Epigenetic targeting drugs potentiate chemotherapeutic effects in solid tumor therapy

Epigenetic therapy is a novel tumor therapeutic method and refers to the targeting of the aberrant epigenetic modifications presumably at cancer-related genes by chemicals which are epigenetic targeting drugs (ETDs). Not like in treating hematopoietic cancer, the clinical trials investigating the potential use of ETDs in the solid tumor is not encouraging. Instead, the curative effects of ETD delivered together with DNA targeting chemo drugs (DTDs) are quite promising according to our meta-analysis. To investigate the synergistic mechanism of ETD and DTD drug combination, the therapeutic effect was studied using both cell lines and mouse engrafted tumors. Mechanically we show that HDAC inhibitors and DNMT inhibitors are capable of increasing the chromatin accessibility to cisplatin (CP) and doxorubicin (Dox) through chromatin decompaction globally. Consequently, the combination of ETD and DTD enhances the DTD induced DNA damage and cell death. Engrafted tumors in SCID mice also show increased sensitivity to irradiation (IR) or CP when the tumors were pretreated by ETDs. Given the limited therapeutic effect of ETD alone, these results strongly suggest that the combination of DTD, including irradiation, and ETD treatment is a very promising choice in clinical solid tumor therapy.

Epigenetic refers to the DNA methylation, histone modifications and the dynamic binding of variable proteins that shapes the chromatin compaction status and therefore, determines the gene expression through locally regulating the accessibility of chromatin to transcriptional factors and the ability to form active transcriptional higher order chromatin organization 1,2 . Although it is the combination of DNA methylation, histone modification and chromatin remodeling which forms the "epigenetic code" to indicate the chromatin status, generally highly compact chromatin is labeled by low histone acetylation while open chromatin is more likely to have high acetylated histones [3][4][5] . Similarly, hypermethylated DNA generally indicates the chromatin silencing and DNA hypomethylation is more common at active transcribing chromatin region 6 .
Global alteration of epigenetic modification is considered as one of the hallmarks of cancer 7 . Hyperacetylation of histones and hypomethylation of DNA are both known to be overwhelmingly abundant in cancer cells 8,9 . However, hypoacetylation of histones and hypermethylation of DNA at specific gene loci are also maintained [10][11][12][13] . Gene locus specific recruitment of epigenetic modifiers such as DNA methyl transferases (DNMT) and histone deacetylases (HDACs) are known to be important in maintaining the modification status of these gene loci in cancer cells 14,15 . These gene loci are frequently discovered to be tumor suppressor genes (TSGs), which raises an interesting supposition that epigenetic silencing of TSG is another important mechanism for cancer formation 13,16 . In the perspective of cancer therapy, the silenced TSG becomes an attractive target because recovery of the expression of TSG is much easier than repairing a mutated TSG gene. The purpose, eventually, is to get the cancer cells back to the differentiation program [17][18][19] , because TSGs may be able to persuade the cancer cell to undergo differentiation rather than continuing cell cycling. Another possible consequence of reactivating TSG is to kill the cancer cells because the cancer cells may not be able to survive when the TSG is back.
Targeting the epigenetic aberrations in cancer is now known as "epigenetic therapy" which mainly relies on the several DNMT inhibitors and HDAC inhibitors. Some promising results have been obtained in treating some types of hematopoietic cancer such as MDS, multiple myeloma and some lymphoma etc. 7,[20][21][22] . However, the clinical trials in treating the solid tumors using epigenetic therapy are now turning out as controversial or no effect at all 23,24 and the mechanism is still not clear. For instance, the most recent approved HDAC inhibitor panobinostat

Results
Beneficial outcome of combined implications between ETD and DTD in solid tumor therapy. Epigenetic markers becomes an attractive therapeutic concept in recent years and substantial progression has been made, especially in treating the hematopoietic malignancies 20,25,[32][33][34][35][36] (Table 1). Many of these epigenetic targeting chemicals were licensed to treat patients with defined cancer subtypes. For example, HDAC inhibitors including suberanilohydroxamic acid (SAHA), also known as Vorinostat, and Romidepsin (Rom), were mainly prescribed to treat some types of B cell lymphoma. Decitabine (DEC) is DNMT inhibitor and was licensed to treat myelodysplastic syndrome (MDS). A well-recognized explanation is that the tumor suppressor genes, which are repressed/silenced in cancer, can be reactivated by these drugs 23,37 . However, the HDACi/DNMTi treatment of leukemia cells or non-malignant cells promoted global gene expression alteration which includes the up-regulation of not only tumor suppressors but also the oncogenes 38 ( Supplementary Fig. 1). Calculation of the gene number suggests no difference between the upregulated tumor suppressors and oncogenes (Fig. 1A). Instead, HDAC1 and DNMT1, the main known HDAC and DNMT factors in cancer have highest expression in hematopoietic malignancies, especially the B-cell lymphoma in the pan-cancer expression analysis (cBioportal, Supplementary  Fig. 2) which may be the reason of the effectiveness of HDACi and DNMTi in treating these cancers. By analyzing the drug response of over 700 cell lines from different tissues, the hematopoietic cancer cell lines show obvious high response to several HDAC inhibitors (Fig. 1B), suggesting hematopoietic cells are more sensitive to this category of drugs specifically. Most of the solid tissue-related cancers show no obvious correlation to highly expressed HDACs and DNMTs, which is consistent with the poor outcome in 23 clinical trials exploring the potential of HDACi and DNMTi in treating solid tumor ( Fig. 1C and Table 2) by comparing to the traditional chemotherapeutic drugs treatment (Fig. 1D) 39 .
