Research ArticleOncology Open Access | 10.1172/jci.insight.159419

Wade R. Gutierrez,1,2,3,4 Amanda Scherer,3,4 Jeffrey D. Rytlewski,4 Emily A. Laverty,4 Alexa P. Sheehan,3,4,5 Gavin R. McGivney,1,3,4,6 Qierra R. Brockman,3,4,5 Vickie Knepper-Adrian,4 Grace A. Roughton,4 Dawn E. Quelle,1,2,3,5,7,8 David J. Gordon,3,9 Varun Monga,3,4 and Rebecca D. Dodd1,2,3,4,5

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Gutierrez, W. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Scherer, A. in: JCI | PubMed | Google Scholar

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Rytlewski, J. in: JCI | PubMed | Google Scholar

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Laverty, E. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Sheehan, A. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by McGivney, G. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Brockman, Q. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Knepper-Adrian, V. in: JCI | PubMed | Google Scholar

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Roughton, G. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Quelle, D. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Gordon, D. in: JCI | PubMed | Google Scholar |

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Monga, V. in: JCI | PubMed | Google Scholar

1Cancer Biology Graduate Program,

2Medical Scientist Training Program,

3Holden Comprehensive Cancer Center,

4Department of Internal Medicine,

5Molecular Medicine Graduate Program,

6Department of Molecular Physiology and Biophysics,

7Department of Neuroscience and Pharmacology,

8Department of Pathology, and

9Department of Pediatrics, University of Iowa, Iowa City, Iowa, USA.

Address correspondence to: Rebecca D. Dodd, Carver College of Medicine, University of Iowa, 375 Newton Rd, 5206 MERF, Iowa City, Iowa 52246, USA. Phone: 319.335.4962; Email: rebecca-dodd@uiowa.edu.

Find articles by Dodd, R. in: JCI | PubMed | Google Scholar

Published October 13, 2022 - More info

Published in Volume 7, Issue 22 on November 22, 2022
JCI Insight. 2022;7(22):e159419. https://doi.org/10.1172/jci.insight.159419.
© 2022 Gutierrez et al. This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Published October 13, 2022 - Version history
Received: February 16, 2022; Accepted: October 11, 2022 View PDF Abstract

The DNA methyltransferase inhibitor decitabine has classically been used to reactivate silenced genes and as a pretreatment for anticancer therapies. In a variation of this idea, this study explores the concept of adding low-dose decitabine (DAC) following administration of chemotherapy to bolster therapeutic efficacy. We find that addition of DAC following treatment with the chemotherapy agent gemcitabine improves survival and slows tumor growth in a mouse model of high-grade sarcoma. Unlike prior studies in epithelial tumor models, DAC did not induce a robust antitumor T cell response in sarcoma. Furthermore, DAC synergizes with gemcitabine independently of the immune system. Mechanistic analyses demonstrate that the combination therapy induces biphasic cell cycle arrest and apoptosis. Therapeutic efficacy was sequence dependent, with gemcitabine priming cells for treatment with DAC through inhibition of ribonucleotide reductase. This study identifies an apparently unique application of DAC to augment the cytotoxic effects of conventional chemotherapy in an immune-independent manner. The concepts explored in this study represent a promising paradigm for cancer treatment by augmenting chemotherapy through addition of DAC to increase tolerability and improve patient response. These findings have widespread implications for the treatment of sarcomas and other aggressive malignancies.

Introduction

Epigenetic drugs have been of scientific and medical interest since their advent in the 1970s. DNA methyltransferase inhibitors (DNMTis) are a promising class of epigenetic modulators that demethylate DNA to reprogram cellular gene expression. Decitabine (DAC) is a deoxycytidine analogue and DNMTi that targets gene methylation through inhibition of DNA methyltransferase 1 (DNMT1). DAC was originally developed as a cytotoxic chemotherapeutic agent for delivery at maximum tolerated doses (MTDs). However, prolonged myelosuppression after DAC treatment limited its clinical applications. Recognition of DAC’s robust DNA-demethylating activity at lower doses led to its reintroduction in the clinic at approximately 5% of the original MTD (1). At these doses, DAC became a mainstay of myelodysplastic syndrome treatment regimens. More recently, interest has grown in leveraging the epigenetic effects of even lower doses of DAC to augment standard-of-care therapies for solid tumors. Studies of combination therapies using this low-dose DAC approach have shown promising results in several cancer types, including melanoma, ovarian cancer, and colorectal cancer (2–5).

