A Combinatorial Strategy for Tageting Papillary Thyroid Carcinoma with MEK Inhibitor and SHP2 Inhibitor

Background: Pharmacologic targeting of components of MAPK/ERK pathway in thyroid carcinoma is often limited due to the development of adaptive resistance. However, the detailed mechanism for MEK inhibitor (MEKi) resistance is not fully understood in papillary thyroid carcinoma (PTC). Methods: RNA-seq was performed in MEKi-resistant PTC cell lines (K1, BCPAP, TPC-1 and KTC-1) to investigate the intrinsic mechanism of drug resistance. Colony formation assay, cell viability assay, cell cycle analysis and various murine models, including xenograft model, long-term MEKi-treated model, and transgenic model were conducted to evaluate the treatment effect of combination therapy (SHP099 and selumetinib). Results: Multiple receptor tyrosine kinases (RTKs) signaling pathways as well as Src-homology 2 domain-containing phosphatase 2 (SHP2) were activated in MEKi-resistant cells. Given the physiological role of SHP2 as the downstream of many RTKs, we rst found that blockage of SHP2 abrogated MEKi resistance in thyroid cancer. Interestingly, we also found MEKi in combination with SHP2 inhibitor remarkably suppress rebound of MEK/ERK pathway compared to that of MEKi treatment alone, which signicantly improved antitumor effects of MEKi. Various murine models conrmed the synergistic suppression on PTC in mice treated with both inhibitors. Conclusion: SHP2 blockade by SHP099 in combination with selumetinib is a promising therapeutic approach for advanced thyroid cancer. were 2-8 and 10 respectively; *P<0.05, **P<0.01, compared with Ctrl, @P<0.05, @@P<0.01, @@@P<0.001, compared with Comb, ANOVA-test. All data were repeated at least three times independently and represented as mean±SD. This showed that the SHP099/selumetinib combination group exhibited a more signicant effect in inhibiting the proliferation in the long-term MEKi-resistant model than other groups. (F) The SHP099/selumetinib combination signicantly induced the G1 phase arrest in the BCPAP-R and TPC-1-R cell lines. The cell cycle distribution was assessed by ow cytometry. The percentage of cells in the G1, S and G2/M phase was plotted. experiments. Compared with single agent, the SHP099/selumetinib combination group exhibited a more signicant effect in inhibiting the progression in four PTC cell lines. SHP099/selumetinib combination signicantly induced the G1 phase arrest PTC cell cell cycle distribution was assessed by ow cytometry. revealed that, knocking down SHP2 expression similarly exhibited the similar as the SHP099 treatment after


Western blot and RTK arrays
Cells were lysed in modi ed RIPA buffer containing 1% PMSF. Equal amounts of total protein were resolved by SDS-PAGE, and transferred onto PVDF membranes (Millipore). Then, these membranes were immunoblotted overnight with the primary antibodies. The primary antibodies used for the western blot were p-MEK, MEK, p-ERK (Cell Signaling Technology, 4370), ERK (Cell Signaling Technology, 4695), SHP2 (ABCAM, ab32083), p-SHP2 (ABCAM, ab62322), and GAPDH (Cell Signaling Technology, 5174). The human phospho-RTK arrays were purchased from R&D Systems (ARY001B), and were used according to manufacturer's guidelines.

Animal experiments
All animal experiments were approved by the Tianjin Medical University Cancer Institute and the Hospital Animal Care and Use Committee, and these were performed according to the IACUC protocol. The thyroid cancer cell line xenografts were established through the subcutaneous injection of 1×10 5

