Endothelin-1 as a Mediator of Heme Oxygenase-1-Induced Stemness in Colorectal Cancer: Influence of p53

Heme oxygenase-1 (HO-1) is an antioxidant protein implicated in tumor progression, metastasis, and resistance to therapy. Elevated HO-1 expression is associated with stemness in several types of cancer, although this aspect has not yet been studied in colorectal cancer (CRC). Using an in vitro model, we demonstrated that HO-1 overexpression regulates stemness and resistance to 5-FU treatment, regardless of p53. In samples from CRC patients, HO-1 and endothelin converting enzyme-1 (ECE-1) expression correlated significantly, and p53 had no influence on this result. Carbon monoxide (CO) activated the ECE-1/endothelin-1 (ET-1) pathway, which could account for the protumoral effects of HO-1 in p53 wild-type cells, as demonstrated after treatment with bosentan (an antagonist of both ETRA and ETRB endothelin-1 receptors). Surprisingly, in cells with a non-active p53 or a mutated p53 with gain-of-function, ECE-1-produced ET-1 acted as a protective molecule, since treatment with bosentan led to increased efficiency for spheres formation and percentage of cancer stem cells (CSCs) markers. In these cells, HO-1 could activate or inactivate certain unknown routes that could induce these contrary responses after treatment with bosentan in our cell model. However more research is warranted to confirm these results. Patients carrying tumors with a high expression of both HO-1 and ECE-1 and a non-wild-type p53 should be considered for HO-1 based-therapies instead of ET-1 antagonists-based ones.


Introduction
Worldwide, colorectal cancer (CRC) annually affects more than one million men and women and causes more than half a million deaths [1]. Drug-resistance remains one of the challenges for the low survival rates of patients in advanced stages of the disease [2].

Patients
This research work was part of a larger, ongoing prospective study that aimed to investigate CSCs regulatory pathways in CRC. Initially, 183 patients were recruited, of which 150 were included in this study [35]. Samples were obtained from patients who underwent surgery for primary sporadic CRC and were provided by the Andalusian Tumor Bank Network (RBTA) ( Table S1). Viable tumor tissues and adjacent normal mucosa were dissected immediately and fresh-frozen in Tissue-Tek1 (Optimal Cutting Temperature Compound, Sakura Finetek Europe B.V., Zoeterwoude, The Netherlands). The inclusion criteria included people over 18 years, without hereditary burden, not treated with neoadjuvant therapy, and not previously diagnosed or treated for cancer. The Ethical Committee of Clinical Research of Granada (project code: PI-067/2013; date of approval: 24 January 2014) approved the study. All patients gave written informed consent for the use of samples in biomedical research.

Transient Transfection of HO-1
Cells at 60-70% of confluence were transiently transfected with lipofectamine 2000 transfection reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. Transfected cells were then collected 24-96 h after transfection. Plasmid expression vector pCMV6 containing the human HMOX1 gene and the corresponding empty vector were purchased from Origene Technologies (Rockville, MA, USA).

ET-1 ELISA
ET-1 was analyzed in 100 µL of conditioned media using a colorimetric ELISA kit (R & D Systems, Minneapolis, MN, USA) in accordance with the manufacturer's protocol using a microplate reader TRIAD Multimode Reader series (Dynex Technologies, Chantilly, VA, USA) at 450 nm. Results were determined by comparison with standard curves and normalized to the number of cells in each well.

RNA Isolation and cDNA Synthesis
Total RNA from tissue samples was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany). The amount of total RNA was determined by UV spectrophotometry, and RNA integrity was assessed by agarose gel electrophoresis. First-strand cDNA was prepared by a reverse transcription cDNA synthesis kit (qScript cDNA Synthesis, Quantabio, Beverly, MA, USA).