Global DNA hypomethylation and reactivation of silenced antitumor genes was initially proposed as the main mechanism of action for DNMTi-based therapies. However, identification of specific gene targets and their relevance to therapeutic outcomes remains unclear, despite many efforts at genome-wide methylation profiling (6–9). Recent data from epithelial tumor models have suggested a robust role for the immune system in the antitumor activity of demethylating agents. These studies demonstrated upregulation of the viral-response pathway through expression of normally silenced endogenous retroviral (ERV) genes, resulting in increased infiltration of antitumor CD8+ T cells (10–13). On the basis of these findings, many groups have combined DNMTi agents with immunotherapy approaches, particularly in ovarian and colorectal cancer models (14, 15). However, few studies have combined DAC with chemotherapy for the treatment of solid tumors. Here, we explore the efficacy and elucidate the mechanism of DAC in combination with gemcitabine (Gem), a chemotherapy commonly used in the treatment of many solid tumors, including sarcoma, pancreatic, bladder, breast, ovarian, head and neck, and non–small cell lung cancers (16–20).

Sarcomas are a heterogenous group of aggressive cancers of mesenchymal origin. The outcome for patients with high-grade sarcoma has remained unchanged for decades. One of the most common forms of adult sarcoma is undifferentiated pleomorphic sarcoma (UPS), an intrinsically chemoresistant tumor that most frequently develops in the large muscles of limbs. Despite broad chemotherapy resistance, patients with advanced sarcoma are routinely treated with chemotherapy when the disease can no longer be adequately treated with surgery or radiation. The relative rarity of individual soft tissue sarcoma (STS) subtypes and widespread chemoresistance has made the study and development of novel chemotherapy regimens extremely difficult. The current standard of care for advanced STS was developed over 40 years ago. Despite a modest 26% overall response rate and 12.8-month increase in survival, a combination of doxorubicin and ifosfamide remains the most effective regimen currently available (21). In addition to having limited efficacy, doxorubicin-based regimens have a cumulative toxicity profile that limits treatment administration and decreases patients’ quality of life (22, 23). Recently, Gem-based chemotherapy has been shown to have similar efficacy as doxorubicin-based regimens but with fewer adverse side effects, making it an attractive candidate for use in new combination therapies (24).

In this study, we explore the therapeutic concept of combining DAC with the widely used chemotherapy Gem. Here, we use a combination of in vitro assays and in vivo approaches to assess fundamental aspects of the molecular and cellular events contributing to the increased efficacy of chemotherapy when combined with DAC in sarcoma. Using a well-established mouse model of aggressive, high-grade sarcoma, we demonstrate that this combination synergistically slows tumor growth and extends survival in vivo. We identify an unexpected, immune-independent mechanism by which DAC augments chemotherapy treatment in a sequence-dependent manner. The concept of leveraging sequential epigenetic therapy to improve initial chemotherapeutic response opens the door to new conceptual paradigms using DAC in the treatment of solid tumors. Furthermore, the potential of adding a well-tolerated epigenetic therapy to lower the effective dose of chemotherapy has strong implications for long-term survivorship and improved quality of life for patients with cancer.

Results

DAC improves chemotherapeutic response in a mouse model of high-grade sarcoma. To examine the ability of DAC to augment chemotherapy in vivo, we used a mouse model of high-grade UPS that resembles human tumors at the molecular, pathological, and physiological levels (25–27). This model has been extensively used for preclinical studies, several of which have advanced into clinical applications (28–30). Importantly, these mice develop tumors that are surrounded by a native, immune-competent microenvironment that evolves in response to cancer growth and treatment. This approach uses Cre-loxP technology to induce tumors in the leg muscle of adult mice through localized deletion of p53 and activation of oncogenic Kras by injection of an adenovirus expressing Cre recombinase into LSL-KrasG12D, p53fl/fx (KP) mice (Figure 1A). High-grade tumors develop within 6 to 12 weeks, surrounded by an intact immune system.

