Host-derived osteopontin maintains an acute inflammatory response to suppress early progression of extrinsic cancer cells

Abstract

The matricellular protein osteopontin (OPN), expressed in various cancer types and elevated in the blood of cancer patients, is thought to have different functions when derived from host versus cancer cells. To assess the effect of host-derived OPN on growth of cancers of epithelial origin, we established a line of cutaneous squamous cell carcinoma (SCC) cells, named ONSC, that lacks the OPN gene and develops SCC in syngeneic wild-type (WT) and OPN-null mice. At 8 and/or 10 wk after subcutaneous injection of ONSC cells in mice, however, there was a lower tumor incidence in WT mice, suggesting that host-derived OPN is associated with suppression of early growth of extrinsic cancer cells. Histological, immunohistochemical, biochemical, and hematological analyses were performed on the tumor microenvironment and blood from tumor-bearing mice during the first weeks after implantation. Host-derived OPN suppression of extrinsic ONSC cell progression is likely mediated through elicitation of an early innate inflammatory response, through its function as a chemoattractant and/or by enhancing survival of inflammatory cells. Further, consistent with a previous report, the serum levels of host-derived OPN, which are elevated during the early phase of tumor growth in mice implanted with ONSC, appear to reflect an anti-tumor progression effect.

Introduction

The survival and growth of malignant cells are influenced by their intrinsic characteristics and through their interaction with the surrounding stroma/matrix environment.¹ In an immune-competent environment, cancer cells are exposed to immune/inflammatory cells, fibroblasts, endothelial cells, and/or adipocytes; as a result, they respond to factors secreted by these host cells and to autocrine signaling by the cancer cells themselves.

Accumulating evidence indicates that osteopontin (OPN), a matricellular protein found in the tumor microenvironment and expressed by host and cancer cells, may regulate tumorigenesis, cancer progression, and metastasis.^2–5 This secreted, acidic glycoprotein interacts with integrins and CD44 variants leading to cellular signaling, which, in the context of cancer, is postulated to modulate cancer cell migration, adhesion, and survival.⁴

There is elevated expression of OPN in cancers of the skin, brain, breast, lung, liver, colon, and prostate. ^{2, 6–13} Additionally, blood levels of OPN are high in many of these cancer patients, and the levels appear to correlate with late-stage cancer metastasis and with poor survival,^14–17 suggesting that OPN is involved in cancer progression and metastasis.^{18, 19}

Although studies with OPN-expressing cancer cells support the role of tumor-derived OPN in cancer progression, in situ hybridization and immunohistochemical analyses on various types of cancers indicate that OPN is not always expressed by cancer cells. Instead, macrophages in the surrounding stroma express OPN,^{6, 20, 21} suggesting that host-derived OPN modulates cancer progression. Moreover, evaluation of cancer progression in one-step carcinogen-induced skin cancers in OPN-null and wild-type (WT) mice suggests that there are distinct functions of OPN derived from host and cancer cells.⁵

Elucidation of the function of host-derived OPN in the progression of cutaneous squamous cell carcinomas (SCCs) has been hindered due to the lack of an appropriate mouse model. Since SCCs that develop in WT mice exposed to a carcinogen express OPN,⁵ it is difficult to determine the function of host-derived OPN in cancer progression. Spindle cell carcinoma cells with an ablated OPN gene develop tumors in nude mice,⁵ but these cannot be used to address the host OPN response in immune-competent mice. To circumvent these issues, we established a murine cutaneous OPN-null SCC cell line (ONSC) with p53 and H-ras mutations, which are common characteristics of SCCs. In athymic nude mice, ONSC cells develop into SCCs that metastasize.²²

In the present study, we show that ONSC cells can also develop into SCCs in syngeneic wild-type (WT) and OPN-null mice. Thus, this model, involving immune-competent mice, allows evaluation of the function of host-derived OPN on extrinsic SCC progression, both in the tumor microenvironment and at the systemic level. ONSC cells injected subcutaneously (s.c.) develop SCC more efficiently in OPN-null mice than in WT mice, suggesting that host OPN suppresses early growth of these extrinsic cancer cells. Data from histological, immunohistochemical, biochemical, and hematological studies indicate that host-derived OPN is necessary for maintaining an early inflammatory response in the tumor microenvironment, thereby suppressing extrinsic tumor growth and leading to a lower tumor incidence in WT mice. Furthermore, consistent with a previous study,²³ the early elevation of serum OPN, seen in mice harboring ONSC, is likely contributed by activated, circulating inflammatory cells and is associated with an anti-progression effect on extrinsic cancer cells.

