Receptor tyrosine kinases (RTK) are key signaling molecules in regulating cancer cell growth and are important cancer drug targets. Despite the success of specific RTK-targeting therapy in certain cancer treatments, the overall response rates are limited to the drug target–stratified populations. We have systematically studied RTK activations in a panel of cancer cell lines, primary cancers, and cancer xenografts and found that different combinations of RTKs were activated in different cancer cells regardless of their tissue origins. Combinations of specific RTK inhibitors (RTKi) preferentially inhibited proliferation of the cancer cells with corresponding RTK activation profiles. We also found that the activations of RTKs were regulated by both cell-autonomous and environment-dependent mechanisms and demonstrated that inhibition of all activated RTKs was essential to completely block cancer cell proliferation. In addition, c-Myc downregulation was identified as an indicator for the effectiveness of the RTKi combination treatments. Our findings demonstrated that the RTK activation profile is a valid biomarker for diagnosis and stratification of cancers, and a corresponding combination of RTKis is a promising strategy to treat cancers, particularly the single RTKi therapy–resistant cancers, selectively and effectively. Mol Cancer Ther; 15(10); 2508–20. ©2016 AACR.
Cancer is a complex multifactor disease (1). Genetic mutations, epigenetic alterations, and multiple growth signaling pathways are involved in the formation and progression of cancers (2, 3). However, targeted anticancer therapies have been mainly focusing on inhibition of single vital oncogenes, so-called oncogenic drivers (4, 5). Thus, the application of targeted therapies is limited. For example, in HER2-overexpressing breast cancers, the overall response rate to Herceptin, a specific mAb against HER2, is only 15% (6). A majority of cancer patients could not benefit from the single-target therapies, because they do not have the genetic alterations in the drug-targeted genes (5). Most of the cancer cells depend on more than one signaling element for their growth.
In addition to the multifactor nature, cancer cells are also highly heterogeneous (7, 8). Different cancer cells contain different sets of mutations and rely on different sets of growth regulators for their growth (9). RNA interference screening experiments have demonstrated that cancer cells could be distinguished from one another on their kinase requirements, even when they were isolated from the same cancer type, indicating that the control of survival and proliferation is significantly different between individual cancers (10, 11).
On the basis of the multifactor and heterogenetic nature of cancer cells, an effective therapeutic strategy to treat a cancer is to target the cell-specific set of growth regulators simultaneously (12, 13). The multitarget drugs, such as Sutent and sorafenib, partially overcome the efficacy problem of the single-target drugs but increase the risk of side effects due to the lack of target selectivity (14–16). On the other hand, the combination of specific single-target drugs may selectively and effectively target specific cancers (17). In fact, combination therapies targeting both hepatocyte growth factor receptor (HGFR or MET) and EGFR in NSCLC patients are now under clinical investigation (ClinicalTrials.gov Identifier NCT01887886, NCT01911507, and NCT02318368). Further investigations into the combination strategy will promote better personalized cancer treatment.
The RTKs are the key cell growth signaling regulators, and many of them are validated drug targets for targeted cancer therapies (18, 19). The expressions and genetic alterations of RTKs are diversified among different types of cancers. Therefore, it is an attractive strategy to identify specific sets of RTKs that are expressed and activated in specific cancer cells, and to use a combination of RTK inhibitors (RTKi) to specifically target the cancer cells (20). Currently, RTK-targeted cancer therapies mainly rely on diagnosis of genetic alterations of the RTKs, that is, gene amplifications, point mutations, and/or translocations. However, genetic alterations do not always correlate with their functions. Furthermore, a large number of cancer cells do not have genetic alterations in their RTK genes (5, 12, 21). All these factors restrict the application of the RTK-targeted therapies. Therefore, better strategies for targeted cancer therapies are urgently needed.
Here, we report an RTK phosphorylation/activation pattern–based personalized diagnosis and treatment of cancers. We found that diversified combinations of RTKs were activated in various types of cancer cells. Combinations of multiple specific RTKis preferentially and effectively inhibited the proliferation of the cancer cells with corresponding RTK activation profiles. Our findings strongly suggest that the RTK activation profile of a cancer is a valid biomarker for diagnosis, and a combination of the corresponding RTKis is a promising therapeutic approach to treat cancers.
