Biological material

Fresh Z. cassumunar rhizomes were obtained from Chatuchak Market in Bangkok, Thailand. The samples were placed in storage in darkness at a temperature of 4°C at the laboratory until required for experimentation. The American Type Culture Collection (ATCC, Manassas, VA, USA) supplied the human intestinal epithelial cell line, Caco-2 cells (ATCC^® HTB-37^™) which were then cultured in RPMI 1640 (Roswell Park Memorial Institute: RPMI) with 10% fetal bovine serum (FBS) provided by Gibco (Rockville, MD, USA).

Chemicals

Acrylamide, ammonium persulfate (APS), ammonium sulfate ((NH₄)₂SO₄), L-ascorbic acid, 2,2’-Azobis(2-methylpropionamidine) dihydrochloride (ABAP), ABTS, bovine serum albumin (BSA), Coomassie Brilliant Blue G, curcumin, 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT), 2’-7’ dichlorodihydrofluorescein diacetate (DCFH-DA), DPPH, Dimethyl sulfoxide, (DMSO), ferric trichloride hexahydrate (FeCl₃.6H₂O), fetal calf serum (FCS), disodium hydrogen phosphate (Na₂HPO₄), monosodium dihydrogen orthophosphate (NaH₂PO₄), hydrochloric acid (HCl), N,N’-methylenebis(acrylamide), naphthylethylenediamine dichloride (NED; C₁₂H₁₆Cl₂N₂), pancreatin from porcine pancreas (8 × USP specifications), pepsin from porcine gastric mucosa (≥250 U/mg), potassium persulphate (K₂O₈S2), quercetin, Sodium acetate, (CH₃COONa), sodium chloride (NaCl), sodium dodecyl sulfate (SDS), sodium hydroxide (NaOH), sodium nitroprusside (SNP; C₅FeN₆Na₂O), sulphanilamide (C₆H₈N₂O₂S), 2,4,6-tri(2-pyridyl)-S-triazine (TPTZ), and tris hydrochloride (Tris-HCl) were obtained from Sigma-Aldrich, Merck (St. Louis, MO, USA). Thermo Fisher Scientific (San Jose, CA, USA) supplied the ethanol, acetonitrile (ACN), formic acid, N,N,N’,N’-tetramethyl ethylenediamine (TEMED), trifluoroacetic acid (TFA), and Tween 20 all of which were of chromatographic grade. All chemicals used in the course of this research were of analytical grade.

Crude protein extraction from Phlai rhizomes

A modified version of Inthuwanarud et al.’s method was utilized to extract protein from Phlai rhizomes. The rhizomes (1.5 kg wet weight) underwent peeling, slicing, and homogenization in 5 L of PBS (0.15 M NaCl with 20 mM phosphate buffer, pH 7.2) using a blender (HR 2061; Philips, Bahasa, Indonesia) [14]. The mixture was agitated overnight at 4°C using an agitator (Model RW-20; IKA. Labortechnik, Staufen, Germany). The resulting suspension was filtered through double-layered cheesecloth and centrifuged for 30 minutes at 15,000 × g. Ammonium sulfate was added to the supernatant, stirred to 80% saturation, and left overnight at 4°C with continued stirring. After centrifugation at 15,000 × g for 30 minutes to collect the precipitate, the supernatant was discarded. The obtained pellet was dissolved in 20 mL of PBS, dialyzed using tubing with a 3,500 Dalton cut-off (SnakeSkin, Thermo Scientific Co., Ltd., USA), and finally freeze-dried as the crude protein preparation.

Phlai protein hydrolysis using pepsin and pancreatin

Initially, Phlai proteins were used as the substrate to produce the protein hydrolysate, following the method described by Petsantad et al. [15] Briefly, freeze-dried protein is dissolved in DDI water at a ratio of 10% (w/v) for 30 min to solubilize and evenly disperse the protein. Sequential enzyme digestion was carried out with pepsin/substrate 1:20 (w/w) enzyme-substrate (E/S) ratios, with the pH adjusted to 2.5 using 1 M HCl. The hydrolysis occurred at 37°C for three hours with constant shaking at 180 rpm. Inactivation was achieved by raising the pH to 7.5 with 1 M NaOH. Subsequently, pancreatin was added by pancreatin/substrate 1:20 (w/w) E/S ratio and incubated for three hours at 37°C with shaking at 180 rpm. The enzyme reaction was terminated by heating to 80°C for 20 minutes. The hydrolysates were then clarified through centrifugation at 15,900 × g for 30 minutes at 4°C. The antioxidant activity of the supernatant was then measured. The protein concentrations for each of the protein hydrolysates could be determined using the Bradford procedure under the BSA standard. The absorbance was then measured at 595 nm using a spectrophotometer (Multiskan GO; Thermo Fisher Scientific, Waltham, MA, USA) [16].

