Schedule dependent synergy of gemcitabine and doxorubicin: Improvement of in vitro efficacy and lack of in vitro‐in vivo correlation

Abstract Combination chemotherapy is commonly used to treat late stage cancer; however, treatment is often limited by systemic toxicity. Optimizing drug ratio and schedule can improve drug combination activity and reduce dose to lower toxicity. Here, we identify gemcitabine (GEM) and doxorubicin (DOX) as a synergistic drug pair in vitro for the triple negative breast cancer cell line MDA‐MB‐231. Drug synergy and caspase activity were increased the most by exposing cells to GEM prior to DOX in vitro. While the combination was more effective than the single drugs at inhibiting MDA‐MB‐231 growth in vivo, the clear schedule dependence observed in vitro was not observed in vivo. Differences in drug exposure and cellular behavior in vivo compared to in vitro are likely responsible. This study emphasizes the importance in understanding how schedule impacts drug synergy and the need to develop more advanced strategies to translate synergy to the clinic.


| I N TR ODU C TI ON
Combination chemotherapy is commonly employed as front-line therapy in many advanced malignancies. 1-5 Chemotherapeutic agents are combined to overcome various mechanisms of single drug resistance and to enhance antitumor activity compared to single drugs. While drug combinations are commonly selected by combining drugs with different mechanisms of dose-limiting toxicity, combination therapy is often only marginally more effective at reducing tumor burden while inducing more toxicity compared to single drug therapy. 6 When establishing how the drugs are administered in combination, dosing regimens are typically based upon the drug administration schedules and maximum tolerated doses (MTDs) determined during phase I trials for the individual drugs. 7 The dose of each drug in combination is administered as close to the single drug MTD as possible, which does not necessarily result in the highest therapeutic index possible for the given drug pair.
To reduce systemic toxicity, synergistic drugs or drugs that work favorably together, are combined in hopes of lowering the required dose to achieve a therapeutic effect. In the past few decades, there have been considerable efforts in developing methods to empirically identify synergistic drug pairs in vitro. [8][9][10][11] Many in vitro studies have shown that drug synergy is dependent on how the drugs are combined, such as the molar ratio of the drugs in the combination; therefore, precision is required in translating these synergistic drug pairs effectively to the clinic. 11 While often neglected, drug schedule plays a significant role in the efficacy of a given combination drug pair. For example, many in vitro studies have shown that sequentially staggering chemotherapeutic agents in vitro can increase drug combination activity and synergy. [12][13][14][15] Often times, temporally staggering drugs can increase levels of apoptotic activity [16][17][18] or induce cell cycle mediated effects 19,20 which result in greater cell death. Although there are many studies which study the mechanisms that are responsible for schedule dependent synergy, few studies also determine if this synergy translates to an in vivo model. Furthermore, even fewer studies directly compare different combination drug schedules in vivo, to evaluate if the optimal schedule identified with in vitro studies is also applicable in the in vivo environment.
Improving combination chemotherapeutic regimens is particularly important in triple negative breast cancer, as cells do not over-express the common receptors (progesterone (PR), estrogen (ER), and human epidermal growth factor (HER2)) which are the common therapeutic targets for other breast cancer types. 1,21 While triple negative breast cancer patients are typically responsive to chemotherapy, patients with metastasized triple negative breast cancer have poor prognosis. Many combination regimens have been evaluated in the clinic to treat these difficult cases; however, toxicity has limited the utility of the combination therapies. 1 Here, using the triple negative breast cancer cell line MDA-MB-231 as a model cell line, we study the impact of drug schedule on the activity of chemotherapeutic combinations in vitro and in vivo. We first identify a synergistic drug pair from a panel of FDA approved drugs, and then show that drug schedule governs the activity of the drug combination in vitro. We further show that the drug combination is effective at treating tumor-bearing mice; however, further work is required to optimize the synergy in vivo.

| Cell culture
Cells were maintained in their proper media supplemented with 50 I.U./ ml penicillin and 50 lg/ml streptomycin (Thermo Fisher Scientific) in a humidified incubator at 378C and 5% CO 2 . MDA-MB-231 human breast cells (ATCC) were cultured in RPMI-1640 media (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum. MCF-10a human nontumorigenic breast epithelial cells (ATCC) were cultured in MEBM media kit supplemented with hydrocortisone, hEGF, insulin, and bovine pituitary extract (Lonza) in addition to 100 ng/ml cholera toxin.

