Inhibition of myeloid cell differentiation in cancer: the role ... - CiteSeerX

Inhibition of myeloid cell differentiation in cancer: the role of reactive oxygen species Sergei Kusmartsev and Dmitry I. Gabrilovich H. Lee Moffitt Cancer Center, University of South Florida, Tampa

Abstract: It is well established that tumor growth is associated with accumulation of immature myeloid cells (ImC). They play an important role in tumor-associated immune suppression. ImC accumulate not only in tumor-bearing hosts but also in immunized, tumor-free hosts or hosts infected with bacterial pathogens. The kinetics of ImC in these mice is different. If in tumor-bearing mice, the number of ImC continues to increase with tumor progression in tumor-free mice after an initial spike, it decreases to the control level. Here, we have investigated the mechanisms of ImC accumulation in tumor-bearing hosts by comparing differentiation of ImC obtained from tumor-free and tumor-bearing mice. In the presence of appropriate growth factors, ImC isolated from tumor-free mice quickly differentiated in vitro into mature dendritic cells (DC), macrophages, and granulocytes. In contrast, differentiation of ImC from tumor-bearing mice was significantly delayed. Similar results were obtained in vivo after adoptive transfer of ImC into naı¨ve, congeneic mice. ImC transferred into tumor-bearing recipients failed to differentiate into DC or macrophages. ImC from tumor-bearing mice had significantly higher levels of reactive oxygen species (ROS) than ImC obtained from tumor-free mice. Hydrogen peroxide (H2O2) but not superoxide radical anions was found to be the major part of this increased ROS production. In vitro experiments demonstrated that scavenging of H2O2 with catalase induced differentiation of ImC from tumor-bearing mice into macrophages. Thus, this is a first demonstration that tumors may prevent differentiation of antigenpresenting cells by increasing the level of endogenous H2O2 in immature myeloid cells. J. Leukoc. Biol. 74: 186 –196; 2003.

MATERIALS AND METHODS

Key Words: macrophages 䡠 dendritic cells 䡠 cell differentiation

Mice and tumors

increased in peripheral blood of cancer patients [3, 4]. Consistent with these findings were recent data on the accumulation of activated granulocytes in peripheral blood of cancer patients [5]. ImC and activated granulocytes directly suppressed T cell responses in cancer patients, suggesting that these cells may play an important role in tumor nonresponsiveness [5, 6]. In mice, tumor growth induces elevated numbers of immature myeloid cells that coexpress Gr-1 and CD11b surface markers [7–10]. These cells express major histocompatibility complex (MHC) class I molecules but do not express markers of mature APCs, MHC class II, or costimulatory molecules. These cells play a critical role in tumor-associated, immune defects. They directly inhibit antigen-specific T cell responses via direct cell– cell contact [7, 9, 10]. Increased production of ImC is a phenomenon that is not limited to cancer. An increase in the Gr-1⫹CD11b⫹ population in the spleen was also demonstrated in tumor-free mice after administration of potent antigen [11] or in mice infected with bacterial pathogens [12, 13]. However, if T cell dysfunction is a common finding in cancer, it is usually absent in immunized individuals or patients with bacterial infections. To explain those differences in T cell dysfunction, we compared the level of ImC accumulation in tumor-bearing and tumor-free, immunized mice and investigated the differentiation of these cells in vitro and in vivo. We demonstrated here, for the first time, that differentiation of ImC from tumor-bearing mice was significantly delayed in comparison with ImC obtained from tumor-free mice. We also showed that differentiation of ImC was halted as a result of hyperproduction of reactive oxygen species (ROS) in these cells. Neutralization of ROS dramatically improved differentiation of ImC, which may suggest a new approach to enhance an antitumor-immune response.

Female BALB/c and C57BL/6 mice (6 – 8 weeks of age) were obtained from the National Cancer Institute (Frederick, MD). B6.SJL-PtrcaPep3b/BoyJ mice

INTRODUCTION Recent studies have demonstrated that tumor growth is closely associated with impaired differentiation of professional antigen-presenting cells (APCs), particularly dendritic cells (DC), and with increased production of immature myeloid cells (ImC; reviewed in refs. [1, 2]). The presence of ImC was dramatically 186

Journal of Leukocyte Biology Volume 74, August 2003

Correspondence: Dmitry Gabrilovich, M.D., Ph.D., H. Lee Moffitt Cancer Center, University of South Florida, MRC-2, Room 2067, 12902 Magnolia Dr., Tampa, FL 33612. E-mail: [email protected] Received January 9, 2003; revised April 1, 2003; accepted April 7, 2003; doi: 10.1189/jlb.0103010.

http://www.jleukbio.org

(CD45.1⫹) were obtained from Jackson Laboratories (Bar Harbor, ME). CT-26 colon carcinoma was established in BALB/c mice by subcutaneous (s.c.) inoculation of 3 ⫻ 105 tumor cells and C3 sarcoma, in C57BL/6 mice by s. c. inoculation of 5 ⫻ 105 tumor cells. Tumor-free C57BL/6 or BALB/c mice were immunized s.c. with 500 ␮g ovalbumin (OVA) protein (Sigma Chemical Co., St. Louis, MO) or 100 ␮g OVA-derived peptide (SIINFEKL, SynPep Corp., Dublin, CA), emulsified in complete Freund’s adjuvant (CFA; Sigma Chemical Co.).

