Background
Tumors avoid immune detection and attack through a variety of mechanisms that circumvent the anti-tumor response. DNA hypermethylation, though reversible, can silence the expression of immunogenic antigens; this makes the immune system less effective, especially during immunotherapeutic interventions [
1]. Tumors also recruit regulatory immune cells, including MDSCs, which dampen the adaptive immune response. Patients with higher levels of circulating MDSCs have increased primary tumor growth [
2], higher metastatic burden [
3], more advanced clinical cancer stage [
4,
5], and shorter overall survival [
3,
6]. Based on these findings and multiple reports in mouse models that implicate MDSCs as key obstacles to successful cancer immunotherapy, there has been much interest in eliminating the suppressive nature of MDSCs to improve patient outcomes [
7‐
10].
The myeloid compartment in cancer has been extensively reviewed, especially MDSCs [
11]. MDSCs are divided into two subsets, monocytic and granulocytic MDSCs. Both types of MDSCs have been shown to be suppressive in both murine tumor models and in several human cancers. Monocytic MDSCs generate nitric oxide as a mechanism of suppression, whereas granulocytic MDSCs express large amounts of reactive oxygen species and arginase-1 that results in suppression of T cell function [
11]. MDSCs have been found in the spleen, blood, liver, and tumor of tumor-bearing animals. These suppressive cells have been found to accumulate in various types of murine tumor models and human cancer, from murine hepatic carcinoma models [
12], breast cancer models [
13], to human ovarian cancer [
14], and many more.
Our lab has previously published on the effects of the DNA methyltransferase inhibitor (DNMTi), decitabine. Using the aggressive murine breast cancer line, 4 T1, we found that decitabine improved the immunogenicity of these cells in vitro, and augmented the effects of adoptive immunotherapy (AIT) in vivo [
10]. While decitabine caused a reduction in tumor-induced MDSC accumulation, the underlying mechanism behind this was never investigated. In our current study, we have expanded upon these findings with the second-generation DNTMi, guadecitabine, and investigated its mechanism of action in tumor reduction. Like the active metabolite decitabine, we found that guadecitabine diminished tumor-induced granulocytosis in 4 T1 tumor-bearing mice. As a result of the reduced MDSC accumulation, guadecitabine rescued immune activation and was able to reduce tumor growth in a T cell-dependent manner. Guadecitabine was similarly effective in the E0771 model of murine breast carcinoma. Finally, we found that guadecitabine in combination with AIT resulted in prolonged survival in both 4 T1 and E0771 breast cancer models. Because of these advantageous effects, guadecitabine could prove to be a beneficial new drug to reduce systemic immune suppression and augment the effectiveness of immunotherapy in cancer patients.
Discussion
There is great difficulty in treating neoplastic disease, especially when the tumor becomes resistant to chemotherapy. Today, the most promising interventions involve boosting the patient’s own immune system to detect and destroy abnormal cells. Immunotherapy has emerged as a highly promising and effective treatment for a variety of cancer types, but still only a minority of patients exhibit strong objective responses. Tumors can employ several “tactics” to avoid recognition and suppress anti-tumor immune response. Through hypermethylation, tumors can silence immunogenic antigens to avoid cell-mediated killing. Additionally, the tumor environment can induce massive myelopoiesis, causing suppressive MDSCs to accumulate in the bone marrow, spleen, circulation, and in the tumor. For these reasons, there is great interest in finding ways to reverse these effects, thus allowing the immune system to clear the tumor.
Several groups have reported on the myelo-depleting properties of demethylating drugs such as decitabine [
28,
32] and 5-Azacitadine (AZA) [
33], as well as other chemotherapy drugs, including gemcitabine [
34,
35], doxorubicin [
8] and docetaxel [
36]. Although MDSCs are not being directly targeted, they seem to be more susceptible to their effects as demonstrated in this study. Guadecitabine, also known as SGI-110, was specifically designed to be resistant to degradation by cytidine deaminase and prolong the exposure of tumor cells to the active metabolite, decitabine. In vivo, guadecitabine treatments resulted in a near-complete absence of MDSCs in tumor-bearing mice (Fig.
