Mutations in TP53 occur commonly in the majority of human tumors and confer aggressive tumor phenotypes, including metastasis and therapy resistance. CB002 and structural-analogs restore p53 signaling in tumors with mutant-p53 but we find that unlike other xanthines such as caffeine, pentoxifylline, and theophylline, they do not deregulate the G2 checkpoint. Novel CB002-analogs induce pro-apoptotic Noxa protein in an ATF3/4-dependent manner, whereas caffeine, pentoxifylline, and theophylline do not. By contrast to caffeine, CB002-analogs target an S-phase checkpoint associated with increased p-RPA/RPA2, p-ATR, decreased Cyclin A, p-histone H3 expression, and downregulation of essential proteins in DNA-synthesis and DNA-repair. CB002-analog #4 enhances cell death, and decreases Ki-67 in patient-derived tumor-organoids without toxicity to normal human cells. Preliminary in vivo studies demonstrate anti-tumor efficacy in mice. Thus, a novel class of anti-cancer drugs shows the activation of p53 pathway signaling in tumors with mutated p53, and targets an S-phase checkpoint.
Tumor suppressor p53 responds to cell stress signals from DNA damage, oncogene activation, oxidative stress, and hypoxia. Upon activation by posttranslational modifications and oligomerization, p53 signals cell cycle arrest, apoptosis, or DNA repair, according to the extent of the cellular stress, thereby controlling cell fate and preventing tumorigenesis (Riley et al., 2008). Thus, it is not surprising that TP53 is the most commonly mutated gene (TCGA, 2020), including in ovarian, colorectal, esophageal, head and neck, lung, and pancreatic cancers that are the most affected sporadic human cancer types (Olivier et al., 2010). TP53 is mutated in over 50% of human cancers and the other 50% involve a biological inactivation of its signaling pathway. Like other tumor suppressors, the mutated p53 protein results in loss-of-function but oligomerization can act in a dominant-negative fashion with regard to the remaining wild-type p53 allele. Unlike other tumor suppressors, mutant p53 protein can also acquire a gain-of-function which contributes to aggressive tumor phenotypes, including enhanced invasion, genomic instability, and therapy resistance (Muller and Vousden, 2014; Dittmer et al., 1993; Lang et al., 2004; Xu et al., 2011). Consequently, patients whose tumors carry p53 mutations have a poor prognosis and decreased overall survival (Wattel et al., 1994).
A common feature of cancer cells is genomic instability due to ineffective cell cycle checkpoint responses. Genomic instability is not necessarily due to defective checkpoints. The checkpoints may be intact but the repair may be deficient. Upon DNA damage, the normal cell cycle checkpoint response is to arrest the cell at the G1-phase. In cancer cells, the majority have an ineffective G1 checkpoint due to p53 mutation but retain a functional G2 checkpoint and thus have the ability to undergo cell arrest at the G2-phase. Cancer cells depend on bypassing intra-S-phase and G2/M checkpoints for unrestrained cell proliferation. Stress signal transduction in the p53 pathway is initiated by activation of kinases ataxia-telangiectasia-mutated (ATM), ataxia telangiectasia and Rad3 (ATR)-related, and downstream checkpoint kinases Chk1/2 that serve as signaling sensors and mediators of p53 activation. It has been a long-standing dogma that ATM/Chk2 and ATR/Chk1 are independently activated but recent studies provide evidence of cross-talk between the kinases (Brown and Baltimore, 2003; Abraham, 2001; Smith et al., 2010). Chk1/2 are kinases that participate in cell cycle checkpoint control, with Chk1 being active in S-phase and G2-phase, whereas Chk2 is active throughout the cell cycle (Smith et al., 2010; Zhao and Piwnica-Worms, 2001; Chehab et al., 2000).
Accumulation of genomic aberrations over time renders cancer cells vulnerable to checkpoint targeting therapy. Since the discovery of checkpoint targets, small molecule inhibitors have been pursued in combination with ionizing radiation and chemotherapy agents in order to deregulate checkpoints, thereby leading to cancer cell death. For example, combination of caffeine, a xanthine derivative, with irradiation or chemotherapy agents was found to deregulate the G2 checkpoint through ATM/ATR inhibition leading to therapy sensitization and enhanced cell death (Russell et al., 1995; Sarkaria et al., 1999). Nonetheless, translational cancer therapeutics studies were discontinued due to unachievable active concentrations in human plasma (Lelo et al., 1986). Thus, for the past two decades, the field has focused on the development of Chk1/2 inhibitors, which are in clinical trials (Fracasso et al., 2011; Huang et al., 2012; Rogers et al., 2020).
Another cancer therapeutic approach we and others have pursued involves restoration of p53 pathway signaling in tumors with mutant p53 or tumors that are null for p53. Despite efforts to restore the p53-pathway, to date, there are no FDA-approved drugs that functionally restore the p53 in tumors with mutated p53. We previously reported a p53-pathway restoring compound CB002 whose mechanism of action was not fully elucidated. We showed that CB002 leads to apoptotic cell death mediated by p53 target Noxa, a pro-apoptotic protein (Hernandez-Borrero et al., 2018). Here, we further evaluated more potent CB002-analog compounds and uncovered a unique mechanism of action suggestive of a novel class of anti-cancer drugs. Based on their molecular structure as xanthine derivatives, the novel class of CB002-analogs, unlike caffeine and other established xanthine derivatives, do not deregulate the G2 checkpoint. By contrast, the novel CB002-analog xanthines perturb S-phase and more importantly they restore the p53-pathway, a property not found with caffeine, pentoxifylline, and theophylline. We sought to characterize and define the new class of small molecules with anti-tumor properties by transcriptomic and proteomic analysis.
We sought to identify more potent analogs of parental xanthine compound CB002. We tested CB002-analogs in the ChemBridge library for the capability to induce the luciferase activity using a p53-regulated luciferase reporter stably expressed in the SW480 colorectal cancer cell line and also determined the IC50 values for the compounds by a CellTiter glow cytotoxicity assay (Figure 1A–B, Figure 1—figure supplement 1). The majority of the CB002-analogs tested, with the exception of analog #12, enhanced p53-reporter activity in a dose-dependent manner within a range of compound concentrations from 0 to 600 μM. We investigated the capability of a set of the CB002-analogs to induce apoptosis as indicated by Propidium Iodide (PI) staining sub-G1 population. As shown in Figure 1C, the treatment of tumor cells with CB002-analogs at an IC50 concentration (100 μM) resulted in a significant increase in sub-G1 content in SW480 cells. Moreover, the most potent CB002-analog #4 was found to increase cleaved-PARP and cytochrome C release from the mitochondria to the cytosol providing further evidence for apoptosis induction in SW480 tumor cells (Figure 1D–E, Figure 1—figure supplement 2). We investigated whether the p53-family member p73 may be a mediator of apoptosis and responsible for inducing p53 transcriptional targets by CB002-analogs.
As we previously showed for CB002 (19), p53-targets Noxa and DR5 were induced independently of p73 and PARP cleavage occurred despite effective p73 knockdown in CB002-analog #4 treated SW480 tumor cells (Figure 1F). Our previously published CB002 data indicated that Noxa plays a key role in mediating CB002-induced apoptosis (19). Thus, we sought to determine if CB002-analogs induce Noxa expression in four human colorectal cancer cell lines. In DLD-1 (p53S241F), SW480(p53R273H,P309S), HCT116(p53WT), and HCT116 p53−/− tumor cells expressing the exogenous R175H p53 mutant, Noxa protein expression was found to be induced, though some variation across cell lines was observed (Figure 1G). As these CB002-analogs are xanthine derivatives, we investigated whether other known xanthine derivatives, that is, caffeine, pentoxifylline, and theophylline can induce Noxa expression. However, we found that only the p53-pathway restoring CB002-analog xanthine compounds and not caffeine, pentoxifylline, and theophylline, induce Noxa protein expression (Figure 1H). Since Noxa can be transcriptionally activated independently of p53, we sought to explore other transcription factors involved in Noxa induction. We performed a knockdown of integrated stress response transcription factors ATF3/4 on SW480 cells. Knockdown of ATF3/4 upon treatment with 100 μM CB002 or 25 μM CB002-analog #4 abrogated Noxa protein induction (Figure 1I). Hence, our data suggests that ATF3/4 play a role in regulating Noxa expression.
