Physiological effects of a novel artificially synthesized antimalarial cyclic peptide: Mahafacyclin B

Mahafacyclin B is a cyclic peptide isolated from the latex of Jatropha mahafalensis and is an antimalarial agent. However, the physiological effects of mahafacyclin B in mammalian cells are not known. Here, we assessed the growth, morphology, and alterations in the transcriptome of CHO-K1 cells exposed to mahafacyclin B (0–22 μM). Mahafacyclin B at 2.2 μM did not affect the proliferation or death of CHO-K1 cells. Mahafacyclin B was not toxic to mammalian cells at 2.2 μM, which represents a normal physiological concentration at which mahafacyclin B retains its antimalarial properties. Interestingly, mahafacyclin B altered the size and morphology of CHO-K1 cells. Comparative transcriptomics revealed that mahafacyclin B modulated the expression of a specific subset of genes.


Introduction
Cyclic peptides are polypeptide chains with a cyclic ring structure. Cyclic peptides have antibacterial, immunosuppressive, and antitumor activities [1]. Cyclic peptides, owing to their increased stability, high resistance to exo-peptidases and endo-peptidases, enhanced binding affinity, and selectivity for target biomolecules, are commonly investigated as therapeutic agents [2]. Biologically active cyclic peptides have been produced using biological and synthetic approaches, both of which have been adopted by pharmaceutical companies to develop novel medications. These novel medications represent a growing share of sales of all those produced worldwide [3].
However, using a biological approach to develop medications includes a high level of risk (e.g. contamination with the porcine circovirus in human vaccines) [4], and sometimes the biological manufacturing process can affect public safety and health [5]. Several therapeutic cyclic peptides are derived from biological processing, but new methods used to synthesize cyclic peptides are under development [6].
Previously, we established a method of chemically synthesizing the cyclic peptide mahafacyclin B, a molecule isolated from the latex of Jatropha mahafalensis [7][8][9][10]. We developed a soluble, tag-assisted peptide head-to-tail cyclization for this compound [11]. Newly synthesized cyclic peptides are requested not only for therapeutic intervention but also in vitro preclinical testing because biological extracts usually contain other molecules. For example, a biological extract and most lipopolysaccharide fractions are contaminated with peptidoglycans because both of them are components of bacterial membranes [12]. Therefore, by synthesising and testing synthetic products in vitro, it is possible to avoid the risk of unintended adverse effects. Using our method, >20 mg of cyclic peptide can be available. Mahafacyclin B is a well-characterized antimalarial agent. Nevertheless, the physiological effects of mahafacyclin B in mammalian cells are not known [13]. Here, we examined the physiological effects of mahafacyclin B in Chinese hamster ovary-derived immortalized mammalian (CHO-K1) cells by measuring their growth, viability, and morphology. We also undertook gene-expression analyses of CHO-K1 cells exposed to mahafacyclin B. Determination of the appropriate concentration of drug to use is critically important for in vitro drug-exposure analyses in preclinical toxicology. High drug concentrations can induce radical transcriptome responses, whereas low concentrations can hamper determination of significant changes in gene expression. A previous study established a quantitative method to determine the appropriate drug concentrations that should be used for in vitro preclinical testing using Shannon's information entropy and/or Kolmogorov complexity to assess transcriptomic changes [14,15]. Comparing the amount of environmental change and the amount of transcriptomic change using phenobarbital in CHO-K1 cells, it was shown that Kolmogorov complexity is a better index of transcriptomic change than Shannon's information entropy [16]. Here, we estimated Shannon's information entropy and the Kolmogorov complexity transcriptomic changes of CHO-K1 cells exposed to mahafacyclin B. It was suggested previously that a medium-sized molecule (e.g., cyclic peptide) can cause drastic regime shifting of gene-expression patterns based on the concentrations used. Here, we aimed to select the most appropriate concentration to analyze differently expressed genes, and found a specific subset of genes that were altered by exposure to mahafacyclin B.

