*2.6. The Effect of Terr, GCB, and Their Combination on the Extracellular Metabolites within Colorectal Cell Llines*

1H-NMR comparative analysis was carried out between the metabolites released from HCT-116 cells after three different drug treatments: Terr only, GCB only, and a combination of both Terr and GCB under normoxic conditions. Three metabolites, namely, ethyl malonate, tyrosine, and methylhistidine were detected extracellularly from HCT-116 cells treated with GCB only. While other metabolites such as hypoxanthine and imidazole were identified in the HCT-116 cells treated with Terr and GCB + Terr but not in the extracellular fluid of the HCT-116 cells treated with GCB only. On the other hand, methionine was detected extracellularly in the HCT-116 cells treated with Terr only. Metabolites such as formate and pipocolate appeared extracellularly in both cell lines treated with Terr and GCB only but were not detected in cell lines treated with (GCB + Terr).

Another similar comparative analysis was performed between the metabolites released from HCT-116 cells after the same three different drug treatments but under hypoxic conditions. Three metabolites, namely, 4-hydroxyphenyl acetate, leucine, and pyruvate were detected only in the extracellular fluid of HCT-116 cells treated with both drugs (GCB + Terr) and were not detected in the cells treated only with single drugs. Furthermore, methionine and phenylalanine were detected in all the treatments except the GCB-treated cell lines.

Then, a third 1H-NMR metabolomic analysis was conducted to compare the extracellular metabolites released by HCT-116 cells in both hypoxic and normoxic conditions. A total of 25 metabolites were detected from HCT-116 cells treated with the three drugs under normoxic and hypoxic conditions. In total, 19 of them were common between the two conditions (Table 2). Three metabolites, namely, 2-hydroxy valerate and 2-phosphoglycerate, were completely absent from all normoxic treatments and detected in all hypoxic ones, while pyruvate was exceptionally detected in the hypoxic cells treated with the combined drugs (GCB + Terr). On the other hand, methylhistidine was uniquely detected in the normoxic cells treated only with GCB drugs.

**Table 2.** Extracellular metabolites identified by 1H NMR-based profiling for HCT-116 cells under both hypoxic and normoxic conditions. The quantification of metabolites was achieved via fitting with its reference spectrum from the library of the Chenomx NMR suite. The mean concentration ± standard error is shown for metabolites from a set of three biological replicates.


Two multivariate statistical analyses, hierarchical cluster analysis (HCA) and partial least squares discriminant analysis (PLS-DA), were utilized to study the overall difference in the metabolites released because of the three different drug treatments under each of the normoxic and hypoxic conditions individually and then were thirdly conducted to compare between the two conditions. The first HCA classified the released metabolites under normoxic conditions after the three drug treatments into clusters based on their abundance (Figure 10). Furthermore, supervised multivariate analysis, PLS-DA, showed that metabolites from different treatments under normoxic conditions are separately clustered (Figure 11A). VIP scores were selected as a criterion for choosing the most important variables of the PLS-DA model (Figure 11B). For normoxic conditions, the significantly different metabolites with a VIP score of greater than 1 are glutamate, histamine, hypoxanthine, phenylalanine, xanthine, dimethylamine, tyramine, 4-hydroxyphenyl alanine, and succinate.

**Figure 10.** Unsupervised hierarchical clustering and heat map analysis of extracellular metabolites released from HCT-116 under normoxic conditions.

**Figure 11.** Supervised partial least squares discriminant analysis (PLS-DA) of the metabolite profiling of HCT-166 cells treated with Terr (T), gemcitabine (G), and Terr + GCB (TG) under normoxic conditions. (**A**) Two-dimensional score plot for HCT-116 cells treatments. (**B**) VIP score plot for treatments of HCT-116 cells.

Similarly, hierarchical cluster analysis (HCA) and partial least squares discriminant analysis (PLS-DA) were conducted to investigate the difference in the metabolites released because of different kinds of drug treatments under hypoxic conditions. HCA clustered the released metabolites based on their abundance (Figure 12). PLS-DA revealed that metabolites from the three treatments are separately clustered (Figure 13A). VIP scores were also selected as criteria for choosing the most important variables of the PLS-DA model. The most significantly different metabolites with a VIP score of more than 1 as shown in (Figure 13B) are histamine, methionine, hypoxanthine, glucose, lactate, imidazole, glutamate, alanine, valine, tyrosine, dimethylamine, pyruvate, 4- hydroxyphenyl alanine, leucine, and succinate.

