Core Molecular Clock Factors Regulate Osteosarcoma Stem Cell Survival and Behavior via CSC/EMT Pathways and Lipid Droplet Biogenesis

The human osteosarcoma cell lines U-2OS, 143B, Saos-2, and MNNG were procured from the American Type Culture Collection (ATCC) and cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Life Technologies, NY, USA). The medium was supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, MO, USA) and 1% penicillin–streptomycin (Gibco, Life Technologies). Cultured OS cells were maintained at 37 °C in a humidified incubator with 5% CO₂ to ensure optimal growth conditions. For gene silencing studies, Lipofectamine RNAiMAX (Thermo Fisher Scientific, Waltham, MA, USA) was used according to the manufacturer’s protocols to transfect the cells with siRNA.

To knock down human CLOCK (SI00069790), BMAL1 (GS406), CRY1 (SI02757370), CRY2 (GS1408), PER1 (SI03076752), PER2 (SI03109687), NR1D1 (SI00130935), and NR1D2 (SI03051965), specific siRNAs were purchased from Qiagen. Transfections of U-2OS cells were performed using 100 pmol of siRNA per well and Lipofectamine RNAiMAX Transfection Reagent (13778075, Thermo Fisher Scientific, Waltham, MA, USA), following the manufacturer’s protocol.

All animal procedures were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the WSU Institutional Animal Care and Use Committee (IACUC; ASAF# 6895). To isolate stem cells from orthotopically induced human osteosarcoma tumor xenografts, the CD133 MicroBead Kit (130-100-857, Miltenyi Biotech, Bergisch Gladbach, Germany) and the MiniMACS™ Kit (130-090-312, Miltenyi Biotech) were utilized, according to the manufacturer’s instructions. Briefly, for each sample, solid OS tumor tissue was harvested from a xenograft generated by the intratibial injection of 143B cells into 6∼8-week-old BALB/c nude immunodeficient mice (CAnN.Cg-Foxn1nu/Crl, Charles River, Wilmington, MA, USA), as described in our previous study [33]. The tissue was mechanically dissociated into a single-cell suspension using a sterile scalpel, followed by enzymatic digestion with 200 U/mL Collagenase type IV (17104019, Thermo Fisher Scientific, Waltham, MA, USA) and 0.6 U/mL Dispase (D4818, Sigma, St. Louis, MO, USA) to facilitate cell release. The resulting cell suspension was then filtered through a 70 µm cell strainer to remove debris and clumps. Cells were then washed with PBS and resuspended in CD133 MicroBead buffer before being incubated with CD133 MicroBeads for 15–30 min to selectively label CD133-positive stem cells. Labeled cells were then separated using the MiniMACS™ Separator. This involved applying the cell suspension to a mini-column placed in the magnetic field of the separator, resulting in CD133-positive cells being retained on the column and unlabeled cells being washed away. Finally, the CD133-positive stem cells were eluted from the column and used for downstream analyses, such as immunofluorescence staining with anti-CD133 antibodies or 3D spheroid formation, as depicted in the figures.

To establish 3D tumor spheroid cultures, osteosarcoma cell lines (U-2OS, 143B, Saos-2, and MNNG) were seeded at a density of 10,000 cells per well in ultra-low-attachment 6-well plates (3471, Corning, NY, NY, USA). The cells were cultured in 3D Tumorsphere Medium XF (PromoCell GmbH, Heidelberg, Germany), a specialized serum-free and xeno-free medium optimized for cancer stem cell (CSC) enrichment and long-term 3D culture. To prevent excessive aggregation, the culture medium was supplemented with 0.5% methylcellulose. The spheroid cultures were maintained in a humidified incubator at 37 °C with 5% CO₂ for one week, after which spheroid formation was assessed via live-cell imaging using the ZOE Fluorescent Cell Imager (1450031, BioRad, Irvine, CA, USA). For siRNA-based experiments, 143B CSCs were initially seeded at a density of 50,000 cells per well in a 24-well plate (353504, Corning, New York, NY, USA). After 24 h, the cells were transfected with siRNA for 48 h according to the experimental design. Following transfection, the culture medium was collected, and adherent cells were detached using 0.25% trypsin-EDTA. The detached cells were combined with the collected medium, centrifuged, and resuspended in fresh culture medium before being reseeded at 10,000 cells per well in ultra-low-attachment 6-well plates. Tumor spheroids were then cultured for an additional seven days, and their morphology was analyzed using fluorescence imaging.

