Distinct CDK6 complexes determine tumor cell response to CDK4/6 inhibitors and degraders

Wu, Xuewei; Yang, Xiaobao; Xiong, Yan; Li, Ruitong; Ito, Takahiro; Ahmed, Tamer A.; Karoulia, Zoi; Adamopoulos, Christos; Wang, Hong; Wang, Li; Xie, Ling; Liu, Jing; Ueberheide, Beatrix; Aaronson, Stuart A.; Chen, Xian; Buchanan, Sean G.; Sellers, William R.; Jin, Jian; Poulikakos, Poulikos I.

doi:10.1038/s43018-021-00174-z

Article
Published: 01 March 2021

Distinct CDK6 complexes determine tumor cell response to CDK4/6 inhibitors and degraders

Nature Cancer volume 2, pages 429–443 (2021)Cite this article

7443 Accesses
25 Citations
32 Altmetric
Metrics details

Subjects

Abstract

Cyclin-dependent kinases (CDKs) 4 and 6 inhibitors (CDK4/6is) are effective in metastatic breast cancer, but they have been only modestly effective in most other tumor types. Here we show that tumors expressing low CDK6 rely on CDK4 function and are exquisitely sensitive to CDK4/6is. In contrast, tumor cells expressing both CDK4 and CDK6 have increased reliance on CDK6 to ensure cell cycle progression. We discovered that CDK4/6is and CDK4/6 degraders potently bind and inhibit CDK6 selectively in tumors in which CDK6 is highly thermo-unstable and strongly associated with the HSP90–CDC37 complex. In contrast, CDK4/6is and CDK4/6 degraders are ineffective in antagonizing tumor cells expressing thermostable CDK6, due to their weaker binding to CDK6 in these cells. Thus, we uncover a general mechanism of intrinsic resistance to CDK4/6is and CDK4/6i-derived degraders and the need for new inhibitors targeting the CDK4/6i-resistant, thermostable form of CDK6 for application as cancer therapeutics.

You have full access to this article via your institution.

Download PDF

Clinical CDK4/6 inhibitors induce selective and immediate dissociation of p21 from cyclin D-CDK4 to inhibit CDK2

Article Open access 07 June 2021

Targeting CDK4 and CDK6 in cancer

Article 18 March 2022

Discovery and resistance mechanism of a selective CDK12 degrader

Article 22 March 2021

Main

CDK4/6 regulate cell cycle progression by phosphorylating and inactivating the tumor suppressor retinoblastoma protein (RB) and thus have been targeted by small-molecule inhibitors for cancer therapy^1,2. Dissociation of hyper-phosphorylated RB alleviates transcriptional repression of E2F promoters and allows initiation of DNA synthesis- and mitosis-related gene transcription^2,3. Recently, CDK4/6is in combination with hormonal therapy showed significant clinical activity in RB-proficient metastatic ER⁺ breast cancers^4,5 and three CDK4/6is, palbociclib, abemaciclib and ribociclib, are now US Food and Drug Administration (FDA)-approved for this indication^4,6,7,8. As the activity of CDK4/6 requires a functional RB protein, tumors that do not express functional RB are resistant to these drugs⁹. However, in many tumor types predominantly expressing wild-type (WT) RB1 (lung adenocarcinomas, melanomas, colon cancers and others) preclinical and clinical studies have shown only modest effectiveness of CDK4/6is^10,11,12, suggesting that other mechanisms limit their efficacy in these tumor types. In ER⁺ breast cancer, CDK6 amplification has been reported to confer acquired resistance to CDK4/6is¹³. However, CDK4/6is are potent inhibitors of CDK6 in vitro and they show effectiveness in some, mostly liquid, tumor types predominantly expressing CDK6 (ref. ¹⁴). In solid tumors a comprehensive analysis of CDK6 expression at baseline and its potential role in tumor response to CDK4/6is has been lacking. Thus, there is a need for better understanding of the role of CDK6 as a potential contributor to tumor resistance to CDK4/6is.

Similarly to most kinases, CDK4 and, to a lesser extent, CDK6 has been previously shown to require the HSP90–CDC37 chaperoning complex for their maturation^15,16,17. Closely related kinases may show largely different dependencies on HSP90–CDC37, for example, CDK4 is highly dependent on HSP90, but the closely related CDK2 is almost completely independent¹⁸. Further, activating mutations have been found to increase the dependency of kinases^19,20, but overall, dependence of a given kinase on HSP90–CDC37 is thought to be uniform across cells and to depend exclusively on its amino-acid sequence. Finally, small-molecule inhibitors have been shown to compete with HSP90–CDC37 for kinase binding¹⁸ and there is evidence from primarily in vitro studies for a connection between the strength of the interaction of CDKs with the HSP90–CDC37 complex and their affinity for inhibitors²¹. However, whether the kinase interaction with the HSP90–CDC37 complex affects tumor response to small-molecule inhibitors remains unknown.

Lately, proteolysis-targeted chimeras (PROTACs), which are hetero-bifunctional small molecules that are aimed at achieving selective degradation of the target proteins have been developed against a number of target kinases, including CDK4/6 (refs. ^22,23,24). We developed a potent and selective CDK4/6-directed PROTAC (CDK4/6 degrader) and we used it as tool to elucidate mechanisms of CDK4/6 regulation and response to CDK4/6is, by monitoring target degradation by CDK4/6 degrader as a surrogate for compound binding to target in cells. Use of this approach provided us with evidence of a critical role of the expression state of CDK6 in affecting tumor response to CDK4/6is.

Results

Intrinsic resistance to CDK4/6is is associated with incomplete inhibition of RB–E2F and expression of CDK6

To gain insight into mechanisms of intrinsic resistance or sensitivity to CDK4/6is, we assessed the concentration-dependent effects of palbociclib on the growth of cancer cell lines derived from a variety of RB-proficient tumor types. We observed large variations in cell line response to CDK4/6is, consistent with previous reports^11,12,25. In the ‘CDK4/6i-sensitive’ (CDK4/6i-S) group, palbociclib concentrations <500 nM and in many cases as low as 80 nM were sufficient for substantial (>90%) growth inhibition. In contrast, in the ‘CDK4/6i-resistant’ (CDK4/6i-R) group, palbociclib concentrations even as high as 2 μM had only modest effects on cell growth (Fig. 1a and Extended Data Fig. 1a). The CDK4/6i-S group included ER⁺ breast cancer cell lines, as expected²⁶, as well as cancer cell lines from several other tumor types. However, most RB-proficient cell lines derived from common solid tumor types, including nonsmall cell lung carcinoma (NSCLC), melanoma and colorectal carcinoma showed a CDK4/6i-R phenotype (Fig. 1a and Extended Data Fig. 1a).

We next assessed whether the observed differences of tumor cell sensitivity to palbociclib were associated with the extent of suppression of RB–E2F output by the drug. Indeed, suppression of both RB phosphorylation and of downstream E2F target proteins, PLK1, cyclin A and DHFR upon palbociclib treatment (RB–E2F output) were consistently of greater magnitude in CDK4/6i-S compared to CDK4/6i-R tumor cells (Fig. 1b and Extended Data Fig. 1b,c). Other US FDA-approved CDK4/6is, including ribociclib⁸ and abemaciclib⁷, showed a similar profile (Fig. 1c,d). When we examined expression of several known CDK4/6 signaling-related proteins, including members of the cyclin D and cyclin E, INK4 and CIP/KIP families, we did not detect any consistent difference in expression between the two groups (Fig. 1e). However, we noticed that most CDK4/6i-S cell lines expressed CDK6 at very low levels, whereas the CDK4/6i-R cell lines in our panel, expressed both CDK4 and CDK6 (Fig. 1e).

Low expression of CDK6 predicts for sensitivity to CDK4/6is in NSCLC

The observation that low CDK6 expression correlated with sensitivity to CDK4/6is, prompted us to assess whether we could identify subsets of CDK4/6i-S cell lines within larger, predominantly RB-proficient, tumor types. We focused on NSCLC, in which previous studies have shown clinical activity of these drugs, but failed to show a significant increase in either progression-free survival (PFS) or overall survival (OS), over current standard of care^27,28. In addition to the NSCLC lines that we examined initially (Extended Data Fig. 1a), we assessed five additional NSCLC lines with a high score for CDK4 dependence (H1792, H2087, H2291, HCC827 and H1915, in addition to H358) based on Achilles RNAi DEMETER scores²⁹. For comparison, we examined two additional NSCLC CDK6-dependent lines (PC9 and H1666, in addition to A549, Calu6). We first confirmed dependence of this NSCLC panel on either CDK4 or CDK6. In CDK4-dependent cells, short interfering (si)RNA-mediated knockdown of CDK4, but not CDK6 suppressed RB–E2F output, whereas downregulation of CDK6, but not of CDK4 suppressed pRB–E2F output in CDK6-dependent cells (Fig. 2a). Further, although expression of CDK4 was variable across our panel of cell lines, CDK4-dependent cells expressed low levels of CDK6 compared to CDK4 (Fig. 2b).

We next assessed whether response to CDK4/6is correlates with low expression of CDK6 in NSCLC tumor cells. In all cases, we observed substantially higher sensitivity of CDK6-low cells to CDK4/6is, both in terms of RB–E2F output (Fig. 2c and Extended Data Fig. 2a) and of cell growth (Fig. 2d) compared to CDK4/6i-R cells with high levels of CDK6 (H358 and A549, Calu6 data are included in Fig. 1a,b and Extended Data Fig. 1a, and a list of the NSCLC cell lines used in our study is in Extended Data Fig. 2b).

To assess the clinical relevance of our observations, we analyzed RNA expression data from tumors from the phase III JUNIPER trial (NCT02152631), in which abemaciclib was evaluated in patients with KRAS-mutated, advanced NSCLC²⁷. In a retrospective assessment, tumors with low CDK6 expression correlated with significantly longer PFS (Fig. 2e) and OS (Fig. 2f) compared to tumors expressing high levels of CDK6. These data confirm in a large clinical dataset, a critical role of CDK6 expression in the tumor in determining outcome of patients with cancer treated with CDK4/6is and suggest that CDK6 expression in the tumor might be used a biomarker to stratify patients with NSCLC for CDK4/6i treatment.

