CDK6-PI3K signaling axis is an efficient target for attenuating ABCB1/P-gp mediated multi-drug resistance (MDR) in cancer cells

Multidrug resistance (MDR) mediated by ATP binding cassette subfamily B member 1 (ABCB1/P-gp) is a major cause of cancer chemotherapy failure, but the regulation mechanisms are largely unknown. Based on single gene knockout, we studied the regulation of CDK6-PI3K axis on ABCB1-mediated MDR in human cancer cells. CRISPR/Cas9 technique was performed in KB-C2 cells to knockout cdk6 or cdk4 gene. Western blot, RT-PCR and transcriptome analysis were performed to investigate target gene deletion and expression of critical signaling factors. The effect of cdk4 or cdk6 deficiency on cell apoptosis and the cell cycle was analyzed using flow cytometry. In vivo studies were performed to study the sensitivity of KB-C2 tumors to doxorubicin, tumor growth and metastasis. Deficiency of cdk6 led to remarkable downregulation of ABCB1 expression and reversal of ABCB1-mediated MDR. Transcriptomic analysis revealed that CDK6 knockout regulated a series of signaling factors, among them, PI3K 110α and 110β, KRAS and MAPK10 were downregulated, and FOS-promoting cell autophagy and CXCL1-regulating multiple factors were upregulated. Notably, PI3K 110α/110β deficiency in-return downregulated CDK6 and the CDK6-PI3K axis synergizes in regulating ABCB1 expression, which strengthened the regulation of ABCB1 over single regulation by either CDK6 or PI3K 110α/110β. High frequency of alternative splicing (AS) of premature ABCB1 mRNA induced by CDK6, CDK4 or PI3K 110α/110β level change was confirmed to alter the ABCB1 level, among them 10 common skipped exon (SE) events were found. In vivo experiments demonstrated that loss of cdk6 remarkably increased the sensitivity of KB-C2 tumors to doxorubicin by increasing drug accumulation of the tumors, resulting in remarkable inhibition of tumor growth and metastasis, as well as KB-C2 survival in the nude mice. CDK6-PI3K as a new target signaling axis to reverse ABCB1-mediated MDR is reported for the first time in cancers. Pathways leading to inhibition of cancer cell proliferation were revealed to be accompanied by CDK6 deficiency.


Introduction
Multidrug resistance (MDR) can develop in cancer cells that survive during chemotherapy. P-glycoprotein (P-gp), revealed as ATP-binding cassette (ABC) subfamily B member 1 (ABCB1) is an ABC transporter and its overexpression by various cancers can produce resistance to various chemotherapeutic drugs that have distinct structures and differing mechanisms of action [1], which hurdles the efficiency of chemotherapy against cancers. Despite the reversal efficacy of various types of ABCB1 inhibitors, mechanisms of inducing and regulating ABCB1-mediated MDR are rarely reported on the level of signaling pathways. Lack of understanding of ABCB1 regulation may lead to increased risk of side effects when treated with inhibitors, most of which are due to the lack of specificity.
An ideal target can exert its efficiency to inhibit cancer cell proliferation when it is used for attenuating MDR in cancer. Through CRISPR/Cas9 gene editing, we recently reported that 110α and 110β subunits of the phosphoinositide 3-kinase (PI3K) signaling pathway are potential targets for reversal of ABCB1/P-gp-and ABCG2/BCRPmediated cancer MDR in a manner independent of AKT expression [2]. With the aim to characterize other efficient and safe targets that can be used to reverse ABCB1mediated MDR in cancers, we further investigated the signaling pathways involved in the PI3K110α and 110β regulated ABCB1 expression.
The cyclin-dependent kinases (CDKs) are members of a family of serine-threonine kinases that regulate the cell cycle by altering cell proliferation and apoptosis [3][4][5][6][7][8][9][10][11][12]. Although many functional studies were based on the inhibitor-triggered variation of protein expression, which were frequently influenced by the cytotoxicity or non-specificity of the inhibitors instead of inhibition of CDKs themselves, some functions are demonstrated by protein-protein interactions. CDKs comprise three domains: the ATP binding and catalytic domain, the cyclin-binding domain, and the P13suc1 binding domain, in which P13suc1 can inhibit the activity of CDK and cell phase alteration [13][14][15]. By mediating a common kinase domain containing peptide, PSTAIRE, CDKs bind to their corresponding cyclin, forming a cyclin-CDK complex, where CDKs catalyze the phosphorylation of certain serine and threonine residues in target proteins. This regulates gene transcription and cell division, further promoting alteration of cell cycle between different phases, e.g., phase G1 to S, G2 to M, and exit from phase M [16].
Currently, whether CDK4 and/or CDK6 have direct regulatory effects on ABCB1 and ABCB1-mediated MDR in cancer, their roles, and mechanisms are unknown. Most of the inhibitors lack specificity, and some of them cannot accumulate inside the cells efficiently because they may be substrates of ABCB1 [29][30][31]. In this study, by gene knockout with CRISPR/Cas9 gene editing technique, we determined the regulation of CDK4 or CDK6 on ABCB1-mediated MDR in human epidermoid carcinoma MDR cell line KB-C2 which overexpresses ABCB1 and H460/MX80 which expresses ABCB1 at low level. Then we focused on the coordination of CDK6 with PI3K 110α/β to regulate ABCB1 expression, and their function to induce alternative splicing (AS) of ABCB1 pre-mRNA. Importantly, this study demonstrated that CDK6 is a novel multi-functional target that may act as an onoff switch for ABCB1 expression, potentially reversing ABCB1-mediated MDR in cancers, and therefore, has significance in combination with cancer chemotherapy.