Real Time PCR
Quantitative analysis of mRNA expression was performed using the PerfeCTa SYBR Green SuperMix Kit (Quantabio, Beverly, MA, USA) on a CFX96 Dx Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) following manufacturer's instructions. Expression of mRNA was evaluated through standard curves generated for each target gene by plotting Ct values versus log cDNA dilution. UBC was used to normalize mRNA levels. PCR products were verified by a melting profile and agarose gel electrophoresis to rule out nonspecific PCR products and primer dimers.

ALDEFLUOR Assay
An ALDEFLUOR™ Kit (Stem Cell Technologies, Vancouver, BC, Canada) was used to detect enzyme ALDH1 activity in cultured cells according to the manufacturer's instructions. After treatments, cells were incubated with BODIPY-aminoacetaldehyde (BAAA), a fluorescent non-toxic substrate for ALDH, which was converted into BODIPY-aminoacate (BAA) and retained inside the cells. Viable ALDH1+ cells were quantified by flow cytometry on a FACS Aria IIIu (BD Biosciences, San Jose, CA, USA). The specific inhibitor of ALDH, diethylaminobenzaldehyde (DEAB), was used to control for background fluorescence.

Isolation and Characterization of CSCs
Enrichment of CSCs was achieved via serial trypsin treatment, as previously described [36,37]. Briefly, cells at 60-80% of confluence were washed with PBS and treated with 0.05% trypsin for 2 min at 37 • C. Detached cells were plated and allowed to attach for 24 h. Then, the above procedure was repeated, and the cells obtained were considered CSCs. After the first trypsin treatment, remaining cells attached to the dishes were washed twice with PBS and incubated with 0.05% trypsin for 4 min at 37 • C. Cells detached from this trypsinization were discarded. Dishes with remaining trypsin-resistant cells were considered non-CSCs Once isolated, cell surface marker levels of CSCs were determined with human anti-CD44-PE, anti-CD326-FITC, and anti-CD133-APC antibodies (Biolegend, San Diego, CA, USA). Samples were measured and analyzed by flow cytometry on a FACS Aria IIIu (BD Biosciences, San Jose, CA, USA).

MTT Assay
Cells were seeded in 96 well-plates at a concentration of 40,000 cells/mL, allowed to attach overnight, and transfected with either pCMV6-HMOX1 or an empty vector. Then, the cells were treated with the vehicle or the corresponding drug, 24 h after transfection. After 72 h, 10 µL of 5 mg/mL MTT were added to each well. Four hours later, cells were lysed with 100 µL buffer (20% SDS in 50% formamide at pH 4.7) at 37 • C overnight. Absorbance was measured in a TRIAD Multimode Reader series, (Dynex Technologies, Chantilly, VA, USA) at 570 nm. Means were estimated from the results of four samples in each experimental group. All experiments were performed two times.

Statistical Analysis
For each patient, mRNA levels of genes in tumor samples were normalized to mRNA levels in normal mucosa. Descriptive statistics were reported as medians with interquartile ranges (IQR) for continuous variables and as whole numbers and percentages for categorical variables. Low and high levels of CSCs markers were obtained through the median of the mRNA expression levels in our cohort of patients. Associations between clinicopathological features of CRC patients and gene expression were analyzed with the Kruskal-Wallis and Mann-Whitney non-parametric tests. For the correlation analysis, the Pearson test was used after transforming the variables applying natural logarithms. All confidence intervals (CIs) were stated at the 95% level. Statistical significance was defined as p < 0.05. All statistical calculations were performed using SPSS software version 15.0 for Windows (IBM, Chicago, IL, USA).