In order to understand how the CB002-analog molecules restore the p53-pathway, we performed a transcriptomic and proteomic analysis in SW480 cells treated with analog #4. Raw data from the transcriptomic and proteomic analysis can be found in Supplementary file 1 and Supplementary file 3, respectively. To verify the quality of our transcriptomic data, the principal component (PC) plot was obtained. PC plots show that the factor with most variability within the samples was the difference between control and treatment (Figure 2—figure supplement 1A–C). Significant differentially expressed genes (DEGs) were defined by a false discovery rate (FDR)<0.05, and a total of 3362 genes met these criteria (Figure 2—figure supplement 1D). We then sought to identify the DEGs involved in the p53 pathway. To do this, a comprehensive known p53 target gene set used for comparison were the genes that have been previously shown to be directly regulated by p53 through chromatin immunoprecipitation assays assays and genes that were protein-coding genes in at least 3 of the 17 genome-wide data sets (from Fisher’s analysis; Fischer, 2017). Out of the 343 genes in the known p53 target gene set, 334 genes were tested in the microarray but only 197 genes met the low expression cutoff. From the 197 genes that met the low expression criteria, 102 genes were found to be differentially expressed (Figure 2A, Supplementary file 2). Gene ontology (GO) analysis of the 102 DEGs indicated that these genes are highly enriched in the regulation of programmed cell death (Table 1). A gene expression heatmap of these genes is shown in Figure 2B, and the majority of the genes are found to be upregulated by analog #4 treatment of tumor cells. We then performed a transcription factor analysis of all 3362 DEGs. Transcription factor analysis defined by direct binding of predictive binding motifs revealed E2F transcription factors as having the highest normalized enrichment score (Figure 2C). Because the transcription factor ATF4 was shown to be important for Noxa induction in Figure 1I, we compared a known ATF4 gene set (Table S3 from Wang et al., 2015), along with an E2F gene set (Table S1 from Ren et al., 2002), together with the known p53 gene set and the DEGs in our analog #4 treatment (Figure 2D). The resulting Venn diagram of this comparison shows that both ATF4 and E2F targets genes are not unique to these transcription factors and also share common targets with p53 (~5%). Analyzing the ratio of DEGs to the transcription factor gene set did not show an obvious gene enrichment regulation of one transcription factor (Table 2). Despite p53 not being the top predictive transcription factor in our analysis, ingenuity pathway analysis (IPA) determined p53 to be activated as an upstream regulator with a z-score value of 3.3 and p-value of 2.9×10−34. A Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis for the p53-pathway signaling was obtained with an adjusted p-value (adjp) equal to 1.18×10−1 that despite not reflecting a significant enrichment of the p53 pathway, it indicates the presence of a total of 31 DEGs out of the 52 genes tested and present in the KEGG analysis. Thus, this accounts for 60% of DEGs in the KEGG p53-pathway analysis. DEGs involved in the KEGG analysis fold change is described by color and additional genes not shown in the p53-pathway figure and yet involved in the KEGG analysis are shown as a heatmap (Figure 2—figure supplement 2 and Figure 2—figure supplement 3). In line with the GO terms results, p53 target genes involved in apoptosis such as Noxa, Puma, and DR5 were upregulated by the analog #4 treatment. Taken together, this data indicates that although a large set of genes differentially expressed are not predicted to be directly regulated through direct p53 binding, a subset of these are enriched in the p53-pathway, indicative of p53-pathway restoration.
We determined the enriched pathways in the whole set of DEGs (3362). To this end, a KEGG analysis was performed. The top four enriched pathways that were obtained from the KEGG analysis namely included cell cycle, DNA repair, mismatch repair, and nucleotide excision repair. The adjp for each KEGG pathway was 2.27×10−6, 2.27×10−6, 5.05×10−3, and 2.18×10−2, respectively. The adjp values indicate a significant enrichment score of each pathway. The fold change of DEGs by analog #4 treatment in the KEGG analysis is reflected by the color legend (Figure 2—figure supplement 4, Figure 2—figure supplement 6, Figure 2—figure supplement 8, and Figure 2—figure supplement 10). Additional genes not shown in the pathway KEGG figures and yet involved in the KEGG analysis are shown as a heatmap (Figure 2—figure supplement 5, Figure 2—figure supplement 7, Figure 2—figure supplement 9, and Figure 2—figure supplement 11). GO terms in biological processes also reflected enrichment of genes that participate in cell cycle regulation (Table 3). Taken together, KEGG analysis and GO ontology both reflected the downregulation of genes involved in the G1/S-phase of the cell cycle in CB002-analog treated cells. E2F is responsible for the induction of genes in DNA initiation and replication, such as minichromosome maintenance (MCM) complexes and origin replication complexes (Bracken et al., 2004). The transcriptomic analysis indicates the downregulation of these genes and this suggests the inhibition of E2F transcriptional activity. In addition, downregulation of Cyclin E and Cyclin A genes further confirmed the delay of cells to S-phase. GADD45, a p53-target gene that can induce cell cycle arrest, was upregulated. Further study is necessary in order to validate the direct implication of E2F’s and p53 target genes in the perturbation of the delay in S-phase. Nonetheless, this data suggests that the identified family of small molecules represent a unique mechanism of action that involves S-phase delay perturbation and p53-pathway restoration.
In order to show that the stimulation of the p53 pathway at the transcriptional level was restoring the p53 pathway at the protein level, a comparative label-free quantitative proteomic analysis of SW480 colon cancer cells in response to DMSO and analog #4 (T4) treated for 24 hr was performed. Figure 3—figure supplement 1A and E shows close clustering of protein abundance of each replicate under the same group and variability among the treatments. Volcano plots of fold change versus q-value of the total of 3743 proteins quantified from SW480 cells in response to DMSO, CB002 (CB), and analog #4 (T4) treatments show differentially expressed proteins determined as significant (p<0.05) up and down (Figure 3—figure supplement 1B–D). At the protein level, pathway analysis did not reflect an enrichment in p53 targets (Figure 3A). Consistent with the microarray data, the proteomic pathway analysis of the differentially abundant proteins shows the downregulation of proteins involved in cell cycle regulation (Figure 3B). In particular, CDK4, CKS1B, ERCC6L, MAPK3, and MAX are significantly decreased in analog #4 treatment than in CB002 (Figure 3C).
As the CB002-analog molecules were discovered as p53 pathway restoring compounds, we compared the proteomic data, with the known p53 target gene set used in our transcriptomic analysis (Table S3 from Fisher’s analysis in Fischer, 2017) together with our in-house p53-proteomic database (Tian et al., 2020). Our in-house proteomic database was derived from a comparison of HCT116 versus HCT116 p53−/− cells treated with 5-Fluorouracil (5-FU). Our results show that out of all significantly upregulated expressed proteins, only four overlapped with the known p53 targets and six proteins with our in-house p53-proteomics (Figure 4A). Eleven proteins were found to be downregulated by analog #4 treatment overlapping with the in-house proteomic database and none with the known p53 target data set (Figure 4B). No upregulated or downregulated proteins were found to overlap in all three data sets: analog #4 treatment and both reference databases (Figure 4A–B). Overall, these results suggest that within the proteins tested in the proteomic analysis, those expressed by analog #4 treatment and involved in the p53 pathway were minimal under the performed experimental conditions. Additional proteins validated by Western blots, such as Noxa and DR5, were not detected in the proteomic analysis indicating that the proteomic analysis should be considered as preliminary and warrants further optimization. Moreover, differences were observed at the level of protein expression between parental compound CB002 and its analog #4 both downregulated and to a lesser extent, upregulated proteins (Figure 4—figure supplement 1). This indicates that these small molecules can have different effects in tumor cells, albeit they have >50% homology in their proteomic composition.
Caffeine is a G2 checkpoint deregulator through inhibition of ATM/ATR. Thus, the combination of chemotherapy agents with caffeine results in enhanced cancer cell cytotoxicity. Nonetheless, it was not pursued due to caffeine’s lack of achievable required concentrations in plasma. We investigated whether CB002 and its analogs can deregulate the G2 checkpoint, like caffeine, pentoxifylline, and theophylline. We synchronized SW480 colon cancer cells using double thymidine block, released and treated with CB002-analog compound alone or in combination with etoposide, and probed for key G2/M-phase cell cycle markers. As expected, we observed that etoposide treatment enhances protein expression of pcdc2(Tyr15) and pcdc25c(Ser16) indicating cell cycle arrest due to DNA damage. The combination of etoposide with caffeine resulted in G2-deregulation as indicated by decreased expression of pcdc2(Tyr15) and pcdc25c(Ser16). Similarly, the combination of etoposide with CB002 or CB002-analog #4 showed a decrease in expression of pcdc2(Tyr15) and pcdc25c(Ser16). Nonetheless, CB002 or CB002-analog #4 do not increase M-phase marker pH3(Ser10) as would be expected for a G2-deregulator like caffeine (Figure 5A). This data suggests that CB002 and CB002-analog #4 either do not deregulate the G2 checkpoint or that these compounds delay cells going into M-phase. Moreover, CB002 and its analogs increase p-Cdc25c and p-Cdc2 in combination with etoposide indicating cell cycle arrest. A similar experiment was performed as a time course after cell synchronization release to further elucidate the cell cycle effects of CB002-analog #4. As seen in Figure 5D, cell cycle markers pcdc2(Tyr15) and pcdc25c(Ser16) expression decreased in CB002-analog #4 compared to DMSO and etoposide and their expression over time increased at 12 hr indicative of a delay of cells in the G2 cell cycle phase. To further elucidate the effect in S-phase, we evaluated Cyclin A and p-RPA-RPA2(S8), the latter as a marker of single-stranded DNA and replication stress that are potentially caused by stalled or collapsed replication forks. Cyclin A expression did not decrease over time in CB002-analog #4 treated cells as compared to DMSO and etoposide indicating that cells were delayed in S-phase. Moreover, p-RPA-RPA2(S8) expression upon CB002-analog #4 treatment was increased compared to DMSO indicating replication stress. The p53 target p21 was also found to increase in CB002-analog #4 treated cells indicating cell cycle arrest. Taken together, these analogs deregulate an S-phase checkpoint and not a G2 checkpoint.