**Figure 12.** Unsupervised hierarchical clustering and heat map analysis of extracellular metabolites released from HCT-116 under hypoxic conditions.

**Figure 13.** Supervised partial least squares discriminant analysis (PLS-DA) of the metabolite profiling of HCT-166 cells treated with Terr (T), gemcitabine (G), and Terr + GCB (TG) under hypoxic conditions. (**A**) Two-dimensional score plot for HCT-116 cells treatments. (**B**) VIP score plot for treatments of HCT-116 cells.

The same multivariate analysis methods (hierarchical cluster analysis and partial least squares discriminant analysis) were applied to assess the difference in the metabolites released from HCT-116 cells under both normoxic and hypoxic conditions. The heat map showed the difference in metabolites' concentration between hypoxic and normoxic treatments (Figure 14). Moreover, PLSD-A separated the hypoxic metabolites and normoxic metabolites into two clusters (Figure 15A). The most significantly different metabolites between hypoxic and normoxic cells were determined using VIP scores (Figure 15B).

**Figure 14.** Unsupervised hierarchical clustering and heat map analysis of extracellular metabolites released from HCT-116 comparing normoxic and hypoxic conditions.

**Figure 15.** Supervised partial least squares discriminant analysis (PLS-DA) of the metabolite profiling of HCT-166 cells treated with Terr (T), gemcitabine(G), and Terr + GCB (TG) under normoxic and hypoxic conditions. (**A**) Two-dimensional score plot for HCT-116 cells treatments. (**B**) VIP score plot for treatments of HCT-116 cells.

#### **3. Discussion**

The need for novel anticancer treatment is an emerging matter as cancer is a major health problem worldwide, and the currently available chemotherapeutic options are becoming increasingly susceptible to resistance [1,2]. Terrein (Terr) is a bioactive marine metabolite isolated from the fungal strain of *Penicillium* species SF-7181 and *Aspergillus terreus* [39]. It exerts its activity via different mechanisms such as angiogenesis inhibition, cell cycle regulation, apoptosis, and autophagy induction [6–9]. However, the exact role of Terr as an antitumor agent remains unclear. Gemcitabine (GCB) is generally considered to be the drug of choice for pancreatic adenocarcinoma and has also been used in other types of cancers such as colorectal cancer; however, its major drawback is its susceptibility to both intrinsic and acquired chemoresistance [40]. Therefore, improving the therapeutic effect of GCB is crucial.

In the current work, Terr showed relatively high IC50s against three different colorectal cancer cell lines under normoxic and hypoxic conditions. Yet, several similar studies showed that Terr induces cancer cell death; however, it is highly cell type-dependent, as well as dose and time dependent [12,16]. On the other hand, GCB showed much higher anticancer potencies against the same set of cell lines compared to Terr under normoxic and hypoxic conditions. The combination indices for GCB with Terr were indicative of a synergistic interaction in HCT-116 and SW620; however, it was antagonistic in HT-29 in both normoxia and hypoxia.

To explain the characteristics of the interaction between GCB and Terr treatments as well as their combination, apoptosis, autophagy, and cell cycle interference were assessed using the flowcytometry technique. This allowed us to determine if the cell death was due to programmed/non-programmed (apoptosis vs necrosis) cell death, autophagy induction/suppression, or simply interference with cell cycle progression (antiproliferative properties). It is worth mentioning that the role of autophagy in cancer is very controversial as it is often referred to as either inducing cell death by suppressing tumorigenesis or facilitating tumorigenesis [41,42]. Apoptotic cell death in GCB singular treatment was significantly higher compared to control untreated cells in all cell lines assessed as expected. Yet, after 24 h of treatment in HCT-116 and SW620 cell lines, the combination treatment significantly increased apoptotic cell death compared to other treatments. The opposite was evident in the HT-29 cell line under both normoxic and hypoxic conditions. Both cases are in alignment with combination indices calculated in these cell lines (synergistic versus antagonistic). It is worth mentioning that apoptosis was induced, and autophagy was suppressed after 24 h in HCT-116 and SW620 cells under combination conditions (GCB + Terr). This might be attributed to suppressed autophagy leading to apoptosis and cell death. Furthermore, the gene expression profile of apoptosis and autophagy genes confirmed this pattern. In the previous literature, cells such as HCT-116 can proceed via apoptosis and suppress autophagy through the regulation of certain pathways, such as PI3K/AKT/mTOR [43]. It is prevalent in colorectal cancer cells and exhibits the antagonistic effect between autophagy and apoptosis as a survival mechanism due to crucial environmental factors [44]. Concerning cell cycle analysis, all cell lines that were tested showed a significant increase in the S phase population after 24 and 48 h of treatment with GCB, while Terr induced a significant increase in G2/M phase arrest in the same set of cell lines. Previous studies showed similar findings and have shown that GCB induces S-phase cell cycle arrest and regulates cell cycle-related proteins, while Terr induces G2/M phase cell cycle arrest [45,46].