HO-1 Overexpression Induces Stemness in CRC In Vitro Regardless of p53 Status
In order to investigate whether HO-1 conferred stem cell properties on CRC cells, we used the pCMV6-HMOX1 plasmid to transiently overexpress HO-1 in HCT-116, HCT-116 p53 null, and HT-29 cells. Overexpression of HO-1 increased aldehyde dehydrogenase 1 positive (ALDH1+) subpopulation, representing the subpopulation of CSCs, in HCT-116 and HCT-116 p53 null cells at 96 h after transfection (Figure 1a,b). In HT-29 cells, the percentage of ALDH1+ cells was lower in pCMV6-HMOX1 versus mock transfected cells at 72 h after transfection, whereas the percentage of this subpopulation was similar in control (mock) and HO-1 overexpressing cells at 96 h ( Figure 1c).
The percentage of cells with high expression of CSCs markers in the total population (TP) of cells increased in pCMV6-HMOX1 versus mock transfected cells at 96 h after transfection. At this time, the percentage of CD133 high /CD44 high /CD326 high cells significantly increased in the TP of pCMV6-HMOX1 versus mock HCT-116 and HT29 transfected cells at 96 h after transfection (Figure 1d). We did not find triple-labeled subpopulation cells in the HCT-116 p53 null line; however, the percentage of CD44 high /CD326 high cells significantly increased in the TP of pCMV6-HMOX1 versus mock HCT-116 p53 null transfected cells at 96 h after transfection (Figure 1e). Similarly, in HT-29 cells, the percentage of the CD44 high /CD326 high subpopulation significantly increased in the TP of pCMV6-HMOX1 versus mock transfected cells at 96 h after transfection (Figure 1e). In addition, the percentage of CD44 high /CD326 high cells increased in the CSCs subpopulation of pCMV6-HMOX1 versus mock transfected cells at 96 h after transfection in the three cell lines studied (Figure 1f). These results indicate that HO-1 overexpression increases the proportion of cancer cells with a CSC phenotype in the different subpopulations analyzed.
Next, we studied the anchorage-independent growth of CSCs and non-CSCs subpopulations extracted from HCT-116 and HCT-116 p53 null after transient overexpression of HO-1 under free-serum conditions. As shown in Figure 2, CSCs and non-CSCs from both cell lines formed spheres with similar efficiency in mock transfected cells; however, CSCs subpopulations formed spheres more efficiently than non-CSCs when parental cells were transfected with the pCMV6-HMOX1 plasmid, indicating that HO-1 overexpression increases the self-renewal capacity of CSCs subpopulations, regardless of p53 status. versus mock transfected cells at 96 h after transfection (Figure 1e). In addition, the percentage of CD44high/CD326high cells increased in the CSCs subpopulation of pCMV6-HMOX1 versus mock transfected cells at 96 h after transfection in the three cell lines studied ( Figure 1f). These results indicate that HO-1 overexpression increases the proportion of cancer cells with a CSC phenotype in the different subpopulations analyzed. Next, we studied the anchorage-independent growth of CSCs and non-CSCs subpopulations extracted from HCT-116 and HCT-116 p53 null after transient overexpression of HO-1 under free-serum conditions. As shown in Figure 2, CSCs and non-CSCs from both cell lines formed spheres with similar efficiency in mock transfected cells; however, CSCs subpopulations formed spheres more efficiently than non-CSCs when parental cells were transfected with the pCMV6-HMOX1 plasmid, indicating that HO-1 overexpression increases the self-renewal capacity of CSCs subpopulations, regardless of p53 status.
It was interesting to find high mRNA and protein content of both HO-1 and ECE-1 in the CSC versus the non-CSCs subpopulation in the three cell lines studied (Figure 3a-c). We also analyzed the mRNA expression of HO-1, ECE-1, and the CSCs markers CD44 and CD133 by quantitative RT-PCR in 150 cases of CRC patients. Tumors showing high expression of CSCs markers (CD133 high CD44 high ) also exhibited higher expression of HO-1 ( Figure 3d) and ECE-1(Figure 3e) than tumors with low expression of these markers (CD133 low CD44 low ). It was very interesting to find that HO-1 and ECE-1 expression highly correlated in both CD133 high CD44 high and CD133 high CD44 high tumors, while no correlations were found between HO-1 and END-1 (Table S3). The status of P53 did not influence any of the results mentioned above ( Figure 3 and Table S3).