cells into 4-
week-old male NSG mice. When the tumor volumes reached approximately 15 mm 3 , these animals were randomly assigned into four groups: rst group (Ctrl), mice were orally treated with DMSO, q.d; second group (SHP099), mice were orally treated with 50 mg/kg of SHP099, q.o.d; third group (Sel), mice received daily oral doses of selumetinib, 20 mg/kg; last group (Comb), mice were treated with the combination therapy. In order to establish the drug resistance model in vivo, ten mice with similar-volume PTC xenografts received oral doses of selumetinib, at 20 mg/kg, once daily. After 15 or 40 days, representative xenografts in two groups were picked for the human phospho-RTK arrays. Remaining xenograft from 40 days treated mice was cut into equal sections, and planted into other mice. After seven days, planted animals were randomly assigned into four groups, and treated as described above. At the indicated time points, these animals were sacri ced, and the tumors were excised for further analysis. A transgenic mouse model of spontaneous PTC was established as previous described. In our conditions, the mice spontaneously developed PTC at 6-12 weeks of age. According to weight, six-week-old TPO-Cre BrafCA mice were randomly assigned into four groups: rst group (Ctrl), mice were orally treated with DMSO, q.d; second group (SHP099), mice were orally treated with 50 mg/kg of SHP099, q.o.d; third group (Sel), mice received daily oral doses of selumetinib, 20 mg/kg; last group (Comb), mice were treated with the combination therapy.

Immunohistochemistry (IHC)
IHC was performed according to standard protocols. The primary antibodies used for the IHC assays were p-ERK (Cell Signaling Technology, 4370) and Ki67 (Cell Signaling Technology, 9027). The stained slides were independently examined by two pathologists, who were blinded to the treatment information.
Hematoxylin and eosin staining were performed by the Department of Pathology of Tianjin Medical University Cancer Institute and Hospital.

RT-PCR
The RT-PCR assays were performed as previously described [17] . The primers were listed as Table 1.

Results
SHP2 is a potential therapeutic target for selumetinib resistant PTC cell lines.
In order to verify the intrinsic mechanism for MEKi-resistance, the resistant models were constructed in vitro with BCPAP and TPC-1. Cells were cultured in increasing concentrations of selumetinib until drugresistant clones were more than 10 times less sensitive to selumetinib than their parental lines, termed BCPAP-R and TPC-1-R, respectively ( Supplementary Fig. S1a). Colony formation assay further con rmed that the MEKi-resistant cell lines exhibited a signi cant resistance (P<0.05) than the primary cell lines, after treatment with selumetinib ( Supplementary Fig. S1b). Then we performed RNA-sequencing analysis to BCPAP (N=4) and BCPAP-R cells (N=4), RNA-seq identi ed 4,270 mRNAs with signi cant change in BCPAP-R relative to BCPAP (Supplementary Fig. S2a and S2b). Based on the total identi ed mRNAs, GSEA analysis revealed that mRNA of RTKs (including VEGFA, VEGF, IGF1 and EGFR) signaling signi cantly increased in BCPAP-R ( Fig. 1A and 1B). These above results indicated that RTK-induced ERK reaction played the main role in acquired MEKi-resistance. SHP2 is a positive RTK downstream signal transducer, which is recently regarded as a novel target for RTKs-driven cancers [18] . Expression of p-SHP2 in both primary and MEKi-resistant cell lines was determined by Western blot. Compared with parental cell lines respectively, p-SHP2 expression was signi cantly upregulated in BCPAP-R and TPC-1-R. At the same time, p-MEK and p-ERK expression were upregulated continuously, not as the negative feedback inhibition in primary cell lines (Fig. 1C). In order to verify whether SHP2 contributes to MEKi-resistance, two MEKi-resistant cell lines (BCPAP-R and TPC-1-R) were treated with DMSO (Ctrl), SHP099, selumetinib (Sel), or the SHP099/selumetinib combination (Comb), respectively. All examined MEKi-resistant cell lines demonstrated susceptibility to SHP099 with suppressed cell colony formation and viability ( Fig. 1D and   1E). Cell cycle analysis revealed that SHP099 arrested MEKi-resistant cell lines at the G1/S phase (Fig.  1F). The co-administration with selumetinib exhibited an increase in e cacy, with additive synergy. SHP099 resensitized BCPAP-R and TPC-1-R to selumetinib, and the combination strategy suppressed the cell colony formation, growth, and cell cycle progression more effectively than either SHP099 or selumetinib alone (Fig. 1D, 1E and 1F). These data indicated that the SHP2-transduced RTKs activation was a dominant mechanism for the MEKi-resistant phenotype, which was synergistically reversed by SHP2 inhibitor SHP099.