For immunofluorescence (IF) analysis, 143B CSCs and non-CSCs isolated from in vivo OS xenografts, as described above, were plated in a 24-well or 98-well plate. The next day, the cells were fixed with 4% paraformaldehyde (PFA, P6148, Sigma-Aldrich, St. Louis, MO, USA) in PBS and incubated with an anti-CD133 antibody (66666-1-Ig, Proteintech, Rosemont, IL, USA) or anti-Ki-67 antibody (9129, cell signaling, Danvers, MA, USA). Following subsequent incubation with a secondary antibody conjugated to Alexa Fluor 568 (A20003, Thermo Fisher Scientific, Waltham, MA, USA) and Hoechst nucleic acid staining dye (62249, Thermo Fisher Scientific, Waltham, MA, USA), the cells were visualized using a Cytation 5 imaging system.

A total of 1.5 × 10⁵pPer2 reporter 143B CSCs and non-CSCs were seeded onto each 35 mm dish. After exposing the cells to chronic 100 nM dexamethasone (dex; D2915, Sigma, St. Louis, St. Louis, MO, USA), they were treated with dex, and real-time bioluminescence of the control and jet-lag cells was monitored using a LumiCycle luminometer (Actimetrics, Wilmette, IL, USA), as previously described [21]. The period and amplitude of the luminescence data were determined using the LumiCycle Data Analysis software program (Actimetrics, Wilmette, IL, USA).

Apoptotic and necrotic cell death was evaluated using the Annexin V-Cy3 Apoptosis Staining/Detection Kit with SYTOX (ab14144, Abcam, Cambridge, MA, USA) according to the manufacturer’s one-step staining protocol. Cells were seeded at a density of 2 × 10⁴ cells per well in 96-well tissue culture-treated plates with complete culture medium. After a 24 h incubation period, cells were transfected with either control or clock gene siRNA for 48 h. Following transfection, the medium was removed, and cells were incubated with the fluorescent dyes for 30 min. Labeled cells were then analyzed using a BioTek Cytation 5 Cell Imaging Multimode Reader (Agilent Technologies, Santa Clara, CA, USA), employing the FITC filter channel (Ex. 488 nm/Em. 530 nm) for SYTOX green dye and the rhodamine filter channel (Ex. 543 nm/Em. 570 nm) for Annexin V-Cy3.

For the migration assays, 143B CSCs were seeded at a density of 50,000 cells per well of a 96-well plate and transfected with siRNA. After 48 h, a scratch was created in the cell monolayer using a 20 µL pipette tip. Live-cell imaging was conducted over the following 48 h, with representative images captured at 0, 12, and 24 h. Wound closure was quantified using ImageJ (1.53C) software, as described in previous studies [45,46]. For the invasion assays, siRNA-transfected cells were seeded 48 h post-transfection onto 24-well plates fitted with transwell inserts (353097, Corning, NY, USA). To assess the effect of oleic acid (OA) (90260, Cayman Chemical, Ann Arbor, MI, USA) treatment, siRNA-treated cells were exposed to 60 µM OA in DMEM, while control wells received 150 µM BSA-containing DMEM, followed by incubation at 37 °C and 5% CO₂ for an additional 24 h, as described in [47], before seeding the cells into transwell inserts. After invasion, cells on the underside of the inserts were stained with 0.5% crystal violet solution (405831000, Thermo Fisher Scientific). Brightfield images were acquired using a Cytation 5 imaging system, and the number of invading cells per field was quantified using ImageJ, following previously published protocols [45,46].