However, some studies and clinical trials applied the combined therapeutic strategy between ETD and DTD in treating solid tumors [40][41][42][43][44][45] , and many of them are still on-going as listed in Table 3. By analyzing the available literature describing such combined strategies, the beneficial effect of ETD in improving the therapeutic effect of DTD can be observed. The objective response rate (ORR) of cancer patients treated by ETD plus DTD is significantly higher than the patients treated by DTD only (Fig. 1E).

ETD Increases the sensitivity to DTD induced apoptosis.
To understand the combined effects of ETD and DTD on solid tumors, several solid tumor cell lines were tested including SKOV3, MCF-7 and A549 etc. representing ovary cancer, breast cancer and lung cancer respectively. The most commonly used DTD is CP which was chosen to combine with SAHA, Rom and DEC. In addition, 2-DG was selected to potentiate the HDAC activity because 2-DG inhibits glycolysis and whole cell abundance of acetyl-CoA, the substrate for histone acetylation. As shown in Fig. 2A and B, SKOV3 cells show increased sensitivity to 30 uM CP treatment when the cells were treated together with increased SAHA or DEC in the MTT assay. 2-DG, on the other hand, increases the cell viability significantly (Fig. 2C). However, treatment of SKOV3 by DEC alone from 0 uM to 4 uM, by SAHA alone from 0 uM to 12 uM, or Rom from 0 uM to 4 uM doesn't generate as significant impact on  Figures 3A, 4B and C). We also tested Dox, another commonly used DTD in treating multiple types of cancer. Again, SKOV3 cells show significantly increased sensitivity to Dox when the cells were pretreated by Rom (Fig. 2D), or show minimally increased sensitivity to Dox by DEC pre-treatment (Fig. 2E). DEC also increases SKOV3 sensitivity to 3 uM CP, an extremely low concentration that hardly kills any SKOV3 cells alone 46,47 (Fig. 2F), which is consistent to the previous studies 48,49 . The sensitivity to CP, when cells were also treated by ETD like Rom, was also tested in MCF-7 and A549 cells. The increased sensitivity of both cells to CP can be observed Fig. 2G and H. Dead cell count also further confirmed this result ( Supplementary Fig. 3D). We further demonstrated that CP, when combined with Rom or DEC, also promotes the expression of p21 and Bax, indicating the ETD and DTD drug combination inhibits cell proliferation and may increase the apoptosis pressure ( Supplementary Fig. 3E). Comparably, A549, which is more resistant to CP, show dramatically increased response to CP when the cells were treated by CP and Rom or CP and DEC ( Fig. 2H and I). However, Rom alone in MCF-7 cells or A549 cells didn't generate a significant impact on cell survival (Supplementary Fig. 3F and G). Similarly, 2-DG increases the resistance of MCF-7 to CP treatment significantly (Fig. 2J). We also tested the combined effect between CP and Dox. The result shows that there is no synergistic effect between these two drugs (data not shown). Finally, MCF10A cell response to CP was evaluated with the presence of ETDs. A clear increased sensitivity was observed in MCF10A cells (Supplementary Fig. 3H and I), suggesting ETD promotes cell sensitivity to DTD is not cancer cell specific.