The combination strategy suppresses the long-term MEKitreated tumor growth in vivo
In order to test the effectiveness of the combined strategy to the long-term MEKi-treated murine model in vivo, mice with K1 xenografts were established and treated as the ow chart indicated ( Fig. 2A). Ten mice with similar-volume PTC xenografts were divided into two groups and received oral doses of selumetinib (20 mg/kg, q.d) for 15 or 40 days, respectively. During the 40 days selumetinib treatment, the tumor growth rate was suppressed in approximately 25 days, and this rapidly climbed in approximately 25-40 days ( Supplementary Fig. S3). Representative xenografts treated with selumetinib for 15 or 40 days were selected for the phospho-RTK array, which revealed that multiple RTKs were consistently activated after 40 days MEKi-treatment (Fig. 2B). In order to analyze the e cacy of the combined strategy to RTKactivated xenografts, the 40 days MEKi-treated xenograft remains were cut into equal sections, and planted into other mice. After seven days, planted mice were randomly assigned into four groups, and treated with DMSO (Ctrl), SHP099, selumetinib (Sel) and both drugs combination (Comb) for 15 days, respectively. Selumetinib had minimal effects on RTK-activated tumors, demonstrating that the tumor models had partial resistance to MEKi. The upregulated susceptibility to SHP099 in long-term MEKitreated tumors indicated that activated RTKs played a major role in PTC cells proliferation. Of note, the SHP099/selumetinib combination caused tumor shrinkage more effectively than the single agent (Fig.  2C, 2D and 2E), demonstrating that developed MEKi-resistance depending on SHP2. As revealed by the IHC assay (Fig. 2F), the SHP099/selumetinib combination treatment group had a lower p-ERK expression, and fewer Ki67-positive cells. Stable weight, normal behavior and appeared health were showed in MEKiresistant models after combination treatment ( Supplementary Fig. S4). These data suggested that the combined SHP2/MEK inhibition continued to be effective for suppressing tumor growth in the long-term MEKi-treated models in vivo.

Suppression of SHP2 abrogates the RTKs-induced MEK/ERK pathway rebound
After treatment with MAPK inhibitor, various cancer models can develop ERK signaling rebound induced by RTKs [19][20][21][22] . In order to examine whether the ERK rebound existed in PTCs after treatment with MEKi, the p-ERK expression was detected after 0, 1, 24 and 48 hours of selumetinib treatment. Western blot revealed that the p-ERK expression initially declined after one hour, and rapidly rebounded to the baseline or to an even higher level at 48 hours (Fig. 3A). In order to verify the relationship between ERK signaling rebound and activated RTKs in PTCs, a human phospho-RTK array was performed to determine the RTK activation status. As expected, the phosphorylation levels of several RTKs were upregulated in K1 after MEKi-treatment (Fig. 3B). Activated by RTKs, p-SHP2 expression signi cantly increased (P<0.001) in all four cell lines treated with selumetinib (Fig. 3C). Hence, it was hypothesized that therapy that co-targets MEK and SHP2 could effectively abolish the ERK signaling reactivation. For further con rmation, four PTC cell lines were treated with DMSO, SHP099 alone, selumetinib alone, or the SHP099/selumetinib combination, and the change in p-ERK expression within 48 hours was detected. Compared with SHP099 or selumetinib alone, the combination of SHP099 and selumetinib persistently inhibited the p-ERK expression (Fig. 3D). As measured by ETV1, 4, 5, and FOSL1 mRNA levels, ERK dependent transcription was also signi cantly inhibited (P<0.01) by combination strategy (Fig. 3E). In addition, the combination of MEKi and SHP2 knockdown had similar effects to the SHP099/MEKi treatment, indicating that SHP099 is "on-target" in PTC cell lines (Fig. 3F, 3G and 3H). Taken together, these data demonstrated that the combination of SHP099 and MEKi could more effectively suppress the p-ERK expression than SHP099 or selumetinib alone, and that the dual SHP2/MEK inhibition caused these observed effects.