Figure 1

Gem + DAC slows tumor growth and extends survival in a primary mouse model of high-grade sarcoma. (A) KP tumors were induced in KP mice using i.m. injection of Cre recombinase adenovirus (Ad-Cre) to locally activate oncogenic Kras and delete p53. After tumor initiation, mice were randomized to 1 of 4 treatment groups: PBS, Gem, DAC, or Gem + DAC. Tumor dimensions were measured by caliper 3 times weekly, and terminal tumor tissue was collected for molecular analyses. (B) At the time of tumor detection, mice were placed in 1 of 4 experimental arms: 6 doses of PBS, 1 dose of Gem (150 mg/kg), 5 doses of DAC (0.2 mg/kg), or 1 dose of Gem followed by 5 doses of decitabine. (C) Treatment with Gem + DAC significantly slowed tumor growth compared with PBS or single-agent controls. Growth rates are reported as the time required for tumors to triple in volume (n = 20–25/group). Boxes represent 25th and 75th percentiles. Whiskers represent minimum and maximum values. Horizontal line represents median; + represents mean. (D) Treatment with Gem + DAC extended survival longer than single-agent treatments. Welch’s ANOVA and Dunnett’s T3 multiple comparison test were used to analyze the data in C. Log-rank (Mantel-Cox) tests with Bonferroni correction were used to analyze the data in D. Adjusted α = 0.00833. *P < 0.05 in C. *P < 0.00833 in D.

After tumors reached 125 to 275 mm3, mice were randomized to receive PBS control, Gem alone, DAC, or Gem + DAC (dosing scheme shown in Figure 1B). No signs of toxicity were observed. Both tumor growth rate and mouse survival were used as metrics of outcome. As previously reported, this model is extremely chemoresistant (29). Treatment with Gem did not significantly extend survival compared with PBS alone (median survival of 16.0 vs 13.0 days) (Figure 1, C and D). Gem also did not slow the rate of tumor growth, with tumors tripling in volume in an average of 12.6 days compared with 10.4 with PBS control. These findings illustrate the aggressiveness of this model, which mimics the poor response of sarcoma to conventional chemotherapy regimens. Similarly, DAC monotherapy did not improve survival or slow tumor growth. DAC-treated mice displayed a median survival of 15.0 days and tumors tripled in volume in 11.4 days. In contrast, treatment with Gem + DAC significantly prolonged survival (median survival, 22.0 days) compared with PBS or single-agent treatment. Tumor growth rate was also slowed in mice treated with Gem + DAC: the time required for tumors to triple in volume was extended to an average of 18.0 days. These robust preclinical data demonstrate that addition of DAC to Gem chemotherapy significantly slows tumor growth and extends survival in an autochthonous, genetically engineered mouse model (GEMM) of high-grade sarcoma.

Immunoprofiling reveals minimal impact of DAC on intratumoral T cell subsets. Several reports on animal models of epithelial cancers suggest that DAC abrogates tumor growth by activating antitumor CD8+ T cells through a viral response pathway (10–13). We therefore hypothesized that T cell infiltration would increase in tumors treated with DAC monotherapy and the Gem + DAC combination. To determine how these therapies affect the immune microenvironment, we examined intratumoral T cell profiles by flow cytometry. We found no change in total immune infiltration between experimental arms, as shown by the pan-immune marker CD45 (Figure 2A). We also observed no changes in overall CD3+ T cell levels across treatment groups (Figure 2B).

