Involvement of Hyaluronan and Its Receptor CD44 with Choroidal Neovascularization

Abstract

purpose. CD44 is a cell-surface adhesion molecule and receptor for hyaluronan (HA), one of the major extracellular matrix components. The purpose of the present study was to clarify a role of HA and CD44 in the development of choroidal neovascularization (CNV).

methods. Laser photocoagulation was used to induce CNV in C57BL/6 mice or CD44-deficient mice. The mRNA expression of CD44 and HA synthase (HAS)-2 in the retinal pigment epithelium (RPE)–choroid complex was evaluated by DNA microarray and real-time RT-PCR analyses 3 days after laser treatment. HA synthesis and CD44 expression were examined by immunohistochemistry 1 week after photocoagulation. Mice with laser-induced CNV were systemically administered the HA synthesis inhibitor 4-methylumbelliferone (MU) or an anti-CD44-neutralizing antibody. The response of CNV was analyzed by volumetric measurements 1 week after photocoagulation. Macrophage infiltration into CNV lesions was evaluated by real-time RT-PCR for F4/80 3 days after laser-induced injury.

results. The induction of CNV led to a significant increase in expression of CD44 and HAS2 mRNA. HA and CD44 were immunopositive in the CNV lesions. Compared with vehicle treatment, the systemic application of MU significantly attenuated CNV volume in a dose-dependent fashion, together with macrophage infiltration into the lesions. Consistently, antibody-based blockade of CD44 resulted in a significant reduction of CNV volume, compared with the isotype control. In contrast, genetic ablation of CD44 significantly augmented CNV formation together with HA accumulation and macrophage infiltration, compared with wild-type mice.

conclusions. These results indicate a significant role of HA and its receptor CD44 in the development of CNV.

Age-related macular degeneration (AMD) is the most common cause of blindness in people older than 50 in the developed countries. ¹ It is complicated by choroidal neovascularization (CNV), during which new choroidal vessels invade the subretinal space through Bruch’s membrane to form fibrovascular proliferative tissue containing vascular endothelial cells, fibroblasts, retinal pigment epithelial (RPE) cells, and various inflammatory leukocytes. ² Retinal neurons are irreversibly damaged by lipid leakage and bleeding from the immature new vessels in the CNV tissue. 

The molecular and cellular mechanisms underlying CNV are not fully elucidated. CNV in AMD develops with chronic inflammation adjacent to the RPE, Bruch’s membrane, and the choriocapillaris. ³ Recent experimental and clinical studies have indicated vascular endothelial growth factor (VEGF) as a critical factor in promoting CNV. ^{4 5} CNV formation is associated with the infiltration of inflammatory cells including macrophages, which are a rich source of VEGF. Pharmacologic depletion of macrophages results in significant suppression of murine CNV. ⁶ Genetic ablation of intercellular adhesion molecule (ICAM)-1 or C-C chemokine receptor (CCR)-2, a receptor for monocyte chemotactic protein (MCP)-1, has been shown to inhibit CNV in the murine model. ^{7 8} We have also shown that CNV-related inflammatory mechanisms are mediated by the renin–angiotensin system ^{9 10} and interleukin (IL)-6 receptor signaling. ¹¹  

The cell-surface adhesion molecule CD44 is a receptor for the major extracellular matrix (ECM) component hyaluronan (HA). ¹² This receptor is shown to contribute to the binding, endocytosis, and metabolism of HA. ¹³ CD44 is constitutively expressed on both parenchymal and hemopoietic cells, including macrophages. Increasing evidence has suggested that CD44 and its ligand HA play a significant role in the development of various inflammatory diseases. ^{14 15 16 17 18 19 20} The cell–ECM interaction between the HA produced in inflammatory foci and the CD44 expressed on macrophages is thought to regulate leukocyte recruitment and function in various pathologic conditions. Therefore, we hypothesized that disruption of HA–CD44 interaction would lead to amelioration of CNV, for which inflammatory processes are required. In the present study, we showed the first evidence of HA–CD44 involvement in the pathogenesis of CNV. 