SHP2 inhibition combined with selumetinib inhibits cell progression in PTC in vitro
In order to determine whether the combined SHP2/MEK inhibition could be more effective in PTC cell lines, colony formation and viability (CCK8) assays were further performed ( Fig. 4A and 4B). As a single agent, SHP099 had a variable effect on colony formation in PTC cell lines with different genetic backgrounds. A minimal effect was detected in three BRAF mutation cell lines (which was consistent with previous studies), while a remarkable effect (P<0.01) was observed in the BRAF wild-type harboring the RET fusion mutation. However, compared with SHP099 or selumetinib treatment alone, few or no detectable colonies in all PTCs were found after SHP099/selumetinib combination treatment (P<0.01). Similar effects were observed in cell viability assays. The combination resulted in intense growth inhibition (P<0.01) in all PTCs. Given the SHP099/selumetinib combination effect on PTC cell lines, it was determined whether the combination treatment suppressed the cell proliferation by regulating cell cycle progression (Fig. 4C). The SHP099/selumetinib combination arrested PTC cell lines at the G1/S phase. Same as the previous study to indicate that SHP099 is "on-target" in PTC cell lines, SHP2 knockdown had similar effects to the SHP099/MEKi treatment ( Fig. 4E and 4F). These data suggested that the combined SHP2/MEK inhibition could inhibit PTC cell proliferation powerfully in vitro.

The combination of SHP099 and selumetinib is a promising therapeutic approach in vivo
To con rm our results in models in vivo, mice carrying K1 or TPC-1 xenografts were utilized to examine the antitumor activity of MEK/SHP2 inhibition. Compared with mice treated with SHP099 or selumetinib treatment alone, mice in the combination group markedly reduced tumor volume and weight (Fig. 5A, 5B and 5C), and more effectively suppressed p-ERK level (Fig. 5D). Xenografts isolated from combinationtreated mice had fewer proliferating cells than those derived from other mice, as measured by Ki67 staining (Fig. 5D). The single-agent effects were similar to the above results in vitro. After treatment with combination strategy, stable weight, appeared healthy and normal behavior and were showed in all mice carrying K1 or TPC-1 xenografts (Supplementary Fig. S5a and S5b).
We further con rmed our results in a transgenic murine model of spontaneous thyroid cancer (Fig. 6A).
Mice treated with SHP099 or selumetinib treatment alone had reduced tumor volume to about 40-50% of the initial volume, though the combination of SHP099 and selumetinib produced a greater reduction in tumor volume and weight than either therapy alone ( Fig. 6B and 6C). IHC assay revealed that the SHP099/selumetinib combination treatment group had a lower p-ERK expression, and fewer Ki67-positive cells (Fig. 6D). These data suggested that the combined SHP2/MEK inhibition could serve as a powerful therapeutic approach in PTCs, in which single-agent targeted therapeutics usually have a limited effect.