Figure 2

Gem + DAC efficacy is immune independent. (A and B) Tumor infiltration of total immune cells (CD45+) and total T cells (CD3+) in primary KP tumors were unchanged by Gem + DAC. Data represent individual tumors and the mean ± SD (n = 7–9 tumors/group). (C) Average frequencies of T cell populations in KP tumors, reported as percentages of total CD3+ T cells. Mean values are calculated from all individual tumors shown in A and B. (D) Generation of immune-competent and immune-deficient allograft models. KRIMS-1 cells derived from an untreated KP tumor were injected orthotopically into the gastrocnemius muscle of 129/SvJae or NSG mice. Mice were treated using the dosing scheme in Figure 1B. (E and F) Gem + DAC slowed tumor growth and prolonged survival in immune-competent 129/SvJae mice (n = 4–5/group). (G and H) Similarly, Gem + DAC efficacy was preserved in immune-deficient allografts in NSG mice (n = 8–9/group). Welch’s ANOVA and Dunnett’s T3 multiple comparison test used to analyze data in A, B, E, and G. Log-rank (Mantel-Cox) tests with Bonferroni correction were used to analyze the data in F and H. Adjusted α = 0.00833. *P < 0.05 in E and G. *P < 0.00833 in F and H.

Closer examination of T cell profiles in our sarcoma model revealed minimal to no affect on T cell subsets between treatment groups (Figure 2C and Supplemental Figure 1, A–D; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.159419DS1). This is in striking contrast to data reported from epithelial cancer models (10–13). In our sarcoma study, CD4+ T helper cells comprised 33%–43% of total CD3+ T cells in all treatment groups. Unlike previous studies that reported DAC-induced increases in CD8+ T cells, we found no change in levels of cytotoxic T cells from sarcomas treated with PBS, Gem, or DAC (32.03%, 30.87%, and 28.26%, respectively). Mice receiving Gem + DAC had a slight decrease in CD8+ T cells (24.76%), although this was not statistically significant. Interestingly, we observed an increase in the proportion of Tregs in Gem + DAC–treated tumors compared with PBS or DAC-treated tumors (24.17% vs 10.60% or 14.75%, respectively). Of note, the effect of DAC and Gem + DAC treatment on nontumor organs, such as spleen, was also minimal (Supplemental Figure 1, E–J). Consistent with these findings, no changes in ERV gene expression or viral response–pathway transcripts were detected in terminally harvested tumors from mice receiving DAC or Gem + DAC (Supplemental Figure 2).

Therapeutic activity of Gem + DAC is independent of T cells. Considering the minimal changes in immune profiles observed in the Gem + DAC–treated tumors, we hypothesized that the therapeutic mechanism of Gem + DAC is immune independent. To test our hypothesis, we generated orthotopic allografts using K-ras induced murine sarcoma 1 (KRIMS-1) cells derived from a primary Kras/p53–mutant UPS tumor. We previously showed that syngeneic KRIMS-1 allografts have similar growth rates, survival, and immune infiltration as the primary KP tumors (31). To test the role of the immune system in Gem + DAC response in vivo, we injected KRIMS-1 cells orthotopically into immune-competent (129/SvJae) or immune-deficient (i.e., NOD/SCID/γ [NSG]) mice (Figure 2D). Results with the immune-competent allograft model closely match the data obtained in the primary GEMM examined in Figure 1 (Figure 2, E and F). Immune-competent tumors in 129/SvJae mice treated with Gem + DAC tripled in volume in approximately 20.0 days, and PBS, Gem-, and DAC-treated tumors tripled within 10.8, 13.6, and 13.0 days, respectively. Gem + DAC also significantly prolonged survival (median survival, 21.0 days) compared with PBS, Gem, or DAC controls (median survival, 13.0, 15.0, and 15.0 days, respectively). In immune-deficient NSG mice, Gem + DAC activity was preserved, despite these mice lacking mature T cells, B cells, and NK cells, and having defective myeloid populations (Figure 2, G and H). Gem + DAC–treated, immune-deficient tumors tripled in 19.4 days, compared with 10–13 days for PBS and monotherapy-treated tumors. Similarly, median survival was extended to 22.0 days in mice receiving Gem + DAC compared with 11.0, 15.5, and 14.0 days for mice receiving PBS, Gem, or DAC, respectively. These findings demonstrate that an intact immune system is not necessary for Gem + DAC activity in vivo and suggest that an alternative, immune-independent mechanism is responsible for the therapeutic benefit observed in these mice.