Materials and Methods
BEL7404, SMMC7721, and FHCC98 were gifts from Prof. Junying Yuan (Shanghai Institute of Organic Chemistry, Shanghai, China). LM3 was a gift from Prof. Hongyang Wang (Eastern Hepatobiliary Surgery Institute, Shanghai, China). The cells were authenticated within 6 months before use according to the provider's recommendations and were tested negative for mycoplasma contamination. All other cell lines were obtained from ATCC from 2006 to 2010, passaged for fewer than 6 months, and maintained as recommended by the provider. ATCC performed cell line authentication using DNA fingerprinting by short tandem repeat analysis. The WiDr, LS174T, BT474, T47D, MDA-MB-453, LM3, SMMC7721, FHCC98, A375, HeLa, and A498 cells were cultured in DMEM (Invitrogen) with 10% FBS. All other cells were cultured in RPMI1640 medium (Invitrogen) with 10% FBS.
The tyrosine kinase inhibitors were purchased from Selleck Chemicals (for in vitro assays) and LC Laboratories (for in vivo assays). The following antibodies were purchased from Cell Signaling Technology: phospho-EGFR (#3777), EGFR (#4267), phospho-Her2 (#2243), phospho-Her4 (#4757), phospho-InsR/IGF1R (#3024), IGF1R (#3018), InsR (#3025), phospho-Met (#3077), Met (#3127), phospho-Akt (#4060), Akt (#9272), phospho-Erk1/2 (#9101), Erk1/2 (#9102), and c-Myc (#5605). The antibody for α-tubulin was purchased from Santa Cruz Biotechnology (#SC-5286), and the antibody for GAPDH was purchased from Shanghai Kangchen (#KC-5G4).
Primary samples of hepatocellular carcinomas, colon cancers, stomach cancers, breast cancers, prostate cancers, and renal cancers were obtained from the Eastern Hepatobiliary Surgery Institute, Renji Hospital, Tongji Hospital, Huashan Hospital, Ruijin Hospital, and Xinhua Hospital (Shanghai, China), respectively. Primary cancers were instantly placed into liquid nitrogen after surgical resection until phospho-RTK analysis or immediately processed to establish patient-derived xenografts. Written informed consents were obtained from the patients, and the studies were approved by the ethics committees of each hospital.
RTK phosphorylation/activation profiling
A total of 5 mg protein lysates of in vitro cultured cells and 10 mg protein lysates of clinical samples and mouse xenografts were analyzed using phospho-RTK arrays (R&D Systems). The arrays were photographed using Image Station 4000MM PRO system (Carestream), and the pixel densities of various spots were collected and quantified with Quantity One Software (Bio-Rad). The average signal (pixel density) of the pair of duplicate spots was determined for each RTK. A signal from the PBS-negative control spots was used as a background value, and signals of reference spots in the corners were used for normalization among different arrays. The relative phosphorylation/activation value of RTKx (phospho-RTKx index) was calculated according to the following formula: phospho-RTKx index = (INTx−INTnc)/(INTref−INTnc). INTx is the pixel density of RTKx, INTnc is the pixel density of background, and INTref is the density of reference spots. Receptors were considered to be phosphorylated/activated when index values were greater than 0.1.
Cells were harvested in RIPA buffer. Protein lysates were separated by SDS-PAGE, transferred to nitrocellulose membranes (GE Healthcare), probed with first antibodies, and then incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies. Immune complexes were detected by Immobilon Western Chemiluminescent HRP Substrate (Millipore) and photographed using Image Station 4000MM PRO system (Carestream).
Female BALB/c athymic (nu/nu) nude mice (age, 4–6 weeks; weight, 18–20 g) were purchased from Vital River Laboratory Animal Technology Co., Ltd. and kept under specific pathogen-free conditions with a 12-hour light/dark schedule at 25°C, and fed with autoclaved chow diet and water ad libitum. Housing and all procedures were performed according to protocols approved by the Institutional Animal Care and Use Committee at Shanghai Institute of Materia Medica (Shanghai, China). Animals were allowed to acclimate for at least 5 days before any handling. For ectopic implantation experiments, 5 × 106 cells per mouse were injected either subcutaneously into the flanks or intraperitoneally into the abdominal cavities of nude mice. For orthotopic implantation experiments, 5 × 105 cells per mouse were transplanted into the liver lobes of nude mice as described previously (22). For metastasis assays, 5 × 106 cells per mouse were injected into the lateral tail veins of nude mice. Tumors growing at various locations (intestinal wall, liver, lung, ovary, or bone) were isolated and performed for phospho-RTK analysis. Patient-derived colon cancer xenografts were established by implanting the cancer cells into the flanks of nude mice subcutaneously as described previously (23).