To investigate further the effects of these CB002-analogs on the cell cycle, we probed for S-phase specific markers and performed PI analysis by flow cytometry upon release of synchronized cells for a time course of 0–48 hr. CB002 and its structural analogs, unlike caffeine, increase single-strand DNA marker p-RPA-RPA2(S8) and p-ATR(Thr1989), indicating that these compounds result in replication stress and activate features of an S-phase checkpoint (Figure 5B–C). PI analysis further confirms that combination of caffeine and etoposide deregulates the G2 checkpoint and that CB002-analogs #4 and #10 treatment results in S-phase accumulation are particularly observed at 8 hr following release from synchronization (Figure 5E). PI and BrdU co-staining confirm that CB002-analog #4 increases by 30% cells in S-phase at 12 hr as compared to DMSO vehicle control and no significant differences are observed in G2-phase cells between etoposide and CB002-analog #4 at 24 hr (Figure 5F). S-phase delays with CB002 and CB002-analog #10 occur at 6–8 hr of treatment, particularly a twofold difference in combination with etoposide. The caffeine-treated S-phase population is comparable to the DMSO vehicle control at all time points indicating that caffeine does not perturb the S-phase. As expected, caffeine decreases the G2-population by 2- to 3-fold at 24 hr in combination with etoposide as compared to etoposide alone, and no other treatment tested decreases the G2-population when combined with etoposide (Figure 5 and Figure 5—figure supplement 1). Haploid cell gating indicates the haploid BrdU-positive cells in Figure 5 and Figure 5—figure supplement 1.
We focused on lead CB002-analog #4 and investigated its therapeutic index in vitro and in vivo. We treated an isogenic HCT116 cell line panel with varying p53 mutation-status were treated with 100 μM CB002 and 25 μM CB002-analog #4 and established IC50 values by the Cell-Titer glow cytotoxicity assay. Across this panel, CB002-analog #4 has a 20- to 30-fold range in IC50 values, independently of the HCT116 p53-status (Figure 6A). Thus, the results indicate that the restoration of the p53-pathway by CB002 or analog #4 is p53-independent. SW480 cells treated with CB002-analog #4 showed a significant increase of sub-G1 content as compared to vehicle control, whereas treatment with CB002-analog #4 of normal human WI38 lung fibroblast cells did not significantly increase the sub-G1 cell population indicating that CB002-analog #4 is safe to normal cells in vitro (Figure 6B).
We further investigated the anti-cancer cytotoxicity potential of CB002-analog #4. We treated a colorectal cancer patient-derived organoid with CB002-analog #4 and performed cellular cytotoxicity analysis in vitro and immunofluorescence staining of ethidium homodimer, calcein, caspase-3, and Ki-67 to distinguish between dead, live, apoptotic, and proliferating cells, respectively. CB002-analog #4 enhances cytotoxicity as compared to the CB002 parent compound in the tested colorectal cancer patient-derived organoid as indicated by the cell viability response curve (Figure 6C). Moreover, the immunofluorescence assay staining for ethidium homodimer and calcein shows an increase of ethidium homodimer staining of CB002 and CB002-analog #4 to a larger extent as compared to vehicle control indicating an enhanced killing of cells. Calcein staining shows that organoids treated with CB002-analog #4 are smaller in size indicating that CB002-analog #4 decreases the growth of the patient-derived organoid (Figure 6D). Cleaved caspase-3 staining indicates that both CB002 and CB002-analog #4 treatment at IC50 doses increases apoptotic cells (Figure 6D). CB002-analog #4 treatment also results in an inverse relationship with Ki-67 staining with respect to drug concentration, indicating that CB002-analog #4 decreases the population of proliferating cells (Figure 6E).
We investigated CB002-analog #4 in vivo for anti-tumor efficacy as well as toxicity in NSG mice. Mice were xenografted with human SW480 colorectal cancer cells treated with CB002-analog #4 at 50 mg/kg by oral gavage three times per week. Our data suggests that CB002-analog #4 is well tolerated as indicated by the mouse body weights during the duration of the experiment (Figure 6F). At 5 weeks of treatment, CB002-analog #4 treated tumors have a statistically significant lower tumor volume as compared to vehicle control (Figure 6G). To determine the importance of Noxa in vivo, mice were xenografted with SW480 cells containing a stable knockdown of Noxa. Mice xenografted with SW480 shNoxa cells did not have a significant difference in tumor volume after CB002-analog #4 treatment compared to vehicle control treated tumors indicating that Noxa is important for reduced tumor volume in vivo (Figure 6H).
We describe a novel class of anti-tumor agents with a unique mechanism of action involving restoration of p53 pathway signaling, independently of p53, in tumors with mutated-p53 and characteristics of an S-phase checkpoint. The defining members of this class that best exemplify the novel mechanistic properties are CB002-analogs #4 and #10. The properties of these CB002-analog xanthine compounds are different from other xanthines, such as caffeine, pentoxifylline, and theophylline, that do not restore p53 pathway signaling in tumors with mutant p53 and which deregulate a G2 checkpoint rather than induce an S-phase checkpoint.
Our approach to discovering p53 pathway restoring compounds involved cell-based screening for functional restoration of p53-regulated reporter activity, coupled with cell death induction. Thus, small molecule lead compounds and structural-analogs were not expected to act directly on mutant p53 or restore binding of mutant p53 to genes normally regulated by p53. In the case of the compounds described here, activation of p53 target genes such as Noxa or DR5 occurred independently of p53 and this was observed in tumor cells with different p53 mutations. Thus, there is no expectation that CB002 or analogs #4 or #10 will cause mutant p53 to bind to DNA or chromatin in the regulatory regions of Noxa or DR5 in a manner that wild-type p53 does. Moreover, the induction of p53 targets occurred independently of p53 family member p73, but in a manner that requires integrated stress response transcription factor proteins ATF3/4. These results provide a molecular mechanism for activation of p53 target genes in a manner that substitutes transcription factors such as ATF3/4 for defective p53. This mechanism results in tumor suppression through induction of pro-apoptotic factors despite p53 mutation, and therefore acts as a bypass mechanism to prevent tumor growth in drug-treated cells.
CB002-analog #4 is 20–30 times more potent and like the CB002 parental compound restores the p53-pathway and induces apoptosis independently of p73. The 12 p53 pathway restoring structural analogs of CB002 tested were similar in that they resemble the structure of a xanthine. Our transcriptional analysis identified 102 genes involved in the p53-pathway and IPA determined p53 to be activated as an upstream regulator with a z-score value of 3.3 and p-value of 2.9×10−34. This data further validates the novel anti-cancer class of small molecules as p53-restoring drugs. Microarray analysis identified approximately 150 genes involved in cell cycle regulation, DNA synthesis, and repair that are significantly decreased compared to DMSO control. These genes include, minichromosome maintenance (MCM) proteins, Cyclin E, CDK, E2F, and Cdc2 (Figure 2—figure supplements 4–11). Proteomic analysis also confirmed a decrease in proteins involved in cell cycle regulation (Figure 3B). Thus, our transcriptomic and proteomic analyses coincide in that CB002-analog #4 significantly reduces key regulators of the cell cycle. Taken together with the fact that known xanthines such as caffeine deregulate the G2 checkpoint, we examined the effects of the CB002-analogs on the cell cycle. Our data indicate that the p53-restoring CB002-analog compounds, unlike known xanthines such as caffeine, pentoxifylline, and theophylline, restore the p53 and do not deregulate the G2 checkpoint. Instead, treatment with these small molecule CB002-analogs results in activation of an S-phase DNA damage response pathway characterized by the increase in p-ATR(Thr1989) and we suggest this ultimately leads to a delay of cells in S-phase and this S-phase perturbation may contribute to cancer cell death. Importantly, the observed S-phase perturbation may lead to new therapeutic regimens such as synthetic lethality in BRCA-deficient cells and combination with PARP inhibitors.
We previously reported that pro-apoptotic protein Noxa plays a critical role in CB002-mediated cell death. Our data shows that CB002-analogs induce Noxa expression across different colorectal cancer cell lines in vitro. More importantly, we show that Noxa appears to be critical in vivo as CB002-analog #4 treatment of SW480 shNoxa tumors does not significantly reduce tumor volume as compared to vehicle control. We have evidence indicating that ATF3/4 play a role in regulating Noxa as knockdown of ATF3/4 results in the decrease of Noxa protein expression. Our proteomic data shows activation of the integrated stress response as indicated by the increase of genes involved in the unfolded protein response, tRNA aminoacylation, and increase of ATF3/4 protein expression by Western blot (Figure 3A, Figure 1I). Whether the S-phase perturbation is a result of cellular stress remains to be addressed.
ATF3/4 can regulate similar targets of that of p53, including p21. Our laboratory has identified a small molecule PG3-Oc which involves the restoration of the p53 pathway independently of p53 through ATF4 (Tian et al., 2021). P53 has been shown to indirectly repress many cell cycle genes through the induction of p21. P21, in turn, binds to the DREAM repressor complex which represses genes controlled by E2Fs and CHR transcription factors (Fischer et al., 2016; Engeland, 2018). We observed many cell cycle genes downregulated at the transcriptional level that are relevant to the p53 signal pathway. Moreover, our bioinformatic analysis predicted E2Fs as one of the transcription factors. We have previously shown that CB002 induces p21 expression (Hernandez-Borrero et al., 2018), as well as analog #4 in this study thus it is possible that the observed S-phase perturbation is through p53-independent p21 stimulation that binds to DREAM complexes. Therefore, it will be interesting to see if ATF3/4 regulate p21 expression and the effect of p21 knockdown on cell cycle genes and affect the S-phase perturbation observed by CB002-analogs.