Media and reagents RPMI-1640 medium was supplemented with 10% fetal bovine serum (FBS), 20 mM HEPES, 200 units/ml penicillin, 50 ␮g/ml streptomycin, 0.05 mM 2-mercaptoethanol, and 2-mM glutamine (all from Life Technologies, Grand Island, NY). Murine recombinant granulocyte macrophage-colony stimulating factor (GM-CSF), M-CSF, and G-CSF were purchased from RDI (Flanders, NJ). Oxidant-sensitive dyes, dichlorodihydrofluoresceine acetate (DCFDA) and dihydroethytium (DHE), were obtained from Molecular Probes (Eugene, OR). Catalase and superoxide dismutase (SOD) were purchased from Calbiochem (San Diego, CA). The following antibodies were used for flow cytometry: CD45.2 (clone 104), Gr-1 (clone RB6-8C5), CD11b (clone M1/70), CD11c (clone HL3), I-Ab (clone AF6-120.1), I-Ad (clone AMS-32.1), CD86 (clone GL1), and B220 [clone RA3-6B2, all from BD PharMingen (San Diego, CA), and F4/80 from Serotec (Raleigh, NC)].

Gr-1⫹ cell isolation and culture Spleens of tumor-bearing or immunized, tumor-free mice were used as a source of Gr-1⫹ cells. Single-cell suspensions were prepared, and red cells were removed using acetate kinase lysing buffer. Splenocytes were resuspended in phosphate-buffered saline (PBS), and 5– 6 ⫻ 106 cells were incubated with 5 ␮g biotinylated anti-Gr-1 monoclonal antibodies (BD PharMingen) for 15 min on ice. Cells were washed with cold PBS twice and then incubated with streptavidin microbeads for 15 min at 4°C. The Gr-1⫹ cell population was isolated using MiniMACS columns (Miltenyi Biotec GmbH, Auburn, CA). The purity of the Gr-1⫹ cell population was evaluated by flow cytometry and exceeded 90%. Isolated Gr-1⫹ cells were resuspended in RPMI-1640 medium supplemented with 10% FBS and different cytokines: GM-CSF (20 ng/ml), M-CSF (20 ng/ml), or G-CSF (20 ng/ml), plated into 24-well plates and cultured for 3–7 days. On days 3 and 5 after culture initiation, half of the medium was replaced with fresh cytokine-supplemented medium.

Flow cytometry One million cells were incubated for 30 min on ice in 100 ␮l PBS with 1 ␮g relevant antibodies and were then washed twice with cold PBS. Flow cytometry data were acquired using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) and were analyzed using CellQuest software (BD Biosciences).

Cell proliferation Proliferation of purified Gr-1⫹ cells was determined by [3H] thymidine incorporation. Cells (1⫻103) in complete RPMI medium were placed into 96-well round-bottom plates and cultured for different times. Each well was pulsed with 1 ␮Ci [3H] thymidine for the final 8 h of incubation. Cells were harvested onto filters, and radioactivity was measured in a liquid scintillation counter.

Analysis of ROS production The oxidation-sensitive dyes DCFDA and DHE were used for the measurement of ROS production by Gr-1⫹ cells. Cells were incubated at 37°C in Dublecco’s modified Eagle’s medium in the presence of 2 ␮M DCFDA for 30 min or 2 ␮M DHE for 60 min, washed twice with cold PBS, and then labeled with APCconjugated anti-Gr-1 antibody and phycoerythrin (PE)-conjugated anti-CD11b antibodies. After incubation on ice for 20 min, cells were washed with cold PBS and analyzed by three-color flow cytometry using FACSCalibur.

Statistical analysis The statistical significance between values was determined by the Student’s t-test. All data were expressed as the mean ⫾ SD. Probability values ⬎ 0.05 were considered nonsignificant.

RESULTS Production of ImC was evaluated in two different tumor models: The CT-26 tumor was established in BALB/c mice and the C3 tumor, in C57BL/6 mice. Mice were killed 10 days and 21 days after inoculation of tumors when tumor sizes reached 1–1.5 cm in diameter. The presence of Gr-1⫹CD11b⫹ ImC in spleens was analyzed using flow cytometry. In parallel, BALB/c or C57BL/6 mice were immunized once with OVA emulsified in CFA. In some experiments, C67BL/6 mice were immunized with an OVA-derived H2Kb-restricted peptide (SIINFEKL). Our preliminary studies have demonstrated that the peak of ImC production occurred on days 8 –11 after immunization. Immunized mice were also killed on days 10 and 21 after immunization, and the level of ImC was compared with that in tumor-bearing mice. Figure 1 illustrates the results of the experiments in C57BL/6 mice. Experiments with BALB/c mice produce similar results (data not shown). Ten days after inoculation of the C3 tumor, the proportion and absolute number of Gr-1⫹CD11b⫹ ImC significantly increased. The proportion and absolute number of ImC in immunized mice on day 10 after immunization were comparable with that in tumor-bearing mice (Fig. 1). Three weeks after tumor inoculation, the presence of ImC in tumor-bearing mice further increased. The proportion of these cells was almost tenfold higher and the absolute number, more than 20-fold higher than those in control mice. In contrast, 3 weeks after immunization with OVA or peptide, the presence of ImC in tumor-free mice returned to normal levels (Fig. 1).

In vivo differentiation of Gr-1⫹ ImC We asked whether ImC derived from tumor-bearing and immunized, tumor-free hosts had equal ability to differentiate into mature myeloid cells. To study ImC differentiation in vivo, Gr-1⫹ cells were isolated from C3 tumor-bearing C57BL/6 mice or immunized with 0.5 mg OVA in CFA tumor-free mice. Splenocytes from C57BL/6 mice have a CD45.2⫹ phenotype. Three million Gr-1⫹ cells were injected intravenously (i.v.) into congenic CD45.1⫹ mice. On days 3 and 5 after adoptive transfer, the recipients’ spleens were isolated, and the donors’ CD45.2⫹ splenocytes were analyzed by multicolor flow cytometry (Fig. 2). Very few CD45.2⫹ cells were detected in lymph nodes of the recipients, even 5 days after the cell transfer, which makes accurate analysis of the donors’ cells in lymph nodes practically impossible. In the first series of experiments, purified Gr-1⫹ splenocytes from tumor-bearing or immunized, tumor-free mice were transferred into naı¨ve, congeneic recipients. Three days after adoptive transfer, the donors’ CD45.2⫹ cells represented 1.1 ⫾ 0.3% of nucleated cells in the recipients’ spleens. Five days after the transfer, their presence was slightly reduced to 0.8 ⫾ 0.3% (P⬎0.1). No differences in the presence of the donors’ CD45.2⫹ cells were found between immunized, tumor-free and tumor-bearing mice at both time points. Myeloid cells isolated from immunized, tumor-free mice quickly lost their immature phenotype (Gr-1⫹CD11b⫹), whereas almost 30% of donors’ cells isolated from tumor-bearing mice remained Gr1⫹CD11b⫹ 5 days after adoptive transfer (Fig. 3A). Most of

