohistochemistry

As no significant difference was observed between untreated control EIU eyes and eyes treated by ET of the pVAX2 plasmid in immunohistochemical analyses (data not shown), only the naked plasmid control was used (Figures 4 and 5).

Figure 4

Effect of electrotransfer (ET) of chimeric hTNFR-Is on ED1+ cells (red), polymorphonuclear leukocytes (PMNs) and inducible nitric oxide synthase (iNOS) (green) expression in the anterior and posterior segments of eyes with endotoxin-induced uveitis (EIU). Numerous ED1+/iNOS cells are observed in the iris/ciliary body of EIU rats (A, C and insets a, c) and in the retina (D, F and insets d, f), where they display a round morphology (inset d). Numerous ED1/iNOS+ cells are infiltrating the anterior segment (B, C, insets b, c) and the retina (DF, inset f) of EIU rats, suggesting that mostly PMNs produce iNOS. In eyes treated with ET of plasmids encoding hTNFR-Is/mIgG1, a reduced number of ED1+ cells is observed in the anterior segment (G, I and insets g, i), and ramified ED1+ cells are observed in the retina of treated eyes (J, L, inset l). A significant reduction in iNOS+ cell is observed in treated eyes (H, K). Nuclei are stained in blue with 46-diamidino-2-phenyl indole (DAPI). Bar, 100 μm; cb, ciliary body; GCL, ganglion cell layer, INL, inner nuclear layer; ONL, outer nuclear layer.

Figure 5

Effect of electrotransfer (ET) of chimeric hTNFR-Is on ED2+ cells (red), polymorphonuclear leukocytes (PMNs) and inducible nitric oxide synthase (iNOS) (green) expression in the anterior and posterior segments of eyes with endotoxin-induced uveitis (EIU). ED2+ cells are found in the ciliary body of eyes with EIU (A, C, inset), and in the retina (D, F, arrows), as well as in the anterior segment (G, I, inset) and in the retina of (J, L) treated eyes. Although cells co-expressing ED2 and iNOS are found in the anterior segment (C, inset arrows) and in the retina (F, f, inset arrow) of eyes with EIU, no cells co-expressing ED2 and iNOS are found in the treated eyes (I, L, l inset). Moreover, in the retinas of treated eyes, ED2+ cells have a ramified shape and remain in the inner layers (J), whereas in EIU eyes, ED2+ cells are round (D, arrows) and invade more outer layers. In EIU eyes, numerous ED2/iNOS+ cells are found in the anterior segment (B, C) and in the retina (E, F), while no iNOS+ cells are found in treated eyes (H, K). Nuclei are stained in blue with 46-diamidino-2-phenyl indole (DAPI). Bar, 100 μm; cb, ciliary body; GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

As shown in Figure 4, numerous ED1+ cells infiltrated the tissues of the anterior and posterior segments of control EIU eyes treated with injection of an empty plasmid combined with ET (Figures 4A, C, D and F). In these control eyes, numerous inducible nitric oxide synthase (iNOS)-positive cells, not labeled with ED1 (characteristic for neutrophils) infiltrated the tissues and media (Figures 4B, C, E and F). In the retina, infiltrating ED1+ macrophages cells did not express iNOS (Figures 4D and F), but numerous neutrophils expressed iNOS. In the rat eyes treated by ET of hTNFR-Is/mIgG1, a marked decrease in the number of infiltrating ED1+ cells in the anterior segment of the eyes (Figures 4G and I) and an almost complete absence of iNOS expression were observed (Figures 4H, I, K and L). However, in the retinas of treated eyes, the density of ED1+ cells displaying a more ramified shape remained high (Figures 4J and L).

In the iris/ciliary body and in the retina of control rat eyes, ED2+ cells (representing resident macrophages) were present (Figures 5A, D, C and F), as well as in treated eyes (Figures 5G, J, I and L). These cells generally did not express iNOS, which was mostly localized in neutrophils of control inflammed eyes (b, e and c, f). In treated eyes, no cells expressing iNOS were found (Figures 5H and K). In the retina, although resident round microglia had invaded the outer retina in pVAX2 control eyes (Figures 5D and F), it remained located in the inner layers and displayed a more ramified shape in the retina of treated eyes, suggesting a reduction of the activation status of these cells (Figures 5J and L).

Quantification of the cellular infiltrate (Figure 2d) revealed that treatment with the pVAX2 hTNFR-Is/mIgG1 plasmid significantly decreased the infiltration of polymorphonuclear leukocytes (PMNs) in the ocular media (414±18 PMNs per eye section) compared to untreated control eyes (1194±111 PMNs) and eyes treated with the pVAX2 naked plasmid (893±61 PMNs) (P<0.0001). No significant effect was observed with the naked plasmid (P>0.05). Similarly, ED1+ cells infiltrating the iris and ciliary body decreased by 59 and 63% in the anti-TNF-treated group in comparison with untreated uveitic eyes and pVAX2-treated eyes, respectively (P=0.0006; data not shown).

Electrotransfer of hTNFR-Is/mIgG1 influences the cytokine expression balance

iNOS, interleukin (IL)-1β, IL-6, IL-10, IFN-γ, TNF-α and chemokines (MIP-1α, Rantes) mRNA expression was tested in iris/ciliary bodies of control eyes (untreated or treated with the plasmid backbone) and eyes treated with injection or ET of 3 or 30 μg of hTNFR-Is/mIgG1. No significant differences in the expression of IL-1β, IFN-γ, TNF-α, MIP-1α and Rantes were found between the control and treated groups (P>0.05; Figure 6). In eyes treated with the only injection of 3 μg of pVAX2 hTNFR-Is/mIgG1, iNOS (1.05±0.28), IL-6 (1.08±0.15) and IL-10 (1.03±0.10) relative mRNA levels were not statistically different compared to untreated control eyes (1.00±0.33, iNOS; 1.00±0.02, IL-6; 1.00±0.09, IL-10) and eyes treated with ET of the pVAX2 backbone (1.19±0.19, iNOS; 1.21±0.15, IL-6; 1.08±0.30, IL-10) (Figures 6a, c and d), correlating well with the lack of clinical efficacy observed in Figure 2a. On the contrary, in eyes treated by ET of the same dose of plasmid, iNOS (0.35±0.08) and IL-6 (0.73±0.04) expression was significantly reduced compared to those treated by ET of the plasmid backbone, whereas IL-10 expression was significantly upregulated (3.73±0.01) (P<0.03). Moreover, a statistical difference was observed between the injected and electrotransferred groups, especially for iNOS and IL-10 mRNA expression (P<0.03). Using a higher dose (30 μg) of pVAX2 hTNFR-Is/mIgG1 (Figure 6b), we observed significantly (P<0.002) decreased level of iNOS relative mRNA in eyes treated by ET (0.33±0.05) compared to untreated control eyes (1.00±0.12). However, with this higher dose of plasmid, injection without ET significantly downregulated iNOS expression (0.40±0.04, P<0.002) (Figure 6b) in agreement with the clinical effect observed (Figure 2a).