We show that CB002-analog #4 induces apoptosis in colorectal cancer patient-derived organoid cells and that it is safe both in vitro and in vivo as indicated by the lack of a statistically significant increase in the sub-G1 population in normal human fibroblasts and also a healthy NSG mice body weight throughout treatment, respectively. The observed decrease in tumor volume was statistically significant at 5 weeks. This effect was suboptimal than desired and further optimization will be required to reach optimal effects. Importantly, the decrease in tumor volume by CB002-analog #4 is dependent on Noxa. As Noxa is not commonly mutated in human cancer, its induction by the CB002-analogs offers a feasible therapeutic advantage leading to tumor cell death and its expression may be used as a pharmacodynamic biomarker to predict therapeutic response. Taken together, our data suggests that CB002-analogs #4 and #10 represent a novel class of anti-tumor agents that provide a unique therapeutic strategy that can be clinically translated.
CB002 structural analogs were obtained from ChemBridge Library and screening was performed in the human SW480 colorectal cancer cell line that stably expresses a p53-regulated luciferase reporter previously generated in our laboratory (Wang et al., 2006). Cells were seeded at a density of 1×104 cells per well in 96-well plates (Greiner Bio-One) and treated with the indicated compound from 0 to 600 μM. p53 transcriptional activity was imaged using an IVIS imaging system at 6 hr. A total of three biological replicates per condition were performed.
DLD-1 (p53S241F) (RRID:CVCL_0248), SW480 (p53R273H,P309S) (RRID:CVCL_0546), and HCT116 (p53WT) (RRID:CVCL_0291) colorectal cancer cell lines and WI38 normal lung fibroblast cells were purchased from ATCC. HCT116 p53−/− (obtained from the Vogelstein Laboratory, Johns Hopkins University), HCT116 R175H p53, and HCT116 R273H p53 were previously described (Hernandez-Borrero et al., 2018). The SW480 cancer cell line that stably expresses a p53-regulated luciferase reporter was previously generated in our laboratory (Ren et al., 2002). Cell lines were authenticated and tested for mycoplasma. Cell lines were maintained in HyClone Dulbecco’s High Glucose Modified Eagles Medium (DMEM, GE Healthcare), HyClone McCoy’s 5A (GE Healthcare) or Eagle’s Minimum Essential Medium (EMEM, ATCC) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (complete media) at 37°C in 5% CO2, as recommended by ATCC.
SW480 cells were seeded in 96-well plates at a density of 5×103 cells per well. A total of three biological replicates per condition were performed. About 20 μL of CellTiter-Glo reagent was added directly to the wells, according to the manufacturer’s protocol, and bioluminescence signal was determined using an IVIS imaging system at a period of 48–72 hr after treatment.
Where indicated, cells were synchronized by double thymidine block. Cells were treated with 2 μM Thymidine for 16 hr, drug was removed and replaced by complete growth media for 8 hr. Cells were treated for the second time with 2 μM Thymidine for 16 hr, at this point, cells were treated and harvested as indicated.
Cells were seeded at a density of 5×105 in a six-well plate and treated for 48–72 hr. A total of two biological replicates per condition were performed. After treatment, floating cells were collected and adherent cells were trypsinized, pelleted, washed with phosphate-buffered saline (PBS) and fixed in 70% ethanol overnight. For PI based sub-G1 apoptosis analysis, cells were spun down after fixation and resuspended in phosphate-citric acid buffer (0.2 M Na2HPO4+0.1 M citric acid, pH 7.8) at room temperature for 5 min. The cell pellet was resuspended for staining with 50 μg/mL PI and 250 μg/mL ribonuclease (RNase A). For BrdU Chase analysis, a final concentration of 10 μM BrdU (Sigma-Aldrich, B9285) was added to the cell culture for 30 min at 37°C prior to cell fixation. Cells were fixed, spun down, and resuspended in 1 mL of 2 N HCL with 0.5% Triton X-100 for 30 min at room temperature. Cells were pelleted, washed with PBS, and resuspended in 20 μL BrdU anti-body (BD Biosciences, cat no. 347580) diluted in 0.5% Tween 20/PBS/5% BSA for 30 min at room temperature. Cells were then spun down and resuspended in 140 μg/mL goat anti-mouse Alexa Fluor 488 (#A-11008, Thermo Fisher Scientific) in 0.5% Tween 20 in PBS/5% BSA for 30 min at room temperature. Cells were then spun down and resuspended in 5 μg/mL PI: 250 μg/mL RNase A solution. Samples were analyzed on an Epics Elite flow cytometer (Beckman Coulter).
For BrdU analysis gating, cell aggregates were gated out in the PI Peak versus DNA PI histogram. BrdU lower limit intensity was set on upper limit of the negative control. No BrdU antibody in Figure 3E and no goat anti-mouse Alexa Fluor 488 antibody in Figure 5—figure supplement 1 were used as the negative controls. Haploid cell gating indicates the haploid BrdU-positive cells. S-phase and G2-phase boundaries were determined by PI staining that indicated G1 and G2 as per DNA content. Gating was held constant throughout the samples within a given experiment.
After treatment, floating cells were collected and adherent cells were trypsinized, washed with PBS, and lysed with RIPA buffer (Sigma-Aldrich) for 30 min to 1 hr at 4°C. Protein lysates were pelleted and supernatant was collected. Total protein per sample was determined using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Proteins were denatured using 1× NuPAGE LDS sample buffer (Thermo Fisher Scientific) and reduced with 2-Mercaptoethanol (Sigma-Aldrich). Protein lysates were boiled for 15 min at 95°C. After protein normalization, samples were loaded into NuPAGE Novex 4–12% Bis-Tris Protein Gels (Thermo Fisher Scientific) and gel electrophoresis was performed with NuPAGE MES SDS Running Buffer, with the exception of ATR that was ran using a NuPAGE Novex 3–8% Tris-Acetate Protein Gels (Thermo Fisher Scientific) and NuPAGE Tris-Acetate SDS Running Buffer. Proteins were transferred onto an Immobilon-P membrane (PVDF, EMD Millipore) using a Bio-Rad system with a 10% Tris-Glycine and 10% methanol transfer buffer diluted in distilled and deionized water. Membranes were blocked with 10% milk in TBST solution and then incubated overnight with primary antibody, washed with TBST, and incubated with secondary antibody for 1 hr. Primary antibody incubations were performed in 5% milk or 5% BSA in TBST solution as per manufacturer instructions. Signal was detected using a Chemiluminescent Detection Kit, followed by autoradiography. The following antibodies were used: cytochrome C (1:1000; #sc-13560; Santa Cruz; RRID:AB_627383), p53 (1:1000; #sc-126; Santa Cruz; RRID:AB_628082), p73 (1:1000; #A300-126A; Bethyl Laboratories), Noxa (1:250; #OP180; EMD Millipore; RRID:AB_564933), DR5 (1:1,000; #3696; Cell Signaling Technology; RRID:AB_10692107), cleaved PARP (1:1,000; #9546; Cell Signaling Technology; RRID:AB_2160593), ATF3 (1:1000, #sc-188, Santa Cruz; RRID:AB_2258513), ATF4 (1:1,000; #11815; Cell Signaling Technology), p-RPA32/RPA2(Ser8) (1:1,000; #54762, Cell Signaling Technology), RPA32/RPA2 (1:1000; 52448; Cell Signaling Technology), p-cdc2(Tyr15) (1:1000; #9111; Cell Signaling Technology), cdc2 (1:1000; #54; Santa Cruz Biotechnology), p-cdc25c(Ser216) (1:1000; #9528; Cell Signaling Technology; RRID:AB_2075150), cdc25c (1:1000; #13138; Santa Cruz Biotechnology; RRID:AB_627227), p-H3(Ser10) (1:1000; #3377; Cell Signaling Technology; RRID:AB_1549592), H3 (1:1000; #14269; Cell Signaling Technology; RRID:AB_2756816), γ-H2AX(Ser139) (1:1000; #2577; Cell Signaling Technology; RRID:AB_2118010), p-ATR(Thr1989) (1:1000; GTX128145, GeneTex; RRID:AB_2687562), ATR (1:1000; #1887; Santa Cruz Biotechnology; RRID:AB_630893), Cyclin A (1:1000; sc-271682, Santa Cruz Biotechnology;), p21 (1:200; #OP64; EMD Millipore; RRID:AB_2335868), Ran (1:10000; #610341; BD Biosciences; RRID:AB_397731), and β-actin (1:10000, A5441, Sigma-Aldrich; AB_476744).
A total of 1×105 cells/well were plated per well in a 12-well plate in a medium with 10% FBS without antibiotic. Forward transfection of p73 siRNA (s14319, Ambion), ATF3 siRNA (sc-29757), and ATF4 (sc-35112) was performed using the Lipofectamine RNAiMAX Transfection Reagent (Life Technologies) and incubated for 48 hr before drug treatments.
SW480 cells were seeded at a density of 1×106 in 10 cm dishes and once adhered, treated with DMSO vehicle control or CB002-analog #4 for a total of two biological replicates per condition. Floating cells were collected and adherent cells were trypsinized at 12 hr of treatment. Cells were pelleted and RNA was isolated using a Quick-RNA MiniPrep (#R1055, Zymo Research) according to the manufacturer’s instructions. RNA quality was tested using an Agilent Bioanalyzer RNA Kit. Once RNA quality was sufficient, RNA was amplified and labeled using the Low RNA Input Linear Amplification Kit (Agilent). Labeled cDNA was hybridized onto Affymetrix Human Gene 2.0 ST array. Significant changes in gene expression were determined as follows: the low expression cutoff of probe signal intensity was set at 50 (unless at least one sample did not meet these criteria for that particular probe). Normalization was performed using the RMA method and Limma eBayes for the statistical method using R studio programming software. Genes with an FDR of <0.05 were determined as significant in DMSO vehicle control versus analog #4.