To analyze lipid droplets (LDs), 143B CSCs were plated at 15,000 cells per well on a high-content imaging 96-well plate (4517, Corning, NY, USA) and transfected with control siRNA (si-CTL) or core clock gene targeting siRNA (si-BMAL1, si-CLOCK, si-CRY1, si-CRY2, si-CRY1/2), as specified in the figures. After 48 h, the cells were fixed with 4% PFA (P6148, Sigma-Aldrich, St. Louis, MO, USA) and stained with Nile Red (1 μg/mL, 30787, Cayman Chemical Company, Ann Arbor, MI, USA) to visualize LDs. Nuclei were counterstained with Hoechst. Representative images were captured using the red fluorescence filter cube (Ex 531 nm/Em 593 nm) for LDs and the DAPI filter cube (Ex. 377 nm/Em 447 nm) for nuclei on a Cytation 5 multi-mode reader. The number of LDs per cell was quantified from the images using ImageJ (1.53C) software.

Total RNA was extracted from 143B CSCs using the RNeasy Plus Mini Kit (74134, Qiagen, Germantown, MD, USA), following the manufacturer’s instructions. Complementary DNA (cDNA) was synthesized from equal amounts of RNA using the Invitrogen Superscript II RT First-Strand Synthesis System, with 2 μL of random hexamers and 9 μL of RNA, as per the manufacturer’s protocol (Life Technologies, Carlsbad, CA, USA). Quantitative PCR (qPCR) was carried out using SYBR Green PCR Master Mix (1708880, BioRad, Hercules, CA, USA) and 10 μM of forward and reverse primers (Appendix Table A1, Table A2 and Table A3). Reactions were cycled as follows: 50 °C for 2 min, 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Reaction specificity was confirmed by melting curve analysis. Gene expression levels were quantified using the comparative Ct method, with normalization to GAPDH as an internal control. All experiments were performed on a ViiA7 Real-Time PCR machine (Thermo Fisher Scientific, Waltham, MA, USA).

To assess the differential expression of the circadian genes BMAL1, CLOCK, CRY1, and CRY2 in osteosarcoma (OS) tissues, publicly available RNA sequencing data (GSE99671↗) from 18 OS patients [48] were analyzed. The dataset consisted of total RNA extracted from 36 fresh–frozen bone tissue samples, including 18 tumoral and 18 paired non-tumoral samples. Fragments Per Kilobase of transcript per Million mapped reads (FPKM) values were calculated for each gene to determine expression differences. Additional differential gene expression in tumor versus normal tissues from human sarcoma (SARC) patients was analyzed using Gene Expression Profiling Interactive Analysis (GEPIA; http://gepia.cancer-pku.cn/↗; accessed on 1 June 2024), a web-based platform that provides differential gene expression (DGE) analysis, survival analysis, and correlation analysis across various cancer types using data from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) projects. Box plots were generated to illustrate the differential expression patterns of core clock genes (BMAL1, CLOCK, CRY1, CRY2) in tumor and normal samples. Survival analyses were conducted for BMAL1, CLOCK, CRY1, and CRY2 using GEPIA. Kaplan–Meier survival curves were generated, and Cox proportional hazard models were applied to evaluate the impacts of gene expression levels on patient outcomes. Finally, correlations between the expression levels of BMAL1, CLOCK, CRY1, CRY2, and genes associated with lipid droplet biogenesis were analyzed in human OS patient samples using the GEPIA platform to explore potential links between clock gene regulation and lipid metabolism in OS.

Statistical analyses for all experiments were conducted using GraphPad Software (version 10). Fluorescence intensities, immunoblots, and spheroid numbers and sizes were quantified using ImageJ software, and statistical evaluation of the data was performed using GraphPad. Significance was assessed using two-way ANOVA, one-way ANOVA, or unpaired Student’s t-tests, as appropriate. When p was less than 0.05, it was considered statistically significant.