Materials and Methods

Mice and ONSC cell line

The OPN-null^{3, 24} and WT mice in 129S6/SvEv background were bred in-house. All experiments were conducted according to Institutional Animal Care and Use Committee guidelines at the University of Alabama at Birmingham’s animal facility.

ONSC cells were derived from an SCC that developed in a female OPN-null mouse subjected to two-stage skin carcinogenesis.²² The cells were maintained in high-glucose, low-calcium DMEM containing 10% fetal bovine serum. Experiments were performed with cells of passages 35–47 and routinely checked to avoid mycoplasma contamination by a previously described procedure.²⁵

ONSC growth in WT and OPN-null mice

OPN-null and WT female mice at 7–8 wk of age were injected s.c. in the lower dorsal region with 2 × 10⁶ ONSC cells in 100 μl of PBS (one injection per mouse). Tumor size was measured with skin calipers (twice per region) weekly and calculated with the formula: V = ab²/2, where V is volume (mm³); a is the maximum tumor diameter; and b is the minimum tumor diameter.³

Histopathological and immunohistochemical analyses

Histopathological analyses of sections stained with hematoxylin and eosin (H&E) were performed independently by a veterinary anatomic (MMJ) and a human anatomic (KJH) pathologist. For immunohistochemical analyses, 5-μm tumor sections were treated as previously described ³. Briefly, serial sections were incubated with antibodies directed to macrophage (rat anti-mouse F4/80, eBioscience, San Diego, CA), OPN (AKm2A1, Santa Cruz Biotechnology, Santa Cruz, CA), K14 or their respective isotype controls, rat IgG2a, IgG₁ and IgG₃, respectively. Visualization of positive stains was determined by the addition of 3′, 3′-diaminobenzidine tetrahydrochloride. Slides were counterstained with hematoxylin and coverslipped. For double staining, slides stained with antibody to macrophages were incubated with antibody to OPN followed by polymer-alkaline phosphatase and development with Warp Red. Photographs were taken of sections with a digital Nikon camera (Coolpix 4500) attached to Olympus CX31 microscope.

Quantitation of ONSC and apoptotic cells

The average numbers of cancer cells per mouse were represented by pixel numbers obtained from morphometric analyses of K14 stained slides (see figure 2B). Apoptotic cells in tumor sections were determined by terminal deoxynucleotide transferase-mediated dUTP-biotin nick end labeling (TUNEL) assay according to procedure of Roche Diagnostics (Indianapolis, IN).³ The average numbers of apoptotic signals around the periphery of the necrotic region per/tumor section and averaged.

Figure 2.

Histology and number of cancer cells in the tumor microenvironment of WT and OPN-null mice injected with ONSC. (A) Representative tumor sections stained with H&E in three WT (top panels) and three OPN-null mice (lower panels) at 1 wk after ONSC injection. Magnification: panels a and h, ×4; panels b, c, d, e, f, and g, and insets, ×10. *, ONSC cells; N, necrotic (B) Dot plot of average pixel numbers correlating the numbers of ONSC cells/section/mouse from tumor sections stained with K14 and hematoxylin and analyzed by morphometry. Wk 1, WT and OPN-null, n = 8; wk 2, WT, n= 9 and OPN-null, n= 7.

Determination of OPN, cytokines, and chemokines from tumor-conditioned media and sera

Tumors obtained at 1 and 2 wk after injection of ONSC cells into WT and OPN-null mice were minced and placed in culture dishes with serum-free media. Tumor-conditioned media were collected after 18 h, centrifuged, and stored at −80 °C until analysis for cytokines and chemokines by Rules-Based Medicine, Inc (Austin, TX). OPN from tumor-conditioned media was determined by ELISA (Assay Design Inc., Ann Arbor, MI). The values for OPN, cytokines, and chemokines were normalized to protein concentrations and tumor sizes. Protein concentrations were determined by the bicinchoninic acid assay, and tumor size was determined as described above. For determination of sera OPN, the blood samples were incubated with protease inhibitors to prevent thrombin cleavage of OPN, prior to sera isolation. Serum OPN and cytokine levels (TNFα, IL-6, and IL-10) were determined by ELISA (for cytokine assay, eBiosciences, San Diego, CA) and normalized to serum protein concentrations.