Next, we studied whether HO-1 could regulate ET-1 production in CRC. As shown in Figure  In the cohort of CRC patients, HO-1 and ECE-1 expressions positively correlated in all cases (Rp = 0.723, p < 0.001; Figure 4e) in tumors harboring a wild-type p53 (Rp = 0.699, p < 0.001; Figure 4f) and in tumors with mutations in this gene (Rp = 0.755, p < 0.001; Figure 4g). END-1, the first gene implicated in the synthesis of ET-1, did not correlate with HO-1 at any of the conditions mentioned above (Rp = 0.042, p = 0.614; Rp = 0.042, p = 0.905; Rp = 0.093, p = 0.506, for all cases, p53 wild-type tumors, and p53 mutated tumors, respectively). As shown in Table S2, none of these genes correlated with any of the clinicopathological features of the patients included in the study.
We next investigated which of the subproducts of the HO-1 reaction is responsible for ECE-1 activation after HO-1 overexpression in CRC cells. To assess this, we treated cells with 10 µM CORM3 (a CO donor) and 10 µM biliverdin. We also used 5 µM hemin and 1 µM CoPPIX treatments as positive controls. CORM3, hemin, and CoPPIX induced ECE-1 protein expression at the doses used in both HCT-116 and HCT-116 p53 null cells, whereas biliverdin was unable to induce it (Figure 5a). We also studied ET-1 production by cells after treatments at the doses specified above (Figure 5b,c). CORM3 induced a significant increase in ET-1 production at all times analyzed in both cell lines. Treatment with hemin induced increased ET-1 production in HCT-116 cells, although only after 24 h, while this effect appeared after 24, 48, and 72 h in HCT-116 p53 null cells. After showing that CO is responsible for the activation of ECE-1, we next investigated the mechanism by which it carries out this effect. As shown in Figure 5d, CO seemed to induce ECE-1 expression through the activation of pNF-kβ and pc-Jun in HCT-116 and HCT-116 p53 null cells, respectively.
It was interesting to find high mRNA and protein content of both HO-1 and ECE-1 in the CSC versus the non-CSCs subpopulation in the three cell lines studied (Figure 3ac). We also analyzed the mRNA expression of HO-1, ECE-1, and the CSCs markers CD44  all cases (Rp = 0.723, p < 0.001; Figure 4e) in tumors harboring a wild-type p53 (Rp = 0.699, p < 0.001; Figure 4f) and in tumors with mutations in this gene (Rp = 0.755, p < 0.001; Figure  4g). END-1, the first gene implicated in the synthesis of ET-1, did not correlate with HO-1 at any of the conditions mentioned above (Rp = 0.042, p = 0.614; Rp = 0.042, p = 0.905; Rp = 0.093, p = 0.506, for all cases, p53 wild-type tumors, and p53 mutated tumors, respectively). As shown in Table S2, none of these genes correlated with any of the clinicopathological features of the patients included in the study. (e) (f) (g) We next investigated which of the subproducts of the HO-1 reaction is responsible for ECE-1 activation after HO-1 overexpression in CRC cells. To assess this, we treated cells with 10 μM CORM3 (a CO donor) and 10 μM biliverdin. We also used 5 μM hemin and 1 μM CoPPIX treatments as positive controls. CORM3, hemin, and CoPPIX induced ECE-1 protein expression at the doses used in both HCT-116 and HCT-116 p53 null cells, hemin induced increased ET-1 production in HCT-116 cells, although only after 24 h, while this effect appeared after 24, 48, and 72 h in HCT-116 p53 null cells. After showing that CO is responsible for the activation of ECE-1, we next investigated the mechanism by which it carries out this effect. As shown in Figure 5d, CO seemed to induce ECE-1 expression through the activation of pNF-kβ and pc-Jun in HCT-116 and HCT-116 p53 null cells, respectively.

