Doxycycline-Mediated Inhibition of Corneal Angiogenesis: An MMP-Independent Mechanism

Abstract

Purpose.: We explored the underlying mechanisms of doxycycline inhibition of corneal angiogenesis.

Methods.: Human umbilical vein endothelial cells (HUVECs) were cultured in vitro in the presence or absence of VEGF, doxycycline, or broad-spectrum matrix metalloproteinase (MMP) inhibitors (1,10-phenanthroline [1PT] and batimastat). HUVEC proliferation was determined using Cell Counting Kit-8. Rats with VEGF-induced corneal neovascularization (CNV) were treated with saline solution, 0.1% doxycycline, 0.1% 1PT, or 50 μM batimastat (n = 7/group). The length and area of CNV were measured on day 7. The activity of MMP-2 and MMP-9 was determined by a fluorogenic peptide substrate. Western blotting and ELISA were used to analyze the expression of phosphorylated eNOS and Akt, and PI3K activity.

Results.: Our results showed that doxycycline inhibited HUVEC proliferation induced by VEGF in a dose-dependent manner in vitro, and decreased CNV induced by VEGF in vivo in terms of vessel length and area. 1PT and batimastat showed similar MMP inhibitory functions with doxycycline in vitro and in vivo, but they had no effects on HUVEC proliferation, and only partially mimicked the inhibitory properties of doxycycline (∼45%) on angiogenesis induced by VEGF. In addition, although doxycycline is capable of modulating the PI3K/Akt-eNOS pathway in vitro and in vivo in an MMP-independent manner, 1PT and batimastat were not.

Conclusions.: The mechanism of doxycycline-mediated inhibition of angiogenesis occurs through MMP inhibitory activity and modulation of the PI3K/Akt-eNOS pathway, an MMP-independent mechanism.

Introduction

Angiogenesis is a double-edged sword, having important roles in physiologic and pathologic situations. In the eye, angiogenesis is essential for the development of many ocular diseases, including diabetic retinopathy, age-related macular degeneration, and corneal disorders, which may result in sight-threatening conditions and even blindness. ^1–6  

Doxycycline, a long-acting, low-cost, semisynthetic tetracycline antibiotic, has been used safely for decades in the clinical setting. In addition to its well-known antibiotic properties, accumulating evidence has shown that doxycycline possesses versatile nonantibiotic properties. Recently, the antiangiogenic property of doxycycline has received increasing attention, particularly in ocular angiogenesis. We and others have reported that oral or topical doxycycline can inhibit ocular angiogenesis effectively. ^7–12 To date, the mechanism of doxycycline-mediated inhibition of ocular angiogenesis remains elusive, although previous studies have suggested that the antiangiogenic effects of doxycycline may be a consequence of simply inhibiting matrix metalloproteinases (MMPs). 

To develop further clinical applications of doxycycline, it is important to elucidate the mechanism of the doxycycline-mediated inhibition of ocular angiogenesis. Thus, we undertook our study to examine the exact molecular mechanism by which doxycycline inhibits VEGF-mediated angiogenesis. Surprisingly, we discovered that the MMP-inhibitory activity is responsible only partially, and the PI3K/Akt/eNOS pathway, an MMP-independent mechanism, is implicated in the doxycycline-mediated inhibition of VEGF-induced angiogenesis. 

Materials and Methods

Cell Culture

A human umbilical vein endothelial cell (HUVEC) line ¹³ (EA.hy 926; American Type Culture Collection, Manassas, VA) was maintained in Dulbecco's modified Eagle's medium (Gibco, Grand Island, NY) containing 10% (vol/vol) fetal bovine serum, 4.5 g/mL glucose (Invitrogen, Carlsbad, CA), and 2% (wt/vol) hypoxanthine/aminopterin/thymidine (Sigma-Aldrich, St. Louis, MO) at 37°C in a 5% carbon dioxide atmosphere. All procedures used conformed to the tenets of the Declaration of Helsinki and were approved by the institutional review board of Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangzhou, China. 