Figure 6

Cytokines expression in iris/ciliary body of rat eyes with endotoxin-induced uveitis (EIU). Levels of inducible nitric oxide synthase (iNOS), interleukin (IL)-6 and IL-10 mRNA in iris/ciliary bodies from EIU rats pretreated with 3 μg (a, c and d; n=4 eyes per group) or 30 μg (b; n=6 eyes per group) of pVAX2 hTNFR-Is/mIgG1 (chimera)±ET. Untreated uveitic eyes or eyes treated with electrotransfer (ET) of 3 μg of pVAX2 plasmid backbone were used as controls. mRNA expression was evaluated by reverse transcription (RT)–PCR and expressed as experimental ratios relative to GAPDH) (arbitrary unit, AU). Results are expressed as mean±s.e.m. Mann–Whitney U-test: *P<0.03 and **P<0.003 vs control; #P<0.03 vs pVAX2+ET; $P<0.03 vs pVAX2 hTNFR-Is/mIgG1−ET. Pretreatment with ET of pVAX2 hTNFR-Is/mIgG1 decreases iNOS and IL-6 mRNA expression and increases IL-10 mRNA expression during EIU.

Long-term efficacy of treatment

For this study, the monomeric form (hTNFR-Is) of the plasmid was used (at the dose of 30 μg). As shown in Figure 7, high levels of hTNFR-Is were detected in the ocular media of eyes treated with the combination of ET (5966.2±674.5 pg ml−1; n=6) 3 months after the ciliary muscle plasmid transfection. Accordingly, low grade of EIU clinical score was seen in these treated eyes (1.2±0.4; n=8) compared to the score recorded in the control eyes (3.3±0.1, P=0.001; n=8). In eyes treated with injection of the plasmid only (without ET), the detected hTNFR-Is levels were very low (5.2±3.5 pg ml−1) with EIU clinical scores as high as those seen in the control eyes (2.8±0.3, P>0.05). This experiment demonstrates that transfection of plasmids to the ciliary muscle using ET induces long-term intraocular production of biologically active anti-TNF molecules.

Figure 7

Clinical efficacy and tumor necrosis factor-α (TNF-α) neutralization in endotoxin-induced uveitis (EIU) 3 months after treatment with the monomeric form of the p55 TNF-α soluble receptor. Line: levels of hTNFR-Is measured in ocular media 3 months after treatment with 30 μg of pVAX2 hTNFR-Is (monomer)±ET (enzyme-linked immunosorbent assay (ELISA), n=6 eyes per group), untreated eyes being used as controls. **P<0.003 vs control; $$P<0.003 vs pVAX2 hTNFR-Is/mIgG1−ET. Histogram, clinical scores evaluated in uveitic eyes in which EIU was induced 3 months after treatment with 30 μg of pVAX2 hTNFR-Is (monomer)±ET (n=8 eyes per group). **P=0.001 vs control; #P=0.01 vs pVAX2 hTNFR-Is/mIgG1−ET. Electrotransfer (ET) allows for a sustained production of hTNFR-Is in ocular media, which still maintain biological efficacy 3 months after treatment. Results are expressed as mean±s.e.m. Statistical analyses: Mann–Whitney U-test.


Different strategies have been used to neutralize deleterious effects of TNF-α both in animal models of experimental intraocular inflammation and in clinical conditions associated with severe and uncontrolled uveitis.1, 7, 8, 9, 10, 21, 23, 24 In rat eye with EIU, the p75 TNF-α soluble receptor-IgG1 fusion protein (etanercept) significantly reduced the uveitis scores and cell infiltration in ocular tissues when administered 24 h before disease induction as subcutaneous injections of 0.4 mg kg−1.23 In a rabbit model of EIU, infliximab at a dose of 20 mg kg−1 was given intravenously 24 h before the lipopolysaccharide (LPS) challenge, resulting in a significant reduction in the clinical scores of uveitis, together with a reduced number of cell infiltration and protein exudation.21 In experimental EAU in rats, neutralization of TNF-α using a p55 TNF-α receptor conjugated with IgG (p55-Rc-IgG fusion protein), injected systemically, resulted in a delayed onset of clinical signs of uveitis without decreasing significantly the severity of the disease. Interestingly, although CD4+ infiltrating cells remain high, a significant rescue of photoreceptors was observed with a decreased macrophages and granulocytes infiltration.11 Moreover, inhibition of iNOS expression together with an increased TGF-β production suggested a suppression of the activation of infiltrating macrophages and resulted in a suppression of retinal damages.25 The ability to suppress disease experimentally has led to the successful translation of anti-TNF therapy, using p55-Rc-IgG for treatment of uveitis. In these patients, altered peripheral blood CD4+ T-cell profiles were observed following each administration of the drug.14 Numerous additional studies have reported beneficial effects of anti-TNF in patients with noninfectious uveitis.8, 26 Studies derived mostly from investigator-sponsored trials and uncontrolled case series of patients have suggested that TNF antagonists, mainly infliximab, are useful in the treatment of ocular inflammation associated with Behçet's disease, rheumatoid arthritis, juvenile arthritis, ankylosing spondyloarthritis, Crohn's disease and sarcoidosis.7, 15 Infliximab was found beneficial also in small series of patients with intraocular inflammation associated with various ailments.7, 8