RT-qPCR analysis for single and combined treatment was conducted for apoptosisrelated genes (BCL2, BIRC5, TP53, and FOXO3), autophagy-related genes (ATG5 and Beclin-1), cell cycle-related genes (CCND1, CDK4, and MCM7), cellular growth-related genes (AKT1, TGF-B1, HIF1-a, and PRKDC) and cellular proliferation-related genes (PCNA and RAD18). Most of the apoptotic genes that were studied were under-expressed in all cell lines that were tested for all treatments under both normoxic and hypoxic conditions. The anti-apoptotic gene, BCL2, which plays a role in programmed cell death as an antiapoptotic protein [47], was over-expressed when HCT-116 cells were treated with Terr alone under normoxic conditions; however, it was significantly downregulated when the cells were treated with the combination treatment (GCB + Terr). This observation is supported by similar combination studies that showed the downregulation of BCL2 when HCT-116 was treated with combination treatment [48].

Similar results were observed with the autophagy-regulating genes; they were found to be under-expressed in all cell lines tested for all treatments under both normoxic and hypoxic conditions. ATG5, which regulates autophagy by forming autophagic vesicles and controls mitochondrial quality after oxidative damage, was over-expressed when HCT-116 cells were treated with Terr under normoxic conditions. According to the literature, Beclin-1 is a gene that plays a significant role in regulating autophagy, proliferation, and apoptosis in colorectal cancer cells (HCT-116 and SW620). The inhibition of Beclin-1 leads to the suppression of autophagy and proliferation as well as the promotion of apoptosis, which are observed in the results [49]. The exact mechanism by which Beclin-1 promotes apoptosis and suppresses autophagy, however, remains unclear.

Concerning cell cycle regulator genes, they followed the same pattern, and most of the genes that were tested were under-expressed in all cell lines that were tested for all treatments under both normoxic and hypoxic conditions. The only exception was CDK4, which is an important gene that encodes proteins for the cell cycle G1 phase progression. CDK4 was only over-expressed when the cells were treated with Terr under normoxic conditions in the HCT-116 cell line. However, the combination treatment (GCB + Terr) led to the downregulation of CDK4 in HCT-116. CDK4 is found to be amplified in colorectal cancer cells compared to normal cells [50,51], and evidence has shown that inhibiting certain CDKs such as CDK1, 2, 4/6, and 9 is useful in enhancing colorectal cancer cell (HCT-116) death [52]. According to previous studies, the inhibition of CDKs has also proven to be beneficial in suppressing colorectal cancer cells from proliferation through cell cycle arrests, and in some cases, can also lead to apoptotic cell death [53].

Most of the cellular growth genes that were tested were observed to be underexpressed in all cell lines in response to treatment. The AKT1 gene was of great interest and a known target for terrein, as it works by regulating various processes such as metabolism, proliferation, cell survival, and cell growth. In HCT-116, AKT1 was over-expressed when the cells were treated with Terr in normoxia and under-expressed for other treatments in the same condition. The contrary was evident in hypoxia. Hypoxia plays a major role in tumor cell behavior and the way it responds to treatment [54]. Stegeman et. al. showed that hypoxia stimulates AKT expression and activation in vivo and in vitro. The current study might prove that terrein-induced pAKT inhibition can overcome the influence of hypoxia and diminish cell survival in hypoxic cells, rather than in normoxic conditions [55]. In HT-29, AKT1 was over-expressed after treatment with GCB in normoxia, which might explain the antagonistic interaction with terrein. In SW620, AKT1 was over-expressed after single as well as combination treatment under both normoxic and hypoxic conditions. Still, this can explain the very high resistance fraction (R-value) to treatment in this cell line. AKT is known to regulate cellular proliferation through the degradation of CDK inhibitors, therefore promoting cell cycle progression and inhibiting apoptosis by inactivating pro-apoptotic molecules [56,57]. As a result, AKT plays a vital role as a signaling biomarker, which integrates many potential oncogenic signals [58]. On the other hand, recent studies showed that overactivation of AKT can also increase cell resistance to oxidative stress and allow cells to be more viable in high reactive oxygen species (ROS) conditions [59]. On the other hand, several studies showed that AKT nucleus translocation can induce cell death via apoptin due to the activity of some anticancer drugs [60]. This proves that the overall outcome of AKT activation or inhibition depends on the signaling context as well as the topological characteristics [58]. Most of the cell proliferation genes that were studied were under-expressed in response to single and combined treatment. Yet, this might indicate the antiproliferative effect rather than the cytotoxic properties of treatments under investigation, especially if we noticed the high resistance fraction in all treatment conditions.

