Polyamine-Blocking Therapy Reverses Immunosuppression in the Tumor Microenvironment

Abstract

Correcting T cell immunosuppression may unleash powerful antitumor responses, however, knowledge about the mechanisms and modifiers that may be targeted to improve therapy remains incomplete. Here we report that polyamine elevation in cancer, a common metabolic aberration in aggressive lesions, contributes significantly to tumor immunosuppression and that a polyamine depletion strategy can exert antitumor effects that may also promote immunity. A polyamine-blocking therapy (PBT) that combines the well-characterized ornithine decarboxylase (ODC) inhibitor difluoromethylornithine (DFMO) with AMXT1501, a novel inhibitor of the polyamine transport system, blocked tumor growth in immunocompetent mice but not in athymic nude mice lacking T cells. PBT had little effect on the proliferation of epithelial tumor cells but it increased the number of apoptotic cells. Analysis of CD45⁺ tumor immune infiltrates revealed that PBT decreased levels of Gr-1⁺CD11b⁺ myeloid suppressor cells and increased CD3⁺ T cells. Strikingly, in a model of neoadjuvant therapy, mice administered PBT one week before surgical resection of engrafted mammary tumors exhibited resistance to subsequent tumor re-challenge. Collectively, our results indicate that therapies targeting polyamine metabolism do not act exclusively as anti-proliferative agents, but also act strongly to prevent immune escape by the tumor. PBT may offer a general approach to heighten immune responses in cancer.

INTRODUCTION

Elevated levels of polyamines (putrescine, spermidine, and spermine) have long been associated with cell proliferation in both normal and neoplastic tissues (1). Multiple oncogenes and tumor suppressors regulate tumor polyamine metabolism resulting in not only increased polyamine biosynthesis in tumor cells but also increased cellular uptake of polyamines via an upregulated polyamine transport system (2, 3). A hallmark of tumorigenesis involves the induction of ornithine decarboxylase (ODC), the initial ratelimiting enzyme in polyamine biosynthesis (4, 5). Use of transgenic mouse models has demonstrated that increased ODC activity is sufficient to promote tumor development following low dose exposure to carcinogens, UV irradiation, or oncogene activation and wounding (6–10). It is commonly thought that tumorigenesis is promoted by elevated polyamine levels because they stimulate cellular proliferation and angiogenesis in tumors (9, 11–14). However, besides this well-known role in proliferation, polyamines have also been reported to exert an immunosuppressive effect that could contribute to the multiple sophisticated mechanisms by which tumors evade the immune response. For instance, spermine inhibits endotoxin- or HMGB1-induced release of TNFα, IL-1, IL-6, MIP-1a, and MIP-1b from human monocytes and peritoneal macrophages (15), and offers protection against lethal sepsis by attenuating local and systemic inflammatory response (16). Additional in vitro studies report that polyamines suppress lymphocyte proliferation and IL-2 production and decrease macrophage-mediated tumoricidal activity, neutrophil locomotion, and IL-2-dependent NK cell activity (17–21). Moreover, polyamines have been shown to suppress adaptive immune responses. Using transgenic mice in which ODC activity is targeted specifically to the epidermis, we have demonstrated that elevated polyamine levels potently suppress a T cell-mediated, hapten-induced contact allergic response (22). In all, these observations suggest that the role of polyamines as local anti-inflammatory effector molecules at sites of infection or wounds may be usurped by tumors to provide a survival mechanism to evade the immune response.

Since a common metabolic shift in many chronic inflammatory conditions and in all tumors is increased polyamine metabolism, targeting polyamine metabolism has long been an attractive approach to cancer chemotherapy. However, treatment with α-difluoromethylornithine (DFMO), a specific inhibitor of ODC activity, has had only moderate success in treating cancer patients (23). Subsequent studies discovered that DFMO-inhibition of ODC leads to upregulation of the polyamine transporter with resulting increased uptake of polyamines derived from the diet and gut flora into the tumor cells. Thus, to polyamine-starve a tumor, both inhibition of polyamine biosynthesis as well as polyamine transport must be achieved. A novel Polyamine Blocker Therapy (PBT) has recently been described that includes the use of DFMO to block polyamine biosynthesis along with AMXT 1501 as an inhibitor of polyamine transport. AMXT 1501 is designed as a polyamine mimetic and consists of a lysine-spermine backbone with a C₁₆ lipophilic substituent added to the ε-amino group of the lysine portion to optimize its ability to block cellular uptake of spermidine in the nanomolar range without crossing the cell membrane (24). Analyses of AMXT 1501 uptake into tumor cells in culture following a 24-hour incubation with 10 µM AMXT 1501 found no intracellular uptake (24). Additionally, no rescue from the growth inhibitory effects of DFMO occurred when AMXT 1501 was given to cells in the absence of exogenous spermidine, suggesting that AMXT 1501 by itself or its metabolites cannot replenish cellular polyamine requirements. Altogether, these results suggest that AMXT 1501 inhibits the polyamine transporter at the plasma membrane and is not internalized within the cell. Initial in vivo experiments have shown that PBT causes complete or near-complete regression in the majority (88%) of carcinogen-induced squamous cell carcinomas (SCC) in ODC transgenic mice in which skin tumors are promoted by elevated epidermal ODC activity (24).

In this study, we tested the therapeutic efficacy of reducing tumor polyamine levels with PBT in animal tumor models that are not as strongly dependent upon polyamine biosynthesis as are the carcinogen-initiated skin tumors in ODC transgenic mice. In addition, we investigated the effects of PBT on the immune response to tumors. Our data highlight a little appreciated role of polyamines as strong modifiers of the inflammatory microenvironment in a tumor, and describe a novel approach to suppress tumor growth and reverse tumor immunosuppression by targeting tumor polyamines.