1c-e). Based on the time-course experiment (Fig.
2), it appears that guadecitabine treatment is
preventing, rather than reversing, MDSC accumulation. We believe guadecitabine targets the bone marrow by diminishing the highly proliferative myeloid progenitors (Fig.
2c,
Supplemental Fig. 3e). This prevents increased MDSC circulation (Fig.
2d) and accumulation within the spleen (Fig.
2b). Surprisingly, we found that the similarly proliferative 4 T1 tumor cells were not vulnerable to cytotoxic effects of in vivo guadecitabine treatments (Fig.
3f).
Within the spleen of tumor-bearing mice we showed an accumulation of MDSCs in the red pulp (
Supplemental Fig. 2d). This perilymphoid localization puts the MDSCs in contact with recirculating CD8
+ CTL. Several tumor studies have portrayed the spleen as an inhibitory environment that can diminish CTL function [
7,
37]. In experiments by Ugel et al, the investigators removed the inhibitory MDSC environment through splenectomies [
7]. Although this did not affect tumor size, they found that T cell activation was recovered despite normal MDSC frequency within the blood and other tissues. This highlights the spleen’s unique role as an isolated region of suppression with the ability to severely dampen the anti-tumor immune response. In the present study, we used guadecitabine to ablate the suppressive splenic environment. We found that IFNγ production within the dLN is comparable between control and guadecitabine-treated mice (Fig.
4a). Upon recirculating through the spleen, however, CTLs from control tumor-bearing mice have diminished activation (Fig.
4b), even though the number of activated cells remained the same (Fig.
4c). These data support the role of the spleen as an important suppressive zone that contributes to tumor progression.
Unlike Ugel’s splenectomy experiments, our treatment additionally resulted in slower tumor growth (Fig.
2a), indicating guadecitabine may have a beneficial impact beyond the removal of regulatory myeloid populations from the spleen. The reduced suppressive activity in the spleen and tumor environments may be the reason for the reduction in tumor growth (
Supplementary Fig. 7a,b). A recent study examining MDSC subsets in the peripheral blood of patients with multiple types of cancer found that arginase1 generated by granulocytic MDSCs was the main T cell inhibition mechanism [
38]. In our study, guadecitabine was able to dramatically reverse the generation of arginase1, as well as dramatically reducing granulocytic MDSCs. Additionally, we demonstrated increased
Ifng expression in the tumors of guadecitabine-treated mice (Fig.
4e). Kim et al. recently demonstrated across several murine models of triple negative breast cancer (TNBC) and in patient samples of TNBC that tumor infiltrating neutrophilic myeloid cells (TINs) were immunosuppressive and contribute to poor prognostic outcomes in patients [
13]. In contrast, tumor infiltrating macrophages or monocytic myeloid cells (TIMs) were associated with increased responsiveness to checkpoint inhibitors in the murine models and better prognosis in humans [
13]. Using guadecitabine treatment, Gr1
+ cells are depleted, which include both MDSCs and neutrophils, but beneficial F4/80
+ TIMs are not affected (
Supplemental Fig. 7b). The enhanced tumor immunity may additionally arise from guadecitabine’s effect on MDSC phenotype. Although the majority of the MDSCs are eliminated, a small percentage of cells remained that are induced to express APC and costimulatory markers such as MHC II and CD80/86 (Fig.
1f, Supplemental Fig.
1b). Since myeloid populations have been previously shown to be extremely plastic [
39], these data suggest that guadecitabine pushes suppressive MDSCs to develop into an immune-stimulatory phenotype that may augment immune activation within the spleen.
The Ugel experiment also emphasizes a significant problem with a popular and promising clinical therapy. Animals that underwent sham surgeries responded poorly to AIT compared to those that received splenectomies. When the antigen-experienced T cells circulate through the suppressive splenin environment, they are inactivated despite being primed to target the tumor. Here we have shown a similar phenomenon; while AIT was effective in slowing the growth rate of tumors, combination therapy with AIT+guadecitabine compounded this effect and resulted in persistent tumor suppression (Fig.