HO-1 Overexpression Induces Stemness in CRC Cell Lines through ECE-1/ET-1 Only in p53 Wild-Type Cells
To study whether HO-1 overexpression induces stemness in our model of CRC in vitro through ET-1 produced by ECE-1, we analyzed the percentage of ALDH1+ cells after HO-1 overexpression in the presence of bosentan, an antagonist of both ETRA and ETRB receptors. In this case, and contrary to what happened in the absence of bosentan, HO-1

HO-1 Overexpression Induces Stemness in CRC Cell Lines through ECE-1/ET-1 Only in p53 Wild-Type Cells
To study whether HO-1 overexpression induces stemness in our model of CRC in vitro through ET-1 produced by ECE-1, we analyzed the percentage of ALDH1+ cells after HO-1 overexpression in the presence of bosentan, an antagonist of both ETRA and ETRB receptors. In this case, and contrary to what happened in the absence of bosentan, HO-1 overexpression was not able to induce an increase in the ALDH1+ subpopulation in HCT-116 cells (Figure 6a), and the percentage of CD133 high /CD44 high /CD326 high cells in the TP was similar in pCMV6-HMOX1 versus mock transfected cells (Figure 6d). However, the percentage of CD44 high /CD326 high cells increased in pCMV6-HMOX1 versus mock transfected cells (Figure 6e). Therefore, HO-1 induces stemness in CRC cells harboring a wild-type p53 trough ET-1. transfected cells (Figure 6d,e), when comparing this result in the absence of bosentan (Figure 1d). According to these results, the blockade of ET-1 receptors induces an increase in the CSCs subpopulation in cells without an active p53 or with a mutated p53 with a gain of function, and ET-1 seems to act as a protective agent in these cells after HO-1 overexpression. The presence of bosentan in the cultured media also influenced the self-renewal capacity of CSCs and non-CSCs subpopulations extracted from transiently transfected HCT-116 and HCT-116 p53 null cells with empty vector (mock) and pCMV6-HMOX1 plasmid. On these working conditions, the capacity of subpopulations to form spheres was similar in mock and pCMV6-HMOX1 transfected HCT-116 cells (Figure 7a). However, CSCs and non-CSCs subpopulations from HCT-116 p53 null cells increased their ability to form spheres (Figure 7b). Considering these results, ET-1 increases the self-renewal capacity in cells with a wild-type p53 and decreases it in cells with an inactive p53 after HO-1 overexpression, and the presence of bosentan eliminates these effects. On the contrary, the addition of bosentan after HO-1 overexpression led to an increase in the percentage of ALDH1+ cells at 72 and 96 h after transfection (Figure 6b) as well as to the appearance of the CD133 high /CD44 high /CD326 high subpopulation in HCT-116 p53 null cells (Figure 6e). In HT-29 cells, the presence of bosentan after HO-1 overexpression did not modify the percentage of ALDH1+ cells, although the CD44 high /CD326 high subpopulation seemed to decrease, while the difference in the percentage of the CD133 high /CD44 high / CD326 high subpopulation increased in pCMV6-HMOX1 versus mock transfected cells (Figure 6d,e), when comparing this result in the absence of bosentan (Figure 1d). According to these results, the blockade of ET-1 receptors induces an increase in the CSCs subpopulation in cells without an active p53 or with a mutated p53 with a gain of function, and ET-1 seems to act as a protective agent in these cells after HO-1 overexpression.
The presence of bosentan in the cultured media also influenced the self-renewal capacity of CSCs and non-CSCs subpopulations extracted from transiently transfected HCT-116 and HCT-116 p53 null cells with empty vector (mock) and pCMV6-HMOX1 plasmid. On these working conditions, the capacity of subpopulations to form spheres was similar in mock and pCMV6-HMOX1 transfected HCT-116 cells (Figure 7a). However, CSCs and non-CSCs subpopulations from HCT-116 p53 null cells increased their ability to form spheres (Figure 7b). Considering these results, ET-1 increases the self-renewal capacity in cells with a wild-type p53 and decreases it in cells with an inactive p53 after HO-1 overexpression, and the presence of bosentan eliminates these effects.