Synthesis of mahafacyclin B
We synthesized mahafacyclin B as reported previously [11]. After purification with semi-preparative reversed-phase high-performance liquid chromatography (HPLC), we finally obtained 20 mg of mahafacyclin B; (purity (UV) 96.8%, epimer <0.9%) (Fig 1). The retention time (RT) of this product was 33.12 min as assessed by HPLC. This RT was slightly longer than that of our product synthesized previously (RT = 31.43 min), but the 1 H NMR spectrum of this product was similar to data published previously [8,13].
Mahafacyclin B dose not modulate the growth of CHO-K1 cells CHO-K1 cells were exposed to mahafacyclin B. Freshly sub-cultured cells (0.8×10 6 /2 ml, viability 97.3723%) were cultured for 12 h in Dulbecco's modified Eagle's medium (DMEM)/F12 medium, and then cultured for 10 h in medium supplemented with 0, 0.22, 2.2, and 22 μM of the cyclic compound. Viable and dead cells were counted using a Vi-cell counter with Trypan Blue staining. In total, 5251 cells were analyzed.
Viable cell densities (in cells/ml (×10 6 )) were: 0.  (Fig 2A). The slope of the regression line was not significantly different than 0 (p = 0.852). The value for the   Fig 2B). The slope of the regression line was significantly different than 0 (p = 0.00496). The value for the slope was −0.41055. The AIC of the linear model on the R function glm was 85.038.
Previously, cell viability was assessed by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay and the effects of cyclic peptides on cell viability were assessed using high concentrations (>50 μM) of each peptide [17]. In the present study, the effects of mahafacyclin B were apparent in terms of cell death, not cell viability. Previously, the antimalarial activity of mahafacyclin B was discovered using a concentration of 2.2 μg/ml [13]. However, cell growth was not inhibited with this concentration of mahafacyclin B.

Mahafacyclin B modulates the morphology of CHO-K1 cells
Cellular morphology was assessed using a Vi-cell microscope and a phase-contrast microscope. Images were analyzed using CellProfiler [18]. We measured length of the major axis and minor axis, and the eccentricity of CHO-K1 cells in response to mahafacyclin B [19,20].
There were differences in the eccentricity of mahafacyclin B-induced cells (Fig 3). The slope of the regression line was significantly different than 0 (p = 0.0177). The value for the slope was −9.167e −5 . The AIC of the linear model on the R function lm was −7721.999. The cell shape is thought to affect plasma membrane signaling [20]. In this context, the importance of the eccentricity had been predicted [20]. This eccentricity changing of mahafacyclin B-induced cells would change plasma membrane signaling and cause transcriptomic changes.
There were weak differences in the length of the major axis of mahafacyclin B-induced cells (Fig 3). The slope of the regression line was weakly significantly different than 0 (p = 0.0658). The value for the slope was −0.003683. The AIC of the linear model on the R function lm was 35952.98.
There were no differences in the length of the minor axis of mahafacyclin B-induced cells (Fig 3). The slope of the regression line was not significantly different than 0 (p = 0.56). The value for the slope was 7.807e −4 . The AIC of the linear model on the R function lm was 31507.26.

