Cancer esearch oenvironment and Immunology luronan Deficiency in Tumor Stroma Impairs Macrophage R fficking and Tumor Neovascularization

Downlo pite the importance of stromal cells in tumor progression, our overall understanding of the molecular s that regulate the complex cellular interactions within tumor stroma is limited. Here, we provide le lines of evidence that tumor-associated macrophages (TAM) preferentially traffic to stromal areas d within tumors in a manner dependent on a hyaluronan (HA)–rich tumor microenvironment. To s the role of stroma-derived HA in macrophage recruitment, we disrupted the HA synthase 2 (Has2) n stromal fibroblasts using conditional gene targeting. The Has2 null fibroblasts showed severe ment in recruiting macrophages when inoculated with tumor cells into nude mice, which shows the bution of stroma-derived HA in intratumoral macrophage mobilization. Furthermore, a deficiency in al HA attenuated tumor angiogenesis and lymphangiogenesis concomitantly with impaired macrophage tment. Taken together, our results suggest that stromal HA serves as a microenvironmental signal for recrui the recruitment of TAMs, which are key regulatory cells involved in tumor neovascularization. Cancer Res; 70(18); 7073–83. ©2010 AACR.


Introduction
Cancers are not autonomous cells that develop independently but rather are cellular masses formed through dynamic interactions with their microenvironment.As carcinomas evolve, the adjacent stroma undergoes profound alterations in composition to provide important support functions in the early stages of cancer establishment and later to facilitate tumor progression and extravasation (1)(2)(3)(4).The tumor stroma is composed of various host cells, including endothelial cells, fibroblasts, and infiltrating inflammatory cells.Notably, tumor-infiltrating fibroblasts and macrophages, known as tumor-associated fibroblasts (TAF) and tumor-associated macrophages (TAM), respectively, have been increasingly implicated as active participants in tumor progression (5)(6)(7).Several lines of evidence have shown that these cells are exposed to activators within tumor stroma to commence de novo synthesis of a repertoire of growth factors and cytokines.Both TAFs and TAMs have the potential to release both angiogenic and lymphangiogenic growth factors and thereby enhance the formation of tumor vasculatures (8)(9)(10)(11)(12).There is therefore no doubt that stromal cells, via complex interactions with tumor cells, play a critical role in the promotion of tumor malignancy, but relatively little is known about the molecular basis regulating the cellular interactions within tumor stroma.
Inside tumor stroma is a dynamic microenvironment where extracellular matrix (ECM) is extensively remodeled during tumor development and progression (13)(14)(15).Hyaluronan (HA) plays a central role in tumor microenvironment formation as a component of the ECM (16), and the increased HA deposition in the stromal compartments of tumors is often associated with tumor aggressiveness and adverse clinical outcome (17)(18)(19).The biosynthesis of HA, which is critical in establishing its biological function, is regulated by three mammalian HA synthases (Has): Has1, Has2, and Has3 (20).Emerging evidence has shown upregulation of Has gene expression in aggressive and metastatic tumors (21)(22)(23)(24).Furthermore, our animal studies using a conditional transgenic mouse model allowing specific expression of Has2 in mammary tumors (25,26) showed that HA production by tumor cells caused rapid development of aggressive breast carcinomas in association with marked stromal induction.Concurrent with stromal HA accumulation, considerable blood and lymphatic vasculatures had been formed within or near the stroma surrounding tumor cell islets.Increased stromal HA may therefore modulate tumor malignancy by controlling the complex cross-talk among host and tumor cells.
To define the roles of stromal HA in tumor malignancy, we developed Has2 conditional knockout mice and showed that this gene disruption in stromal fibroblasts attenuated trafficking of TAMs to tumor stroma.Along with the reduction of TAM recruitment, a deficiency in stromal HA also suppressed tumor angiogenesis and lymphangiogenesis.On the basis of our current findings, we postulate that stromal HA, via TAM recruitment, remodels the local microenvironment to promote the formation of tumor vasculatures.

Antibodies
The antibodies used in this study are listed in Supplementary Table S1.

Animals
The details of animals used in this study are described in Supplementary Materials and Methods.

