DKK1 is upregulated in oxaliplatin-resistant colorectal cancer cells
OR CRC cell lines (LoVo-OR, SW48-OR, and CA01-OR) were created following established protocols [7]. Utilizing MTS-based assays revealed reduced sensitivity to oxaliplatin compared to parental cells (P) (Fig. 1A). Further analysis via tumor sphere formation (Supplementary Fig. 1A) and xenograft assays confirmed the robust growth of LoVo-OR cells in the presence of oxaliplatin (Fig. 1B and Supplementary Fig. 1B). To explore molecular factors influencing oxaliplatin responses, we conducted proteomic analysis comparing LoVo-OR and LoVo-P tumor spheres. Utilizing the Proteome Profiler Human XL Oncology Array, we identified 3 downregulated and 4 upregulated proteins in LoVo-OR spheres, notably FGF2 and DKK1 (Fig. 1C and Supplementary Fig. 1C). Dysregulation of FGF signaling is associated with aggressive cancer traits, drug resistance, and poor clinical outcomes [37]. However, analysis from the UALCAN database showed that FGF2 upregulation wasn't significantly linked to colon adenocarcinoma (COAD) tissues, tumor stages, lymph node status, or clinical outcomes (Supplementary Fig. 1D-G).
DKK1 is frequently upregulated in various tumors and is associated with poor prognosis in patients [25, 38, 39]. Analysis of the UALCAN database showed a significant increase in DKK1 expression in COAD tissues (P = 0.0019) across different tumor stages and lymph node statuses (Supplementary Fig. 1H-J), suggesting its potential as a diagnostic marker for COAD patients. Although the association between DKK1 mRNA expression and overall survival (OS) in COAD patients was not statistically significant (Supplementary Fig. 1K), we aimed to investigate its role in oxaliplatin resistance in CRC.
Western blotting analysis of CRC tumor spheres demonstrated upregulated DKK1 expression OR cell lines (LoVo-OR, SW48-OR, and CA01-OR) compared to their oxaliplatin-sensitive counterparts (P) (Fig. 1D). As DKK1 is a secreted protein, we examined its levels in the cell culture media. Daily collection of medium samples showed detectable DKK1 protein in both LoVo-P and LoVo-OR cells, with higher levels in LoVo-OR cells (Fig. 1E). Enzyme-linked immunosorbent assay (ELISA) quantification normalized against cell numbers confirmed more efficient DKK1 secretion in LoVo-OR cells, with a concentration of secreted DKK1 (sDKK1) measured at 0.084 ± 0.023 ng/µl in LoVo-OR cell culture medium on day 3 (Fig. 1F).
Downregulation DKK1/CKAP4/AKT signaling facilitates oxaliplatin resistance in colorectal cancer cells
DKK1 is known to bind, cytoskeleton-associated protein 4 (CKAP4, a membrane protein), activating downstream AKT signaling and facilitating cancer cell proliferation and malignant transformation [17, 38]. To explore the involvement of DKK1 signaling in oxaliplatin resistance in CRC, we examined AKT activation, finding a positive correlation with increased DKK1 expression in OR CRC tumor spheres (Fig. 1D). Using RNA interference (RNAi), we reduced DKK1 expression with two independent DKK1-specific shRNAs (#85 and #88). SW48-OR cells expressing DKK1 shRNAs showed decreased DKK1 expression, leading to reduced phospho-AKT levels at Thr308 and Ser473 (Fig. 2A). DKK1 knockdown inhibited SW48-OR colony formation in soft-agar assays (Fig. 2B) and significantly reduced the number of SW48-OR tumor spheres in the presence of oxaliplatin (Fig. 2C). Similar results were observed in LoVo-OR and CA01-OR cells (Supplementary Fig. 2A-D), underscoring DKK1's crucial role in activating AKT and modulating oxaliplatin responses in resistant CRC cells.