Determination of cell viability: MTT assay
Exponentially growing cells were seeded into 96 well plates at 5 × 10 3 cells/well. After 72 h of incubation, 20 μL of MTT (5 mg/mL) was added to each well. After incubation for an additional 4 h, the medium containing MTT was discarded and replaced with 150 μL of DMSO to dissolve the dark blue-purple crystal. The absorbance was measured at a wavelength of 490 nm, using an ELx 800 Universal Microplate Reader (Bio-Tek, Inc. Winooski, VT). The relative survival rate for the cells was analyzed using the SPSS 20 program (SPSS Inc., Chicago, IL) and the survival rate -drug concentration curves were generated using Origin 9.0 software (OriginLab Corporation, Northampton, MA). The concentration of drug required to inhibit cell viability by 50% (IC 50 value) was determined using Origin 9.0 software.

Western blot assay and immunofluorescence (IF) analysis
Parental KB-3-1 cells and MDR KB-C2 cells were incubated with 9 μM of ribociclib for 2 h and co-cultured with paclitaxel or colchicine for 24 to 72 h. For the Western blot assay, the cells were lysed with SDS lysate reagent, separated on a gradient polyacrylamide gel and transferred to a PVDF membrane. After blocking with 5% milk and washed with TBST buffer, the membrane was incubated with mouse anti-ABCB1 antibody at 4 °C for 2 h, adequately washed with TBST, and incubated with goat anti-mouse IgG-HRP at RT for 2 h. The membrane was then washed with TBST and exposed to the SignalFire ™ ECL Reagent developing reagent (Cell Singling Technology, Danvers, MA), and the results were quantified using a AI600 RGB GEL Imaging System (GE, Fairfield, CT) set for the chemiluminescence mode.

Analysis of regulation of CDK6-PI3K axis on ABCB1 mediated drug resistance
CRISPR/Cas9 technique was performed in KB-C2 cells to knockout cdk6 or cdk4 gene. Western blot, RT-PCR and transcriptome analysis were performed to investigate target gene deletion and expression of critical signaling factors. Detailed procedure and other methods were described in supplemental materials and methods.

CRISPR/Cas9 knockout and characterization for target gene deficiency
The cells were seeded into 48-well plates with serumfree DMEM. A polymeric GenePORTER transfection reagent was mixed with sgRNA/Cas9 all-in-one expression plasmid HCP254656-CG12-1-10 or HCP288574-CG12-1-10 to target AGT GTG AGA GTC CCC AAT GG in cdk4 (NM_000075.3) and GAA GAA CGG AGG CCG TTT CG in cdk6 (NM_001259.7), respectively (GeneCopoeia Inc., Rockville, MD). After incubation at room temperature for 10 min, the mixture was added to the cells. After incubation for 4 h, FBS was added to the cultures at 20% of the final volume ratio. To achieve a high transfection ratio, the same well of cells were transfected again after 48 h of cultivation. The drugs used in this study were not used during transfection and cell stabilization, and RT-PCR and Western blotting were performed to determine the transcription and expression of the target genes after stabilizing the gene deficient cells for 1 month via continuous cell cultivation.

Analysis of CDK4 or CDK6 to regulate ABCB1 expression and ABCB1-mediated anti-drug efficacy
In the KB-C2 cells, either the cdk4 or cdk6 gene was deleted and the cells were cultured for 48 h and lysed with SDS lysate reagent. Western blot analysis of cell lysate was performed as described above for the determination of the effect of cdk4 or cdk6 knockout on the expression of the ABCB1 protein. The gene deficient cells were cultured with gradient concentrations (0.001, 0.01, 0.1, 1, 10 or 100 μM) of either colchicine or paclitaxel for 72 h. The MTT assay and IC 50 determination was carried out as described above.

Differential expression analysis and identification of AS events
Cells were lysed and Total RNA was extracted using the TRIzol ® Reagent, according to the manufacturer's instructions (Invitrogen). Genomic DNA was then removed using Dnase I (Takara Bio, Otsu, Japan). RNA quality was determined using a 2100 Bioanalyser (Agilent Technologies, Palo Alto, CA, USA) and quantified using ND-2000 (NanoDrop Technologies, Inc. Wilmington, DE, USA). RNA samples (5 μg) were used to construct the RNA-seq transcriptome library. The messenger RNA (mRNA) was fragmentated using a fragmentation buffer and isolated by the polyA selection method using oligo (dT) beads. According to Illumina's library construction protocol, cDNA originated double-strands DNA (dscDNA) were synthesized, subjected to endrepair, phosphorylation and ' A' base addition, selected for cDNA target fragments of 200-300 bp, followed by PCR amplification using Phusion DNA polymerase (NEB) for 15 PCR cycles, and then sequenced using the Illumina HiSeq 4000 (2 × 150 bp read length). After map reading [33], differential expression genes (DEG) between two different samples were identified according to the fragments per kilobase of exon per million mapped reads (FRKM) method [34]. Network of gene clusters and protein-protein interaction were analyzed on metascape (http:// metas cape. org) [35] or STRING (http:// stringdb. org) platform, choosing Homo Sapiens as the control species. The data was visualized by drawing heatmaps using TBtools (https:// github. com/ CJ-Chen/ TBtoo ls). For the identification of AS events. All of the AS events that occurred in the samples were identified by using the program, Multivariate Analysis of Transcript Splicing (MATS, http:// rnaseq-mats. sourc eforge. net/) [36].