ETD potentiates the DNA damage induced by DTD. Lethal    CP or IR-induced DNA damage represented by phosphorylated γH2AX foci detected by immunocytochemistry ( Fig. 3A and B). On the contrary, 2-DG decreases the CP or IR-induced γH2AX foci ( Fig. 3A and B). Western blot of total phosphorylated γH2AX also indicates that ETD increases the DNA damage generated by CP in both SKOV3 and MCF-7 cells ( Fig. 3C and D), suggesting the CP-induced DNA damage is under influence of ETDs without tumor specificity. Again, phosphorylated γH2AX show mild decrease in 2-DG treated samples which represent mild protection to the cells exposed to CP ( Fig. 3C and D). Similarly, DEC and Rom also increases the DNA damage generated by CP or IR in MCF-7 cells significantly and 2-DG decreases the sensitivity to CP or IR in MCF-7 cells ( Fig. 3E and F). To further demonstrate that ETD and DTD drug combination increases the DNA damage, the phosphorylated p53 and phosphorylated ATM were detected to monitor if there is activation of DNA damage repair pathways. As shown in Fig. 3G and H, both MCF-7 and SKOV3 cells show increased phosphorylated p53 and phosphorylated ATM upon treatment by CP and Rom or CP and DEC. However, there is no dramatic increase when the treatment was extended from 12 hours to 24 hours, suggesting the DNA damage response is within a short period of time. These results are consistent to the previous report that SAHA treatment increases the nuclease sensitivity and intercalating agent sensitivity 50 .
ETD increases the DTD accessibility. Since DNA damage generated by CP and Dox relies on the attachment of CP and Dox to DNA directly, we attempted to observe if ETD increases the CP and Dox integration into DNA. The antibody recognizing specifically the CP-DNA adducts was used to detect the CP molecules which covalently bound to DNA, but not the free CP. By extracting genomic DNA from CP treated cells and detecting the CP-DNA adducts via dot-blotting, the retention of CP is revealed ( Supplementary Fig. 4A). As shown in Fig. 4A, increasing the SAHA, Rom and DEC concentration in treating SKOV cells results in the increased CP-DNA adduct abundance and 2-DG, on the contrary, reduces the CP-DNA adducts abundance. Dox is an auto-fluorescent dye with an excitation/emission wavelength at 488/580 nm. However, the fluorescence signal of Dox extinguishes once forming stable bound with DNA 51 . By isolating nucleus and measuring their fluoresce signal at 580 nm through flow cytometry assay, the amount of Dox that forms stable covalent bound with DNA in each individual nucleus can be estimated. As shown in Fig. 4B, the vertical straight line in each panel indicates the Dox fluorescence signal to be high (right) or low (left). Given the total nucleuses that were counted, the treatment dosage of Dox are all the same across all the panels, and the total nuclear absorption of Dox is not influenced by the drug treatment ( Supplementary Fig. 4B), the increase of nucleus population with low Dox signal in DEC, SAHA and Rom pretreatment groups indicates that more Dox forms stable bound to DNA. 2-DG pretreatment, however, increases the population of nucleuses with high Dox signal. These results suggest ETDs increase the retention of both CP and Dox to the chromatin which is the direct reason of increased DNA damage.
ETD decompact the chromatin globally. Chromatin compaction, either locally or globally, is the main target of ETDs 52 . We hypothesized that ETD may potentiate the effectiveness of DTD through loosening chromatin globally which is required for DTD incorporation and DNA damage as seen in Fig. 4A and B. DAPI, a commonly used dye in DNA staining, was chosen to demonstrate the compactness of nuclear DNA using SKOV3 cell 53 . By limiting the dosage and time of DAPI staining, the florescence emission shows patchy distribution within the nucleus in control cells, indicating co-existence of highly compacted chromatin and less compacted chromatin (Fig. 5A). 2-DG increases the abundance of heavily stained chromatin dramatically, whilst DEC and Rom almost erase all the heavily stained chromatin (Fig. 5B), suggesting these drugs are capable of modulating the chromatin compaction globally 54,55 , which is consistent with the effect generated by TSA treatment 56 .