Isolation and treatment of peritoneal exudate cells (PECs)

These cells were harvested from WT and OPN-null 7–8 wk-old female mice. Briefly, mice were sacrificed, and their abdomens were flushed with 3 ml of RPMI 1640 medium. Cells were collected from the peritoneal lavage fluid by centrifugation and resuspended in RPMI containing 5% FBS. Cells were seeded at 3×10⁵/well in 24-well plates overnight, rinsed, and treated with PBS or LPS (15 μg/ml), quadruplicate wells per treatment for 24 h. The conditioned media were collected and cytokines were assessed by ELISA. PECs from control and tumor-bearing mice were performed similarly as above except with no LPS treatment.

Hematology

EDTA plasma samples obtained from OPN-null and WT mice without or with ONSC tumors at 1 wk after inoculation were analyzed by use of a VetScan HMII® (Abaxis, Union City, CA). For each analysis, 25 μl of plasma was used.

Statistical analyses

Results were expressed as mean ± SD. Statistical evaluations of the variables for studies in vivo were performed by ANOVA, Student’s t-test, or Fisher’s exact probability test. Statistical significance was defined as p<0.05. To compare the values for number of cancer cells and for cytokines and chemokines from the tumor-conditioned media between WT and OPN-null mice, Statistical Analyses SAS R Proprietary Software 9.2 (TS1M0) was used to perform the exact Wilcoxon test. The one-sided p-value is 0.05, assuming a 5% overlap.

Results

ONSC cells develop into SCCs with a greater incidence in OPN-null mice

To determine if ONSC cells develop into SCCs in syngeneic mice in addition to nude mice as previously shown²², we first assessed tumor incidence at 8 and/or 10 wk after cells were injected s.c. into OPN-null and WT mice. Although SCCs developed in both groups, the incidence of SCCs, was significantly lower in WT (average 70%) than in OPN-null mice (100%), p< 0.05, n = 50 per group. Histopathology of tumors at 10 wk showed well-differentiated SCCs in both types of mice (Fig. 1A).

Figure 1.

Tumor histology and tumor growth curve in WT and OPN-null mice injected with ONSC cells. (A) Histology of SCCs at 10 wk derived from injection of ONSC cells into WT and OPN-null mice. Arrow head, SCC; arrow, keratin pearl. Magnification, ×10 (inset, ×40).

(B) Tumor growth curve (1–10 wk) of WT and OPN-null mice injected s.c. with ONSC cells into the lower dorsal regions (single injection per mouse). Only measurable tumors are plotted. OPN, n = 3/wk; WT, n = 5–6 mice/wk. (C) Injection site volume from 1–4 wk was measured each week. Data shown represent mean ± SD, n = 5~6/group. OPN+/+, WT; OPN−/−, OPN-null. Figures A, B, and C represent independent experiments.

A separate study on tumor growth starting from injection to 10 wk of both WT and OPN-null mice indicated that the tumors in OPN null mice were significantly larger than those of WT mice at 7–9 wk (Fig. 1B). Since the lower tumor incidence in WT compared to OPN-null mice could be partially due to a differential response in the microenvironment during the early period of extrinsic ONSC cells growth, an independent study was performed to examine the volume at the injection site during the first 4 wk. We observed an initial increase in volume at the injection site (wk 1) followed by decreasing volume at wk 1–3, which is a tissue reaction to extrinsic ONSC cells, resulting in edema and inflammation that gradually subsides (Fig. 1C). At wk 4, this decrease in volume continued in WT mice; in contrast, those in OPN-null mice were significantly larger (p<0.04) than those in WT mice (Fig. 1C).

In the early phase of tumor growth, ONSC cells survive and proliferate better in the OPN-null environment compared to that of WT mice

To better understand the changes in the microenvironment of the injection site (designate as tumor section), tumors obtained at wk 1–4 after injection of ONSC into WT and OPN-null mice were stained with H&E, and histopathological analyses were performed. The dermal region of the injection site in both groups of mice contained a central area with necrotic cells (Fig. 2A). In the connective tissue around the periphery of the necrotic region, however, there were significantly more ONSC cells in the OPN-null mice at wk 1, Fig. 2A and 2B. By wk 2, the number of ONSC cells in both group were increased with more ONSC cells on the average in OPN-null mice compared to WT mice, however, there were no significant differences, Fig. 2B.