Mouse treatment studies
In subcutaneous xenograft studies, once tumors reached an average volume of approximately 50 mm3, mice were randomized into treatment groups (n = 6/group) and treated orally with vehicle, lapatinib (10 or 20 mg/kg), osi-906 (OSI; 20 mg/kg), crizotinib (1 or 2 mg/kg), or drug combinations (10 mg/kg lapatinib + 1 mg/kg crizotinib, 20 mg/kg lapatinib + 2 mg/kg crizotinib, or 20 mg/kg lapatinib + 20 mg/kg OSI). Tumor volume and body weight were assessed twice weekly [tumor volume = 1/2 (length × width2)]. The studies were terminated when the tumors of vehicle group reached approximately 1,500 mm3.
Cell growth inhibition analyses
IC50 values of the compounds were calculated from 72-hour cell culture MTT assays. To determine IC50 ratios, the IC50 value of the most sensitive treatment (MST) was used as a common denominator. As lapatinib was a component of all combinations, the IC50 ratio for each combination was derived by dividing the IC50 value for lapatinib by the MST IC50 for lapatinib. The IC50 ratios for single drugs were derived by dividing the IC50 value for that single drug by the MST IC50 value for that drug (if present in the MST) or for lapatinib (if not present in the MST). To evaluate the synergy effects of RTKi combinations, the combination index values were calculated by using the method of Chou and Talalay (24). The CIs <0.9 indicate synergy effects, the CIs between 0.9 and 1.1 indicate additive effects, and the CIs >1.1 indicate antagonistic effects. For measuring real-time cell growth, the xCELLigence RTCA system (Roche Applied Science) was used.
Statistical analyses were performed by using Prism version 6.01 (GraphPad Prism Software). Data were graphically represented as mean ± SEM. Two-way ANOVA with Tukey post hoc test was used for calculating tumor volumes in mouse xenograft studies. Statistical significance was established for P < 0.05 (*), P < 0.01 (**), and P < 0.001(***).
Multiple RTKs in different combinations were activated in cancer cells
To understand the efficacy and selectivity of specific RTKis, we chose four commonly used RTKi anticancer drugs, that is, lapatinib targeting the EGFR family (25), osi-906 (OSI) targeting the insulin receptor (IR) family (26), jnj38877605 (JNJ) targeting the HGFR family (27), and axitinib targeting the VEGFR family (28) and evaluated their effects on the growth of a panel of cancer cells of different tissue origins (Supplementary Table S1). Only the cancer cells that had genetic alterations and were addicted to the altered kinases were sensitive to a single RTKi treatment. The cancer cells without known genetically altered RTKs were relatively insensitive to the single RTKi treatment. These results indicated that very limited cancer cells can be treated by single RTK-targeted drugs at the concentrations that are sufficient to inactivate the targeted RTKs. Apparently, additional molecules were responsible for mediating the growth signals to support the cancer cell growth.
To understand the nature of the resistance of the cancer cells to the single-RTKi treatment and to identify the key molecules responsible for the resistances, we analyzed the RTK phosphorylation/activation profiles of a panel of cancers. The samples were from 10 different tissue origins, including breast, lung, liver, colon, prostate, kidney, intestine, stomach, skin, and cervix. Similar results were obtained from the cancer cell lines, the primary cancer samples, or the xenograft samples. A total of 90% (31/33) of the cancer cell lines, 80% (17/21) of the cancer xenografts, or 60% (14/24) of the primary cancer tissues had more than one RTK activated (Supplementary Fig. S1A–S1C). The number of the phosphorylated RTKs in the primary samples could be underestimated because of the limited sizes of the samples. We also collected the RTK activation information of cancer samples from published literatures (Supplementary Tables S2 and S3). More than 80% of the 86 cancer cell lines and the 35 primary cancers had more than one RTK activated (Supplementary Fig. S1D and S1E).