Isolation of primary mammary tumor cells and TAFs
Primary mammary tumor cells and TAFs were established from mouse mammary tumors developed spontaneously in Has2 conditional transgenic mice as described previously (26).These cells were verified morphologically and by immunostaining for mesenchymal-specific vimentin and epithelial cell-specific cytokeratin.Both tumor cells and TAFs were subcultured and used within the nine passages.

Tumor transplantation
Mammary tumor cells derived from Has2 ΔNeo , Has2 +Neo , and nontransgenic mice were transplanted together with stromal fibroblasts into the mammary fat pads of nude mice (8-week-old female BALB/c nu/nu mice; SLC) as described previously (26).Control Has2 +Neo tumor cells were also transplanted together with 100 μg of native HA or an equivalent amount of HA-versican aggregates into the mammary fat pads of mice (8-week-old female BALB/c nu/nu mice).Animals were sacrificed at an end point when either tumor size reached a mean diameter of 10 mm or on day 15 after the inoculation of tumor cells.Transplanted tumors were dissected out and weighed at that time.

Histologic and morphometric analyses
Excised tumors were immediately fixed in neutralized 10% formalin or in Tris-buffered zinc fixative, dehydrated, and embedded in paraffin wax.Deparaffinized sections (5-μm thickness) were rehydrated and stained with Azan-Mallory staining.Immunofluorescence staining was carried out using one or more of the following primary antibodies: anti-F4/80 (diluted 1:500) for macrophages, anti-type I collagen (diluted 1:100) for stromal compartments, anti-CD31 (diluted 1:100) for blood vessels, and anti-podoplanin (diluted 1:200) for lymphatic vessels.Some sections were also stained with the biotinylated HA-binding region of aggrecan (b-HABP, Seikagaku Corp.) or with anti-versican antibody (diluted 1:200) to identify HA-rich ECM as described previously (25).Immunolocalization of the antigens was performed using Alexa Fluor-conjugated secondary antibodies under a Leica TCS SP2 AOBS confocal microscope (Leica Microsystems CMS GmbH).Images were exported in TIFF format and analyzed using the Metamorph software (Universal Imaging).The average positive cell density or area per objective field was calculated after immunohistochemical staining.

Preparation of tumor-infiltrating cells and peritoneal exudate macrophages
A single-cell suspension of tumor-infiltrating cells was obtained from tumor transplants as previously described (27).Peritoneal exudate cells (PEC) were obtained from green fluorescent protein (GFP) mice as described previously (28).Fluorescence-activated cell sorting (FACS) analysis revealed the isolated PECs to be 84% of F4/80 + CD11b + monocyte/ macrophage lineage.

FACS analysis
Freshly prepared tumor-infiltrating cells were suspended in cold PBS supplemented with 1% fetal bovine serum.To minimize false positive staining, cell surface Fc receptors were blocked by incubation with 10 μg/mL FcBlock at 4°C for 5 minutes before antibody staining.Tumor-infiltrating cells (1 × 10 6 per sample) were incubated with anti-F4/80 PE-Cy5 and anti-CD206 Alexa Fluor 488 at 4°C for 30 minutes.A total of 100,000 viable cells were analyzed by FACSCaliber (BD Bioscience) with CellQuest software version 5.1.Gates were determined with the use of appropriate isotype controls.Results are given as the positive percentage minus background from appropriate isotype controls.

Selective depletion of macrophages using clodronate liposomes
Clodronate liposomes, termed clodrolip, were prepared as described previously (29).Clodrolip was transplanted into the mammary fat pads of nude mice the day before tumor cell injection.Has2 ΔNeo or Has2 +Neo tumor cell suspension was then injected into the mammary fat pads of nude mice as described above.The mice received additional clodrolip every 5 days during the experimental period (30).Control groups received liposomes containing PBS (PBSlip) at the same time points.The mice were sacrificed on day 15 after the inoculation of tumor cells, and excised tumor transplants were fixed and subjected to histologic examination.

