Intratumor heterogeneity and cell secretome promote chemotherapy resistance and progression of colorectal cancer – Cell Death & Disease

3D tumor spheroid formation of CMS cell lines

HCT116 and LoVo have been classified as CMS1 cells [29], while SW620 and MDST8 have been classified as CMS4 cells [30]. In order to assess the link between CMS and chemotherapy sensitivity, the half-maximal inhibitory concentrations (IC₅₀) of 5-FU acting on these cell lines were determined using viability assays. HCT116 cells exhibited the highest sensitivity to 5-FU with IC₅₀ = 3.83 ± 0.76 µM for 3 days, whereas SW620 cells were the most resistant with a 32.6-fold higher IC₅₀ = 124.68 ± 27.09 µM (Supplementary Table S1), in line with the previous observations that CMS4 cells are relatively resistant against chemotherapy [30, 31].

Tumor spheroids composed of different CMS cell lines were generated using microwell-based cultures with ultralow attachment surfaces. HCT116 cells formed compact spheroids after two days with a diameter of ~100 μm and grew into ~450 μm structures on day 4, representing a physiologically relevant size (Supplementary Fig. S1a) [32]. The cells maintained viability for 4 days in culture and exhibited increased 5-FU resistance with IC₅₀ = 15.00 ± 3.84 μM (Supplementary Fig. S1b) as compared to 2D monolayer cultures, as previously described [33]. This increased drug resistance is believed to be largely due to the restriction of 5-FU diffusion into 3D structures, as well as due to the molecular concentration gradients in oxygen, pH, nutrients, and cellular metabolites [33, 34]. Although LoVo and MDST8 cells also showed the potential to form spheroids, these structures were rather loose resulting in non-spherical shape (Supplementary Fig. S2). Indeed, the morphology of LoVo spheroids is suggestive of loosely aggregating structures that fail to organize into organoids. MDST8 cells tended to form aggregates of multiple small sub-spheroids that failed to generate compact, fully integrated spheroids. SW620 cells did not adopt a spheroidal conformation at all (Supplementary Fig. S2). Therein, when grown in suspension, distinct CMS cell lines differ in their propensity to generate spheroids.

Spatial distribution of CMS cells in cocultured 3D spheroids

To model the intercellular interactions of CMS populations in vivo, 3D cocultured tumor spheroid models that reflect the ITH of CRC were established. 10% of CMS4 cells, which were either SW620 and MDST8 cells labeled with the CMTRA cell tracker (red), were cocultured with 90% of CMS1 cells such as HCT116 labeled with the CMFDA cell tracker (green) or LoVo expressing green fluorescent protein (GFP). We observed that the CMS1/CMS4 cocultures formed spheroidal structures and that CMS1 cells (HCT116 or LoVo) grew at the center of such spheroids, while CMS4 cells (SW620 or MDST8) preferentially localized in the periphery (Fig. 1a). Similar spheroid morphologies were observed when coculturing CMS1 and CMS4 cells at a 1:1 ratio. Collectively, these data suggest that CMS1 cells present a core-like location while CMS4 cells organize at the edges of mixed spheroids.

Cell growth and drug resistance in cocultured 3D spheroids

Next, we explored the effects of CMS interactions on cell growth in cocultured tumor spheroids composed of 10% CMS4 and 90% CMS1 cells. Interestingly, coculture did not appear to exert a strong effect on the cell growth in either population (Fig. 1b). To further assess the effect of coculture on drug resistance, spheroids were treated with 5-FU after their initial formation on day 1 post-seeding. The subsequent growth of each cell population was monitored by fluorescence microscopic imaging. When added to cultures comprising HCT116 cells alone, 10–50 μM of 5-FU decreased the volume of spheroids, accompanied by decreased compactness and shape (Fig. 2a). In contrast, HCT116 cells cocultured with SW620 cells were protected against 5-FU, resulting in an 18.3% and 32.2% increase in HCT116 survival rate with 10 and 50 μM of 5-FU, respectively (Fig. 2b). Moreover, the number of admixed SW620 cells was largely increased, by up to 91% (10 μM of 5-FU), in the coculture compared to that in the monoculture (Fig. 2b). This suggests that CMS1 cells confer 5-FU resistance to CMS4 cells in coculture conditions and that mixed CMS1/CMS4 spheroid possess a collective survival advantage in adverse conditions. Indeed, MDST8 cells were also conferred 5-FU resistance by HCT116 cells. In these mixed spheroids, MDST8 cells showed a 130% increase in survival rate (50 μM of 5-FU) when compared to those in monocultures (Fig. 2c), without being in comprehensive contact with HCT116, but rather forming several small spheroids on their own (Supplementary Fig. S3). The overall survival of HCT116 cells was again supported by MDST8 cells (Fig. 2c). In addition, coculture with HCT116 cells stimulated outgrowth of CMS4 cells against 5-FU treatments, showing a maximum increase of 36% (5 μM of 5-FU) and 22% (2.5 μM of 5-FU) in cell number for SW620 and MDST8, respectively, when compared to the vehicle-treated control.