Subsequently, UALCAN database analysis revealed elevated CKAP4 expression in COAD tissues (P = 0.0016), significant across tumor stages and lymph node status (Supplementary Fig. 2E-G). Although the association between CKAP4 mRNA expression and overall survival (OS) in COAD patients wasn't statistically significant (Supplementary Fig. 2H), we investigated CKAP4's role in mediating DKK1-stimulated AKT signaling. SW48-OR cells expressing CKAP4 shRNAs exhibited decreased CKAP4 expression, reduced AKT activity, and impaired colony and tumor sphere formation compared to controls with scrambled shRNAs (Fig. 2D-F). Similar effects were seen in LoVo-OR and CA01-OR cells (Supplementary Fig. 2I-L). In conclusion, our findings suggest that DKK1/CKAP4/AKT signaling axis is upregulated in oxaliplatin-resistant CRC cells and plays a crucial role in promoting cell growth in the presence of oxaliplatin.
Secreted DKK1-mediated activation of AKT signaling via CKAP4 in colorectal cancer cells
The DKK1/CKAP4/AKT signaling axis's impact on cancer biology has been previously demonstrated by manipulating cellular DKK1 expression levels [19, 38]. We aimed to address the function of secreted DKK1 (sDKK1) in CRC cells. sDKK1-containing conditioned media (CM) was obtained from 293 cells expressing 3xFlag-tagged DKK1 (DKK1-3xF, Supplementary Fig. 3A) after a 3-day serum-free medium incubation. Levels of DKK1-3xF in CM were assessed via western blotting (Supplementary Fig. 3B) and ELISA. Time-course CM treatment of SW48 cells revealed significant AKT activation at the 30-minute mark with DKK1-3xF, contrasting with negligible effects observed with mock CM (Fig. 3A). To assess sDKK1’s interaction with the CKAP4 receptor, we employed a truncated form of DKK1 lacking CRD1 (DKK1ΔCRD1-3xF; Supplementary Fig. 3A), responsible for CKAP4 interaction [40]. CM containing DKK1ΔCRD1-3xF (ΔCRD1-3xF) was collected and leveled (Supplementary Fig. 3B), revealing reduced DKK1-induced AKT phosphorylation in SW48 cells (Fig. 3A), similarly observed in CA01 cells (Supplementary Fig. 3C). Subsequently, we explored CKAP4’s role in mediating sDKK1-induced AKT signaling. SW48 cells expressing shRNAs against CKAP4 treated with DKK1-3xF-containing CM showed partially attenuated DKK1-3xF-induced phospho-AKT levels (Fig. 3B), suggesting CKAP4’s involvement in mediating sDKK1-activated AKT.
Additionally, to visualize sDKK1's interaction with CKAP4 on the plasma membrane, we generated a DKK1-GFP fusion protein, where GFP was tagged at DKK1’s C-terminus. The signal peptide (SP) [41] derived from DKK1 was fused to the N-terminus of GFP (SP-GFP) for GFP secretion guidance, serving as a control (Supplementary Fig. 3A). CM collected from 293 cells expressing SP-GFP (SP), DKK1-GFP (DKK1), or DKK1ΔCRD1-GFP (ΔCRD1) was analyzed and leveled (Supplementary Fig. 3D). Flow cytometry analysis of SW48 cells incubated with CM revealed a significant decrease in GFP + cells with ΔCRD1 CM compared to DKK1 CM (Fig. 3C and Supplementary Fig. 3E). Plasma membrane isolation assays confirmed DKK1 presence in the plasma membrane fraction, with reduced levels for ΔCRD1, supporting sDKK1’s binding to the cell plasma membrane, attenuated by CRD1 truncation (Fig. 3D). To validate CKAP4’s role in sDKK1-cell interactions, flow cytometry analysis of SW48 cells expressing shRNAs against CKAP4 incubated with DKK1 CM revealed reduced GFP + cells (Fig. 3E and Supplementary Fig. 3F). Immunocytochemistry (ICC) analysis further confirmed sDKK1's association with the plasma membrane and CKAP4, with colocalization observed and reduced by CRD1 truncation (Fig. 3F). These findings support CRD1’s critical role in sDKK1 interaction with CKAP4 on the cell surface to activate AKT signaling.