5b) and prolonged survival (Fig.
5e). It is interesting to note that in the AIT experiments, guadecitabine was administered earlier at days 3, 4, 5, and 6 (Fig.
5), rather than days 10, 11, 12, and 13 (Figs.
1,
2,
3 and
4). This dosing schedule still resulted in slower tumor growth through day 16, although the reasons why are unclear. Further, we tested the effectiveness of delaying adoptive T cell transfer until the final guadecitabine treatment rather than being delivered concomitantly with the initial treatment (
Supplemental Fig. 7). The delayed AIT alone was more effective at days 6 and 7 than at days 3 and 4, perhaps because the tumor has become more immunogenic and vascularized by the later date.
Chimeric antigen receptor (CAR) T cells are genetically engineered to be specific for a designated target tumor antigen. These cells have been shown to be successful in the clinic against CD19
+ B cell neoplasms. New CARs are being generated against antigens on solid tumors [
40]. One of the biggest blocks to CAR treatment in solid tumors has been the suppressive environment created by MDSCs [
40]. The addition of guadecitabine to CAR therapy may hold the key to favorably altering this environment and, as seen with the AIT experiments in this study, help augment the efficacy of adoptively transferred T cells.
Finally, we showed the efficacy of guadecitabine in slowing the growth of another tumor line on a different background strain. There were no observable strain differences noted relating to guadecitabine treatment between the C57Bl/6 J and Balb/cJ mice. The tumor line E0771 is not known to elicit a robust leukemoid reaction, but studies still indicate a suppressive role for MDSCs in this model [
41,
42]. Overall, we observed a similar and persistent reduction in tumor growth with guadecitabine alone, or in combination with AIT.
Methods
Animals
Wildtype (WT) female Balb/cJ, C57Bl/6 J, and athymic NU/J mice 8–10 weeks old were purchased from Jackson Laboratory. The health report was consistent with that of our Barrier Vivarium facility. Balb/cJ mice weighed an average of 20 g prior to the start of experiments and C57Bl/6 J mice weighed an average of 20 g prior to the start of experiments. ADAM10Tg mice were generated by the VCU Transgenic Core and maintained in the Barrier Vivarium facility. All mice were housed within Virginia Commonwealth University vivarium facilities, specifically the Massey Cancer Center Barrier Vivarium, in accordance with the humane treatment of laboratory animals set forth by the NIH and the American Association for the Accreditation of Laboratory Animal Care (AAALAC). All animal experiments were conducted with the permission and oversight of the Virginia Commonwealth University Institutional Animal Care and Use Committee (IACUC) under the protocols AM10065, and AM10256.
All animals were housed with 12 h light and dark cycle in NexGen Cages from Allentown (194 mm × 181 mm × 398 mm) on ventilated racks with corncob bedding (Shepard’s Specialty Corn Cob Plus). The temperature maintained in the cages is between 68 to 76 degrees Fahrenheit. Five animals are housed per cage. Animals are fed Envigo Teklad 2919 and given water ad libitum via Lab Product’s Hydropacs.
Animals were euthanized by CO2 inhalation followed by cervical dislocation.
Experimental models and guadecitabine treatment
50,000 4 T1 or 200,000 E0771 cells in 50 μL PBS were injected subcutaneously into the flank at day 0. Cagemates were randomly assigned to differing groups at the start of the experiment. Appropriate groups received i.p. injections of 50 μg guadecitabine (kindly provided by Astex Pharmaceuticals, Inc.) on days 10, 11, 12, and 13, unless otherwise indicated. All injections took place between the hours of 10 am to 2 pm, working from the home cage. For each treatment, the control group was initially treated, followed by the non-control groups. Mice were euthanized on day 16, when we collected blood by cardiac puncture, bone marrow from femurs and tibias, tumors, spleens, and inguinal lymph nodes.