Antioxidant capacity measurement

Isolation and enrichment of bioactive peptides

Identification and synthesis of peptides

The identification of the peptide contained within the RP-HPLC fraction (F₅ sub-fraction), exhibiting the highest antioxidant activity, was achieved through the utilization of an electrospray ionization and an UltiMate 3000 nano-liquid chromatography system (Thermo Fisher Scientific), coupled with a micrOTOF-Q electrospray ionization quadrupole time-of-flight MS (Bruker Daltonics, Billerica, MA, USA). Subsequently, the data were analyzed using a de novo sequencing approach. For sample injection, mobile phase A comprised 2% acetonitrile and 0.1% formic acid in water, while mobile phase B consisted of 0.1% formic acid in acetonitrile, flowing at a rate of 300 nL/min for 60 minutes. Data acquisition was managed using Hystar software (Bruker Daltonics Data). The peptide spectra were obtained within two mass ranges: m/z 400–3,000 and 50–1,500. LC-MS/MS data were utilized for de novo sequencing, and the MS data files were scrutinized using Mascot database software (Matrix Science, London, UK) against the NCBI database. The resulting amino acid sequence was then evaluated via the NCBI database using the BLASTp program, allowing the identification and statistical assessment of matches found.

Following peptide identification, further synthesis was carried out via Fmoc solid-phase synthesis utilizing an Applied Biosystems Model 433A Synergy peptide synthesizer (Applied Biosystems, Foster City, CA, USA). The peptide purity was subsequently assessed using analytical mass spectrometry employing a quadrupole ion trap Thermo Finnigan^™ LXQ^™ LC-ESI-MS (San Jose, CA, USA) connected to a Surveyor HPLC (Thermo Fisher Scientific, San Jose, CA, USA). The HPLC results indicated that the synthetic peptides should exhibit a purity of no less than 98%. The sequences were determined to be Asp-Gly-Ile-Phe-Val-Leu-Asn-Tyr (DGIFVLNY or DY-8; m/z = 940.478) and Ile-Pro-Thr-Asp-Glu-Lys (IPTDEK or IK-6; m/z = 702.367), with a molecular weight of 1,014.20 Da and HPLC purity exceeding 98%. The antioxidant capabilities of these synthetic peptides were also evaluated.

Bioinformatics tools supporting the annotation of peptide properties

To guide future potential applications, it is necessary to understand the properties of different antioxidant peptides. Data are available online, so in this case, the potential profiles of DY-8, and IK-6 can be found using the BIOPEP database, which is available at: www.uwm.edu.pl/biochemia/index.php/en/biopep. BIOPEP offers the additional advantage of classifying the possible sensory qualities of different peptides, which is useful for the food industry. The peptide solubility can be assessed using the Innovagen server (www.innovagen.com/proteomics-tools), while hydrophobicity information is available from Peptide2 (www.peptide2.com). Meanwhile, in silico toxicity of the peptides can be calculated using the ToxinPred server (crdd.osdd.net/raghava/ toxinpred), which uses an SVM threshold score below zero to indicate non-toxicity, or AllerTOP (http://www.ddg-pharmfac.net/AllerTOP/), which makes use of auto cross covariance (ACC) transformation to make in silico predictions concerning toxicity and allergenic properties. The 3D structure of peptides can be assessed in silico using the PEP-FOLD tool V3.5 (http://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD3/) which takes into consideration the amino acid sequences in making the prediction, while the 3D structure images were created with Discovery Studio software 2016 (Accelrys Software Inc., San Diego, CA, USA).

Cellular antioxidant activity assay (CAA)

Quantitative real-time reverse transcription-polymerase chain reaction (qPCR)