HO-1 Overexpression Induces Resistance to 5-FU Treatment in CRC In Vitro
Given the implication of CSCs on therapy resistance in cancer, we analyzed whether HO-1 overexpression induced this effect on CRC in vitro and the mediation of the ECE-1/ET-1 system on it. The treatment of mock and pCMV6-HMOX1 HCT-116, HCT-116 p53 null, and HT-29 transfected cells with 5-fluorouracil (5-FU) at different doses (0-4 μM) showed that HO-1 overexpression induced resistance to this treatment in HCT-116 cells (Figure 8a). We did not find different 5-FU sensitivity after HO-1 overexpression in HCT-116 p53 null nor in HT-29 cells (Figure 8b,c), at least at the doses of 5-FU used. The presence of bosentan in the culture media sensitized HCT-116 cells overexpressing HO-1 to 5-FU ( Figure 8a); however, no effect was found in HCT-116 p53 null cells (Figure 8b). On the contrary, in HT-29 cells transfected with pCMV6-HMOX1, the presence of bosentan induced resistance to 5-FU (Figure 8c). We could conclude that HO-1 overexpression induces 5-FU resistance in cells with a wild-type p53 and that ET-1 is responsible, at least in part, for this effect. On the contrary, in cells with a non-wild-type p53, HO-1 overexpression seemed to not affect the response to 5-FU treatment at the doses used in this study. In addition, ET-1 increases the response to this treatment in cells with a mutated p53 with a gain of function.

HO-1 Overexpression Induces Resistance to 5-FU Treatment in CRC In Vitro
Given the implication of CSCs on therapy resistance in cancer, we analyzed whether HO-1 overexpression induced this effect on CRC in vitro and the mediation of the ECE-1/ET-1 system on it. The treatment of mock and pCMV6-HMOX1 HCT-116, HCT-116 p53 null, and HT-29 transfected cells with 5-fluorouracil (5-FU) at different doses (0-4 µM) showed that HO-1 overexpression induced resistance to this treatment in HCT-116 cells (Figure 8a). We did not find different 5-FU sensitivity after HO-1 overexpression in HCT-116 p53 null nor in HT-29 cells (Figure 8b,c), at least at the doses of 5-FU used. The presence of bosentan in the culture media sensitized HCT-116 cells overexpressing HO-1 to 5-FU ( Figure 8a); however, no effect was found in HCT-116 p53 null cells (Figure 8b). On the contrary, in HT-29 cells transfected with pCMV6-HMOX1, the presence of bosentan induced resistance to 5-FU (Figure 8c). We could conclude that HO-1 overexpression induces 5-FU resistance in cells with a wild-type p53 and that ET-1 is responsible, at least in part, for this effect. On the contrary, in cells with a non-wild-type p53, HO-1 overexpression seemed to not affect the response to 5-FU treatment at the doses used in this study. In addition, ET-1 increases the response to this treatment in cells with a mutated p53 with a gain of function.