To establish a CSC research model for human OS, we assessed the feasibility of using a 3D spheroid culture system, which mimics the native tumor environment and supports CSC propagation [49,50,51], with various human OS cell models (U-2OS, 143B, MNNG, Saos-2). Interestingly, our results showed that among the cell lines tested, 143B cells had the highest tumor spheroid formation capacity under 3D stem cell culture conditions (Figure S1). This is consistent with previous studies that characterized 143B cells as a representative model for studying human OS disease due to their enhanced tumorigenic potential and metastasis in mouse xenotransplantation studies in vivo [52,53]. In a recent study, we established a 143B bioluminescent reporter cell line that stably expresses a Per2 promoter-driven luciferase construct (pPer2-dLuc) and verified that these cells form in vivo OS tumors 4–6 weeks after orthotopic xenotransplantation [33]. To isolate CSCs from in vivo OS tumors, we intratibially injected 143B reporter OS cells into the mice. When OS tumors became detectable, they were harvested, dissociated into single-cell suspensions, and separated into CSC and non-CSC populations using the automated magnetic-activated cell sorting (autoMACS) system, which utilizes magnetic beads to isolate CSCs based on the expression of the CSC surface marker CD133 (Figure 1A) [54,55].

To validate our separation of the CSC and non-CSC populations in our tumor samples, we performed immunofluorescence (IF) analysis by staining for CD133 using a specific anti-CD133 antibody. We observed significantly brighter CD133 staining in CSCs compared to non-CSCs (Figure 1B,C). In addition, examining spheroid formation in 3D tumor cell cultures revealed that CSCs exhibited significantly higher rates of spheroid formation and generated larger spheroids compared to non-CSCs (Figure 1D,E). To assess the differential circadian profiles of CSC and non-CSC populations, we performed real-time bioluminescence recording analysis. Our results showed that both CSCs and non-CSCs exhibited distinct circadian oscillations of Per2 promoter activity, with similar periods (Figure 1F,G). However, the non-CSCs had a moderately higher amplitude of pPer2-dLuc expression than the CSCs (Figure 1H). These data suggest that circadian rhythms are present in 143B CSCs isolated from in vivo xenograft OS tumors.

To characterize the functional role of the circadian clock in CSCs, we used RNA interference (RNAi) to specifically reduce the expression of several circadian clock genes (e.g., CLOCK, BMAL1, PER1, PER2, CRY1, CRY2, NR1D1, NR1D2) in 143B CSCs. Significant knockdown of the mRNA expression of each gene was validated by quantitative real-time PCR (qPCR) (Figure 2A).

Once we had confirmed sufficient knockdown, we cultivated the clock gene-depleted cells under 3D culture conditions. Notably, knocking down genes in both the positive (BMAL1, CLOCK) and negative (PER1, PER2, CRY1, CRY2) limbs of the core clock machinery significantly reduced the number and size of 143B CSC-derived spheroids compared to control siRNA-treated cells (si-CTL) (Figure 2B–D, Figure S2A,B). In contrast, individual depletion of NR1D1 and NR1D2 (encoding REVERB-α and REVERB-β, respectively) (Figure 2B), which act as BMAL1 repressors in the clock stabilizing loop, had minimal effects on spheroid formation rate and size (Figure 2C,D). To further investigate the mechanism underlying the reduced spheroid formation observed upon clock gene inhibition, immunostaining analysis using a Ki-67 antibody, a marker of cell proliferation, revealed a moderate (e.g., si-CRY2) to significant (e.g., si-BMAL1, si-CRY1) decrease in Ki-67-positive nuclei in clock gene KD cells compared to control (si-CTL) cells (Figure S3A,B). Furthermore, live-cell fluorescence imaging with SYTOX necrosis and Annexin V-Cy3 apoptosis staining demonstrated that clock gene inhibition resulted in moderate to significant increases in apoptosis and necrosis in 143B CSC cells (Figure S4A,B). These results align with recent studies showing that BMAL1 and CLOCK downregulation induces cell cycle arrest and apoptosis in glioblastoma stem cells (GSCs) [27] and hepatocellular carcinoma cells [56], while Cry1/Cry2 knockout reduces MYC proto-oncogene expression [57], impairs self-renewal in induced pluripotent stem cells (iPSCs) [58], and promotes apoptosis in cervical cancer cells by suppressing NANOG, inhibiting Signal Transducer and Activator of Transcription 3 (STAT3) signaling, and activating tumor suppressor p53 [59]. These findings suggest the critical role of the core clock machinery in the tumorigenic properties of 143B CSCs, likely through a combination of decreased proliferation and increased cell death.