Analysis of cell apoptosis using flow cytometry
The KB-C2 cells and the derivative cdk4-or cdk6-deficient cells were seeded into a 24-well plate at 1 × 10 5 cells per mL of medium and cultured with colchicine (1.25 μM) for 24 h. The cells were treated with 0.25% trypsin, followed by DMEM to terminate trypsin activity and washed two times with cold PBS. The cells incubated with the medium were used as a control group. Cell apoptosis was determined using an Annexin V-FITC Apoptosis Detection Kit (Solarbio Science & Technology Co., Ltd., Beijing, China), following the manufacturer's instructions. The cellular apoptosis and necrosis ratios were analyzed using a CytoFLEX flow cytometer (Beckman Coulter, Brea, CA).

Determining the effect of cdk4 or cdk6 deficiency on the cell cycle
The KB-C2 cells and cdk4-or cdk6-deficient cells derived from KB-C2 were seeded into a 24-well plate at 1 × 10 5 cells per mL of medium and cultured with colchicine (1.25 μM) for 24 h. Next, the cells were treated with 0.25% trypsin, followed by DMEM and washed two times with PBS. The cells incubated with medium were used as a control group. The stage of the cell cycle was determined using a Cell Cycle Staining Kit (Multi Sciences (LianKe) Biotech, Co., Ltd., Hangzhou, China), following the manufacturer's instructions. The cell populations were analyzed using a CytoFLEX flow cytometer (Beckman Coulter, Brea, CA) and the population range corresponding to the untreated KB-C2 cells in the best status (approximately 65% among all cell events), was selected as the gate for all the tested cell groups. All experiments were repeated in triplicate.

In vivo studies
We established two tumor-bearing mouse models to evaluate the reversal of MDR tumor effect of cdk6 deficiency in vivo. To establish the KB-C2 and KB-C2-k.o.cdk6 tumor-bearing mouse models, the aliquots (0.1 mL) of cancer cells in sterilized saline solution (1.0 × 10 7 cells/ mL) were subcutaneously inoculated into the back neck of the nude mice (BALB/c-nu). The mice were randomly divided into 4 groups (9 mice per group) with equivalent average starting tumor size (~ 50 mm 3 ) and body weight (~ 23 g) treated with 0.9% saline (set as control) or DOX (2 mg Kg − 1 ), via the intraperitoneal route of administration for each tumor-bearing mouse model. Animals were examined daily, and their body weights and tumor sizes were determined using the formula: Tumor size = Length×Width×Height×π/6 [37]. The increased tumor sizes were calculated by subtracting the tumor sizes in the initial group from that of saline-or DOXtreated group at the subsequent measurement time. Animals were sacrificed 19 days after treatment.
The pathological morphology of liver, lung, spleen, heart, stomach and kidney of nude mice after drug withdrawal was analyzed by HE staining. Based on transcriptome analysis, genes with high level of expression in both KB-C2 and KB-C2-k.o.cdk6 are screened. Among them, cytoskeleton-associated protein 4 (CKAP4) was used as a biomarker for determination of distribution of KB-C2 or KB-C2-k.o.cdk6 in different organs including liver, lung and spleen. Immunofluorescence staining was then performed to analyze tumor cell migration in each group of tumor-bearing nude mice.

Statistical analysis
All experiments were performed three times and the results were analyzed using Student's unpaired t-test using SPSS 20.0 software (SPSS Inc., Chicago, IL). Data were expressed as the mean ± standard error of the mean (SEM). Results showing P values less than 0.05 were considered statistically significant.

Experimental procedures
Reversal of MDR in cancer cells was studied by MTT assay. The expression level of ABCB1 was analyzed by Western blot assay and immunofluorescence (IF). the transcription and expression of the target genes were determined by RT-PCR and Western blotting, respectively, after stabilizing the gene deficient cells for 1 month via continuous cell cultivation. Up-or down-regulation of relative proteins and signaling pathways was studied by differential expression analysis and the mechanism was explored by identification of AS events and proteinprotein interaction (PPI). Cell apoptosis and the cell cycle were analyzed by flow cytometry.

Ribociclib, an inhibitor for CDK4/6, could downregulate ABCB1 expression
During screening the agents for reversal of ABCB1mediated MDR, we found that ribociclib (LEE011), an inhibitor for CDK4/6, could significantly downregulate ABCB1 expression in MDR KB-C2 cells, as shown by immunofluorescence results and Western blot (Fig. 1). To our knowledge, most of the CDK4/6 inhibitors did not show the capability to reverse ABCB1-mediated MDR or downregulate ABCB1 expression. This could possibly be related to the low efficacy of CDK4 or CDK6 to regulate ABCB1 expression, or to the low intracellular accumulation efficiency and relatively poor specificity of many known inhibitors. Therefore, it is necessary to perform gene knockout to elucidate whether CDK4 or CDK6 deficiency has a role in reversing ABCB1-mediated drug resistance.
The knockout of cdk4 and cdk6 genes using the CRISPR/ Cas9 gene editing technique To explore the potential mechanism of CDK4 or CDK6 in regulating ABCB1 expression, cdk4 and cdk6 genes were knocked out by transfecting KB-C2 cancer cells with the all-in-one CRISPR/Cas9 plasmid harboring the sgRNA that targets either the cdk4 or cdk6 genes (Fig. 2). The transfected cells were stabilized through continuous culture for at least 2 weeks. The deletions of cdk4 or cdk6 genes in KB-C2 produced a significant decrease in the expression of the CDK4 and CDK6 proteins, respectively ( Fig. 2B). Western blot analysis (Fig. 2B), RT-PCR analysis ( Fig. 2B) and transcriptome sequencing and quantification (Table S1) indicated that cdk4 or cdk6 gene were deleted by chromosomal recombination. The cdk4 or cdk6 deficient KB-C2 cells were named as KB-