sDKK1 modulates oxaliplatin responses in CRC cells
Next, we investigated the impact of sDKK1 on oxaliplatin responses in CRC cells via the CKAP4/AKT pathway. CM containing sDKK1 was obtained from 293 cells expressing DKK1-3xF, DKK1ΔCRD1-3xF (ΔCRD1-3xF), or an empty vector (Mock). SW48 cells treated with 0.1 µM oxaliplatin for 24 hours, followed by a 30-minute CM incubation. Western blotting showed increased phospho-AKT levels with DKK1-3xF CM, partially restoring oxaliplatin-inhibited AKT phosphorylation, weakened by CRD1 truncation (Fig. 4A). MTS assays indicated no significant effect on cell proliferation, but DKK1-3xF CM partially rescued oxaliplatin-inhibited cell growth (Fig. 4B). Similar results were observed in CA01 cells (Fig. 4C and D). Apoptosis assays in SW48 cells showed DKK1-3xF CM minimized oxaliplatin-induced early apoptosis compared to mock or ΔCRD1-3xF CM (Fig. 4E and Supplementary Fig. 4A). Long-term focus formation assays in SW48 cells stably expressing DKK1-3xF and ΔCRD1-3xF indicated DKK1 expression partially rescued cells from oxaliplatin treatment (Fig. 4F and Supplementary Fig. 4B). Anchorage-independent cell growth assays in SW48 cells demonstrated DKK1 expression partially restored oxaliplatin-suppressed sphere growth, whereas ΔCRD1 did not (Fig. 4G and Supplementary Fig. 4C). These findings collectively suggest sDKK1 contributes to modulating oxaliplatin responses through the CKAP4/AKT pathway.
DKK1 sustains CRC tumor growth in the presence of oxaliplatin
In a study on CRC's oxaliplatin resistance, CA01 cells expressing SP-GFP (SP) or DKK1-GFP (DKK1) were xenografted into mice. Mice received mock or oxaliplatin injections (5 mg/kg) weekly for 3 weeks. DKK1-GFP expression had no significant effect on tumor volume. However, oxaliplatin notably reduced tumor volume in SP-GFP tumors, whereas DKK1-GFP tumors-maintained growth (Fig. 5A and B). By treatment end, SP-GFP tumors decreased while DKK1-GFP tumors increased in size (Fig. 5C). Immunohistochemistry (IHC) analysis revealed a decrease in Ki67 + cells to 33.7 ± 8.5% and an increase in TUNEL + cells to 9.4 ± 1.5% in SP-GFP tumors, whereas these responses were attenuated in DKK1-GFP tumors (Fig. 5D and E, and Supplementary Fig. 5A and B). To confirm DKK1-GFP secretion in tumor-bearing mice, blood plasma underwent immunoprecipitation with anti-GFP antibodies and protein A/G beads. Western blotting confirmed GFP fusion protein presence, indicating SP-GFP and DKK1-GFP secretion from tumors into the blood (Supplementary Fig. 5C).
Findings from SW48 cells expressing DKK1-3xF (WT), DKK1ΔCRD1-3xF (ΔCRD1), or empty vector (Vec) mirrored those in CA01 xenograft tumors. DKK1 or ΔCRD1 expression didn't significantly enhance SW48 xenograft growth. Oxaliplatin notably reduced tumor volume in ΔCRD1 or Vec tumors, while DKK1-expressing tumors remained stable (Supplementary Fig. 5D and E). By treatment end, ΔCRD1 and Vec tumors decreased in size, while DKK1 tumors increased (Supplementary Fig. 5F). Animal body weights remained consistent throughout treatment (Supplementary Fig. 5G). These findings support elevated DKK1 sustaining CRC tumor growth in oxaliplatin presence, indicating its key role in oxaliplatin resistance in colorectal cancer.
To associate DKK1 expression with oxaliplatin resistance onset in CRC patients, we conducted IHC on 12 pairs of pre-treatment and post-relapse tumor samples from oxaliplatin-treated patients (Fig. 5F). DKK1 expression was quantified using the IHC H-score. Statistical analysis revealed a significant difference in DKK1 expression between pre-treatment and post-relapse CRC samples (p = 0.0197, Supplementary Fig. 5H). Normalizing DKK1 H-scores from 12 post-relapse samples against their pre-treatment counterparts yielded fold changes in DKK1 expression (Fig. 5G). Remarkably, increased DKK1 H-scores (fold change > 1.0) were found in 9 OR CRCs, constituting 75% of the total 12 CRCs. Among these, 6 OR CRCs (50% of the total 12 CRCs) exhibited over a 2-fold increase in DKK1 H-score (Fig. 5G). These results strongly suggest a positive association between DKK1 expression and oxaliplatin resistance development in CRC.