Compared to WT mice, the lack of OPN expression in OPN-null mice is associated with lower cytokine and chemokine expression in the tumor microenvironment and is indicative of a weak inflammatory response

To determine why there were significantly fewer ONSC cells present in the WT tumor sections compared to OPN-null at wk 1, we determined whether there were differences in the levels of cytokines and chemokines, which are likely expressed by infiltrating neutrophils and macrophages during the inflammatory responses observed in the early weeks, Fig. 1C. In the tumor-conditioned media from ONSC tumors of WT mice, the inflammatory cytokine IL-6 was significantly (p<0.05) higher in the first 2 wk relative to values for OPN-null group (Fig. 3A). IL-10 levels were also significantly higher at wk 1 in WT compared to that of OPN-null group (Fig. 3B); there were no apparent differences in the levels of TNF-α and IL-1β (Fig. 3C and E, respectively). Nevertheless, for both weeks, the levels of IL-1α in the tumor-conditioned media were significantly higher for OPN-null compared to WT mice (Fig. 3D). Furthermore, the cytokines IL-12p70, INFγ and IL-17 were not detectable in the tumor-conditioned media of OPN and WT mice (data not shown).

Figure 3.

Cytokine and chemokine levels in conditioned media of tumors from OPN-null and WT mice. Tumors grown for 1 or 2 wk in WT and OPN-null mice were measured, excised, and then cultured in serum-free medium for 18 h. Tumor-conditioned media were analyzed for cytokines (A–E, IL-6, IL-10, TNFα, IL-1α, and IL-1β, respectively) and chemokines (F–M, MCP-1, MCP-3, MCP-5, MIP-1α, MIP-1β, MIP-1γ, MIP-2, and MIP-3β, respectively). Values were normalized to protein concentrations and tumor sizes (see Materials and Methods). Each dot or circle represents one mouse. WT, wk 1, n = 5; wk 2, n = 5; OPN-null, wk 1, n = 3; wk 2, n = 5. p<0.05, significantly different.

The weaker cytokine responses in OPN-null mice, compared to WT mice, were associated with significantly lower production of chemokines, such as MCP-1, at 1 and 2 wks (Fig. 3F), and of MIP-1γ and MIP-3β at wk 2 (Fig. 3K and M, respectively). In contrast, there were no significant differences between the two groups in the levels of MCP-3, MCP-5, MIP-1α, MIP-1β, and MIP-2 (Fig. 3G–J, and L, respectively). Thus, the generally lower levels of cytokines and chemokines in the OPN-null tumor microenvironment compared to WT mice suggest that there is a weak inflammatory response in OPN-null mice, which provides a favorable environment for SCC development.

OPN-null mice retain the ability to trigger inflammatory responses

To determine if the weak inflammatory tumor microenvironment in the OPN-null mice might be due to their inability to trigger inflammatory response, two experiments were performed. In the first experiment, we tested the cytokine expression of isolated peritoneal exudates cells (PECs) upon treatment with lipopolysaccharide (LPS). Data showed that LPS-treated OPN-null PECs stimulated TNFα and suppressed IL-10 expression as well as WT cells, Fig. S1 A. Interestingly, there were significant differences in both the basal level and LPS-induced expression of IL-6 in OPN-null PECs compared to WT cells.

Since the first experiment was an in vitro response of PECs to LPS, the second experiment was performed to determine whether PECs from tumor-bearing OPN-null mice will result in induced-IL-6 expression compared to non-tumor-bearing mice. Conditioned media from PECs of tumor-bearing and control mice from both groups were analyzed. Data showed that OPN-null PECs from tumor-bearing mice were able to express IL-6 as well as WT PECs, but the basal level of IL-6 expressed by OPN-null PECs is significantly lower compared to WT mice, Fig. S1 B. Thus, these data demonstrate that although the basal level of IL-6 expressed by PECs of OPN-null mice is lower than those of WT, these mice retain the ability to trigger inflammatory cytokines in tumor-bearing mice.