SW480 cells were seeded at a density of 1×106 in 10 cm dishes and treated with DMSO vehicle control or CB002-analog #4 for 24 hr. A total of three biological replicates per condition were performed. Floating cells were collected and adherent cells were trypsinized. Cells were spun down, wash with PBS, and pelleted cells were flash frozen with liquid N2 and subjected for for LC-MS/MS analysis.
Briefly, cell pellets were lysed with a lysis buffer (8 M urea, 1 mM sodium orthovanadate, 20 mM HEPES, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, pH 8.0, 20 min, 4°C) followed by sonication at 40% amplification by using a microtip sonicator (QSonica, LLC, Model no. Q55) and cleared by centrifugation (14,000×g, 15 min, 15°C). Protein concentration was measured (Pierce BCA Protein Assay, Thermo Fisher Scientific) and a total of 100 μg of protein per sample was subjected for trypsin digestion. Tryptic peptides were desalted using C18 Sep-Pak plus cartridges (Waters, Milford, MA) and were lyophilized for 48 hr to dryness. The dried peptides were reconstituted in buffer A (0.1 M acetic acid) at a concentration of 1 μg/μL and 5 μL was injected for each analysis.
The LC-MS/MS was performed on a fully automated proteomic technology platform that includes an Agilent 1200 Series Quaternary HPLC system (Agilent Technologies, Santa Clara, CA) connected to a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific, Waltham, MA). The LC-MS/MS setup was used as described earlier (Ahsan et al., 2017). Briefly, the peptides were separated through a linear reversed-phase 90 min gradient from 0% to 40% buffer B (0.1 M acetic acid in acetonitrile) at a flow rate of 3 μL/min through a 3 μm 20 cm C18 column (OD/ID 360/75 μm, Tip 8 μm, New objectives, Woburn, MA) for a total of 90 min run time. The electrospray voltage of 2.0 kV was applied in a split-flow configuration, and spectra were collected using a top 9 data-dependent method. Survey full-scan MS spectra (m/z 400–1800) were acquired at a resolution of 70,000 with an AGC target value of 3×106 ions or a maximum ion injection time of 200 ms. The peptide fragmentation was performed via higher-energy collision dissociation with the energy set at 28 normalized collision energy. The MS/MS spectra were acquired at a resolution of 17,500, with a targeted value of 2×104 ions or maximum integration time of 200 ms. The ion selection abundance threshold was set at 8.0×102 with charge state exclusion of unassigned and z=1, or 6–8 ions and dynamic exclusion time of 30 s.
Peptide spectrum matching of MS/MS spectra of each file was searched against the NCBI Human database (TaxonID: 9606, downloaded on 02/19/2020) using the Sequest algorithm within Proteome Discoverer v 2.4 software (Thermo Fisher Scientific, San Jose, CA). The Sequest database search was performed with the following parameters: trypsin enzyme cleavage specificity, two possible missed cleavages, 10 ppm mass tolerance for precursor ions, and 0.02 Da mass tolerance for fragment ions. Search parameters permitted dynamic modification of methionine oxidation (+15.9949 Da) and static modification of carbamidomethylation (+57.0215 Da) on cysteine. Peptide assignments from the database search were filtered down to a 1% FDR. The relative label-free quantitative and comparative among the samples were performed using the Minora algorithm and the adjoining bioinformatics tools of the Proteome Discoverer 2.4 software. To select proteins that show a statistically significant change in abundance between two groups, a threshold of 1.5-fold change with p-value (0.05) was selected.
30,000 cells/well were seeded in eight-chamber slides. Cells were washed with PBS at the harvesting time point and fixed with 4%parafornaldehyde for 25 min. Cells were then washed with PBS and permeabilized with 0.2% Triton X-100 for 5–10 mins. Cells were then washed with PBS and incubated overnight 1:100 with the indicated primary antibody cytochrome C (#sc-13560; Santa Cruz; RRID:AB_627383), Tom-20 (#42406, Cell Signaling Technology; RRID:AB_2687663), cells were washed with PBS and incubated with secondary antibody 1:200 goat anti-mouse Alexa Fluor 488 (#A-11008, Thermo Fisher Scientific; RID:AB_143165) and Cy3 AffiniPure Donkey anti-rabbit (#711-165-152, Jackson Immuno Research) for 1 hr followed by PBS washed, 1:400 DAPI staining, washed with PBS and imaged. Organoid viability imaging was determined by CellTrace Calcein Green (#C34852, Thermo Fisher Scientific), Ethidium Homodimer-1 (#E1169, Thermo Fisher Scientific), Ki-67 (#9449; Cell Signaling Technology; RRID:AB_2797703) incubated at 37°C for 1 hr then washed with PBS and imaged. Imaging was done using a Leica Confocal Microscope. Experiments were performed at least twice and more than three technical replicates were obtained, a representation of one is shown.
In vivo drug efficacy studies were performed on 10 NSG (RRID:IMSR_JAX:005557) female randomized mice per cohort. Mice tested negative for pathogens listed on Indexx Bioanalytics Laboratory IMPACT I testing including Mycoplasma spp., Mycoplasma pulmonis, mouse hepatitis virus, pneumonia virus, murine norovirus, sendai virus, and Corynebacterium bovis. Tumor inoculation was induced by subcutaneous injection in the left and right dorsal flank, each with a 150 μL suspension of 1–5×106 human colon cancer cells in PBS with Matrigel (1:1). Once tumor size reached 100 mm3, mice were treated 3×/week with DMSO vehicle or compound #4 via oral gavage (22 gauge 1 in. needle) in a solution of 10% DMSO, 20% Kolliphor EL (Sigma-Aldrich, cat. no. C5135) and 70% PBS. Mouse weight and tumor measurements were recorded 1–2 times per week. Tumor volume was calculated as V=0.5*L*W^2, were L is length and W is width of the tumor. At the end of the experiment, mice were euthanized by CO2. All in vivo procedures were performed according to an approved Institutional Animal Care and Use Committee (IACUC) protocol #14–17 at Fox Chase Cancer Center.
To assess the statistical significance, two-way ANOVA or unpaired t-test for two comparisons was performed, with p<0.05 defined as statistically significant. Data are presented as means ± SEM (three biological replicates). Comparisons were made against the DMSO vehicle control.
WSE-D is a Founder of p53-Therapeutics, Inc, a biotech company focused on developing small molecule anti-cancer therapies targeting mutant p53. WSE-D has disclosed his relationship with p53-Therapeutics and potential conflict of interest to his academic institution/employer and is fully compliant with NIH policies and institutional policies that is managing this potential conflict of interest.
All data generated or analysed during this study are included in the manuscript.
Cell cycle checkpoint signaling through the ATM and ATR kinasesGenes & Development 15:2177–2196.https://doi.org/10.1101/gad.914401
Essential and dispensable roles of ATR in cell cycle arrest and genome maintenanceGenes & Development 17:615–628.https://doi.org/10.1101/gad.1067403
Chk2/hCds1 functions as a DNA damage checkpoint in G(1) by stabilizing p53Genes & Development 14:278–288.https://doi.org/10.1101/GAD.14.3.278
Cell cycle arrest through indirect transcriptional repression by p53: i have a DREAMCell Death & Differentiation 25:114–132.https://doi.org/10.1038/cdd.2017.172
A Phase 1 study of UCN-01 in combination with irinotecan in patients with resistant solid tumor malignanciesCancer Chemotherapy and Pharmacology 67:1225–1237.https://doi.org/10.1007/s00280-010-1410-1
Discovery of a novel series of CHK1 kinase inhibitors with a distinctive hinge binding modeACS Medicinal Chemistry Letters 3:123–128.https://doi.org/10.1021/ml200249h
Assessment of caffeine exposure: caffeine content of beverages, caffeine intake, and plasma concentrations of methylxanthinesClinical Pharmacology and Therapeutics 39:54–59.https://doi.org/10.1038/clpt.1986.10
TP53 mutations in human cancers: origins, consequences, and clinical useCold Spring Harbor Perspectives in Biology 2:a001008.https://doi.org/10.1101/cshperspect.a001008
Transcriptional control of human p53-regulated genesNature Reviews Molecular Cell Biology 9:402–412.https://doi.org/10.1038/nrm2395
The ATM-Chk2 and ATR-Chk1 pathways in DNA damage signaling and cancerAdvances in cancer research 108:73–112.https://doi.org/10.1016/B978-0-12-380888-2.00003-0
Gain of function of mutant p53 by coaggregation with multiple tumor suppressorsNature Chemical Biology 7:285–295.https://doi.org/10.1038/nchembio.546
ATR-mediated checkpoint pathways regulate phosphorylation and activation of human Chk1Molecular and Cellular Biology 21:4129–4139.https://doi.org/10.1128/MCB.21.13.4129-4139.2001
Mone ZaidiSenior and Reviewing Editor; Icahn School of Medicine at Mount Sinai, United States
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
In this paper, a sub-group of xanthine analogues with single agent anti-tumor properties has been recognized to act through a unique mechanism of action involving significant restoration of the p53 pathway transcriptome, independently of p53, in tumors with mutated-p53 and the compounds trigger features of an S-phase checkpoint. The properties of this sub-class of xanthine compounds are different from classical xanthines such as caffeine, pentoxifylline, and theophylline that do not restore p53 pathway signaling in tumors with mutant p53 and which deregulate a G2-checkpoint rather than induce an S-phase checkpoint. The novel mechanism appears to involve transcription factors in the integrated stress response such as ATF3 and ATF4 leading to p53 and p73-independent pro-apoptotic Noxa upregulation and tumor cell death.https://doi.org/10.7554/eLife.70429.sa1
[Editors' note: we include below the reviews that the authors received from another journal, along with the authors’ responses.]