Macrophage adhesion assay
Quantification of macrophage adhesion to fibroblasts was done using PECs isolated from GFP mice as described Research.
on April 13, 2017.© 2010 American Association for Cancer cancerres.aacrjournals.orgDownloaded from Published OnlineFirst September 7, 2010; DOI: 10.1158/0008-5472.CAN-09-4687 previously (31).Fluorescence intensity was measured using a CytoFluor Multiwell Plate Reader Series 4000 (Applied Biosystems) with an excitation of 485 nm and an emission of 530 nm.The number of adherent cells was calculated using a standard curve, and adhesion index was expressed as a percentage of total inoculated cells.All measurements were carried out in triplicate.

Statistical analysis
Statistical analysis was performed using the two-tailed Student's t test.All results were expressed as mean ± SD.A P value of <0.05 was considered to be statistically significant.

TAMs preferentially traffic to stromal compartments formed within HA-producing tumors
To define the molecular basis of the complex cellular interactions within tumor stroma, we first focused our investigation on the effect of a HA-rich tumor microenvironment on the recruitment and function of TAMs.For this purpose, we used a conditional transgenic mouse model allowing specific expression of Has2 in mammary tumors, where Has2 ΔNeo transgenic mice actively produce HA in spontaneous mammary tumors by specifically expressing exogenous Has2, whereas control Has2 +Neo mice, with a silent transgene cassette, do not (25,26).HA overproduction resulted in marked stromal induction and HA accumulation within the stromal compartments of the aggressive tumors (Fig. 1A, top).Concurrent with stromal HA accumulation, considerable blood and lymphatic vasculature had formed within or near the stroma surrounding tumor cell islets (Fig. 1A, top).
Mammary tumor tissues from Has2 ΔNeo and control Has2 +Neo transgenic mice were initially evaluated by immunostaining with the pan-macrophage marker F4/80.F4/80positive (F4/80 + ) macrophages were far more numerous in the stroma of Has2 ΔNeo tumors compared with control Has2 +Neo tumors (Fig. 1A, top), and quantitative analysis showed that the average density of infiltrating F4/80 + macrophages was 7-fold greater in Has2 ΔNeo tumors than in controls (Fig. 1B, left).Notably, macrophages were seen to preferentially traffic to HA-enriched and type I collagenenriched stromal structures in spontaneous Has2 ΔNeo mammary tumors (Fig. 1A, top), which is strongly suggestive of the stromal involvement in macrophage recruitment.To specify the participants responsible for macrophage recruitment, we established primary tumor cells from Has2 ΔNeo and control Has2 +Neo mice and transplanted them with or without TAFs into the mammary fat pads of nude mice.When Has2 ΔNeo tumor cells were inoculated with TAFs, F4/80 + macrophages trafficked to tumor transplants as they did in the spontaneous mammary tumors (Fig. 1A, bottom and B, right).In spite of single inoculation, the macrophages trafficked to a similar degree to stromal structures in Has2 ΔNeo tumors, which may have been due to their high capability of stromal induction.On the other hand, Has2 +Neo tumors induced much less macrophage mobilization (Fig. 1A, bottom and B, right).With TAFs, however, the number of in-filtrating macrophages was markedly increased, suggesting a critical contribution of TAFs to intratumoral macrophage mobilization.
TAMs, which have little cytotoxicity against tumor cells and can actually promote tumor aggressiveness, share a phenotype similar to that of immunosuppressive M2 macrophages (32)(33)(34).To evaluate TAM infiltration into tumor stroma, we examined the M2 macrophage population in both Has2 ΔNeo and Has2 +Neo tumors.FACS analysis revealed that F4/80 + CD206 + M2 macrophages were more abundant in Has2 ΔNeo tumors than in Has2 +Neo tumors (Fig. 1C and D).We further examined the mobilization of F4/80 + total and F4/80 + CD206 + M2 macrophages using tumor cells from nontransgenic mice and found that the control experiment gave similar results to those of Has2 +Neo tumor cells (Fig. 1D).