| Cell viability assays
Cell viability was measured by MTT assay after exposure to various chemotherapeutic agents (LC Laboratories). Cells (5 x 10 3 MDA-MB-231 or 1 x 10 4 MCF-10a) were allowed to adhere overnight in 96-well plates.
Cells were then incubated with single drugs or drug combinations for 72 hr (unless specified otherwise). When sequentially exposing drugs to cells, the first drug was aspirated and replaced with the second drug in media by pipetting along the side of the wells to avoid physically disrupting the cells. After drug incubation, cells were incubated with media containing 0.5 mg/ml MTT (Thermo Fisher Scientific) for 3.5-4 hr. After replacing the MTT solution with DMSO, the plate was shaken for 30 min to solubilize the formazan crystals, and the absorbance of each well was measured at 570 nm (Tecan Infinite M200 Pro). Fractional cell inhibition was then calculated by subtracting the viability of the treated cells from the viability of the control cells, normalized by the viability of the control cells.
Fractional cell inhibition was measured after exposure to drugs at a range of concentrations to generate dose-response curves using the median effect model. 22 The dose response curves were used to calculate IC 50 values and the combination index (CI), which determines synergy of drug combinations. 22 Error bars for the CI were calculated by propagating the confidence intervals from the individual drug fits and the error in toxicity for the combination treatment. When calculating the CI, single drug dose response curves were used with the identical schedule as given in the combination treatment.

| Caspase 3 fluorometric assay
Induction of apoptosis was determined by measuring caspase 3 activity using the EnzChek caspase 3 assay kit (Thermo Fisher Scientific) according to manufacturer's instructions. Briefly, cells were seeded into T25 flasks (7.5 ml; 1 x 10 5 cells/ml) and allowed to adhere overnight. Following incubation with drug formulations, both floating and adherent cells were washed with PBS and lysed. Cell lysates were then centrifuged at 5,000 rpm for 5 min and the supernatant was collected. Caspase 3 activity was quantified by incubating 50 ml of the supernatant with 50 ml of the 23 working solution containing the substrate Z-DEVD-AMC for 30 min at 378C. The fluorescence intensity was then measured of each well at an excitation/emission of 342/441 nm (Tecan Plate Reader). Caspase 3 activity was calculated with respect to an AMC standard curve.
Caspase activity was then normalized to the total protein content, using the micro BCA assay (Thermo Scientific). Briefly, 15 ml of the leftover supernatant was diluted with 135 ml of DI water and incubated with 150 ml of the BCA solution for 1.5 hr at 378C. After incubation, protein content was determined by reading the absorbance at 562 nm (Tecan Plate Reader) with internal protein standards of bovine serum albumin.

| Identification of synergistic chemotherapeutic drug pairs
To identify synergistic drug pairs, MDA-MB-231 cells were exposed to a panel of FDA approved chemotherapeutic agents with different mechanisms of action, including doxorubicin (DOX), paclitaxel (PTX), ixabepilone (IXA), and gemcitabine (GEM). Cell viability was measured after exposure to single drugs using the MTT assay to generate dose response curves and determine IC 50 concentrations (Figure 1). Cells were responsive to both mitotic inhibitors PTX and IXA, with IC 50 concentrations of 19 6 5 nM and 15 6 5 nM, respectively, and to DOX, with an IC 50 concentration of 0.28 6 0.02 mM. Cell growth was inhibited by GEM at low concentrations (nM range); however, high GEM concentrations (>1 mM) were ineffective at killing more cells, resulting in a very large IC 50 concentration.
After establishing the IC 50 concentration of each drug, drugs were tested in combination. First, cells were exposed concurrently to a combination of drugs at single drug concentrations between the respective IC 25  hr, respectively (Supporting Information Figure S1). Furthermore, the IC 50 is increased to 0.42 6 0.03 or 1.6 6 0.3 mM, if exposure to DOX is delayed by 4 or 24 hr, respectively. Therefore, to understand the impact that drug schedule has on drug-drug interactions, drug synergy must be evaluated using dose response curves for the single drugs, at identical exposure schedules which were given in the combination treatment, as previously done by Chou. 13

| Schedule dependent synergy between GEM and DOX
Because GEM and DOX were both synergistic when given concurrently and showed a strong schedule dependence, this drug pair was studied in detail. MDA-MB-231 cell viability was measured after incubating cells with different sequences of GEM and DOX at different drug doses ( Figure 3A). Sequentially exposing the cells to GEM ! DOX or concurrently to GEM and DOX inhibited more cell growth than exposing the cells to DOX ! GEM at each drug dose tested. Combination indices were calculated for each GEM and DOX combination ( Figure 3A) based on single drug dose response curves (Supporting Information Figure S1), which were made with identical exposure schedules compared to the combination treatments. It was more synergistic (lower CI) to incubate the cells with GEM prior to DOX than to expose the cells to GEM and DOX simultaneously or to DOX prior to GEM.
In addition to the triple negative breast cancer cell line (MDA-MB-231), the nontumorigenic breast epithelial cell line MCF-10a, which served as a control, was exposed to the schedule of GEM (24 hr) ! DOX (48 hr). The effect of molar ratio on drug synergy is shown for each cell line in Figure 3B. The schedule of GEM ! DOX was synergistic on MCF-10a cells for GEM : DOX molar ratio less than one; however, it was antagonistic for GEM : DOX molar ratios greater than one. On the contrary, the schedule of GEM : DOX was extremely synergistic (CI < 0.2) on the triple negative breast cancer cells at all of the molar ratios tested.  It is important to note that the repeated injections of the low doses of GEM and DOX caused no observable toxicity as evident by no change in body weight throughout the course of the study ( Figure 6D).