However, to date, no controlled, randomized study has identified the optimal regimen and the clear indications of anti-TNF treatments in the management of intraocular inflammation. Also, no study is available to support the use of one specific compound in these indications, even if infliximab appears more efficient than etanercept, and associated to reduced numbers of induced uveitis.27 However, some reports suggested that pulmonary tuberculosis is more frequent in patients under infliximab than under etanercept treatment.28

The need for repeated systemic injections of these anti-TNF drugs for long periods of time enhances the rate of undesirable side effects of anti-TNF therapies.28, 29 Therefore, simple local administration would have indisputable advantages especially for a ‘closed’ organ as the eye. Moreover, the induction of self-continuous and sustained production of therapeutic anti-TNF-α proteins by the eye tissues was a most appealing possibility. We investigated the feasibility of these two ideas and combined them for the possible implementation of an efficient treatment for uveitis.

In a preliminary study, we succeeded in demonstrating that the ocular ciliary muscle can be used as an intraocular ‘producer’ for protein molecules.20 Extending this preliminary proof-of-principle findings, we undertook the present study. Transfecting the ocular ciliary muscle with plasmids encoding for different variants of hTNFR-Is (p55), we observed a high and long-term production of these proteins in the intraocular fluids. Moreover, the intraocular production of these proteins was also associated with a decrease in TNF-α release within the ocular media and a marked inhibition of clinical disease of treated eyes in the EIU rat model of ocular inflammation.

Several TNF blockers are available in clinical practice using either antibodies or soluble receptors. Clinical, uncontrolled studies seem to favor the use of infliximab over those of etanercept for uveitis control. Etanercept is a dimeric fusion protein consisting of two extracellular domains of the human p75 (75 kDa) TNF receptor (sTNFR-II), linked to the Fc portion of a type 1 human immunoglobulin (IgG1). It therefore does not have any effect on the intrinsic activity of the sTNFR-I. Indeed, besides its basic functions of neutralizing circulating TNF-α, sTNFR-I specifically induces apoptosis in cells bearing transmembrane TNF-α, indicating that sTNFR-I acts as a ligand. The mechanisms underlying these phenomena have remained elusive. Interestingly, only infliximab, but not etanercept, induces apoptosis in transmembrane TNF-α-bearing cells. However, etanercept neutralizes soluble TNF-α as effectively as infliximab. Therefore, the induction of apoptosis by high-affinity TNF-α-binding agents such as sTNFR-I (p55) or anti-TNF-α antibodies is due to their ligation to transmembrane TNF-α and not to the neutralization of secreted TNF-α.30 This may explain why p55 soluble receptors may be more potent than p75 soluble receptors. Most of the molecules under development are variants of soluble p55 TNF-α receptors, such as onercept, or pegsunercept, which is a truncated form of soluble, natural TNF p55 (type I) receptor molecule, bound to a 30FD PEG to increase its half-life.31

This study demonstrates that the neutralization of ocular TNF-α, using the local production of p55 soluble receptors, efficiently suppresses clinical EIU. Decreasing only the ocular TNF-α levels reduced the rolling and the number of cells infiltrating ocular tissues, and the protein exudation in aqueous humor, correlating well with the ocular effects observed after systemic TNF-α blockade using etanercept or infliximab in EIU models.2, 21, 32 We could also demonstrate that after local treatment, the activation state of infiltrating and resident monocytes was impaired as demonstrated by a reduction of iNOS expression and a change in the shape and distribution of those cells. The downregulation of IL-6 pro-inflammatory mediator, associated with an upregulation of the IL-10 anti-inflammatory molecule, favors the hypothesis of a subsequent modulation of the cytokines ocular profile by the local blockade of TNF-α. Interestingly, systemic treatment with p55 TNFR-Ig suppressed NO production thus preventing its deleterious effect on the retina in a rat model of EAU.33 Moreover, the increased fraction of peripheral blood CD4+ T cells expressing IL-10 observed in patients treated systemically with a p55 TNFR-Ig was associated with a recovery of visual function.14 Other results from our laboratory have confirmed that in EAU, ET of pVAX2 hTNR-Is/mIgG1 induced a downregulation of inflammatory cytokines and chemokines and upregulation of anti-inflammatory cytokines, suggesting that this treatment could be beneficial in different types of uveitis.34

Because no hTNFR-Is was detected in the serum, even in the rats treated with the higher dose of plasmid (30 μg), the anti-inflammatory changes that were observed resulted mainly from the local blockade of TNF, without any effect on the circulating TNF levels. Our study highlights that permanently neutralizing the local production of TNF in the eye using ciliary muscle as a ‘factory’ for the needed soluble receptors can have a very high therapeutic index without the potential systemic side effects associated with the systemic administration of these drugs. However, although no clinical effects of sustained production of any forms of the TNF-α soluble receptors tested in this study were observed, and particularly no induced inflammation was detected, we cannot exclude that potential local side effects may result from permanent ocular TNF-α blockade. Toxicological studies will be carried out in the prospect of further clinical application of this technology.

Because the eye is a closed organ, protected from the systemic circulation by several barriers (external hemato-retinal barrier and blood–ocular barrier), low molecular weight protein fragments that have very short half-life in the serum may reach the therapeutic index if produced in the eye in a continuous manner by the ciliary muscle. Another beneficial effect of the local production of anti-TNF is that it is not influenced by the status of the ocular barriers, contrary to a systemic administration.

Using the monomeric form, we demonstrated that the ciliary muscle allowed for its sustained local expression and release in the ocular media, for at least 3 months. When LPS challenge was undertaken at this time point, uveitis did not develop, indicating that this treatment is not only able to cure evolving acute disease, but also to prevent recurrences as often observed in patients with uveitis. Ongoing kinetic studies in our laboratory show that production can be achieved for at least 9 months after initial transfection of plasmid to the ciliary muscle (not shown). As in arthritis,22, 35 the chimera exhibited the best clinical therapeutic outcome. However, unlike in experimental arthritis, the monomer was efficient in reducing uveitis scores. Its lack of efficiency in arthritis is explained by its instability in the circulation.22 In the eye, the high level of production of the monomer molecule counterbalances its slightly lower efficacy compared to the chimeric form. Moreover, a possible better diffusion in ocular tissues due to its small size could also contribute to its effectiveness. This opens a wide field of opportunities for the use of small simple proteins into the ocular media, using their continuous production by the ciliary muscle machinery.