Selective macrophage depletion reduces tumor angiogenesis and lymphangiogenesis in HA-producing tumors
Recent clinical and experimental studies have indicated that TAMs promote malignant tumor progression through their proangiogenic activity (10,11), so we explored whether TAMs are responsible for angiogenic promotion in HAproducing tumors.Because liposome-encapsulated clodronate (clodrolip) is commonly used for selective macrophage depletion (29, 30), we conducted experiments in which local TAMs were depleted by peritumoral administration of clodrolip.HA-producing Has2 ΔNeo tumor cells were inoculated into the mammary fat pads of nude mice and then evaluated for tumor growth after macrophage depletion.Clodrolip treatment significantly suppressed tumor growth and caused a significant reduction in tumor dry weight compared with PBSlip treatment (Fig. 2A), suggesting a pivotal role of macrophages in growth acceleration of HA-producing tumors.The degree of macrophage depletion was verified by F4/80 staining of transplanted tumors (Fig. 2B and C), showing successful elimination of local TAMs in clodroliptreated tumors.
To investigate whether macrophage depletion had an effect on neovascularization, tumor-associated vasculatures were visualized by immunostaining with an antibody to endothelial cell-specific CD31.In agreement with a previous study (25), the blood vessels in HA-producing Has2 ΔNeo tumors were mostly small and fragmented (Fig. 2B).CD31positive blood vessels were significantly decreased by clodrolip treatment (Fig. 2C).Macrophages in tumor stroma have been suggested to be the major cellular sources of lymphangiogenic growth factors and participate in the promotion of tumor lymphangiogenesis (8), and our previous study also suggested that tumor lymphangiogenesis was closely associated with the formation of tumor stroma in HA-producing tumors (26).Quantitative evaluation of podoplanin-positive areas showed a significant reduction in lymphatic vessels in clodrolip-treated tumors (Fig. 2B and C).
To further investigate the link between HA and macrophage mobilization and neovascularization, a similar experiment was performed using Has2 +Neo tumor cells.Clodrolip treatment suppressed macrophage mobilization and vascular formation in the transplanted tumors but to a lesser extent than those of HA-overproducing Has2 ΔNeo tumors (Fig. 2C), suggesting that HA works synergistically with macrophage recruitment to promote vascular formation.It was also considered that clodrolip treatment may suppress macrophage mobilization and subsequent tumor neovascularization by decreasing HA production.To clarify this possibility, we measured the HA content in Has2 ΔNeo tumors and found a significant reduction after clodrolip treatment (Fig. 2A).HA distribution was also examined with biotinylated HABP (b-HABP), a specific probe for HA.HA staining was detected at the intercellular boundaries of tumor cells and was prominent in the tumor stroma of Has2 ΔNeo tumors (Fig. 2B).Interestingly, HA accumulation had disappeared from the stromal compartments after clodrolip treatment, suggesting that macrophages are crucial for the maintenance of HA-rich stromal structures.On the other hand, macrophages preferentially trafficked to HA-rich stromal compartments (Fig. 1A), which may indicate that a paracrine loop between macrophage mobilization and accumulation of stromal HA is required for sustained expansion of tumor stroma and vasculatures.