The most frequently activated RTK was EGFR, which was activated in more than 80% of the cancer cells (Supplementary Fig. S1F). Her2, Her3, InsR, IGF1R, and Met were among the top frequently activated RTKs in cancer cells (higher than 20%). VEGFR, FGFR, AXL, PDGFR, EPHR, and RET families were also activated in some of the cancer cells. The phosphorylation patterns of RTKs were very heterogeneous in both the in vitro and in vivo growing cancer cells without apparent tissue-specific patterns (Fig. 1A and B). These data demonstrated that multiple RTKs in different combinations were activated in specific cancers, which may account for the resistance of cancer cells to the single-RTKi treatments.
Combinations of multiple RTKis synergistically inhibited cancer cell growth by inducing G1 cell-cycle arrest
To address the function and contribution of each of the activated RTKs in supporting cancer cell growth, we inhibited the RTKs using specific RTKis individually or in combinations and analyzed the effects of the RTKis on cancer cell growth. We selected cancer cell lines with activations in one or more of the four inhibitor-targeted RTKs to test the effects of the RTKis in different combinations on the cell growth and the downstream signaling molecules.
The first cell line we chose was the human lung cancer cell line H522, in which two members of the EGFR family (Her2 and Her4) were activated (Fig. 2A). Lapatinib dose dependently inhibited the activation of the two receptors, the downstream signaling molecules, and the cell growth (Fig. 2B and Supplementary Fig. S2A–S2C). Lapatinib was about 40 times more potent at inhibiting Her2 than inhibiting Her4 in in vitro enzymatic activity assays (25). Indeed, lower doses of lapatinib (<2 μmol/L) inhibited only Her2 and partial cell growth. A complete inhibition of both RTKs by higher doses (>10 μmol/L) was required to stop cell growth. These data suggest that both Her2 and Her4 activations are required for the growth of H522 cells.
The second cell line we studied was the human liver cancer cell line BEL7404, which had three RTKs from two families activated (EGFR of EGFR family, InsR and IGF1R of IR family; Fig. 2C). Lapatinib or OSI inhibited their respective receptors effectively (Supplementary Table S4), but had little effects on cell growth (Fig. 2D). However, a combination of them synergistically inhibited the cell growth in a dose-dependent manner (Fig. 2D and E; Table 1). Further experiments demonstrated that the growth inhibition was caused by induction of G1 cell-cycle arrest (Fig. 2F), but not by cell death when the drugs were used at the concentrations that completely inhibited the phosphorylation of the two RTKs (Supplementary Fig. S2D), suggesting that the activated RTKs mainly supported the proliferation, but not survival, of the cancer cells.
|.||.||CI at .|
|Cell line .||Combination treatment (ratio) .||IC50 .||IC75 .||IC90 .|
|BEL7404||LAP + OSI (1:1)||0.3476||0.6000||1.0459|
|SMMC7721||LAP + OSI + JNJ (10:10:1)||0.1230||0.0215||0.0039|
|WiDr||LAP + OSI + JNJ (10:10:1)||0.1005||0.1042||0.1129|
|DU145||LAP + OSI + JNJ (10:10:1)||0.2901||0.0279||0.0027|
|.||.||CI at .|
|Cell line .||Combination treatment (ratio) .||IC50 .||IC75 .||IC90 .|
|BEL7404||LAP + OSI (1:1)||0.3476||0.6000||1.0459|
|SMMC7721||LAP + OSI + JNJ (10:10:1)||0.1230||0.0215||0.0039|
|WiDr||LAP + OSI + JNJ (10:10:1)||0.1005||0.1042||0.1129|
|DU145||LAP + OSI + JNJ (10:10:1)||0.2901||0.0279||0.0027|
Abbreviations: CI, combination index; LAP, lapatinib.
aCombination index values for combination treatments were generated for the concentrations at which 50%, 75%, or 90% of the cell growth were inhibited (IC50, IC75, and IC90, respectively). CI < 0.9, synergy; 0.9 < CI < 1.1, additive.
The third cell line was the human liver cancer cell line SMMC7721, in which three RTKs from three different families were activated (EGFR of the EGFR family, IGF1R of the IR family, and Met of the HGFR family; Fig. 2G). Lapatinib, OSI, or JNJ inhibited their respective RTKs (Supplementary Table S4), but only the combination of the three synergistically and dose dependently inhibited the cell growth in either two- or three-dimensional cultures (Fig. 2H and I and Supplementary Fig. S2E; Table 1). These results suggested that all of the activated RTKs contributed to the cancer cell growth, and inactivating all of them was essential to completely stop the cancer cell growth.