Drug sequence is critical for synergistic Gem + DAC activity in vitro. To explore the mechanisms driving the activity of Gem + DAC, we treated KRIMS-1 cells in vitro with a similar dosing strategy used for in vivo studies described above (Figure 3A). Dose-response curves revealed KRIMS-1 cells are moderately sensitive to Gem, with IC50 values in the nanomolar range. In contrast, these cells are resistant to DAC, with IC50 values approaching micromolar levels (Supplemental Figure 3 and Supplemental Figure 4). Using the Bliss independence model to assess drug synergy (32, 33), we investigated increasing concentrations of Gem and DAC after 4 days of incubation (Figure 3B). Gem + DAC treatment generally was additive (δ score, 0–10), with a strong synergistic interaction (δ score, >10) occurring with 15 nM Gem and 128 nM DAC (δ score, 29.99). The combination of 15 nM Gem and 128 nM DAC was identified as being strongly synergistic, using 3 different synergy analyses: Bliss independence, highest single agent (34), and zero-interaction potency (35) (Supplemental Figure 5, A–D). Similarly, analysis of human sarcoma and carcinoma cell lines identified areas of synergistic interaction with Gem + DAC treatment, particularly in the embryonal rhabdomyosarcoma line RD, the alveolar rhabdomyosarcoma cell line SJRH30, and the pancreatic ductal adenocarcinoma cell line MIA PaCa-2 (Supplemental Figures 3, 4, 6, and 7).

Figure 3

Drug sequencing is critical for Gem + DAC efficacy. (A) In vitro treatment scheme. KRIMS-1 cells were treated with Gem or media control on day 1, followed by DAC or DMSO control on days 2-3. (B) Representative synergy plot of Gem + DAC identifying concentrations that synergistically inhibit cell growth (maximum Bliss synergy score of 29.99 for gemcitabine 15 nM and DAC 128 nM). (C–H) Longitudinal viability and day 4 measurements of KRIMS-1 cells using different sequences of delivery for Gem (15 nM) and DAC (128 nM). (C and D) Sequential administration of Gem followed by DAC. (E and F) Concurrent administration of Gem + DAC treatment. (G and H) Reversed-sequence DAC + Gem treatment, with DAC preceding Gem treatment. Individual viability measurements and statistical analysis for data in C, E, and G are available in Supplemental Figure 8. For C–H, data represent independent experiments (n = 3) and the mean ± SEM. Ordinary 1-way ANOVA and Tukey’s multiple comparisons test used to analyze data in C–H. *P < 0.05.

We next assessed, through a series of longitudinal studies, how the order of drug delivery influenced these synergistic effects. First, we tested sequential Gem + DAC and observed decreased viability at days 3 and 4 in cells receiving the combination treatment compared with monotherapy (Figure 3, C and D; Supplemental Figure 8A; and Supplemental Table 1). These data further support our finding of synergy identified in Figure 3B. Gem and DAC monotherapies reduced viability to 73.4% and 87.7%, respectively, compared with DMSO control. Sequential treatment with Gem + DAC reduced viability to 44.9%. We then tested the effects of concurrent Gem + DAC treatment, because coadministration of drugs is a more logistically feasible treatment scheme to use in the clinic (Figure 3, E and F; Supplemental Figure 8B; and Supplemental Table 1). The synergistic effects of Gem + DAC were heightened with concurrent administration: viability was reduced to 17.5%. Finally, we tested the reverse order of drug delivery by treating first with DAC on days 1 and 2, followed by Gem on day 3 (Figure 3, G and H, Supplemental Figure 8C, and Supplemental Table 1). Unlike the sequential and concurrent Gem + DAC treatments, DAC + Gem (i.e., DAC given first) only reduced viability to 68% and did not perform better than DAC alone. Similar results were seen in vivo, with the Gem + DAC sequence having a stronger impact on overall survival and tumor growth compared with DAC + Gem (Supplemental Figure 9, A–C). These in vitro and in vivo findings suggest that the synergistic interaction of Gem + DAC is sequence dependent, relying on the initial presence of Gem to modulate response to DAC.

