Systemic blockade of the Hyaluronan Receptor for Endocytosis (HARE) prevents lymph node metastasis of prostate cancer

Abstract

Tumor progression and metastasis are promoted by the remodeling of organized tissue architecture and engagement of molecular interactions that support tumor cell passage through endothelial barriers. Prostate tumor cells that secrete and turn over excessive quantities of pericellular hyaluronan (HA) exhibit accelerated growth kinetics and spontaneous lymph node metastasis in mice. The HA Receptor for Endocytosis (HARE) is an endocytic clearance receptor for HA in the liver that is also highly expressed in sinusoidal endothelium of lymph nodes and bone marrow, which are frequent sites of prostate cancer metastasis. In this study, we tested the hypothesis that HARE can act as an endothelial receptor for metastatic tumor cells with pericellular HA. In an orthotopic mouse model of prostate cancer, we delivered a monoclonal antibody against HARE that specifically blocks HA binding and internalization. This treatment fully blocked the formation of metastatic tumors in lymph nodes. No effects on primary tumor growth were observed and the antibody did not induce toxic outcomes in any other tissue. Our results implicate HARE for the first time in potentiation of tumor metastasis and suggest a novel mechanism by which tumor cell-associated HA could promote tissue-specific dissemination.

Introduction

Hyaluronan (HA) is a secreted glycosaminoglycan, the functions of which in adults primarily include facilitating joint or tissue lubrication, and coordinating wound healing and immune responses ¹. HA is continually synthesized by most tissues, and turned over systemically via the lymphatics and vasculature ² through the action of receptor-mediated endocytosis in the sinusoidal endothelial cells (SECs) of the lymph nodes (LN) and liver ³. Accumulation of HA has been extensively reported in epithelial cancers, such as prostate, where the presence of tumor cell HA matrices is both histopathologically correlated with aggressive cancer ⁴, and functionally implicated in metastatic spread ⁵.

In prostate cancer, metastatic sites include LN, liver and lungs, but bone metastasis and its complications are the most frequent cause of death ⁶. Studies have shown that bone tropism is facilitated by specific cell surface receptor-ligand interactions between tumor cells and bone marrow SECs. A static adhesion assay using monolayers of immortalized bone marrow SECs demonstrated that tumor cell HA matrix is essential for specific and rapid adherence of prostate tumor cells ⁷. Among several commercially available prostate tumor cell lines, only lines with metastatic potential in mice have surface HA matrices, and only those with surface HA adhere to bone marrow cells. Non-adherent cells become adherent upon transfection with HA synthase (HAS) constructs, and HA-mediated adhesion of metastatic cells is blocked by competition with exogenous HA, enzymatic degradation of the matrix, or HAS knockdown constructs ⁸. Cells stably selected for HAS knockdown are poorly tumorigenic in mice ⁹, and do not spontaneously metastasize to LN following prostate capsule injections as is observed for vector-transfected controls ¹⁰. Attempts to block HA-mediated adhesion to bone marrow SECs using antibodies against the primary HA receptor, CD44, were unsuccessful ⁷, leaving the identity of the bone metastasis counter receptor in question.

The HA Receptor for Endocytosis (HARE), also termed Stabilin 2, is a type I transmembrane scavenger receptor expressed highly in the SECs of LN, liver, spleen and bone marrow, and detectable in other specialized cell and tissue types ^{11, 12}. HARE is the major clearance receptor for circulating HA shed by other tissues ^{2, 3}, and functions as a coated pit-mediated recycling receptor that binds to its ligands in a large external domain ¹³. HA binding and uptake triggers an intracellular signaling cascade that activates additional pathways impacting growth and survival of prostate tumor cells. HARE is highly specific for HA but also constitutively facilitates the internalization and clearance of heparin, chondroitin sulfates, acetylated LDL, apoptotic cells and other ligands ¹⁴.

Since HARE is most prominently expressed in liver, LN and bone marrow, which are tissues most frequently colonized by metastatic tumor cells of epithelial origin bearing surface HA matrices, we hypothesized that HARE could serve as a receptor for HA-mediated tumor cell metastasis. In this study, we tested this hypothesis using the well-characterized mouse orthotopic model of prostate cancer in which metastatic tumor cells that are injected in the prostate develop both primary and spontaneously metastatic LN tumors. Animals were treated with a specific monoclonal antibody to block the HA binding function of HARE and metastasis was evaluated. We found that blocking HARE function limits HA-mediated LN metastasis, suggesting that HARE may contribute to tissue-specific metastasis of prostate cancer.