Discussion
In the recent years, numerous research works have tried to elucidate the role of HO-1 in colorectal cancer. This has not been an easy task, since this enzyme is found in both cancer and stromal cells, including immune cells [38][39][40][41]. In addition, HO-1 in CRC cells is able to regulate immune-mediated cytotoxicity against CRC cells [18]. In this complex scenario, the results from the scientific literature have shown contradictory effects of HO-1 in this type of cancer [15][16][17][18][19][20], which highlights the need to continue this field of research. The metabolic status of cancer cells influences how heme-degrading enzymes modulate tumor growth. Conversely, CO and biliverdin can modulate lipid and glucose metabolism [42]. Specifically, CO can promote increased mitochondrial biogenesis and induce an anti-Warburg effect [43], which has been linked to a lower rate of proliferation and cell differentiation [44]. In addition, the presence of HO-1 in cytosol or its translocation into the nucleus can also affect its activity and, therefore, its effects on tumors [42].
We used the CRC cell lines HCT-116, HCT-116 p53 null, and HT-29 to conduct our in vitro model. These cells have the ability to form tumorospheres within 3-5 days after being seeded in appropriate plates and medium [45][46][47], which allowed us to carry out our study even though the transformation of the parental cells was performed by transient transfection [48,49].
In this study, we found that HO-1 regulates the proportion and phenotype of CSCs in CRC, a result that is not surprising considering that we also found a higher expression of this protein in the CSCs subpopulation compared to non-CSCs in cell cultures. In tumors from CRC patients, samples with a higher expression of CSCs markers also showed a higher expression of HO-1. These results agree with previous works that reported a direct implication of HO-1 in regulating CSCs in leukemia, glioma, melanoma, breast, pancreatic, and lung cancers [24,[49][50][51]. According to our results, HO-1 could regulate stemness in CRC independently of p53; however, previous reports have demonstrated an important relationship between them [16,52]. In fact, HO-1 is a p53-dependent target gene that is responsible for p53-dependent cellular survival after oxidative treatment [52]. Other works have suggested that HO activity is involved in the regulation of p53 expression in normal [53] and cancer cells [54].
In our in vitro model, CO produced after HO-1 overexpression led to increased expression of ECE-1 and ET-1 secretion by CRC cells through the activation of NF-κB and AP-1 in cells harboring a wild-type p53 and in cells with a mutated p53, respectively. These transcription factors are essential for ECE-1 activation in endothelial cells [55]. The CO releasing molecule CORM3 has been reported to activate AP-1 in vitro [56,57]. The