a Representative live cell confocal fluorescence microscopy images showing the spheroid morphology of HCT116 and SW620 after 3 days of 5-FU treatment. Cells were stained with either cell tracker CMFDA (green) or CMRA (red) fluorescent probes. Scale bars = 100 μm. b–e 5-FU response of CMS1 and CMS4 cells in spheroids. Monocultured (white bars) and cocultured spheroids (gray bars) of b HCT116 and SW620, c HCT116 and MDST8, d LoVo and SW620 and e LoVo and MDST8 were exposed to different concentrations of 5-FU for 3 days. Cell viability was measured by means of image analysis quantifying the fluorescence intensity of cell trackers that represent the cell number, and normalized to the vehicle control. The bars represent the average of viability and the error bars represent the standard deviation (n = 3). Statistical significance was calculated using a one-way ANOVA followed by Student’s t-test. P-values of <0.05 and 0.01 were considered significant (*) and highly significant (**), respectively, when compared to the monoculture.

Finally, LoVo cells stably expressing GFP were cocultured with either SW620 or MDST8 cells. The resulting spheroids were then exposed to different 5-FU concentrations using the same experimental setup as above (Supplementary Figs. S4 and S5). As observed for HCT116 cells, coculture significantly sustained the survival of LoVo and enhanced the resistance of CMS4 cells to 5-FU (Fig. 2d, e). Once again, this effect appeared independent of close contact of one cell population to another (Supplementary Figs. S4 and S5). Collectively, these results suggest that CMS1/CMS4 coculture increases 5-FU resistance.

The effect of the CMS1 secretome on CMS4 drug resistance

We next investigated the potential mechanisms involved in the interplay between CMS1 and CMS4 cells. Recently, Bastola and colleagues reported that the secretome from the glioblastoma core promoted malignancy of cells at the tumor edge [28]. Based on this finding, we examined whether the secretome of 5-FU treated CMS1 cells would influence the drug response of CMS4 cells to 5-FU. Conditioned (CM) were derived from HCT116 cells cultured in the absence (DMSO, CM_vehicle) or presence of 2.5 µM of 5-FU (CM_5-FU) for 3 days. Recipient SW620 cells were then cultured in HCT116 CM and their own culture medium at a 1:1 ratio and exposed to increasing concentrations of 5-FU for 3 days (Fig. 3a). Both CM_vehicle and CM_5-FU dramatically reduced the toxic effect of 500 µM 5-FU on SW620, yielding a threefold increase in viable cells (Fig. 3b). Enhanced resistance to 5-FU used at 30–100 µM was also observed with MDST8 received HCT116 CM. In contrast, the drug response of neither HCT116 nor LoVo was altered by HCT116-derived CM. These data suggest that the HCT116 secretome can promote 5-FU resistance of CMS4 cells specifically.

a Schematic of recipient cells treated with CM of donor cells. Recipient cells were treated with either CM_vehicle or CM_5-FU of b HCT116 or c LoVo doner cells, and were exposed to different concentrations of 5-FU for 3 days. Cells treated with media only were taken as a control. Cell viability was measured using MTS assays. The squares, circles, and triangles represent the average viability normalized to the vehicle control and the error bars represent the standard deviation (n = 3). Statistical significance was calculated using a one-way ANOVA followed by Student’s t-test. P-values of <0.05 and 0.01 were considered significant (*) and highly significant (**), respectively, when compared to the control.