Gem + DAC treatment induces apoptosis and cell cycle arrest. We next explored the mechanisms by which Gem + DAC reduces cell growth. We examined multiple mechanisms that could be responsible for reduced cellular viability, including markers of DNA damage, apoptosis, senescence, and cell cycle arrest. As expected, DAC monotherapy and Gem + DAC decreased DNMT1 protein levels and 5-methylcytosine levels in genomic DNA, demonstrating that canonical DAC activity is preserved with Gem + DAC treatment (Figure 4, A and B, and Supplemental Figure 10A). Both DAC monotherapy and Gem + DAC reduced DNMT1 and 5-methylcytosine levels to a similar extent, indicating that Gem + DAC efficacy is not primarily driven by decreased levels of DNMT1 or DNA hypomethylation. Immunoblot studies further demonstrated that Gem + DAC treatment does not appreciably alter levels of full-length PARP or cleaved PARP, which are markers of DNA damage and apoptosis, respectively (Figure 4, C and D). Similarly, levels of γH2AX, a marker of DNA double-strand breaks, were only modestly increased by Gem + DAC treatment (Supplemental Figure 10B).

Figure 4

Gem + DAC induces apoptosis and cell cycle arrest. (A–D) Representative Western blot and quantification of lysates collected on day 4. Data represent independent experiments (n = 4) and the mean ± SD. (B) Levels of DNMT1 are decreased in cells treated with DAC and Gem + DAC. (C and D) Levels of full-length PARP (fPARP) and cleaved PARP (cPARP) are not altered across treatments. (E) Day 4 measurement of annexin V staining shows increased apoptosis in Gem + DAC–treated cells. (F) Longitudinal cell cycle analysis using EdU/PI staining shows increased cell cycle arrest and accumulation in G2/M in Gem + DAC–treated cells. Complete statistical analysis is available in Supplemental Figure 10. Data in E and F represent independent experiments (n = 3) and the mean ± SD. Ordinary 1-way ANOVA and Tukey’s multiple comparisons test were used for analysis of data in B–F. *P < 0.05.

To directly measure the impact of Gem + DAC on apoptosis, we performed annexin V staining. Using this method, we observed that Gem + DAC induced apoptosis in 8.1% of cells, compared with 2.0%–3.1% in control or monotherapy-treated cells (Figure 4E). Despite the significant increase in the number of apoptotic cells, the relatively low magnitude of apoptosis in the Gem + DAC population could not account for the large decreases in viable cells identified in Figure 3. Next, using analysis of cell morphology by brightfield microscopy, we examined potential induction of cellular senescence (Supplemental Figure 11, A–E). Some cell enlargement and flattening, features classically associated with senescence, were observed with DAC and Gem + DAC treatments. However, these changes were minimal compared with the effects of strongly prosenescent MEK and CDK4/6 inhibitor therapies (36).

Finally, we assessed the impact of Gem + DAC on cell cycle progression during the 4-day treatment. Longitudinal cell cycle analysis revealed that in addition to promoting apoptosis, Gem + DAC caused cell cycle arrest (Figure 4F and Supplemental Figure 10, C–E). The effects of Gem + DAC on cell cycle progression varied over time. A significant 16.3% increase in S-phase arrest was detected at day 2 of treatment, as expected from Gem on the basis of previous literature (37). After removing Gem and adding DAC, an increase in G2/M arrest was observed on days 3 and 4. Gem + DAC caused greater than a 2-fold increase in G2/M compared with vehicle and Gem controls, and a 1.6-fold increase compared with single-agent DAC. These results strongly suggest that a key feature of the Gem + DAC combination therapy is slowed proliferation due to sustained cell cycle arrest.