To further validate the global chromatin compaction status influenced by ETD, we extracted histones from ETD treated nucleuses 57 . Acetylated histone H3, a marker of loose chromatin, show consistent upregulation by SAHA, Rom and DEC but not by 2-DG (Fig. 5B Top). Another independent protein to indicate the chromatin compaction is HP1α which is a heterochromatin binding protein and its abundance correlates with the global change of heterochromatin and was used for evaluating global chromatin compaction 58  (bottom), chromatin-bound HP1α significantly decreased by SAHA, Rom and DEC but dramatically increased by 2-DG. These data suggest that ETDs treatment alters both the active chromatin and repressive chromatin simultaneously and globally. In addition, the SKOV3 cell stably expressing fusion H2B-GFP was used to evaluate the effect of ETDs on nuclear chromatin compaction 58,60 . As shown in Fig. 5C, H2B-GFP signal in nuclei of Rom and DEC treated cells are more flatten and even, comparably, 2-DG treatment seems increasing the local compaction of chromatin in some area.
Finally, to directly demonstrate the change of chromatin compaction at DNA level, MNase sensitivity assay was performed 61 . As shown in Fig. 5D and Supplementary Fig. 5, SAHA, Rom and DEC consistently increase the chromatin sensitivity to MNase and the mono-nucleosome bound DNA but not 2-DG, suggesting ETD is able to decompact the chromatin globally.

ETD improves the therapeutic effect of DTD.
To further validate the hypothesis that ETD increases the ability of DTD in killing cancer cells, the breast tumor model was established using SCID mice. 4T1 is a malignant breast cancer cell line originated from mice 62 and was used to establish the tumor models. Both DEC and SAHA significantly retard the tumor growth along the treatment by IR in these engrafted breast tumors (Fig. 6A,B and C). For the tumors exposed to CP treatment, both DEC and SAHA improves the CP effect in shrinking the tumor significantly (Fig. 6D,E and F). Statistics of the tumor weight using the box plot indicates that DEC and SAHA significantly improve the tumor sensitivity to CP (Fig. 6G). Importantly, the western blotting of acetylated H3 increased in tumors treated by CP plus SAHA or DEC, and IR plus SAHA or DEC dramatically, along with the increased DNA damage response signals such as γ-H2AX and cleaved PARP. Also, the apoptosis signal is dramatically increased in these tumors treated with drug combination (Fig. 6I). Like CP, Dox also show a significant synergistic effect when either DEC or SAHA were used to treat the tumor bearing mice (Supplementary Fig. 6). Together, these data strongly indicate that ETD potentiates the therapeutic effect of DTD via increasing the DNA damage generated by DTD.

Discussion
Application of ETD in tumor therapy is supported by several important observations including re-initiating the expression of tumor suppressor genes 10, 23 , improving the radio-sensitivity of cancer cells 31, 43, 63 , blocking the DNA repair activity 64 , and repressing expression of DNA repair genes etc. 9 . However, the investigation on the clinical effectiveness of ETD suggests only very limited subtypes of cancer can be treated by this category of drugs effectively and these cancer mainly is hematopoietic cancers, suggesting that ETD treatment only minimally influences the cell survival 24 . Theoretically, ETD mainly targets the factors that regulate the chromatin status and their effect can be much mild by comparing to DTD. However, up-regulation of HDACs and DNMT genes is very common among the human cancers. So ETD becomes the attractive drug that can be broadly used to treat different types of cancer and many clinical trials are still on-going 23 .
Recovery of tumor suppressor expression was particularly addressed in the cancer subtypes indicated by the ETD drug licenses. However, quite many studies point out the discrepancy that ETD may not only turn on the tumor suppressors but also turn on other genes such as the pro-metastasis genes 65 . Simultaneously, DNA repair genes are repressed by ETD treatment in multiple types of cancer cells 7,9,66,67 , suggesting tumor suppressor genes are not the only target of ETD. These combined effects generated by ETD treatment also raises confusion in understanding the mechanism through which ETD treatment increases the therapeutic effect of DTD 17, 63, 67-70 . In our current study, we propose that ETD globally influences chromatin structure. Although previous studies have pointed out the possibility that global change of chromatin organization is responsible for the increased therapeutic effect generated by ETD and DTD combination, the evidence to directly demonstrate that ETD potentiates DNA damage induced by DTD is absent. For example, TSA has been shown to induce the global chromatin decompaction and leads to a more homogenous distribution 56 . Importantly, such chromatin decompaction has no gene locus specificity because image analysis indicates that the chromatin reorganization occurs to chromatin domains crossing several megabases of DNA, which equivalent to 200 nm to 1 um of change under microscope 56 . In another independent study, VPA treatment evacuates several important factors that are critical to maintaining the higher order chromatin structure such as the members of cohesion and condensins including SMC proteins, SMC-associated proteins and some heterochromatin proteins 50 .