Discussion
In addition to RTKs reactivation, tumors evade the long-term MEK blockage therapies via various resistance mechanisms [23] , such as the BRAF gene ampli cation in colorectal cancers harboring BRAF V600E [24] , the increased formation of Raf-1/B-Raf dimers in melanoma cells [25] , and the enhanced activation of the PI3K/AKT pathway in melanoma cells [26] . Upregulated multiple RTKs signaling was revealed by RNA-seq in MEKi-resistant model in vitro and identi ed by RTK assay in long-term MEKitreated models in vivo. Activated by RTKs, SHP2 interacts with Ras, participating in signal transducer and activator of MAPK pathway to promote tumor progression [27] . A higher SHP2 activity was found in all resistant clones, which indicated that SHP2 can be activated by RTKs, and plays a major role in MEKiresistance. The combination of selumetinib and SHP099 signi cantly blocks the tumor growth and ERK activation in all MEKi-resistant models, demonstrating that PTCs with either BRAF V600E mutation or RET activation acquire MEKi-resistance through activated RTKs, and this mainly depended on SHP2. Hence, the combination strategy may be a promising treatment approach for MEKi-resistant PTC. A schematic model summarizing the mechanism of the SHP099/selumetinib combination strategy in PTC was presented in (Supplementary Fig. S6).
Although a synergy has been widely observed in various tumor types, the combination of SHP099 and selumetinib has been reported to be less effective for certain tumors. RAS is the direct downstream signal for SHP2, and the GTPase activity of different RAS mutants positively correlated with the sensitivity to SHP099. Certain RAS mutations (G13D and Q61X) have the lowest intrinsic GTPase activity than other mutations, thereby emerging the resistance to this combination [28][29][30] . Most RTKs reactivate the RAS/MEK/ERK pathway through SHP2 to form adaptive resistance against MAPK signaling inhibitors, while remaining RTKs, including FGFR, could transduce signals in a SHP2-independent fashion [22] .
Depend on FGFR activation, a subset of ATC cells harboring BRAF V600E has been proven to be insensitive to the combination of SHP099 and MAPK inhibitor [22] . However, although derived from thyroid follicular cells as well, our study demonstrated SHP2/MEK inhibition effectively suppressed tumor progression in all tested PTC models both in vivo and in vitro. In addition, RTK assay also revealed that MEKi remarkably induced SHP2-dependent RTKs activation in PTC models. Hence, distinct with ATC, the combination of SHP099 and selumetinib is a promising therapeutic approach for PTC.
RET is an RTK involved in the development of PTC and medullary thyroid carcinomas (MTC) [31,32] . RET/PTC fusions were presented in PTC, and these can reach 50-70% in pediatric patients, and in cases that have experienced radioiodine exposure [33] . In contrast with the PTC cell line harboring BRAF V600E mutation, high SHP2 activity and SHP099 sensitivity were found in TPC-1, a PTC cell line harboring RET Gain-of-function fusion. Resemble trend also occurred in TT ( Supplementary Fig. S7a and S7b), a MTC cell line harboring constitutive-activation RET C634W mutation [34] . Above data suggested that SHP099 signi cantly suppresses RET activation-driven PTC progression, though SHP2/MEK blockage produced a greater inhibition in tumor development.
The present study has several limitations. First, the effect of the combination of SHP099 and selumetinib in PTC cell lines harboring RAS mutations was not evaluated. This was limited by the low mutation rate of RAS in PTCs [35] , and the well-known relationship between RAS mutation and the SHP099 effect [30,36,37] . Second, the impact of the combination strategy on the microenvironment of PTC could not be evaluated. Both SHP099 and selumetinib have been reported to be correlated with various immune cell activations [31,38,39] . Our study revealed that SHP099 can effectively suppressed the tumor growth in the immunocompetent transgenic mouse model of spontaneous PTC, which was consistent with the above studies. Hence, further studies are needed to determine the synergy of SHP099 and selumetinib in shaping the microenvironment of PTCs.

Conclusions
In conclusion, the combination of SHP099 and selumetinib abolished the RTK-mediated ERK rebound, resulting in growth inhibition. Compared with the use of SHP099 or selumetinib alone, the combination strategy overcame the intrinsic resistance in the long-term and extended the application range in MAPKactivated PTCs more effectively. The combination of SHP2 inhibitor SHP099 and selumetinib may be a promising and meaningful therapeutic strategy against PTC. (D) and colony formation (E) assays were assessed at 2-8 and 10 days, respectively; *P<0.05, **P<0.01, ***P<0.001, compared with Ctrl, @P<0.05, @@P<0.01, @@@P<0.001, compared with Comb, ANOVA-test.
All data were repeated at least three times independently and represented as mean±SD. This showed that the SHP099/selumetinib combination group exhibited a more signi cant effect in inhibiting the proliferation in the long-term MEKi-resistant model than other groups. (F) The SHP099/selumetinib combination signi cantly induced the G1 phase arrest in the BCPAP-R and TPC-1-R cell lines. The cell cycle distribution was assessed by ow cytometry. The percentage of cells in the G1, S and G2/M phase was plotted.    The combination of SHP099 and selumetinib was a promising therapeutic approach for advanced PTC in vivo. (A) Images of tumors dissected from NSG mice injected with the K1 and TPC-1 cells, respectively.