In conclusion, our experiments have shown that using the ciliary muscle as a biofactory, the intraocular production of therapeutic proteins can be controlled by the dose of transfected plasmids. ET allows for the long-term expression of TNF-α soluble receptors, preventing uveitis as long as 3 months after initiation of treatment. Because no anti-TNFs were detected in the serum, even with the highest plasmid dose, no systemic effects of long-term systemic TNF blockade should be expected. However, the local TNF blockade produced biological ocular effects similar to those observed after systemic administration of anti-TNF. Unexpectedly, the monomeric variant, inefficient in experimental arthritis, produced continuously in the ocular media by the ciliary muscle exerted its anti-TNF activity. Designing ‘smart’ TNF-α antagonists does not only rely on the structure of the molecule, but also on the method to deliver it properly to a specifically affected tissue organ. We believe that the principle of ciliary muscle transfection for the production of therapeutic molecules can also be applied for additional intraocular diseases. The extension of this principle is now under extensive evaluation in our laboratory.

Materials and methods


To localize the expression after ciliary muscle ET, we used the commercially available plasmid pVAX1-LacZ containing the LacZ gene (Invitrogen, Carlsbad, CA, USA). The pVAX2 construct consists of a pVAX1 plasmid (Invitrogen, Carlsbad, CA, USA) in which the cytomegalovirus (CMV) promoter was replaced by the pCMVβ plasmid promoter (Clontech, Palo Alto, CA, USA). The pVAX2 hTNFR-Is, pVAX2 hTNFR-Is/mIgG1 and pVAX2 (hTNFR-Is)2 constructs are 3.6, 4.3 and 4.3 kb plasmids encoding for a monomeric (hTNFR-Is), chimeric (hTNFR-Is/mIgG1) and dimeric (hTNFR-Is)2 form of the human TNF-α soluble receptor type I (hTNFR-Is), also called p55 TNF-α receptor.22 These forms are drawn schematically in Figure 3 and described in the Results section.


Female Lewis rats (6- to 8-week old) weighing 230–250 g (Elevage Janvier, Le Genest Saint Isle, France) were used and handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Rats were anesthetized with intramuscular injection of ketamine (75 mg kg−1; Virbac, Carros Cedex, France) and Largactil (0.5 mg kg−1; Sanofi-aventis, Paris, France) before ocular injection and ET. At the end of the experiments, rats were anesthetized by intraperitoneal injection of pentobarbital (30 mg kg−1; Sanofi-aventis) before blood collection by intracardiac puncture and then killed with a lethal dose of pentobarbital.

Electrotransfer to rat ciliary muscle

Electrotransfer was performed as previously described,20 with a slight modification of the route of injection. Briefly, the plasmid (3–30 μg in 10 μl saline) was injected in the ciliary muscle using a 30-gauge disposable needle (BD Micro-Fine syringe, NM Médical, Asnière, France) transsclerally posterior to the limbus (Figure 1a).

For electric pulse delivery, a specially designed iridium/platine electrode (500 μm in diameter), naked up to 1 mm and rest covered with Teflon, was introduced into the preformed intrasscleral tunnel. The semi-annular platine return anode electrode was placed around the cornea, facing the active electrode (Figure 1b). ET was performed with eight electrical pulses (200 V cm−1, 20 ms, 5 Hz), generated by the 830 BTX electropulsator (Genetronics, San Diego, CA, USA) as previously described.20 These conditions were safe for ocular tissues.

To localize the expression, we treated two rats (two experimental eyes and the contralateral eyes used as controls) with the pVAX1 plasmid containing the LacZ gene (30 μg/10 μl). Two other rat eyes were treated similarly with the pVAX1 backbone. After 7 days, the six eyes were enucleated, incised at the limbus and fixed for 1 h at 4 °C in 2% paraformaldehyde and 0.2% glutaraldehyde in phosphate-buffered saline (PBS). They were rinsed three times in PBS before being incubated overnight at room temperature with 1 mg ml−1 X-gal (5-bromo-4-chloro-3-indolyl galactopyranoside; Sigma-Aldrich, Saint-Quentin-Fallavier, France) in PBS containing 5 mM of K3Fe(CN)6, 5 mM of K4Fe(CN)6, 2 mM of MgCl2 and 0.02% NP-40. After washing with PBS, direct imaging from the outside the eyes was carried out using a numerized camera (Coolpix; Nikon, Fnac, Paris, France). Then, the eyes were dissected by transversal section at 1 mm from the limbus and iris/ciliary body complexes were dissected, flatmounted in PBS/glycerol (1:1) and photographed under Aristoplan light microscope (Leica, Rueil-Malmaison, France) with the high-resolution Leica DFC480 R2 digital camera (Leica, Rueil-Malmaison, France).

Induction and scoring of endotoxin-induced uveitis

Endotoxin-induced uveitis was induced by footpad injection of 0.1 ml sterile pyrogen-free saline containing 500 μg kg−1 LPS (from Salmonella typhimurium; Sigma-Aldrich). Animals were examined by slit lamp during the maximal severity of EIU developing in this model 24 h after the footpad injection. The intensity of clinical ocular inflammation was scored on a scale from 0 to 5 for each eye as described previously36: grade (0) indicates no inflammation; grade 1 indicates the presence of a minimal iris and conjunctival vasodilation but without the observation of flare or cells in the anterior chamber (AC); grade 2 indicates the presence of moderate iris and conjunctival vessel dilation but without evident flare or cells in the AC; grade 3 indicates the presence of intense iris vessels dilation, flare and less than 10 cells per slit lamp field in the AC; grade 4 indicates the presence of more severe clinical signs than grade 3, with more than 10 cells in the AC with or without the formation of a hypopyon; grade 5 indicates the presence of intense inflammatory reaction, fibrin formation in the AC and total seclusion of the pupil.

