CHEK1 and circCHEK1_246aa evoke chromosomal instability and induce bone lesion formation in multiple myeloma

Gene expression profiling (GEP) cohorts were collected using the GEO database as described previously [22, 23]. The Total therapy 2 (TT2) and TT3 patient cohorts, the Dutch-Belgian Cooperative Trial Group for Hematology Oncology Group-65 (HOVON65) trial (GSE19784) patient cohort, and the Assessment of Proteasome Inhibition for Extending Remission (APEX) patient cohort (GSE9782) were included in analyses, which used publicly available gene expression profile data for each of these patient cohorts [3].

Antibodies used were as follows: CHEK1 (sc-8408; Santa Cruz Biotechnology, USA); rabbit IgG (a7016); mouse IgG (a7028; Beyotime Institute of Biotechnology, China); FLAG (F-4020; Merck KGaA, Germany); PARP (9542S), Caspase-3 (9662S), β-actin (4970S; Cell Signaling Technology, USA); MYC (16286–1-AP), CEP170 (18899–1-AP; ProteinTech Group, China); and α-Tubulin (ab7291; Abcam, UK).

Doxycycline (DOX) was purchased from the Beyotime Institute of Biotechnology. Puromycin was purchased from Merck KGaA. Bortezomib (BTZ), Adriamycin (ADR), dexamethasone (DEX), LY2603618, and other reagents were purchased from Selleck Chemicals (Houston, TX). The rapid Giemsa staining kit was obtained from BBI Life Sciences (Shanghai, China).

Human MM cell lines, including the BTZ-resistant cell lines ARP1, H929, ANBL6 wild-type (WT) and ANBL6/BTZ-resistant, and the DEX-resistant cell lines MM1S and MM1R, were cultured in RPMI-1640 (Biological Industries, Israel). HEK-293 cells were cultured in DMEM (Thermo Fisher Scientific, USA). All media were supplemented with 10% fetal bovine serum (Gibco, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin (Sigma Aldrich, Germany). All cells were cultured at 37 °C in 5% CO₂.

Plasmids containing human CHEK1 cDNA and CHEK1 shRNA cassettes were purchased from Generay Biotech Co., China. The construct number of CHEK1 shRNA that used in the functional assay was 1168–2. The CHEK1-coding sequence was cloned into a BTZ-resistant flag-tagged lentiviral vector, CD513B-1. CHEK1-targeting shRNA under the control of a DOX-inducible promoter was cloned into the pTRIPZ vector. Lentiviruses were produced by co-transfection of the expression vector of interest with the packaging plasmids PLP1, PLP2, and VSVG into HEK293 cells using Lipofectamine™2000 Transfection Reagent (Invitrogen, USA). Virus supernatant was collected after 48 h. Transfected MM cells were selected by puromycin resistance. Transduction efficiency was determined by western blotting (WB).

This study was conducted in accordance with the Government-published recommendations for the Care and Use of Laboratory Animals, and were approved by the Institutional Ethics Review Boards of Nanjing University of Chinese Medicine (Ethics Registration no. 201905A003).

WT and CHEK1-overexpressing cells (1 × 10⁶) were injected subcutaneously into the left and right abdominal flanks, respectively, of 6–8-week-old SCID/NOD mice, which were treated with intraperitoneal (IP) administrations of BTZ (1 mg/kg) or ADR (1 mg/kg) twice weekly.

WT and CHEK1 knockdown (KD) cells (5 × 10⁶) were injected subcutaneously on the flanks of 6–8-week-old SCID/NOD mice. On day 3 after MM cell transfer, DOX (2 mg/mL) was added to the drinking water to induce CHEK1 shRNA expression.

Tumor diameter was measured 2–3 times weekly using calipers. Once the tumor diameter reached 20 mm, mice were sacrificed, and tumor tissues were collected, weighed, and photographed.