Kusmartsev and Gabrilovich ROS prevent differentiation of immature myeloid cells in cancer

187

Fig. 1. Accumulation of Gr-1⫹CD11b⫹ ImC in spleens of tumor-bearing and tumor-free mice. C3 tumors were inoculated into naı¨ve C57BL/6 mice as described in Materials and Methods. For immunization, naı¨ve mice were injected s.c. into the flank with OVA protein (0.5 mg/mouse) or OVA-derived peptide SIINFEKL (100 ␮g/mouse) emulsified in 100 ␮l CFA. Splenocytes were isolated on days 10 and 21 after immunization or tumor inoculation. (Upper) Cumulative results (average⫾SD) from five mice in each group. (Lower) Actual data from one representative experiment.

the ImC differentiated in vivo into CD11c⫹IAb⫹B7-2⫹ DC or F4/80⫹ macrophages. The next group of experiments investigated the fate of Gr-1⫹ cells transferred into tumor-bearing hosts. Gr-1⫹ cells isolated from immunized, tumor-free or tumor-bearing mice were injected i.v. into the CD45.1⫹ congenic, mice-bearing, syngeneic C3 tumor (18 –20 days after inoculation; tumor size, 1–1.5 cm in diameter). After adoptive transfer, no differences in the proportion of the CD45.2⫹ donors’ cells in the recipients’ spleens were found between cells isolated from tumor-bearing and tumor-free mice (data not shown). Three days after adoptive transfer of Gr-1⫹cells isolated from immunized, tumor-free mice, only few double-positive ImC (Gr-1⫹CD11b⫹) were 188


found among the donors’ CD45.2⫹ cells, whereas after transfer of Gr-1⫹ cells isolated from tumor-bearing mice, almost 40% of the donors’ cells expressed markers of ImC (Fig. 3B). By day 5, Gr-1⫹CD11b⫹ cells represented less than 5% of the donors’ cells in both groups. Only few donors’ cells had a phenotype of DC (CD11c⫹IAb⫹) 5 days after transfer of ImC from tumorbearing mice, whereas more than 1/4 of donors’ cells had a DC phenotype after the transfer of Gr-1⫹ cells from immunized, tumor-free mice (5.1⫾1.2% and 28.4⫾3.4%, respectively; P⬍0.05; Fig. 3B). Transfer of ImC isolated from immunized, tumor-free mice into naı¨ve and tumor-bearing recipients revealed a clear shift in the direction of myeloid cell differentiation. In naı¨ve reciphttp://www.jleukbio.org

Fig. 2. Scheme and the example of a typical experiment with in vivo differentiation of Gr-1⫹ cells. Detailed results are provided in Figures 3–5. FITC, fluorescein isothiocyanate.

ients, by day 5, most of transferred Gr-1⫹ cells differentiated into DC (50%) and macrophages (30%) with a relatively minor percentage of Gr-1⫹CD11b– cells (10%; Fig. 4). Almost all DC expressed a costimulatory molecule B7-2, which reflects their relatively mature state. In contrast, in tumor-bearing recipients, less than 30% of the donors’ cells became CD11c⫹IAd DC, and less than 5% of cells were CD11c⫹B7-2⫹ DC. Only 10% of donors’ cells became macrophages, and more than 50% of donors’ cells had a Gr-1⫹CD11b– phenotype (Fig. 4).

Immature myeloid cells in site of the tumor We have also evaluated the fate of ImC in the recipients’ tumor site. Three and 5 days after the transfer of Gr-1⫹ cells into congeneic tumor-bearing hosts, tumors were excised and digested for 30 min with collagenase. Cells were collected, washed, labeled with antibodies, and analyzed by flow cytometry. Three days after the transfer of Gr-1⫹ cells from immunized, tumor-free mice, the donors’ CD45.2⫹ cells represented 0.28 ⫾ 0.2% of the total cells in the tumor site. This proportion was slightly increased by day 5 to 0.9 ⫾ 0.3%. Three days after the transfer of Gr-1⫹ cells from tumor-bearing mice, the proportion of the donors’ CD45.2⫹ cells in the tumor site was

0.8 ⫾ 0.3%. It remained the same 2 days later (0.8⫾0.3%). Most of the Gr-1⫹ donors’ cells (60%) that were able to reach the tumor site retained an immature phenotype (Gr1⫹CD11b⫹; Fig. 5). About 40% of these cells acquired a F4/80 antigen and became Gr-1⫹CD11b⫹F4/80⫹ triple-positive cells (data not shown). Very few donors’ cells transferred from immunized, tumor-free mice had a phenotype of Gr1⫹CD11b– cells, whereas a significantly higher proportion of these cells was found after the transfer of Gr-1⫹ cells from tumor-bearing mice (Fig. 5). Mature CD11c⫹B7-2⫹ DC were practically undetectable in the tumor site after the transfer of Gr-1⫹ isolated from tumor-bearing mice. In contrast, a substantial proportion of DC derived from donors’ cells could be found after the transfer of Gr-1⫹ cells from immunized, tumorfree mice (Fig. 5). Taken together, our data demonstrated that ImC from tumorbearing mice had an impaired ability to differentiate into mature myeloid cells. This was evident after the transfer of these cells into a tumor-free host. This effect became more prominent when cells were transferred into a tumor-bearing host. What could be the mechanisms of the impaired ability of ImC to differentiate into mature myeloid cells in cancer? It is known that ROS play an important role in the function of