In an attempt to determine whether the secretome of other CMS1 cells could also induce 5-FU resistance, or whether this phenomenon exclusively applies to HCT116, CM were collected from LoVo cells under the same conditions and added to CMS4 cells. LoVo-derived CM significantly sustained the viability of SW620 and MDST8 cells against 5-FU at concentrations from 10 to 500 μM (Fig. 3c). LoVo CM_vehicle induced minimal or no increase in viability of these cell lines, suggesting that the observed effect is largely the result of specific secretome changes induced by 5-FU. Unlike SW620 and MDST8 cells, HCT116 and LoVo cells were as insensitive to LoVo CM as they were to HCT116 CM (Fig. 3b).

The effect of the CMS1 secretome on CMS4 migration and invasion

We next examined the capacity of MDST8 to migrate through the matrix of the basement membrane after exposure to the CMS1 secretome. This was determined using transwell inserts coated with a Matrigel layer onto which MDST8 were cultured. These transwell inserts were then placed on top of HCT116 or LoVo cells. Exposure to soluble signals emanating from HCT116 or LoVo cells modestly increased MDST8 migration through the transwell membrane by 1.15 and 1.23 fold, respectively (Fig. 4a, Fig. S6). DMSO-treated HCT116 and LoVo cells significantly increased MDST8 invasion rate by 1.64 and 1.45 fold, causing 6.82% and 16.54% MDST8 cells to cross the Matrigel barrier, respectively (Fig. 4). The invasion capacity of MDST8 was further promoted by the addition of 5-FU to the system by 16.53% (in response to HCT116 cells) and 26.53% (in response to Lovo cells) (Fig. 4). Therefore, we may surmise that, in response to 5-FU, CMS1 cells secrete factors that promote the invasion capacity of CMS4 cells.

a Representative widefield fluorescence microscopy images showing MDST8 migration and invasion through transwell membrane with and without Matrigel coating after 2 day exposure to HCT116 or LoVo in the bottom wells. MDST8 exposed to only media without cells were taken as a control. Cells were treated with either DMSO vehicle or 2.5 μM of 5-FU. Cell nuclei were stained with Hoechst (blue). Scale bars = 100 μm. b Invasion rate of MDST8 presented as the percentage of cell invasion through Matrigel-coated transwell membrane relative to the cell migration through the non-Matrigel coated transwell membrane. The bars represent the average and the error bars represent the standard deviation (n = 3). Statistical significance was calculated using a one-way ANOVA followed by Student’s t-test. P-values of <0.05 and 0.01 considered significant (*) and highly significant (**), respectively, when compared to the control.

The effect of metabolites on CMS4 drug resistance

During tumor progression and metastasis, tumor cells undergo rapid metabolic adaptations and coordinate with their surroundings to maintain biosynthetic and bioenergetic demands while escaping immunosurveillance or therapeutic interventions, which are now recognized as hallmarks of cancer [35]. Thus, we investigated whether metabolites in the CMS1 secretome are responsible for the observed effects. CM were collected from HCT116 or LoVo cultured in the absence (DMSO, CM_vehicle) or presence of 2.5 µM of 5-FU (CM_5-FU) for 3 days. Metabolites of these CM were dialyzed into fresh media (Metabolite_vehicle, Metabolite_5-FU) using dialysis membranes with a cut-off of 3.5 kDa and applied to CMS4 cell lines as previously. Similar to CM_5-FU, Metabolite_5-FU greatly sustained the viability of both SW620 and MDST8 cells against 5-FU at concentrations from 5 to 300 µM (Fig. 5a). The same effect was observed with LoVo metabolites (Fig. 5b). Moreover, we observed the 5-FU resistance-promoting effect with the remaining CM_5-FU after the dialysis of metabolites (Fig. S7). These results suggest that dialyzable metabolites (rather than extracellular vesicles or proteaceous factors) are the key communicators in the CMS1 secretome that can promote 5-FU resistance of CMS4 cells.

CMS4 cells were treated with either metabolite_vehicle or metabolite_5-FU of a HCT116 or b LoVo CMS1 cells, and were exposed to different concentrations of 5-FU for 3 days. Cells treated with media only were taken as a control. Cell viability was measured using MTS assays. The squares, circles, and triangles represent the average viability normalized to the control and the error bars represent the standard deviation (n = 3). Statistical significance was calculated using a one-way ANOVA followed by Student’s t-test. P-values of <0.05 and 0.01 were considered significant (*) and highly significant (**), respectively, when compared to the control.