The RTK activation profile-based combinational treatment was cell type independent but RTK phosphorylation pattern dependent
To test whether the RTK activation profile-based combinational treatment is cell type dependent, we chose three cell lines with similar patterns of RTK activations but from different tissue origins, the colon cancer cell line WiDr, the prostate cancer cell line DU145, and the liver cancer cell line SMMC7721. All of them had EGFR, IR family, and Met activated (Figs. 3A and 2G). We examined the effects of the same combinational treatment on the RTK phosphorylation and the growth of these cells. The concentrations of the RTKis required for the complete inhibition of the corresponding RTKs (IC99s) were cell type independent, around 500 nmol/L for lapatinib, 500 nmol/L for OSI, or 50 nmol/L for JNJ (Supplementary Fig. S3A–S3C; Supplementary Table S4), despite the differences in the levels of expression and activation of the RTKs (Supplementary Fig. S3D). We therefore combined the three inhibitors, lapatinib, OSI, and JNJ, with a ratio of 10:10:1, to treat the cancer cells. The combination synergistically and potently inhibited the growth of all three types of cancer cells by inducing G1 cell-cycle arrest, while single inhibitors had little effect (Fig. 2H and I and Fig. 3B–E). Another Met inhibitor crizotinib showed similar effects when combined with lapatinib and OSI (Supplementary Fig. S3E). These data suggested that the RTK phosphorylation profile-based combinational treatment was cell type independent.
To further find out whether the effectiveness of the RTKi combination treatments on the cancer cell growth was determined by their RTK activation profiles, we reclassified the cancer cells according to their RTK activation profiles (Supplementary Fig. S3F) and examined the effects of different RTKi combinations on the growth of different cancer cells with different RTK activation patterns (Table 2). The cytotoxic drug vincristine was used as a nonspecific drug control. The growth-inhibitory effects of vincristine on these cells were all similar, the IC50 of which ranged from 1.3 to 6.4 nmol/L. The growth-inhibitory effects of RTKi combinations, on the other hand, were rather specific. The treatments that covered all of the activated RTKs (shadowed areas in Table 2) were more potent than the treatments that did not (Table 2, compared horizontally), suggesting that the activated RTKs were critical for growth of cancer cells and that selective inhibition of the activated RTKs was an efficient strategy to inhibit cancer cell growth. Moreover, cells with the same RTK activation pattern demonstrated similar specificity and potency for the RTKi treatment. Therefore, RTKi combinations were selective inhibitors of the cancer cell growth, and their selectivity was determined by the RTK activation pattern, instead of the tissue origin, of the cancer cells.
Dual inhibition of Erk and Akt and reduction of c-Myc were essential for effective inhibition of cancer cell growth by the RTKi combinations
To understand the mechanisms of RTKi combinations on inhibiting cancer cell growth and to search for a molecular marker to predict the growth-inhibitory capability of the combinations, we examined the effects of RTKis on three key downstream signaling molecules, Erk, Akt, and c-Myc (29). The effects of the RTKis on the phosphorylation of Erk and Akt appeared to be different in cells with different RTK activation patterns.
In cells that had only the EGFR family members activated, the EGFR inhibitor lapatinib was sufficient to inhibit the phosphorylation of both Erk and Akt, the protein level of c-Myc (Fig. 4A), and the growth of the cancer cells (Fig. 2B). In the EGFR and IR family dual-activated BEL7404 cells, lapatinib inhibited only the phosphorylation of Erk, while OSI inhibited only the phosphorylation of Akt, and neither of the inhibitors alone reduced the c-Myc level and the cancer cell growth (Figs. 4B and 2D). But the combination of lapatinib and OSI inhibited the phosphorylation of both Erk and Akt, decreased the protein level of c-Myc, and inhibited the growth of the cancer cells. In the EGFR/IR/MET family triple-activated cells, the combination of lapatinib, OSI, and JNJ inhibited the phosphorylation of both Erk and Akt, reduced the protein level of c-Myc (Fig. 4C and Supplementary Fig. S4A and S4B), and inhibited the cancer cell growth (Figs. 2H and 3C). A combination of two of the inhibitors had less inhibitory effects on Akt or Erk activation, c-Myc level, and the cell growth (Table 2; Fig. 4C and Supplementary Fig. S4A–S4C). These results indicated a good correlation between the dual inhibition of Erk and Akt, the c-Myc reduction, and the inhibition of cancer cell growth. C-Myc protein level appeared to be a potential marker to indicate the growth-inhibitory effects of the RTKis.