189

Fig. 3. Differentiation of Gr-1⫹ cells transferred into a congenic recipient. Analysis of recipients’ splenocytes. Gr-1⫹ splenocytes were purified from immunized, tumor-free (open bars) or C3 tumor-bearing CD45.2⫹ C57BL/6 (shaded bars) mice and were transferred i.v. into (A) naı¨ve or (B) C3 tumor-bearing CD45.1⫹ congenic mice (3⫻106 cells/mouse). On days 3 and 5 after transfer, recipient mice were killed, and the presence of myeloid cells within CD45.2⫹ donor cell population in spleens was evaluated by multicolor flow cytometry. Day 0, Purified Gr-1⫹ cells before transfer. Each group included three mice. Average ⫾ SD is shown. *, Statistically significant differences between the groups (P⬍0.05).

190



Fig. 4. Comparative analysis of donor cells transferred into naive versus tumor-bearing recipients. Gr-1⫹ splenocytes purified from immunized, tumorfree mice were transferred i.v. into naı¨ve or C3 tumor-bearing, congenic recipients. On day 5 after the transfer recipient mice were killed, the presence of myeloid cells within the CD45.2⫹ donors’ cell population in the spleen was evaluated by multicolor flow cytometry. Each group included three mice. *, Statistically significant differences between the groups (P⬍0.05).

myeloid cells, and it has also been documented that tumor cells have increased ROS production. Here, we investigated the possible role of ROS in differentiation of ImC.

ROS production in Gr-1⫹ cells To measure ROS generation by myeloid cells, we used two dyes: DHE and DCFDA. DHE is selectively oxidized by superoxide anion, and the fluorescence of DCFDA indicates oxidation by hydrogen peroxide (H2O2), peroxynitrite, or hydroxyl radical. Superoxide anions also can contribute to DCFDA oxidation, albeit at a lesser degree. Freshly isolated splenocytes from tumor-bearing or from tumor-free, immunized mice were loaded with these dyes and

then labeled with anti-Gr-1-APC and CD11b-PE antibodies. The fluorescence of those dyes was evaluated within the population of gated double-positive Gr-1⫹CD11b⫹ myeloid cells. No difference in superoxide production (DHE oxidation) was found between two groups of cells, whereas the level of DCFDA oxidation by ImC from tumor-bearing mice was significantly (threefold) higher than their counterparts from immunized, tumor-free mice (Fig. 6A). We compared the levels of DCFDA-mediated fluorescence in Gr-1⫹CD11b⫹ ImC and Gr1–CD11b⫹ macrophages in the same spleens. ImC generated three- to fourfold more ROS than Gr-1–CD11b⫹ macrophages (Fig. 6B). Next, we investigated the nature of ROS produced by ImC. Superoxide and H2O2 are the main factors contributing to ROS activity. To evaluate their role, we used specific inhibitors: SOD and catalase. Catalase dramatically reduced the ROS level in tumor-bearing, mice-derived ImC (more than fourfold), indicating that H2O2 contributed greatly into the overall level of ROS in these cells. As expected, SOD did not significantly affect the levels of ROS in these cells (Fig. 7).

Involvement of endogenous H2O2 in differentiation of Gr-1⫹ cells As Gr-1⫹ cells derived from tumor hosts produce elevated levels of H2O2, we asked whether neutralization of H2O2 might affect differentiation of these cells. Gr-1⫹ ImC were isolated from immunized, tumor-free or tumor-bearing mice. Flow cytometry after isolation has shown that more than 92% of all cells had the phenotype Gr-1⫹CD11b⫹F4/80–, consistent with the phenotype of ImC (data not shown). Neither ImC from tumor-bearing nor from immunized, tumor-free mice survived more than 72 h culture without the presence of cytokines (data not shown). Therefore, cells were cultured with 20 ng/ml GM-CSF to maintain their viability. After 7 days of culture in the presence of GM-CSF, most of ImC from immunized mice differentiated into CD11b⫹F4/80⫹ macrophages

Fig. 5. Differentiation of Gr-1⫹ cells transferred into a tumor-bearing host. Analysis of recipients’ tumor-infiltrated cells. Gr-1⫹ splenocytes were purified from immunized, tumor-free (open bars) or C3 tumor-bearing CD45.2⫹ C57BL/6 (shaded bars) mice and were transferred i.v. into C3 tumor-bearing CD45.1⫹ congenic mice (3⫻106 cells/mouse). On days 3 and 5 after the transfer recipient mice were killed, tumors were excised, connective tissue digested, and single-cell suspension prepared. The presence of myeloid cells within the CD45.2⫹ donor cell population in the tumor site was evaluated by multicolor flow cytometry. Each group included three mice. Average ⫾ SD is shown. *, Statistically significant differences between the groups (P⬍0.05).


191

Fig. 6. Gr-1⫹CD11b⫹ myeloid cells from tumor-bearing mice demonstrate an increased level of ROS. Splenocytes from tumor-bearing mice were incubated in serum-free medium at 37°C in the presence of DCFDA (2 ␮M, 30 min) or DHE (2 ␮M, 60 min), washed with cold PBS, and then labeled with Gr-1-APC and CD11b-PE antibodies. After incubation on ice for 20 min, cells were washed and analyzed by three-color flow cytometry. Two experiments with similar results were performed. The intensity of fluorescence (Geo Mean) in a gated population of cells for each histogram is shown.