T cell depletion was performed as previously described [
20]. Briefly, mice were injected
i.p. with 200 μg of monoclonal antibodies on days 6, 7, 8, 9, and 14. Upon sacrifice, T cell depletion was confirmed in the spleen by flow cytometry. CD4
+ and CD8
+ T cells were depleted using the clones GK1.5 and 2.43, respectively (antibodies generated in house). Rat IgG was used as an isotype control.
AIT was performed as previously described [
10]. Briefly, donor Balb/cJ or C57Bl/6 mice were injected with 5 × 10
5 4 T1 or E0771 cells, respectively, into the hind footpad; popliteal lymph nodes were collected at day 10 and activated overnight with bryostatin (5 nM, Calbiochem) and ionomycin (1 μM, Calbiochem) in the presence of recombinant IL-2 (Peprotech). Cells were then washed and expanded in IL-7 and IL-15 (both 10 ng/mL, Peprotech) for one week. On the indicated day, tumor-bearing recipient mice were treated
i.p. with 100 mg/kg of cyclophosphamide (CYP). 24 h later, 50 million expanded lymphocytes were infused intravenously. All groups, except control mice, received a single CYP treatment. Tumor areas were measured through the skin on live animals using digital calipers as length x width; tumor volumes represent length x width x height of excised tumors. Animals were observed at least three times per week, according to IACUC standards. Animals were euthanized upon reaching a humane endpoint, including tumor area > 100 mm
2, severe ulceration, or weight-loss.
ADAM10Tg animals and C57Bl/6 animals used for MDSC and T cell isolation were euthanized and spleens were harvested.
Organ processing and cell counts
Blood volume collected by cardiac puncture was recorded and used to calculate the normalized number of cells per milliliter. Whole spleens were crushed to obtain a single cell suspension, then red blood cells were removed with ACK lysing buffer (Quality Biological). Femurs and tibias from each mouse were cleaned of connective tissue and spun at 350×g for 5 min to collect marrow before removing red blood cells. Viable cell counts were performed using trypan blue exclusions.
Magnetic cell isolation, T cell suppression assay, and MMT cytotoxicity assay
Splenic MDSCs were purified by CD90.2 and CD11c magnetic depletion (Miltenyi Biotec). In experiments where MDSCs are considered pretreated, ADAM10Tg mice received four consecutive daily i.p. injections of vehicle or guadecitabine before collecting splenic MDSCs. CD90.2+ T cells were purified by magnetic column (Miltenyi Biotec) then labeled with Track-It Violet (Biolegend). Cells were co-cultured with MDSCs from control or guadecitabine-treated ADAM10Tg mice at a 1:2 (T/MDSC) ratio in the presence of anti-CD28 and plate-bound anti-CD3 (both 1 μg/mL, Biolegend) for 96 h. Cells were then harvested and analyzed by flow cytometry for T cell division.
The MTT assay was performed according to manufacturer’s instructions (Abcam). Both 4 T1 cells and MDSCs were cultured with increasing doses of guadecitabine for either 24, 48, or 72 h prior to performing the assay.
Flow cytometry
Single cell suspensions were obtained and stained with the fixable live/dead stain ZombieAqua (Biolegend) per manufacturer’s instructions. Samples were then Fc-blocked with 2.4G2 [
43] for 5 min and stained for 30 min on ice. Flow samples that included multiple Brilliant Violet antibodies were stained in the presence of Brilliant Stain Buffer (BD Biosciences) per manufacturer’s instructions. All cells were then fixed in 4% paraformalydehyde (PFA) fixation buffer (Biolegend) for 15 min at room temp. For intracellular staining, fixed cells were permeabilized with PermWash Buffer (Biolegend) per manufacturer’s instructions. Flow data were collected using a BD LSRFortessa running BDFACSDiva™ 8.0 software, and analyzed with FlowJo (10.4.2). Total MDSCs were characterized as both monocytic and granulocytic populations combined. Gating for MDSC populations are as follows: CD11b
+ Ly6C
hi (monocytic-MDSCs) and CD11b
+ Ly6C
intLy6G
+ (granulocytic MDSCs). B cells were gated as MHC II
+ B220
+, and T cells were gated as B220
− CD4
+ or CD8
+. CMP, GMP, and MEP were gated as previously described [
44]. The antibody clones were as follows: Ly6C (clone HK1.4), Ly6G (1A8), IFNγ (XMG1.2), CD80 (16-10A1), CD86 (GL-1), I-A/I-E (M5/114.15.2), Sca-1 (D7), CD16/32 (2.4G2), cKit (2B8) Lineage cocktail (B220 (RA3-6B2), CD4 (GK1.5), CD8a (53–6.7), Gr1 (RB6-8C5), CD11b (M1/70), Ter119 (Ter119)), IL7Rα (A7R34), CD34 (MEC14.7) all from Biolegend, and CD45R/B220 (RA3-6B2), CD4 (GK1.5), CD11b (M1/70), all from BD Biosciences.