We then looked into additional cell lines that had different RTK phosphorylation patterns (Supplementary Fig. S4D–S4G and our unpublished observations for LS174T, FHCC98, BT474, T47D, EBC-1, H460, and PC3 cells). In all of the cases, the RTKis that led to the inactivation of Erk or Akt alone were insufficient to decrease the c-Myc level and stop the cell growth. Only the RTK inhibitions that inactivated both Erk and Akt could decrease the c-Myc level (Supplementary Fig. S4E and S4G) and block the cancer cell growth (Table 2). However, a combination of the MEK inhibitor U0126 and the Akt allosteric inhibitor MK2206 failed to dual inhibit the Erk and Akt activation and to decrease the c-Myc level (Supplementary Fig. S4H). It was also insufficient to stop the cancer cell growth (Table 2). Cotreatment of lapatinib with U0126 and MK2206 reduced both of the Erk and Akt activation and the c-Myc level (Supplementary Fig. S4I) and significantly inhibited the cancer cell growth (Supplementary Fig. S4J). These results suggested that an MEK-independent Erk signal downstream of the RTKs might be involved in supporting cancer cell growth, and a combinational inhibition of the RTKs was more effective than the inactivation of MEK and Akt to stop the cancer cell growth.
It is interesting to note that in cells that had more than one activated RTKs, a specific RTKi inhibited the phosphorylation of its target RTK but increased the phosphorylation of the remaining nontargeted RTKs. For example, lapatinib inhibited the phosphorylation of EGFR but increased the phosphorylation of InsR/IGF1R and Met, OSI inhibited the phosphorylation of InsR/IGF1R but increased the phosphorylation of EGFR and Met, while JNJ inhibited the phosphorylation of Met but increased the phosphorylation of EGFR and InsR/IGF1R (Fig. 4C and Supplementary Fig. S4A and S4B). More interestingly, the RTKis did not seem to induce the activation of new RTKs but only increased the phosphorylation of the existing activated RTKs to compensate the reduced phosphorylation of Erk or Akt to keep the protein level of c-Myc (Fig. 4D). A combination of the RTKis completely inactivated all of the RTKs and their downstream signaling molecules (Fig. 4C and Supplementary Fig. S4A and S4B) and stopped the growth of cancer cells (Figs. 2H and 3C). These data suggested that there might be a feedback mechanism among the activated RTKs to keep the c-Myc protein level, further supporting that a combination of RTKis targeting all of the activated RTKs is necessary to stop cancer cell growth.
Both cell-autonomous and environment-dependent mechanisms were involved in the activation of RTKs
To understand the regulation of the RTK activation in cancer cells, we serum starved the cells and analyzed the RTK activation in the absence of the serum-provided growth signals. Serum deprivation eliminated phosphorylation of all three RTKs in the human prostate cancer PC3 cells, did not change the RTK phosphorylation in the lung cancer H522 cells, while eliminated part of the RTK phosphorylation in the liver cancer SMMC7721 cells (Fig. 5A), suggesting that both cell-autonomous and environment-dependent mechanisms were involved in regulating the RTK activation in cancer cells.