CDK6 is a potential target for on-off regulation of ABCB1 expression and ABCB1-mediated MDR
Amazingly, the gene deletion experiments indicated that the cdk4 and cdk6 genes have different roles in regulating ABCB1 expression and ABCB1-mediated MDR in KB-C2 cells. Compared with the KB-C2 cells, the deletion of the cdk6 alone induced significant downregulation in the expression of the ABCB1 transporter in the KB-C2-k.o.cdk6 cells (Fig. 2C). The Western blot indicated that wild-type ABCB1 was remarkably downregulated in KB-C2-k.o.cdk6 cells (Fig. 2C) whereas trace amounts of ABCB1 homologous protein with a higher molecular weight might exist. Due to the decreased ABCB1 transporter expression, cdk6-deficient KB-C2 cells achieved remarkable reversal of MDR, showing significant increase in sensitivity to ABCB1 substrates colchicine (Fig. 2D) and paclitaxel (Fig. 2E), demonstrating that CDK6 regulates cancer MDR mediated by ABCB1 specifically.
Similarly, ABCB1 positive protein signal in Western blot was upregulated in KB-C2-k.o.cdk4 population, in which cdk4 gene was knocked out (Fig. 2C). In line with this phenomenon, MDR was increased in the KB-C2-k.o.cdk4 (cdk4 deficient KB-C2) cells incubated with low-to-medium concentrations of colchicine (< 2.5 μM) or paclitaxel (< 4 μM). However, MDR was not enhanced at higher concentrations of these drugs, indicating that KB-C2 cells deficient in CDK4 were more susceptible when incubated with higher concentrations of the drugs. Thus, our results clearly indicate that CDK6 and CDK4 function as possible on-off switches that regulate the expression of the ABCB1-mediated MDR, although CDK4 is homologous to CDK6. Native CDK6 may have a role in maintaining ABCB1-mediated MDR of cancers.

Determining potential pathway(s) for CDK6 to regulate ABCB1 expression
The downregulation of ABCB1 induced by CDK6 deficiency in KB-C2-k.o.cdk6 attracted our attention. In order to uncover the pathway for CDK6 to downregulate ABCB1 expression and reverse MDR, we analyzed differential gene expression based on the transcriptome sequencing database of the KB-C2-k.o.cdk6 cell population. Among the 8974 genes with significant variance in gene expression (p-adjust < 0.001, differential folds > 1.3), the Top10 KEGG pathways involved in the differential gene expression caused by knockout of CDK6 are intimately related to advanced glycation end product (AGE)-AGE receptor (AGE-RAGE) signaling, transforming growth factor-β signaling, endocrine resistance (acquired drug resistance to endocrine therapy of cancers), tumor necrosis factor signaling, et al., among which CDK6 is one of the top10 factors (Fig. 3A). Among the proteins most correlated with endocrine resistance, neurogenic locus notch homolog protein 3 (NOTCH3), transcription factor AP-1 (JUN), and mitogen-activated protein kinase 10 (MAPK10) were downregulated, whereas phosphatidylinositol 3-kinase regulatory subunit gamma (PIK3R3), G-protein coupled estrogen receptor 1 (GPER1), cyclindependent kinase inhibitor 2C (CDKN2C), cytochrome P450 2D6 (CYP2D6), cAMP-dependent protein kinase catalytic subunit beta (PRKACB), phosphatidylinositol 3-kinase regulatory subunit beta (PIK3R2) and notch 2 N-terminal like B (NOTCH2NLB) were upregulated. Interactions between 1) MAPK10 and JUN, 2) PIK3R2 and PIK3R3 were revealed (Fig. S1). We then selected the "drug resistance" associated genes from KEGG category. Network analysis showed that they belong to one or more groups associated with different types of resistance or response to drugs or stimuli (Fig. 3B). The differential expression of these genes in KB-C2 and KB-C2-k.o.cdk6 are shown in Fig. 3C. Protein-protein interaction (PPI) analysis revealed: 1) the signaling pathways involving CDK4, FOS, PRKACB, brain V-type proton ATPase subunit B2 (ATP6V1B2), mitochondrial ATP synthase subunit e (ATP5ME) and mitochondrial ATP synthase-coupling factor 6 (ATP5PF), and 2) PI3K/HRAS signaling pathways. However, no regulation between these genes and ABCB1 expression was retrieved, either using metascape or STRING platform (Fig. 3D, Fig. S2). Nevertheless, the experimental data showed that a series of endocrine resistance-related signaling factors were upregulated or downregulated in the cdk6-deficient cells. Among them, FOS-promoting cell autophagy [38] was upregulated, whereas PI3K (PIK3CA/PIK3CB), which is a key signaling factor in cancer and was found to regulate ABCB1-mediated MDR in our recent study [2], was downregulated accompanying the remarkable downregulation of ABCB1 (Fig. 4A, B and C).