The qPCR analysis required Caco-2 cells to be cultured with different concentrations of DY-8 and IK-6 peptides using 6-well plates for durations of 24 hours and 48 hours. The cells were then harvested through EDTA-free trypsinization and rinsed thoroughly two times in cold PBS. Subsequently, a MasterPure RNA Purification Kit (Epicentre, USA) was used to obtain the RNA from the cells in line with the instructions of the manufacturer. The reverse transcription of the total RNA (1 μg) was conducted with Oligo-dT primers using a precision nanoScript II reverse transcription kit (PrimerDesign, Camberley, UK). A NanoDrop 2000 UV-Vis spectrophotometer (Thermo Fisher Scientific, Inc.) was then used to measure the RNA concentration at 260 nm. The different primers used in the qPCR assays are shown in S1 Table, while all assays were conducted using MyGo Pro^® Real time PCR apparatus (IT-IS International Ltd., Stokesley, UK). Meanwhile, the 20 μL PCR reaction was carried out using 1 μL of cDNA, 1 μL of each primer (10 mM), 7 μL of ultrapure water, and 10 μL of 2× qPCRBIO SyGreen Mix (PCR Biosystems Ltd., London, UK). The reaction was started through a 2-minute activation step at 95°C, whereupon 40 ten-second cycles followed at 95°C, and then cycles of 20 seconds at 60°C, and 30 seconds at 72°C, to form a melting curve in the range of 55–95°C using one minute for each stage. Determination of the The cycle threshold (C_t) was determined through experimentation in triplicate, whereupon the data were normalized using β-actin as the internal control. The obtained C_t value could then be employed to determine the relative gene expression via the method of Livak and Schmittgen [22], as shown below in Eq (4): (4) where ΔΔC_t represents the heightened target gene threshold cycle when the threshold cycle is rising for the housekeeping gene (β-actin). Application of the formula generates a value, which is indicative of no change if the value equals one. If the value is less than one, gene expression is understood to have been reduced, whereas a value exceeding one indicates an increase.

Apoptosis

To examine apoptosis, dual staining of cells with Annexin V-FITC and propidium iodide (PI) was conducted following the protocol of the Annexin V-FITC/PI detection kit (BioLegend Inc., San Diego, CA, USA). Caco-2 cells were initially seeded in 25 cm² culture flasks (1 × 10⁷ cells/flask) in CM with 2.0 mM L-glutamine. After overnight incubation, DY-8 and IK-6 peptides at various concentrations and 0.5 mg/mL of doxorubicin as a positive control were added. Incubation continued for 24, 48, or 72 hours at 37°C. Cells were harvested using a scraper, washed in cold phosphate-buffered saline (pH 7.2) with 1% (v/v) FCS, and centrifuged. The cell pellets were then resuspended in Annexin V-binding buffer (100 μL), and 100 μL of the cell suspension was mixed with 5 μL of PI solution and 2.5 μL of Annexin V-FITC in a microcentrifuge tube. After vortexing and incubation in darkness for 15 minutes at room temperature, 200 μL of Annexin V-binding buffer was added. Apoptosis was detected using a Cytomics EC500 MPL flow cytometer (BD FACSCalibur, BD Biosciences, Singapore), and data were analyzed using FlowJo software version 7 (FlowJo LLC, Ashland, OR) [23].

Western blot analysis

The expression levels for the apoptosis signaling pathway protein caspase-3, cleaved caspase-3, caspase-9, Bax, and Bcl-2 were analyzed via Western blot analysis. To do so involved treating the Caco-2 using different concentrations of the DY-8 and IK-6 peptides over a duration of 48, and 72 hours. Harvested cells were subsequently lysed using NP40 Cell Lysis Buffer supplemented with a protease inhibitor cocktail. Protein concentrations were measured using the BSA method. In every treatment scenario, the protein extract was exposed in equal amounts to electrophoresis with 12% sodium dodecyl sulfate poly-acrylamide gel electrophoresis (SDS-PAGE) gel before moving to polyvinylidene fluoride (PVDF) membranes (Millipore, Bedford, MA, USA). The membranes were then blocked with 3% skimmed milk (w/v) in tris-buffered saline TBST (1× TBS buffer and 0.5% Tween 20) for one hour while maintaining room temperature. The primary antibody (S1 Text) was used to probe the membranes, followed by incubation with gentle overnight shaking at 4°C. The secondary antibodies (1:5000) were then washed thrice in TBST, and added to the membranes for 1 hour of room temperature incubation. The three washes in TBST were then repeated before the bands were developed using enhanced chemiluminescence HRP substrates (GeneDirex, Taiwan). Band images were then collected using an ImageQuant LAS 500 (GE Healthcare, Uppsala, Sweden), while the variations in protein expression were evaluated via ImageJ software.