At 2 wk, more apoptotic cells were present in the tumors of WT mice compared to OPN-null mice

Since a more robust inflammatory response in the tumor microenvironment could lead to cell apoptosis, TUNEL analyses of the tumor sections were performed. At 1 wk, the numbers of apoptotic cells in the non-necrotic regions of the tumor were not different between OPN-null and WT mice. At wk 2, there were significantly fewer apoptotic cells in both groups, compared to wk 1 (Fig. 4A), but there were significantly more apoptotic cells present in tumors of the WT compared to OPN-null group. Similar results were observed at wk 3, but the numbers of apoptotic cells were not different between groups.

Figure 4.

Numbers of apoptotic cells and immunohistochemical analyses of macrophages in tumors of WT and OPN-null mice. (A) The histogram shows the average number of apoptotic cells in ONSC tumor/field/section or mouse as analyzed by the TUNEL assay. Only apoptotic signals in the connective tissue around the periphery of the necrotic regions of tumor sections from WT or OPN-null mice were scored. 40x field, n = 5 mice/group. (B) Parallel tumor sections were stained with an antibody to mouse macrophages, F4/80 (lower panels), or IgG2a,κ (top panels), N, necrotic; CT, connective tissue region where ONSC cells are located in most of the OPN-null tumor sections and in some WT tumor sections; □, the location of the image in the small panels. Magnification: panels a, and e, ×4; panels b, c, f and g, ×100; insets- macrophages, ×200; panels d and h, ×10).

The numbers of infiltrating macrophages, but not neutrophils, in the tumor microenvironment are significantly different for WT and OPN-null mice at wk 1

The differences between the levels of various cytokines and chemokines from the conditioned media of tumors from WT and OPN-null mice suggested differences in the number of neutrophils and macrophages in the tumor microenvironments of these two groups. The small size of the tumors at 1 and 2 wk (ranging from < 20 to 80 mm³, see Fig. 1C), and their large central necrotic region, did not permit the use of flow cytometry to identify various cell populations. As an alternative method, the numbers of neutrophils and macrophages were scored by histopathological examination and immunohistochemical analyses, respectively. At 1 wk, there were no differences between the two groups in the numbers of neutrophils in ONSC tumor sections, Table 1.

Table 1.

Average numbers of neutrophils, macrophages and OPN expressing cells in the tumors of WT and OPN-null mice and percent of macrophage expressing OPN

Cell type	Wk post ONSC injection	WT	OPN-null
Neutrophils	1	160 ± 791 (n=5)	132 ± 79 (n=5)
Macrophages	1	43.1 ± 9.02 (n=5)	27.9 ± 4.7* (n=5)
	2	46.4 ± 10.2 (n=3)	55.5 ± 6.1** (n=4)
Cells expressing OPN	2	45.1 ± 6.63 (n=3)
% of Macrophage expressing OPN	2	45.4 ± 8.64 (n=3)

¹To determine the numbers of neutrophils, four fields at 40X magnification were selected per slide/mouse, and the scored values were averaged.

²To determine the numbers of infiltrating macrophages, tumor sections were stained with the antibody F4/80. Photos of 5 to 15 fields (depending on the tumor section) at 40X magnification were taken per slide/mouse for enumeration of these cells by two individuals blinding to the identification of the sections. Their numbers were recorded as the averages of macrophages per field/mouse.

³To determine the numbers of OPN expressing cells, tumor sections were double stained with antibodies direct to macrophage and OPN. Four fields at 40X magnification were selected per slide/mouse, and cells stained positive for OPN were scored and values were averaged/field/mouse.

^*p<0.01, significantly different from WT at 1 wk

^**p<0.001, significantly different from OPN-null at 1 wk.

Immunohistochemical staining, however, showed differences in the localization of macrophages between tumors of WT versus OPN-null mice. Because there were more SCC cells at 1 wk for OPN-null compared to WT mice, and the macrophages in the OPN-null tumor section were localized around the periphery of the SCC region (Fig. 4B). In WT mice, where tumor cells were less prominent, more macrophages were concentrated in the connective tissues around the necrotic region. Scoring of macrophages in the tumors of OPN-null mice indicated that there was a significant decrease (p<0.001) in the average number of macrophages at 1 wk compared to 2 wk, Table 1. Further, compared to the average numbers of macrophages in tumors of WT mice, there was a significant decrease (p<0.01) at 1 wk, but not at 2 wk, in OPN-null tumors, Table 1. This suggests that, at wk 1, there is a delay in infiltration of macrophages into the tumor microenvironment of OPN-null mice.