Cross-downregulation between CDK6 and PI3K 110α/110β suppresses ABCB1 expression in cdk6 deficient cancer cells
Our previous study revealed that P110α and P110β could be potential targets that regulate ABCB1-mediated MDR in cancers [2]. It is of great interest whether cdk6 gene downregulation may cause co-downregulation of PI3K 110α and 110β expression. In so doing, both P110α and P110β expression levels are significantly downregulated with the knockout of cdk6 gene ( Fig. 4A and C, Table  S1). Therefore, downregulation of ABCB1 expression and reversal of ABCB1-mediated MDR in KB-C2-k.o.cdk6 could possibly, at least partially be induced by the downregulation of P110α and P110β resulted from knocking out the cdk6 gene. Furthermore, knockout of cdk6, PIK3CA or PIK3CB in KB-C2 indicated that CDK6 and P110α/P110β could positively regulate each other (Tables S1, S2). This reverse regulation of P110α/P110β on CDK6 was also found in H460/MX80 cells via knockout of PIK3CA or PIK3CB genes (Table S2), and further strengthened the possibility that cross regulation exists between CDK6 and P110α/ P110β, which co-interact and synergize to regulate ABCB1 expression and ABCB1-mediated MDR in KB-C2 and H460/MX80 cells (Fig. S2).
To study the mechanisms of the altered ABCB1 expression induced by CDK6 or CDK4, we analyzed the AS of the pre-mature mRNA of ABCB1 gene, which expresses protein ABCB1. The results demonstrated that in either KB-C2 cell population with cdk6 or cdk4 deficiency, a high frequency of skipped exon (SE) splicing was detected in the ABCB1 pre-mRNA (Table 1). Among the 14 SE events in ABCB1 pre-mRNA of KB-C2-k.o.cdk6 cells, 12 of them were novel chromosomal AS events. Thirteen SE were found in KB-C2-k.o.cdk4 cells, where 10 were novel chromosomal AS. The high frequency of AS events in ABCB1 pre-mRNA in response to cdk6 or cdk4 deficiency generated altered expression level of the ABCB1 protein, further demonstrating that CDK6 and CDK4 may regulate ABCB1 expression by generating responsive AS in ABCB1 pre-mRNA. The knockout of either PIK3CA (encoding P110α) or PIK3CB (encoding P110β) in KB-C2 cells also resulted in a number of novel AS events in ABCB1 pre-mRNA and downregulation of ABCB1 protein expression (Table 1). Interestingly, 10 common SE positions in ABCB1 were found in the four types of gene-deficient cell populations, i.e., KB-C2 cells deficient in cdk6, cdk4, PIK3CA and PIK3CB (  (Table 1). These positions could be responsible for the high frequency of AS in ABCB1 pre-mRNA, leading to changes in ABCB1 expression in these gene-deficient cell populations.
Among the significantly downregulated proteins, KRAS is a GTPase and a proto-oncogene product relaying early signals from outside the cell to the cell nucleus [39]. It is one of the effectors downstream of EGFR activation and triggers intracellular signaling cascades such as MAPK and PI3K [40]. It is interesting that KRAS was significantly downregulated in either CDK6 or P100α/ P110β deficient cell populations. The JUN product, proto-oncoprotein c-JUN, in combination with c-Fos, forms oncogenic transcription factor that interacts with specific target DNA sequence to regulate gene expression and is involved in both translocations and deletions in human malignancies [41]. Mitogen-activated protein kinase 10 (MAPK10), also known as c-Jun N-terminal kinase 3 (JNK3) is a c-Jun N-terminal kinases (JNKs) that may prevent cell apoptosis. MAPK10/JNK3 signaling may induce carcinogenesis [42]. Knocking down MAPK10 suppressed ovarian cancer cell growth and migration [43]. Fibronectin 1 (FN1) is a glycoprotein of the extracellular matrix that binds to integrins (a type of membrane-spanning receptor proteins) [44]. It can suppress apoptosis and promoted viability, invasion, and migration in cancers [45]. SERPINE1 gene in human encodes serpin E1, also known as endothelial plasminogen activator inhibitor or plasminogen activator inhibitor-1 (PAI-1), which has significant functions in the cancer occurrence, relapse and multidrug resistance [46]. Therefore, the knockout of cdk6 gene inducing remarkable downregulation of these genes (Fig. 5) may contribute to the inhibition on the survivability or proliferation of the KB-C2 cells. Among the significantly upregulated proteins, ATP2A2 as one of the P-type pumps important for higher order physiological processes, is an isoform of the SERCA Ca 2+ -ATPases that catalyze the hydrolysis of ATP, maintain nanomolar cytoplasmic Ca 2+ levels, and provide high levels of Ca 2+ in the endoplasmic reticulum, Golgi, and secretory vesicles which store Ca 2+ for a wide range of signaling functions [47,48]. cAMP-dependent protein kinase catalytic subunit beta (PRKACB) is a member of the serine/ threonine protein kinase family and is a catalytic subunit of PKA. By activating the protein kinase A (PKA), which transduces the signal through phosphorylation of variant target proteins, it functions as a signaling molecule important for a variety of cellular activities [49]. As reported, down-regulation of PRKACB is related to the occurrence of various human malignancies [50]. the PI3K regulatory subunits PIK3R2/p85beta and PIK3R3/p55gamma, are frequently overexpressed in various cancers [51,52]. The chemokine (C-X-C motif) ligand 1 (CXCL1) is a small peptide belonging to the CXC chemokine family. It promotes arteriogenesis through enhanced monocyte recruitment into the peri-collateral space and has relationship with tumor progression [53,54]. For these significantly upregulated genes that have central regulation function and are involved in multiple signaling pathways (Fig. 4D), ATP2A2 and PRKACB upregulation may contribute to inhibiting the viability of KB-C2 cells. It is also the first time to unveil the significant upregulation of some malignancy related genes including PIK3R2, PIK3R3 and CXCL1. These results indicated that responsive upregulation or downregulation of a number of genes and signaling pathways occurred that affected multiple cell functions after cdk6 knockout. The mostly changed signaling pathways downregulating ABCB1 expression and inhibiting cancer cell proliferation were concluded in Fig. 5.