| D I SCUSSION
There have been extensive efforts to identify synergistic chemotherapeutic agents to improve the efficacy of chemotherapy. While many studies have identified synergistic drug pairs for specific cancer cells in vitro, few report if the synergy offers improved therapeutic benefits in vivo. Here, we specifically studied the impact that drug schedule has on synergy in vitro, and if optimal drug schedules identified in vitro can be used to more effectively treat tumor-bearing mice in vivo.
To identify a synergistic drug pair, four FDA approved chemotherapeutic agents were screened to treat MDA-MB-231 cells. The dependence of synergy on schedule was studied in detail for DOX and GEM, which was synergistic when given concurrently and demonstrated a strong schedule dependence on cell growth inhibition. Previous clinical trials have shown benefits to combining DOX and GEM while treating advanced and metastatic breast cancer patients; however, neutropenia was often reported as limiting the administered dose. [23][24][25][26] Therefore, understanding how to improve the potency of GEM and DOX at lower drug doses is of interest.
Giving GEM prior to DOX was significantly more synergistic than giving the two drugs concurrently or in the reverse sequence. The increase in synergy upon giving GEM prior to DOX was consistent with an increase in caspase activity ( Figure 4). Previous reports have also shown that the schedule in which cancer cells are exposed to chemotherapeutic drugs can significantly impact the induction of proapoptotic pathways. 16 Furthermore, the advantages of exposing cells in vitro to GEM prior to a topoisomerase inhibitor have previously been reported. 27 50 of DOX (0.28 mM) in vitro was significantly lower than GEM (>10 mM); however, GEM was much more effective than DOX at inhibiting tumor growth in vivo. Although GEM was given at a higher dose than DOX, 20 mg/kg compared to 2 mg/kg, GEM is cleared faster than DOX. 36,37 The drugs were also administered sequentially in combination (GEM ! (1 day) DOX and DOX ! (1 day) GEM) for one injection and four repeated injections. Both combination drug regimens were more effective at inhibiting tumor growth than the individual drugs on their own. In fact, four repeated cycles of both drug combination regimens stopped tumor growth 40 days after the first injection. In combination, the cumulative dose of both DOX and GEM is much lower than the single drug doses required to inhibit tumor growth to the same extent. [32][33][34] Interestingly, the sequential administration of DOX ! GEM (1 cycle or 4 cycles) is essentially just as effective as the sequential administration of GEM ! DOX, contradictory with the in vitro cell viability and apoptosis assays.
Previous studies have shown that optimal drug exposure schedules identified in vitro can offer improved therapeutic benefits in vivo 16,38,39 ; however, we demonstrate here that it can be nontrivial to translate optimal drug schedules in vivo. The difference in how drug schedule impacts GEM and DOX activity in vivo compared to in vitro is interesting and is likely due to differences in the tumor environment and drug exposure. In vivo, the cancer cells will likely develop different characteristics and respond to the drug combination differently, due to the complex tumor microenvironment consisting of stromal and immune cells. [40][41][42] In addition, tumor growth in response to chemotherapy can be impacted by the immune response in vivo 43 ; however, the impact of drug combination schedule on the activation of the immune system was not accounted for in the in vitro studies. Many advanced in vitro models have been developed, which attempt to replicate the complex tumor environment, and they should be used in future studies to further study the impact of drug schedule in combination therapies. 44 Another significant challenge with regards to optimizing schedule in combination therapies, is the fast and variable clearance of each drug in vivo. For example, here, GEM and DOX are cleared with elimination half-lives of 30 min and 10 hr, respectively. 36,37 The impact of these pharmacokinetic parameters on the in vitro conclusions about drug synergy is unknown and challenging to accurately study in vitro.
To take full advantage of the benefits of combining multiple chemotherapeutics, it is necessary to precisely control how cells are exposed to the drugs in vivo.
The development of combination delivery platforms has made it possible to control the relative concentration and exposure sequence of multiple drugs in vivo. In fact, various delivery systems have already been engineered to deliver DOX and GEM simultaneously in the same carrier. [45][46][47][48][49] By delivering multiple therapeutics in a single vehicle, drug ratios, and release kinetics can be finely controlled to more closely mimic optimal drug conditions discovered in vitro. 50,51 As the field continues to develop, an emphasis needs to be placed on translating synergistic drug interactions by optimizing drug ratio and release rates from these delivery systems.
The challenge in being able to predict the activity of a drug combination in vivo on a single cell line, which was studied extensively in vitro, reiterates the difficulty in improving combination chemotherapy regimens in human patients. Future improvements in combination therapies will not only depend on the ability to improve translation from in vitro assays to animal models, but also on the translation from animal models to human patients. The development of tools to quickly VOGUS ET AL. study the response of patient tumors to combination treatments can help improve the optimization of drug combinations. 52,53 Moving forward, it is critical to simultaneously study the impact of drug schedule on tumor regression in multiple tumor phenotypes, while also studying the impact of drug schedule on off-site toxicity.

| C ONC LUSI ON S
The use of synergistic drug pairs can improve the efficacy of combination chemotherapy by lowering required drug doses and reducing toxicity; however, synergistic drug interactions often require very specific conditions. The impact of drug schedule on synergy for DOX and GEM