and CD11c⫹IAd⫹ DC (Table 1). Only relatively few cells (9.4⫾1.8%) remained ImC with phenotype Gr-1⫹CD11b⫹F4/ 80–. A substantially lower proportion of macrophages and a slightly lower proportion of DC were found after 7 days culture of ImC from tumor-bearing mice (Table 1). A substantial proportion of these cells retained the phenotype of immature myeloid cells (23.1⫾4.5%, P⬍0.05). These data are consistent with the results of in vivo experiments, indicating that Gr-1⫹ ImC from tumor-bearing mice had decreased the ability to differentiate into macrophages or DC in comparison with their counterparts from tumor-free, immunized mice. To evaluate the role of H2O2 in ImC differentiation, Gr-1⫹ splenocytes isolated from tumor-bearing mice were cultured in the presence of GM-CSF and catalase. First, we analyzed the possible toxic effects of catalase. Cell viability and apoptosis were measured using double-staining with AnnexinV and aminoactinomycin D (7-AAD). At a range of concentrations known to exert a biological effect (500 –2000 U/ml), catalase did not affect the viability of Gr-1⫹ cells during the first 2 days in culture. Starting from day 3, catalase at concentrations 1000 U/ml and 2000 U/ml induced apoptosis in Gr-1⫹cells and slightly decreased the proportion of viable cells (Fig. 8, A and B). Prolonged incubation of cells with GM-CSF and catalase at these concentrations decreased cell viability even further (Fig. 192


8A). During the first 4 days in culture with GM-CSF, catalase at a dose of 500 U/ml did not affect the viability of Gr-1⫹ cells (Fig. 8A). Based on these preliminary data, we used 500 U/ml catalase during the first 4 days of culture in further experiments. GM-CSF induced substantial proliferation of ImC seen clearly on days 3 and 4 of culture. The presence of SOD only slightly decreased cell proliferation, whereas catalase completely inhibited this proliferation (Fig. 8C). Neutralization of endogenous H2O2 by catalase during the first 4 days in culture dramatically changed GM-CSF-induced differentiation of immature myeloid cells. After 4 days of culture in GM-CSF alone, 50% of the ImC isolated from CT-26 tumor-bearing mice retained the ImC phenotype (Gr-1⫹CD11b⫹F4/80–). The proportion of relatively mature macrophages was slightly more than 15%, and CD11c⫹IAd⫹ DC was ⬃20% (Fig. 8D). Addition of catalase dramatically reduced the proportion of ImC (from 50% to 15%, P⬍0.01) and dramatically increased the macrophage cell population from 15% to almost 60% (Fig. 8D). SOD did not significantly affect the differentiation of ImC (Fig. 8D). Similar effects were observed in the other animal tumor models, Gr-1⫹ ImC isolated from C3 tumor-bearing C57BL/6 mice (Fig. 8E). In both models, inhibition of H2O2 prevented differentiation to CD11c⫹ MHC class II⫹ DC. Similar effects of http://www.jleukbio.org

Fig. 7. Catalase decreases ROS production by immature myeloid cells. Purified Gr-1⫹ splenocytes were preincubated with catalase (500 U/ml), SOD (300 U/ml), or medium alone for 15 min at room temperature before adding DCFDA (2 ␮M). After 30 min incubation in the presence of DCFDA, cells were washed with cold PBS and analyzed by flow cytometry. A typical result of one out of three performed experiments is shown.

were observed after treatment of ImC isolated from immunized, tumor-free mice with catalase (data not shown).

DISCUSSION It is well-established that tumor growth caused dramatic expansion of the population of immature myeloid cells [3, 6, 7, 9, 14, 15]. In mice, these cells have a phenotype of GrTABLE 1.

Source of ImC Immunized, tumor-free mice (N ⫽ 3) Tumor-bearing mice (N ⫽ 3)

1⫹CD11b⫹ cells, and in mice and humans, these cells have been shown to directly inhibit T cell responses [10, 11, 16] and therefore, may play an important role in tumor nonresponsiveness. Their presence may also be responsible for the failure of cancer vaccines. However, a significant increase in the presence of ImC has also been reported under the conditions usually not associated with immune suppression: immunization with potent antigens and bacterial infection [11–13]. We asked what could explain this paradox. Our experiments demonstrated that 10 days after tumor inoculation (in two tested animal models) or immunization of tumor-free mice with potent antigens (OVA protein or peptide in CFA), the number of ImC was equally increased. The differences between these mice appeared at a later time. In tumor-bearing mice, the presence of ImC further increased and 10 days later, was more than 20-fold higher than in control mice. In contrast, the number of ImC in immunized mice reversed to the normal level by that time. There are two not mutually exclusive explanations of these data: ImC from tumor-free and tumor-bearing mice may differ in their ability to differentiate into mature myeloid cells, or tumor-bearing mice produce factors stimulating myelopoiesis, which increase with tumor growth. In immunized mice, the production of these factors by macrophages, lymphocytes, and endothelial cells, among others, is limited in time by the presence of antigens, which may explain the differences noted in accumulation of ImC. To evaluate differentiation of ImC in vivo, we used adoptive transfer of Gr-1⫹ cells to congenic recipients. Differentiation of ImC obtained from tumor-free mice was severely impaired after the transfer into tumor-bearing recipients. Differentiation of DC and macrophages was significantly inhibited, and most of the remaining cells expressed a Gr-1⫹CD11b– phenotype, which is characteristic of granulocytes. These data were consistent with previously published observations that tumor-derived factors inhibit differentiation of DC from hematopoietic progenitor cells [17–21]. It appears that tumor-derived factors may affect later stages of myeloid cell differentiation as well. A very high (more than 60%) proportion of donor cells found in the tumor site was Gr-1⫹CD11b⫹ ImC. All these experiments were performed on splenocytes. Accurate analysis of donors’ cells in lymph nodes was technically impossible as a result of the fact that very few donors’ cells migrated into lymph nodes after i.v. injection. However, our preliminary data demonstrated a significant increase in the presence of ImC in lymph nodes of tumor-bearing mice (S. Kusmartsev and D. Gabrilovich, unpublished observation). In addition, a dramatic increase of Gr-1⫹ cells was found in lymph nodes from mice

Proportion of Cells after 7-Day Differentiation of ImC In Vitro Gr-1⫹ CD11b⫹F4/80– ImC

CD11c⫹IAd⫹ DC

CD11b⫹F4/80⫹ macrophages

9.4 ⫾ 1.8% 23.1 ⫾ 4.5%, P ⬍ 0.05

14.8 ⫾ 2.1% 9.2 ⫾ 1.5%, P ⬍ 0.05

55.4 ⫾ 4.6% 34.5 ⫾ 2.3%, P ⬍ 0.05

Gr-1⫹ ImC were isolated from control or tumor-bearing mice and were cultured in vitro for 7 days with 20 ng/ml GM-CSF in complete culture medium. Cells were collected and labeled with appropriate antibodies as described in Materials and Methods. A proportion of cells was analyzed using multicolor flow cytometry and was presented as average ⫾ SD. Three experiments were performed.