Preparation of the system and molecular docking

Analysis of molecular docking establishes the binding modes for the DY-8 peptide to the three principal Bcl-2 families of antiapoptotic proteins: Bcl-2, Bcl-xL, and Mcl-1. The Protein Data Bank provided the details of the 3D structures of Bcl-2, Bcl-xL, and Mcl-1 based on the respective Protein Data Bank accession codes of 2XA0 [24], 3FDL [25], and 2NLA [26]. The tleap module used in AMBER20 [27] was employed to create the peptide structure, whereupon the 3D structure of the DY-8 peptide underwent energy-minimization via 2000 steps of steepest descent and 1000 steps of conjugated gradient methods, on the basis of the implicit solvent calculation (igb = 2). Docking simulation was then carried out via the HDOCK webserver (http://hdock.phys.hust.edu.cn/) [28] to examine the molecular docking interactions of the prepared peptides and proteins. Those docked complexes which achieved the lowest HDOCK scores were then selected for further analysis of the protein-peptide interactions via the Discovery Studio Visualizer (BIOVIA, San Diego, CA, USA) and also the ChimeraX program [29] which enables the creation of 3D visualizations.

Statistical analysis

Assays were all performed in triplicate, while data were shown as mean ± standard deviation (S.D.). Analysis relied upon the Statistical Package for Social Sciences (SPSS) version 15.0 (SPSS, Inc., USA). Data from the experiments were evaluated via Duncan’s multiple range tests in order to compare data at a statistical significance level of p < 0.05, or in some cases, p < 0.01.

Pepsin-pancreatin hydrolysis of Phlai protein isolates

For rapid screening of new foods or delivery systems, in vitro digestion techniques can replace animal or human digestion models, offering cost-effective, fast, and safe testing procedures. Unlike industrially produced bioactive peptides, which tend to degrade rapidly and lose their properties upon digestion, the preparation of protein hydrolysates is often employed in vitro gastrointestinal (GI) digestion models. This process is known to generate antioxidant peptides through enzymatic digestion in the gastrointestinal tract [30]. Many oligopeptides with 2–10 amino acids can cross the upper gastrointestinal absorption channel. Dietary proteins containing antioxidant peptide segments may explain their health benefits beyond regular nutrition. Human digestion primarily involves pepsin, an endopeptidase with strong cleavage activity for peptide bonds linked to hydrophobic amino acids, particularly phenylalanine, tyrosine, tryptophan, aromatic properties, and carboxyl-side amino acids like leucine and glutamic acid. In the small intestine, pancreatin, comprising exopeptidases (carboxypeptidases A and B) and endopeptidases (trypsin, α-chymotrypsin, and elastase), cleaves peptide bonds involving amino acids such as arginine and lysine due to its broad specificity [31].

The findings in this research indicate the importance of Phlai protein to provide bioactive peptides which are released when conditions approximating gastrointestinal digestion are replicated in vitro. Table 1 confirms that sequential incubation with pepsin and pancreatin produced potent digests with significant antioxidant activity. Hydrolyzed samples outperformed non-hydrolyzed ones in terms of antioxidant capacity. Enzymatic hydrolysis releases bioactive peptides with hydrogen donors that react with free radicals, terminating chain reactions. Antioxidant peptides in PPH obtained through pepsin-pancreatin digestion likely contain strong hydrophobic residues in their amino acid sequences, contributing to their potent antioxidant activity. In previous studies investigating the antioxidant activity of peptides obtained through pepsin-pancreatin hydrolysis, Saisavoey et al. [32] showed that pepsin hydrolysates derived from protein in rice bran could successfully scavenge free radicals. In addition, sunflower protein was hydrolyzed using pepsin and pancreatin by Megías et al. [33] who reported that the resulting hydrolysate demonstrated notable copper-chelating characteristics because the peptide sequence contained histidine and arginine. Meanwhile, Girgih et al. [34] hydrolyzed hemp protein isolate, initially with pepsin, and then with pancreatin in order to create biopeptides. It was possible to identify 23 peptide sequences from this particular protein hydrolysate, whereupon in vitro and in vivo testing showed that the identified peptides offered excellent antioxidant properties.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of values for ABTS, DPPH, NO, and FRAP indicating the IC₅₀ values, for crude protein, PPH, and the five PPH-fractions derived through pepsin-pancreatin hydrolysis (using 10 kDa, 5 kDa, 3 kDa, and 0.65 kDa MWCO membranes).

https://doi.org/10.1371/journal.pone.0304701.t001

Free radical scavenging activity after PPH size fractionation using ultrafiltration