Mahafacyclin B modulates the expression of a subset of genes in CHO-K1 cells
We investigated the contribution of Shannon's information entropy and Kolmogorov complexity to the transcriptomic changes of CHO-K1 cells by sequencing the transcriptomes of cells exposed to vehicle or mahafacyclin B. We measured the Shannon's information entropy and the Kolmogorov complexity of those transcriptomes.
The Kolmogorov complexities were: 0.0607232, 0.0609433, and 0.0610302 with 0 μM mahafacyclin B; 0.0610032, 0.0609453, and 0.0610109 with 0.22 μM mahafacyclin B; 0.0597326, 0.0612445, and 0.0611866 with 2.2 μM mahafacyclin B; 0.0610824, 0.0611094, and 0.0613025 with 22 μM mahafacyclin B. With 2.2 μM of mahafacyclin B, we found an outlier; the transcriptomic data showed extremely low values for Kolmogorov complexity. We, therefore, removed the transcriptomic data from this sample before further analyses. The slope of the regression line was significantly different than 0 (p = 0.0288) (Fig 4). The value for the slope was 2.238e −4 . The AIC of the model was −134.0088. The AIC of the model which utilized Kolmogorov complexity was lower compared with that utilizing Shannon's information entropy. Kolmogorov complexity was a better index of transcriptomic data than Shannon's information entropy. The AIC (Akaike's information criterion) is an index for selecting statistical models. To increase the numbers of parameters of statistical models makes models fit better to the data and phenomenon. However, fitting model having too much model is useless. The AIC solves this problem. We chose the model having the lowest AIC. The Shannon's information entropy is an index of data. The transcriptome which a few genes occupy major region represents small information entropy. The transcriptome which many genes occupy major region represents large information entropy. Therefore, the Shannon's information entropy was used as a quantitative marker of cellular differentiation/dedifferentiation [14]. The Kolmogorov complexity is also an index of data. The elementary theories of Shannon's information and Kolmogorov complexity have a common purpose. Kolmogorov complexity is the minimum number of bits from which a particular message or file can effectively be reconstructed [15].
Studies comparing Shannon's entropy and Kolmogorov complexity have shown that Kolmogorov complexity is, in fact, a better indicator of regime shifting of transcriptomic data than Shannon's information entropy. In one study, cells cultured with drug concentrations lower than the tipping point showed uniformly high Kolmogorov complexity, whereas those cultured with concentrations higher than the tipping point showed uniformly low Kolmogorov complexity using primary cultured tissues and phenobarbital [14,15]. In another study, cells cultured with low drug concentrations showed low Kolmogorov complexity, whereas those cultured with higher drug concentrations showed low Kolmogorov complexity using mammalian cell lines and phenobarbital [16]. In primary cultured cells and cell lines, the relationship between drug concentration and Kolmogorov complexity was reversed. In the present study, cells cultured with low drug concentrations showed low Kolmogorov complexity and those cultured with high drug concentrations showed low Kolmogorov complexity. Increased Kolmogorov complexity with increasing drug concentration is observed in cell lines. Changes in the Kolmogorov complexity indicate drastic regime shifting of genomic expression systems. Hence, a medium-sized molecule (e.g., cyclic peptide) can cause drastic regime shifting of genomic expression systems in a similar manner to that of small molecules.
Using traditional RNA-sequencing methods with a heterogeneous population of cells can induce variance in RNA-sequencing data pools. Treatment with phenobarbital and mahafacyclin B could increase the heterogeneity of immortalized cell lines. To address this possibility, we analyzed the morphological heterogeneity of CHO-K1 cells in response to mahafacyclin B using Shannon's information entropy. Shannon's information entropy for morphological eccentricity was compared with the concentration of mahafacyclin B. The use and interpretation of Shannon's information entropy as a measure of diversity and homogeneity is justified by mathematically founded arguments [21]. The slope of the regression line was not significantly different than 0 (p = 0.673). Shannon's information entropy for major-axis length was compared with the concentration of mahafacyclin B. The slope of the regression line was not significantly different than 0 (p = 0.192). Shannon's information entropy for minor-axis length was compared with the concentration of mahafacyclin B. The slope of the regression line was not significantly different than 0 (p = 0.366). These results indicate that increased cellular heterogeneity does not explain the observed phenomenon: increasing Kolmogorov complexity is correlated with increasing drug concentration.

Identification of genes modulated by mahafacyclin B
To investigate the effects of mahafacyclin B on gene expression, we undertook two comparisons of transcriptomic changes that occurred in response to mahafacyclin B (0 vs. 0.22 μM, 0 vs. 22 μM). We obtained 19 statistically significant differentially expressed genes (false discovery rate (FDR) < 0.05) using data from cells exposed to a low concentration of mahafacyclin B (0 vs. 0.22 μM, Fig 5A, Table 1) and 14 from data used with a high concentration of the cyclic peptide (0 vs. 22 μM, Fig 5B, Table 2). Six genes were found to overlap in both comparisons, and these were identified as LOC100754005, LOC100689425, LOC103164497, LOC1007 52965, Kmt2a, and Ino80d. Among these genes, LOC103164497 (NCBI Reference Sequence: XP_007652779.1) expression was increased in response to mahafacyclin B, but its function is not known. However, XP_007652779.1 shared sequence similarity with XM_007643794.2, a predicted protein similar to the Cricetulus griseus retrovirus-related Pol polyprotein from transposon 17.6-like (LOC103163056) (eValue, 0.0; identities, 756/758 (99%); gaps 0/758 (0%): tblastn). Pol polyprotein was predicted to have aspartic-type endopeptidase activity, and could be used to degrade mahafacyclin B intracellularly. Pol polyprotein was also predicted to be involved in viral entry into host cells. Endogenous retroviral sequences are present in high copy numbers within the genomes of all species, whereas endogenous production of retroviruses can be induced chemically [22][23][24]. Therefore, mahafacyclin B could be used (in theory) to discover novel retroviruses. Expression of chromatin-remodeling genes (Kmt2a, Ino80d [25], Ash1l [26], and BOX [27]) decreased in response to mahafacyclin B, indicating that epigenetic inheritance could occur to prolong the effects of the peptide. Reduced expression of genes related to cell division (ORC1-like [28], Cdk6, PRC1 [29], RPA1, and Cdk11B) indicate that cell-cycle progression was slowed. However, we did not observe a reduction in cell number in response to mahafacyclin B. Expression of genes related to regulation of ubiquitin-mediated proteolysis (Wdfy3 [30], FBXW-2 like: LOC100754385 [31]) decreased. Interestingly, expression of genes related to ATP synthesis (LOC100762406:ATP5E, LOC100767351: COX7A2) increased in response to mahafacyclin B. The CHO cell line has been used for protein production. Recently, it was shown that ATP has the properties of a biological hydrotrope [32]. Therefore, mahafacyclin B may increase ATP concentrations intracellularly and improve protein-folding efficiency. Reduced expression of a gene encoding a structural component of microtubules (Tubb) and a gene related to the regulation of cell-cell junctions (Heg1 [33]) could also be associated with morphological changes in response to high concentrations of mahafacyclin B. No apoptotic genes were induced by mahafacyclin B.