Generation of Has1 and Has2 double-knockout mice and establishment of Has-deficient stromal fibroblasts
To define the relative contribution of stromal HA, we explored whether the absence of HA production in stromal fibroblasts, achieved by deleting the Has genes, would abrogate macrophage mobilization and subsequent tumor neovascularization.Our previous report has shown that, of the three Has genes, Has1 and Has2, but not Has3, are endogenously expressed in stromal fibroblasts (26).To ablate the HA synthesis of stromal fibroblasts, Has1 and Has2 doubleknockout mice were obtained by intercrossing Has1-null and Has2 conditional knockout mice, and stromal fibroblast cells were established from founder lineages having a different combination of the two knockout alleles (Supplementary Figs.S1 and S2).
Genomic PCR verified that a part of exon 5 of the Has1 locus was replaced by a neomycin-resistant gene in Has1deficient fibroblasts (Supplementary Fig. S2A).To selectively suppress HA synthesis in stromal fibroblasts, the Has2 floxed allele was excised in vitro by infecting Has2 flox/flox fibroblasts with a recombinant adenovirus carrying the Cre recombinase gene.As a result, exon 2 of the Has2 locus was effectively deleted by Cre-mediated recombination in Has2 flox/flox stromal fibroblasts (Supplementary Fig. S2A).In contrast, an intact floxed allele remained in Has2 flox/flox fibroblasts after infection of a control virus carrying the LacZ gene.Real-time quantitative reverse transcription-PCR confirmed that the expression of intact Has1 transcripts was lost in Has1-deficient fibroblasts and that Has2 expression was greatly diminished by Cre-mediated recombination in Has2 flox/flox fibroblasts (Supplementary Fig. S2B).Deletion of a single allele, however, did not change Has2 expression, suggesting the possibility that Has2 gene expression is compensated for by an unknown mechanism to maintain a constant level.We then compared HA content in the conditioned medium of stromal fibroblasts and found that deletion of Has1 gene did not affect HA synthesis in stromal fibroblasts (Supplementary Fig. S2B).In contrast, Cremediated recombination of Has2 locus induced an almost complete failure in HA synthesis.The current results, together with those of our previous study showing that Has2 is more catalytically stable than Has1 (35), suggest that Has2 accounts for most of the HA production from stromal fibroblasts.HA staining revealed that no cablelike ECM structures of HA could be detected in Has2deficient (Has2 Δ/Δ ) fibroblasts, but such structures could be seen in wild-type and Has2 flox/flox fibroblasts (Supplementary Fig. S2C).

A deficiency in stromal HA impairs macrophage trafficking
To test whether a deficiency in stromal HA influences macrophage trafficking, Has2 +Neo tumor cells were inoculated together with Has-deficient stromal fibroblasts into the mammary fat pads of nude mice.The generated transplants were then processed for FACS analysis, which showed that F4/80 + macrophages trafficked as expected to the tumor transplants generated by coinoculation of Has2 +Neo tumor cells and Has2 flox/flox fibroblasts (termed Has2 flox/flox tumors; Fig. 3A and B).Heterozygous mutation of the Has2 allele (Has2 +/Δ ) had a similar effect on macrophage trafficking.However, a significantly lower number of macrophages was recorded in tumor transplants composed of Has2 +Neo tumor cells and Has2 Δ/Δ fibroblasts (termed Has2 Δ/Δ tumors).A decrease in the M2 population was also seen in Has2 Δ/Δ Figure 1.TAM trafficking toward tumor stroma.A, histopathologic and immunofluorescence analyses of spontaneous (top) and transplanted (bottom) mammary tumors.Azan-Mallory staining of stromal compartments formed within tumors.Scale bar, 100 μm.Tumor sections were stained with HA-specific b-HABP (red), anti-CD31 antibody (green) for blood vessels, anti-podoplanin antibody (red) for lymphatic vessels, anti-F4/80 antibody (green) for macrophages, and anti-type I collagen (Col.I) antibody (red) for tumor stroma.Cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; blue).Scale bars, 40 μm.For transplantation experiments, Has2 ΔNeo and Has2 +Neo tumor cells (termed Has2 ΔNeo T and Has2 +Neo T, respectively) were transplanted with (+) or without (-) TAFs into the mammary fat pads of nude mice, and macrophage recruitment was evaluated in transplanted tumors.B, quantitative analyses of macrophage infiltration in spontaneous (left) and transplanted tumors (right).The combinations of tumor cells and TAFs are described under the graphs.Sections from each tumor were immunostained with anti-F4/80 antibody, and the average positive cell density per objective field was calculated.Values represent the mean ± SD of five random fields from seven tumors per group.**, P < 0.01.NS, not significant.C, FACS analysis of macrophages mobilized into transplanted tumors.Cell suspensions from transplanted tumors were stained with Alexa Fluor 488-CD206 and PE-Cy5-F4/80 monoclonal antibodies and then subjected to FACS analysis.The total number of macrophages was determined by counting cell number in gates.M2 macrophages were localized on the right half of the circle.D, quantitative analyses of F4/80 + total macrophages (left) and F4/80 + CD206 + M2 macrophages (right).Tumor cells from nontransgenic mice were also analyzed for macrophage mobilization as controls.Graphs represent the mean ± SD from five tumors per experimental group.**, P < 0.01.