It is important to produce peptides of smaller size since they can be beneficial for health. This can be done via ultrafiltration, whereby a membrane allows fractionation to create smaller sized peptides. This approach works better than traditional techniques, such as gel chromatography, due to the lower cost, scalability, efficiency, and energy savings. In addition, the approach is environmentally friendly and can be carried out at room temperature. As a result, there are no undesirable precipitates formed during the process of separation [35]. In this study, the pepsin-pancreatin PPH underwent fractionation via ultrafiltration using MWCO membranes of 0.65, 3, 5 and 10 kDa. The best radical scavenging potential according to the findings shown in Table 1 came from the smallest MWCO fraction (MW < 0.65 kDa), which was selected ahead of the other fractions for further examination. Peptides with low molecular weight typically offer strong antioxidant activity due to lower steric hindrance, while the hydrophilic and hydrophobic amino acid residues form the active components of short chains and are therefore readily available for reactions with free radicals. They can also act as carriers capable of being absorbed through the intestinal barrier, thus enabling them to perform their antioxidative role [36]. It has also been shown in the literature that the antioxidant activity of many fractions is enhanced via ultrafiltration, as confirmed by Aursuwanna et al. [17] who reported that hydrolysates drawn from the proteins of lingzhi mushrooms showed better scavenging ability in the case of the fractions of smaller molecular size. Similarly, Sonklin et al [37]. found that hydrolysates from the proteins of mungbean meal offered superior DPPH scavenging capabilities for the fractions smaller than 1 kDa, while Wu et al. [38] found higher scavenging ability for fractions of low molecular weight obtained from Douchi hydrolysates. It may be the case that the fractions of lower molecular size contain a larger proportion of hydrophobic amino acids when compared to bigger peptides, and these are able to interact with peroxyl DPPH radicals. Zou et al. [36] have argued that peptide size is crucial when considering antioxidant capability, with lower molecular mass associated with greater antioxidant activity. In addition, the DPPH radical scavenging activity performed by fractions of low molecular weight may also be affected by other factors such as the solubility of the peptides, which permits the peptides to bind to the free radicals much more easily than is the case for bigger peptides which do not dissolve so readily.

RP-HPLC fractionation of MW < 0.65 kDa and assessment of the free radical scavenging activity

The fractionation of PPHs in molecular weight lower than 0.65 kDa were used RP-HPLC technique with RP-C18 column and gradient elution while allow interaction with hydrophobic analyses within a polar mobile phase. Seven sub-fractions could be obtained as indicated in Fig 1. The fractions (F₁-F₇) were collected, lyophilized, concentrated, and examined antioxidant activity (ABTS, DPPH, NO, and FRAP), the sub-fraction (F₅) showed only high potential antioxidant activity with IC₅₀ values was 5.959 ± 0.204, 7.856 ± 0.040, 22.197 ± 5.254 mg/mL, and 0.759 ± 0.002 mM per Fe²⁺ mg, respectively. The yield of each purification procedure is shown in S2 Table. The F₅ sub-fraction was then subjected to peptide amino acid sequencing through the use of Q-TOF-ESI-MS/MS analysis. According to the property of C₁₈ column, the eluted order of each fraction peak was correlated to the polarity of samples. Generally, the hydrophobic components stay firmly within the column and are harder to be eluted out. The hydrophobic qualities and time taken are important to determine their anti-oxidative peptides that occur in chromatography column [39]. The longer time of elution can refer to high hydrophobicity and rise in anti-oxidative activity those depend on amino acid composition and peptides sequence in PPHs. It was previously reported for the peptides obtained from the protein hydrolysates of perilla seed protein hydrolysate, fraction c with the longest retention time and lowest molecular weight showed the highest DPPH radical-scavenging activity (87.7%) at a concentration of 0.5 mg/mL [40]. Meanwhile, Peng et al. [41] reported the elution of two peptides include bombinin-like peptide (BLP) and bombinin hydrophobic-type peptide (BH) were 92 and 147 minutes, respectively. These two peptides were able to act biologically against both microorganism and cancer because of that effect from their sequence and α-helical cationic structure, the capabilities for anticancer was demonstrated in BH peptide due to its greater hydrophobicity and high antioxidant activity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The RP-HPLC profile for the active fraction (< 0.65 kDa) from PPH.