MATERIALS AND METHODS

Animals

ODC-ER transgenic mice, in which an involucrin promoter directs the expression of the inducible ODC cDNA fused in frame to a 4-hydroxytamoxifen-responsive mutant estrogen receptor ligand binding domain to the suprabasal epidermis, has been described previously (9). ODC-ER transgenic mice and their normal littermates were backcrossed into either the Balb/c or C57BL/6 background for at least 10 generations. ODC activity was induced in ODC-ER transgenic mice beginning one week prior to tumor cell injection or intradermal immunizations by topical application of 4-hydroxytamoxifen (4OHT) (Sigma, St. Louis, Missouri) dissolved in ethanol (1.0 mg/100 µl) applied each day to a shaved area of the dorsal skin. Additional mice used in tumor experiments included C57Bl/6, Balb/c, and athymic nude mice obtained from Charles Rivers/NCI and FVB mice purchased from Taconic Labs. Protocols using animals for this study were approved by Institutional Animal Care and Use Committee of the Lankenau Institute for Medical Research in accordance with the current US Department of Agriculture, Department of Health and Human Service regulations and standards.

Cell Culture

EL4 murine lymphoma, EG7 (an ovalbumin (OVA)-expressing EL4 variant (25)) cell lines, and murine macrophage RAW 264.7 cells, all purchased from the American Type Culture Collection (Rockville, MD), were cultured in RPMI 1640 medium (Mediatech, Manassas, VA) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Atlanta Biologicals, Lawrenceville, GA), 100 µg/ml streptomycin and 10 mM L-glutamine; all purchased from Mediatech (Manassas, VA). B16-F10 murine melanoma cells were purchased from the American Type Culture Collection (Rockville, MD). CT26.CL25 (American Type Culture Collection, Rockville, MD) is a subclone of CT26 colon carcinoma cells that have been transduced with the Escherichia coli β-gal gene, which have been shown to be equally as lethal in normal mice as the parental clone CT26.WT. Neu02 (26), a mammary carcinoma cell line derived from an autochthonous mammary gland tumor from a MMTV-Neu.FVB mouse, was generously provided by Dr. Alexander Muller (Lankenau Institute for Medical Research, Wynnewood, PA). B16F10, CT26.CL25, and neu02 cells were cultured in DMEM (Mediatech, Manassas, VA) with 10% FBS. Primary cultures of epidermal keratinocytes were prepared from 3–4 day old FVB mice by a trypsin flotation method (27). Cells were plated at 3.5 × 10⁶ cells per 60 mm dish in low calcium keratinocyte media (EMEM w/o calcium, BioWhittaker Walkersville, MD) supplemented with 8% chelex-treated FBS and 0.05 mM calcium and grown at 35°C with 5% CO₂. Keratinocytes were transduced with a replication-defective retrovirus containing the v-Ha-Ras oncogene (28) to generate Ker/Ras cells.

In vivo tumor models

Tumor models were established by subcutaneous (s.c.) injection of 1 × 10⁶ EG7 cells or 5 × 10⁵ B16F10 cells in C57BL/6 mice; s.c. injection of 5 × 10⁵ CT26.CL25 cells in Balb/c mice; and s.c. injection of 3 × 10⁶ Ker/Ras cells in FVB mice. Mice were monitored twice a week for tumor growth, and treatment with 0.5% (w/v) DFMO in their drinking water and AMXT 1501 (i.p., 3 mg/kg (as its 4HCl salt), bid Mon-Fri and once a day on weekends) was initiated when the tumors were palpable (50–100 mm³ in size). Tumor growth was assessed morphometrically using calipers, and tumor volumes were calculated according to the formula V (mm³) = π/6 × A × B² (A is larger diameter, B is the smaller diameter) (29).

For tumor re-challenge experiments, a total of 3 × 10⁶ neu02 cells were orthotopically injected in the mammary fat pad of FVB mice. When tumors were 50–80 mm³ in size, mice were treated with DFMO and AMXT 1501 using the protocol above, for 7 days. At that time, treatment was halted, tumor size measured, and all mammary tumors were surgically resected. After 2 weeks, the mice were orthotopically injected with the same number of neu02 cells as before (3 × 10⁶) in a mammary fat pad on the opposite side. Mice were monitored for tumor growth, and tumors were measured with calipers.

Antigen-Specific T-Cell Response Detection by IFN-γ ELISpot

LacZ- replication-deficient human type 5 recombinant adenovirus (LacZ-Adeno-X) (Clontech, Mountain View, CA) was expanded and purified using the Adeno-X Virus Purification kit (Clontech, Mountain View, CA) and titered using the Adeno-X Rapid Titer kit (Clontech, Mountain View, CA). For immunization, ODC-ER.Balb mice and their normal Balb/c littermates were first topically treated once a day with 4OHT (1.0 mg/100 µl ethanol) to induce ODC activity in ODC-ER transgenic mice, and then one week later the mice were immunized via a single intradermal injection of 1 ×10⁸ IFU adenoLacZ on the shaved dorsal flank (30). Two weeks later, mice were sacrificed, and splenocytes were analyzed for IFN-γ producing cells by ELISpot assay.