193

Fig. 8. Scavenging H2O2 inhibits GM-CSF-driven proliferation and induces differentiation of Gr-1⫹ cells into F4/80⫹ macrophages. Gr-1⫹ splenocytes were purified from tumor-bearing mice. All cells were cultured in the presence of GM-CSF (20 ng/ml) for 4 –7 days. (A) Viability of Gr-1⫹ cells cultured in the presence of catalase. On days 1– 4 and 7, cells were recovered, washed, and stained with Annexin V and 7-AAD. Viability was evaluated by flow cytometry. (B) Apoptosis of Gr-1⫹ cells cultured in the presence of catalase. Cells were collected on days 1– 4, washed, and stained with Annexin V and 7-AAD. Apoptosis was evaluated by flow cytometry as the proportion of Annexin V-positive, 7-AAD-negative cells. (C) Catalase suppresses GM-CSF-induced proliferation. Gr-1⫹ cells were cultured for 1– 4 days in the presence of 500 U/ml catalase and 200 U/ml SOD. Each well was pulsed with 1 ␮Ci [3H] thymidine for the final 8 h of incubation. Cells were harvested, and radioactivity was measured using a scintillation counter. CPM, Counts per minutes. (D) Catalase increases the proportion of macrophages and inhibits GM-CSF-induced DC differentiation. Gr-1⫹ cells isolated from CT-26 tumor-bearing BALB/c mice were cultured in the presence of 500 U/ml catalase or 200 U/ml SOD. On day 4, cells were recovered, washed, and stained with Gr-1-APC, CD11b-FITC, F4/80-PE, as well as CD11c-APC and IAd-PE. After incubation on ice, expression of indicated markers was analyzed by flow cytometry. Results of three experiments are shown. *, Statistically significant differences between the groups (P⬍0.05). (E) Gr-1⫹ cells were isolated from C3 tumor-bearing C57BL/6 mice and were cultured and analyzed as described in D. Preliminary experiments demonstrated that Gr-1⫹ cells from C57BL/6 mice were more resistant to catalase than the cells from BALB/c mice. Therefore, a higher concentration of catalase (3000 U/ml) was used. The viability of C57BL/6 myeloid cells incubated with 3000 U/ml catalase was equal to the viability of BALB/c cells incubated with 500 U/ml catalase.

treated with tumor-derived factor, vascular endothelial growth factor [22]. Taken together, these data suggest that a similar process of ImC differentiation is taking place in spleen and lymph nodes. Thus, our experiments confirmed that in the presence of tumor-derived factors, differentiation of ImC accumulated in tumor-bearing hosts is severely affected. However, our data also demonstrated that ImC from tumor-free and tumor-bearing mice significantly differ in their ability to differentiate in a tumor-free environment. This was evident from the experi194


ments in vitro, where after 7 days in culture with GM-CSF, a substantially higher proportion of cells isolated from tumorbearing mice retained their immature phenotype when compared with cells isolated from tumor-free mice. The proportion of DC and macrophages generated from tumor-bearing micederived ImC was substantially lower than those differentiated from tumor-free mice-derived ImC. Similar results were obtained in an experiment in vivo. Almost all ImC obtained from immunized, tumor-free mice differentiated into mature DC and macrophage within 5 days after the transfer into congenic mice. http://www.jleukbio.org

In contrast, a substantial proportion of ImC derived from tumor-bearing mice retained the phenotype of immature cells (Gr-1⫹CD11b⫹), and differentiation of macrophages was significantly decreased. After the transfer into tumor-bearing recipients, ImC from tumor-bearing mice practically failed to differentiate into DC, whereas a substantial proportion of ImC from immunized, tumor-free mice generated DC. These data suggested that differentiation of ImC obtained from tumorbearing mice was markedly impaired. What could be the mechanisms of this defect? Myeloid cell function is closely associated with production of ROS [23]. Previous studies have shown that oxidative stress by activated granulocytes in cancer patients [5] or by tumor-derived macrophages in mice [16] contributes to the suppression of T cell function. However, the possible role of ROS in differentiation of myeloid cells has not been identified yet. Our experiments demonstrated that ImC isolated from immunized, tumor-free mice had significantly higher levels of ROS than ImC isolated from tumor-free, immunized mice. Gr-1⫹CD11b⫹ ImC had a significantly higher level of ROS than Gr-1–CD11b⫹ macrophages. An increase of the specific oxidation of DCFDA but not DHE indicated the preferential contribution of H2O2 to the ROS pool in tumor-bearing mice-derived Gr-1⫹ splenocytes. This fact was confirmed in direct experiments where ROS was inhibited by SOD and catalase. SOD, which neutralizes superoxide, did not decrease the level of ROS in these cells, whereas catalase, which neutralizes H2O2, decreased the ROS level more than fourfold. Apparently, in ImC, superoxide is quickly reduced to H2O2, thus diminishing its contribution to the total ROS pool. It is possible that these cells may have high endogenous SOD activity or decreased catalase activity. We are currently investigating these possibilities. We hypothesized that increased levels of ROS might interfere with differentiation of these cells and could be responsible for the impaired differentiation of ImC isolated from tumorbearing mice. To test this hypothesis, we incubated Gr-1⫹ cells isolated from tumor-bearing mice with GM-CSF and catalase or SOD. As expected, GM-CSF induced proliferation of Gr-1⫹ ImC, which was clearly evident after 3 and 4 days in culture. SOD slightly decreased that proliferation, whereas catalase, at nontoxic concentrations, completely blocked cell proliferation. These results, although novel, are not entirely unexpected. It is known that H2O2 at low concentrations induces cell proliferation via an increase of Ca2⫹ influx and activation of Ras and extracellular-regulated kinase 1/2 pathways [23–25]. As GMCSF is known to stimulate ROS production [26], it is conceivable to suggest that neutralization of H2O2 could result in inhibition of GM-CSF-induced proliferation of ImC. Neutralization of H2O2 not only inhibited proliferation of ImC but also stimulated their differentiation. After a 4-day culture of ImC from tumor-bearing mice, the proportion of Gr-1⫹CD11b⫹F4/ 80– ImC decreased more than threefold, and the proportion of F4/80⫹ macrophages increased more than threefold. It is interesting that neutralization of H2O2 blocked differentiation of DC. The mechanism of this process is unclear. It is known that ROS regulate transcription of many genes via their effect on several transcription factors, including nuclear factor (NF)-␬B, activated protein-1, c-myb, specificity protein-1, and others [23]. Hyperproduction of ROS may alter the balance of expres-