Discussion
In the recent years, numerous research works have tried to elucidate the role of HO-1 in colorectal cancer. This has not been an easy task, since this enzyme is found in both cancer and stromal cells, including immune cells [38][39][40][41]. In addition, HO-1 in CRC cells is able to regulate immune-mediated cytotoxicity against CRC cells [18]. In this complex scenario, the results from the scientific literature have shown contradictory effects of HO-1 in this type of cancer [15][16][17][18][19][20], which highlights the need to continue this field of research. The metabolic status of cancer cells influences how heme-degrading enzymes modulate tumor growth. Conversely, CO and biliverdin can modulate lipid and glucose metabolism [42]. Specifically, CO can promote increased mitochondrial biogenesis and induce an anti-Warburg effect [43], which has been linked to a lower rate of proliferation and cell differentiation [44]. In addition, the presence of HO-1 in cytosol or its translocation into the nucleus can also affect its activity and, therefore, its effects on tumors [42].
We used the CRC cell lines HCT-116, HCT-116 p53 null, and HT-29 to conduct our in vitro model. These cells have the ability to form tumorospheres within 3-5 days after being seeded in appropriate plates and medium [45][46][47], which allowed us to carry out our study even though the transformation of the parental cells was performed by transient transfection [48,49].
In this study, we found that HO-1 regulates the proportion and phenotype of CSCs in CRC, a result that is not surprising considering that we also found a higher expression of this protein in the CSCs subpopulation compared to non-CSCs in cell cultures. In tumors from CRC patients, samples with a higher expression of CSCs markers also showed a higher expression of HO-1. These results agree with previous works that reported a direct implication of HO-1 in regulating CSCs in leukemia, glioma, melanoma, breast, pancreatic, and lung cancers [24,[49][50][51]. According to our results, HO-1 could regulate stemness in CRC independently of p53; however, previous reports have demonstrated an important relationship between them [16,52]. In fact, HO-1 is a p53-dependent target gene that is responsible for p53-dependent cellular survival after oxidative treatment [52]. Other works have suggested that HO activity is involved in the regulation of p53 expression in normal [53] and cancer cells [54].
In our in vitro model, CO produced after HO-1 overexpression led to increased expression of ECE-1 and ET-1 secretion by CRC cells through the activation of NF-κB and AP-1 in cells harboring a wild-type p53 and in cells with a mutated p53, respectively. These transcription factors are essential for ECE-1 activation in endothelial cells [55]. The CO releasing molecule CORM3 has been reported to activate AP-1 in vitro [56,57]. The regulation of NF-κB by CO has been previously described in several models of disease [58,59]. This molecule is also able to induce ROS production in vitro [60]. Interestingly, ROS production and NF-κB activation are critical events in CSCs appearance and CRC initiation [61]. Biliverdin was unable to activate the ECE-1/ET-1 pathway in our study. Biliverdin is converted to bilirubin by bilirubin reductase [62]. The ROS scavenging properties of both heme metabolites contribute to their antitumoral activities [11,63]. We did not test the possible role of ferrous ions on ECE/ET-1 induction by HO-1 overexpression. Similarly to bilirubin, literature to date has shown protective effects of ferrous iron against cancer, mainly due to the induction of ferroptosis and cell death [11,41,64,65].
The endothelin system regulates stem properties of cells in CRC. The END-1 gene is highly expressed in CSCs isolated from cultured CRC cells [66]. ET-1 regulates important pathways for the maintenance of the stemness phenotype, such as MAPK, PI3K/Akt, and Wnt/β-catenin [29]. However, given the short half-life of ET-1, its biological effects are totally dependent on the maintenance of a critical concentration by ECE-1 [67]. Of the four described isoforms of ECE-1 (a, b, c and d), ECE-1c has been implicated in CSCs generation in CRC [67]. In our study, we did not analyze which isoform of ECE-1 increased after HO-1 overexpression. However, ETRA and ETRB receptors blockaded with bosentan led to a decrease in the ALDH1+ population, in the percentage of CD133 high /CD44 high /CD326 high cells, and in the resistance to 5-FU treatment in the total population and a decrease in the ability for self-renewal of the CSCs subpopulation of wild-type p53 cells. Previous studies have reported that ETRA receptors are responsible for ET-1-mediated stemness in CRC [29].
On the other hand, treatment with bosentan after HO-1 overexpression induced an increase in the ALDH1+ and the appearance of CD133 high /CD44 high /CD326 high subpopulations in the total population, and it increased the ability to form spheres in the CSCs subpopulation of cells lacking p53. Similarly, in cells harboring a mutated p53, the ETR blockade induced a loss of CD44 high /CD326 high cells and an increase in CD133 high /CD44 high /CD326 high ones. These results could indicate that even though HO-1 overexpression induces increased expression of ECE-1 and ET-1 production regardless of p53, in cells without an active p53 or even a p53 with gain of function, like HT-29 cells, another mechanism that does not involve the ECE-1/ET-1 system is responsible for the increase in the stem subpopulation. More interestingly, the HO-1-induced ECE-1/ET-1 system seems to exert a protective effect in these p53 not-active or with gain of function CRC cells. Contrary to these results, previous reports have shown that treatment with bosentan is proapoptotic and antiproliferative in HT-29 [68] cells and sensitizes them to Fas-L-induced cell death [69,70]. An ET-1independent mechanism of ECE-1 in CRC and other cancers has also been described [67], but in these cases it acts as a protumoral protein [67]. On the contrary, clinical trials with approved endothelin receptor antagonists, including bosentan, were not so promising, with no statistically significant results, even though the drugs were well-tolerated [71]. Although more research is warranted to confirm our results, HO-1 overexpression probably activates or inactivates certain unknown pathways that could induce these contrary results after treatment with bosentan in our cell model.

Conclusions
HO-1 regulates stemness and resistance to 5-FU treatment in CRC in vitro. CO produced after HO-1 overexpression seemed to activate the ECE-1/ET-1 pathway, which could be the final effectors of HO-1 protumoral events in p53 wild-type cells. Surprisingly, in cells with a non-active p53 or a gain-of-function mutated p53, ECE-1-produced ET-1 seems to act as a protective molecule. These results could implicate that patients carrying tumors with high expression of both HO-1 and ECE-1 and a non-wild-type p53 should be considered for HO-1 based-therapies instead of ET-1 antagonists-based ones.