https://doi.org/10.1371/journal.pone.0304701.g001

Identification of peptide sequences and prediction of bioinformatic properties

Following RP-HPLC, the best antioxidant F₅ sub-fraction from PPH was selected and the amino acid sequence was determined via ESI-Q-TOF-MS/MS analysis. The mass spectra (MS) of the antioxidant peptides obtained from this F₅ sub-fraction can be seen in Fig 2. The mass spectral analysis presented evidence of a number of previously unreported peptides: (1) Asp-Gly-Ile-Phe-Val-Leu-Asn-Tyr (DGIFVLNY or DY-8; m/z = 940.478), and (2) Ile-Pro-Thr-Asp-Glu-Lys (IPTDEK or IK-6; m/z = 702.367). Alignment of the fragments using the SwissProt database (de novo deducing) was then conducted to determine the homologous region. In the case of DY-8, 100% (8/8) similarity was revealed for the amino acid sequence with the cytosolic endo-β-N-acetylglucosaminidase 1-like from Z. officinale (SwissProt accession number XP_042429169.1), along with hypothetical protein ZIOFF_004822 from Z. officinale (SwissProt accession number KAG6531052.1) which can be seen in S3 Table. IK-6 was shown to be similar to a hypothetical protein ZIOFF_061232, whereas the identity of the amino acid sequence provided a 100% (6/6) match with Z. officinale (SwissProt accession number KAG6477800.1), indicating that the protein could potentially act as a signaling or transport protein, as evidenced in S4 Table. However, the peptide is very short, and it is therefore difficult to find exact conflict sites, so the peptide could be replaced in the polymorphic regions of proteins derived from divergent subunit proteins. Innovagen’s server data affirms poor water solubility, as shown in Table 2, but our research indicates DY-8 boasts good water solubility below 5.0 mg/mL. Solubility declined with increasing concentration. DY-8’s hydrophobicity measured 50%, while IK-6 registered at 33.33%. Hydrophobicity, alongside charge and molecular size, influences peptide bioavailability in vitro. Xie et al. [42] explored the link between casein peptide hydrophobicity and in vitro bioavailability, revealing highly hydrophobic peptides exhibit strong bioavailability. However, it’s noted that excessive hydrophobicity can hinder in vitro bioavailability. Thus, while hydrophobicity is vital for transcellular transport, moderation is key, considering peptidase hydrolysis during absorption in in vitro peptide bioavailability assessment.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The mass fragmentation spectrum for the F₅ sub-fraction obtained via reversed-phase high-performance liquid chromatography (see Fig 1): (a) DY-8, and (b) IK-6.

https://doi.org/10.1371/journal.pone.0304701.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Peptide property profiles for the DY-8 and IK-6 peptides.

https://doi.org/10.1371/journal.pone.0304701.t002

Where peptides are to be used in functional foods, it is necessary to screen for undesirable qualities such as toxicity and allergenicity. This can best be achieved using bioinformatics. This is important because the use of peptides depends upon safety as well as bioavailability, so virtual screening is a quick and cost-effective way to check for safety concerns such as toxicity. ToxinPred, a tool utilizing information on amino acid composition and location, supports in silico predictions for peptide toxicity. Machine learning models, aided by a quantitative matrix, predict toxicity based on known properties of the peptides [43]. The SVM scores in Table 2, falling below zero, affirm the non-toxic nature of the peptides. Simultaneously, the rising concern of food-related allergies affecting approximately 8% of children and up to 2% of adults underscores the need for proactive measures. As many common allergens are present in foods derived from plants or animals, the European Food Safety Authority (EFSA) advocates in silico testing to predict potential issues stemming from exposure to food proteins. AllerTOP, employing a k-nearest neighbor algorithm, classifies peptides using a training set comprising well-known allergens and non-allergens from various species [44]. AllerTOP analysis assures minimal risk of allergenic problems with DY-8 and IK-6 peptides. As a means of assessing the other activities of peptides, the cheapest and fastest method is to involve computer simulations, such as BIOPEP, [27] although it must be acknowledged that these results may be somewhat limited in terms of the biological activities the tool might consider. In the case of DY-8 and IK-6, it was indicated that these peptides should be bioactive, with BIOPEP indicating the likely inhibition of dipeptidyl peptidase-IV (DPP-IV) and angiotensin-converting enzyme I (ACE), in addition to activity which would stimulate the uptake of glucose. These predications have their basis in specific short sequences, primarily comprising just two or three amino acid residues, which are a part of the sequences as shown in Table 2.

Antioxidant activity confirmation and the structure-activity association in synthesized peptides