CSCs are frequently associated with the EMT, a process that enhances the motility and invasiveness of cancer cells, traits commonly observed in malignant metastatic progression [13,14,15,16,17]. To further explore the role of core clock genes in CSC behaviors, we performed in vitro scratch wound healing assays to monitor real-time cell migration in live 143B CSCs after siRNA knockdown of BMAL1 or CRY1/2, which were selected as representative positive and negative clock regulators, respectively. Interestingly, si-CTL-treated cells closed the scratch wound completely within 12 h (Figure 3A,B). However, wound closure was significantly delayed in cells treated with siRNAs targeting BMAL1 (si-BMAL1), CLOCK (si-CLOCK), CRY1 (si-CRY1), CRY2 (si-CRY2), or both CRY1 and CRY2 together (si-CRY1/2), with the delay being less pronounced for CRY2 knockdown alone and most pronounced for CRY1/2 double knockdown (Figure 3A,B, Figure S5).

To further assess the migration defects associated with core clock gene knockdown, we conducted transwell invasion assays with 143B CSCs transfected with clock gene-targeting siRNAs (Figure 3C,D). Similar to the results of the wound healing assays, knockdown of BMAL1 and both CRY1 and CRY2 together significantly reduced the number of cells that traversed the transwell membrane. Notably, the inhibitory effect of CRY2 knockdown alone was less pronounced than what was observed with BMAL1 or CRY1 knockdown (Figure 3C,D). These findings indicate that BMAL1 and CRY1 play more critical roles than CRY2 in regulating the migratory and invasive properties of 143B CSCs.

To determine the molecular mechanisms underlying the effects of core clock gene perturbations on CSC phenotype and behavior, we examined the expressions of well-characterized stem cell (ALDH1, ALDH2, SOX2, OCT4, NANOG) and EMT markers (CDH1, CDH2, ZEB1, VIM) in core clock gene-depleted 143B CSCs using qPCR analysis. The results showed that among the CSC markers, ALDH2 and OCT4 were significantly reduced in BMAL1- or CRY1-depleted cells compared to controls (Figure 4A), while only OCT4 was significantly reduced in CLOCK-depleted cells (Figure S6). In contrast, CRY2 knockdown had little effect on ALDH2 or OCT4 expression but significantly upregulated the expression of the other CSC markers (ALDH1, SOX2, NANOG) (Figure 4A). We also found that knockdown of BMAL1, CLOCK, or CRY1 upregulated the expression of the epithelial marker (CDH1) while downregulating the expression of the mesenchymal markers CDH2, ZEB1, and VIM (Figure 4B, Figure S6). However, CRY2 knockdown upregulated the expression of most EMT marker genes (CDH1, CDH2, ZEB1) (Figure 4B). This explains, in part, our data showing the weaker inhibitory effect of CRY2 knockdown, compared to BMAL1, CLOCK, or CRY1 knockdown, on the migratory and invasive capacities of 143B CSCs (Figure 3, Figure S5). Taken together, these findings suggest that components of the core clock machinery have distinct regulatory roles in modulating CSC and EMT marker gene expression that partially contribute to the regulation of CSC properties and behaviors.

A growing body of evidence suggests that lipid metabolism is intricately linked to cancer progression, with LD accumulation contributing to enhanced CSC tumorigenicity and invasiveness [35,36,37,38,39,40]. To investigate how core clock genes influence LD formation in 143B CSCs, we performed LD staining assays using Nile Red, a lipophilic fluorescent dye commonly used to visualize intracellular LDs [60,61]. Remarkably, knockdown of BMAL1, CLOCK, CRY1, and/or CRY2 significantly reduced LD numbers in the cells in similar patterns (Figure 5A,B, Figure S7A,B). Further quantification analysis of LD diameters and volumes revealed no substantial differences, except for a moderate but significant reduction in the CRY1/2 knockdown condition (si-CRY1/2) compared to the control (si-CTL) (Figure 5C,D, Figure S7C,D).