Material and Methods

Cell culture and reagents

The highly metastatic PC3 derived cell line, PC3M-LN4 (human prostate adenocarcinoma cells), was kindly provided by Dr I.J. Fidler (M. D. Anderson Cancer Center, Houston, TX), and maintained as described ¹⁵. Anti-HARE monoclonal antibody mAb-174 was generated and purified as described ¹². Nonimmune total mouse IgG was purified and used identically to mAb-174 as an isotype control for animal treatments. Low endotoxin content of antibody preparations was confirmed using the Limulus Lysate Endosafe Kit (Charles River).

Mouse liver perfusion and quantification of HA uptake

Liver perfusions were performed as for rats ¹² with modifications for smaller livers. SECs were purified by Percoll gradients and adhesion to glass to remove hepatocytes and Kupffer cells, respectively. Primary cultures were established in 24-well plates and used within 1 day. Cells were treated in serum-free medium for 1 hr at 37°C, washed, incubated with 1.5 µg/ml ¹²⁵I-HA in medium/BSA with the indicated amount of purified mAb-174 for 4 hr ¹². Specific uptake was measured in the presence of 100-fold excess unlabeled HA. HA uptake was determined with a gamma counter.

Orthotopic injection and animal treatments

Male NOD/SCID mice (Jackson Labs) were cared for and maintained under the supervision and guidelines of the University of Nebraska-Lincoln Institutional Animal Care and Use Committee. Six-week-old mice were intraperitoneally injected two days prior to tumor cell injection with 0.5 mg of either non-immune control or mAb-174 IgG. For tumor cell injections, animals were anesthetized with 2% isoflurane and injected in the dorsal prostate with 1x10⁵ PC3M-LN4 cells (10 mice per group) suspended in serum free medium, via a lower abdominal incision. Beginning at day two following surgical injection of tumor cells, animals received 0.3 mg of control or mAb-174 IgG intraperitoneally at three day intervals for 28 days, at which point all animals were sacrificed. Primary prostate tumors were removed and weighed. Paraaortic and subiliac LN (four total LN per animal) were harvested and cryopreserved in OCT compound for sectioning. We previously determined that the use of these injection conditions and the 28-day endpoint with the PC3M-LN4 cell line produces robust primary tumor growth and nearly 100% frequency of metastasis in paraaortic LN, while metastasis to other organs or lymphatic sites (e.g.; subiliac LN) is rarely detectable by hematoxylin and eosin stain (H/E). Necropsy of all animals in the study confirmed the absence of tumor or other morphological changes in organs other than prostate and paraaortic LN: liver, spleen, kidney, pancreas, intestine, lung, and uninvolved axillary and superficial cervical nodes were examined. In addition, no animals in the study presented with altered appetite, body weight, voiding function or overall activity that would have indicated toxicity.

Data analysis and statistics

All cryopreserved LN were cut into 8 µm sections (8–10 sections per node), ethanol fixed and stained with H/E. Each section was digitally imaged at 63x magnification. After assessing whether tumor tissue was present in each LN section, paraaortic LN areas were determined from the digital images using the histogram function of Adobe Photoshop. Mean node area for each animal group was normalized to the mean area of similarly analyzed paraaortic nodes from a separate cohort of animals containing no tumors (sham surgeries, n=5). For each group, this ratio (mean relative tumor burden) was plotted as fold increase in area ± SEM. Statistical significance was determined by Student’s unpaired t-test.

Results

Identification of a reagent to block HA binding by HARE in mice

To test whether HARE is involved in HA-mediated tumor cell metastasis in mice, we first determined the efficacy of mAb-174 for blocking HA uptake in mouse SECs. mAb-174 is a monoclonal antibody we previously characterized for its specificity against rat HARE, ability to block formation of HA•HARE complexes, and thereby prevent clathrin-mediated endocytosis of HA in rat perfused livers and primary SECs ¹². Here, we used a similar perfusion technique to isolate mouse liver SECs. Addition of mAb-174 to SECs inhibited radiolabeled HA uptake in a dose dependent manner, with maximum inhibition of ≈75% (Fig.1A). Specificity of mAb-174 was supported further by the inhibition of HA uptake with mAb-235 (Fig.1B), which only partially blocks HA binding to rat HARE and HA uptake by rat SECs ¹². No effect was seen with control IgG. Data correlate well with the degree of inhibition in perfused rat liver and SECs, suggesting mAb-174 is a valid tool for examining HARE function in HA-mediated processes in the mouse.