LZ protein suppresses oxaliplatin-resistant CRC cell growth
Our findings suggest that the crucial role of the DKK1/CKAP4 interaction in regulating oxaliplatin responses in CRC cell lines, indicating a potential therapeutic strategy for suppressing OR CRC cell growth. Given the reported requirement for the LZ domain in CKAP4 and the CRD1 in DKK1 for their interaction [19, 40], we investigated the disruptive potential of an LZ domain-containing protein.
We cloned a one-hundred-amino-acid region overlapping the defined LZ domain (aa 468–503) of CKAP4 [38], generating a secreted form of the SP-LZ-mCherry fusion protein (LZ protein) (Supplementary Fig. 6A). CM containing LZ protein or SP-mCherry was collected, confirmed, and normalized for consistency (Supplementary Fig. 6B). To assess the interaction of the LZ protein with sDKK1 via CRD1, immunoprecipitation assays were performed using anti-DsRed antibodies with CMs containing DKK1-GFP and LZ protein. The results revealed GFP fusion proteins in the anti-DsRed immunoprecipitation complex in DKK1-GFP CM, with diminished levels in DKK1ΔCRD1-GFP CM (Supplementary Fig. 6C), suggesting that the CRD1 mainly facilitates the interaction between the LZ protein and sDKK1. Further validation of the specificity of the LZ protein/sDKK1 interaction was achieved, demonstrating the specific interaction of the LZ protein with both exogenously expressed DKK1-GFP and endogenous sDKK1 proteins (Fig. 6A and B). To assess the impact of the LZ protein on the binding of sDKK1 to CKAP4 on the cell surface, flow cytometry and ICC analyses were performed using SW48 cells incubated with DKK1-GFP + SP-mCherry CM or DKK1-GFP + LZ protein CM for 30 minutes. The percentage of GFP + cells (Fig. 6C and Supplementary Fig. 6D) and the association of sDKK1-GFP with the cell surface (Fig. 6D) were reduced by the presence of LZ protein CM, contrasting with SP-mCherry. These results suggest that the LZ protein functions by sequestering sDKK1, thereby interfering with the binding of sDKK1 to CKAP4 on the cell surface.
We then evaluated the LZ protein's therapeutic potential in inhibiting OR CRC growth. Long-term focus formation assays demonstrated significantly inhibited colony formation in LoVo-OR cells when exposed to LZ protein CM compared to SP-mCherry (Fig. 6E). To validate these findings, LoVo-OR cells were xenografted into mice, divided into SP and LZ groups, and administrated SP-mCherry CM and LZ protein CM three times per week for 3 weeks. After CM withdrawal, tumor growth was monitored for an additional 3 weeks with continuous oxaliplatin administration (Supplementary Fig. 6E). Notably, LoVo-OR tumor growth was significantly suppressed in the LZ group compared to the SP group, and this effect persisted for up to 3 weeks after discontinuing LZ protein CM administration (Fig. 6F and Supplementary Fig. 6F). Importantly, there were no discernible effects on the mice’s body weight (Supplementary Fig. 6G), suggesting the safety of CM administration. Additionally, IHC analysis of tumor samples revealed a reduced percentage of actively proliferating tumor cells (96.9 ± 0.9% in SP, and 59.4 ± 6.5% in LZ) and an increased percentage of apoptotic cells (0.8 ± 0.4 in SP, and 12.5 ± 4.8% in LZ) in the LZ group (Fig. 6G, and Supplementary Fig. 6H and I). These findings collectively support the LZ protein's inhibitory effect on oxaliplatin-resistant CRC growth, suggesting a promising therapeutic strategy for CRC characterized by elevated DKK1 levels.