For WT mice, the elevation of serum OPN is associated with an increased expression of this protein in tumor-conditioned media

Because OPN is expressed by activated immune cells and other cells in the stroma surrounding the cancer cells ^{6, 20, 21, 26, 27}, there was a possibility that ONSC lacking OPN expression elicited changes in the levels of host-derived OPN in both the tumor microenvironment and systemically, i.e., in the blood. Relative to amounts in WT mice not bearing ONSC cells, serum host-derived OPN was significantly increased at 1 and 3 wk after injection of ONSC cells (Fig. 5A); serum OPN was not detected in OPN-null mice. The reason for the decrease in serum host-derived OPN at wk 2 relative to wk 1 in WT mice is not clear.

Figure 5.

OPN expression in the sera and tumor microenvironment of WT and OPN-null mice harboring ONSC tumors. (A) Time course of serum OPN expression in mice harboring ONSC tumors. Serum OPN was analyzed by ELISA, and values were normalized to serum protein concentrations, n = 3 mice/group. (B) Time course of OPN expression in conditioned media of cultures of tumor explants. Media were collected from tumors grown for 1–3 wk, as described above, and analyzed for secreted OPN by ELISA, n = 4–5 mice/group. (C) Immunohistochemical analyses of OPN in tumors of WT mice at wk 1 and 2. Tumor sections were incubated with anti-mouse OPN (AKm2A1) or IgG₁. Tumor sections from wk 1 of OPN-null mice served as negative controls. N, necrotic. (D) Double-labeling of OPN and macrophage expressing cells in wk 2 of WT tumor. Macrophages stain brown and OPN expression cells are pink. OPN expressing macrophages are light pink indicated by *. Magnification, ×40.

In the tumor microenvironment of WT mice, the levels of host-derived OPN were low but detectable at wk 1 and 3; there was a pronounced rise of host-derived OPN (10x) at wk 2 compared to wk 1 and 3 (Fig. 5B). This is consistent with the immunohistochemical analyses of OPN expression, which showed more intense staining for OPN at wk 2 compared to wk 1 (Fig. 5C). The increase in host-derived OPN expression in the tumors at wk 2 complemented the elevated serum OPN at wk 1. Thus, the increased concentration of OPN in the tumor microenvironment may be derived in part from the increased level in the blood produced by activated, circulating inflammatory cells. Double-labeling analyses from wk 2 tumor sections showed that 45.4 ± 8.6 % of the OPN-expressing cells are infiltrating macrophages, Fig. 5D and Table 1.

The numbers of circulating inflammatory cells are not different between tumor-bearing and non-tumor-bearing mice and between OPN-null and WT tumor-bearing mice

To determine if the elevated level of serum OPN observed in wk 1 of WT tumor-bearing mice compared to WT non-tumor bearing mice was due to an increase in the number of circulating leukocytes, blood samples from these mice were analyzed. The numbers of circulating leukocytes and their subpopulations (lymphocytes, monocytes, and granulocytes) in WT tumor-bearing mice did not differ from those of WT non-tumor-bearing mice (Fig. S2). Thus, the elevated serum OPN levels observed in wk 1 of WT tumor-bearing mice compared to non-tumor bearing mice are likely derived from increased expression of secreted OPN by the activated leukocytes.

Furthermore, to determine if the decrease in the numbers of macrophages infiltrated into the tumor microenvironment in OPN-null mice is due to a decrease in the number of circulating monocytes in OPN-null compared to WT-tumor bearing mice, blood samples from these mice were analyzed and compared. The numbers of circulating monocytes from blood samples of WT and OPN-null mice at 1 wk did not show significant differences between groups with or without tumors (Fig. S2). Therefore, the decrease in the numbers of macrophage infiltrated into the tumor microenvironment of OPN-null relative to WT mice at wk 1 is not due to the decrease in the number of circulating monocytes.

Discussion

Studying the effects of host-derived OPN on cancer progression of cutaneous SCC in an immune-competent environment has been hampered by the lack of appropriate in vivo model. ONSC cells,²² which lack the Spp1 gene and have the capacity to develop SCCs in syngeneic WT and OPN-null mice, can be used to determine the function of host-derived OPN in tumor progression at the systemic level and at the level of the tumor microenvironment of these mice.