Decreased cellular deoxycytidine levels drive Gem + DAC activity. Gem, a deoxycytidine analogue (2′,2′-difluoro-2′-deoxycytidine), induces cytotoxic effects through 2 well-established mechanisms: direct termination of DNA polymerization and irreversible inhibition of ribonucleotide reductase (RNR) (Figure 5A). Gem-induced RNR inhibition decreases cellular deoxyribonucleotide pools, particularly 2′-deoxycytidine 5′-triphosphate (dCTP) levels (38). The decrease in cellular dCTP potentiates the effects of Gem by decreasing its competition with dCTP for incorporation into DNA (39). Like Gem, DAC is a deoxycytidine analogue (5-aza-2′-deoxycytidine). Inhibition of RNR augments DAC efficacy and increases its incorporation into DNA (40). On the basis of these known mechanisms of action and our data demonstrating the importance of drug sequencing and cell cycle inhibition (Figures 3 and 4), we hypothesized that Gem + DAC activity is based on initial inhibition of RNR by Gem, which primes cells for treatment with DAC. We further predicted that DAC activity is more effective in combination-treated cells because of an increased DAC to dCTP ratio, resulting in elevated and sustained cell cycle arrest. To test the first part of this hypothesis, we blocked RNR activity with an alternative compound, substituting thymidine (Thy) for Gem in the sequential Gem + DAC dosing strategy (Figure 5B). Like Gem, Thy is a potent inhibitor of RNR that specifically depletes dCTP levels (41, 42). However, in contrast to Gem activity, Thy does not directly inhibit DNA polymerization (Figure 5A), making it a useful tool compound for dissecting the mechanistic activity of Gem in Gem + DAC therapy.

Figure 5

Decreased cellular dCTP drives increased efficacy of Gem + DAC. (A) Gem causes direct termination of DNA polymerization and irreversible inhibition of RNR, resulting in decreased levels of dCTP. Thy inhibits RNR and causes a similar decrease in dCTP levels. (B) Treatment schematic for Gem + DAC and Thy + DAC single-agent and combination approaches. (C) At day 4, Thy and Thy + DAC decrease viability to the same extent as Gem and Gem + DAC, respectively. (D) Day 4 G2/M analysis using EdU/PI staining shows similar cell cycle arrest in Thy-containing treatments. Complete statistical analysis of C and D is available in Supplemental Figure 12. (E) Gem inhibits RNR, decreasing cellular dCTP and increasing the ratio of DAC to dCTP. Levels of dCTP can be directly augmented by addition of exogenous deoxycytidine (dC), but not uridine (U), due to inhibition of RNR by Gem. (F) Uridine (30 μM) or dC (30 μM) was added during DAC or DMSO treatment on days 2 and 3. (G and H) Addition of dC, but not uridine, rescued the effects of DAC on viability and G2/M arrest in cells treated with both DAC and Gem + DAC treatments. Complete statistical analysis of the data in G and H is available in Supplemental Figures 13 and 14. Data in C, D, G, and H represent independent experiments (n = 3) and the mean ± SEM. Ordinary 1-way ANOVA and Tukey’s multiple comparisons test used for analysis. *P < 0.05.

As expected, Thy treatment decreased cell viability to a similar extent as Gem (Figure 5C and Supplemental Figure 12A). In combination with DAC, Thy has similar effects as Gem, with Thy + DAC treatments reducing cellular viability to 25.4%, which is nearly identical to the 26.8% viability observed after Gem + DAC treatment. Similarly, Thy + DAC also alters cell cycle progression by promoting G2/M arrest, as seen with Gem + DAC treatment (Figure 5D and Supplemental Figure 12, B–D). Approximately 8% of cells were in G2/M arrest after treatment with Thy or DAC alone. This increased to 13.1% with Thy + DAC, similar to 10.9% with the Gem + DAC combination. Although Thy had a greater impact on G2/M arrest than did Gem (8.4% vs 5.8%), this ~2% difference persisted when combined with DAC, suggesting that the magnitude of DAC’s impact is similar in Gem + DAC and Thy + DAC treatments. These data demonstrate that blocking RNR with a tool compound prior to DAC treatment has similar effects to Gem + DAC on cell viability and G2/M arrest, supporting the conclusion that Gem primes cells for DAC activity through inhibition of RNR.

As a deoxycytidine analogue, DAC must compete with the intracellular dCTP pool for incorporation into DNA. Because both Gem and Thy specifically decrease cellular dCTP levels, the second part of our hypothesis predicted that the augmented effectiveness of Gem + DAC therapy results from an elevated DAC to dCTP ratio after Gem treatment. To test that idea, we treated KRIMS-1 cells with sequential Gem + DAC in the presence of supplemental deoxycytidine or uridine du