Fig. 1. Anti-HARE mAb-174 blocks HA binding to mouse liver SECs.

(A) Isolated primary mouse liver SECs were incubated with ¹²⁵I-HA and the indicated amount of purified mAb-174. Cells were washed and specific ¹²⁵I-HA uptake was determined. Values are the average of duplicates. (B) Cultured SECs were incubated with ¹²⁵I-HA as in (A) with no addition (none), 40 µg/ml of mouse IgG (mIgG), 150 µg/ml of HA, or 20 µg/ml of mAb-174 or mAb-235. Cells were washed and total ¹²⁵I-HA uptake was determined (values are mean ± SD, n = 3–5).

Lymph node tumor burden is reduced by treatment with anti-HARE mAb-174

We previously showed that PC3M-LN4 cells synthesize and retain abundant cell surface HA ⁷, metastasize readily to LN from the prostate in orthotopic injections, and that stable knockdown of HA synthesis eliminates metastasis ¹⁰. Here, we used this model system to test the effect of systemically delivered anti-HARE mAb-174 on prostate tumor growth and metastasis. Animals were preinjected with either control non-immune or mAb-174 IgG prior to orthotopic injection, and then continuously treated with non-immune or mAb-174 IgG. As expected, sizes of excised orthotopic tumors were statistically identical in all groups (Fig. 2), consistent with the absence of HARE expression in prostate tissue (¹¹ and our unpublished data).

Fig. 2. HA-blocking anti-HARE mAb-174 treatment does not impact primary prostate tumor growth.

Mice were treated with non-immune (IgG) or mAb-174 IgG (mAb174) 24 h prior to orthotopic injection of PC3M-LN4 prostate tumor cells. Animals were subsequently given injections of IgG or mAb-174 at 3-day intervals. Mean primary prostate tumor wet weight ± SEM was plotted for each group, along with values from animals implanted with tumor cells that received no IgG treatments (No Ab). Values are indicated on the plot. Pre: indicates the antibody used in the preinjection. Post: antibody given in subsequent injections.

Animals pretreated and continuously treated with non-immune IgG exhibited 100% frequency of metastasis to paraaortic LN (20/20) and 5% to subiliac LN (1/20), equivalent to the level observed in untreated animals (not shown). Since paraaortic LN were completely occupied by tumor cells, we quantified the increase in LN area to compare metastatic tumor burden among groups (Fig.3A). Interestingly, animals pretreated with control IgG and subsequently treated with mAb-174 beginning two days post-tumor cell injection bore equivalent frequency (18/20 positive paraaortic LN) and area of LN metastatic burden relative to control treated groups (Fig. 3A, 3B, 3C). In contrast, mice pretreated and then continuously treated with mAb-174 IgG had no evidence of tumor cell colonization in LN (0/20 positive paraaortic LN). LN areas in these animals were identical to those measured in non-tumor-bearing animals, which were ≈5% of the LN areas in tumor bearing animals given only control antibody. (Fig 3A, 3B, 3D; p<0.0001). Collectively, these results support a model in which human metastatic prostate tumor cells with cell-surface HA invade and colonize the LN via a HARE-mediated and HA-mediated mechanism.

Fig. 3. Antibody inhibition of HA binding to HARE decreases spontaneous LN metastasis of prostate tumors.

(A) Areas of paraaortic LN were used to compare LN tumor burden among tumor-injected untreated (No Ab) and treated animals, relative to sham-injected animals containing no tumors (NT). Mean normalized LN area per group is plotted ± SEM. Relative values are indicated on the plot (* p<0.0001 relative to the IgG/IgG group). Representative images of paraaortic LN from each treatment group are shown: (B) IgG/IgG; (C) IgG/mAb174; (D) mAb174/mAb174. The main images show the nodes in situ, indicated by an ellipse, and the inset panel illustrates a representative H/E stained section from that node.

Discussion

Prostate cancer progression is hallmarked by metastasis to tissues accessible via sinusoidal vasculatures, such as LN and bone marrow, in which expression of HARE is enriched on specialized SECs. Our results demonstrate that inhibition of HA binding to HARE can block tumor cell invasion from prostate to LN and suggest that HARE may be a key receptor that underlies the observed tissue tropism of tumor cells. The involvement of HARE also provides a novel mechanism to explain how HA-encapsulated tumor cells exhibit a metastatic advantage over those that lack HA. Our results are consistent with a model in which HARE functions both to facilitate arrest of tumor cells in a tissue-specific fashion and to transmit signals that support colonization of the tissue.