Cell proliferation rate and viability were detected using a trypan blue exclusion assay, and counted using a hemocytometer.

For colony formation assays, clonogenic growth was determined by plating 1 × 10⁴ cells in 0.5 mL of 0.33% agar/RPMI 1640 supplemented with 10% FBS. Medium was replaced twice weekly, and cells were cultured for around 14 days. Clusters of cells were considered to be a clonogenic colony if > 40 cells were present. Colonies were imaged, and colony numbers were counted in blinded images using ImageJ.

For cell cycle assays, samples were washed with PBS and treated with propidium iodide (PI) solution (Yeasen, China) for 30 min. Samples were analyzed using flow cytometry (Merck Millipore, Germany).

WB was performed as previously described [24]. Co-IP was conducted using a Pierce Direct Magnetic IP/Co-IP kit (Thermo Scientific) per the manufacturer’s instructions.

Cells were fixed with 4% paraformaldehyde, permeabilized with PBS containing 0.1% Triton X-100, quenched with 50 mM NH₄Cl (xx min), and blocked with 1% BSA. After overnight incubation with primary antibodies at 4 °C, slides were incubated with corresponding secondary antibodies. Images were captured using a confocal microscope (TCS SP8; Leica, Germany).

SDS-PAGE was used to separate proteins, and gel bands at the expected size were excised and digested with sequencing-grade trypsin (Promega, USA). The resulting peptides were analyzed using a QExactive mass spectrometer (Thermo Fisher Scientific). Fragment spectra were analyzed according to the National Center for Biotechnology Information nonredundant protein database.

We first examined CHEK1 expression in MM GEP cohorts. Intriguingly, CHECK1 mRNA was significantly increased in MM cells compared with normal plasma (NP) cells and monoclonal gammopathy of undetermined significance (MGUS) cells (Fig. 1A). Further, higher CHEK1 expression was associated with poor outcome in the TT2 (Fig. 1B), HOVON65 (Fig. 1C), and GMMG-HD (Figure S1) patient cohorts, which included over 1200 MM patients. Taken together, these findings suggested that increased CHEK1 expression was associated with poor MM outcome [3, 5].

The protein level of CHEK1 endogenously expressed in commonly used MM cell lines was measured by WB (Fig. 1D), revealing that all cell lines tested expressed CHEK1. To further determine if CHEK1 was a contributing factor to MM rather than an artifact of other oncogenes, CHEK1 was overexpressed (OE) in MM cells using a lentiviral system, which was validated by WB (Fig. 1E). Interestingly, proliferation was increased in CHEK1-OE cells relative to WT in both ARP1 and H929 cells, as demonstrated by a trypan blue dye exclusion assay (Fig. 1F), suggesting that CHEK1 promoted MM proliferation. Conversely, CHEK1 was knocked down (KD) by three distinct CHEK1-targeting shRNAs, which were all validated by WB in both ARP1 and H929 cells (Fig. 1G). Cell proliferation was decreased by CHEK1 KD in both ARP1 and H929 cells (Fig. 1H). Moreover, a clonal formation assay revealed that CHEK1 OE increased clonal formation, while CHEK1 KD inhibited clonal formation in both ARP1 and H929 cells (Fig. 1I). Consistently, flow cytometric cell cycle analysis demonstrated that in CHEK1-OE ARP1 and H929 cells, an increased proportion of cells were in the G2/M phase relative to WT cells (Fig. 1J), with a decreased proportion of G2/M phase cells with CHEK1 KD in both cell lines (Fig. 1K). Taken together, these findings suggested that CHEK1 promoted MM proliferation and clonal expansion.