Materials and Methods

Animals

Male C57BL/6 mice (CLEA, Tokyo, Japan) and CD44-deificient mice (based on the C57BL/6 strain; Jackson Laboratories, Bar Harbor, ME) at the age of 6 weeks were purchased and maintained in the specific pathogen-free Animal Facility of the Research Park, Keio University School of Medicine. All animal experiments were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Induction of CNV

Laser-induced CNV is a widely used animal model reflecting the pathogenesis of the inflammation-related CNV in human AMD. In this model, new vessels from the choroid invade the subretinal space after laser photocoagulation. Laser photocoagulation was performed around the optic nerve with a slit lamp delivery system (Novus Spectra; Lumenis, Tokyo, Japan), as described previously. ⁶  

Quantification of Laser-Induced CNV

One week after laser photocoagulation, the eyes were enucleated and fixed with 4% paraformaldehyde. Eye cups obtained by removing anterior segments were incubated with 0.5% fluorescein isothiocyanate (FITC)-isolectin B4 (Vector Laboratories, Burlingame, CA). CNV was visualized with a blue argon laser (wavelength, 488 nm) and a scanning laser confocal microscope (FV1000; Olympus, Tokyo, Japan). Horizontal optical sections of CNV were obtained at 1-μm intervals from the surface to the deepest focal plane. The area of CNV-related fluorescence was measured by NIH Image (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). The summation of the whole fluorescent area was used as the volume of CNV, as described previously. ⁶  

DNA Microarray Analysis

Tissue processing, microarray procedure, and bioinformatic and statistical analyses were performed according to published methods, with slight modifications. ^{21 22 23} Briefly, murine RPE-choroids were separated and placed in RNA lysis buffer, and RNA was extracted (RNeasy Minikit; Qiagen, Valencia, CA). Each RPE-choroid sample was prepared from four eyes. A gene microarray procedure was performed according to the manufacturer’s instructions (Affymetrix, Santa Clara, CA). Data analyses were performed with the microarray software (Microarray Suite 5.0; Affymetrix). The average of the all-probe set signals was normalized to 100. We compared the relative gene expression levels in the RPE-choroid from CNV mice with those from normal control animals. The changes were considered to be significant if there was a greater than threefold increase in the expression level of these two RPE-choroid tissues and the signal of the tissue from CNV animals was greater than 100. 

Real-Time RT-PCR

We performed quantitative RT-PCR of the selected genes to validate the microarray data. Commercially available primers and probe sets of the genes were prepared for the analysis. HA synthase (HAS)-2 and CD44 were selected for the experiment, while glyceraldehydes-3-phosphate dehydrogenase (GAPDH) was selected as the positive control for monitoring PCR reactions. Total RNA was isolated from the RPE-choroid complex of normal control mice and CNV animals on day 3 after photocoagulation (RNeasy Kit; Invitrogen, Carlsbad, CA). The extracted RNA was then reverse transcribed (First-Strand cDNA Synthesis Kit; GE Healthcare, Piscataway, NJ). Quantitative RT-PCR analyses were performed in a real-time PCR system (model 7900 HT Sequence Detection system; Applied Biosystems [ABI] Foster City, CA, in combination with TaqMan chemistry), as described previously. ^{24 25}  

Quantification of Infiltrating Macrophages

We measured the expression level of F4/80, a specific marker for macrophages, to quantify the amount of infiltrating macrophages. Total RNA was isolated from the RPE–choroid complex 3 days after photocoagulation and reverse transcribed. Quantitative RT-PCR analyses for F4/80 and GAPDH were performed as just described. Relative expression was then calculated as the density of the product of F4/80 divided by that for GAPDH from the same cDNA. 