Proliferation Assay

Grown on tissue culture plates, EA.hy 926 cells were pre-incubated with different doses of doxycycline or a synthetic broad-spectrum MMP inhibitor (1,10-phenanthroline, [1PT]; Sigma-Aldrich) for 1 hour, followed by incubation with or without VEGF (20 ng/mL), and then were allowed to proliferate for 24 hours. After 72 hours, 100 μL of cells from each well were transferred to a new 96-well plate with 10 μL of Cell Counting Kit-8 solution (Dojindo Laboratories, Kumamoto, Japan). The absorbance at 450 nm was measured with a microplate reader. Cell viability was assessed before and after treatment using trypan blue exclusion, and examined using phase contrast microscopy. 

Animals

Female Sprague-Dawley rats (6–8 weeks, 180–200 g) were obtained from the Guangzhou Animal Testing Center. The studies were approved by the Institutional Animal Care and Use Committee of Zhongshan Ophthalmic Center, Sun Yat-sen University. All procedures involving animal eye studies were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and were performed consistently by the same investigator. The right eye of each animal was used for the experiments, and the left eye was left intact. 

VEGF-Induced Corneal Neovascularization (CNV) Model

For the in vivo assessment of CNV induced by VEGF, we used a corneal micropocket assay. ¹⁴ In brief, the slow-release polymer Hydron (Sigma-Aldrich) was dissolved in absolute ethanol (12% wt/vol) in a rotator at 37°C overnight and stored at room temperature before pellet preparation. A stock solution of VEGF (PeproTech, Rocky Hill, NJ) at a concentration of 200 ng/μL was prepared by dissolving the lyophilized product in sterile PBS. A sucralfate (sucrose octasulfate aluminum complex; Sigma-Aldrich) stock solution was prepared by suspending sucralfate in sterile PBS at 100 μg/μL and stored at 4°C. The sucralfate suspension, VEGF solution, sterile PBS, and Hydron solution were mixed at a volume ratio of 4:5:21:30. The mixture, at 3 μL per drop, was deposited onto an autoclaved 2-mm Teflon sheet (Small Parts, Inc., Miami Lakes, FL) using a micropipette. The drops became solid pellets after drying overnight in a desiccator. Each pellet contained approximately 50 ng of VEGF. 

A small vertical incision was made into the center of the cornea at a depth of approximately 50%, and a micropocket was created by separating the lamella of the stroma toward the temporal limbus using a modified iris spatula. The shortest distance between the bottom of the micropocket and the limbus was approximately 1.4 mm. A prepared, sterile Hydron pellet was rehydrated with a drop of sterile saline and placed into the corneal micropocket. After surgery, tobramycin ointment (S.A. Alcon-Couvreur N.V., Rikjsweg, Belgium) was applied at the incision site to prevent dehydration and inflammation of the cornea. 

Quantification of CNV

CNV was quantified as in our previous study. ¹⁰ In brief, 6 days after corneal pocket surgery, the rats were euthanized. A total of 10 mL of Higgins waterproof India ink (Sanford, Bellwood, IL) was injected to visualize the corneal vessels. The eyes were enucleated and fixed in 10% neutralized, buffered formaldehyde, and the corneas were dissected and mounted on slides. Slide-mounted corneal images were taken using a slit-lamp microscope (SL120; Carl Zeiss, Oberkochen, Germany) with a 25× objective magnification. The CNV length and area were measured (Image Pro-Plus 5.1; Media Cybernetics Company, Silver Spring, MD). 

MMP Activity Assays

MMP activity was measured by a fluorogenic peptide substrate (R&D Systems, Inc., Minneapolis, MN) used to assess MMP activity (MMP-2 and MMP-9) following the protocol recommended by the manufacturer. Briefly, the MMP substrate was diluted in TCN buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 10 mmol/L CaCl₂, pH 7.5) and was added to the supernatants (preactivated by aminophenylmercuric acetate for 1 hour) before incubation at 37°C. After 30 minutes, total MMP activity was determined on a fluorometer (FLX 800 Microplate Fluorescence Reader; Bio-Tek Instruments, Winooski, VT). 