Conclusions
We characterized the physiological effects of the antimalarial agent mahafacyclin B. Exposure to mahafacyclin B had no effect on proliferation but, at certain concentrations, induced cell death. Mahafacyclin B is, therefore, not toxic at 2.2 μM, which is a physiologically relevant concentration. We found that mahafacyclin B altered the expression of a specific subset of genes; no apoptotic genes were induced by mahafacyclin B. The increase of peptidase enzyme was not observed in this study. These findings also indicated the safety and stability of mahafacyclin B as the antimalarial drug. Mahafacyclin B induced expression of ATP synthesis genes; mahafacyclin B could be used for an industrial protein production enhancement not only an antimalarial agent. Kolmogorov complexity was a better index than Shannon's information entropy in assessing the comparative transcriptomics of CHO-K1 cells exposed to mahafacyclin B.
The reaction mixture was filtered with Celite™ 454, and DIPE was added to the evaporated filtrate for product precipitation. The precipitability of the product was decreased significantly as the volume of TFA for washing the filter increased (filtrate 40 ml after evaporation), so the supernatant was centrifuged three times at 3000 rpm for 10 min at −9˚C. After the precipitate was dissolved in CH 3 Cl/MeOH, evaporated, and dried, 75.0 mg of crude product was obtained. The purity of the desired product (as assessed by HPLC) was decreased to 62%, thereby increasing the rate of epimer formation during alkaline hydrolysis.

Induction assay
All chemicals used in the induction assay were of analytical grade. Mahafacyclin B was dissolved in solvent (water and acetonitrile, 1:1) to make a stock solution, which was diluted further and added to the culture medium to yield final concentrations of 0.22, 2.2, and 22 μM. CHO-K1 growth medium was replaced with mahafacyclin B-containing medium before all induction assays. For each experiment, CHO-K1 cells (American Type Culture Collection, Manassas, VA, USA) were cultured with mahafacyclin B or vehicle for 10 h. Cells were counted using a Vi-cell counter (Beckman Coulter, Brea, CA, USA). CHO-K1 cells were exposed to mahafacyclin B. Freshly sub-cultured cells (0.8×10 6 /2 ml, viability = 97.37%) were cultured for 12 h in DMEM/F12 medium, and then cultured for 10 h in medium supplemented with 0, 0.22, 2.2, and 22 μM of the cyclic compound. Viable and dead cells were counted using a Vicell counter with Trypan Blue staining. In total, 5251 cells were analyzed.
All data were processed using bash v3.2 and visualized using R. The percentage of viable and dead cells was analyzed using R (R project for statistical computing, Vienna, Austria) and Fitting Linear Models (lm). The number of viable cells, the eccentricity of cells, the length of the major axis of cells, the length of the minor axis length of cells, the Shannon's information entropies of transcriptomes and the Kolmogorov complexities of transcriptomes in response to mahafacyclin B were analyzed using simple linear regression analyses. The simple linear regression analyses were performed using R. The rate of cell death in response to mahafacyclin B was analyzed using generalized linear models (glm).
Induction cultures were terminated using a QIAshredder; (Qiagen, Hilden, Germany) and the cells were kept at −80˚C before analyses. RNA was extracted using an RNeasy mini kit (Qiagen). mRNA libraries were prepared for RNA sequencing using a TruSeq RNA Sample kit (Illumina, San Diego, CA, USA) per manufacturer protocols. These libraries were sequenced using a NextSeq 500 sequencer (Illumina) per manufacturer protocols. Short-read data were deposited in the Short Read Archive (project ID, DRA005920) of the DNA Data Bank of Japan. All raw sequencing reads were mapped to the CHO-K1 RefSeq assembly (ID GCF_000223135.1) using rsem [34].
Kolmogorov complexity was measured as described previously [15] and analyzed using R and lm. Analyses of differentially expressed genes were done using R and the TCC v1.1.99 [35]. We estimated library normalization factors using TbT method with TCC package. Statistically significant differentially expressed genes were extracted with false discovery rate (FDR).