Immunohistochemistry for HA, CD44, and F4/80

Immediately after the mice were killed with deep anesthesia, the eyes were enucleated, fixed with 4% paraformaldehyde, and embedded in paraffin. Deparaffinized sections were incubated over night at 4°C with a rat monoclonal antibody against mouse CD44 (clone IM7; BioLegend, San Diego, CA), biotinylated HA-binding protein (Seikagaku, Tokyo, Japan) or FITC-conjugated rat polyclonal antibody against F4/80 (MCA497F; Serotec, Oxford, UK). The sections were then washed three times with phosphate-buffered saline (PBS) before incubation for 60 minutes at room temperature with a secondary antibody labeled with Texas Red (Invitrogen), Alexa 546 (Molecular Probes, Eugene, OR), or FITC-conjugated avidin (Invitrogen). The sections were finally washed with PBS and mounted (Vectashield mounting medium with DAPI; Vector Laboratories), and visualized with a confocal microscope (model LSM710; Carl Zeiss MicroImaging, Tokyo, Japan). 

Inhibition of HA Synthesis with 4-Methylumbelliferone

4-Methylumbelliferone (MU; Wako, Osaka, Japan), which is a known inhibitor of HA synthesis, was suspended in 1% gum arabic at a density of 20%. ^{26 27} The mice were pretreated with MU suspension or 1% gum arabic solution as a control daily from 1 day before laser injury until the end of the study. MU was orally administered by a gastric feeding needle at a dose of 0.33, 1.0, or 3.0 g/kg body weight (BW). 

CD44 Blockade with the Neutralizing Antibody

To confirm the contributory role of HA–CD44 interaction in the development of CNV, we examined the effect of the antibody-based CD44 blockade. ²⁸ Mice received intraperitoneal injection of either the anti-CD44 antibody (clone IM7) or an isotype nonimmune antibody (BD Biosciences-Pharmingen, San Diego, CA) at a dose of 5 mg/kg BW 1 day before and on three consecutive days after laser photocoagulation. 

Statistical Analyses

All results are expressed as the mean ± SD. The values were processed for statistical analyses by Mann-Whitney test. Differences were considered statistically significant at P < 0.05. 

Results

Upregulated Expression of HAS2 and CD44 after CNV Induction

Based on the criteria described herein, the expression of 449 genes was significantly upregulated on day 3 after CNV induction compared with normal RPE-choroid complex (data not shown). Of these genes, Table 1shows representative molecules related to cell–ECM interaction, inflammation, and proliferation, some of which are known regulators of CNV, including VEGF, ⁴ MCP-1, ⁷ MMP-9, ²⁹ integrin β2 (CD18), ⁸ and αvβ3/αvβ5. ³⁰ Notably, the signal log ratios of HAS2 and CD44 are equivalent to those of the molecules responsible for CNV generation. 

To confirm CNV-associated induction of HAS2 and CD44 in the DNA microarray data (Table 1) , we performed real-time RT-PCR for these two molecules (Fig. 1) . The gene expression of HAS2 and CD44 was significantly (P < 0.01 for both) upregulated on day 3 after CNV induction, suggesting the contributions of the active synthesis of HA and HA–CD44 interaction to the pathogenesis of CNV. 

Positive Immunostaining of HA and CD44 in CNV Tissue

To confirm the quantitative RT-PCR data on HAS2 and CD44 (Fig. 1) , we performed immunohistochemistry for HA and CD44 in CNV specimens on day 7 after laser injury (Fig. 2A) . The immunostaining for HA was positive adjacent to and within the CNV lesion, but it tended to be diminished by the systemic administration of the HA synthesis inhibitor MU. CD44 immunoreactivity was intensely present in the interphotoreceptor matrix and was expressed in CNV lesions, consistent with previous reports showing CD44 localization to the interphotoreceptor matrix in normal eyes ³¹ and to macrophages, RPE, and vascular endothelial cells in CNV tissues. ³² Indeed, F4/80-positve macrophages infiltrating the CNV lesions expressed CD44 on day 3 after laser treatment (Fig. 2B) . 

Suppression of CNV Development by HA Synthesis Blockade with MU

CNV volume was analyzed to evaluate the effect of MU on the development of the lesions (Fig. 3) . Of importance, systemic application of MU led to significant (P < 0.01 for 1.0 and 3.0 g/kg BW) suppression of CNV volume in a dose-dependent manner, compared with vehicle treatment. 