Western Blot

Rat corneal homogenates (50–100 μg of total protein) were separated on a polyacrylamide-SDS gel and electroblotted onto a nitrocellulose membrane (Bio-Rad, Hercules, CA). After blocking with TBS/5% nonfat dry milk, the membrane was incubated with antibodies against rat phosphorylated eNOS and Akt (Cell Signaling Technology, Inc., Danvers, MA) followed by incubation with a horseradish peroxidase (HRP)-conjugated secondary antibody, and the signals were visualized by enhanced chemiluminescence detection (Pierce, Rockford, IL). The blots also were reprobed with a specific antibody against β-actin (Sigma-Aldrich). 

Nitric Oxide (NO) Assay and ELISA

Supernatant NO levels were measured with the Griess reaction. ¹⁵ In brief, 50 μL of the samples were mixed with 0.1% N-1-napthylethylenediamine dihydrochloride and 1% sulfanilamide at room temperature for 10 minutes. The absorbance at 550 nm was measured with a microplate reader. The activity of PI3K was determined using a PI3K Activity ELISA: Pico kit (Echelon Biosciences, Inc., Salt Lake City, UT). The levels of intracellular cGMP were measured with the cGMP enzyme immunoassay kit supplied by Sigma-Aldrich. 

Statistical Analysis

The statistical analysis was performed using SPSS software (SPSS 16.0; SPSS, Inc., Chicago, IL). Statistically significant differences were defined as P < 0.05. Experimental data were compared using a one-way ANOVA and an independent two-sample t-test. 

Results

Doxycycline Suppresses HUVEC Proliferation Induced by VEGF In Vitro

To explore the mechanisms of doxycycline-mediated suppression of angiogenesis, we first tested whether doxycycline is capable of inhibiting the proliferation of vascular endothelial cells. To this end, HUVECs were cultured for 24 hours and then stimulated with VEGF in the absence or presence of doxycycline at different concentrations for another 24 hours. After 24 hours of VEGF stimulation, HUVEC proliferation was determined by Cell Counting Kit-8. Our results demonstrated clearly that doxycycline inhibits HUVEC proliferation induced by VEGF in a dose-dependent manner (Fig. 1A). VEGF-stimulated HUVEC proliferation was inhibited by 40 μM of doxycycline to levels observed in unstimulated cultures, but it did not affect HUVEC cell viability (Figs. 1A, 1B). Compared to 40 μM doxycycline, 80 μM doxycycline inhibited VEGF-stimulated HUVEC proliferation more effectively, but HUVEC cell viability was affected (Figs. 1A, 1B). 

Figure 1. 

 

Doxycycline inhibits the VEGF-induced proliferation of HUVECs. (A) Doxycycline inhibits the VEGF-induced proliferation of HUVECs in a dose-dependent manner. (B) The effects of doxycycline at different concentrations on cell viability. The results are reported as the mean ± SEM. *P < 0.05, **P < 0.01.

Inhibition of MMP Activity Is Not Required for the Doxycycline-Mediated Inhibition of HUVEC Proliferation Induced by VEGF In Vitro

Previous studies have reported that doxycycline may inhibit ocular angiogenesis by inhibiting MMP activity. ^7–12 Thus, we wanted to test whether the inhibition of MMP activity is responsible for the doxycycline-mediated inhibition of HUVEC proliferation induced by VEGF. To this end, HUVECs were cultured with VEGF with or without doxycycline (40 μM) or the broad-spectrum MMP inhibitors 1PT (40 μM) and batimastat (1 μM). After 24 hours, we analyzed MMP activity in different experimental groups. As expected, our results showed that doxycycline inhibited the activity of MMPs produced by HUVECs (Figs. 2A–C). Unexpectedly, however, 1PT and batimastat, which showed similar inhibitory effects on MMP activity, had no inhibitory effects on HUVEC proliferation induced by VEGF (Figs. 2A–C). Taken together, these findings indicated that there is an MMP-independent mechanism of doxycycline-mediated inhibition of HUVEC proliferation. 

Figure 2. 

 

Inhibition of MMP activity is not required in the doxycycline-mediated inhibition of VEGF-induced HUVEC proliferation in vitro. (A, B) MMP2 and MMP9 activity in the supernatants was measured by a fluorogenic peptide substrate. (C) The effects of 1PT and batimastat on VEGF-induced cell proliferation. The results are reported as the mean ± SEM. *P < 0.05, **P < 0.01.