Knockout of cdk6 or cdk4 gene promotes apoptosis, but only cdk6 deficiency reverses MDR in KB-C2 cells
In the cdk4-or cdk6-deficient KB-C2 cells cultured with or without colchicine, the proportion of non-apoptotic cells was half declined compared with the original KB-C2 cells (Fig. 6A). The deletion of the cdk6 gene induced transition of more cells from slightly apoptotic status to heavily apoptotic status (Fig. 6). The culture of KB-C2 cells with colchicine (1.25 μM) only produced minor increase of apoptotic cells because KB-C2 cancer cells are highly resistant to colchicine (Fig. 6). However, cdk6 knockout remarkably increased the drug sensitivity and induced significantly promoted transition from moderately apoptotic cells to heavily apoptotic cells (15.6%) in KB-C2-k.o.cdk6 cells incubated with colchicine compared with control cells incubated without colchicine (8.35%) (Fig. 6). Slightly increase of non-apoptotic cells and decrease of moderately apoptotic cells was detected in the drug-treated KB-C2-k.o.cdk4 cell populations (cdk4 deleted KB-C2) compared with control cells (Fig. 6).
The deletion of cdk6 or cdk4 accelerates cell senescence in the G 1 phase, whereas cdk6 deletion remarkably decreased the drug resistance of KB-C2 cells to colchicine Deletion of either cdk4 or cdk6 accelerated senescence in G 1 phase and promoted apoptosis of KB-C2 cells (Fig. 6B, Table S3). However, the deficiency of cdk6 but not cdk4, caused significant inhibition on cell division in G2-M phase in the presence of colchicine compared with nongene-deficient KB-C2 cells. This further demonstrated that the knockout of ckd6 gene but not cdk4 reversed the MDR of KB-C2 via a mechanism of inhibiting cell division. As reported, colchicine binds to microtubules that form the mitotic spindle during cell division, thereby leads to mitotic arrest and ultimately cause cell death [55]. Downregulation of ABCB1 by the deficiency of CDK6 may reduce the efflux of colchicine which results in the increase of colchicine accumulation within the cells and enhanced inhibition on cell division.

Loss of gene cdk6 remarkably increased the sensitivity of KB-C2 tumors to DOX by increasing DOX accumulation inside the tumor cells
DOX remarkably inhibited the cdk6 deficient KB-C2, i.e., KB-C2-k.o.cdk6 tumors. Among the 9 KB-C2-k.o.cdk6 tumors, 6 were completely eliminated, and only 3 with very small sizes were detected ( Fig. 7A and B) by the 19th day after DOX treatment. Tumor enlargement was most rapid in DOX-untreated KB-C2 group, followed by DOXtreated KB-C2 and DOX-untreated KB-C2-k.o.cdk6 groups. This result indicated that loss of gene cdk6 remarkably increased the sensitivity of KB-C2 tumors to DOX. Meanwhile, cdk6 deficiency can inhibit tumor growth to some extent because this gene is important for cancer cell proliferation.
To find out the mechanisms of the enhanced inhibition in KB-C2-k.o.cdk6 tumors, the drug sensitivity to DOX was compared between the cdk6-present and cdk6deficient tumors. All remaining KB-C2-k.o.cdk6 tumors showed strong fluorescence of DOX, while only a small part of the KB-C2 tumors had detectable DOX accumulation (Fig. 7C). This result demonstrated that cdk6 deficiency greatly reversed the drug resistance caused by ABCB1, of which DOX is its substrate, as more amount of DOX was concentrated within the tumor cells where ABCB1 was downregulated by cdk6 deficiency.
With the inhibition of tumor growth, pathological changes including focal infiltration of peri-bronchial inflammatory cells and expansion of the germinal center in spleen white pulp was remarkably reduced in the DOX-treated nude mice bearing KB-C2-k.o.cdk6, as compared with the KB-C2 groups or the DOX-untreated KB-C2-k.o.cdk6 group (Fig. 7D).

Combining knockdown of cdk6 and DOX-chemotherapy inhibited growth and metastasis of KB-C2 tumor in vivo
Gene differential expression analysis and immunofluorescence analysis indicated that KB-C2 cells with small cell nuclei and spindle-shaped morphology showed strong CKAP4-positive signal. High levels of tumor cell metastasis were detected in the liver, lung and spleen of KB-C2 tumor-bearing nude mice, and more tumor cells were distributed inside the organs, as compared with KB-C2-k.o.cdk6 tumor-bearing nude mice (Fig. 8). In KB-C2-bearing nude mice treated with DOX, the tumor cells in the organs were only slightly less than those without DOX treatment, and in the lungs, the tumor cells showed obvious inward invasion and growth. In KB-C2-k.o.cdk6-bearing nude mice with DOX treatment, the tumor cells that had invaded in the interior of the organ were the least (Fig. 8), and for the lungs, they were detected only around one outside and no apparent signals were detected inside the lung tissues. These results demonstrated that combining knockdown of cdk6 and DOX chemotherapy inhibited the metastatic ability