Suppression of Macrophage Infiltration by Treatment with MU

To examine the cellular mechanisms underlying the MU-induced suppression of CNV (Fig. 3) , we quantified macrophage infiltration into CNV by real-time RT-PCR for F4/80, a specific marker of murine macrophages (Fig. 4) . MU treatment at a dose of 3.0 g/kg BW led to a significant (P < 0.05) decrease in the expression of F4/80 in the RPE–choroid complex on day 3 after CNV induction, compared with vehicle treatment. 

Suppression of CNV Development by Antibody-Based Blockade of CD44

To confirm the significance of HA–CD44 involvement with CNV generation, we systemically administered the neutralizing antibody against CD 44 (clone IM7) to the mice with CNV (Fig. 5) . Compared with treatment with an isotype control antibody, IM7 treatment significantly (P < 0.01) reduced CNV volume (Fig. 5)to the level observed in the MU treatment at the maximum-effect dose (3.0 g/kg BW; Fig. 3 ), suggesting a crucial role for HA–CD44 interaction in the pathogenesis of CNV. 

Enhancement of CNV Development, HA Accumulation, and Macrophage Infiltration by Genetic Ablation of CD44

To further confirm the involvement of HA–CD44 interaction with CNV formation, we induced CNV in CD44-deficient mice. CNV volume was significantly (P < 0.01) augmented in CD44-deficient mice compared with wild-type control animals (Fig. 6A) . In parallel, CD44-deficient mice exhibited enhanced HA accumulation on day 7 (Fig. 6B)and a significant (P < 0.05) increase in macrophage infiltration on day 3 after laser injury (Fig. 6C) . 

Discussion

The present study revealed, for the first time to our knowledge, that the major ECM component HA and its receptor CD44 contribute to the development of laser-induced CNV. First, induction of CNV led to marked upregulation of HAS2 and CD44 expression (Fig. 1) , in consistence with positive immunostaining of HA and CD44 in the CNV tissue (Fig. 2) . Second, inhibition of HA synthesis with MU substantially diminished HA accumulation in the CNV tissue (Fig. 2)and resulted in statistically significant suppression of CNV volume (Fig. 3)and of macrophage infiltration into CNV lesions (Fig. 4) . Moreover, antibody-based blockade of CD44 significantly reduced CNV volume (Fig. 5) , which further confirms the involvement of HA–CD44 interaction with CNV. In contrast, genetic ablation of CD44 significantly augmented CNV generation together with HA accumulation and macrophage infiltration (Fig. 6) . 

Our present data on HAS2 and CD44 (Figs. 1 2)are supported by and comparable with previous results ³³ showing the presence of HAS2 and CD44 mRNA in the retina and choroid of normal rabbits via quantitative RT-PCR. Of the three HAS isoforms, HAS2 was shown to be predominantly expressed in the normal RPE-choroid. Immunohistochemical data using the rat model of laser-induced CNV have shown that CD44 is localized to macrophages, RPE, and vascular endothelial cells in the CNV tissues. ³² In our murine CNV model as well, F4/80-positive macrophages expressed CD44 (Fig. 2) . Compared with other HAS isoforms, HAS2 is known to have relatively potent enzymatic activity producing high-molecular-weight HA and play a major role in wound healing and several biological stresses. ³³ In the laser-induced CNV model, thermal burn-related inflammation is likely to enhance the expression of HAS2 and subsequent accumulation of HA, leading to CD44-bearing macrophage infiltration. Indeed, the present data revealed that the HA synthesis inhibitor MU mediated decline of CNV generation (Fig. 3) , together with macrophage infiltration into the CNV (Fig. 4) , which has recently proven to be a critical cellular factor that facilitates the development of laser-induced CNV in mice. ⁶  

Although MU is safely and widely used in patients with biliary disorders as a choleretic (bile flow stimulant) in clinical practice, another new function of MU has recently been shown as the inhibitor of cell-surface HA formation. ^{26 27 34} HA is a polymer of disaccharides composed of glucuronic acid (GlcA) and N-acetylglucosamine (GlcNAc), both of which are repeatedly added by HAS to lengthen the HA chain in the presence of the substrates uridine 5′-diphosphate (UDP)-GlcA and UDP-GlcNAc. The mechanism for MU-mediated inhibition of HA synthesis in vitro was shown to involve the competitive production of MU-GlcA (i.e., glucuronidation of MU) by UDP-glucuronyltransferase resulting in decline of the substrate UDP-GlcA. ³⁵ Recent in vivo data on the murine model of liver metastatic tumor have revealed that oral administration of MU decreased tissue HA content together with metastasis, ²⁶ consistent with the currently observed inhibitory effect of MU in the eye (Figs. 2 3 4) . 