The PI3K/AKT-eNOS-NO Pathway Is Involved in the Doxycycline-Mediated Inhibition of HUVEC Proliferation

Endothelial eNOS-derived NO is a key mediator in vascular endothelial cell proliferation and angiogenesis. ^16,17 Previous studies have reported that doxycycline is capable of inhibiting NO production. ¹⁸ Thus, to explore the mechanism of doxycycline-mediated inhibition of HUVEC proliferation, we tested whether doxycycline is capable of modulating the expression of phosphorylated eNOS and NO production by HUVECs. For this purpose, HUVECs were cultured with VEGF in the absence of doxycycline (40 μM), 1PT (40 μM), or batimastat (1 μM). After 1 hour, the supernatants and endothelial cells (ECs) were collected for further analysis. As shown in Figure 3A, doxycycline significantly reduced NO production by HUVECs stimulated with VEGF, but not 1PT and batimastat. Consistent with the previous result, the Western blot results also showed that doxycycline significantly decreased eNOS phosphorylation, while 1PT and batimastat did not (Fig. 3B). 

Figure 3. 

 

The PI3K/AKT-eNOS-NO pathway is involved in the doxycycline-mediated inhibition of HUVEC proliferation. (A) The NO concentration in the supernatants under different conditions was determined using the Griess reaction. (B, H) p-eNOS and p-Akt expression was analyzed by Western blot. (C, E) The level of intracellular cGMP under different conditions was determined with the cGMP enzyme immunoassay kit. (D, F) Cell proliferation under different conditions. (G) The activity of PI3K was determined using a PI3K activity ELISA kit. The results are reported as the mean ± SEM. *P < 0.05, **P < 0.01.

NO has been shown to act via the NO-cGMP pathway to modulate endothelial responses. ¹⁹ Therefore, we measured the level of cGMP in HUVECs under different conditions. Accompanied by HUVEC proliferation induced by VEGF, we found that VEGF also significantly increased the cGMP level in HUVECs, whereas a NO synthase inhibitor, Nomega-nitro--arginine methyl ester (L-NAME), significantly reversed the VEGF-mediated effects on the cGMP level and HUVEC proliferation. Similar to L-NAME, doxycycline also significantly inhibited the VEGF-mediated effects on the cGMP level and HUVEC proliferation, whereas 1PT and batimastat did not. These results confirmed further that NO-cGMP signaling was critical in VEGF-induced HUVEC proliferation, suggesting that doxycycline may inhibit HUVEC proliferation by modulating eNOS-NO-cGMP signaling. 

To explore further the role of eNOS-NO signaling in the doxycycline-mediated effects on VEGF-induced HUVEC proliferation, SNAP (an NO donor) and 8Br-cGMP (an analog of cGMP), were used in the next experiments. Our results showed that 100 μM SNAP and 50 μM 8Br-cGMP restored the intracellular cGMP level of HUVECs induced by VEGF, and significantly reversed the doxycycline-mediated inhibitory effects on VEGF-induced HUVEC proliferation. These results indicated that eNOS-NO-cGMP signaling is involved in the doxycycline-mediated inhibition of VEGF-induced HUVEC proliferation. 

Previous studies have shown that the PI3K/Akt pathway is critical for endothelium eNOS-derived NO release and VEGF-induced angiogenesis. ²⁰ Thus, to explore further the mechanism of doxycycline-mediated inhibition of angiogenesis, PI3K activity and the expression of phosphorylated Akt were analyzed by ELISA and Western blot. Our results showed that treatment with doxycycline significantly reduced PI3K activity and the expression of phosphorylated Akt in ECs (Figs. 3C, 3D). Taken together, these results implicated the PI3K/Akt-eNOS pathway, an MMP-independent mechanism, in the doxycycline-mediated inhibition of HUVEC proliferation and vascular hyperpermeability in vitro. 