For IFN-γ ELISpot assays, multiscreen filtration plates (Millipore, Billerica, MA) were coated with 0.5 ug/ml purified anti-mouse IFN-γ capture antibody (BioLegend, San Diego, CA) overnight at 4°C. Single cell suspensions of splenocytes were plated at 1 ×10⁶ and 5 ×10⁵ cells per well and then peptide-stimulated with a known H-2d restricted β-galactosidase peptide (TPHPARIGL) (ChemPep) (that is presented by the MHC class I L^d-molecule on the surface of CT26.CL25 cells) or with the dominant adenovirus epitope DBP_412–420 (31) peptide at 20ug/ml. For ELISpot assays with splenocytes from EG7 tumor-bearing mice, cells were stimulated with SIINFEKL peptide (AnaSpec, Fremont, CA), addition of EL4 or EG7 cells, or with the dominant adenovirus epitope DBP_418–426 peptide. After 16 hr of stimulation at 37°C, the cells were removed by washing, and spots were developed with biotinylated anti-IFN-γ detection antibody and Strepavidin-HRP conjugate (BioLegend, San Diego, CA) followed by Nitro-blue tetrazolium chloride and 5-Bromo-4-chloro-3’-indoylphosphate p-toluidine salt substrate (Sigma, St. Louis, MO). Spot numbers were counted, and data reported as IFN-γ-spot forming cells (SFC) per 10⁶ cells or per well.

Immunohistochemistry

Mouse tumor tissues were fixed in 4% p-formaldehyde in PBS overnight and embedded in paraffin. Skin sections were deparaffinized, hydrated, and then heated in 0.01 M sodium citrate buffer (pH 6.0) in a steamer for 8 min. Sections were incubated with primary antibodies followed by biotinylated secondary antibody, and then an avidin horseradish peroxidase complex (Vectastain Elite ABC kit, Vector Laboratories, Inc., Burlingame, CA). Primary antibodies used were a rat anti-mouse neutrophil antibody recognizing the 7/4 antigen (Clone 7/4, Cedarlane, Burlington, NC) and a rat monoclonal anti-mouse F4/80 antigen (AbD Serotec, Raleigh, NC). Immunoreactive cells were localized by incubating the sections with a chromagen solution containing diaminobenzidine and peroxide and then counterstaining with hematoxylin. Apoptotic cells were visualized by staining for terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) (Roche Diagnostics, Mannheim, Germany). Bromodeoxyuridine (BrdU) incorporation in cells undergoing DNA synthesis was detected in skin sections using a rat monoclonal anti-BrdU antibody (Zymed Laboratories, San Francisco, CA). Pictures were acquired using a Zeiss Axiophot microscope (Carl Zeiss Inc., Oberkochen, Germany), with a digital color camera and corresponding software (Axiocam, Zeiss). All images were processed for printing using Adobe Photoshop software.

Flow Cytometry Analysis of Immune Cell Infiltrates

Tumor or spleen tissue was digested in a 0.3% collagenase/0.1% hyaluronidase solution, pressed through a nylon mesh filter, and a single cell suspension prepared for flow cytometric analysis. Equal numbers of viable cells were stained with combinations of the following: CD45-ApcCy7, Gr1-PECy7, CD11b-PE, F4/80-Apc, CD3-FITC, CD8-PECy5, CD44-PE (all antibodies from eBioscience, San Diego, CA). Flow cytometric data were acquired on a BD FACSCanto II flow cytometer and analyzed using FACSDiva software (BD Biosciences, San Jose, CA) or FloJo7 software (Tree Star, Ashland, OR). Viable cells were gated based on forward and side scatter profiles. A minimum of 10⁶ viable cells were analyzed for each condition.

Determination of Arginase Activity

Arginase activity was measured in cell lysates with slight modifications, as previously described (32). Briefly, cells were lysed in 0.1% Triton X-100 and 25 mM Tris HCl, pH 8.0. To 100 µl of this lysate, 10 µl of 10 mM MnCl₂ was added, and the enzyme was activated by heating for 10 min at 56°C. Arginine hydrolysis was conducted by incubating the lysate with 100 µl of 0.5 M L-arginine (pH 9.7) at 37°C for 60 to 120 min. The reaction was stopped with 900 µl of H₂SO₄ (96%)/H₃PO₄ (85%)/H₂O (1:3:7, v/v/v). The urea concentration was measured at 540 nm after addition of 40 µl α-isonitrosopropiophenone (dissolved in 100% ethanol, Sigma, St. Louis, MO) followed by heating at 95°C for 30 min. The rate of urea production was used as an index for arginase activity.

NO Measurement

NO production was measured as nitrite using the Griess reagent (Sigma, St. Louis, MO). Culture supernatant (50 µl) was mixed with 50 µl of 1% sulfanilamide, 0.1% N-(1-naphthyl)ethylenediamine dihydrochloride, and 2.5%H₃PO₄. Absorbance was measured at 540 nm in a microplate reader.

Radiolabeled Spermidine Transport Assays

Polyamine transport in tumor cells was evaluated essentially as described previously (33, 34). Cells were plated in triplicate and grown to approximately 70% confluence. After washing with PBS, ³H-spermidine (Net-522, spermidine trihydrochloride, [terminal methylenes-³H(N)], specific activity 16.6 Ci/mmol; Perkin Elmer, Boston, MA) was added at 1.0 µM and incubated for 60 minutes at 37°C. The cells were washed with cold PBS containing 50 µM cold spermidine one time followed by two washes with PBS, lysed in 500 µl of 0.1 N NaOH at 37°C for 30 minutes, sonicated for 10 sec, and then an equal volume of 0.1N HCL was added to neutralize pH. The ³H radioactivity in each cell lysate was measured by scintillation counting and normalized to protein.

Statistical Analysis

All in vitro experiments were performed at least in triplicate, and data were compiled from 2–3 separate experiments. Analyses were done using SAS Version 9.2 using a two-tailed Student t-test. In vivo studies were carried out using multiple animals (n = 4–8 per treatment group). Tumor growth curves were analyzed with a Generalized Linear model with fixed effects of treatment and time. Data were examined for the interaction between treatment group and day of observation, testing whether the slopes of the growth curves (tumor volume versus day of observation) were significantly different for the control and treatment groups. In all cases, values of p ≤ 0.05 were regarded as being statistically significant.