Research.
on April 13, 2017.© 2010 American Association for Cancer cancerres.aacrjournals.orgDownloaded from Published OnlineFirst September 7, 2010; DOI: 10.1158/0008-5472.CAN-09-4687 tumors (Fig. 3A and B), which was supportive of the pivotal role of Has2-derived stromal HA in intratumoral TAM mobilization.As the influence of a Has1 deficiency was marginal on macrophage trafficking (data not shown), Has1 null fibro-blasts having different combinations of Has2 knockout alleles were used for further experiments.
We next sought to address exactly how the absence of HA synthesis in stromal fibroblasts attenuated macrophage trafficking.Because recent studies have suggested that cablelike HA structures are primarily responsible for the binding of mononuclear leukocytes (31,36), stromal HA might mobilize macrophages into tumors through direct interaction.We thus examined macrophage adhesion to Has2-deficient fibroblasts to verify the role of stromal HA in macrophage trafficking.When PECs were seeded onto monolayers of stromal fibroblasts having different combinations of the Has2 knockout alleles, the adhesion rate of PECs was almost 2-fold less to Has2 Δ/Δ fibroblast monolayers compared with that of Has2 flox/flox fibroblasts (Fig. 3C), which was identical to the results obtained from the in vivo transplant study.

A deficiency in stromal HA attenuated tumor angiogenesis and lymphangiogenesis
We next investigated whether an HA deficiency in stroma had an effect on tumor neovascularization.In accordance with the massive formation of HA-rich stroma, Has2 flox/flox fibroblasts gave rise to highly vascularized tumors when coinoculated with Has2 +Neo tumor cells (Fig. 4A).In morphometric analyses of CD31-positive blood vessels, the mean vessel density was over 20 vessels per field in Has2 flox/flox tumors (Fig. 4B).Conversely, CD31-positive microvessels were significantly fewer in Has2 Δ/Δ tumors than in control Has2 flox/flox tumors.Immunofluorescence staining of podoplanin was Top, sections from tumors treated with clodrolip were immunostained with anti-F4/80 antibody (green) to identify infiltrating macrophages.Middle, blood (arrowheads) and lymphatic vessels were visualized by immunostaining with anti-CD31 antibody (green) and anti-podoplanin antibody (red), respectively.Bottom, the distribution of HA was evaluated by staining with b-HABP (red).Cell nuclei were counterstained with DAPI (blue).Scale bars, 40 μm.C, quantitative analysis of macrophage density and morphometric analyses of blood and lymphatic vessels.Has2 ΔNeo or Has2 +Neo tumor cells were transplanted into nude mice, and the animals were treated with clodrolip (white columns) or control PBSlip (black columns).Graphs represent the mean ± SD from more than six tumors per experimental group (at least three sections and five fields were counted for each tumor).**, P < 0.01.also carried out on tumor transplants, where attenuated expression of podoplanin in Has2 Δ/Δ tumors confirmed that a Has2 deficiency resulted in impaired tumor lymphangiogenesis (Fig. 4A and B).

The cooperative effect of HA and versican on macrophage trafficking
Because a wide variety of HA-binding molecules contribute to the assembly of pericellular HA-ECM and tightly regulate HA functions, we hypothesized that such molecules, when produced by stromal fibroblasts, can modulate stromal HA to enhance macrophage recruitment and subsequent neovascularization.Versican, a member of the HA-binding chondroitin sulfate proteoglycan family, is important for the stabilization of pericellular HA coats.Consistent with the previous observation that versican is enriched in stromal compartments (25), we found that this molecule was produced mainly by TAFs and not by tumor cells (Fig. 5A).As shown in Fig. 5B, a deficiency in HA synthesis resulted in the destruction of HA-rich and versican-rich pericellular ECM, and only small pieces of versican cables were detected on the surface of Has2-deficient stromal fibroblasts despite versican levels being unaffected by an HA deficiency (data not shown).The deformed cable-like structures were partially restored by the addition of exogenous HMW-HA (Fig. 5B).
In the final set of experiments, we investigated whether versican exerted cooperative effects with HMW-HA in terms of potentiation of macrophage trafficking.Has2 +Neo tumor cells were implanted together with exogenous HMW-HA, versican, or their aggregates into the mammary fat pads of nude mice.Administration of HA or versican had little effect on tumor growth or the promotion of macrophage recruitment in the transplanted model (Fig. 5C and D).In contrast, HA-versican aggregates accelerated the mobilization of both F4/80 + total and F4/80 + CD206 + M2 macrophages as well as tumor growth (Fig. 5C and D).Histologic staining of HA and versican showed that these were prominent in the tumor stroma and were accompanied with macrophages (Supplementary Fig. S3).Together with our previous notion that HA-versican aggregates accelerate angiogenesis and lymphangiogenesis (25,26), all of the above data provide evidence that a HA-rich stromal microenvironment is crucial for TAM trafficking and subsequent tumor neovascularization.