This manuscript has been reviewed by three experts in the field, and all have similar concerns with lack of clarity of some of the figures and figure legends, and more particularly with lack of clarity as the mechanism of action of some of the compounds, and the conclusion that the compounds are lethal due to p53 pathway reactivation. Unfortunately the recommendation of all three Reviewers was reject. I know this decision will be disappointing for the authors, but it is my hope that these comments are useful to them as they plan next steps; please discuss with them the possibility of addressing these concerns and possibly transferring with reviews to MCT.
We believe the manuscript previously submitted was misinterpreted by the reviewers. We never stated that the small molecules in this study activated p53, were p53 dependent, or rescued the mutant p53 to wild-type p53 activity. Instead, we claimed that these small molecules restore the p53 pathway by bypassing direct p53 activation. Our approach to discovering p53 pathway restoring compounds has over the last two decades involved live cell-based screening for functional restoration of p53-regulated reporter activity, coupled with cell death induction as a function of dose and time when we have tested chemical libraries. Thus, small molecule lead compounds and structural-analogues are not for the most part expected to act directly on mutant p53 or restore binding of mutant p53 to genes normally regulated by p53. Thus, there is no expectation that CB002 or its analogues will cause mutant p53 to bind to DNA or chromatin in the regulatory regions of p53 target genes in a manner that wild-type p53 does.
We showed that the effect of parental compound CB002 and its more potent analogue #4 IC50 values were similar in an isogenic panel of HCT116 cells (p53 null, p53 wild-type and p53 mutant), consistent with the fact that these small molecules are p53-independent. We describe the small molecules as compounds that restore the p53 signaling pathway through the involvement of other pathways. We showed that the integrated stress response ATF3 and ATF4 transcription factors play a role in the stimulation of pro-apoptotic p53 target gene, Noxa, a key mediator of apoptosis by these small molecules. Using the published Fischer p53 data-set (343 genes), we detected 197 genes and out of those, 102 were differentially expressed by analogue #4 indicating that nearly 50% of the established p53 target genes were altered.
Reviewer #1 (Reviewer Comments to the Author):
In this manuscript, Hernandez Borrero et al. have extended their studies of the compound CB002 that they initially published on in Cell Cycle in 2018. In that paper they showed that this compound causes p53R175H degradation and a cell death mechanism that was dependent on Noxa. Here they attempt to elucidate this mechanism further as well as test analogues of the molecule that were synthesized in an attempt to improve potency. They show using a small number of cell lines induce Noxa independently of p73. They also show by transcriptomic and proteomic analysis that the compounds inhibit the cell cycle. Cell cycle analysis shows that compound cause an arrest in S-phase. Overall the paper has a number of major and minor problems that cause it not to be suitable for publication in the Journal. The biggest issue is the claim that the compounds restore the p53 signaling when the proteomic and transcriptomic data only weakly demonstrate that this. It s not clear if they are trying to claim that the compounds restore wild type function to mutant p53 by restoring site specific DNA binding (though it seems the compound activity is independent of p53). Another conclusion one can draw from this data is that the compounds have anti-cancer properties that kill cancer cells (more so than non-cancer cells) by an apoptotic mechanism. They also arrest cells in the Sphase of the cell cycle. Much of the transcriptomic and proteomic data are poorly annalzyed and poorly reported.
The structures of compounds are missing as are the rationale and depiction of the synthesis scheme.
We thank the reviewer for their comments. We would like to clarify that the analogues tested were not synthesized by our laboratory but rather were commercially available and categorized as analogues within the commercially available Chembridge library, where the parental CB002 compound was originally obtained from. Thus, the synthesis scheme is not provided.
We never claimed that the small molecules in this study rescued the mutant p53 to wildtype p53 activity. Instead, we attest that these small molecules restore the p53 pathway by bypassing direct p53 activation.
The updated manuscript included a thorough analysis of the transcriptomic and proteomic date. At the transcriptome level, we showed that using the Fischer p53 data set (343 genes), we detected 197 genes and out of those, 102 were differentially expressed by analogue #4 indicating that nearly 50% of the established p53 target genes were altered. We discussed in the manuscript the limitations in our proteomic study including the detection of proteins. For example, DR5 and Noxa were not detected in the proteomics analysis but were validated by Western blot analysis.
Figure 1a – the p53 reporter is not sufficiently described in the results or the methods section (what is the construct of this? ). Is this done in cells? What cells?
The doses of the compounds are not provided here.
Further description about the reporter, cells used and drug concentrations were added in the results and methods sections.
Figure 1b,c – again, what are the cells used here?
Further description about the cells used and drug concentrations were added in the Results section and figure legends.
Figure 1f – need to state what the p53 status is for these cells line? And need to perform cell growth inhibition on these cells lines to see if the response to some of the agents is dependent on p53 (should use WT, null, mutant +/- siRNA)
The p53 status for each cell line on Figure 1F now figure 1G has been added to the methods section. We do not claim that the effects of the small molecules is dependent on direct p53 activation. In the cell lines tested, restoration of the p53 pathway by the small molecules is independent of p53, thus it is not expected that Knockdown of p53 will have an effect. Moreover, our experiments in an isogenic panel of HCT116, including p53 null cells suggests that the mechanism of action is independent of p53. The idea that restoration of the p53 transcriptome by pathways involving alternative transcription factors and the extent to which we have demonstrated this in our paper we believe is at the forefront of the field.
Figure 1g – belongs in the supplemental.
We thank the reviewer for this comment nonetheless, we believe this key finding pertains as a main figure. The manuscript has been re-arranged and original figure 1G is now shown as figure 1H.
Figure 2A – Again, it is striking that the cell line (and p53 status) used here is not mentioned in the Results section or the figure legend. there are only 6 p53 targets shown here (there should be many more). Further, the increase in their expression is only mild.
Figure 2A is now figure 2A-B. Microarray data has been reanalyzed. Analysis of p53-responsive genes that were differentially expressed by the analogue #4 showed that, using the Fischer p53 data set (343 genes), we detected 197 genes and out of those, 102 were differentially expressed by analogue #4 indicating that nearly 50% of the established p53 target genes were altered. This is a significant overlap of p53 regulated genes. Details regarding the cell line used for transcriptomics are located in the figure legend and result section. P53 status for cell line used are located in the materials and method section.
Use of the term "reactome" is confusing. In the Results section they state they use both transriptomic and proteomic data, so which are "reactome" are they both?
The term reactome was edited to either transcriptome or proteome when applied to be more clear on what is meant.
Figure 2A the data showed in the pathway analysis of genes upregulated and down regulated is not discussed in the Results section. Nor really is the proteomic data in Figure 2B. The only mention is that the cell cycle pathway is down. Again, this data is very poorly explained. I do not see how this really supports the conclusion that mechanism of cell growth inhibition is through the p53 pathway.
The points about the role of the p53 protein versus the p53 pathway are explained more in the response to reviewers and in the manuscript. The manuscript has been re-arranged and Figure 2A and 2B are now Figure 2 and Figure 3, respectively. The transcriptomic data has been re-analyzed. The discussion from the results of the transcriptomic and proteomic analysis has been extended and incorporated into the main text of the Results section (now also including Figure 4). We have highlighted in the manuscript the restoration of the p53 pathway at the transcriptional level. Our discussion includes limitations of the results observed at the transcriptional and protein level. For example, DR5 and Noxa were not detected in the proteomic analysis but were validated by Western blot analysis.
Figure S5 – the venn diagrams show actually very few p53 target proteins in common when looking at jus the in house database or the published database, and even fewer that are in common with all three. This is not strong data to support their conclusion.
We thank the reviewer for this comment. We have edited the manuscript to highlight the observed restoration of the p53 pathway at the transcriptional level. Our discussion includes limitations on the results observed at the protein level. Of note, differences in protein overlaps are influenced by the type of cell line and treatment. In house proteomics was performed in HCTT116 cell lines with wild-type p53 versus null for p53. Upon treatment with 5-FU we compared our profiles with the known p53 targets from the Fischer database which is a compilation of transcriptomic data from a variety of studies using different cell lines and treatments.
Figure 3A – this blot is very hard to follow. CB002 some of the data appear to contradict their statements in the results for example they say Lines 195, 196 – "Similarly, the combination of etoposide with CB002 or CB002-analogue #4 showed a decrease in expression of pcdc2(Tyr15) and pcdc25c(Ser16)." However the blot shows that the pcdc2(tyr15) levels increase in combo with etoposide.
We have re-run the western blot from Figure 3A now figure 5A to improve the clarity of the result. Both proteins, pcdc2(Tyr15) and pcdc25c(Ser16) decrease in single treatments compared to DMSO control. A slight increase is observed after small molecule treatment with etoposide as compared to single small molecule treatment we believe due to an etoposide effect on G2/M.