To further investigate the environmental dependency of the RTK activations, we constructed several xenograft models by implanting cancer cells into different locations of the animals (Fig. 5B). The BEL7404 cells, which had the EGFR and IR family members activated in vitro, had EGFR, Her2, and Met activated when implanted subcutaneously; had only EGFR activated when implanted near the intestine by intraperitoneal injection; and had Her3 activated when implanted in liver (Fig. 5C). Similar phenomena were observed with other cancer cells (Fig. 5C, human lung cancer cell line H460, and our unpublished observations for LM3, SW1116, and HT29 cells). When the in vivo growing cancer cells were cultured back into in vitro conditions, the RTK activation pattern changed back (Fig. 5C, ex vivo). These results suggested, at least in our xenograft models, that the change of RTK activation profiles in vivo was due to environmental induction but not clonal selection. Similar phenomena were observed for the primary cancer cells. The RTK activations of the same primary colon cancer cells varied depending on their culturing environments (Fig. 5D and E), suggesting that RTKs in primary cancers were partially activated by environmentally supplied growth signals. However, the RTK activation is not simply determined by the environment, because different cancer cells showed differential RTK activation profiles when growing in the same environment (Fig. 5C, BEL7404 vs. H460; Fig. 5E, colo002 vs. colo004). These data suggested that intrinsic factors of cancer cells were also involved, and a complex interaction between the cancer cells and their growing environments determined the final activation patterns of the RTKs.
Combinations of multiple RTKis synergistically inhibited cancer cell growth in vivo
To test the effects of RTKi combinations on the cancer cell growth in vivo, we administrated the BEL7404 xenograft mice with different RTKis individually or in combinations. The in vitro and in vivo patterns of the RTK phosphorylation were different for the BEL7404 cells. The EGFR, Her2, and Met were activated in vivo, while the EGFR, InsR, and IGF1R were activated in vitro (Fig. 5C). A combination of EGFR family inhibitor lapatinib and Met inhibitor crizotinib significantly inhibited cancer cell growth in vivo (Fig. 6A and B), while a combination of lapatinib and IR family inhibitor OSI had little effect (Fig. 6C and D). None of the treatments had significant effects on the weight of the mice (Supplementary Fig. S5A and S5B), and no organ injuries were observed, suggesting that all of the treatments were relatively safe.
To confirm the function of activated RTKs in vivo, we repeated the in vivo experiment using the liver cancer SMMC7721 cells, which had EGFR, IGF1R, and Met activated in in vitro cultures, and was sensitive to the corresponding RTKis combination treatment (Fig. 5C; Table 2). However, EGFR, Her2, and Met were activated when the cells were growing in vivo (Fig. 5C). A combination treatment with lapatinib (targeting EGFR and Her2) and crizotinib (targeting Met) significantly inhibited the tumor growth in vivo (Fig. 6E and F). No significant decrease of the body weight was observed (Supplementary Fig. S5C). In contrast, the same combination of lapatinib and crizotinib had little effect on the cell growth in vitro (Supplementary Table S5).
In summary, all of these results suggested that a correct combination of the RTKis based on their in situ RTK activation pattern is an effective strategy to treat cancer.
Although targeted therapies are superior to cytotoxic chemotherapies in their side effects, the limited efficacies of the targeted drugs in the biomarker-stratified patients and their restricted range of beneficial populations remain as major clinical challenges (4, 5, 30, 31). Here, we report a RTK activation profile-based functional stratification strategy and an RTKi combinational therapy for personalized diagnosis and treatment of cancers. We demonstrated that multiple RTKs in different combinations were activated in cancer cells and functioned together to control their growth. Combinational RTKi treatments led to potent and selective growth inhibition of cancer cells in vitro and in vivo. Our study provided a novel personalized strategy for using the RTKis to treat cancer cells.
The RTKs are key cell growth regulators and the RTK-mediated cell growth signals may come from autocrine, paracrine, or environmentally supplied growth factors, or from genetic alterations that activate RTKs directly (32). We hypothesized that the RTK phosphorylation/activation profiles may be a rational stratification strategy for personalized cancer treatment. Indeed, all genetically altered and activated RTKs had prominent phosphorylation, such as the amplified MET in EBC-1 cells, HER2 in BT474 cells, and IGF1R in colo205 cells. Treatments of these cancer cells with corresponding single RTKis are consistent with the biomarker-guided therapies (33–35). However, these biomarker-positive cells contained additionally activated RTKs beside the genetically altered receptors, and a combinational treatment exerted better effects compared with the single RTKi treatment (our unpublished observations for EBC-1 and COLO205 cells). Our study provides possible explanations to some of the clinical observations. For example, MET deregulation is frequently observed in human cancers and it drives tumorigenesis in animal models (36, 37), but it is insufficient to use MET as a predictive marker to stratify patients for the targeted treatment (38). Our studies suggest that the growth of these cancer cells may depend on more RTKs in addition to the amplified MET.