Discussion
The present work on the effect of stromal HA on tumor progression has helped to identify the molecular basis of TAM mobilization toward stromal compartments within malignant tumors.To our knowledge, this is the first report showing the HA-dependent interactions between TAMs and stromal fibroblasts to enhance tumor neovascularization by modulation of the local stromal microenvironment.
HA accumulation varies considerably depending on its origin and the histologic type of the cancer.Clinically, increased deposition of HA in stromal compartments within tumors often correlates with tumor aggressiveness and adverse clinical outcome (17)(18)(19), and HA has been frequently observed at elevated levels around actively migrating and proliferating tumor cells (37,38).For these reasons, clarifying the functional aspects of HA on tumor cell surfaces and in the stroma constitutes an indispensable step toward our overall understanding of the roles of a HA-rich tumor microenvironment in tumor progression.Our knockout experiments clearly showed that HA species released by stromal fibroblasts participate in TAM trafficking into tumor masses, thereby boosting tumor neovascularization.Because a wide variety of HA binding molecules contribute to the assembly of pericellular HA-ECM and tightly regulate HA functions (39), the diverging properties of tumoral and stromal HA species may reflect the wide spectrum of molecules constituting HA-ECM.In the present study, we showed that HA and versican derived from stromal fibroblasts promoted macrophage mobilization in a cooperative fashion.Although our work did not fully define the functional diversity of tumoral and stromal HA species, the presence of versican may account for the specific action of stromal HA.
HA is an important ligand for leukocyte recruitment and retention (40,41), and recent studies suggest that cable-like HA structures are primarily responsible for the binding of mononuclear leukocytes (31,36).Accordingly, the HA-rich tumor microenvironment may serve as a suitable scaffold for macrophage trafficking.The current study suggests at least a partial involvement of HA-enriched and versican-enriched pericellular ECM in macrophage recruitment.Consistent with our observations, Potter-Perigo and colleagues have also recently reported an important role of versican in the HA-dependent binding of monocytes to the ECM of lung fibroblasts stimulated with poly I:C (42).Additionally, Kim and colleagues showed that versican activates macrophages through the activation of the Toll-like receptor (TLR) family member TLR2 (43).Thus, stromal HA may indirectly activate and recruit TAMs by In conclusion, our results defined the molecular basis for TAM recruitment and provided evidence that stromal HA, via macrophage recruitment, remodels the local microenvironment to promote the formation of tumor vasculatures.A more detailed understanding of the mechanisms regulating cellular interactions within the HA-rich tumor microenvironment will further contribute to the development of novel therapeutic strategies for the prevention of tumor neovascularization.