Histology and Immunoflourescence
Some spleens and tumors were fixed with 4% PFA for 15 min then equilibrated with successive incubations in 10, 20, and 30% sucrose before being mounted in Optimal Cutting Temperature (OCT) medium. 10 μm cryo-sections were briefly fixed in ice-cold acetone then in 4% PFA prior to staining. Light microscopy slides were stained using ApopTag Plus Peroxidase In Situ Apoptosis Detection Kit (Millipore Sigma) followed by counterstaining with Meyer’s Hematoxylin then imaged using an Olympus BX41. Immunoflourescence slides were stained as previously described [
45] with antibody clones arginase1 (clone C-2–Santa Cruz Biotechnology), F4/80 (BM8), IFNγ (XMG1.2), Gr1 (RB6-8C5), and CD8a (53–6.7) all from Biolegend. They were imaged on a Zeiss LSM 700 confocal microscope and images were processed on Zeiss 3.1 Blue edition software.
qRT-PCR
For RNA isolation from tumors, TRIzol (Invitrogen) was added to frozen tissue and homogenized. Subsequent RNA isolation was performed according to the manufacturer’s instructions. RNA was quantified using an ND-100 NanoDrop spectrophotometer. One microgram of total RNA was reverse transcribed using SuperScript IV (Thermo Fisher) with oligo (dT20). Primers used in quantitative PCR (qPCR) analysis are as follows: Hprt_forward 5′-CAGGGATTTGAATCACGTTTGTG-3′, Hprt_reverse 5′-TTGCAGATTCAACTTGCGCT-3′, Ifng_forward 5′-TGCCAAGTTTGAGGTCAACAAC-3′, Ifng_reverse 5′-TCATTGAATGCTTGGCGCTG-3′, In short, qPCR was conducted using a QuantStudio 3 Real-Time PCR System with 45 cycles using PowerUp SYBR Green Master Mix (both from Applied Biosystems). Primers were tested for specificity using melt curve analysis.
Cell lines
4 T1 (ATCC® CRL-259™) and E0771 cell lines were purchased from ATCC and CH3 Biosystems, respectively. Cell lines were maintained at low passage numbers and ATCC-recommended tests were performed, including morphology checks and mycoplasma screening.
Ex vivo restimulation for IFNγ production
All cells from each individual tumor-draining lymph node (dLN) or 106 total splenocytes were plated in 2 mL media and restimulated with PMA (250 ng/mL) and ionomycin (1 μM) in the presence of monensin and brefeldin A (Biolegend). After three hours, the cells were washed and permeabilized before being stained for intracellular IFNγ for flow cytometry.
Statistics
Each figure depicts one representative experiment of at least three independently conducted experiments. n values vary between 3 and 5 animals per group, per experiment. The number of experimental animals was determined based on data from spread in previous tumor experiments, while taking into account the need to reduce the use of unnecessary animals in research. For un-paired comparisons, a Student’s t test was performed. Where appropriate, a one-way ANOVA with Tukey’s or a two-way ANOVA with Sidak’s multiple comparison tests was used when analyzing 3+ normally distributed data-sets (specific test indicated in legend). All statistical analyses were performed using GraphPad Prism 7. Significance is indicated in individual figure legends.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.