Doxycycline Inhibits VEGF-Induced Neovascularization in a Rat Corneal Pocket Model

We next wanted to test whether the inhibitory capacity of doxycycline in vitro translates into the suppressive activity on angiogenesis in vivo. For this purpose, VEGF was used to induce corneal neovascularization in a rat corneal pocket model. After 6 days, all tested animals were euthanized to quantify angiogenesis. Representative images of the rat corneas are shown in Figure 4A. With doxycycline treatment, corneal neovascularization was decreased significantly in terms of vessel length and vessel area (Figs. 4B, 4C). 

Figure 4. 

 

Doxycycline inhibits VEGF-induced neovascularization in a rat corneal pocket model. (A) Typical images of CNV in different groups; quantitative measurement of vascular networks in the cornea 6 days after VEGF pellet implantation. (B) Average vessel length and (C) area of new vessels on the surface of the cornea (n = 5). The results are reported as the mean ± SEM. *P < 0.05, **P < 0.01.

Inhibition of MMP Activity Is Responsible Only Partially for the Doxycycline-Mediated Inhibition of CNV Induced by VEGF

We next wanted to test whether the inhibition of MMP activity is critical for the doxycycline-mediated inhibition of VEGF-induced CNV. To this end, 1% doxycycline solution, 1% 1PT solution, and 50 μM batimastat were used in the VEGF-induced CNV model. As shown in Figure 5, 1PT and batimastat showed inhibitory effects similar to doxycycline on MMP activity, but resulted in only a portion of the inhibitory effect (∼45%) of doxycycline on VEGF-induced CNV (Figs. 5A–D). These findings suggested that there is an MMP-independent mechanism in vivo in the doxycycline-mediated inhibition of CNV. 

Figure 5. 

 

The inhibition of MMP activity is only partially responsible for the doxycycline-mediated inhibition of VEGF-induced CNV. (A, B) The activity of MMP2 and MMP9 in corneal lysis was measured by a fluorogenic peptide substrate. (C, D) The effects of 0.1% doxycycline, 0.1% 1PT, and 50 μM batimastat on VEGF-induced CNV (n = 5). (E, F) The effects of SNAP (5 mM) on VEGF-induced CNV (n = 5). The results are reported as the mean ± SEM. *P < 0.05, **P < 0.01.

The PI3K/Akt-eNOS Pathway Is Implicated in the Doxycycline-Mediated Inhibition of CNV

Because our in vitro studies showed that the PI3K/Akt-eNOS pathway is involved in the doxycycline-mediated inhibition of HUVEC proliferation, we next asked whether the PI3K/Akt-eNOS pathway also is implicated in the doxycycline-mediated inhibition of CNV. To this end, the levels of phosphorylated eNOS and Akt in the cornea were determined by Western blot analysis. Our results showed that doxycycline significantly decreased eNOS and Akt phosphorylation, but 1PT and batimastat did not (Fig. 6). In addition, the topical application of SNAP (5 mM) also partially reversed the inhibitory effects of doxycycline on CNV. Taken together, these results suggested that the PI3K/Akt-eNOS pathway, an MMP-independent mechanism, is involved in the doxycycline-mediated inhibition of CNV. 

Figure 6. 

 

The PI3K/Akt-eNOS pathway is involved in the doxycycline-mediated inhibition of CNV. (A, B) p-eNOS and p-Akt expression was analyzed by Western blot. The results are reported as the mean ± SEM. *P < 0.05, **P < 0.01.