Clinical evaluation was performed in a masked manner by two observers. The mean clinical score of the two observers was recorded.

Plasmid dose–response

To perform a biological dose–response curve, three doses (3, 15 and 30 μg in 10 μl saline) of plasmid encoding for the chimeric variant (hTNFR-Is/mIgG1) were injected in the ciliary muscle of right eyes of 68 rats. 16 eyes were treated with 3 μg, 24 eyes with 15 μg and 28 eyes received 30 μg of plasmid pVAX2 hTNFR-Is/mIgG1. Injection was immediately followed by ET in half of the treated eyes (8 eyes for 3 μg, 12 for 15 μg, 14 eyes for 30 μg). Contralateral eyes not receiving any injection nor ET pulses (n=8, 12 or 14) or receiving ET pulses after injection of the plasmid backbone pVAX2 (3 μg, n=8; 15 μg, n=12 and 30 μg, n=14) were used for controls. Six days after group assignment and treatment, EIU was induced in all of the 68 experimental rats. Clinical evaluation of uveitis intensity was performed 24 h after induction of disease. For TNF-α determination, the ocular media from rats treated with 3 μg (n=8) and 30 μg (n=8) of plasmid were immediately centrifuged and the cell-free fraction was collected, frozen at −20 °C before analysis and assayed using a specific ELISA for rat TNF-α (Duoset; R&D Systems, Abingdon, UK), according to the manufacturer's instructions (detection threshold estimated around 50–100 pg ml−1). To enhance detection abilities of TNF-α, we pooled the ocular media of two eyes within the same group of treatment and with the same clinical grading scores (eight eyes divided in four samples per condition). The level of hTNFR-Is/mIgG1 chimeric receptor was measured by ELISA in the group treated with 30 μg of plasmid (n=6) using a human receptor type I specific kit (Duoset; R&D Systems), according to the manufacturer's instructions (detection threshold estimated around 5–10 pg ml−1). Total proteins in the aqueous humor were used as a marker of blood–ocular barriers breakdown, as already used by others.21 The level of proteins within the aqueous humor was determined using the Bradford protein assay in the group treated with 15 μg of plasmid (Bio-Rad Laboratories, Marnes-la-Coquette, France). For this assay, the aqueous humors of two eyes within the same group of treatment and with the same clinical grading scores were pooled (12 eyes divided in six samples per condition).

Evaluation of hTNFR-Is variants production after plasmid electrotransfer in the ciliary muscle

The production of the three TNF-α soluble receptor I variants (the chimeric (hTNFR-Is/mIgG1), dimeric (hTNFR-Is)2 and monomeric (hTNFR-Is)) in the ocular media was evaluated 7 days after ciliary muscle ET of 10 μg of plasmids in the right eye of 18 rats (6 rats per variant and untreated contralateral eyes as control) using the previously described injection and electric conditions. The levels of the different TNF receptor variants were measured in the ocular media of treated (and untreated control) eyes and in the rats’ serum using a commercially available ELISA kit for human TNFR-Is (Duoset; R&D Systems) following the manufacturer's instructions. Results were expressed as mean±standard error of the mean (s.e.m.).

Biological effects of ciliary muscle electrotransfer of pVAX2 hTNFR-Is/mIgG1 on EIU

Because ET of 3 μg of plasmid encoding for the chimeric variant already led to an optimal clinical efficacy (Figure 2a), this dose was further used for the evaluation of cytokines and iNOS expression and immunohistochemistry. A total of 20 rats were used for this experiment: one group of 4 rats received the injection of 3 μg (in 10 μl of saline) of pVAX2 hTNFR-Is/mIgG1 in the ciliary muscle without ET, one group of 8 rats received the same injection followed by ET and one group of 8 rats received the injection of 10 μl of plasmid backbone (pVAX2) in the ciliary muscle followed by ET. Untreated contralateral eyes were used as control. EIU was induced 6 days after treatment and evaluation was performed 24 h later. Iris/ciliary bodies were snap-frozen for RNA extraction and reverse transcription (RT)-PCR gene expression analysis (n=4 for each group of treatment). The four remaining rats treated with ET of the plasmid encoding the chimeric variant and the four other rats treated with pVAX2 ET were used for cryosections and immunohistochemistry.

Twelve additional uveitic rats treated by injection followed or not by ET of 30 μg of pVAX2 hTNFR-Is/mIgG1 (n=6 for each group) were used for RT-PCR gene expression analysis as described above. Untreated control eyes served as control.

In vivo confocal microscopy

For this experiment, rats were treated with ET of 30 μg of plasmids. Control animals received ET of 30 μg of empty plasmid pVAX2 (n=4 in each group). EIU was induced in all rats 6 days after treatment and the eyes were examined by in vivo confocal microscopy 24 h after the LPS challenge. The HRT II/RCM (Heidelberg Engineering GmbH, Heidelberg, Germany) in vivo confocal microscope for animal studies was used as reported.37 Briefly, after anesthesia, the rats were positioned sideways under the objective. Several confocal microscopic images of the conjunctival vessels of each eye were taken. Images consisted of 384 × 384 pixels covering an area of 400 × 400 μm, with lateral and vertical resolutions of 1 and 2 μm, respectively. In selected cases, short videos were recorded to analyze the rolling phenomenon of inflammatory cells in blood vessels. The image acquisition time was 0.024 s with two-dimensional 384 × 384 pixel digitized images.

Tissue collection and processing for immunohistochemistry

Immediately after killing, we collected the eyes and fixed them for 1 h at 4 °C in PBS containing 4% paraformaldehyde before being rinsed overnight in PBS. The next day, samples were embedded and frozen in optimal cutting temperature compound (Tissue-Tek; Sakura Finetek, Zoeterwoude, the Netherlands) and stored at –80 °C. Frozen anteroposterior sections of eyes (10 μm thick) were performed at the optic nerve level using a cryostat (Leica CM3050S; Leica, Wetzlar, Germany) and mounted on superfrost slides for immunohistochemical analysis.