Tumors expressing both CDK4 and CDK6 depend selectively on CDK6

We next focused on understanding the basis of resistance of tumor cells expressing both CDK4 and CDK6 to CDK4/6is. Overexpression of a target is a mechanism that can drive resistance to small-molecule inhibitors. We thus assessed whether short hairpin (sh)RNA-mediated knockdown of CDK4 or CDK6 would sensitize palbociclib-unresponsive cells to CDK4/6is. As expected, in cell lines expressing predominantly CDK4 and low levels of CDK6, knockdown of CDK4, but not of CDK6, reduced RB–E2F output (Extended Data Fig. 3a). In CDK4/6i-R tumor cells, inducible CDK6 knockdown suppressed RB–E2F output (Fig. 3a), as well as cell growth (Fig. 3b), which were further suppressed by treatment with palbociclib. Notably, knockdown of CDK4 in these same cells had no effect on either phosporylated RB (pRB) levels or sensitivity to palbociclib (Fig. 3a,b). CDK6 (but not CDK4) knockdown also suppressed RB–E2F output as well as growth in additional CDK4/6i-R cell lines (Fig. 3c). Moreover, ectopic expression of an shRNA-resistant mutant of CDK6 rescued cell growth and RB–E2F output inhibition promoted by CDK6 shRNA-mediated knockdown (Fig. 3d), thus establishing the specific role of CDK6 in driving RB–E2F signaling and cell growth in these cells. Finally, phosphorylation of CDK4-T172, an established marker of CDK4 activation³⁰, was readily detected in the CDK4/6i-S cells, but was substantially lower in the CDK4/6i-R cells (Fig. 1e). All of these findings suggested that despite expression of both CDK4 and CDK6, RB–E2F signaling was driven selectively by CDK6 in CDK4/6i-R tumor cells.

**Fig. 3: Tumor cells expressing both CDK4 and CDK6 depend selectively on CDK6.**

To assess the generality of our observation, we analyzed the Avana dependency map (DepMap) dataset to look at the correlation between the expression and protein levels to the dependency profiles of CDK4 and CDK6^{31,32,33,34,35,36}. First, similar to the cell lines used in our study, CDK4 expression was relatively uniform across cancer cell lines, whereas CDK6 expression was highly variable. Further, CDK6 expression and protein levels were correlated with dependence on CDK6, however CDK4 protein or mRNA expression levels were not predictive of dependence on CDK4 (Fig. 3e and Extended Data Fig. 3b). The data were similar whether dependency was interrogated using shRNA or single guide (sg)RNA-mediated loss of function³¹. Thus, the analysis revealed that our observation applies across tumor lines of various cancer types.

Development of MS140, a potent and selective CDK4/6 degrader (PROTAC)

The inactivity of CDK4/6is in cells expressing both CDK4 and CDK6, led us to postulate that targeted degradation²² of CDK4/6 might result in more potent inhibition of RB–E2F output and growth suppression of CDK4/6i-R tumor cells. As an approach, we developed hetero-bifunctional small molecules³⁷ that would both inhibit CDK4/6 kinase activity as well as target CDK4/6 proteins for degradation. On the basis of the crystal structure of CDK6 in complex with palbociclib (Protein Database code 5L2I)³⁸, we used solvent-exposed piperazine as the linker attachment point. We synthesized hetero-bifunctional potential CDK4/6-directed PROTACs by linking palbociclib to pomalidomide, a moiety with high affinity for the E3 ligase cereblon (CRBN), a component of a cullin-RING ubiquitin ligase complex^{39,40,41,42,43}, using linkers of various lengths and types. By screening and optimizing synthesized compounds for their ability to degrade CDK4/6 and inhibit RB–E2F output in cells, we identified MS140 (Fig. 4a), as a highly potent CDK4/6 kinase inhibitor in vitro (Extended Data Fig. 4a) that markedly suppressed RB–E2F signaling and reduced CDK4 and CDK6 protein levels in a concentration and time-dependent manner (Fig. 4b,c). Degradation of CDK4/6 proteins by MS140 was specific, as shown by its abrogation upon pretreatment with excess palbociclib or pomalidomide (Fig. 4d and Extended Data Fig. 4b). We also confirmed that CDK4/6 protein degradation by MS140 was specifically mediated by the proteasome by its abrogation upon pretreatment with the proteasome inhibitor bortezomib or with the Nedd8-activating enzyme inhibitor MLN4924 (Fig. 4d and Extended Data Fig. 4b). Further, MS140 failed to degrade CDK4 in cells in which CRBN had been knocked out using CRISPR/Cas9 technology (Fig. 4e), confirming that target degradation by MS140 was specifically mediated by CRBN. As a negative control, we designed and synthesized MS140-ve (Extended Data Fig. 4c), a methyl analog of MS140 that is predicted not to bind CRBN^42,44. Treatment with MS140-ve did not reduce CDK4/6 protein levels (Fig. 4f), further confirming that MS140 degrades its targets by linking them to the ubiquitin–proteasome machinery. Although palbociclib has been reported to bind a number of kinases in addition to CDK4/6, including CDK9 (ref. ⁴⁵), we observed that MS140 did not degrade CDK7 or CDK9 (Fig. 4b,f). Finally, global analysis of protein degradation using mass spectrometry (MS) revealed an impressively selective protein downregulation profile of MS140, with very few hits other than CDK4 and CDK6 (Fig. 4g). Thus, MS140 potently and selectively both inhibited and degraded CDK4/6 kinases in CDK4/6i-S tumor cells by targeting them to the CRL4–CRBN–E3 ubiquitin complex.

CDK4/6-directed degradation is more effective than CDK4/6is in CDK4/6i-S tumor cells

Treatment of various RB-proficient tumor cell lines with MS140 resulted in 3–30-fold greater suppression of both RB–E2F output and cell growth, compared to palbociclib, in the CDK4/6i-S cells identified in our first screen (Fig. 5a–d). A notable exception was H358, in which MS140 only modestly decreased CDK4 levels or inhibited RB signaling as compared to palbociclib (Fig. 5e). However, this cell line exhibited very low endogenous levels of CRBN (Fig. 1e), consistent with the increased effectiveness of MS140 being CRBN-dependent. We next assayed additional cell lines of hematopoietic origin. All mantle cell lymphoma (MCL) lines tested showed sensitivity to palbociclib and higher sensitivity to MS140 (Fig. 5f), associated with more potent and durable inhibition of RB–E2F output by MS140 (Fig. 5g and Extended Data Fig. 5a), consistent with these cells being predominantly CDK4-driven and expressing low levels of CDK6.

Of note, a smaller group of cancer cell lines expressing high levels of CDK6 compared to CDK4 are sensitive to CDK4/6is and highly sensitive to MS140 (Fig. 5h,i). These are mostly cells from hematopoietic origin, but also include the colorectal cancer cell line Colo205. These tumor cells dependent on CDK6 (refs. ^12,29) (Extended Data Fig. 5b–d), indicating that in this subgroup of tumor cells, CDK6 is sensitive to CDK4/6is. Our data on growth response of hematopoietic tumor lines to palbociclib and MS140 are summarized in Extended Data Fig. 5e. Thus, MS140 was more effective than palbociclib in inhibiting RB–E2F output and cell growth of many CDK4/6i-S tumor cell lines. CDK4/6i-S cells are predominantly those expressing CDK4 and low CDK6, but we also identified a smaller subset of CDK6-dependent tumor cells that are also sensitive to CDK4/6is and more sensitive to the CDK4/6 degrader.

The increased effectiveness of MS140, as compared to palbociclib in CDK4/6i-S cells, prompted us to assess its biochemical and antitumor effects in vivo. Treatment of mice with MS140 promoted degradation of CDK4/6 kinases and suppression of RB–E2F output in JeKo-1 (MCL) and Colo25 (colorectal carcinoma) xenografts in vivo (Fig. 5j,k), resulting in equivalent or greater suppression of tumor growth compared to palbociclib administered at the same total daily concentration (Fig. 5l,m and Extended Data Fig. 5f) without significant body weight loss (Extended Data Fig. 5g) or other apparent toxicities.

The increased potency of MS140 in tumor cells compared to palbociclib, prompted us to assess the effect of MS140 on normal tissue to obtain some evidence on its potential therapeutic index. To this end, we determined the effect of treatment of the same daily dose of either MS140 or palbociclib on expression of known E2F targets in tumor and normal tissue (kidney and liver). In fact, the analysis revealed that treatment with MS140 resulted in more potent inhibition in the tumor than in normal tissue of several E2F-target transcripts, compared to palbociclib (Fig. 5n), suggesting that the strategy of combined inhibition and degradation of CDK4/6 may show higher therapeutic index than current clinical CDK4/6is. Finally, as a known toxicity induced by CDK4/6is is a reduction on white blood cells and neutropenia⁴, we performed complete blood counts in mice after 3 weeks of treatment with either palbociclib or MS140 at the same total daily concentration. Consistent with the clinical experience⁴ and previous studies in mice⁴⁶, treatment with palbociclib resulted in a severe (>50%) reduction in neutrophil count (Fig. 5o). Total white blood cell count and lymphocytes were also reduced in response to palbociclib treatment, whereas red blood cells were slightly reduced (Extended Data Fig. 5h). However, neither the white blood cell nor lymphocytes count was decreased after treatment with MS140 (Fig. 5o), suggesting that the compound can be better tolerated at doses that are comparably effective with palbociclib. Thus, our results indicate that CDK4/6-directed degraders may be both a highly effective and well-tolerated therapeutic option for patients with CDK4-driven tumors, as well as for patients with tumors dependent on CDK4/6i-S CDK6.

In CDK4/6-R cells, CDK4/6 degraders fail to degrade CDK6 due to weak binding of compound

As MS140 both inhibits and degrades CDK4/6, we would expect it to show increased potency compared to the parent CDK4/6is in all cell lines. Notably however, in all CDK4/6i-R tumor lines tested, treatment with MS140 suppressed RB–E2F output and cell growth less effectively than palbociclib (Fig. 6a,b). This result was unexpected, as shRNA-mediated knockdown of CDK6 sensitized these same tumor cells to palbociclib (Fig. 3a,b), suggesting that MS140 downregulated CDK6 expression and activity with different potency in the two groups of cells lines. As MS140 is a bifunctional compound, the difference could be the result of either differential binding to CDK6 or difference in CRBN expression between the two groups of cell lines. As shown in Fig. 1e, with the exception of H358, CRBN is expressed at comparable levels in CDK4/6i-S and CDK4/6-R cell lines. Further, ectopic expression of CRBN in a CDK4/6i-R cell line did not increase CDK6 protein degradation or inhibition of downstream RB–E2F output by compound (Extended Data Fig. 6a). We thus concluded that differences in CRBN expression cannot be the underlying basis for the difference in sensitivity to MS140 and focused on the effect of MS140 directly on CDK6.