Metabolite analyses of CMS1 conditioned media

To evaluate the metabolic adaptation of CMS1 cells in response to 5-FU, as well as to identify the relevant mediators and pathways involved in the reactive secretome, the CM of CMS1 cells were analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS). A total of 146 metabolites involved in a broad range of metabolic pathways were quantified, including amino acids, organic acids, nucleotides, and cofactors (Table S2). The relative steady-state levels of 52 metabolites were significantly altered in the CM_vehicle of HCT116 compared to the control media without cells (Supplementary Fig. S8 and Table 1). Pathway analysis indicated that the levels of 13 metabolites involved in aminoacyl-tRNA biosynthesis (amino acids) were consumed by HCT116 cells, representing the highest pathway significance (Supplementary Table S3). Of note, phenylalanine, tyrosine, and tryptophan biosynthesis, as well as linoleic acid metabolism showed the highest pathway impact of 1.0 among downregulated metabolites. On the other hand, upregulated metabolites in the HCT116 secretome were mainly involved in alanine, aspartate, and glutamate metabolism, including N-acetylaspartate, asparagine, glutamine, fumarate, pyruvate, and alpha-ketoglutarate, representing the highest pathway significance (Supplementary Table S4). Moreover, D-glutamine and D-glutamate metabolism and vitamin B6 metabolism showed the highest pathway impact of 0.50 and 0.49, respectively.

The levels of 51 metabolites were significantly altered in the CM_vehicle of LoVo compared to the control media without cells (Supplementary Fig. S9 and Table 2). Similar to HCT116, significantly downregulated metabolites were involved in the aminoacyl-tRNA biosynthesis, representing the highest pathway significance (Table 2, Supplementary Table S5). Once again, phenylalanine, tyrosine, and tryptophan biosynthesis together with linoleic acid metabolism showed the highest pathway impact. Unlike HCT116 cells, upregulated metabolites were mainly involved in citrate cycle (TCA cycle), representing the highest pathway significance (Supplementary Table S6). Riboflavin metabolism, D-glutamine and D-glutamate metabolism, and vitamin B6 metabolism showed the highest pathway impact of 0.50, 0.50, and 0.49, respectively. Overall, the metabolite profile of LoVo CM_vehicle largely overlaps with that of HCT116 CM_vehicle (Supplementary Table S7).

We next examined the influence of 5-FU on the metabolite composition of CMS1 secretome. A total of 37 soluble metabolites exhibited differential patterns in the secretome induced by 5-FU compared to vehicle for both HCT116 and LoVo (Supplementary Figs. S10 and S11, Table S8 and S9). Among these metabolites, we observed a significant overlap of 22 (19 upregulated and 3 downregulated) compounds between HCT116 and LoVo (Table 3). Pathway analysis on these upregulated metabolites revealed that 5-FU treatments impacted several metabolic pathways, including aminoacyl-tRNA biosynthesis, which showed the highest pathway significance. Phenylalanine, tyrosine, and tryptophan biosynthesis and linoleic acid metabolism had the highest pathway impact of 1.0 (Supplementary Table S10). These data suggest that such differentially regulated factors in the CMS1 secretome, induced by 5-FU, could stimulate drug resistance, outgrowth, and invasion capacity of CMS4 cells.

The effect kynurenine pathway (Kyn) metabolites on CMS4 drug resistance

Following the metabolite analyses of CMS1 CM, we further tested the functional significance of the corresponding metabolites. We observed that phenylalanine, tyrosine and tryptophan biosynthesis and linoleic acid metabolism were highly upregulated in the CMS1 secretome in response to 5-FU. Tryptophan metabolism occurs mainly via the Kyn pathway, which has been shown to promote colorectal cancer progression, especially by enhancing cellular proliferation and resistance to apoptosis [36, 37]. Therefore, we evaluated the effect of 5 metabolites involved in the Kyn pathway on CMS4 drug resistance [38]. Nicotinamide, kynurenine, and quinolinic acid sustained the viability of SW620 cells against 150 µM of 5-FU (Fig. S12). A similar drug resistance-promoting effect was also observed with nicotinamide and kynurenic acid on MDST8 cells responding to 50 and 150 µM of 5-FU (Fig. S13). These results suggest that the metabolites from Kyn pathway in the CMS1 secretome might contribute to promote 5-FU resistance of CMS4 cells.