RESULTS

Elevated polyamine levels suppress antigen-specific T cells in the tumor microenvironment

To mimic the increased polyamine biosynthesis found in tumors, we have generated an inducible ODC-ER transgenic mouse model in which the expression of a switchable form of the ODC protein fused to a modified estrogen receptor (ODC-ER) is targeted to the epidermis using an involucrin promoter (9). A single, topical, sub-threshold exposure to a carcinogen, then induction of epidermal ODC enzyme activity by topical treatment with the inducing agent 4-hydroxytamoxifen (4OHT), leads to rapid development of squamous cell carcinomas in transgenic mice (9). Wound repair following skin abrasion in ODC-ER transgenic mice is prolonged and characterized by exaggerated epidermal hyperplasia, persistent inflammation, and wound-induced tumors (10).

Based on earlier observations of a reduced T cell-mediated contact hypersensitivity response in mice harboring elevated epidermal polyamine levels (22), we hypothesized that polyamines might contribute to T cell immunosuppression in the tumor microenvironment. To investigate the polyamine-dependent microenvironmental effects on growth of tumor allografts, cutaneous ODC activity was induced in ODC-ER.B6 mice with 4OHT treatment and then EG7 thymoma cells expressing ovalbumin (OVA) (25, 35) were injected intradermally into the syngeneic ODC-ER.B6 transgenic mice and normal littermates. The increased EG7 tumor growth in ODC-ER transgenic mice compared to their normal littermates was accompanied by significantly fewer IFN-γ-producing splenocytes as measured by ELISPOT assay following ex vivo stimulation with irradiated EG7 cells or the OVA_257–264 (SIINFEKL) peptide (Fig. 1A, B). Thus, an antigen (OVA)-specific T lymphocyte response was suppressed in ODC-ER transgenic mice, perhaps permitting greater tumor growth of the OVA-expressing EG7 tumor cells.

Fig. 1. Activation of antigen-specific T cells is suppressed in ODC-ER transgenic mice.

ODC-ER.B6 transgenic mice and their normal littermates were s.c. injected with 1 × 10⁶ EG7 cells in their lower dorsal flank. A) Tumor volumes of mice bearing EG7 tumors (n = 5 mice/group; mean volume per time point ± SEM).

B) Splenocytes from EG7 tumor-bearing mice adoptively transferred with wild-type C57BL/6 CD8⁺ T cells were stimulated with SIINFEKL or irradiated OVA-expressing EG7 cells or parental EL-4 tumor cells. The frequency of IFNγ-producing T cells was measured by the ELISpot assay. Results are given as mean ± SD (n = 5 mice/group). *, P < 0.01 vs splenocytes from normal littermate mice (t-test). C) ODC-ER.Balb transgenic mice and their normal littermates (n = 5 mice/group) were immunized by intradermal injection with adenovirus encoding the LacZ peptide, and the LacZ-specific response of splenocytes was evaluated 14 days after immunization in an IFN-γ ELISPOT assay. *, P < 0.01 vs splenocytes from immunized normal littermate mice (t-test). D) ODC-ER.Balb transgenic mice and their normal littermates (n = 5 mice/group) were intradermally injected in their lower dorsal flank with 5 × 10⁵ CT26.CL25 tumor cells expressing LacZ antigen. Eleven days after tumor cell injection, the LacZ-specific response of splenocytes was evaluated in an IFN-γ ELISPOT assay. *, P < 0.01 vs splenocytes from normal littermate mice (t-test).

Following induction of ODC activity with daily 4OHT treatment, ODC-ER transgenic mice and normal littermates were then immunized by intradermal injection with adenovirus encoding β-galactosidase. The antigen-specific response of splenocytes was evaluated 14 days after immunization in an IFN-γ ELISPOT assay. Following stimulation with a LacZ peptide or peptide derived from the adenovirus vector, there was significantly (p < 0.01) fewer IFN-γ-producing splenocytes from ODC-ER transgenic mice compared to normal littermates (Fig. 1C). This suppression of antigen-specific T-cell cytokine production was also seen in ODC-ER.Balb transgenic mice injected intradermally in the flank with syngeneic CT26.CL25 tumor cells that express the LacZ antigen (Fig. 1D). Interestingly, palpable CT26.CL25 tumors were only detected in ODC-ER transgenic mice and not in normal littermate mice at 11 days following tumor injection when the splenocytes were assayed in the IFN-γ ELISPOT.

Antitumor efficacy of polyamine-blocking therapy relies on host immune competence

Because increased tumor growth in ODC-ER transgenic mice was accompanied by a suppressed antigen-specific T-cell cytokine response, we tested the therapeutic efficacy of polyamine depletion via a combination of DFMO to block polyamine biosynthesis and AMXT1501 to inhibit the polyamine transport system (polyamine blocker therapy, PBT, Fig. 1A). In these experiments we analyzed the responses of two different murine tumor cell lines, CT26.CL25 colon carcinoma and B16F10 melanoma cells. In both models, treatment was initiated with 0.5% DFMO in the drinking water with or without i.p. injection of AMXT 1501 (3 mg/kg, bid) when the tumors were 50–100 mm³ in size. Whereas treatment with either DFMO or AMXT1501 alone retarded B16F10 tumor growth in syngeneic C57BL/6 mice, co-treatment with both DFMO and AMXT1501 significantly inhibited tumor growth more than treatment with a single agent (Fig. 2A). HPLC analysis of polyamine content in tumors revealed greatly elevated levels of polyamines in tumor tissue compared to non-tumor bearing skin (Fig. 2C). Moreover, only treatment with both DFMO and AMXT1501 significantly decreased putrescine and spermidine levels in the tumors.

Fig. 2. Tumor growth inhibition by DFMO + AMXT1501 is dependent on T cells.