Injection of ONSC cells into OPN-null compared to WT mice resulted in a greater incidence of SCC development. This suggests that during the early progression of extrinsic cancer cells, expression of host-derived OPN suppresses ONSC growth, leading to a lower incidence of SCC formation in WT mice. The present results show that the lack of OPN in the tumor microenvironment of OPN-null tumor bearing mice in the first 2 wk is associated with decreases in the levels of cytokines and chemokines. These decreases do not appear to be due to the inability of OPN-null macrophages (PECs) to trigger inflammatory responses. The lower cytokine and chemokines could be due in part to a lower infiltration of macrophages, since there is a significant decrease in the average number of macrophages in the tumor microenvironment of OPN-null compared to WT mice at wk 1. Thus, these findings implicate an early, but relatively weak, inflammatory response in OPN-null mice bearing the OPN-null tumors. In this favorable microenvironment, ONSC cells, lacking OPN expression but have both H-ras and p53 mutations,²² are able to survive, proliferation, and differentiation in OPN-null mice much better than those in WT mice.

Studies with other models support the association of OPN with inflammatory responses, chemotaxis, and the survival of inflammatory cells ^28–32. The present results are consistent in indicating that during the first two weeks of post ONSC injection, host-derived OPN likely maintains an innate inflammatory response in the tumor microenvironment, since cytokines (IL-12p70, INFγ and IL-17) representative of acquired immune responses were not detectable.

OPN is postulated to function as a chemoattractant for infiltrating monocytes and neutrophils ^{29, 31, 33–35}; however, its role in vivo remains controversial.^{24, 28, 36} The contradictory findings are likely due to differences in the models used, the lengths of treatment, and the times at which the assays were performed. In our model, OPN appears to function as a chemoattractant for macrophages, but not neutrophils. At 1 wk after implantation of ONSC cells, there were no differences in the number of infiltrating neutrophils or in the level of myeloperoxidase (not shown) in the tumor microenvironment of WT versus OPN-null mice.

In regard to OPN as a chemoattractant for macrophages, the lack of OPN in the tumor microenvironment of OPN-null mice bearing ONSC cells led to a significant decrease in the number of macrophages at wk 1, but not at wk 2, when compared to OPN-expressing WT mice. Thus, ablation of OPN or the lack of basal OPN expression in the tumor microenvironment appears to delay macrophage infiltration/migration to the tumor site, despite the fact that there were no differences in the numbers of circulating monocytes between the two groups of mice. This is consistent with the significantly lower amounts of cytokines and chemokines in the tumor microenvironment of OPN null mice. Thus, based on these findings, OPN expression appears to elicit a prompt macrophage infiltration to the tumor site.

In addition to functioning as a chemoattractant, OPN prevents apoptosis of various cells in vivo.^{3, 30, 32} Because ONSC cells develop SCCs in some WT mice, the tumor-protective function of host-derived OPN, in the early weeks, appears to be effective in the presence of increased inflammation. TUNEL analyses at 1 wk after ONSC injection in WT mice showed significantly higher apoptotic cells compared to 2 wk. The higher number of apoptotic cells at 1 wk correlated with lower number of cancer cells, greater inflammatory response, and lower level of OPN expression relative to wk 2. Markedly elevated levels of the proinflammatory cytokines IL-6 and IL-10 and chemokines were present in the tumor microenvironment of WT mice. IL-6 possess anti-tumor properties in vivo ³⁷ and IL-10 activates macrophage activity³⁸ and consequently, in the presence of lower OPN level can result in increase cell apoptosis. By wk 2, in WT mice there was a marked decrease in the number of apoptotic cells compared to wk 1. This decrease could be partly due to the increase in cancer cell number and some factor(s) preventing further apoptosis. This factor may be OPN, because the decrease in apoptosis in WT mice coincided with the 10-fold increase of OPN in the tumor microenvironment. The surge of OPN levels may protect cells in the stroma and, possibly, some ONSC cells from being destroyed in the stronger inflammatory environment in WT mice. The host-derived OPN is likely contributed by the activated infiltrating inflammatory cells, such as macrophages and pre-existing histiocytes comprising approximately 45% with the rest of the 55% possibly from activated dermal dendritic cells³⁹, fibroblasts, and other cells in the tumor microenvironment. The putative protective role of OPN on tumor cells is supported by the fact that 70% of the WT mice injected with ONSC cells develop SCCs. Additionally, ONSC cells express integrins and a CD44 variant that can bind to OPN, and experiments in vitro indicate that OPN delays anoikis of ONSC cells (not shown).