We first compared the extent of reduction of CDK6 protein expression in response to shRNA-mediated CDK6 knockdown or MS140 treatment. As shown in Fig. 6c,d, MS140 was less potent in promoting CDK6 protein degradation and in suppressing cell growth, compared to shRNA-mediated CDK6 knockdown. These results prompted us to directly compare the extent of CDK4/6 degradation by MS140 in CDK4/6i-S and CDK4/6i-R cell lines. Treatment with MS140 resulted in potent degradation of CDK4/6 in both CDK4-driven and CDK6-driven CDK4/6i-S cells, but promoted only minimal CDK4/6 degradation in CDK6-driven CDK4/6i-R cells (Fig. 6e). To assess whether this difference applied more generally to CDK4/6 degraders, we tested two recently reported CDK4/6-directed PROTACs, YKL-06-102 (refs. ^23,24) and BSJ-02-162 (ref. ²⁴). We found that, similarly to MS140, treatment with either compound resulted in more potent degradation of CDK6 in CDK4/6i-S as compared to CDK4/6i-R cells (Extended Data Fig. 6b,c). Of note, one of these compounds has been reported to be a CDK6-selective inhibitor, compared to CDK4. We confirmed that, as reported, that compound was a relatively more potent degrader of CDK6 than CDK4 in CDK4/6-S cells, compared to other CDK4/6 degraders (data not shown). However, as shown in Extended Data Fig. 6b,c, YKL-06-102 was similarly ineffective to the rest of CDK4/6 degraders in degrading CDK6 in CDK4/6-R cells. Thus, failure to degrade CDK4/6 in CDK4/6i-R cells is a general property of CDK4/6 degraders.

To directly assess binding of CDK4/6is to CDK6 in cells, we compared the shift of the CDK6 melting temperature (T_m) promoted by CDK4/6is in CDK4/6i-S versus CDK4/6i-R tumor cells (all driven by CDK6) by cellular thermal shift assay (CETSA)^47,48. Treatment of CDK4/6i-S, but not CDK4/6i-R, cells with palbociclib resulted in a shift in T_m, indicating strong inhibitor binding to CDK6 in these cells (Fig. 6f). Similar results were obtained with MS140-ve, used instead of MS140 to make sure there is no CRBN-mediated protein degradation that could confound the experiment (Fig. 6g). Finally, to assess inhibitor binding in cells using a second method, we employed a probe-based chemoproteomic method using biotinylated phosphates of ATP or ADP that irreversibly react with protein kinases on conserved lysine residues in the ATP binding pocket⁴⁹, followed by western blot analysis. We found that palbociclib prevented binding of the probe in CDK4/6i-S cells but not in CDK4/6i-R tumor cells (Fig. 6h). Together these data show that CDK6-dependent tumor cells can be divided in two groups. In most tumor cells that express CDK6, CDK4/6is bind weakly to CDK6 and these cells are resistant to CDK4/6is (and consequently to CDK4/6 degraders). In a smaller group of CDK6-dependent cells, CDK6 binds strongly to CDK4/6is and they are thus sensitive to CDK4/6is and degraders.

CDK4/6i-R cells express CDK6 as a thermostable, weak HSP90 client protein

To gain mechanistic insight on the basis of the difference in binding of CDK6 to CDK4/6is, we carried out protein expression analysis of the main known CDK4/6-related regulators (Fig. 7a), as well as immunoprecipitation of CDK6 followed by MS analysis of interacting partners in a CDK4/6-S (KMS-12-PE) and in a CDK4/6i-R (Calu6) cell line (both expressing high levels of and dependent on CDK6) (Fig. 7b). Association of CDK6 with known CDK6 interactors, including members of the cyclin, INK4 or CIP/KIP families correlated well with their relative basal expression in the two cell lines (Fig. 7c and Extended Data Fig. 7a), suggesting that additional factors affect tumor cell sensitivity to CDK4/6is. For example, p27 was found to be expressed at high levels and associate strongly with CDK6 in a CDK4/6i-S cell line (KMS12PE) as well as in a CDK4/6i-R line (A375) (Fig. 7c). We however noticed a threefold to fivefold higher interaction of CDK6 with multiple components of the HSP90–CDC37 chaperoning complex in CDK4/6i-S compared to CDK4/6i-R cells (Fig. 7b and Extended Data Fig. 7a). The observation was further confirmed by direct co-immunoprecipitation experiments showing a much stronger interaction of CDK6 with HSP90 and CDC37 in CDK4/6i-S as compared to CDK4/6i-R tumor cells (Fig. 7d and Extended Data Fig. 7b).

Association of a kinase with the HSP90–CDC37 complex has been shown to correlate with kinase dependence on HSP90 for its folding and stability¹⁵. In fact, treatment with the HSP90 inhibitor ganetespib resulted in much less potent CDK6 degradation in CDK4/6i-R cells, than in CDK4/6i-S cells (Fig. 7e and Extended Data Fig. 7c). A structurally distinct HSP90 inhibitor, luminespib showed the same results (Extended Data Fig. 7d). Moreover, CDK6 protein levels remained unchanged after siRNA-mediated knockdown of CDC37 in a CDK4/6-R line (Extended Data Fig. 7e), further confirming that the HSP90–CDC37 chaperoning activity is not required for CDK6 protein stability in these cells. As expected, treatment of CDK4-only-expressing cells with ganetespib resulted in potent degradation of CDK4 (Extended Data Fig. 7f), a known strong HSP90 client^18,50. Other known HSP90 clients were degraded to a similar degree upon ganetespib treatment (Extended Data Fig. 7g), indicating that the HSP90–CDC37 chaperone system operates similarly in the two settings, but CDK6 in CDK4/6i-S cells is expressed as a strong HSP90 client, whereas in CDK4/6i-R cells it is predominantly expressed as a weak HSP90 client.

Strong HSP90 client kinases have been shown to be highly thermo unstable¹⁵. We compared the thermostability of CDK6 in CDK6-driven CDK4/6i-S versus CDK4/6i-R tumor cells using CETSA. Consistent with the relative association of the two CDK6 forms with HSP90, we found much lower thermostability (lower T_m) of CDK6 protein in CDK4/6i-S cells as compared to that of CDK6 expressed in CDK4/6i-R cells (Fig. 7f). CDK6 in the two cellular settings also differed remarkably in its overall stability. Cycloheximide treatment resulted in significantly faster degradation of CDK6 in CDK4/6i-S compared to CDK4/6i-R cells (Fig. 7g).

Stronger association of a kinase with HSP90 has been reported to shift the conformational ensemble of different activation states toward a more active distribution^18,51. Crystallographic data show that CDK4/6is are type I, in that they bind the active conformation of CDK4/6 (αC-IN/DFG-IN)⁵². Our findings support a model by which, in CDK6-driven and CDK4/6i-S tumor cells, the CDK6 protein population is enriched in highly active conformations that have strong binding affinity for current CDK4/6is (and consequently for CDK4/6 degraders). In contrast, in CDK4/6i-R cells, CDK6 is expressed as a thermostable, weak HSP90 client, with lower affinity for CDK4/6is. To further test this idea, we inserted a previously reported activating mutation in CDK6 (S178P)⁵³ and we confirmed that it activated CDK6 (Extended Data Fig. 7h). Treatment of cells expressing either WT or mutationally activated CDK6 with MS140 resulted in substantially more potent degradation of CDK6 (S178P) compared to CDK6 WT (Fig. 7h), consistent with a highly active conformation of CDK6 being more sensitive to degradation by MS140. Finally, a previous report using biotinylated phosphates of ATP or ADP followed by MS has shown that in RB-null cells, palbociclib does not bind CDK4/6 kinases⁵⁴. Consistent with those results, we found that in BT549, RB-null cells, CDK6 is also highly thermostable and resistant to degradation by MS140, similar to the CDK6 state in palbociclib-resistant cells (Fig. 7i and data not shown).

Discussion

We show here that the expression state of CDK6 is a critical regulator of tumor response to CDK4/6is. Cells universally express CDK4, but CDK6 expression varies substantially across tumor types as well as within the same tumor type. We found that tumors with low CDK6 expression relative to CDK4 are sensitive to CDK4/6is. Such tumor types include luminal breast cancer, Ewing sarcomas, MCLs and others, as well as subgroups within tumor types, such as a portion of lung adenocarcinomas. Among lung adenocarcinoma models, we found a subgroup with a low ratio of CDK6 to CDK4 protein expression that was sensitive to CDK4/6is. The association of low CDK6 expression with increased clinical response to CDK4/6is was further confirmed by analyzing clinical data in patients with NSCLC²⁷. Thus, collectively our data indicate that a low ratio of CDK6 to CDK4 expression in the tumor may be used to stratify patients who are likely to benefit from CDK4/6-based therapies. Further, it is becoming increasingly apparent that in various contexts of intrinsic or acquired resistance, combinatorial targeting of parallel or convergent growth factor signaling pathways, such as the PI3K–PTEN–AKT^55,56,57 or RAS–MAPK^25,58 pathway may be necessary for effective and durable tumor response. Our findings suggest that selection of patients with low CDK6 relative to CDK4 expression, in combinations of CDK4/6is or CDK4/6 degraders with inhibitors of parallel pathways, such as PI3K, AKT, MEK or ERK inhibitors are likely to be most effective.

We further found that in most tumor cells, CDK6 expression confers intrinsic resistance to CDK4/6is. The finding is consistent with CDK6 overexpression being a reported mechanism of acquired resistance to CDK4/6is in luminal breast cancer^13,59. However, CDK4/6is have been shown here (Extended Data Fig. 4a) and by others¹ to potently bind purified CDK6 in vitro. In addition, a proportion of tumors (mostly of certain types of leukemias, such as acute myeloid leukemia) have been shown to require CDK6 (ref. ¹⁴) and are sensitive to CDK4/6is, suggesting that CDK6 may or may not confer resistance to CDK4/6is depending on cellular context. We used a potent and selective CDK4/6 degrader that we developed (MS140) as a tool to resolve the discrepancy between the in vitro data and our findings in cells. By employing in-cell binding assays, we found that CDK6 is expressed in different tumors in either of two distinct forms, only one of which is accessible to binding by CDK4/6is (Fig. 7h). We further found that the two CDK6 forms differ dramatically in their thermostability and their dependence on the HSP90–CDC37 chaperone complex. The data are consistent with previous studies showing kinase dependence on HSP90 to be associated with an enrichment of folded, highly active states of the kinase¹⁸, which, in the case of CDK6, is structurally predicted to bind strongly to CDK4/6is. The conformation of the active site of the highly active CDK6 in cells likely resembles that of the highly active purified CDK6, commonly used in in vitro kinase and binding assays. Thus, the current clinical CDK4/6is have been developed to strongly bind and inhibit CDK6 selectively in the subgroup of tumor cells that express CDK6 similar to its purified, highly active state.

Recently, two studies have proposed the interaction of CDK4 to either p27 (ref. ⁶⁰) or p16 (ref. ⁶¹) as mechanisms promoting resistance to CDK4/6is. In another study, formation of complexes with members of the cyclin D family or with endogenous inhibitors of CDKs was shown to affect the interaction of different members of the CDK family of proteins with inhibitors and HSP90–CDC37 in vitro²¹. However, we observed p16 to be expressed at undetectable levels in most cancer cell lines tested in our experiments, including the ones that are insensitive to CDK4/6is (Figs. 1e and 7a) and we did not observe a correlation of endogenous expression of members of the cyclin D family with either CDK6 sensitivity to CDK4/6is or HSP90 dependence. Further, p27 is expressed in most CDK4/6i-S and CDK4/6-R cells at similar levels (Figs. 1e and 7a) and knockdown or overexpression of p27 did not affect sensitivity of cell lines to CDK4/6is (data not shown). Finally, we found that all cell lines tested expressing CDK4 and low levels of CDK6 are universally sensitive to CDK4/6is, regardless of p16 or p27 status or expression levels. We therefore conclude that although the reported mechanisms may affect target engagement by CDK4/6is or the interaction of CDK4 or CDK6 with HSP90–CDC37 in certain contexts, they are unlikely to be the predominant mechanism by which large subsets of RB-proficient tumors diverge in terms of their sensitivity to CDK4/6is.