A) Structural representations of the inhibitor of the polyamine transport system, the D-C₁₆-acyl Lys-spm conjugate, AMXT1501, and the ODC inhibitor, α-difluoromethylornithine (DFMO). B) C57BL/6 mice were injected s.c. with 5 × 10⁵ B16F10 tumor cells. When the tumors were 50–100 mm³ in size, treatment was initiated with either saline, 0.5% DFMO (w/v) in the drinking water, AMXT1501 (i.p., 3 mg/kg, bid Mon-Fri and once a day on weekends), or both DFMO and AMXT1501. Graph shows tumor growth rate and volume in mice bearing B16F10 tumors under different treatments (n = 6–7 mice/treatment group; mean volume per time point ± SEM). C) Upon sacrifice, nontumor-bearing normal skin and tumor tissues (n = 6–7 mice/treatment group) were frozen in liquid nitrogen, ground to a fine powder, homogenized in 0.2 N perchloric acid, and incubated at 4°C overnight. Dansylated polyamines were separated on a reversedphase C18 HPLC column (52). Polyamine values were normalized for the amount of DNA in the tissue extracts and are expressed as nmol/ mg DNA. Values are means ± SEM, with an asterisk marking significant differences between polyamine levels in control mice and treated mice with p-values ≤ 0.02. D) Balb/c mice were injected s.c. with 5 × 10⁵ CT26.CL25 tumor cells. When the tumors were 50–100 mm³ in size, treatment was initiated with either saline, 0.5% DFMO (w/v) in the drinking water, or co-treatment with 0.5% DFMO and AMXT1501 (i.p., 3 mg/kg, bid Mon-Fri and once a day on weekends). Graph shows tumor growth rate and volume in mice bearing CT26.CL25 tumors under different treatments (n = 5 mice/treatment group; mean volume per time point ± SEM). E) Athymic nude mice were injected s.c. with 5 × 10⁵ CT26.CL25 tumor cells. When the tumors were 50–100 mm³ in size, treatment was initiated with either saline or co-treatment with 0.5% DFMO in the drinking water and AMXT1501 (i.p., 3 mg/kg, bid Mon-Fri and once a day on weekends). Graph shows tumor growth rate and volume in mice bearing CT26.CL25 tumors under different treatments (n = 5 mice/treatment group; mean volume per time point ± SEM). *, P < 0.01.

Similar to B16F10 tumors, CT26.CL25 tumor growth in syngeneic Balb/c mice was not significantly inhibited by DFMO treatment alone, but treatment with both DFMO and AMXT 1501 significantly inhibited tumor growth (Fig. 2D). Because T-cell responses were suppressed in ODC-ER transgenic mice with elevated polyamine levels, we hypothesized that the PBT-mediated repression of tumor growth via polyamine deprivation may be dependent on T-cells. To test this hypothesis, athymic nude mice lacking T cells were injected with CT26.CL25 tumor cells, and PBT was initiated when the tumors were 50–100 mm³. In contrast to that seen in immunocompetent tumor-bearing Balb/c mice, combined treatment with DFMO and AMXT 1501 had no effect on CT26.CL25 tumor growth in athymic nude mice (Fig. 2E). These results suggest that PBT-tumor inhibition is dependent on T cell function.

PBT inhibition of CT26.CL25 tumor growth was accompanied by a dramatic increase in the number of apoptotic tumor cells as shown by the increased TUNEL stained cells in tumors from DFMO and AMXT 1501-treated mice compared to untreated mice (Fig. 3A). Mice were injected with bromodeoxyuridine (BrdU) 2 hours before sacrifice, and the relative proliferation index was determined by counting tumor epithelial cells that incorporated BrdU. DFMO treatment had no effect on the proliferation index, whereas treatment with DFMO and AMXT 1501 moderately decreased proliferation, albeit not to a statistically significant extent (Fig. 3B).

Fig. 3. Inhibition of CT26.CL25 tumor growth with DFMO ± AMXT1501 is accompanied by increased apoptosis but little change in the proliferation index.

Balb/c mice were injected s.c. with 5 × 10⁵ CT26.CL25 tumor cells. When the tumors were 50–100 mm³ in size, treatment was initiated with either saline, 0.5% DFMO (w/v) in the drinking water, or co-treatment with 0.5% DFMO and AMXT1501 (i.p., 3 mg/kg, bid Mon-Fri and once a day on weekends). All mice were injected i.p. with bromodeoxyuridine (BrdU, 100 µg/g body weight) 2 h before sacrifice. A) Cell death was analyzed by TUNEL staining, and nuclei were stained with DAPI. Representative images are shown for tumor sections from mice not treated or treated with both DFMO and AMXT1501. B) Tumor sections were stained for cells that incorporated BrdU by immunohistochemical staining. BrdU-positive cells/1000 tumor epithelial cells were counted in 4–5 sections for each treatment group. The percentage of BrdU-positive cells is expressed as the mean ± SD. ns, not significant.

We expanded the above studies to a genetically defined tumor model by using primary keratinocytes transformed via v-Ha-ras retroviral infection (Ker/Ras cells). Using a similar methodology, Ker/Ras cells were used to generate orthotopic skin tumors by intradermal injection in the flank of syngeneic FVB mice. PBT treatment was initiated at 13 days following injection of Ker/Ras tumor cells when the tumors were between 50–100 mm³ in size. Co-treatment with both DFMO and AMXT 1501 significantly suppressed Ker/Ras tumor growth compared to that in untreated mice, demonstrating better therapeutic efficacy than treatment with either DFMO or AMXT1501 alone (Fig. 4A). There was no significant difference in tumor growth in mice treated with AMXT1501 alone compared with that in control treated mice. Notably, DFMO alone was a better inhibitor of Ker/Ras tumor growth compared to its tumor suppressive ability with B16F10 or CT26.CL25 tumors. These results underscore that PBT inhibits two main sources of polyamines for tumors. Some tumors, such as the Ker/Ras tumors, may be more dependent on polyamine biosynthesis, and DFMO is sufficient to significantly retard tumor growth. However, growth of tumors with an upregulated polyamine transporter will be suppressed only with the addition of an inhibitor of polyamine transport such as AMXT1501. The relative spermidine transport activity in each of these different tumor cells was measured using an in vitro assay of tumor cells for uptake of [³H] spermidine. As expected, treatment of cells with AMXT1501 completely blocked uptake of [³H] spermidine (Fig. 4B). With no addition of a polyamine transport inhibitor, B16F10 tumor cells demonstrated the greatest polyamine transport activity compared to CT26.CL25 cells or Ker/Ras tumor cells (Fig. 4B). These data reflect our observed in vivo effect that treatment with AMXT1501 alone reduced the growth of B16F10 tumors but not that of Ker/Ras tumors.