Figure 2 .
Figure 2. Macrophage depletion affects tumor growth and neovascularization.A, macroscopic appearance and growth curve of Has2 ΔNeo tumors treated with clodrolip and control PBSlip.Clodrolip and PBSlip were given the day before the injection of Has2 ΔNeo tumor cells and every 5 d thereafter.Animals were sacrificed at the end point, and the transplanted tumors were excised and evaluated for tumor growth, tumor weight, and HA content.Scale bar, 10 mm.Graphs represent the mean ± SD (PBSlip, n = 6; clodrolip, n = 8).*, P < 0.05; **, P < 0.01.B, photomicrographs of immunofluorescence images.Top, sections from tumors treated with clodrolip were immunostained with anti-F4/80 antibody (green) to identify infiltrating macrophages.Middle, blood (arrowheads) and lymphatic vessels were visualized by immunostaining with anti-CD31 antibody (green) and anti-podoplanin antibody (red), respectively.Bottom, the distribution of HA was evaluated by staining with b-HABP (red).Cell nuclei were counterstained with DAPI (blue).Scale bars, 40 μm.C, quantitative analysis of macrophage density and morphometric analyses of blood and lymphatic vessels.Has2 ΔNeo or Has2 +Neo tumor cells were transplanted into nude mice, and the animals were treated with clodrolip (white columns) or control PBSlip (black columns).Graphs represent the mean ± SD from more than six tumors per experimental group (at least three sections and five fields were counted for each tumor).**, P < 0.01.

Figure 3 .
Figure 3. Deficiency of stromal HA impairs macrophage trafficking.A, FACS analysis of macrophages mobilized into transplanted tumors.Has2 ΔNeo or Has2 +Neo tumor cells were inoculated into the mammary fat pads of nude mice (top).Has2 +Neo tumor cells were inoculated with Has2 flox/flox or Has2 Δ/Δ stromal fibroblasts into the mammary fat pads of nude mice (bottom).B, quantitative analyses of F4/80 + total macrophages (top) and F4/80 + CD206 + M2 macrophages (bottom).The combinations of tumor cells and stromal fibroblasts are described under the graphs.The genotypes of fibroblast clones are as described in Supplementary Fig. S2.Data represent the mean ± SD of five tumors for each group.**, P < 0.01.C, adhesion index of PECs to confluent monolayers of fibroblasts having different combinations of the Has2 knockout alleles.The adhesion index was compared between Ad-Cre and control Ad-LacZ infectants.Data represent the mean ± SD of three independent experiments.**, P < 0.01.

Figure 4 .
Figure 4.A deficiency in stromal HA impairs tumor neovascularization.A, HA localization and tumor neovascularization in Has2 flox/flox and Has2 Δ/Δ tumors.HA distribution was evaluated by staining with b-HABP (red).Blood (arrowheads) and lymphatic vessels were visualized by immunostaining with anti-CD31 antibody (green) and anti-podoplanin antibody (red), respectively.Scale bars, 40 μm.B, morphometric analyses of blood vessels and lymphatic vessels formed in transplanted tumors.The combinations of tumor cells and stromal fibroblasts are described under the graphs.The genotypes of fibroblast clones were as described in Supplementary Fig. S2.Graphs represent the mean ± SD from five tumors per experimental group (at least three sections and five fields were counted for each tumor).**, P < 0.01.

Figure 5 .
Figure 5.The cooperative effect of HA and versican on macrophage trafficking.A, dot blot analysis and quantification of versican in the culture medium of tumor cells and TAFs.B, immunofluorescence images of the pericellular ECM of cultured stromal fibroblasts.HA and versican were visualized by staining with b-HABP (red) and anti-versican antibody (green), respectively (top).Cell nuclei were counterstained with DAPI (blue).Has-deficient (Has1 −/− Has2 Δ/Δ ) stromal fibroblasts were incubated with different concentrations of exogenous HMW-HA for 3 d, and the formation of HA-rich pericellular ECM (arrowheads) was examined (bottom).Scale bars, 40 μm.C, effects of exogenous HA and versican on the growth of transplanted tumors.Has2 +Neo tumor cells were implanted together with exogenous HMW-HA, versican, or their aggregates into the mammary fat pads of nude mice.Excised tumors were evaluated for tumor volume and weight.Data represent the mean ± SD from five tumors per experimental group.**, P < 0.01.D, quantitative analyses of F4/80 + total macrophages (black columns) and F4/80 + CD206 + M2 macrophages (white columns).The number of macrophages that had mobilized into transplanted tumors was determined by FACS analysis.Data represent the mean ± SD from five tumors per experimental group.†, P < 0.05; † †, P < 0.01; **, P < 0.01.