The duality of CDK6 when expressed in different cells with respect to its dependence on HSP90–CDC37 challenges the current paradigm on the molecular determinants of kinase dependence on HSP90. HSP90–CDC37 dependency of a given kinase has been shown to be altered by genetic alterations, as in the cases of c-Src and v-Src¹⁵, WT and mutated BRAF¹⁹ and WT and mutated EGFR²⁰. Previous studies using ectopically expressed proteins, have shown that CDK6 exhibited an intermediate dependency on HSP90 (HSP90 interaction score of 2.23), lying in between the highly dependent CDK4 (score of 3.15) and the entirely HSP90-independent CDK2 (score of 0)¹⁵. However, we unexpectedly found that the same kinase (CDK6) can be expressed as a strong or weak HSP90–CDC37 client depending on the cellular context. As regulation of kinase function by the HSP90–CDC37 complex remains incompletely understood¹⁵, in the case of CDK6 may be affected by post-translational modification(s), interacting partner(s) or localization, in different cellular environments.

Current clinical CDK4/6is have been optimized to bind and inhibit CDK4 as well as the thermo-unstable, strong HSP90 client form of CDK6. However, based on our data, it is expected that a large number of tumors express thermostable CDK6, which is resistant to these drugs. Drug development efforts directed in potently targeting this newly identified state of CDK6 may bring about more effective therapies for a large number of patients with RB-proficient tumors.

Methods

Compounds

Commercially available compounds palbociclib, ganetespib, luminespib and bortezomib were obtained from Selleckchem. Abemaciclib, ribociclib and MLN4924 were obtained from Medchem Express. Blasticidin S, CHX and DOX were purchased from Sigma. Chemical compounds were dissolved either in DMSO or water and stored at −20 °C.

Cell culture

Cell lines HEK293T, HEK293FT, Huh-7, U87MG, A375 and A673 were maintained in DMEM supplemented with 10% fetal bovine serum, 2 mM glutamine and penicillin/streptomycin, MCF7, T47D, ZR-75-1, CAMA-1, NCI-H358, Colo205, SK-MEL-1, IGROV-1, TC-71, Mino, JeKo-1, Granta 519, Z-138, REC-1, KMS-12-PE, MV4-11, MOLM-14, Pfeiffer, SK-MEL-2, A549, Calu6, NCI-H1792, NCI-H2087, NCI-H2291, NCI-H1915, NCI-H1666, NCI-H1395, HCC827, PC9, HCT15, HCT116, RKO, LoVo, HT-29, SW620 and BT549 were maintained in RPMI1640 supplemented with 10% fetal bovine serum, 2 mM glutamine and penicillin/streptomycin (Thermo Fisher Scientific).

MCF7, T47D, ZR-75-1, CAMA-1, NCI-H358, Colo205, A375, A549, SK-MEL-2, Calu6, NCI-H1792, NCI-H2087, NCI-H2291, NCI-H1915, NCI-H1666, NCI-H1395, HCC827, RKO, HCT15, HCT116, SW620, LoVo, HT-29 and HEK293T were purchased from the American Type Culture Collection. PC9 cells were purchased from Sigma-Aldrich. U87MG, IGROV-1 and BT549 were provided by R. Parsons. Huh-7 cells were provided by A. Lujambio. Z-138 cells were provided by E. P. Reddy. REC-1, KMS-12-PE and Pfeiffer were gifts from S. Parekh. MV4-11 and MOLM-14 cells were provided by I. Aifantis (New York Univeristy). Mino, JeKo-1 and Granta 519 were gifts from S. M. Buckley (University of Nebraska). A673 and TC-71 were provided by C. A. Pratilas (John Hopkins). HEK293FT cells were a gift from W. Kaelin Jr (Harvard Medical School).

Plasmids

Tet-pLKO-puro (cat. no. 21915) and pcDNA3-FLAG-CRBN (cat. no. 107380) was purchased from Addgene. shCDK4 (TRCN0000018364 and TRCN0000010520), shCDK6 (TRCN0000010081 and TRCN0000000488) and shCDC37 (TRCN0000116633 and TRCN0000116632) (Supplementary Table 1) were subcloned into Tet-pLKO plasmid. Human CDK6 coding sequences were amplified from HEK293H cDNA using a SuperScript III First-Strand kit (Thermo Fisher Scientific). C-terminal V5-tagged CDK6 were subcloned using pcDNA3 as a backbone. shRNA-resistant form of V5-CDK6 and S178P V5-CDK6 were generated with a QuikChange II XL Site-Directed Mutagenesis kit (Agilent).

siRNA knockdown

A673, TC-71 and NSCLC cell lines were transfected with SMARTpool ON-TARGETplus Non-targeting pool or SMARTpool ON-TARGETplus CDK4-siRNA (L-003238-00-0005), or SMARTpool ON-TARGETplus CDK6-siRNA (L-003240-00-0005) (Dharmacon) using Lipofectamine RNAiMAX (Thermo Fisher Scientific).

Western blot and immunoprecipitation

Cells were washed with PBS and lysed on ice for 10 min in NP40 buffer (50 mM Tris (pH 7.5), 1% NP40, 150 mM NaCl, 10% glycerol, 1 mM EDTA) supplemented with protease and phosphatase inhibitors (Roche). Lysates were centrifuged at 15,000 r.p.m. for 10 min and the protein concentration was quantified using BCA (Thermo Fisher Scientific). Protein G agarose (Thermo Fisher Scientific) was used for immunoprecipitations. The following antibodies were used: phospho-RB (Ser807/811, 1:1,000 dilution), RB (4H1, 1:1,000 dilution), PLK1 (208G4, 1:1,000 dilution), cyclin D2 (D52F9, 1:1,000 dilution), cyclin E2 (1:1,000 dilution), Akt1 (2H10, 1:1,000 dilution), JAK2 (D2E12, 1:1,000 dilution), DHFR, ß-actin (13E5, 1:1000), anti-rabbit, HRP (1:1000) and anti-mouse, HRP (1:1000) (Cell Signaling), CDK4 (H-22), CDK6 (C-21; DCS-83, 1:1,000 dilution), CDK2 (D-12, 1:1,000 dilution), CDK7 (C-19, 1:2,000 dilution), CDK9 (D-7, 1:3,000 dilution), p15/16 (C-7, 1:400 dilution), p16 (C-20, 1:500 dilution), p18 (118.2, 1:250 dilution), p57 (KP39, 1:500 dilution), cyclin A (H-432, 1:2,000 dilution; B-8, 1:1,000 dilution), cyclin D1 (M-20, 1:400 dilution; A-12, 1:250 dilution), cyclin D3 (D-7, 1:500 dilution), cyclin E1 (HE12, 1:2,000 dilution), CDC37 (H-271; E-4, 1:2,000 dilution), Hsp70 (W27, 1:8,000 dilution) and Hsp90α/ß (F-8, 1:8,000 dilution) (Santa Cruz), p27 (1:1,000 dilution) and c-Raf (1:2,000 dilution) (BD Transduction Laboratories), p21 (1:1,000 dilution) and phospho-CDK4-T172 (1:1,000 dilution) (ABclonal), CRBN (1:500 dilution; Novus Biologicals), V5 (1:5,000 dilution; Thermo Fisher Scientific) and FLAG M2 (1:3,000 dilution; Sigma).

In vitro CDK4 and CDK6 kinase assay

Palbociclib, MS140 and ribociclib were assayed by Reaction Biology Corporation for in vitro kinase activity. Briefly, kinase substrates were added to base reaction buffer (20 mM HEPES (pH 7.5), 10 mM MgCl₂, 1 mM EGTA, 0.02% Brij35, 0.02 mg ml⁻¹ BSA, 0.1 mM Na₃VO₄, 2 mM dithiothreitol and 1% DMSO). Differential CDK4/6 complexed with cyclin D1 were then diluted in the substrate solution. A range of concentrations of compounds were incubated with the kinase reaction mixture by acoustic technology (Echo550; nanoliter range) and incubated for 20 min at room temperature followed by the addition of 10 μM ³³P-ATP. Reactions were incubated for 2 h at room temperature and radioactivity was detected by filter binding methods. Kinase activity data were presented as percentage kinase activity in samples relative to DMSO control. IC₅₀ values were calculated using GraphPad Prism 5.

Lentivirus transduction and stable cell lines

Lentivirus was produced by transfecting HEK293FT cells with a lentiviral transfer vector, psPAX2 and pMD.G at a 5:4:1 ratio using Lipofectamine 2000 (Thermo Fisher Scientific). The viral supernatant was collected 72 h after transfection and filtered through a 0.45-µm filter unit (Millipore). For Dox-inducible stable CDK4/6 knockdown cell lines, cells were transduced with shCDK4 or shCDK6 lentivirus in the presence of 8 µg ml⁻¹ polybrene (EMD Millipore) and selected with 2 µg ml⁻¹ puromycin (Thermo Fisher Scientific) for 5–7 d.

Cellular thermal shift assay

For basal CDK6 thermal stability assay, cells resuspended in 1 ml PBS containing protease inhibitor cocktail (Roche) were aliquoted with 100 µl cell suspension each into nine PCR tubes (~3 × 10⁶ cells). PCR tubes were heated at different temperature end points (35–59 °C) for 3 min in the Veriti 96-well thermal cycler (Thermo Fisher Scientific). Tubes were removed and incubated at room temperature for 3 min. Samples were immediately snap-frozen in liquid nitrogen. Cells were frozen and thawed three times using liquid nitrogen and a thermal cycler at 25 °C. Cell lysates were centrifuged at 20,000g for 15 min at 4 °C. Supernatants were collected for western blots.

For palbociclib or MS140-ve-induced thermal shift assay, cells were treated with 1 µM palbociclib or 15 µM MS140-ve for 2 h. Cells resuspended in PBS containing protease inhibitor cocktail (Roche) were aliquoted with 100 µl cell suspension each into nine PCR tubes. Samples were heated at different temperature end points for 3 min using the Veriti 96-well thermal cycler (Thermo Fisher Scientific), followed by incubation at room temperature for 3 min and snap-freezing in liquid nitrogen. After three freeze–thaw cycles, samples were centrifuged at 20,000g for 15 min at 4 °C. Supernatants were collected for western blot analysis.