Fig. 4. PBT retards Ker/Ras tumor growth and stimulates macrophage infiltration in Ker/Ras tumors.

Primary murine keratinocytes (3 × 10⁶) transduced with v-H-Ras retrovirus were s.c. injected into syngeneic FVB mice. When the tumors were 50–100 mm³ in size, treatment was initiated with either saline, 0.5% DFMO (w/v) in the drinking water, AMXT1501 (i.p., 3 mg/kg, bid Mon-Fri and once a day on weekends), or co-treatment with DFMO and AMXT1501. A) Graph shows tumor growth rate and volume in mice bearing Ker/Ras tumors under different treatments (n = 5 mice/treatment group; mean volume per time point ± SEM). *, P < 0.01. B) To compare basal polyamine transport in cultures of Ker/Ras cells with B16F10 and CT26.CL25 cells, ³H-spermidine was added, and cells were incubated for 1 hr at 37°C or on ice. Uptake of ³H-spermidine (cpm) was measured using a scintillation counter and is normalized to protein. Values are the mean of triplicate samples ± SD. *, P < 0.01. (C–H) Flow cytometry analyses of CD45⁺ inflammatory infiltrates in Ker/Ras tumors. Mice with Ker/Ras tumors were treated for two weeks with DFMO or with both DFMO and AMXT1501, and tumors were analyzed by flow cytometry for C) % CD45⁺ immune cell infiltrates in total viable tumor cell population. Subpopulations of viable CD45⁺ cells were also stained for the following immune cell markers: D) F4/80⁺ macrophages; E) CD11b⁺ myeloid cells; F) CD3⁺ lymphocytes; G) Gr-1⁺ granulocytes; and H) Gr-1⁺CD11b⁺ MDSCs. *, P < 0.01.

After 17 days of treatment, Ker/Ras tumors were analyzed for inflammatory infiltrates by immunohistochemical staining and by flow cytometry analysis. Although F4/80⁺ stained macrophages were detected in all tumors regardless of treatment, there was very little infiltration of macrophages in the epithelial compartment of tumors from untreated mice compared to tumors from mice treated with DFMO and AMXT1501 (Supplementary Fig. 1S). Following PBT treatment, F4/80⁺ stained macrophages were no longer marginalized outside the tumor epithelial compartment (Supplementary Fig. 1S). Flow cytometry analyses of tumor infiltrating cells revealed significant differences in the subpopulations of CD45⁺ immune cells between mice that were treated and untreated. Whereas there were no significant differences in the numbers of CD45⁺ inflammatory cells in disaggregated, viable tumor cells, treatment with DFMO and AMXT1501 (PBT) increased the percentage of CD3⁺ T cells and decreased the percentage of F4/80⁺ macrophages and Gr-1⁺ (neutrophil/monocyte) subpopulations in CD45⁺ tumor infiltrating cells compared to that in CD45⁺ tumor infiltrating cells from untreated mice (Fig. 4C – H). In particular, PBT significantly decreased the population of Gr-1⁺ CD11b⁺ cells, which have been characterized as myeloid-derived suppressor cells (MDSC) in many tumor types. Although PBT decreases the numbers of tumor myeloid cells, the increased tumor infiltration of both myeloid cell populations and T cells suggests a phenotypic change in the tumor myeloid population with PBT treatment.

Polyamine-blocking therapy limits induction of an M2 macrophage phenotype associated with tumor immunosuppression

Multiple subpopulations of tumor-infiltrating immune cells, including Gr-1⁺CD11b⁺ MDSCs, granulocytes, immature dendritic cells, and regulatory T cells, have been shown to profoundly suppress T cell functions via arginase-mediated arginine depletion (36–38). Inhibition of myeloid cell-associated arginase activity eliminates tumor-associated immune dysfunction and suppresses tumor growth (37, 38). Arginase hydrolyzes L-arginine to the products urea and L-ornithine which is further metabolized via ODC to polyamines (Supplementary Fig. 2S). In addition to induction by Th2 cytokines, we have observed that arginase activity in tumor myeloid cells may also be induced by increased tumor levels of polyamines. Figure 5A shows that arginase activity is significantly increased in primary keratinocytes with elevated ODC activity and polyamine levels, potentially explained by an increased metabolic utilization of L-ornithine. Conversely, arginase activity is decreased with DFMO treatment. While DFMO itself has been described as an arginase inhibitor (39), an alternative explanation for the decreased arginase activity upon DFMO treatment of these cells could be the feedback inhibition of arginase by the increased levels of ornithine resulting from DFMO-inhibition of ODC.

Fig. 5. DFMO blocks both ODC induction of arginase activity in primary keratinocytes and M2 polarization of IL-4 treated macrophages.