In concert with MU-induced suppression of CNV, parallel experiments using the anti-CD44 neutralizing antibody (Fig. 5)further confirmed the significant role of HA–CD44 interaction in the pathogenesis of CNV. Previous reports indicated the anti-inflammatory effect of antibody-based CD44 blockade in various in vivo models including cutaneous delayed-type hypersensitivity, ¹⁴ collagen type II-induced arthritis, ¹⁶ experimental autoimmune encephalomyelitis, ¹⁷ and IL-2-induced vascular leak syndrome in the lung and liver, ¹⁸ in accordance with the current data on inflammation-associated CNV (Fig. 5) . In addition, genetic ablation of CD44 led to significant suppression of ischemic brain injury in mice, together with cytokine expression and microglia activation. ¹⁹  

In contrast, however, in the murine model of bleomycin-induced lung injury CD44-deficient mice exhibited prolonged and enhanced inflammation due to persistent accumulation of HA and impaired clearance of infiltrating leukocytes including macrophages. ²⁰ CD44 functions not only as an adhesion molecule but also as a scavenger of HA. ¹³ As molecular and cellular mechanisms underlying the proinflammatory nature of CD44-deificient mice, RHAMM (receptor for HA-mediated motility), a redundant HA receptor distinct from CD44, compensated for the loss of CD44 in binding HA, supporting cell migration and exacerbating inflammation in collagen-induced arthritis. ³⁶ Of importance, in collagen-induced arthritis as well as bleomycin-induced lung inflammation, the loss of CD44 in knockout mice allowed increased accumulation of HA, leading to significant augmentation of the proinflammatory functions of RHAMM. In accordance with these knockout data, the present study revealed that CD44 gene ablation led to enhancement of CNV development together with increased HA accumulation and macrophage infiltration (Fig. 6) . Actually, the expression of RHAMM as well as CD44 was substantially induced in the RPE-choroid of wild-type mice after induction of CNV (Table 1) . Thus, the divergence between antibody-based blockade and genetic ablation of CD44 is seemingly contradictory, but it may in fact be complementary in terms of molecular redundancy. 

Another possible mechanism for promoting HA-mediated CNV is likely to include the proangiogenic effects of HA degradation products in stimulating the proliferation, migration and tube formation of vascular endothelial cells through intracellular signal transduction and activation mediated by CD44 and RHAMM. ³⁷ On the other hand, our new in vitro and in vivo data show that HA–CD44 interaction is essential for tumor necrosis factor (TNF)-α-induced epithelial–mesenchymal transition (EMT) in RPE cells (Takahashi E, et al., manuscript submitted, 2008). During EMT, ECM production and cell-ECM adhesion are enhanced, which leads to promotion of cell motility and fibrotic tissue formation. CNV development, complicated by fibrosis, is associated with upregulation of TNF-α and transforming growth factor (TGF)-β, ^{38 39 40} both of which are the major stimulators of EMT in various cells. Previous in vitro studies have indicated that TGF-β induces transdifferentiation in RPE cells. ^{41 42 43} Reasonably, these findings suggest that blockade of HA–CD44 interaction affects not only macrophages (Fig. 4)but also parenchymal cells, such as RPE and endothelial cells, thus resulting in the suppression of CNV formation. 

In summary, the present data are the first to show that inhibition of HA–CD44 interaction leads to significant suppression of inflammation-related CNV. These findings indicate the possibility of HA–CD44 blockade as a novel therapeutic strategy for inhibiting CNV in vision-threatening AMD. 

 Supported by Japanese Ministry of Education, Culture, Sports, Science and Technology Grant-in-Aid for Scientific Research 20791271 (TK).