Figure 3B – I assume that this was done using flow cytometry but that is not stated.
Details from figure 3B now figure 5E, in addition to be included in the methods section, have been added to in the Results section and figure legends.
Figure 3E – including 7 time points is excessive when the data can be communicated with three and sometimes just two time points. It makes the figures very difficult to read.
We thank the reviewer for this comment. Figure 3E is now figure 5E. We kindly disagree with this comment, showing the effects of the drug with just two time point would not capture how the cells are passing through the cell cycle. Therefore, to clearly show the effect of drug treatments in the cell cycle our data includes several time points from these very carefully done experiments.
Figure 4B – there does appear to be a almost two fold increase in toxicity in the WI38 cells. This is an increase. (it may not be statistically significant) but that not how it is written in the manuscript. Furthrmore, drug concentrations on the organoid in 4C go up to 100 uM, here they only go up 25 μm in the Wi38 cells. They would need to show the data in doses at 50 and 100 uM.
We thank the reviewer for this comment. Figure 4B is now figure 6B. We have edited the result section to specifically mention that the increase in Sub-G1 cell content in the WI38 cell line is not statistically significant. The treatment with CB002-analogue #4 in this experiment was performed at 25uM since this is the IC50 concentration, which was also used throughout the experiments, unless otherwise indicated.
Figure 4D should be quantified in addition to the micrographs.
Quantification of Figure 4D, now figure 6D has been included in the manuscript.
Figure 4G – the tumor growth inhibition is seen at only one time point and is mild.
We thank the reviewer for this comment. Figure 4G is now figure 6G in the updated manuscript. The current manuscript specifically states that at 5 weeks the differences between control and treated tumors are statistically significant. We have also included in our discussion that the effects are suboptimal and warrant optimization.
Reviewer #2 (Reviewer Comments to the Author):
The manuscript by Hernandez Borrero et al. describes the characterization of a set of small molecules derived from a previously identified lead compound CB002 that the authors described as p53 pathway restoring compound. The new derivatives are more potent in inducing cell cycle disruption and inducing cell death in cell culture and tumor spheroid culture. The authors suggest this class compounds may be useful as novel cancer therapeutics based on their ability to restore p53 activity in tumor cells expressing mutant p53.
Identifying new compounds that restore the transcriptional function of mutant p53 is a challenging but attractive strategy of targeting mutant p53 in cancer. Currently there is a set of basic criteria that true mutant p53 rescue compounds are expected to meet, such as restoring mutant p53 binding to consensus DNA sequence in vivo and activating representative wt p53 target genes. Knockout of endogenous mutant p53 should eliminate the effect, and reintroduction of mutant p53 to a p53-null cell should restore the effect. The rescue compound is likely to interact with mutant p53 protein either non-covalently or covalently to change its conformation and DNA binding activity.
The compounds described in this study have not been rigorously tested in this regard and do not fit any of the above criteria. They induce a small number of genes that overlap with p53responsive genes, but it is unclear whether the induction is mediated by mutant p53. But curiously strong wt p53 target genes p21, MDM2, PUMA etc are not induced by these compounds. The results do not rule out the possibility that the small overlap in gene expression profile is coincidental, since p53 target genes can be regulated by other pathways and factors. Based on the data presented, it is premature to call these compounds mutant p53-rescue compounds or p53-pathway activating drugs.
The data mainly focused on analyzing the cell cycle and toxicity profile of the compounds, which is not very informative in terms of drug development. One would have like to see evidence of how the compounds restore mutant p53 function, whether they interact with p53 protein directly, if they do not act on p53 directly, what is their cellular target. Given the absence of these information, the manuscript is not suitable for publication.
We thank the reviewer for their comments. We would like to clarify that we did not state in the manuscript that the small molecules activate p53, are p53-dependent, or rescue the mutant p53 to wild-type p53 activity. Rather, we attest that these small molecules restore the p53 pathway by bypassing direct p53 activation. Thus, it is not expected that CB002 or its analogues will cause mutant p53 to bind to DNA or chromatin in the regulatory regions of p53 target genes in a manner that wild-type p53 does. Therefore, the small molecules are not mutant p53-rescue compounds but rather p53-pathway restoring or bypass compounds.
We performed a microarray analysis to identify p53-responsive genes that were differentially expressed by the treatment of the most potent analogue #4. At the transcriptomic level, we showed that using the Fischer p53 data set (343 genes), we detected 197 genes and out of those, 102 were differentially expressed by analogue #4 indicating that nearly 50% of the established p53 target genes were altered. This is a significant overlap of p53-regulated genes. We validated several p53 target genes including DR5 and Noxa. In the original manuscript p21 was not shown but we extended this observation in the updated manuscript (Figure 5D). MDM2 and Puma were not shown therefore it is not accurate to state that these target genes were not induced, rather they were not part of the scope of the shown experiment.
Although the molecular target was not identified in this study, we provide evidence that the integrated stress response ATF3 and ATF4 transcription factors play a role in the stimulation of pro-apoptotic p53 target gene, Noxa, a key mediator of apoptosis by these small molecules.
Some specific comments
Figure 1A. compound concentration range in the p53 reporter assay was not provided. Is mutant p53 expression needed to observe the activation of p53 reporter? Does crispr knockout of mutant p53 abrogate the reporter activation by the compounds? #4 has strong cell kill but modest p53 reporter activation.
Compound concentration ranges were added to the results and methods sections. Mutant p53 is not required to activate the p53 reporter thus a CRISPR knockout would not be expected to abrogate the p53 reporter activity. At the conditions tested (6 hr) for the p53 reporter assay, CB002 analogue #4 had a modest reporter activation but showed the strongest cell killing in the cytotoxicity assays performed at 48 hrs. Because the p53 reporter assay is based on early transcriptional effects, we believed that at 6 hrs the signaling effect for analogue #4 may have already ceased given that it was the most potent drug and most likely the signal transduction was initiated and terminated compared to the other compounds tested.
Fig1F. Text mentioned HCT116 and HCT116-p53- with exogenous p53 mutant, but data does not contain these cells.
Figure 1F now Figure 1G now includes data from HCT116 and HCT116-p53- with exogenous p53 R275H mutant.
FigS5 Compound 4 induced genes only have very few overlap with known p53 targets. Does not support the notion that the compound activates p53 pathway, possibly coincidental overlap.
We thank the reviewer for this comment. FigS5 has been updated to Figure 2A due the microarray re-analysis. Analysis of p53-responsive genes that were differentially expressed by CB002 analogue #4 showed that, using the Fischer p53 data set (343 genes), we detected 197 genes and out of those, 102 were differentially expressed by analogue #4 indicating that nearly 50% of the established p53 target genes were altered. This is a significant overlap of p53 regulated genes.
Reviewer #3 (Reviewer Comments to the Author):
The manuscript by Hernández Borrero et al. describes a small molecule, CB002, that is said to activate the p53 pathway in cancers with mutant p53. The authors observed that the compound leads to S-phase checkpoint activation, with increased phosphorylation of RPA, ATR, cyclin E and decreased levels of phosphorylated H3 and proteins related to DNA synthesis. Moreover, the authors report that CB002 and its analogues induce apoptosis in cells, as seen through higher proportion of sub-G1 cells, increased PARP cleavage and a release of cytochrome c from the mitochondria. Noxa was shown to be upregulated upon treatment with CB002 and its analogues. Importantly, Noxa depletion inhibited the anti-tumor effects exerted by analogue #4 in vivo, suggesting an important role of Noxa in mediating apoptosis downstream of this compound. The compound tested showed limited toxicity in normal cells, as shown in cell culture and mouse studies, further suggesting a potential of this compound to be brought forward for further clinical studies. The authors propose this molecule as a potential therapeutic approach in targeting tumors that have a p53 mutation.
The potential of CB002 and its derivatives can be appreciated from this manuscript. However, the currently reported data do not provide evidence that these compounds act through p53. Without thorough additional experimentation, it cannot be concluded that p53 reactivation is the major mechanism by that these drugs are acting. As it stands, it is at least equally possible (if not more conceivable) that these are cytotoxic drugs that act independent of p53.
Overall, the manuscript text itself was well written, and easy to read. However, the figures were slightly messy and not very intuitive. Often, figure legends lack sufficient details that help readers understand the figures. Certain labels are cropped off and not aligned very well with the results. The use of many different CB002 analogues is appreciated, but the lack of consistency between the analogues shown in the results is also confusing. The manuscript would benefit from mentioning why a particular analogue (and not the others) was chosen and shown for a particular experiment.
We thank the reviewer for their comments. We would like to clarify that we did not state in the manuscript that the small molecules activate p53, are p53-dependent, or rescue the mutant p53 to wild-type p53 activity or conformation. Rather, we claim that the mechanism of these drugs involves the restoration of the p53 pathway by bypassing direct p53 activation. We provide evidence that the integrated stress response ATF3 and ATF4 transcription factors play a role in the stimulation of pro-apoptotic p53 target gene, Noxa, a key mediator of apoptosis by these small molecules.
CB002 analogues were prioritized by their ability to induce Noxa expression as shown in Figure 1G. We then show that a subset of these analogues in Figure 1C induce apoptosis. As we started studying the cell cycle effects, we focused on the analogues that fit the criteria of the new class of small molecules i.e. perturbation of the S-phase of the cell cycle and restoration of the p53 pathway. We then concentrated the majority of the manuscript on the most potent CB002 analogue #4.