Only HCT-116 cells showed a synergistic versus antagonistic interaction between terrein and GCB under normoxic versus hypoxic conditions, respectively. Yet, it was further assessed via a metabolomic study using 1H-NMR comparative analysis and profiling under normoxic versus hypoxic conditions. Lately, many studies have been trying to investigate colorectal cancer metabolic profiling compared to normal tissues [61]. Hirayama et al. found that due to hypoxia, there is a significant variation in the energy metabolism in the colorectal cancer tissues [62]. Denkert et al. reported that intermediates of the lipids and tricarboxylic acid (TCA) cycle were downregulated in tumor tissue, while urea cycle metabolites, purines, pyrimidines, and amino acids were upregulated compared to normal tissue [63]. In the current study, the extracellular metabolites of the different treatment conditions (single versus combination treatment) were found to be significantly different

from one another when compared under different oxygen conditions. According to the VIP scores, the metabolites that were believed to exhibit a significant role in the metabolic shift when comparing the different oxygen conditions as well as when testing normoxia and hypoxia per se were leucine, tyramine, 4-hydroxyphenyl acetate, xanthine, and tyrosine. A study carried out by Hirayama et al. showed that tumors can have a tumor-specific metabolism that grants them more prevalent proliferation, at the same time keeping some metabolic characteristics of the tissues from which they originated. To put it in another way, cancer cells are progressed via metabolic adaptation that includes the upregulation of glucose consumption and increase in amino acids, while preserving the tissue-specific dependency of aerobic respiration characterized by TCA intermediate and nucleotide levels [62]. Another study performed by Frezza et al. suggests that hypoxic HCT116 cells could depend on catabolic processes to make up the energetic defect created by the loss of mitochondrial activity and that cannot be made up by the increased glycolytic flux [64]. Herein, these metabolic profiling data are in alignment with autophagy and apoptosis results. Mitochondrial membrane and integrity are among the early flip points in apoptosis/autophagy balance [44,65]. The significantly different profiles of energy metabolites in both normoxic and hypoxic conditions mirrored the different profiles in autophagy/apoptosis balance in the HCT-116 cells [66]. In our study, according to the PLS-DA, there was no overlap in metabolites clustering among the different treatments. Further analysis was conducted via a metabolite–gene–disease interaction network, which illustrated that some of the genes that were tested were directly linked to colorectal cancer in alignment with the metabolites that were found earlier, such as AKT1 and BCL2.

In conclusion, terrein possesses a controversial role in influencing the anticancer properties of gemcitabine in colorectal cancer cells under normoxic versus hypoxic conditions ranging from antagonism to synergism. However, this influence is evident via significant changes in cell proliferation patterns, cell cycle progression, apoptosis, and autophagy with mirrored confirmatory gene expression profiles. AKT1 seems crucial in all these processes in terms of activation and expression. On top of all these, the metabolic profile of energy and mitochondrial function was significantly and differentially affected by single and combined treatments under normoxia versus hypoxic conditions. It is recommended to further study these effects under more complicated tissue culture conditions such as a 3D culture system or even in vivo animal models to add the dimension of tissue penetration to the current research outcome.

## **4. Materials and Methods**

### *4.1. Chemicals and Drugs*

Gemcitabine (GCB), terrein (Terr), and sulforhodamine-B were purchased from Sigma Chemical Co. (St. Louis, MO, USA). RPMI-164 media, DMEM, fetal bovine serum, and other cell culture materials were purchased from ATCC (Houston, TX, USA). Other reagents used were of the highest analytical grade.