ATP-biotin competition assay

Cell lysates were prepared and labeled according to the manual instructions for the Pierce Kinase Enrichment Kits and ActivX Probes (Thermo Fisher Scientific). Briefly, cells were pretreated with palbociclib for 2 h, lysed and centrifuged at 16,000g for 10 min at 4 °C. The supernatant was desalted through Zeba Spin Desalting Columns. Then, 1 mg of total cell lysates in 500 µl were used for ATP competition reaction with a final concentration of 5 µM desthiobiotin-ADP probe for 10 min at room temperature. Samples were mixed with 500 µl 8 M urea and 50 µl streptavidin agarose for 1 h at room temperature on a rotator. Beads were washed with 4 M urea/lysis buffer and collected by centrifugation at 1,000g for 1 min. Proteins were eluted with 2× sample buffer at 95 °C for 5 min. Samples were analyzed by immunoblotting.

Generation of CRBN ^−/− ZR-75-1 cell line by CRISPR/Cas9

Human CRBN gRNA (5′-CACCGTAAACAGACATGGCCGGCGA-3′) was designed using the following website: http://crispr.mit.edu/ and cloned into BsmBI-digested lentiCRISPRv2 (Addgene, 52961). ZR-75-1 cells were transduced with sgCRBN lentivirus in the presence of 8 µg ml⁻¹ polybrene followed by selecting with 2 µg ml⁻¹ puromycin for 5 d. Cells were plated at 0.3 cells per well in a 96-well plate. After 2–3 weeks, individual clones were expanded. CRBN homozygous knockout clones were validated by genotyping with primers (F: 5′- AAG TCA TGC TAA GGG CTG GAA C -3′, R: 5′- GGA TGG GTT TCC TGT TCT TAA TAG -3′) and western blotting.

Protein extraction and digestion

Colo205 cells were treated with 0.3 µM MS140 or MS140-ve for 5 h in duplicate. Cell pellets were lysed in lysis buffer (8 M urea, 100 mM Tris-HCl pH 8.0). Sonication (5 s on, 5 s off; 2 × 30 s) was performed to shear genome DNA. Lysates were centrifuged for 30 min at 3,500g at 4 °C, and the supernatants were transferred to clean tubes. Protein concentration was determined (BCA assay), and protein was reduced with 5 mM dithiothreitol (DTT), alkylated with 15 mM iodoacetamide (IAA) in the dark, then diluted with 3× volume of 25 mM Tris (pH 8.0), 1 mM CaCl₂. The final urea concentration was 2 M. Trypsin was added into the protein solution with 1:100 ratio (trypsin:protein) and digested for 12–16 h or overnight at room temperature.

Mass spectrometry analysis

Peptides were cleaned up by C18 stage tips, and peptide concentration was determined (Peptide assay, Thermo 23275). The clean peptides were dissolved in 0.1% formic acid and analyzed on a Q-Exactive HF-X coupled with an Easy nanoLC 1200 (Thermo Fisher Scientific). 0.5 µg of peptides was loaded onto an Acclain PepMap RSLC C18 column (150 mm × 75 μm inner diameter (ID), C18, 2 μm, Thermo Fisher). Analytical separation of all peptides was achieved with 100 min gradient. A linear gradient of 5 to 30% buffer B over 80 min was executed at a 300 nl/min flow rate following by a ramp to 100% buffer B in 5 min, and a 15-min wash with 100% buffer B, where buffer A was aqueous 0.1% formic acid and buffer B was 80% acetonitrile and 0.1% formic acid. LC-MS experiments were performed in a data-dependent mode with full MS (externally calibrated to a mass accuracy of <5 p.p.m. and a resolution of 60,000 at m/z 200) followed by high-energy collision-activated dissociation-MS/MS of the top 20 most intense ions with a resolution of 15,000 at m/z 200. High-energy collision-activated dissociation-MS/MS was used to dissociate peptides at a normalized collision energy of 27 eV in the presence of nitrogen bath gas atoms. Dynamic exclusion was 30.0 s. There were two biological replicates for one treatment, and each sample was subjected to two technical LC-MS replicates.

MS data analysis

Mass spectra processing and peptide identification were performed on the Andromeda search engine in MaxQuant software (version 1.6.10.43) against the human UniProt database (UP000005640). All searches were conducted with a defined modification of cysteine carbamidomethylation, with methionine oxidation and protein amino-terminal acetylation as dynamic modifications. Peptides were confidently identified using a target-decoy approach with a peptide false discovery rate (FDR) of 1% and a protein FDR of 1%. A minimum peptide length of seven amino acids was required, maximally two missed cleavages were allowed, initial mass deviation for precursor ion was up to 7 p.p.m., and the maximum allowed mass deviation for fragment ions was 0.5 Da. Data processing and statistical analysis were performed on Perseus (version 1.6.10.50). Protein quantitation was performed on biological and technical replicate runs, and two-sample t-test statistics was used to report statistically significant expression fold changes.

MS to identify CDK6-interacting proteins

Preparation of samples for MS

The affinity purified proteins were reduced, alkylated and digested with trypsin directly on beads. Briefly, beads were resuspended in 100 μl 100 mM ammonium bicarbonate. Proteins were reduced with 2 μl of 0.2 M dithiothreitol (Sigma) for 1 h at 57 °C (pH 7.5), alkylated with 2 μl of 0.5 M iodoacetamide (Sigma) for 45 min at room temperature (RT) in the dark and digested using 200 ng sequencing-grade trypsin (Promega) overnight at RT with gentle shaking. The solution was transferred to a new tube and digestion was stopped by adding 100 μl of a 5% formic acid and 0.2% TFA R2 50 μm Poros (Applied Biosystems) beads slurry in water. The samples were allowed to shake at 4 °C for 3 h. Beads were loaded onto C18 ziptips (Millipore), equilibrated with 0.1% TFA, using a microcentrifuge for 30 s at 6,000 r.p.m. Beads were washed with 0.5% acetic acid. Peptides were eluted with 40% acetonitrile in 0.5% acetic acid followed by 80% acetonitrile in 0.5% acetic acid. Organic solvent was removed using a SpeedVac concentrator and the sample was reconstituted in 0.5% acetic acid.

MS analysis

One-tenth of each sample was loaded onto an Acclaim PepMap trap column (2 cm × 75 µm) in line with an EASY-Spray analytical column (50 cm × 75 µm ID PepMap C18, 2-μm bead size) using the auto sampler of an EASY-nLC 1200 HPLC (Thermo Fisher Scientific) with solvent A consisting of 2% acetonitrile in 0.5% acetic acid, and solvent B consisting of 80% acetonitrile in 0.5% acetic acid. Peptides were gradient-eluted into a Thermo Fisher Scientific Q Exactive HF-X Mass Spectrometer using the following gradient: 5–35% in 60 min, 35–45% in 10 min, followed by 45–100% in 10 min. High-resolution full MS spectra were recorded with a resolution of 45,000, an AGC target of 3×10⁶, with a maximum ion time of 45 ms and a scan range from 400 to 1,500 m/z. MS/MS spectra were collected using a resolution of 15,000, an AGC target of 1×10⁵, maximum ion time of 120 ms, one microscan, 2-m/z isolation window and normalized collision energy of 27.

Data processing

MS/MS spectra were searched against the UniProt human reference proteome database containing common contaminant proteins using Sequest within Proteome Discoverer 2.3. Search parameters were as follows: precursor mass tolerance ±10 ppm, fragment mass tolerance ±0.02 Da, digestion parameters trypsin allowing two missed cleavages, fixed modification of carbamidomethyl on cysteine, variable modification of oxidation on methionine, variable modification of deamidation on glutamine and asparagine and a 1% peptide and protein FDR searched against a decoy database. The results were filtered to only include proteins identified by at least two unique peptides.

Cell viability assay

Cells were seeded at 3–10 × 10³ cells per well in 96-well plates. At 24 h after seeding, cells were treated with palbociclib and MS140 at a range of concentrations for 72 h. Ten microliters of 0.1 mg ml⁻¹ resazurin (Sigma-Aldrich) or 20 µl of CellTiter 96 AQueous One Solution Reagent (Promega) was added to cells and incubated for 2–3 h at 37 °C. Cell viability was determined by measuring the fluorescence at 560 nm excitation wavelength and 590 nm emission wavelength (resazurin) or 490 nm (CellTiter 96 AQueous One Solution Reagent) using a Molecular Devices Spectramax M5 plate reader. IC₅₀ values were calculated using log-transformed, normalized data in GraphPad Prism 5.0.

Crystal violet cell growth assay

Cells were seeded at 1–10 × 10³ cells per well in six-well plates. The next day, cells were treated with the increasing concentrations of palbociclib and MS140 for 10–15 d. Cell culture medium was replaced every 2 d in the presence or absence of inhibitors. Cells were fixed with 10% formalin solution (Sigma-Aldrich) for 5 min at RT followed by 0.05% crystal violet for 25 min. cells were de-stained with tap water and air dried.

Quantitative real-time PCR

Total RNA was extracted using Trizol Reagent (Thermo Fisher Scientific). Complementary DNA was synthesized with a SuperScript IV First-Strand kit (Thermo Fisher Scientific). Quantitative real-time PCR was performed using a Fast SYBR Green Master Mix kit (Thermo Fisher Scientific) with a 7500 Fast real-time PCR system (Applied Biosystems). PCR primers are listed in Supplementary Table 1. Differences in expression were calculated by the ΔΔCt method.

Genetic dependency data and genomics data

CRISPR dependency data were obtained from the 20Q1 public Avana dataset containing genome-scale CRISPR knockout screens for 18,333 genes in 739 cell lines. The gene dependencies were estimated for each gene and cell lines by the CERES algorithm³¹. RNA interference (RNAi) dependency data were derived from combination of the Broad Institute Project Achilles, Novartis Project Drive and Marcotte et al. database^32,33,34,35. Genetic dependencies were estimated using the DEMETER2 model across 712 unique cancer cell lines³⁶. Cancer cell line mRNA expression was taken from the DepMap 20Q1 data release. Cancer Cell Line Encyclopedia proteomics data were obtained by quantitative profiling of proteins by MS across 375 cell lines¹. Normalized protein quantification data and cell line omics data can be downloaded at DepMap (https://depmap.org/portal/).

Animal experiments

To determine in vivo degradation efficacy, 5–7-week-old female athymic Nude-Foxn1^nu mice obtained from Envigo Laboratories were injected subcutaneously with 1 × 10⁷ JeKo-1 cells in 1:1 PBS/Matrigel GFR membrane matrix (Corning) or 5 × 10⁶ Colo205 cells in PBS. Mice were treated with vehicle (5% DMSO and 95% PEG 300) or MS140 (25 or 30 mg kg⁻¹) intraperitoneally, twice daily or palbociclib (50 or 60 mg kg⁻¹) orally once daily for 3 d when tumors reach around 100 mm³. At 5 h after the last dose, tumors were collected for further analysis. Liver and kidney were collected for qPCR analysis.