Sections were double-stained with the following combinations: ED1/iNOS and ED2/iNOS. Briefly, after permeabilization with 0.1% Triton X-100 in PBS for 30 min, specimens were rinsed and saturated for 30 min with 5% skimmed milk in PBS. They were then incubated with a 1:50 mouse monoclonal anti-macrosialin CD68 (clone ED1), directed against a cytoplasmic antigen in rat monocytes, macrophages and dendritic cells or a mouse monoclonal anti-CD163 (clone ED2), recognizing a cell membrane antigen of resident macrophages (both purchased from Serotec Ltd., Oxford, UK). After washing, sections were incubated with a secondary Alexa 594 (red)-conjugated donkey anti-mouse mAb (1:250e; Invitrogen, Cergy-Pontoise, France). Sections were washed and then incubated sequentially with a polyclonal rabbit anti-iNOS (1:75e; Transduction Laboratories, Lexington, KY, USA) followed by a secondary Alexa 488 (green)-conjugated goat anti-rabbit mAb (Interchim, Montluçon, France) at dilution 1:250. Each antibody was diluted in PBS (1% skimmed milk) 0.1% Triton X-100 and incubated for 30 min at room temperature. Different controls were included in every staining run: negative controls without primary antibodies and isotype controls with addition of normal mouse or rabbit serum Ig in place of primary antibodies. Sections were mounted with an anti-fade medium with 46-diamidino-2-phenyl indole (DAPI, Vectashield; Vector Laboratories, Burlingame, CA, USA) and observed by fluorescence photomicroscopy (Microphot FXA; Nikon, Melville, NY, USA). Digitized micrographs were obtained using a digital camera (Spot; BFI Optilas, Evry, France).

Infiltration of PMNs, identified by their nucleus shape, and ED1+ cells was quantified on immunolabeled sections. The analysis was performed on four eyes per experimental group, with three different sections per eye at the optic nerve head level. Results were expressed as mean±s.e.m.

RNA isolation and semiquantitative RT-PCR for cytokines and chemokines

Four or six eyes per group were used for this analysis. Immediately after dissection, iris/ciliary body complexes extracted from each eye were separately snap-frozen and stored at −80 °C until use. Total RNA was extracted from tissues (RNeasy minikit; Qiagen, Courtaboeuf, France) according to the manufacturer's instructions. RT was performed on 1 μg of total RNA in a total volume of 30 μl using Superscript II Reverse Transcriptase (Invitrogen, Cergy-Pontoise, France) following the manufacturer's instructions.

PCR was conducted in a total volume of 25 μl containing 2 μl of first-strand reaction product, 0.4 μM forward and 0.4 μM reverse primers, 0.4 μM dNTP Mix, 1.5 mM MgCl2, 1 × PCR buffer and 2.5 U Taq DNA polymerase (Invitrogen, Cergy-Pontoise, France). Forward and reverse primers (Table 1) were obtained from Invitrogen. After an initial denaturation (3 min at 94 °C), PCR cycles (30–35) were performed on a GeneAmp PCR System 2400 (PerkinElmer, Courtaboeuf, France) as follows: 30 s of denaturation (94 °C), 1 min of annealing, 30–60 s of elongation (72 °C). The final cycle was completed by 5 min of elongation at 72 °C. The PCR fragments were analyzed by 2% agarose gel electrophoresis and visualized by ethidium bromide staining under UV light. To verify that equal amounts of RNA were added in each PCR reaction within an experiment and to verify a uniform amplification process, we also transcribed and amplified GAPDH) mRNA for each sample. For each gene of interest, the relative band intensity was calculated in comparison to that for GAPDH.

Table 1 Overview of the primers used in the RT-PCR

Evaluation of the efficacy of p55 TNF receptor variants on EIU

The biological effects of the three different variants on the clinical scores of EIU and on the TNF-α levels in the ocular media were compared. For this purpose, 32 rats were used. In 24 rats, a ciliary muscle injection of 30 μg of plasmid encoding for one of the three variants construct (n=8 per variant) was administered. The injection was followed by ET. Eight additional rats were used as untreated controls. In all of the 32 rats, EIU was induced 6 days after the ET procedure. Clinical evaluation was performed 24 h later. After masked clinical scoring, two ocular media from rats within the same treatment group and with the same clinical EIU score were pooled for determination of TNF-α level. Ocular fluids were immediately centrifuged and the cell-free fraction was collected, frozen at −20 °C before analysis, then assayed using a specific ELISA for rat TNF-α (Duoset; R&D Systems), according to the manufacturer's instructions.

Efficacy of plasmid electrotransfer to the ciliary muscle for the long-term prevention of EIU and extended production of TNFR

The potential for long-term prevention of clinical disease and prolonged production of hTNFR-Is (monomeric form) in the rat eye was also studied. For this experiment, 42 Lewis rats were used. Among them, 28 rats received an injection of 30 μg pVAX2 hTNFR-Is to the right eye ciliary muscle. In 14 rats, the injection was followed by ET whereas 14 rats received the plasmid injection only. Fourteen rats remained untreated and were used as negative controls for hTNFR-Is level measurements in the ocular fluids. After 3 months (90 days), ocular media were collected from 18 rats: hTNFR-Is levels were evaluated by ELISA in the ocular media of eyes treated either by simple plasmid injection (n=6 eyes), by injection combined with ET (n=6) or in untreated eyes (n=6).

EIU was induced in the 24 remaining rats (8 receiving only the plasmid ciliary muscle injection, 8 rats receiving the ciliary muscle injection combined with ET and 8 naive rats). After 24 h, clinical disease development was assessed as described in the above sections.

Statistical analysis

All experiments were repeated twice leading to similar results. The results presented for each one came from the experiment for which the severity of EIU was the highest in the control group or in which the number of experimental points was the more numerous. Results are expressed as means±s.e.m. Data were compared using the nonparametric Mann–Whitney U-test. P<0.05 was considered statistically significant.