We further employed RNA-sequencing (RNA-seq) to assess activation of CHEK1-related signaling pathways, revealing activation of two pathways related to CHEK1 and MM progression, cell cycle regulation and osteoclast differentiation (Fig. 2A–B). Because high-risk MM is characterized by aggressive proliferation, we measured CHEK1 mRNA expression in MM subgroups, and found that CHEK1 expression was highest in the PR subgroup, considered the highest-risk MM subgroup (Fig. 2C) [22]. MM patients in the PR group are characterized by high MM proliferation rate and poor clinical outcomes, and increased CHEK1 mRNA levels in this subgroup suggested that CHEK1 could be a biomarker for high-risk MM. Furthermore, CHEK1 expression was increased in MM relapse samples relative to first-diagnosis MM samples in 88 paired patient samples (Fig. 2D). In patients who experienced relapse, increased CHEK1 expression was significantly associated with decreased overall survival (OS) relative to patients with lower CHEK1 expression in two independent cohorts, the TT2 (Fig. 2E) and APEX (Fig. 2F) cohorts [25].

Because high-risk MM is generally associated with drug resistance, we measured CHEK1 expression in two pairs of drug-susceptible and -resistant cell lines, the MM1.S and MM1.R lines, which are susceptible and resistant to dexamethasone, respectively, and ANBL6 WT and BTZ-resistant cells. WB analysis revealed that CHEK1 protein levels were increased in both drug-resistant cell lines compared with paired drug-susceptible controls, suggesting an association between CHEK1 upregulation and multiple drug resistance (Fig. 2G).

To determine if CHEK1 induced drug resistance, we performed MTT and WB assays on CHEK1 WT and OE cells to measure the IC_50s of ADR and BTZ, as well as the protein levels of apoptotic markers in drug-treated cells. The IC_50s for both ADR and BTZ were significantly higher in CHEK1-OE cells relative to WT cells (Fig. 2H), while cleavage of the apoptotic markers PARP and Caspase 3 was decreased in drug-treated CHEK1-OE cells relative to WT (Fig. 2I). Contrastingly, treatment with either CHEK1 shRNA (Fig. 2J) or the selective CHEK1 inhibitor LY2603618 (Fig. 2K) increased apoptosis in ADR- and BTZ-treated cells. Taken together, these findings suggested that CHEK1 was a high-risk MM marker associated with relapse and drug resistance in MM patients, and induced drug resistance in cultured MM cells.

We next sought to investigate the mechanisms by which CHEK1 promoted MM proliferation, malignancy, and drug resistance. Our prior study reported that CHEK1 was included in the chromosomal instability gene list for cancer cells [26, 27]. We therefore explored whether CHEK1 prompted MM CIN, resulting in MM proliferation and drug resistance.

Gimsa staining revealed that CHEK1 OE increased the separation error rate and numbers of multiple nuclear cells, two key features of CIN [28, 29], in ARP1 and H929 cells (Fig. 3A–B). Immunofluorescent (IF) staining for α-Tubulin and DAPI was next used to further evaluate the extent of CHEK1-induced CIN. In both cell lines, CHEK1 OE increased chromosomal plate width and decreased mitotic bipolar spindle length, two additional indicators of CIN in MM cells [28, 30‐32] (Fig. 3C–D).

We subsequently performed a comparative genomic hybridization (CGH) array to directly assess the effect of CHEK1 on MM chromosomal composition [26], which identified significant gains and losses of multiple chromosomal segments in CHEK1-OE ARP1 and H929 cells relative to the corresponding WT cells (Fig. 3E). Taken together, these data suggested that increased CHEK1 expression promoted CIN in MM cells.

CIN contributes to drug resistance in multiple types of cancer. We therefore determined if CHEK1 OE could overcome BTZ sensitivity by inducing CIN. CHEK1-OE ARP1 and H929 cells were resistant to BTZ treatment (Fig. 2H–I). IF staining for α-Tubulin and DAPI revealed that chromosomal plate width increased and mitotic spindle length decreased in BTZ-treated CHEK1-OE ARP1 and H929 cells relative to both vehicle-treated CHEK1-OE cells and WT cells, suggesting that CHEK1-induced CIN was an important contributor to MM drug resistance (Fig. 3F).