Ligation of HARE to HA may facilitate metastasis by allowing tumor cells to adhere to the sinusoids of the LN. LN have a primary function of accommodating the passage of lymphocytes and other immune cell types. HA is a well-recognized substrate for circulating cell adhesion, as evidenced by its ability to arrest lymphocytes in high endothelial venules ¹⁶. It has been postulated that tumor cells mimic the mechanisms used by lymphocytes, entering the lymphatics passively through shedding from an aggressive primary tumor ¹⁷. This may be accomplished by passage through pre-existing perforations in the afferent lymph vessels that drain the tumor ¹⁸. Tumor growth in the TRAMP mouse model of prostate cancer ¹⁹ and in prostate orthotopic implantations ²⁰ occurs with the activation and involvement of peritumoral lymphatic vessels that lead to tumor cell accumulation in the subcapsular sinus of the proximal paraaortic and subiliac LN. The requirement for a pre-blocking treatment is consistent with reports showing that tumor cells enter the lymphatics within the first few days following orthotopic injection ²¹. Binding of tumor cell HA to sinusoidal endothelial HARE would provide sufficient support for subsequent engagement of stronger adhesion mechanisms via the reticular matrix proteins that form the node architecture, followed by initiation of cell division, additional tumor cell shedding, and/or invasion of the parenchyma. In our system, the anti-HARE antibody may simply block HA-mediated adhesion of the tumor cells. However, the endocytic function of HARE may also be critical for endothelial transmigration of the tumor cell into the node and/or for subsequent colonization. This outcome could be indirectly impacted by altered crosstalk between HARE and other receptors localized to clathrin-coated pits. We know that HARE constitutively and rapidly cycles between internalized vesicles and the plasma membrane, so it is possible that antibody-mediated crosslinking of HARE at the plasma membrane preempts its clearance activity in the LN, rendering parenchymal invasion more challenging.

An alternative or complementary explanation could be that altered endothelial sinusoidal morphology or function in the presence of antibody blocking impacts processes such as tumor-induced lymphangiogenesis, which is implicated in progression of prostate cancer. LYVE-1, a member of the link module family of HA receptors, is primarily expressed on lymphatic endothelial cells, can support HA adhesion of lymphocytes, and mediates a low level of HA uptake through the lymphatic system during inflammation ²². Unlike HARE, however, LYVE-1 is not a coated pit-mediated recycling receptor that can mediate systemic HA clearance. Expression of LYVE-1 is used as a marker for lymphangiogenesis and lymphatic content of human tumors. Interestingly, lymph vessels are destroyed in the transition from benign to malignant prostate cancer, rather than created ²³. Studies show a correlation between VEGF-C and LN metastasis, suggesting lymph vessel permeability, though not lymph vessel density, may be the key factor underlying LN metastasis ²⁴. Since endothelial permeability is observed during HA-mediated bacterial infections, it is tempting to speculate that tumor cell surface HA would act similarly to bacterial HA by inducing or exacerbating lymphatic endothelial permeability. LYVE-1 null mice have normal lymphatic development, and form tumors comparably to wild-type mice ²⁵, suggesting involvement of a different lymphatic endothelial HA receptor, such as HARE. No prior studies have addressed the roles of HA and HARE in lymphangiogenesis, so these mechanisms remain a focus of ongoing study.

Finally, an implication of this work for other sites of prostate cancer metastasis is that HARE expression on any endothelial cell could facilitate the arrest of HA-bearing tumor cells. Metastasis of prostate tumor cells to bone is partly driven by physical arrest in the microvasculature, followed by engagement of receptor-ligand interactions that signal transendothelial migration ⁶. Hematopoietic cells use HA to arrest in the bone marrow sinusoids ¹⁶. In vitro adhesion assays confirm that HA on tumor cells can mediate specific interactions with bone marrow derived endothelial cells ⁷. Additional studies are in progress to evaluate the HARE-mediated model of metastasis to bone and other tissues.

Abbreviations

HA

hyaluronan

HARE

Hyaluronan Receptor for Endocytosis

HAS

hyaluronan synthase

SEC

sinusoidal endothelial cell

LN

lymph node