For efficacy of MS140 on tumor xenograft, 5 × 10⁶ Colo205 cells in PBS or 10 × 10⁶ JeKo-1 cells in 1:1 PBS/Matrigel GFR membrane matrix were injected subcutaneously on the left flank in 6-week-old female athymic Nude-Foxn1^nu mice. Tumors were allowed to reach 100 mm³ in size before the animals were randomized into two groups of 5–8 mice per group. Mice were treated with vehicle (5% DMSO and 95% PEG 300) or MS140 (25 or 30 mg kg⁻¹) intraperitoneally, twice daily or palbociclib (50 or 60 mg kg⁻¹) orally once daily for 3 weeks. Tumor size was measured using calipers every 3 d and tumor volume was calculated using the following formula: (length × width²) / 2. Tumor samples and organs were collected at the end of treatment for further analysis. All experiments were conducted under a protocol approved by Mount Sinai School of Medicine Institutional Animal Care and Use Committee (IACUC-2016-0066). All mice were maintained in specific-pathogen-free barrier conditions under a 12-h-light–12-h-dark cycle and with temperature and humidity set points at 20–25% and 30–70%, respectively.

Complete blood count

Female 5–7-week-old C57BL/6 mice (Envigo Laboratories) were dosed intraperitoneally with 25 mg kg⁻¹ MS140 twice daily or 50 mg kg⁻¹ palbociclib once daily via oral gavage for 3 weeks. Mouse blood samples were collected in K₃EDTA tubes before treatment and last treatment. Complete blood counts were performed with a Coulter Ac•T 5diff Hematology Analyzer (BECKMAN).

Trial design

The JUNIPER study was a randomized phase III study of abemaciclib with best supportive care versus erlotinib with best supportive care in patients with stage IV NSCLC with a detectable KRAS mutation whose disease had progressed after platinum-based chemotherapy.

RNA sequence

A total of 148 patients had evaluable formalin-fixed paraffin-embedded tumor sections for CDK6 RNA sequencing and further statistical analysis. RNA exome sequencing was performed by Almac with the Illumina TruSeq RNA Sample Preparation kit v2. CDK6 sequencing data were quantile-normalized across samples.

Statistical method and cutoff point selection

The Kaplan–Meier product limit method was used to estimate the survival curve (OS or PFS) for treatment arms in both CDK6 expression low and high subgroup. Unstratified Cox proportional hazard model were performed to compare treatment effect in both CDK6 expression low and high subgroup. HR estimates and 95% CIs from the Cox model were reported on the Kaplan–Meier plot. The cutoff to define CDK6 expression low and high subgroups was derived from an unstratified Cox proportional hazard interaction model by minimizing the treatment by CDK6 interaction P value in terms of OS.

Software

R programming language was used for analysis.

Chemical synthesis

Chemistry general procedures

HPLC spectra for all compounds were acquired using an Agilent 1200 Series system with a diode-array detection detector. Chromatography was performed on a 2.1 × 150 mm Zorbax 300SB-C₁₈ 5-μm column with water containing 0.1% formic acid as solvent A and acetonitrile containing 0.1% formic acid as solvent B at a flow rate of 0.4 ml min⁻¹. The gradient program was as follows: 1% B (0–1 min), 1−99% B (1–4 min) and 99% B (4–8 min). High-resolution mass spectra data were acquired in positive ion mode using an Agilent G1969A API-TOF with an electrospray ionization (ESI) source. NMR spectra were acquired on a Bruker DRX-600 spectrometer (600 MHz ¹H, 150 MHz ¹³C) or a Varian Mercury spectrometer (400 MHz ¹H, 100 MHz ¹³C). Chemical shifts are reported in ppm (δ). Preparative HPLC was performed on Agilent Prep 1200 series with ultraviolet detector set to 254 nm. Samples were injected into a Phenomenex Luna 75 × 30 mm, 5-μm, C₁₈ column at RT. The flow rate was 40 ml min⁻¹. A linear gradient was used with 10% (or 50%) of methanol (A) in H₂O (with 0.1 % TFA) (B) to 100% of methanol (A). HPLC was used to establish the purity of target compounds. All final compounds had >95% purity using the HPLC methods described above.

(2-(2,6-Dioxopiperidin-3-yl)-1,3-dioxoisoindolin-4-yl)glycine

A solution of 2-(2,6-dioxopiperidin-3-yl)-4-fluoroisoindoline-1,3-dione (1.38 g, 5.0 mmol), tert-butyl glycinate (0.66 g, 5.0 mmol) and N,N-diisopropylethylamine (1.31 ml, 7.5 mmol) in DMF (10 ml) was heated to 85 °C in a microwave reactor for 40 min. After cooling to RT, the reaction was diluted with water and extracted with ethyl acetate (3×). Combined organic phase was dried over anhydrous sodium sulfate and concentrated under reduced pressure. The resulting residue was purified by silica gel chromatography to give the desired ^tBu ester intermediate as oil. This intermediate was treated with a solution of hydrogen chloride in dioxane (10 ml, 4.0 M) for 16 h. The reaction was concentrated under reduced pressure to give the desired acid product (0.24 g, 14%). 1H NMR (600 MHz, CD₃OD) δ 7.57 (dd, J = 8.5, 7.1 Hz, 1H), 7.11 (d, J = 7.1 Hz, 1H), 6.95 (d, J = 8.5 Hz, 1H), 5.07 (dd, J = 12.6, 5.5 Hz, 1H), 4.12 (s, 2H), 2.86 (ddd, J = 18.0, 14.4, 5.4 Hz, 1H), 2.74 – 2.67 (m, 2H), 2.15 – 2.08 (m, 1H). MS (ESI) m/z 332.1 [M + H]⁺.

4-((2-(4-(6-((6-Acetyl-8-cyclopentyl-5-methyl-7-oxo-7,8-dihydropyrido(2,3-d)pyrimidin-2-yl)amino)pyridin-3-yl)piperazin-1-yl)-2-oxoethyl)amino)-2-(2,6-dioxopiperidin-3-yl)isoindoline-1,3-dione (MS140)

To a solution of (2-(2,6-dioxopiperidin-3-yl)-1,3-dioxoisoindolin-4-yl)glycine (0.017 g, 0.051 mmol) in DCM (10 ml) and DMSO (2 ml) were added palbociclib (0.023 g, 0.051 mmol), 4-methylmorpholine (0.020 g, 0.20 mmol), 1-hydroxy-7-azabenzotriazole (0.0090 g, 0.066 mmol) and N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (0.013 g, 0.066 mmol). The reaction mixture was allowed to stir at RT for 16 h, before being concentrated under reduced pressure. The resulting residue was purified by reverse-phase prep-HPLC to yield the product (0.028 g, 72%) as a yellow solid. 1H NMR (600 MHz, CD₃OD) δ 9.08 (s, 1H), 8.21 (dd, J = 9.6, 2.8 Hz, 1H), 7.86 (d, J = 2.8 Hz, 1H), 7.55 – 7.50 (m, 2H), 7.03 (d, J = 7.0 Hz, 1H), 6.99 (d, J = 8.5 Hz, 1H), 6.04 – 5.97 (m, ¹H), 5.08 (dd, J = 12.7, 5.5 Hz, 1H), 4.24 (s, 2H), 3.88 – 3.73 (m, 4H), 3.42 – 3.31 (m, 4H), 2.92 – 2.82 (m, 1H), 2.80 – 2.67 (m, 2H), 2.50 (s, 3H), 2.43 (s, 3H), 2.35 – 2.27 (m, 2H), 2.15 – 2.06 (m, 3H), 1.94 – 1.87 (m, 2H), 1.73 – 1.66 (m, 2H). High-resolution mass spectra (ESI-TOF) m/z: [M + H]⁺ calculated for C₃₉H₄₁N₁₀O₇, 761.3160; found: 761.3150.

Statistics and reproducibility

No data were excluded in our analysis. All data represent single experiments. In animal experiments, the tumors were randomized across treatments, the investigators were not blinded to allocation during experiments and outcome assessment. Statistical analysis was performed using a two-tailed, paired or unpaired Student’s t-tests between two groups. Variation is indicated by mean ± s.d. unless otherwise specified. GraphPad Prism software (v.5) and R v.3.6.3 were used for statistical calculations.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

MS raw files for CDK6 complex analysis are available at the Mass Spectrometry Interactive Virtual Environment (https://massive.ucsd.edu) under ID MSV000086571.

MS files for global protein degradation have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD023137.

Normalized protein quantification data and cell line omics data can be downloaded at DepMap (https://depmap.org/portal/).

Reagents generated in this study will be made available on request, but we may require a payment and/or a completed Materials Transfer Agreement if there is potential for commercial application. Source data are provided with this paper.

Code availability

Data analysis was performed in R v.3.6.3 using custom-made or publicly available R packages. The code is available from GitHub (https://github.com/lijin0303/CDK46_expression_dependency). Figure 3e and Extended Data Fig. 3b were generated with R 3.6.3 (Packages taigr, dplyr, tidyr, ggplot2, ggpubr) and corresponding code can be found in https://github.com/lijin0303/CDK46_expression_dependency. Source data are provided with this paper.