To further determine how CHEK1 promoted MM CIN, we performed a Co-IP assay followed by mass spectrometry (MS) to determine which proteins interacted with CHEK1. Hundreds of proteins were identified by MS, and these candidate proteins were screened against the CIN-related gene list, and genes associated with poor outcome in the TT2 cohort (Fig. 4A). Centrosomal Protein 170 (CEP170) was identified as a candidate CIN gene that could potentially interact with CHEK1 (Fig. 4A–B), and high expression of CEP170 mRNA was significantly correlated with decreased OS in the TT2 MM cohort (Fig. 4C). Physical interaction between CHEK1 and CEP170 was identified using a Co-IP assay in CHEK1-OE ARP1 and H929 cells (Fig. 4D). CEP170 is a centrosomal component, and is required for centriole appendage assembly [33]. IF staining revealed that CEP170 OE significantly increased chromosomal plate width and decreased mitotic bipolar spindle length in ARP1 and H929 MM cells, suggesting that CEP170 evoked MM CIN (Fig. 4E–F).

Our findings suggested that CHEK1 induced MM CIN by directly interacting with CEP170. CHEK1 belongs to the kinase family, and we hypothesized that CHEK1 could phosphorylate CEP170. Consistent with this notion, a Co-IP assay revealed that the phosphorylated form of CEP170, as detected by an anti-phospho-serine antibody, was increased in CHEK1-OE cells relative to WT cells in both cell lines (Fig. 4G). The CEP170 CHEK1 phosphorylation site was Ser1260, as identified by Thermo Q-Exactive (MS) (Fig. 4H). To further confirm that CEP170 Ser1260 was the CHEK1 phosphorylation site, we mutated Ser1260 to Ser1260Ala. A Co-IP assay confirmed that the interaction between mutant Ser1260Ala CEP170 and CHEK1 protein, linked with Myc and Flag, respectively, was attenuated dramatically in CHEK1-OE cells compared with WT cells (Fig. 4I). Further, Ser1260Ala mutant CEP170 OE decreased CIN markers, as indicated by decreased chromosomal plate width and increased mitotic spindle length (Fig. 4J–K). Collectively, these data demonstrated that CHEK1 induced MM CIN by phosphorylating CEP170 at the Ser1260 site.

Because RNA-seq analysis revealed that CHEK1 expression was correlated with osteoclast differentiation (Fig. 2A–B), we evaluated MRI data from MM patients of the TT2 cohort and found that CHEK1 expression was higher in MM patients with bone lesions than in MM patients without bone lesions, as detected by MRI (Fig. 5A). To evaluate the potential mechanism for CHEK1-promoted bone lesion formation, we overexpressed murine Chek1 cDNA in cultured murine RAW264.7 macrophages. Tartrate-resistant acid phosphatase (TRAP) staining revealed that increased Chek1 expression promoted osteoclast differentiation in macrophages treated with RANKL (50 ng/mL) or M-CSF (15 ng/mL) for 10 days (Fig. 5B–C). When the concentrations of RANKL and M-CSF were decreased, exogenous m-Chek1 cDNA expression was still able to prompt osteoclast differentiation in a RANKL and M-CSF dose-dependent manner (Fig. 5D–E), indicating that CHEK1 was an important activator of osteoclast differentiation. This finding was verified in human primary peripheral blood mononuclear cells (PBMCs). PBMCs transfected with human CHEK1 cDNA developed significantly more osteoclasts than vehicle-transfected control cells (Fig. 5F–G). Inversely, the CHEK1 inhibitor LY2603618 prevented RAW264.7 cells from differentiating into osteoclasts in a dose-dependent manner, and decreased expression of NFATc1, which is the key factor for osteoclast differentiation (Fig. 5H–I). We then performed a Co-IP assay in m-Chek1-OE RAW264.7 cells to determine if CHEK1 directly interacted with NFATc1 (Fig. 5J). Further, the expression of NFATc1 was increased in m-Chek1-OE RAW264.7 cells relative to WT cells (Fig. 5K) indicating CHEK1 promotes osteoclasts formation through upregulating NFATc1 expression. 5TMM3VT model eventually confirmed this in vivo and demonstrated that 5TMM3VT-KD cells induced less bone damage characterized by increased bone volume and trabecular numbers (data not shown) compared to the control group by microCT (Fig. 5L).