Discussion
MDR of cancers can induce failure of chemotherapy. As one major reason of MDR in cancers, ABC transporters are highly expressed in various MDR cancer cell lines [56,57]. Today, a number of ABC transporter inhibitors have been explored to attenuate the activity of the ABC transporters to pump out chemotherapeutic drugs [2,58]. However, most of the inhibitors may have several targets, some of which have important function in normal cells [31]. Furthermore, the regulatory mechanisms are greatly unknown; as such risks of side-effects may be increased. Seeking safer targets, often based on tumor-specific  proteins that regulate ABC transporters, has become an ideal approach to improve MDR-reversal in cancer therapy. As ABCB1 is a vital ABC transporter for the cancer cells to extrude a variety of anti-cancer drugs [31,59,60], we investigated the targets to reverse ABCB1-mediated MDR based on screening of effective reversal agents and gene deletion technology, and revealed a novel targeting pathway. In KB-C2, CDK6 may induce on-off regulation of ABCB1 expression through coordination with PI3K 110α/β and inducing AS in the ABCB1 pre-mRNA. In a parallel study, similar regulation of ABCB1-mediated MDR by CDK6 was observed in H460/MX80-k.o.cdk6, in which cdk6 was knocked out from the NSCLC H460/ MX80 MDR cells (Fig. S3). Therefore, CDK6 may be potential and secure targets for reversing ABCB1-mediated MDR, since its overexpression has been shown to stimulate cancer cell proliferation [61]. At present, the project focusing on the function of ribociclib in inhibiting the ABCB1 mediated-MDR in cancers are undergoing in our laboratory. Because the mechanisms for ribociclib to downregulate ABCB1 and inhibit P-gp mediated MDR in cancers could be associated with many aspects, we are currently performing experiments, including real-time PCR of ABCB1 in ribociclib treated cells, or transcriptome sequencing and quantification, in addition to Western blot and immunofluorescence which we have done, to explore the ABCB1 expression at the transcription level. We will also try to explore other potential pathway(s) through the participation of which, ribociclib might downregulate ABCB1 expression. In addition, our unpublished data showed that the structure of ABCB1 could be changed by ribociclib at effective MDR-reversal concentration (e.g., 9 μM) under physiology condition, causing inhibition of the ATPase and drug-efflux activity of ABCB1, indicating that ABCB1 was less stable through interaction with ribociclib. Due to the intricacy of the function of ribociclib within the MDR cancer cells, specific gene deletion was performed to analyze whether CDK6 could be an effective target for overcoming ABCB1-mediated MDR in cancer cells.
We recently realized that the knockout of mutant PI3K 110α or 110β achieved inhibition of ABCB1 expression as well as reversal of MDR mediated by ABCB1 [2]. We further found in this study that downregulation of CDK6 expression through gene knockout downregulated the expression of PI3K P110α and P110β subunits, and vice versa. This cross regulation between CDK6 and P110α/ P110β could explain the remarkable decrease in ABCB1 expression and MDR in cdk6 or P110α/P110β deleted KB-C2 cells.
Upon further analysis of the regulatory mechanism of ABCB1 expression by CDK6, CDK4 and PI3K 110 α/β, we surprisingly found that the TPM levels of the ABCB1 transcripts expressing the ABCB1 protein and truncated ABCB1 peptides varied significantly between the gene-deficient KB-C2 cell populations and non-genedeficient KB-C2 cells. In addition, we found that a high frequency of SE splicing events had occurred during AS of the ABCB1 pre-mRNA that expresses ABCB1. This could explain the markedly reduced ABCB1 level in KB-C2-k.o.cdk6 cells and enhanced expression of ABCB1 or its isoform in KB-C2-k.o.cdk4 cells, which resulted in different efficacy on ABCB1-mediated MDR. Interestingly, a knockout experiment indicated that cdk6 was downregulated along with the knockout of cdk4 in KB-C2 cells. To our knowledge, it is the first time that the responsive AS occurrence in the pre-mRNA of ABCB1 was found to be induced by the deletion of cdk6, cdk4 or PI3K 110α/β subunits. This will surely have significance in reversing ABCB1-mediated MDR by modulating cdk6 or cdk4 gene expression.
CDK6 has been suggested to be a redundant homolog of CDK4 in the past [62]. For the first time, our study demonstrated that CDK6 and CDK4 may be two homologous proteins that differentially regulate ABCB1 expression in MDR KB-C2 cells. Indeed, our results indicated that the knockout of cdk6 reversed ABCB1-mediated MDR in KB-C2 cells (Fig. 2, Fig. 6). In this study, differential gene expression revealed that, along with the deletion of CDK6, a series of signaling pathways altered, for example, JUN/MAPK10 which is crucial in regulating tumorigenesis [42,63], NOTCH3 which is involved in carcinoma and immune tolerance [64], were downregulated in KB-C2-k.o.cdk6 cell population, whereas CYP2D6 which was responsible for the metabolism of many drugs [65], was upregulated. Combining our findings that cdk6-deficient cancer cells lost ABCB1mediated MDR via co-downregulating PI3K 110α/β, we deem that cdk6 could be a promising target for reversing ABCB1-mediated MDR and inhibiting cancer cell proliferation simultaneously. It may be hopeful to develop strategies based on targeting CDK6 for combined chemotherapy of MDR cancers with improved accuracy. In this aspect, modernized functional biomaterials can be integrated for further enhancing the targeting inhibition on CDK6 in cancer cells.