Falk et al., for the first time, correlates the chromatin condensation to irradiation sensitivity directly 31 . This study suggested highly compacted chromatin is more resistant to DNA damage induced by irradiation. This study was followed by others which recap the similar results using different ETDs such as SAHA, 5-aza-CdR, DEC, and TSA etc. 17,43,68 . In fact, several recent studies validated that DTD is influenced by ETD for their DNA damage induction and cell death induction. For example, SAHA increases the apoptosis of Rhabdomyosarcoma cell death when treated together with Dox 70 . A recent study also systematically investigated the synergistic effect of CP and HDAC inhibitors using several cell models and the result is also very promising 71 . Interestingly, another independent study observed the enhanced sensitivity of DNA to nucleases and increased interaction of DNA with intercalating agents simultaneously after the cells were treated by ETDs 50 . All these observations strongly support that ETDs are able to potentiate the therapeutic effect of DTD by loosening the chromatin and increasing the DNA damage sensitivity.
The other benefit of ETD and DTD combined treatment of tumor is to reduce the drug resistance as observed in our data (Fig. 2). Traditionally, the main reason of drug resistance is the appearance of DTD resistant cancer cell population after a period of tumor therapy. The mechanism of drug resistance is complicated, but the dosage limitation of DTD is critical since the patient tolerance have to be counted. The dosage lower than the threshold to kill all the tumor cells will leave some cells alive which eventually grow as drug resistance colonies. ETD treatment, however, will program the cells to a more vulnerable status that can be killed by even low dosage of DTD. In other words, the threshold to kill the majority of tumor cells is lowered by ETD co-treatment. Therefore, an extra benefit to the patients will be reduced DTD toxicity and the associated side effect because of the lower DTD dosage in addition to the higher efficiency in killing tumor cells.

Materials and Methods
Gene enrichment analysis. The total tumor suppressors and oncogenes were obtained by searching  integrated all these trials and collected 119 cases in total. We calculated the pooled unadjusted odds ratio with 95% CIs for each study using a random-effects model. Analysis was performed with Revman software.
MNase Assay. Briefly, the MNase assay was performed according to liu et al. 31    Animal assay. Experimental setup for irradiation therapy part: 1*10 6 4T1 breast cancer cells were inoculated into subcutaneous of SCID mice hind limb on Day 1. Mice were randomly divided into three groups on Day 14. For each group, N = 12 tumors. The mice in ETDs combined with IR groups received 0.4 mg/kg Dec or 30 mg/ kg SAHA intravenous injection via tail vein on Day 15,16,17, at the same time control and IR group receive solvent injection. All mice received 10 Gy irradiation on D18. Experimental terminal is D22. Tumor volume was measured and calculated by the formula V = (length × Width^2) × π/6. Tumors were surgically removed from mice on Day 22. For cisplatin therapy part: 1*10 6 4T1 breast cancer cells were inoculated into subcutaneous of SCID mice hind limb on Day 1. Mice were randomly divided into three groups on Day 14. For each group, N = 12 tumors. The mice in ETDs combined with cisplatin groups received 0.4 mg/kg Dec or 30 mg/kg SAHA intravenous injection via tail vein on Day 14,15,17,18, at the same time control and cisplatin group received solvent injection. All mice received 5 mg/kg cisplatin on D16 and D18. Experimental terminal is D21. Tumor volume was measured and calculated by the formula V = (length × Width^2) × π/6. Tumors were surgically removed from mice on Day 22. For Doxorubicin therapy, the Doxorubicin dosage was 10 mg/kg intravenous injection via tail vein. Other time and process was similar with cisplatin therapy design.
All the animal related experimental procedures were approved by Animal Ethics Committee of University of Macau (AECUM) and the related experiments were conducted in accordance with the guideline of AECUM.
MTT assay, Typan blue staining assay and Western blot assay are all in supplementary.

Experimental Statistical Analyses.
Results for continuous variables are presented as means ± standard deviation (SD) unless stated otherwise. All data are evaluated by 2-tailed student t-test and P < 0.05 was considered statistically significant.