After identifying the two peptides, DY-8 and IK-6, they underwent synthesis and testing to assess their antioxidant properties, as detailed in Table 3. In this study, we chose peptide synthesis post-isolation to ensure precision and reproducibility. Isolation from natural sources often yields complex mixtures, while synthesis provides well-defined and purified peptides with precise sequences. This approach enhances control, facilitates targeted investigations, and improves result reliability. Both peptides share sequences containing hydrophobic amino acids like Pro and Tyr, known for promoting antioxidant activity. The aromatic rings of Phe, Trp, and Tyr, play a role in pro-oxidant metal ion chelation, contributing to the observed antioxidant activity. Sequences with N-terminal hydrophobic amino acids (Val or Leu) and specific residues within the sequence (Pro, His, or Tyr) exhibit robust antioxidant activity. Additionally, a C-terminal with hydrophilic (Asn, Gln, Ser, and Thr) and polar residues (Arg, Asp, Glu, His, and Lys) further enhances antioxidant potential. Hydrogen-donating residues near the C-terminal support chelation and free radical scavenging, heightening antioxidant activity. Among the better residues for antioxidant activity, Trp and Tyr have been claimed to be excellent because of their hydrogen-donating indole groups, in the case of Trp, and phenolic groups in the case of Tyr. If these amino acids were located at the N- and C-terminus positions, the performance would be even better [45]. The 3D structures of the identified peptides, predicted using the PEP-FOLD tool V3.5 (S1 Fig), revealed diverse spatial structures and folding patterns, predominantly featuring β-turns (75%) or α-helices (60%) instead of β-sheets or random coils. The α-helix plays a significant role in peptide antioxidant activity due to its contextual constraints [46]. The DY-8 and IK-6 peptides exhibited α-helix and random coil structures, with no observed β-sheets. However, since peptides with good antioxidant activity typically have low molecular weight, the secondary structural factors might not exert a particularly powerful influence upon antioxidant activity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Free radical scavenging activities for the DY-8 and IK-6 peptides.

https://doi.org/10.1371/journal.pone.0304701.t003

The assessment of radical scavenging activity through DPPH and ABTS revealed distinctions due to radical solubility in varied solvents. Alcoholic solvents dissolve DPPH, enabling it to accept electrons and hydrogen ions, while ABTS dissolves in organic media and water. This allows antioxidant assays with DPPH and ABTS to explore compounds with both lipophilic and hydrophilic characteristics [47]. In contrast, NO reacts with oxygen, generating nitrite and proxy nitrite, leading to inflammation and disease. The FRAP assay evaluates the reduction of ferric ions to ferrous ions in the presence of antioxidants [48]. DY-8 and IK-6 donate electrons, aiding in the reduction of ferric ions, correlating with their scavenging abilities in DPPH and ABTS assays. Results indicate that in aqueous solutions, DGIFVKNY exhibited lower scavenging activity, while IK-6 showed greater activity in the ABTS assay than in alcoholic solvents. This suggests that the peptides’ conformation in aqueous solutions enhances electron and hydrogen acceptance to eliminate free radicals. Additionally, IK-6’s smaller size may contribute to its superior antioxidant properties compared to DGIFVKNY.

Antioxidant peptides can have their abilities enhanced by hydrophobic amino acids, which lead to improvements in the interactions of peptides which can act against cancer. These peptides enter target organs via hydrophobic interactions with membrane lipid bilayers, scavenging radicals in the outer leaflets of tumor cell membranes, effectively targeting cancer cells. Sequences with charged (Glu), heterocyclic (Pro), and aromatic amino acids (Trp, Tyr, Phe) exhibit enhanced anticancer properties. Sae-Leaw et al. [49] reported that the gelatin peptides obtained from seabass skins contained hydrophobic amino acids which would improve the antioxidant activity due to the hydrophobic cellular targets, such as polyunsaturated chains of fatty acids, becoming more readily accessible to the peptides. Furthermore, Chalamaiah et al. [50] stated that low molecular weight peptides exhibited greater anticancer activity, since this increases the quantity of hydrophobic amino acids. Understanding the structural properties of peptides from food sources is crucial in deciphering anticancer potential due to differences in cytotoxicity linked to molecular weight.

Cytotoxicity and CAA of DY-8 and IK-6 peptides on Caco-2 cells

The cytotoxicity analysis determined that the peptides DY-8 and IK-6 were both non-toxic at concentrations below 5 μM (with cell viability greater than 50%, as determined by MTT assay), as shown in Fig 3(a). Subsequent analysis revealed that both peptides demonstrated toxicity levels that exceeded 5 μM. No cytotoxic effects were detected below this threshold, suggesting a secure range for cellular viability. Specifically, cell survival was compromised at concentrations exceeding 5 μM, indicating the onset of toxicity. In the following experiments, concentrations of 0.156, 0.312, 0.625, 1.25, 2.5, and 5 μM were therefore selected. The antioxidant properties of DY-8 and IK-6 were next assessed with a popular in vitro oxidative stress model involving Caco-2 cells. This study used DCFH as a fluorescent probe to measure intracellular ROS production. The hydrolyzed DCFH-DA probe indicates oxidative stress in cells by transforming into a highly fluorescent DCF in the presence of ROS. Digesting phosvitin phosphopeptide under simulated gastrointestinal conditions decreased pro-inflammatory interleukin secretion in Caco-2 cells, suggesting its potential in improving gut health and preventing oxidative stress [51]. From these findings, it is likely that DY-8 and IK-6 would be able to prevent ROS from causing oxidative damage to cells, and therefore the peptides might in future be employed to prevent ROS formation. Where the DY-8 and IK-6 peptides were involved in pre-treatment, DCF fluorescent intensity declined significantly. Results in Fig 3(b) show that DCF fluorescent intensity was similar for the peptides DY-8 and IK-6 at every concentration up to 5 mM, suggesting a cellular radical scavenging effect from both peptides.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