Discussion

For decades, doxycycline has been used successfully as a broad-spectrum antibiotic in the clinical setting. Recently, much attention has been focused on the doxycycline-mediated inhibitory effect on ocular angiogenesis. Our group first reported, to our knowledge, that oral doxycycline can reduce the CNV induced by an alkali burn. ⁹ In later studies, we showed that topical doxycycline not only inhibits CNV, ¹⁰ but it also enhances the inhibitory effects of bevacizumab on CNV in a rat corneal pocket model and an alkali burn model. ¹¹ Our findings are supported further by Aydin et al., who reported that topical doxycycline decreased CNV induced by a chemical burn. ⁶ In addition, Samtani et al. ⁷ and Cox et al. ⁸ further explored the doxycycline-mediated inhibitory effects on ocular angiogenesis. Taken together, these studies provide compelling evidence that doxycycline is a good alternative in the treatment of ocular angiogenesis, but they did not elucidate the mechanisms of the doxycycline-mediated inhibition of ocular angiogenesis. Defining the mechanisms is very important for the clinical application of doxycycline in ocular angiogenesis. Herein, we further explored the mechanisms underlying the doxycycline-mediated inhibition of angiogenesis. First, our results showed that doxycycline is capable of inhibiting HUVEC proliferation induced by VEGF and MMP activity in vitro. However, we unexpectedly found that the inhibition of MMP activity is not required for the doxycycline-mediated inhibition of VEGF-induced HUVEC proliferation. Furthermore, our study found that the PI3K/Akt-eNOS pathway is essential for the doxycycline-mediated inhibition of HUVEC proliferation. Using a VEGF-induced CNV model, we confirmed further in vivo that the PI3K/Akt-eNOS pathway, an MMP-independent mechanism, also is implicated in the doxycycline-mediated inhibition of angiogenesis. These findings provided the first evidence to our knowledge that the mechanism of doxycycline-mediated inhibition of angiogenesis occurs through MMP inhibitory activity and that the PI3K/Akt-eNOS pathway, an MMP-independent mechanism, also is involved. 

MMPs are among the most potent proangiogenic regulators, and they have important roles in CNV. ^21–23 MMPs can facilitate the angiogenic factor–stimulated migration of endothelial cells by disrupting cell–cell and cell–extracellular matrix connections, which ultimately leads to the formation of new vasculature. ^24,25 In our study, 1PT and batimastat, two broad-spectrum MMP inhibitors, showed inhibitory effects on MMP activity similar to doxycycline in vitro and in vivo. Interestingly, 1PT and batimastat do not have an inhibitory effect on VEGF-induced HUVEC proliferation, and only partially mimic the inhibitory effect of doxycycline on CNV in vivo. These findings, combined with the aforementioned MMP function, indicated that the combination of MMP inhibitory function and MMP-independent activities is responsible for the doxycycline-mediated inhibition of CNV. It is possible that doxycycline inhibits vascular endothelial cell growth through MMP-independent activities in vivo. Meanwhile, doxycycline reduces MMP activity to suppress vascular endothelial cell migration and invasion. Taken together, our results confirm further the therapeutic property of doxycycline in CNV, and reveal that MMP inhibitor functions and MMP-independent activities are involved in the doxycycline-mediated inhibition of CNV. 

In contrast to MMPs, NO, a critical mediator in angiogenesis, is an endothelial survival factor that inhibits apoptosis ^26,27 and increases endothelial cell proliferation. ²⁸ Furthermore, NO enhances endothelial migration. ^29,30 A number of angiogenic factors, including VEGF, upregulate the endothelial expression of NO synthase (eNOS) and stimulate the release of endothelium-derived NO. ³¹ The PI3K/Akt pathway has an important role in the process of eNOS activation by angiogenic factors. ¹⁸ Considering the long-standing assumption that doxycycline inhibits angiogenesis principally by inhibiting MMP activity, it is noteworthy that the PI3K/Akt-eNOS pathway, an MMP-independent mechanism, also was implicated in the doxycycline-mediated inhibition of VEGF-induced angiogenesis in vitro and in vivo. These findings can be used to explore further the application of doxycycline as a long-acting, low-cost, FDA-approved drug in the clinical setting. 

In summary, our study provides the first evidence to our knowledge that the mechanism of the doxycycline-mediated inhibition of angiogenesis occurs through MMP inhibitory activity and the PI3K/Akt-eNOS pathway, which is an MMP-independent mechanism. These findings confirmed further the therapeutic property of doxycycline in CNV, and implicated MMP inhibitory functions and MMP-independent involvement in the doxycycline-mediated inhibition of CNV. 

Footnotes

 Supported by the Guangdong Provincial Department University-Industry Cooperation Projects (2010B090400415) and the Technological Project of Guangzhou City, China (3030901016030).

Footnotes

 Disclosure: W. Su, None; Z. Li, None; F. Li, None; X. Chen, None; Q. Wan, None; D. Liang, None