In summary, we demonstrated that regulation of ABCB1 can be realized through targeting CDK6-PI3K axis. Knockout of the cdk6 gene unveiled a novel targeting pathway for overcoming ABCB1-mediated MDR in cancers. PI3K 110α/β signaling was downregulated in the cdk6-deficient KB-C2 cell populations. A high frequency of responsive AS, mainly SE events, were induced by CDK6, CDK4, and PI3K 110α/β level changes, and these AS events finally led to the alteration of the ABCB1 expression and ABCB1-mediated MDR in cdk6 or cdk4 deficient KB-C2 cancer cells. Cross-regulation between CDK6 and PI3K110α/110β was discovered, and knockout of either one of these two genes could induce downregulation of the other and remarkable inhibition of ABCB1 expression, implying that they may synergize with each other to regulate ABCB1 expression. This will greatly strengthen the efficiency during reversing ABCB1 when using either of them as target.
With the knockout of cdk6, quite a few of signaling factors leading to inhibition of cancer cell proliferation were found to be promoted. For example, FOS-promoting cell autophagy [38] was upregulated. It is possible that these genes are quite variable in expression to adapt to CDK6 deficiency.
In the Western blot analysis, the positive signal band showed certain peptide with a reduced molecular weight in the cdk4 or cdk6 deficient cell populations. This should be induced by DNA repairment during CRISPR/Cas9 gene editing, that is, truncated proteins with similarity to wild-type CDK4 or CDK6 was produced after chromosomal repairment. To confirm this, we performed RT-PCR and certified deficiency of the objective region in cdk4 or cdk6 genes on chromosomes, and the results demonstrated knockout of wild-type cdk4 or cdk6 genes.
In vivo experiment demonstrated the function of cdk6 in maintaining ABCB1-mediated drug resistance in cancers. Since cdk6 deficiency or PIK3CA/PIK3CB deletion downregulated each other, and PIK3CA/PIK3CB knockout also downregulated ABCB1, we deem that CDK6-PI3K signaling axis is an efficient target for attenuating ABCB1-mediated MDR. To some degree, cdk6 deficiency inhibited KB-C2 tumor growth, which showed similar efficacy with DOX treated KB-C2 tumors. However, the efficacy was far less than the DOX-treated KB-C2-k.o.cdk6 tumor group. ABCB1-mediated drug resistance appeared to be a dominant barrier for inhibition on KB-C2 tumor.
Most known cancer cells have increased PIK3CA expression or more active PIK3CA which may lead to enhanced cell survivability, proliferative or anti-apoptosis ability, and overexpression of ABCB1 as we reported [2]. therefore, realizing downregulation of PIK3CA and ABCB1 by knocking out CDK6 in MDR cancer cells overexpressing ABCB1 might benefit combined chemotherapy against cancer cells with ABCB1-mediated MDR.
Through transcriptome analysis in this study, we found that protein CKAP4 is expressed at especially high level in both KB-C2 and KB-C2-k.o.cdk6 cells. It was reported that CKAP4 was minimally expressed in non-tumor cells [66], therefore, it was adopted as a biomarker for investigation of KB-C2 and KB-C2-k.o.cdk6 metastasis in the tissues of nude mice. We performed transcriptome sequencing and quantification to test the protein expression at the transcription level. The results reported that CDK6 expression in MDR KB-C2 cells (over expressing ABCB1, Fig. S4) was elevated to approximately 7 folds (Fig. S5) compared to drug sensitive parental KB-3-1 cells which did not express detectable amount of ABCB1 (Fig. S4). This provided a clue for clarifying the regulatory effect of CDK6 on ABCB1 by specific gene knockout experiment. By knocking down cdk6 in KB-C2 cell population, metastasis and survival of KB-C2-k.o.cdk6 tumor cells were remarkably inhibited in the DOX-treated nude mice, mainly because the drug resistance of KB-C2-k.o.cdk6 was inhibited compared with KB-C2.
Although a combination of prospective and retrospective cohort study involving 131 cases of advanced stage invasive breast cancer (which have received neoadjuvant chemotherapy) indicated that the expression of ABCB1 had no significant statistical correlation to metastases (p = 0.659) [67], whether ABCB1 has any effect on tumor cell migration, which has not been clearly studied, is an undergoing project we are studying.
Based on these findings, combined targeted chemotherapy with improved efficacy can be realized through downregulating CDK6-PI3K signaling axis in cancers overexpressing ABCB1, which is a common ABC member overexpressed in most malignant tumors and functions to generate drug resistance by efflux of its substance drugs.

Conclusion
In conclusion, we deem that CDK6-PI3K axis can be ideal target for reversing ABCB1 mediated MDR in cancers without inducing cancer cell proliferation, and this finding is of great significance for the combined anticancer chemotherapy which can reverse multidrug resistance and inhibit the growth of tumor cells at the same time. This study revealed the cross downregulation between CDK6 and PI3K in cdk6 deficient cancer cell populations, demonstrated that CDK6-PI3K axis could be new target for inhibiting ABCB1-mediated MDR. Meanwhile, knockout of cdk6 could inhibit cancer cell proliferation and malignancy. These new findings will surely benefit exploration of new drugs targeting cdk6 or PIK3CA/ PIK3CB genes or gene products through which the therapeutic effect could be optimized.