A) Primary keratinocytes were isolated from the skin of K6/ODC transgenic mice and their normal littermates. Half of the keratinocytes were cultured in medium containing with 2 mM DFMO, and after 24 hr the cell lysates were assayed for arginase activity. All cell cultures were assayed at least in triplicate. (B and C) Murine RAW264.7 macrophage cells were treated with IL-4, IFNγ, or LPS for 24 hr in the presence or absence of DFMO (0.1 or 0.5 mM). B) Cell lysates were assayed for arginase activity (urea production) and C) the cell supernatants were assayed for NO production using the Griess reagent. All results are given as mean ± SD; *, P < 0.01.

Because macrophages have been shown to polarize to an arginase-expressing M2 phenotype following exposure to Th2 cytokines such as IL-4 or IL-10 (40), we tested the effect of inhibition of polyamine biosynthesis using DFMO on the M1/M2 polarization of cultured murine RAW264.7 macrophage cells. As expected, IFN-γ or LPS-induction of a pro-inflammatory M1 phenotype was accompanied by increased NOS activity as indicated by increased NO production, and IL-4 induction of a M2 phenotype was marked by an induction of arginase activity (Fig. 5 B and C). DFMO treatment significantly blocked IL-4 induction of arginase activity (Fig. 5B). However, the same concentrations of DFMO had no significant inhibitory effect on LPS induction of NOS activity in RAW264.7 cells (Fig. 5C). All together, these data suggest that inhibition of polyamine biosynthesis reverses the immunosuppressive tumor microenvironment by skewing myeloid cell polarization from a M2 pro-growth phenotype to a M1 pro-inflammatory phenotype.

Polyamine-blocking therapy promotes durable protection against tumor re-challenge

If reversal of a polyamine-dependent immunosuppressive tumor microenvironment contributes to PBT inhibition of tumor growth, then it is possible that PBT generates a protective T cell memory to protect against tumor recurrence. We used a model of HER2/neu-mediated breast cancer to functionally test whether short term treatment with DFMO and AMXT 1501 before surgical excision of a primary tumor can afford protection against re-challenge with the same tumor. Neu02 mammary tumor cells isolated from a mammary adenocarcinoma in a transgenic MMTV-Neu mouse (26) were orthotopically injected in a right side mammary fat pad of syngeneic female FVB mice. When the tumors were between 50–80 mm³ in size, half the mice were treated with both DFMO and AMXT 1501 for 7 days while the other half were treated with tap water and injected with control vehicle. After 7 days of treatment, the tumors were surgically resected, and treatment was discontinued. After 2 weeks, the mice were orthotopically re-injected with Neu02 cells in a left side mammary fat pad. As expected, Neu02 tumors grew significantly bigger in naïve FVB mice than in mice previously challenged with Neu02 tumor cells with surgical removal of their first Neu02 tumor. Tumors grew only in surgically resected mice that had not received prior PBT treatment (Fig. 6). In contrast, there was no tumor growth in any mice that had been previously treated with PBT before surgery (Fig. 6). Overall, these results revealed that PBT may elicit a durable immune memory when administered before surgical resection of the primary tumor.

Fig. 6. Initial PBT treatment leads to protection against tumor re-challenge.

FVB female mice were orthotopically injected with syngeneic Neu02 mammary adenocarcinoma cells. When the tumors were 50–80 mm³ in size, mice (n = 5 mice/ treatment group were treated with saline (Control) or co-treatment with AMXT1501 (3 mg/kg i.p., bid) and 0.5% DFMO in the drinking water for 7 days. Treatment was stopped, and all tumors were surgically resected. A) Graph shows tumor growth rate and volume in mice bearing neu02 tumors during the 7 day treatment (mean volume per time point ± SEM). B) Two weeks following the tumor resection, all mice were re-challenged with the same Neu02 tumor cells via orthotopic injection in a mammary fat pad on the opposite side. Mice were not treated after the second tumor injection. Graph shows tumor growth rate and volume in mice after being re-challenged with Neu02 tumor cells (mean volume per time point ± SEM). Tumors did not grow in animals that had received prior PBT. *, P < 0.01.

DISCUSSION

Tumors have evolved a variety of survival mechanisms to suppress the innate and the effector arms of the immune system, thus compromising most cancer immunotherapy and vaccine therapies in clinical trials. An exception is the recent targeting of the immune checkpoint molecules, CTLA-4 or PD-1, which holds great promise by unleashing the inherent immune reactivity to the tumor (41). The development of additional, complementary therapeutic approaches to defeat the tumor’s immune checkpoint mimicking ability is urgently needed. The availability of broad-based immunochemotherapies that reverse the immunosuppressive microenvironment inherent in growing tumors offers profound clinical implications. For instance, inhibition of the pathways controlled by indoleamine 2,3-dioxygenase (IDO) (42), adenosine (43), and arginase (44), all shown to impede T cell-mediated tumor elimination, is being vigorously pursued. Our data raise the possibility that tumor production and release of polyamines into the tumor microenvironment may contribute to immune-editing or sculpting of tumor cells and therefore represent an important mechanism for selection of immune cloaked cells inherent to the immunosuppressive tumor microenvironment. Based on the data discussed here, we propose adding polyamines to the list of immunosuppressive small molecule metabolites. Our study is significant because it highlights this immunosuppressive function of tumor polyamines as an important factor contributing to tumor development and survival that is in addition to the commonly held view that polyamines are needed for tumor cell proliferation. Importantly, PBT-mediated reduction of tumor polyamine levels dramatically retards tumor growth by reversing this immunosuppression.