sion of different genes, which may affect differentiation of myeloid cells. Specifically, it is known that ROS activate the NF-␬B transcription factor [27], which plays a critical role in differentiation of DC [28, 29]. It is not know whether hyperactivation of NF-␬B affects DC differentiation, but it is established that inhibition of NF-␬B activation blocks DC differentiation [30]. It is possible that in our in vitro experiments, catalase inhibited ROS-inducible NF-␬B activation and thus blocks DC differentiation from ImC. How can a tumor activate ROS in myeloid cells? It is known that a number of cytokines and growth factors induce ROS production. They include interleukin (IL)-1, IL-6, IL-3, tumor necrosis factor ␣, platelet-derived growth factor, transforming growth factor-␤, GM-CSF, and fibroblast growth factor (reviewed in ref. [23]). Tumor cells produce many of those factors. It is likely that hyperproduction of some of these factors by tumor cells may result in constant stimulation of ROS in myeloid cells, which in turn, prevents their effective differentiation. Constant production of these factors in tumor-bearing mice can explain the different fate of ImC transferred into naive and tumor-bearing recipients in our experiments. Apparently, exposure of ImC isolated from immunized, tumor-free mice to tumor-derived factors after adoptive transfer into tumor-bearing recipients resulted in increased ROS production that affected their differentiation. Conversely, ImC isolated from tumor-bearing mice retained increased ROS levels for some time after the transfer into naı¨ve recipients or during in vitro incubation. This may explain the delayed and decreased differentiation of these cells. The factor(s) responsible for the induction of ROS in myeloid cells in cancer are currently under investigation. In conclusion, our study, for the first time, has demonstrated that accumulation of immature myeloid cells in tumor-bearing hosts is in part caused by the inability of these cells to differentiate into mature myeloid cells. Increased production of ROS, specifically H2O2, which is induced by tumor-derived factors, may be responsible for this phenomenon. This may suggest approaches to improve immune response in cancer by neutralization of ROS production.

ACKNOWLEDGMENT This study was supported by NIH grant CA 84488 to D. I. G.

REFERENCES 1. Bronte, V., Serafini, P., Appoloni, E., Zanovello, P. (2001) Tumor-induced immune dysfunctions caused by myeloid suppressor cells. J. Immunother. 24, 431– 446. 2. Kusmartsev, S., Gabrilovich, D. I. (2002) Immature myeloid cells and cancer-associated immune suppression. Cancer Immunol. Immunother. 51, 293–298. 3. Almand, B., Resser, J., Lindman, B., Nadaf, S., Clark, J., Kwon, E., Carbone, D., Gabrilovich, D. (2000) Clinical significance of defective dendritic cell differentiation in cancer. Clin. Cancer Res. 6, 1755–1766. 4. Salvadori, S., Martinelli, G., Zier, K. (2000) Resection of solid tumors reverses T cell defects and restores protective immunity. J. Immunol. 164, 2214 –2220. 5. Schmielau, J., Finn, O. J. (2001) Activated granulocytes and granulocytederived hydrogen peroxide are the underlying mechanism of suppression


195

6.

7.

8.

9. 10. 11.

12.

13. 14. 15. 16.

17.