More importantly, the RTK activation profile-based treatments are also effective in treating cancers without known genetic mutations in their RTK genes. These cancers represent majority of clinical cases, and they do not respond to the single-RTKi treatment. Our studies demonstrated that very few cancer cells relied on the activation of only one single RTK. Most of the cancer cells contained multiple RTK activations and depended on more than one of the activated RTKs for their growth. Combinations of the corresponding RTKis potently and selectively inhibited the cancer growth both in vitro and in vivo. Thus, the RTK activation profile-based combinational treatment strategy is particularly useful for treating the single RTKi–resistant cancers.
We also found that the RTK phosphorylation patterns were determined by both cell-autonomous and environment-dependent mechanisms. The cancer cells from the same tissue origin did not guarantee the same RTK activation pattern, while cancers from different tissue origins might share the same RTK activation pattern. Furthermore, the same cancer cells changed their RTK phosphorylation profiles when grown in different in vitro and in vivo environments. These data indicated that both intrinsic and extrinsic growth signals contributed to the RTK activations.
We also noticed that there appeared to be two types of RTKs to promote cancer cell growth, one mainly activating Erk signaling and the other activating Akt signaling. Inactivation of one type of RTK compensatively induced the activation of the other. Both Erk and Akt activation upregulates the protein level of c-Myc but with different mechanisms. Erk increases c-Myc mRNA level (39, 40), while Akt inhibits the c-Myc protein degradation (40, 41). Inactivation of both Erk and Akt is required to reduce the c-Myc protein level. The transcription factor c-Myc plays a central role in regulating cell proliferation and cell-fate decision, and its oncogenic function was well documented (42–44). Consistently, our data suggest that c-Myc is an effective indicator of success of an anticancer drug treatment. A proper combination of RTKis is necessary to inhibit both types of the RTKs to reduce the c-Myc protein level and to inhibit cancer cell growth.
RTKs have been reported to regulate both cell survival and growth (19, 45). However, our data indicated that inhibition of the activated RTKs by the combination of RTKis inhibited cancer cell proliferation but did not induce significant cell death. We noticed that most of the previous studies on the functions of RTKs in cancers were on the activated RTKs caused by gene mutations or amplifications. The RTKs that we were studying, however, were wild-type RTKs and were activated by their physiologic ligands. Our data suggest an intriguing possibility that the activities and functions of the mutated RTKs and the wild-type RTKs, although both are activated, may be distinct. The mutated RTKs may provide the cancer cells with survival as well as growth signals, while the growth factor–activated RTKs provide mainly cell proliferation signals. It would be interesting to further investigate whether the genetic mutation–activated RTKs transduce signals different from that of the growth factor–activated RTKs.
In summary, we have developed a personalized cancer diagnosis and treatment strategy, based on the RTK activation profile of cancer cells. A combination of the corresponding RTKis selectively and effectively inhibited the growth of the cancer cells in vitro and in vivo by arresting the cells in the G1 phase of the cell cycle. This combinational treatment is more effective than the single drug treatment and is particularly useful for treating the single RTKi–insensitive cancer cells. Our study also revealed a complex regulation of the RTK activations, which is mediated by both cell-autonomous and environment-dependent mechanisms. A better understanding of the regulation, function, and interaction of RTKs is essential for using combinations of RTKis to treat cancer patients.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: X. Sun, Q. Song, Q. Yu
Development of methodology: X. Sun
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Sun, Q. Song, L. He, L. Yan, J. Liu, Q. Zhang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Sun, Q. Song, Q. Yu
Writing, review, and/or revision of the manuscript: X. Sun, Q. Yu
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Sun, Q. Song, Q. Yu
Study supervision: Q. Yu
We thank Drs. Junying Yuan, Hongyang Wang, Lei Chen, and Yexiong Tan for providing liver cancer cell lines and primary cancer samples. We also thank Dr. Xun Huang for the generation of the mouse liver cancer xenografts.
This work was supported by the China Ministry of Science and Technology Key New Drug Creation and Manufacturing Program (no. 2013ZX09102015 and 2013ZX10002010-009 to Q. Yu), the National Natural Science Foundation of China (no. 81302792 to X. Sun; no. 81373447, 91413121, and 91213304 to Q. Yu), and the China National Key Basic Research Program (no. 2012CB910704 and 2013CB910904 to Q. Yu).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.