  1. 1

    Dick AD, Forrester JV, Liversidge J, Cope AP . The role of tumour necrosis factor (TNF-alpha) in experimental autoimmune uveoretinitis (EAU). Prog Retin Eye Res 2004; 23: 617–637.

    CAS  Article  Google Scholar 

  2. 2

    Koizumi K, Poulaki V, Doehmen S, Welsandt G, Radetzky S, Lappas A et al. Contribution of TNF-alpha to leukocyte adhesion, vascular leakage, and apoptotic cell death in endotoxin-induced uveitis in vivo. Invest Ophthalmol Vis Sci 2003; 44: 2184–2191.

    Article  Google Scholar 

  3. 3

    Santos Lacomba M, Marcos Martín C, Gallardo Galera JM, Gómez Vidal MA, Collantes Estévez E, Ramírez Chamond R et al. Aqueous humor and serum tumor necrosis factor-alpha in clinical uveitis. Ophthalmic Res 2001; 33: 251–255.

    CAS  Article  Google Scholar 

  4. 4

    Ahn JK, Yu HG, Chung H, Park YG . Intraocular cytokine environment in active Behçet uveitis. Am J Ophthalmol 2006; 142: 429–434.

    CAS  Article  Google Scholar 

  5. 5

    El-Shabrawi Y, Wegscheider BJ, Weger M, Renner W, Posch U, Ulrich S et al. Polymorphisms within the tumor necrosis factor-alpha promoter region in patients with HLA-B27-associated uveitis: association with susceptibility and clinical manifestations. Ophthalmology 2006; 113: 695–700.

    Article  Google Scholar 

  6. 6

    Kuo NW, Lympany PA, Menezo V, Lagan AL, John S, Yeo TK et al. TNF-857T, a genetic risk marker for acute anterior uveitis. Invest Ophthalmol Vis Sci 2005; 46: 1565–1571.

    Article  Google Scholar 

  7. 7

    Hale S, Lightman S . Anti-TNF therapies in the management of acute and chronic uveitis. Cytokine 2006; 33: 231–237.

    CAS  Article  Google Scholar 

  8. 8

    Lindstedt EW, Baarsma GS, Kuijpers RW, van Hagen PM . Anti-TNF-alpha therapy for sight threatening uveitis. Br J Ophthalmol 2005; 89: 533–536.

    CAS  Article  Google Scholar 

  9. 9

    Biester S, Deuter C, Michels H, Haefner R, Kuemmerle-Deschner J, Doycheva D et al. Adalimumab in the therapy of uveitis in childhood. Br J Ophthalmol 2007; 91: 319–324.

    Article  Google Scholar 

  10. 10

    Galor A, Perez VL, Hammel JP, Lowder CY . Differential effectiveness of etanercept and infliximab in the treatment of ocular inflammation. Ophthalmology 2006; 113: 2317–2323.

    Article  Google Scholar 

  11. 11

    Dick AD, McMenamin PG, Körner H, Scallon BJ, Ghrayeb J, Forrester JV et al. Inhibition of tumor necrosis factor activity minimizes target organ damage in experimental autoimmune uveoretinitis despite quantitatively normal activated T cell traffic to the retina. Eur J Immunol 1996; 26: 1018–1025.

    CAS  Article  Google Scholar 

  12. 12

    Dick AD, Duncan L, Hale G, Waldmann H, Isaacs J . Neutralizing TNF-alpha activity modulates T-cell phenotype and function in experimental autoimmune uveoretinitis. J Autoimmun 1998; 11: 255–264.

    CAS  Article  Google Scholar 

  13. 13

    Murphy CC, Greiner K, Plskova J, Duncan L, Frost A, Isaacs JD et al. Neutralizing tumor necrosis factor activity leads to remission in patients with refractory noninfectious posterior uveitis. Arch Ophthalmol 2004; 122: 845–851.

    CAS  Article  Google Scholar 

  14. 14

    Greiner K, Murphy CC, Willermain F, Duncan L, Plskova J, Hale G et al. Anti-TNFalpha therapy modulates the phenotype of peripheral blood CD4+ T cells in patients with posterior segment intraocular inflammation. Invest Ophthalmol Vis Sci 2004; 45: 170–176.

    Article  Google Scholar 

  15. 15

    Ohno S, Nakamura S, Hori S, Shimakawa M, Kawashima H, Mochizuki M et al. Efficacy, safety, and pharmacokinetics of multiple administration of infliximab in Behçet's disease with refractory uveoretinitis. J Rheumatol 2004; 31: 1362–1368.

    CAS  PubMed  Google Scholar 

  16. 16

    Calabrese L . The yin and yang of tumor necrosis factor inhibitors. Cleve Clin J Med 2006; 73: 251–256.

    Article  Google Scholar 

  17. 17

    Kempen JH, Gangaputra S, Daniel E, Levy-Clarke GA, Nussenblatt RB, Rosenbaum JT et al. Long-term risk of malignancy among patients treated with immunosuppressive agents for ocular inflammation: a critical assessment of the evidence. Am J Ophthalmol 2008; 146: 828–836.

    Article  Google Scholar 

  18. 18

    Bongartz T, Sutton AJ, Sweeting MJ, Buchan I, Matteson EL, Montori V . Anti-TNF antibody therapy in rheumatoid arthritis and the risk of serious infections and malignancies: systematic review and meta-analysis of rare harmful effects in randomized controlled trials. JAMA 2006; 295: 2275–2285.

    CAS  Article  Google Scholar 

  19. 19

    Tugal-Tutkun I, Mudun A, Urgancioglu M, Kamali S, Kasapoglu E, Inanc M et al. Efficacy of infliximab in the treatment of uveitis that is resistant to treatment with the combination of azathioprine, cyclosporine, and corticosteroids in Behçet's disease: an open-label trial. Arthritis Rheum 2005; 52: 2478–2484.

    CAS  Article  Google Scholar 

  20. 20

    Bloquel C, Bejjani R, Bigey P, Bedioui F, Doat M, BenEzra D et al. Plasmid electrotransfer of eye ciliary muscle: principles and therapeutic efficacy using hTNF-alpha soluble receptor in uveitis. FASEB J 2006; 20: 389–391.