References

Sherr, C. J., Beach, D. & Shapiro, G. I. Targeting CDK4 and CDK6: from discovery to therapy. Cancer Discov. 6, 353–367 (2016).
Article CAS PubMed Google Scholar
O’Leary, B., Finn, R. S. & Turner, N. C. Treating cancer with selective CDK4/6 inhibitors. Nat. Rev. Clin. Oncol. 13, 417–430 (2016).
Article PubMed CAS Google Scholar
Otto, T. & Sicinski, P. Cell cycle proteins as promising targets in cancer therapy. Nat. Rev. Cancer 17, 93–115 (2017).
Article CAS PubMed PubMed Central Google Scholar
Finn, R. S. et al. Palbociclib and letrozole in advanced breast cancer. N. Engl. J. Med. 375, 1925–1936 (2016).
Article CAS PubMed Google Scholar
Hortobagyi, G. N. et al. Ribociclib as first-line therapy for HR-positive, advanced breast cancer. N. Engl. J. Med. 375, 1738–1748 (2016).
Article CAS PubMed Google Scholar
Ingham, M. & Schwartz, G. K. Cell-cycle therapeutics come of age. J. Clin. Oncol. https://doi.org/10.1200/JCO.2016.69.0032 (2017).
Goetz, M. P. et al. MONARCH 3: abemaciclib as initial therapy for advanced breast cancer. J. Clin. Oncol. 35, 3638–3646 (2017).
Article CAS PubMed Google Scholar
Im, S. A. et al. Overall survival with ribociclib plus endocrine therapy in breast cancer. N. Engl. J. Med. 381, 307–316 (2019).
Article CAS PubMed Google Scholar
Knudsen, E. S. & Witkiewicz, A. K. The strange case of CDK4/6 Inhibitors: mechanisms, resistance, and combination strategies. Trends Cancer 3, 39–55 (2017).
Article CAS PubMed PubMed Central Google Scholar
Hamilton, E. & Infante, J. R. Targeting CDK4/6 in patients with cancer. Cancer Treat. Rev. 45, 129–138 (2016).
Article CAS PubMed Google Scholar
Gong, X. et al. Genomic aberrations that activate D-type cyclins are associated with enhanced sensitivity to the CDK4 and CDK6 inhibitor abemaciclib. Cancer Cell 32, 761–776 (2017).
Article CAS PubMed Google Scholar
Kim, S. et al. The potent and selective cyclin-dependent kinases 4 and 6 inhibitor ribociclib (LEE011) is a versatile combination partner in preclinical cancer models. Oncotarget 9, 35226–35240 (2018).
Article PubMed PubMed Central Google Scholar
Yang, C. et al. Acquired CDK6 amplification promotes breast cancer resistance to CDK4/6 inhibitors and loss of ER signaling and dependence. Oncogene https://doi.org/10.1038/onc.2016.379 (2016).
Placke, T. et al. Requirement for CDK6 in MLL-rearranged acute myeloid leukemia. Blood 124, 13–23 (2014).
Article CAS PubMed PubMed Central Google Scholar
Taipale, M. et al. Quantitative analysis of HSP90-client interactions reveals principles of substrate recognition. Cell 150, 987–1001 (2012).
Article CAS PubMed PubMed Central Google Scholar
Vaughan, C. K. et al. Structure of an Hsp90-Cdc37-CDK4 complex. Mol. Cell 23, 697–707 (2006).
Article CAS PubMed PubMed Central Google Scholar
Stepanova, L., Leng, X., Parker, S. B. & Harper, J. W. Mammalian p50Cdc37 is a protein kinase-targeting subunit of Hsp90 that binds and stabilizes CDK4. Genes Dev. 10, 1491–1502 (1996).
Article CAS PubMed Google Scholar
Taipale, M. et al. Chaperones as thermodynamic sensors of drug-target interactions reveal kinase inhibitor specificities in living cells. Nat. Biotechnol. 31, 630–637 (2013).
Article CAS PubMed PubMed Central Google Scholar
Grbovic, O. M. et al. V600E B-Raf requires the Hsp90 chaperone for stability and is degraded in response to Hsp90 inhibitors. Proc. Natl Acad. Sci. USA 103, 57–62 (2006).
Article CAS PubMed Google Scholar
Sawai, A. et al. Inhibition of Hsp90 down-regulates mutant epidermal growth factor receptor (EGFR) expression and sensitizes EGFR mutant tumors to paclitaxel. Cancer Res. 68, 589–596 (2008).
Article CAS PubMed PubMed Central Google Scholar
Hallett, S. T. et al. Differential regulation of G1 CDK complexes by the Hsp90-Cdc37 chaperone system. Cell Rep. 21, 1386–1398 (2017).
Article CAS PubMed PubMed Central Google Scholar
Lai, A. C. & Crews, C. M. Induced protein degradation: an emerging drug discovery paradigm. Nat. Rev. Drug Discov. 16, 101–114 (2017).
Article CAS PubMed Google Scholar
Brand, M. et al. Homolog-selective degradation as a strategy to probe the function of CDK6 in AML. Cell Chem. Biol. 26, 300–306 e309 (2019).
Article CAS PubMed Google Scholar
Jiang, B. et al. Development of dual and selective degraders of cyclin-dependent kinases 4 and 6. Angew. Chem. Int. Ed. Engl. 58, 6321–6326 (2019).
Article CAS PubMed PubMed Central Google Scholar
Ruscetti, M. et al. NK cell-mediated cytotoxicity contributes to tumor control by a cytostatic drug combination. Science 362, 1416–1422 (2018).
Article CAS PubMed PubMed Central Google Scholar
Finn, R. S. et al. PD 0332991, a selective cyclin D kinase 4/6 inhibitor, preferentially inhibits proliferation of luminal estrogen receptor-positive human breast cancer cell lines in vitro. Breast Cancer Res. 11, R77 (2009).
Article PubMed PubMed Central CAS Google Scholar
Goldman, J. W. et al. A randomized phase 3 study of abemaciclib versus erlotinib in previously treated patients with stage IV NSCLC with KRAS mutation: JUNIPER. J. Clin. Oncol. 36, 9025 (2018).
Article Google Scholar
Pacheco, J. & Schenk, E. CDK4/6 inhibition alone and in combination for non-small cell lung cancer. Oncotarget 10, 618–619 (2019).
Article PubMed PubMed Central Google Scholar
Ghandi, M. et al. Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature 569, 503–508 (2019).
CAS PubMed PubMed Central Google Scholar
Kato, J. Y., Matsuoka, M., Strom, D. K. & Sherr, C. J. Regulation of cyclin D-dependent kinase 4 (CDK4) by CDK4-activating kinase. Mol. Cell. Biol. 14, 2713–2721 (1994).
CAS PubMed PubMed Central Google Scholar
McFarland, J. M. et al. Improved estimation of cancer dependencies from large-scale RNAi screens using model-based normalization and data integration. Nat. Commun. 9, 4610 (2018).
Article PubMed PubMed Central CAS Google Scholar
Meyers, R. M. et al. Computational correction of copy number effect improves specificity of CRISPR-Cas9 essentiality screens in cancer cells. Nat. Genet. 49, 1779–1784 (2017).
Article CAS PubMed PubMed Central Google Scholar
Tsherniak, A. et al. Defining a cancer dependency map. Cell 170, 564–576 (2017).
Article CAS PubMed PubMed Central Google Scholar
McDonald, E. R. 3rd et al. Project DRIVE: a compendium of cancer dependencies and synthetic lethal relationships uncovered by large-scale, deep RNAi screening. Cell 170, 577–592 (2017).
Article CAS PubMed Google Scholar
Nusinow, D. P. et al. Quantitative proteomics of the Cancer Cell Line Encyclopedia. Cell 180, 387–402 (2020).
Article CAS PubMed PubMed Central Google Scholar
Marcotte, R. et al. Functional genomic landscape of human breast cancer drivers, vulnerabilities, and resistance. Cell 164, 293–309 (2016).
Article CAS PubMed PubMed Central Google Scholar
Gadd, M. S. et al. Structural basis of PROTAC cooperative recognition for selective protein degradation. Nat. Chem. Biol. 13, 514–521 (2017).
Article CAS PubMed PubMed Central Google Scholar
Chen, P. et al. Spectrum and degree of CDK drug interactions predicts clinical performance. Mol. Cancer Ther. 15, 2273–2281 (2016).
Article PubMed CAS Google Scholar
Bondeson, D. P. et al. Catalytic in vivo protein knockdown by small-molecule PROTACs. Nat. Chem. Biol. 11, 611–617 (2015).
Article CAS PubMed PubMed Central Google Scholar
Winter, G. E. et al. Drug development. Phthalimide conjugation as a strategy for in vivo target protein degradation. Science 348, 1376–1381 (2015).
Article CAS PubMed PubMed Central Google Scholar
Chamberlain, P. P. et al. Structure of the human Cereblon-DDB1-lenalidomide complex reveals basis for responsiveness to thalidomide analogs. Nat. Struct. Mol. Biol. 21, 803–809 (2014).
Article CAS PubMed Google Scholar
Zhang, C. et al. Proteolysis targeting chimeras (PROTACs) of anaplastic lymphoma kinase (ALK). Eur. J. Med. Chem. 151, 304–314 (2018).
Article CAS PubMed PubMed Central Google Scholar
Fischer, E. S. et al. Structure of the DDB1-CRBN E3 ubiquitin ligase in complex with thalidomide. Nature 512, 49–53 (2014).
Article CAS PubMed PubMed Central Google Scholar
Lebraud, H., Wright, D. J., Johnson, C. N. & Heightman, T. D. Protein degradation by in-cell self-assembly of proteolysis targeting chimeras. ACS Cent. Sci. 2, 927–934 (2016).
Article CAS PubMed PubMed Central Google Scholar
Sumi, N. J., Kuenzi, B. M., Knezevic, C. E., Remsing Rix, L. L. & Rix, U. Chemoproteomics reveals novel protein and lipid kinase targets of clinical CDK4/6 inhibitors in lung cancer. ACS Chem. Biol. 10, 2680–2686 (2015).
Article CAS PubMed PubMed Central Google Scholar
Bisi, J. E. et al. Preclinical development of G1T38: a novel, potent and selective inhibitor of cyclin dependent kinases 4/6 for use as an oral antineoplastic in patients with CDK4/6 sensitive tumors. Oncotarget 8, 42343–42358 (2017).
Article PubMed PubMed Central Google Scholar
Martinez Molina, D. et al. Monitoring drug target engagement in cells and tissues using the cellular thermal shift assay. Science 341, 84–87 (2013).
Article PubMed CAS Google Scholar
Jafari, R. et al. The cellular thermal shift assay for evaluating drug target interactions in cells. Nat. Protoc. 9, 2100–2122 (2014).
Article CAS PubMed Google Scholar
Patricelli, M. P. et al. In situ kinase profiling reveals functionally relevant properties of native kinases. Chem. Biol. 18, 699–710 (2011).
Article CAS PubMed PubMed Central Google Scholar
Verba, K. A. et al. Atomic structure of Hsp90-Cdc37-Cdk4 reveals that Hsp90 traps and stabilizes an unfolded kinase. Science 352, 1542–1547 (2016).
Article CAS PubMed PubMed Central Google Scholar
Boczek, E. E. et al. Conformational processing of oncogenic v-Src kinase by the molecular chaperone Hsp90. Proc. Natl Acad. Sci. USA 112, E3189–E3198 (2015).
Article CAS PubMed PubMed Central Google Scholar
Zhang, J., Yang, P. L. & Gray, N. S. Targeting cancer with small molecule kinase inhibitors. Nat. Rev. Cancer 9, 28–39 (2009).
Article PubMed CAS Google Scholar
Bockstaele, L., Bisteau, X., Paternot, S. & Roger, P. P. Differential regulation of cyclin-dependent kinase 4 (CDK4) and CDK6, evidence that CDK4 might not be activated by CDK7, and design of a CDK6 activating mutation. Mol. Cell. Biol. 29, 4188–4200 (2009).
Article CAS PubMed PubMed Central Google Scholar
Nomanbhoy, T. K. et al. Chemoproteomic evaluation of target engagement by the cyclin-dependent kinase 4 and 6 inhibitor palbociclib correlates with cancer cell response. Biochemistry 55, 5434–5441 (2016).
Article CAS PubMed Google Scholar
Romano, G. et al. A preexisting rare PIK3CA(E545K) subpopulation confers clinical resistance to MEK plus CDK4/6 inhibition in NRAS melanoma and is dependent on S6K1 signaling. Cancer Discov. 8, 556–567 (2018).
Article CAS PubMed PubMed Central Google Scholar
Herrera-Abreu, M. T. et al. Early adaptation and acquired resistance to CDK4/6 inhibition in estrogen receptor-positive breast cancer. Cancer Res. 76, 2301–2313 (2016).
Article CAS PubMed PubMed Central Google Scholar
Costa, C. et al. PTEN loss mediates clinical cross-resistance to CDK4/6 and PI3Kα inhibitors in breast cancer. Cancer Discov. 10, 72–85 (2020).
Article CAS PubMed Google Scholar
de Leeuw, R. et al. MAPK reliance via acquired CDK4/6 inhibitor resistance in cancer. Clin. Cancer Res. 24, 4201–4214 (2018).
Article PubMed PubMed Central Google Scholar
Li, Z. et al. Loss of the FAT1 tumor suppressor promotes resistance to CDK4/6 inhibitors via the hippo pathway. Cancer Cell 34, 893–905 (2018).
Article PubMed PubMed Central CAS Google Scholar
Guiley, K. Z. et al. p27 allosterically activates cyclin-dependent kinase 4 and antagonizes palbociclib inhibition. Science https://doi.org/10.1126/science.aaw2106 (2019).
Green, J. L. et al. Direct CDKN2 modulation of CDK4 alters target engagement of CDK4 inhibitor drugs. Mol. Cancer Ther. 18, 771–779 (2019).
Article CAS PubMed Google Scholar
Nusinow P. & Gygi, S. P. A guide to the quantitative proteomic profiles of the Cancer Cell Line Encyclopedia. Preprint at bioRxiv https://doi.org/10.1101/2020.02.03.932384 (2020).