Further evidence on the protective role of OPN is the fact that, despite the weaker inflammatory responses in the tumor microenvironment at wk 1 of OPN-null tumor bearing mice, elevated numbers of apoptotic cells were present. It is likely that the lack of OPN decreases the survival of non-cancer cells in the tumor microenvironment, since the tumor sections showed significantly greater numbers of cancer cells compared to those in WT mice. ONSC cells have H-ras and p53 mutations, which make them resistant to apoptosis in a weak inflammatory environment. The significant decrease in the number of apoptotic cells by wk 2 in the tumors in OPN-null mice is largely due to the growth of ONSC cells, indicated by an increase of average number of cancer cells compared to wk 1. Thus, these data suggest that, in addition to OPN functioning as a chemoattractant of macrophages, it may also enhance cell survival.

OPN enhances cell survival of monocytes/macrophages, endothelial cells, and cancer cells in vitro ^40–43 and of tubular epithelium cells, basal keratinocytes, and brain infiltrating lymphocytes in vivo.^{3, 30, 32} Further, OPN added to OPN-null peritoneal exudate cells (PECs) rescues cellular apoptosis mediated through the caspase-3 pathway.²⁹ In the present study, ONSC cells in the microenvironment of WT mice may trigger the activation of fibroblasts, endothelial cells, and pre-existing histiocytes to secrete OPN, which in turn enhances survival of these host cells. In WT mice, activated histiocytes then act to remove ONSC cells through phagocytosis or by a direct cytotoxic effect. Impaired phagocytosis and an anti-tumor cytotoxic effect of macrophages have been noted in OPN-null/deficient mice.^{28, 44, 45} Further, OPN facilitates phagocytosis, possibly through the αxβ2 integrin.^{46, 47} Moreover, the host-derived OPN in blood may enhance the survival of circulating and infiltrating inflammatory cells, resulting in the greater inflammatory response in the tumor microenvironment of WT mice.

With respect to the systemic/blood levels of OPN in WT mice bearing ONSC, elevated OPN serum levels are observed at 1 and 3 wk relative to control mice. We postulate that the elevated serum OPN is contributed largely by activated circulating monocytes, lymphocytes, and/or neutrophils, since non-activated leukocytes express minimal levels of OPN. This is supported by our finding that activated PECs derived from ONSC-bearing WT mice at 2 wk also express elevated OPN in the conditioned media (not shown). Furthermore, there is elevated expression of host-derived OPN in WT mice injected s.c. with Ras-transfected embryonic OPN-null fibroblast cells ²³. Collectively, these findings suggest that, shortly after implantation of ONSC into the microenvironment of WT mice, the initial increase in bloods levels of OPN reflect an anti-tumor response. This anti-tumor effect of host-derived serum OPN is in response to extrinsic cancer cell progression and may not necessarily reflect its role in cell progression at later stages, as observed in cancer patients with metastases⁴⁸. How host-derived OPN affects metastasis remains to be elucidated.

The present study may appear to contradict our previous finding that ablation of OPN delays papilloma development³. In fact, these are two different models addressing the same function of OPN in enhancing cell survival on two different cell types. In contrast to the present study, the previous study examined OPN’s function on tumorigenic transformation of precancerous cells using the two-stage chemically-induced skin carcinogenesis. Precancerous cells or initiated cells usually possess only single mutation (i.e. Ras) where as cancer cells may have two or more mutations rendering them more resistance to apoptosis. Therefore, the lack of induced OPN expression stimulated by carcinogen and tumor promoter, which concomitantly can also induce cell apoptosis on OPN-null precancerous cells, will result in decreased cell survival and thereby, less papilloma development compared to WT mice.³

In conclusion, our SCC cell line with ablated OPN is useful in providing insight into the function of host-derived OPN at the systemic and tumor microenvironment level in the context of extrinsic cancer cell progression in immune-competent mice.

Abbreviations

ONSC

osteopontin-null squamous cell carcinoma cell line

OPN

osteopontin

s. c

subcutaneous

SCCs

cutaneous squamous cell carcinomas

WT

wild-type