To explore how MM cells disrupted cells of the normal BM microenvironment, genomic structure analysis was performed, revealing the presence of a secreted circCHEK1 circular RNA fragment (738 bp) containing six exons (Supplementary Figure 2). Use of a divergent primer in cDNA samples and Sanger sequencing confirmed that back-splicing occurred in the CHEK1 exons (Fig. 6A). We then designed convergent and divergent primers to detect linear mRNA and circular RNA, respectively. RNase R treatment significantly diminished linear CHEK1 mRNA, while circCHEK1 was resistant to RNase R digestion (Fig. 6B), indicating that circCHEK1 was more stable than its linear counterpart.

Emerging studies have identified the presence of circRNAs with protein-coding capacity [34]; we therefore analyzed the putative open reading frame of circCHEK1. Bioinformatics analysis revealed that circCHEK1 contained a putative internal ribosome entry site (IRES) sequence that encoded a novel CHEK1 isoform with 246 amino acids, termed “circCHEK1_246aa” in the present study. The predicted size of this isoform was 28.1 kDa, so we adopted the mass spectrometer to confirm the presence of this novel isoform in MM cells. We first used a CHEK1 antibody that specifically recognizes the CHEK1 N-terminus to conduct a Co-IP experiment that enriched CHEK1 protein isoforms containing the N-terminus sequence. WB analysis confirmed that the CHEK1 antibody recognized circCHEK1_246aa at the expected size (Fig. 6C). The enriched protein was separated by SDS-PAGE, excised from the gel, and subjected to MS to detect circCHEK1_246aa. The specific peptide fragments from circCHEK1_246aa were successfully identified by MS analysis, as marked in yellow (Fig. 6C), confirming the expression of circCHEK1_246aa in MM cells. To further examine the kinase function of circCHEK1_246aa, a Co-IP assay was conducted, revealing that circCHEK1_246aa strongly interacted with native CEP170, which was significantly diminished in cells expressing mutant CEP170 (Fig. 6D). In addition, circChek1_246aa expression induced features of CIN in MM cells (Fig. 6E–F), and promoted osteoclast differentiation in PBMCs (Fig. 6G). Together, these findings indicated that the newly identified circular RNA circCHEK1_246aa exacerbated MM by evoking CIN and inducing bone lesion formation (Fig. 6H).

We next evaluated the effect of CHEK1 on MM progression and dug resistance in vivo. ARP1 CHEK1 WT or OE cells were injected subcutaneously into the right or left flanks of NOD-SCID mice, respectively. Mice were then divided into three groups (n = 8 mice/group), including untreated control, ADR-treated, and BTZ-treated. After 28 days, we visually observed that tumors derived from CHEK1-OE cells grew faster than tumors derived from WT cells (Fig. 7A & C), with significantly increased tumor volume and weight (Fig. 7B & D). In addition, tumors derived from CHEK1-OE cells were resistant to both ADR and BTZ, whereas WT cells were sensitive to the treatment (Fig. 7A–D), suggesting that CHEK1 induced MM drug resistance in vivo. Conversely, targeting CHEK1 by doxycycline-inducible h-CHEK1 shRNA significantly inhibited tumor growth when NOD-SCID mice were administered doxycycline through drinking water to induce h-CHEK1 shRNA expression (Fig. 7E–H). Collectively, these data suggested that targeting CHEK1 had therapeutic effects in an in vivo MM murine xenograft model.