of T-cell function in advanced cancer patients. Cancer Res. 61, 4756 – 4760. Almand, B., Clark, J. I., Nikitina, E., English, N. R., Knight, S. C., Carbone, D. P., Gabrilovich, D. I. (2001) Increased production of immature myeloid cells in cancer patients. A mechanism of immunosuppression in cancer. J. Immunol. 166, 678 – 689. Bronte, V., Chappell, D. B., Apolloni, E., Cabrelle, A., Wang, M., Hwu, P., Restifo, N. P. (1999) Unopposed production of granulocyte-macrophage colony-stimulating factor by tumors inhibits CD8⫹ T cell responses by dysregulating antigen-presenting cell maturation. J. Immunol. 162, 5728 –5737. Bronte, V., Apolloni, E., Cabrelle, A., Ronca, R., Serafini, P., Zamboni, P., Restifo, N., Zanovello, P. (2000) Identification of a CD11b(⫹)/Gr-1(⫹)/ CD31(⫹) myeloid progenitor capable of activating or suppressing CD8(⫹) T cells. Blood 96, 3838 –3846. Kusmartsev, S., Li, Y., Chen, S. (2000) Gr-1⫹ myeloid cells derived from tumor-bearing mice inhibit primary T cell activation induced through CD3/CD28 costimulation. J. Immunol. 165, 779 –785. Gabrilovich, D. I., Velders, M., Sotomayor, E., Kast, W. M. (2001) Mechanism of immune dysfunction in cancer mediated by immature Gr-1⫹ myeloid cells. J. Immunol. 166, 5398 –5406. Bronte, V., Wang, M., Overwijk, W., Surman, D., Pericle, F., Rosenberg, S. A., Restifo, N. P. (1998) Apoptotic death of CD8⫹ T lymphocytes after immunization: induction of a suppressive population of Mac-1⫹/Gr-1⫹ cells. J. Immunol. 161, 5313–5320. Atochina, O., Daly-Angel, T., Piskorska, D., Harn, D. (2001) A shistosome-expressed immunomodulatory glycoconjugate expands peritoneal Gr1(⫹) macrophages that suppress naı¨ve CD4(⫹) T cell proliferation via an IFN-gamma and nitric oxide-dependent mechanism. J. Immunol. 167, 4293– 4302. Cauley, L., Miller, E., Yen, M., Swain, S. (2000) Superantigen-induced CD4 T cell tolerance mediated by myeloid cells and IFN-gamma. J. Immunol. 165, 6056 – 6066. Young, M. R. I., Newby, M., Wepsic, T. H. (1987) Hematopoiesis and suppressor bone marrow cells in mice bearing large metastatic Lewis lung carcinoma tumors. Cancer Res. 47, 100 –106. Jaffe, M., Arai, H., Nabel, G. (1996) Mechanisms of tumor-induced immunosuppression: evidence for contact-dependent T cell suppression by monocytes. Mol. Med. 2, 692–701. Otsuji, M., Kimura, Y., Aoe, T., Okamoto, Y., Saito, T. (1996) Oxidative stress by tumor-derived macrophages suppresses the expression of CD3 zeta chain of T-cell receptor complex and antigen-specific T-cell responses. Proc. Natl. Acad. Sci. USA 93, 13119 –13124. Gabrilovich, D. I., Chen, H. L., Girgis, K. R., Cunningham, T., Meny, G. M., Nadaf, S., Kavanaugh, D., Carbone, D. P. (1996) Production of vascular endothelial growth factor by human tumors inhibits the functional maturation of dendritic cells. Nat. Med. 2, 1096 –1103.

196


18. Gabrilovich, D. I., Nadaf, S., Corak, J., Berzofsky, J. A., Carbone, D. P. (1996) Dendritic cells in anti-tumor immune responses. II. Dendritic cells grown from bone marrow precursors, but not mature DC from tumorbearing mice are effective antigen carriers in the therapy of established tumors. Cell. Immunol. 170, 111–120. 19. Gabrilovich, D., Cheng, P., Fan, Y., Yu, B., Nikitina, E., Sirotkin, A., Shurin, M., Oyama, T., Adachi, Y., Nadaf, S., Carbone, D. P., Skoultchi, A. (2002) H1(0) histone and differentiation of dendritic cells. A molecular target for tumor-derived factors. J. Leukoc. Biol. 72, 285–296. 20. Shurin, G., Shurin, M., Bykovskaia, S., Shogan, J., Lotze, M., Barksdale, E. J. (2001) Neuroblastoma-derived gangliosides inhibit dendritic cell generation and function. Cancer Res. 61, 363–369. 21. Menetrier-Caux, C., Montmain, G., Dieu, M., Bain, C., Favrot, M., Caux, C., Blay, J. (1998) Inhibition of the differentiation of dendritic cells from CD34(⫹) progenitors by tumor cells: role of interleukin-6 and macrophage-colony-stimulating factor. Blood 92, 4778 – 4791. 22. Gabrilovich, D., Ishida, T., Oyama, T., Ran, S., Kravtsov, V., Nadaf, S., Carbone, D. P. (1998) Vascular endothelial growth factor inhibits the development of dendritic cells and dramatically affects the differentiation of multiple hematopoietic lineages in vivo. Blood 92, 4150 – 4166. 23. Sauer, H., Wartenberg, M., Hescheler, J. (2001) Reactive oxygen species as intracellular messengers during cell growth and differentiation. Cell. Physiol. Biochem. 11, 173–186. 24. Bladier, C., Wolvetang, E. J., Hutchison, P., de Haan, J. B., Kola, I. (1997) Response of a primary human fibroblast cell line to H2O2: senescencelike growth arrest or apoptosis. Cell Growth Differ. 8, 589 –598. 25. Abe, J., Berk, B. C. (1999) Fyn and JAK2 mediate Ras activation by reactive oxygen species. J. Biol. Chem. 274, 21003–21010. 26. Sattler, M., Winkler, T., Verma, S., Byrne, C. H., Shrikhande, G., Salgia, R., Griffin, J. D. (1999) Hematopoietic growth factors signal through the formation of reactive oxygen species. Blood 93, 2928 –2935. 27. Schreck, R., Albermann, K., Baeuerle, P. A. (1992) Nuclear factor kappa B: an oxidative stress-responsive transcription factor of eukaryotic cells (a review). Free Radic. Res. Commun. 17, 221–237. 28. Burkly, L., Hession, C., Ogata, L., Relly, C., Marconi, L. A., Olson, D., Tizard, R., Cate, R., Lo, D. (1995) Expression of relB is required for development of thymic medulla and dendritic cells. Nature 373, 531– 536. 29. Wu, L., D’Amico, A., Winkel, K. D., Suter, M., Lo, D., Shortman, K. (1998) RelB is essential for the development of myeloid-related CD8 alpha (–) dendritic cells but not of lymphoid-related CD8 alpha(⫹) dendritic cells. Immunity 9, 839 – 847. 30. Oyama, T., Ran, S., Ishida, T., Nadaf, S., Kerr, L., Carbone, D., Gabrilovich, D. (1998) Vascular endothelial growth factor affects dendritic cell maturation through the inhibition of nuclear factor-kappaB activation in hemopoietic progenitor cells. J. Immunol. 160, 1224 –1232.