To uncover the molecular mechanisms behind the suppression of LD formation in 143B CSCs, we conducted qPCR analysis to identify any changes in the expressions of lipid metabolism and LD biogenesis genes (Table 1) [35].

Our results revealed that the expressions of most of the LD-associated genes tested (DGAT1, FASN, ACSL4, CHKA, SREBP1), with the exception of DGAT2 and PLIN2, were significantly reduced in BMAL1-, CLOCK-, or CRY1/CRY2-depleted cells compared to cells treated with si-CTL (Figure 5E, Figure S7E). Interestingly, PLIN2, a key structural component essential for LD integrity, was significantly upregulated in most clock gene knockdown conditions. This suggests that core clock genes predominantly influence lipid synthesis and metabolism without compromising LD structural integrity. Notably, the effects of CRY2 knockdown were relatively less, as it did not significantly impact SREBP1 expression or upregulate DGAT2 (Figure 5E). Importantly, treatment with oleic acid (OA), an inducer of LD formation [74], failed to rescue the reduced invasion capacity of 143B CSCs following clock gene knockdown (si-CLOCK, si-CRY1), despite the marked increase in invasion under control siRNA conditions (si-CTL) (Figure S8). This indicates that core clock genes regulate cell invasion through mechanisms that extend beyond or are independent of LD formation which OA alone cannot restore. Collectively, these findings suggest that components of the core clock machinery are critical for regulating lipid metabolism and LD biogenesis in 143B CSCs, which may be directly linked to clock-regulated CSC/EMT phenotypes and functions.

The presence of CSCs and EMT phenotypes is closely associated with malignant tumor progression and poor prognosis in clinical patients [75,76,77]. Based on our in vitro data showing the critical role of core clock components in CSC/EMT signatures, we performed integrated transcriptomic analyses on publicly available datasets to investigate the impacts of these factors on human OS progression and prognosis. To examine the differential expressions of BMAL1, CLOCK, CRY1, and CRY2 in normal versus tumor OS tissues, we analyzed previously published RNA sequencing data (GSE99671↗) from 18 OS patients [48]. Interestingly, despite individual variability (Figure 6A,C,E, Figure S9A), BMAL1 expression remained unchanged, while CRY1 was significantly upregulated, CRY2 showed a downregulation trend (Figure 6B,D,F), and CLOCK exhibited an upregulation trend in tumors compared to normal tissues, though not statistically significant (Figure S9B).

Similar to our OS gene expression analysis, complementary analysis of sarcoma (SARC) patient data from Gene Expression Profiling Interactive Analysis (GEPIA) revealed that CRY1 and CLOCK exhibited higher expression in tumors compared to BMAL1 or CRY2 (Figure 7A,C,E, Figure S9C). Furthermore, corresponding survival analysis showed that elevated CRY1 and CLOCK expression in tumor tissues was strongly correlated with poorer patient prognosis, compared to what was observed for BMAL1 or CRY2 (Figure 7B,D,F, Figure S9D).

In additional gene correlation analysis, CRY1 was significantly positively associated with several lipid metabolism/LD-associated genes, including DGAT1, FASN, ACSL4, and CHKA, which was more than what was observed for BMAL1, CLOCK, and CRY2 (Figure 8A–C, Figure S10, bold red boxes).

While these correlations were statistically significant, the correlation coefficients were relatively low (r = 0.12–0.14) with modest p-values (p = 0.02–0.04), suggesting that these associations, though present, may be influenced by the large sample size. Notably, CRY2 exhibited a unique positive correlation with DGAT2, contrasting with the negative correlation of DGAT2 with BMAL1 and CRY1 (Figure 8C). This observation is reminiscent of our in vitro data showing the opposing effects of knockdown of CRY2 and BMAL1/CRY1 on DGAT2 expression (Figure 5C). Collectively, these findings indicate that CRY1 may serve as a prominent pro-tumorigenic factor among the core clock components, potentially contributing to OS progression and prognosis by modulating lipid metabolism and LD biogenesis pathways.

The original contributions presented in this study are included in the article/. Further inquiries can be directed to the corresponding author. Supplementary Materials