    CAS  Article  Google Scholar 

  21. 21

    Diaz-Llopis M, García-Delpech S, Salom D, Udaondo P, Bosch-Morell F, Quijada A et al. High-dose infliximab prophylaxis in endotoxin-induced uveitis. J Ocul Pharmacol Ther 2007; 23: 343–350.

    CAS  Article  Google Scholar 

  22. 22

    Bloquel C, Bessis N, Boissier MC, Scherman D, Bigey P . Gene therapy of collagen-induced arthritis by electrotransfer of human tumor necrosis factor-alpha soluble receptor I variants. Hum Gene Ther 2004; 15: 189–201.

    CAS  Article  Google Scholar 

  23. 23

    Avunduk MC, Avunduk AM, Oztekin E, Baltaci AK, Ozyazgan Y, Mogolkoc R . Etanercept treatment in the endotoxin-induced uveitis of rats. Exp Eye Res 2004; 79: 357–365.

    Article  Google Scholar 

  24. 24

    Theodossiadis PG, Markomichelakis NN, Sfikakis PP . Tumor necrosis factor antagonists: preliminary evidence for an emerging approach in the treatment of ocular inflammation. Retina 2007; 27: 399–413.

    Article  Google Scholar 

  25. 25

    Robertson M, Liversidge J, Forrester JV, Dick AD . Neutralizing tumor necrosis factor-alpha activity suppresses activation of infiltrating macrophages in experimental autoimmune uveoretinitis. Invest Ophthalmol Vis Sci 2003; 44: 3034–3041.

    Article  Google Scholar 

  26. 26

    Imrie FR, Dick AD . Biologics in the treatment of uveitis. Curr Opin Ophthalmol 2007; 18: 481–486.

    Article  Google Scholar 

  27. 27

    Lim LL, Fraunfelder FW, Rosenbaum JT . Do tumor necrosis factor inhibitors cause uveitis? A registry-based study. Arthritis Rheum 2007; 56: 3248–3252.

    CAS  Article  Google Scholar 

  28. 28

    Carmona L, Gómez-Reino JJ, Rodríguez-Valverde V, Montero D, Pascual-Gómez E, Mola EM et al. Effectiveness of recommendations to prevent reactivation of latent tuberculosis infection in patients treated with tumor necrosis factor antagonists. Arthritis Rheum 2005; 52: 1766–1772.

    CAS  Article  Google Scholar 

  29. 29

    Lin J, Ziring D, Desai S, Kim S, Wong M, Korin Y et al. TNFalpha blockade in human diseases: an overview of efficacy and safety. Clin Immunol 2008; 126: 13–30.

    CAS  Article  Google Scholar 

  30. 30

    Waetzig GH, Rosenstiel P, Arlt A, Till A, Bräutigam K, Schäfer H et al. Soluble tumor necrosis factor (TNF) receptor-1 induces apoptosis via reverse TNF signaling and autocrine transforming growth factor-beta1. FASEB J 2005; 19: 91–93.

    CAS  Article  Google Scholar 

  31. 31

    Wong M, Ziring D, Korin Y, Desai S, Kim S, Lin J et al. TNFalpha blockade in human diseases: mechanisms and future directions. Clin Immunol 2008; 126: 121–136.

    CAS  Article  Google Scholar 

  32. 32

    Avunduk MC, Avunduk AM, Oztekin E, Baltaci AK, Ozyazgan Y, Mogolkoc R . Etanercept treatment in the endotoxin-induced uveitis of rats. Exp Eye Res 2004; 79: 357–365.

    Article  Google Scholar 

  33. 33

    Robertson M, Liversidge J, Forrester JV, Dick AD . Neutralizing tumor necrosis factor-alpha activity suppresses activation of infiltrating macrophages in experimental autoimmune uveoretinitis. Invest Ophthalmol Vis Sci 2003; 44: 3034–3041.

    Article  Google Scholar 

  34. 34

    Kowalczuk L, Touchard E, Camelo S, Naud MC, Castaneda B, Brunel N et al. Local ocular immunomodulation results from electrotransfer of plasmid encoding soluble TNF receptors in the ciliary muscle. Invest Ophthalmol Vis 2009; 50: 1761–1768.

    Article  Google Scholar 

  35. 35

    Bloquel C, Denys A, Boissier MC, Apparailly F, Bigey P, Scherman D et al. Intra-articular electrotransfer of plasmid encoding soluble TNF receptor variants in normal and arthritic mice. J Gene Med 2007; 9: 986–993.

    CAS  Article  Google Scholar 

  36. 36

    Behar-Cohen FF, Parel JM, Pouliquen Y, Thillaye-Goldenberg B, Goureau O, Heydolph S et al. Iontophoresis of dexamethasone in the treatment of endotoxin-induced-uveitis in rats. Exp Eye Res 1997; 65: 533–545.

    CAS  Article  Google Scholar 

  37. 37

    Labbé A, Liang H, Martin C, Brignole-Baudouin F, Warnet JM, Baudouin C . Comparative anatomy of laboratory animal corneas with a new-generation high-resolution in vivo confocal microscope. Curr Eye Res 2006; 31: 501–509.

    Article  Google Scholar 

Download references


We thank Dr Jean-Louis Bourges for his help to perform the acquisition of confocal microscopy videos. We also thank Christophe Klein for his precious technical skills regarding videos retouching and Dr Serge Camelo for helpful discussions and critical reading of the paper. This work was funded in part by the ANR EMERGENCE project ANR-05-EMPB-001-02 and the European project EVI-GENORET LSHG-CT-2005-512036.

Author information



Corresponding author

Correspondence to F Behar-Cohen.

Additional information

Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Touchard, E., Bloquel, C., Bigey, P. et al. Effects of ciliary muscle plasmid electrotransfer of TNF-α soluble receptor variants in experimental uveitis. Gene Ther 16, 862–873 (2009). https://doi.org/10.1038/gt.2009.43

Download citation


  • nonviral gene therapy
  • eye
  • inflammation
  • p55 TNFR
  • electroporation
  • drug delivery

Further reading