In this study we observed several intriguing phenomena using multiple murine tumor models: i) elevated polyamine-mediated suppression of an antigen-specific immune response creates a more permissive microenvironment for tumor growth; ii) inhibition of tumor growth by polyamine depletion in tumors via co-treatment with DFMO and AMXT 1501 is T-cell-dependent; and iii) treatment with DFMO and AMXT 1501 before surgical removal of the primary tumor leads to protective immunity to tumor re-challenge. These data are consistent with earlier reports (15, 16, 19, 20) that polyamines inhibit induction of proinflammatory cytokines in monocytes, protect against lethal sepsis, and suppress immune cell activity. Polyamine levels are elevated in tumors and are released by dying cancer cells found in hypoxic and necrotic areas of the tumor following chemotherapy or radiotherapy. Our data provide compelling in vivo evidence that the anti-inflammatory effects of polyamines released into the tumor microenvironment contribute to the immunosuppressive milieu commonly found in most tumors. Furthermore, it is intriguing to speculate on the immune suppressive capabilities of each individual polyamine. The majority of prior research explored the potency of spermine as an immune suppressive metabolite; our data support a profound biological outcome following predominately putrescine and spermidine depletion from the tumor microenvironment (Fig. 2C). Only minor changes in the levels of spermine were observed with any of the treatments described. Clearly, subtle changes in the balance of individual polyamines have a significant impact on the immune biology of the tumor microenvironment. Furthermore, the potential for toxic consequences of drastic spermine depletion and associated offtarget effects should also be considered when designing a polyamine anti-metabolite approach.

PBT suppresses tumor growth in several in vivo tumor models by, at least in part, stimulating an immune antitumor effect that is T cell-dependent. Our results demonstrate that PBT antitumor efficacy is dependent on T cells and perhaps decreases M2 macrophage/monocyte cells that have pro-tumor growth properties. One possible mechanism is via inhibition of tumor myeloid cell arginase activity. We have shown that increased polyamine levels induce arginase activity in epithelial cells and may also contribute to the induction of immunosuppressive arginase-expressing tumor myeloid cell populations. Inhibition of polyamine biosynthesis with DFMO blocks the IL-4-induced arginase activity (marker of M2 polarization) of macrophage cells. Smoldering, nonresolving inflammation characterizes the tumor microenvironment in which tumor infiltrating myeloid cells acquire M2 properties that promote cancer cell proliferation, stromal deposition, and angiogenesis and that inhibit adaptive immunity (45). Interestingly, we have shown previously that wound-induced skin tumor formation in ODC-ER transgenic mice is dependent on a polyamine-induced persistent inflammation following wounding (skin abrasion) (10). PBT may modulate the chronic, “smoldering” inflammation that characterizes many tumors by skewing tumor myeloid cell polarization from a M2 pro-tumor growth phenotype to a M1 pro-inflammatory phenotype. In addition, short term PBT prior to surgical tumor resection induced a protective immune response to subsequent tumor re-challenge. Since the function of T cells is severely compromised when the extracellular arginine concentrations are low (46), PBT inhibition of arginase activity may be in part responsible for stimulating a tumor responsive immune response. Ongoing studies are investigating possible mechanisms by which PBT reverses the immunosuppressive microenvironment in tumors to boost protection against tumor growth and perhaps induces tumor-infiltrating cytotoxic T cells.

Targeting polyamine metabolism has long been an attractive approach to cancer chemotherapy. However, treatment with DFMO has had only moderate success in treating cancer patients (23). Subsequent studies highlighted that DFMO-inhibition of ODC leads to upregulation of the polyamine transporter with resulting increased uptake of polyamines from the diet and gut flora into the tumor cells (47, 48). Thus, to polyamine-starve a tumor, both inhibition of endogenous polyamine biosynthesis as well as uptake of exogenous polyamines must be achieved. Interestingly, the therapeutic efficacy of DFMO was reported to be much greater in mice bearing mutant L1210 leukemia cells deficient in their polyamine transport system than in mice injected with parental L1210 tumor cells (49). Subsequently, Moulinoux et al. (20, 50) reported a 40% inhibition in tumor progression and metastases and an antitumor immune stimulation in animals treated with DFMO and fed a polyamine deficient diet and neomycin in the drinking water to partially reduce gut bacterial polyamine production. The discovery of an effective, non-toxic inhibitor of the polyamine transport system (AMXT 1501) offers a direct strategy to block the polyamine transport system that is upregulated in many malignant tumors. PBT is a novel approach to selectively starve tumors via the combined use of a key inhibitor of polyamine biosynthesis and a potent inhibitor of the polyamine transport system. Previous studies with PBT have shown complete regression of carcinogen-induced squamous cell carcinomas in ODC transgenic mice (24). Whereas these skin tumors were completely dependent on polyamine biosynthesis driven by a strong ODC transgene, we demonstrate here the power of PBT to suppress malignant tumor growth in wild type, immunocompetent animals. Our expectation for increased tumor association of AMXT 1501 following polyamine biosynthesis inhibition by DFMO in animals will be the subject of future studies.

The capacity of PBT to target both polyamine biosynthesis as well as the polyamine transport system effectively deprives tumor cells of polyamines necessary for their growth and survival. Treatment with DFMO alone is cytostatic, not cytotoxic in tumors (23, 51). We have shown that PBT inhibition of tumor growth was accompanied by increased apoptosis but no significant decrease in the proliferation index in tumor cells, indicating that the antitumor efficacy of PBT depends more on its ability to re-condition the tumor microenvironment than its activity as an anti-proliferative agent. With the use of PBT that more effectively starves tumors of polyamines compared to DFMO alone, the essential function of tumor polyamines as immunosuppressive modifiers can for the first time be appreciated along with the commonly viewed pro-proliferation role of polyamines in tumor development. The higher polyamine levels found in most tumor types reflects their common requirement for polyamines to mold a tumor microenvironment that suppresses antitumor immune responses and stimulates tumor proliferation, survival, and angiogenesis. Thus, PBT immunochemotherapy offers a much needed general approach to treat multiple types of cancer. Polyamine deprivation via PBT offers exciting potential as adjunct cancer treatment both with conventional chemotherapeutic agents and in stimulating antitumor immune responses in tumor immunotherapies.