1. Although the clinical potential of these compounds is interesting, especially considering its lack of toxicity in normal cells, the link to p53 is weak. As the authors are presenting these compounds as "activators" of p53, this link would need to be strengthened, unless the authors re-consider their concept and describe CB002 and its derivatives as cytotoxic drugs with as yet unknown mechanisms of action.
We appreciate these comments. We respectfully disagree with the reviewer and would like to clarify that we did not state in the manuscript that the small molecules activate p53. Rather, we claim that the mechanism of these drugs involves the restoration of the p53 pathway by bypassing direct p53 activation. We have edited the manuscript to highlight the restoration of the p53 pathway at the transcriptional level. We showed that using the Fischer p53 data set (343 genes), we detected 197 genes and out of those, 102 were differentially expressed by analogue #4 indicating that nearly 50% of the established p53 target genes were altered.
It seems unclear if the compounds activate p53 as the major mechanism to induce apoptosis, S-phase arrest, decreased proliferation (Ki67) and therefore reduction in tumor growth. Testing for p53 reporter activity is not sufficient. It would be important to observe the entire set of common p53-responsive genes (e.g. the CDKN1A gene discovered to be p53-responsive by the senior author, or Mdm2, Puma, etc., perhaps referring to the catalogues presented by M. Fischer, doi: 10.1038/onc.2016.502, upon addition of these compounds).
We appreciate the reviewer’s comment. We wish to clarify that we do not claim that the mechanism of these drugs involves the activation of p53 for the restoration of the p53 pathway. Rather they restore the p53 pathway through bypassing the direct activation of p53 via other molecular pathways. We provide evidence that the integrated stress response ATF3 and ATF4 transcription factors play a role in the stimulation of pro-apoptotic p53 target gene, Noxa, a key mediator of apoptosis by these small molecules. We do not conclude p53 pathway restoration solely on the p53-responsive luciferase reporter which was a preliminary result years ago at an early point in the history of this work. We incorporate a microarray analysis and compare the differentially expressed genes to the Fischer data set. In this comparison, we detected 197 p53-responsive genes and out of those, 102 were differentially expressed by CB002 analogue #4 indicating that nearly 50% of the established p53 target genes were altered. We have also validated some of these targets at the protein level by western blot analysis showing increased expression of p53 target genes p21, Noxa, and DR5.
Adding on to this point, it would also be important to prove that the induction of Noxa was through p53 activation. Knocking down p53, and testing for Noxa expression upon treatment with these compounds would be essential to make this claim.
We wish to clarify that we do not claim that the mechanism of these drugs involves the activation of p53 for the restoration of the p53 pathway. Restoration of the p53 pathway in mutant p53 harboring cells by these small molecules is independent of p53, thus it is not expected that knockdown of p53 will have an effect. Moreover, our experiments in an isogenic panel of HCT116, including p53-null cells suggests that the mechanism of action is independent of p53.
I find it puzzling that the transcriptome data comparison with in-house p53 proteomic database and known p53 targets had such little overlap. Does this not further suggest that the compounds are acting through a different pathway, independent of p53?
We thank the reviewer for this comment. We have highlighted in the manuscript the restoration of the p53 pathway at the transcriptional level. Our discussion includes limitations in the results observed at the transcriptional and protein level. For example, DR5 and Noxa were not detected in the proteomics analysis but were validated by Western blot analysis.
We have also included in our discussion how a large number of differentially expressed genes involved in the cell cycle can be connected to p53 pathway restoration. The p53 target p21 is involved in the downregulation of cell cycle genes through its association with the DREAM complex, consequently E2F gene targets become downregulated. This connection was suggested as a future direction and was beyond the scope of this manuscript.
According to Supp. Figure 10C, kd of ATF3/4 completely abolishes Noxa induction. This can be due to one of two reasons; the compounds activate p53 which leads to induction of ATF3/4, then to Noxa, or these compounds lead to the activation of other pathways that induces ATF3/4 (i.e: the Integrated stress response). Therefore, the p53 knockdown experiment is important here to prove the role of p53 in their system. On the same note, I am curious to know if these cells in Supp. Figure 10C with ATF3/4 kd survive CB002/ analogue #4 treatment. If Noxa is the key mediator of cell death here, one would expect that these cells escape the apoptosis-inducing effects of the compounds.
Supp Figure 10C is now Figure 1I. We thank the reviewer for this comment. Restoration of the p53 pathway in mutant p53-harboring cells by these small molecules is independent of p53, thus it is no expected that knockdown of p53 will have an effect on Noxa expression. Studies from our lab suggest novel involvement of integrated stress response transcription factors ATF3/4 in the induction of p53 target genes (10.1016/j.neo.2021.01.004) and this manuscript suggests similar based on its findings with a different class of compounds that bypass the p53 pathway.
2. Analogue #4 appears to exert its toxicity in tumor cells irrespective of the p53 status (Figure 4A) – roughly the same IC50 in HCT116 p53-/- or p53 +/+ or the R273H mutant.
Could the authors explain how the compound exerts its effects in the p53 null cell line, and why these cells would have similar IC50s for this analogue? Does this not further suggest an alternative pathway of activation downstream of the compound? The integrated stress response again was mentioned in the discussion (line 301). It would be of high interests to test if these compounds lead to apoptosis or an S-phase checkpoint activation because of the stress response rather than p53. This would provide an entirely different explanation for the observations presented in the rest of the paper.
We thank the reviewer for this comment. We want to clarify that the restoration of the p53 pathway by these compounds in mutant p53-harboring cells is independent of p53. Along these lines, our experiments in an isogenic panel of HCT116, including p53-null cells suggest that the mechanism of action is independent of p53.
We showed that the integrated stress response ATF3 and ATF4 transcription factors play a role in the stimulation of pro-apoptotic p53 target gene, Noxa, a key mediator of apoptosis by these small molecules. In regards to the S-phase activation, we have also included in our discussion how a large number of differentially expressed genes involved in cell cycle control can be connected to the p53 pathway restoration. The p53 target p21 is involved in the downregulation of cell cycle genes through its association with the DREAM complex, consequently E2F gene targets become downregulated. ATF3/4 have been implicated in the regulation of p21. This connection was suggested as a future direction and was beyond the scope of this manuscript.
1. Could the authors test a non-53 reporter construct as a negative control in the experiment in Figure 1A.
We appreciate the reviewer’s comment. The specificity of the reporter used in this study has been validated in the original PNAS paper published in 2006 from our laboratory including the changes at endogenous targets including p21, DR5, Noxa (as well as in the original paper of the p21 (WAF1) discovery by El-Deiry et al., Cell, 1993). We understand that reporters can be misleading and we thus validated endogenous targets in this study including p21, DR5, and Noxa which is really very important.
2. I am not too convinced just by the pictures in Figure 1E. Could the authors show the different channels separately as well? Would it be possible to provide the quantification of the green intensity in the IF pictures?
We have enhanced the image to improve clarity of the result.
3. The authors described on the manuscript text (line 135 – 137) that they observed Noxa induction upon treatment with the CB002 analogues in DLD-1, SW480, HCT116, and HCT116 p53-/- tumor cells expressing the exogenous R175H p53 mutant. However, the figure only shows Noxa expression in DLD-1 and SW480 cells.
Figure 1F now Figure 1G includes data from HCT116 and HCT116-p53- with exogenous p53 R275H mutant.
4. The heatmap of Figure 2B confused me. According to the legends, the authors treated the cells with analogue #4 (T4) for 12 hours but CB2 for 24 hours. Are they comparing these 2 treatments of different timepoints here?
The figure legend has been edited for clarification. Transcriptomics with analogue #4 were done at 12 hrs, proteomics for both CB002 and analogue #4 were performed at 24 hrs.
5. Could the authors include the total levels of H3 and total cdc25 in Figure 3A and also the total levels of RPA and ATR in Figure 3B?
Protein totals have been added to the requested proteins in the western blot that is shown.
6. The gating used for the flow cytometry (as shown in Figure 3E and Supp. Figure 9) does not appear the same across the different samples. This would of course lead to bias in analysis. Please check and re-gate/analyze if necessary.
We appreciate the reviewer’s comment. Figures 3E and supplemental Figure 9, now Figure 5F and S14, respectively were independent experiments and compared to each negative control. Gating was maintained constant on each set of experiments and gates were set as per the negative control for each one.
7. The labels in the figures: Please change the μm to μM.
The symbols for μM have been corrected.https://doi.org/10.7554/eLife.70429.sa2
- Liz Hernandez Borrero
- Wafik S El-Deiry
- Wafik S El-Deiry
- Wafik S El-Deiry
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
This work was presented in part at the annual American Association for Cancer Research (AACR) meetings in 2017, 2018, 2019, and 2020. LJHB received the AACR Minority Scholar Research Award in 2017. This work was supported in part by NIH grants CA176289 (WSE-D). WSE-D is an American Cancer Society (ACS) Research Professor and is supported by the Mencoff Professorship in Medical Science at Brown University. This work was supported by the American Cancer Society and by the Teymour Alireza P’98, P’00 Family Cancer Research Fund established by the Alireza Family.
Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocol at Fox Chase Cancer Center (protocol #14-17).
- Mone Zaidi, Icahn School of Medicine at Mount Sinai, United States
- Received: May 17, 2021
- Accepted: June 9, 2021
- Version of Record published: July 29, 2021 (version 1)
© 2021, Hernandez Borrero et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.