 Submitted for publication October 21, 2008; revised February 20, 2009; accepted June 30, 2009.

 Disclosure: H. Mochimaru, None; E. Takahashi, None; N. Tsukamoto, None; J. Miyazaki, None; T. Yaguchi, None; T. Koto, None; T. Kurihara, None; K. Noda, None; Y. Ozawa, None; T. Ishimoto, None; Y. Kawakami, None; H. Tanihara, None; H. Saya, None; S. Ishida, None; K. Tsubota, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

DNA Microarray Data on Laser-Induced Gene Expression in the RPE-Choroid

	Gene	Signal Log Ratio
Collagen families	Collagen type 1, α1, α2	1.9, 1.6
	Collagen type 3, α1	2.0
	Collagen type 4, α3, α4	2.1, 2.8
	Collagen type 5, α1, α2, α3	1.6, 1.9, 2.0
	Collagen type 8, α1	2.2
Other ECM components	Fibronection 1	2.5
	Laminin B1 subunit 1	1.8
	Syndecan 1	1.8
Integrin families	Integrin αv	1.9
	Integrin αx	2.5
	Integrin β2 (CD18)	1.6
Hyaluronan synthase and receptors	HAS2	2.5
	RHAMM	2.0
	CD44	1.6
Inflammatory cytokines	VEGF-A	1.7
	CCL2 (MCP-1)	2.4
	CCL9 (MIP-1γ)	1.6
Matrix metalloproteinases (MMPs)	MMP-3	2.3
	MMP-9	2.9
	MMP-12	2.1

Figure 1.

 

Upregulated expression of HAS2 and CD44 after CNV induction. The expression of HAS2 (A) and CD44 (B) was significantly upregulated on day 3 after CNV induction. n = 4 for all. *P < 0.01.

Figure 2.

 

Positive immunostaining of HA and CD44 in CNV tissue. (A) Positive immunostaining of HA (green in left panels) adjacent to and within CNV area (arrowheads) was substantially diminished by application with MU (bottom left). CD44 immunoreactivity (red in right panels) in IPM and CNV lesions was unaltered after MU treatment. IPM, interphotoreceptor matrix; SC, sclera. (B) Green fluorescence from an anti-F4/80 antibody (left) and red fluorescence from an anti-CD44 antibody (middle) identified the F4/80-positive macrophages as expressing CD44 when the images were superimposed (arrows, right). Scale bar: (A) 25 μm; (B) 30 μm.

Figure 3.

 

Suppression of CNV development by HA synthesis blockade with MU. The development of CNV was significantly suppressed, in a dose-dependent manner, by treatment with MU. (A) Volume of CNV. (B) Flatmounted choroids from vehicle- and MU-treated mice. Arrowheads: lectin-stained CNV. n = 48, 39, 39, and 44 for vehicle and MU treatment (0.33, 1.0, and 3.0 g/kg BW), respectively. *P < 0.01. Scale bar, 100 μm.

Figure 4.

 

Suppression of macrophage infiltration by treatment with MU. MU treatment at the dose of 3.0 g/kg BW led to a significant decrease in the expression of F4/80 in the RPE-choroid complex on day 3 after CNV induction. n = 4 for all. **P < 0.05.

Figure 5.

 

Suppression of CNV development by antibody-based blockade of CD44. Compared with an isotype nonimmune antibody, the anti-CD44 neutralizing antibody significantly reduced the CNV volume. n = 44 for all. *P < 0.01.

Figure 6.

 

Enhancement of CNV development, HA accumulation, and macrophage infiltration by genetic ablation of CD44. (A) CNV volume was significantly augmented in CD44-deficient mice compared with wild-type control animals. The CD44-deficient mice exhibited enhanced HA accumulation (B) and a significant increase in macrophage infiltration (C). Arrowheads: CNV lesions. ONL, outer nuclear layer; SC, sclera. n = 52 and 58 for wild-type and CD44-defficient mice, respectively. *P < 0.01, **P < 0.05. Scale bar, 25 μm.

The authors thank Fumiyoshi Ishidate (Carl Zeiss MicroImaging) for technical assistance.