Download references

Acknowledgements

We thank A. Lujambio, S. Parekh, R. Parsons, E.P. Reddy, R. Sachidanandam, (Icahn School of Medicine at Mount Sinai), I. Aifantis (New York University), S.M. Buckley (University of Nebraska), C.A. Pratilas (Johns Hopkins University) and W.G. Kaelin Jr. (Harvard Medical School) for sharing cell lines and reagents. P.I.P. is supported by the US National Institutes of Health (NIH)/NCI (R01CA204314, R01 CA240362 and R01CA238229), the Irma T. Hirschl Trust, the Manhasset Women’s Coalition against Breast Cancer, the Breast Cancer Alliance, the Melanoma Research Foundation, the Melanoma Research Alliance and Tisch Cancer Institute developmental awards. T.A.A. has been supported by grant T32CA078207 and Z.K. would like to acknowledge the 2017 Robin Chemers Neustein Postdoctoral Fellowship. W.R.S. acknowledges support from the NIH (1R01CA233626) and the Ludwig Center at Harvard, T.I. is supported by the Department of Defense Peer Reviewed Cancer Research Program Horizon Award (W81XWH-19-1-0271). X.C. is supported by the NIH (R01-GM133107). S.A.A. acknowledges support from a grant from the Breast Cancer Research Foundation. This work was supported in part by the P30CA196521 grant (to J.J.) from the NIH and an endowed professorship from the Icahn School of Medicine at Mount Sinai (to J.J.). This work utilized the AVANCE NEO 600 MHz NMR Spectrometer System that was upgraded with funding from an NIH SIG grant 1S10OD025132-01A1.

Author information

Authors and Affiliations

Department of Oncological Sciences, The Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA
Xuewei Wu, Tamer A. Ahmed, Zoi Karoulia, Christos Adamopoulos, Stuart A. Aaronson & Poulikos I. Poulikakos
Department of Pharmacological Sciences, The Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA
Xiaobao Yang, Yan Xiong, Jing Liu & Jian Jin
The Broad Institute of Harvard and MIT, Dana-Farber Cancer Institute and Harvard Medical School, Boston, MA, USA
Ruitong Li, Takahiro Ito & William R. Sellers
Eli Lilly and Company, Indianapolis, IN, USA
Hong Wang & Sean G. Buchanan
Department of Biochemistry and Biophysics, University of North Carolina, Chapel Hill, NC, USA
Li Wang, Ling Xie & Xian Chen
Department of Biochemistry and Molecular Pharmacology, New York University, New York, NY, USA
Beatrix Ueberheide

Authors

Xuewei Wu
View author publications
You can also search for this author in PubMed Google Scholar
Xiaobao Yang
View author publications
You can also search for this author in PubMed Google Scholar
Yan Xiong
View author publications
You can also search for this author in PubMed Google Scholar
Ruitong Li
View author publications
You can also search for this author in PubMed Google Scholar
Takahiro Ito
View author publications
You can also search for this author in PubMed Google Scholar
Tamer A. Ahmed
View author publications
You can also search for this author in PubMed Google Scholar
Zoi Karoulia
View author publications
You can also search for this author in PubMed Google Scholar
Christos Adamopoulos
View author publications
You can also search for this author in PubMed Google Scholar
Hong Wang
View author publications
You can also search for this author in PubMed Google Scholar
Li Wang
View author publications
You can also search for this author in PubMed Google Scholar
Ling Xie
View author publications
You can also search for this author in PubMed Google Scholar
Jing Liu
View author publications
You can also search for this author in PubMed Google Scholar
Beatrix Ueberheide
View author publications
You can also search for this author in PubMed Google Scholar
Stuart A. Aaronson
View author publications
You can also search for this author in PubMed Google Scholar
Xian Chen
View author publications
You can also search for this author in PubMed Google Scholar
Sean G. Buchanan
View author publications
You can also search for this author in PubMed Google Scholar
William R. Sellers
View author publications
You can also search for this author in PubMed Google Scholar
Jian Jin
View author publications
You can also search for this author in PubMed Google Scholar
Poulikos I. Poulikakos
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

X.W. conceived of the study, performed most of the experiments, analyzed the data and helped write the manuscript. X.Y. and Y.X. synthesized chemical compounds. T.A.A, Z.K. and C.A. performed experiments and analyzed data. S.A.A designed research, analyzed data and helped write the manuscript. J.L. and J.J. designed chemical compounds and analyzed data. L.X. carried out MS analysis. B.U. and X.C. supervised MS analysis. R.L., L.W., T.I. and H.W. performed data analysis. S.G.B. and W.R.S. designed the research, analyzed the data and helped to write the manuscript. P.I.P conceived of the study, designed research, analyzed the data and wrote the manuscript.

Corresponding author

Correspondence to Poulikos I. Poulikakos.

Ethics declarations

Competing interests

X.W., X.Y., Y.X., T.A., Z.K., J.L., J.J. and P.I.P. hold a patent on CDK4/6-directed degradation (WO/2018/106870). The Jian Jin laboratory received research funds from Celgene Corporation, Levo Therapeutics and Cullgen, Inc. J.J. is an equity shareholder and consultant of Cullgen, Inc. H.W. and S.G.B. are employees of Eli Lilly. All other authors declare no competing interests.

Additional information

Peer review information Nature Cancer thanks John Brognard, Keiran Smalley and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Reporting Summary

Supplementary Table 1

Oligonucleotide sequences

Cite this article

Wu, X., Yang, X., Xiong, Y. et al. Distinct CDK6 complexes determine tumor cell response to CDK4/6 inhibitors and degraders. Nat Cancer 2, 429–443 (2021). https://doi.org/10.1038/s43018-021-00174-z

Download citation

Received: 17 July 2020
Accepted: 21 January 2021
Published: 01 March 2021
Issue Date: April 2021
DOI: https://doi.org/10.1038/s43018-021-00174-z

This article is cited by

ERK hyperactivation serves as a unified mechanism of escape in intrinsic and acquired CDK4/6 inhibitor resistance in acral lentiginous melanoma
- Kasturee Jagirdar
- Marie E. Portuallo
- Vito W. Rebecca
Oncogene (2024)
PROTAC-mediated CDK degradation differentially impacts cancer cell cycles due to heterogeneity in kinase dependencies
- Vishnu Kumarasamy
- Zhe Gao
- Erik S. Knudsen
British Journal of Cancer (2023)
The RNA-binding protein LRPPRC promotes resistance to CDK4/6 inhibition in lung cancer
- Wei Zhou
- Wenxi Wang
- Xiaohong Fang
Nature Communications (2023)
LIMK2 promotes melanoma tumor growth and metastasis through G3BP1-ESM1 pathway-mediated apoptosis inhibition
- Parmanand Malvi
- Dhana Sekhar Reddy
- Narendra Wajapeyee
Oncogene (2023)
Cellular mechanisms underlying response and resistance to CDK4/6 inhibitors in the treatment of hormone receptor-positive breast cancer
- April C. Watt
- Shom Goel
Breast Cancer Research (2022)

Subjects

Abstract

Similar content being viewed by others

Main

Results

Intrinsic resistance to CDK4/6is is associated with incomplete inhibition of RB–E2F and expression of CDK6

Low expression of CDK6 predicts for sensitivity to CDK4/6is in NSCLC

Tumors expressing both CDK4 and CDK6 depend selectively on CDK6

Development of MS140, a potent and selective CDK4/6 degrader (PROTAC)

CDK4/6-directed degradation is more effective than CDK4/6is in CDK4/6i-S tumor cells

In CDK4/6-R cells, CDK4/6 degraders fail to degrade CDK6 due to weak binding of compound

CDK4/6i-R cells express CDK6 as a thermostable, weak HSP90 client protein

Discussion

Methods

Compounds

Cell culture

Plasmids

siRNA knockdown

Western blot and immunoprecipitation

In vitro CDK4 and CDK6 kinase assay

Lentivirus transduction and stable cell lines

Cellular thermal shift assay

ATP-biotin competition assay

Generation of CRBN −/− ZR-75-1 cell line by CRISPR/Cas9

Protein extraction and digestion

Mass spectrometry analysis

MS data analysis

MS to identify CDK6-interacting proteins

Preparation of samples for MS

MS analysis

Data processing

Cell viability assay

Crystal violet cell growth assay

Quantitative real-time PCR

Genetic dependency data and genomics data

Animal experiments

Complete blood count

Trial design

RNA sequence

Statistical method and cutoff point selection

Software

Chemical synthesis

Chemistry general procedures

(2-(2,6-Dioxopiperidin-3-yl)-1,3-dioxoisoindolin-4-yl)glycine

4-((2-(4-(6-((6-Acetyl-8-cyclopentyl-5-methyl-7-oxo-7,8-dihydropyrido(2,3-d)pyrimidin-2-yl)amino)pyridin-3-yl)piperazin-1-yl)-2-oxoethyl)amino)-2-(2,6-dioxopiperidin-3-yl)isoindoline-1,3-dione (MS140)

Statistics and reproducibility

Reporting Summary

Data availability

Code availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Additional information

Extended data

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

This article is cited by

Search

Quick links

Generation of CRBN ^−/− ZR-75-1 cell line by CRISPR/Cas9