(a) Caco-2 cell viability after treatment at different concentrations of (yellow circle) DY-8 and (black circle) IK-6 for 72 hours, and (b) the cellular antioxidant activity of (yellow circle) DY-8 and (black circle) IK-6. Bars are used to indicate the standard deviation (n = 3).

https://doi.org/10.1371/journal.pone.0304701.g003

It is understood that the DY-8 and IK-6 peptides have hydrophobic and antioxidant amino acid residues, and these hydrophobic amino acid residues. Accordingly, it might be the case that these amino acid residues support DY-8 and IK-6 in controlling the oxidative damage caused by ROS. The production of ROS is widely considered to be cytotoxic, and has been linked to various adverse health conditions. As the concentration of ROS increases at the gastrointestinal level, intestinal pathologies are more likely to develop, so it would be useful to discover food antioxidants capable of protecting the intestinal epithelium when oxidative stress occurs. Both DY-8 and IK-6 have shown such abilities in maintaining cell viability and preventing intestinal oxidative damage when tested using the Caco-2 cells. Similar findings have been reported for the split gill [51] and lingzhi edible mushrooms, [17] which deliver protection in ABAP-stimulated Caco-2 cells. These compounds are known to be both healthy and safe, with a low molecular weight. Their activity levels are high and they can be readily absorbed. They also have the ability to combine well with non-peptide antioxidants to produce a better protective outcome.

Analysis of flow cytometry

To assess the apoptotic capacity of DY-8 and IK-6 peptides, varying concentrations and incubation periods were applied to Caco-2 cells. Apoptosis rates were determined using annexin V-FITC/PI double staining analysis, with concentrations below 5 μM chosen for assessment. Notably, while the peptides induced apoptosis, no significant cytotoxicity was observed, underscoring the controlled nature of apoptosis compared to general cell damage. These findings underscore the effectiveness of the selected peptide concentrations in inducing apoptosis without compromising overall cell viability. The results revealed a significant increase in apoptosis rates in Caco-2 cells, particularly with higher peptide concentrations and longer incubation durations. With an incubation period of 72 hours at different peptide concentrations, the apoptotic cell percentages shown in Fig 4 for the Caco-2 cells in the case of both the DY-8 peptide and also the IK-6 peptide, confirming that cell apoptosis induced by these peptides occurred in a dose-dependent manner. In earlier research, human cervical cancer cells were examined for cell apoptosis when exposed to the ZXR-1 peptide (128 μg/mL) for 6 hours. Early apoptosis results were 55.1%, compared to 2.47% for the control, while late apoptosis measured 5.8%, compared to 0.3% for the control. It was apparent that the ZXR-1 peptide was able to promote apoptosis through the targeting of the mitochondrial membrane which released apoptotic factors via the intrinsic pathways [52]. Similarly, the work of Ortiz-Martinez et al. [53] compared the antiproliferative effects of the peptide fractions obtained from albumin alcalase hydrolysates of white hybrid (Asgrow-773) and those obtained from quality protein (CML-502) maize. A model of human liver cancer was created in vitro, and it was found that the apoptosis rates of HepG2 cells exposed to Asgrow-773 were greater than was the case for CML-502 according to the flow cytometry results. In this experiment, the control represented 100% of the cells, whereupon every fold change indicated an increase by 100%. For Asgrow-773, the rise in late apoptotic events was evaluated as 484%, while for CML-502 the value was 400%. Furthermore, CML-502 saw a rise in early apoptotic events amounting to 567%. The findings indicated that both genotypes exerted antiproliferative effects by inducing the apoptosis of HepG2 cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The flow cytometry analysis for Caco-2 cells which had been cultured for 24 hours (A–F), 48 hours (G–L), or 72 hours (M–R) in CM.

Cells are labeled with FITC-annexin V and PI following culture with different (a) DY-8, and (b) IK-6 peptide concentrations. (A, G, and M) nothing (negative control), (B, H, and N) 0.625 μM, (C, I, O) 1.25 μM, (D, J, P) 2.5 μM, (E, K, Q) 5 μM of DY-8 peptide and (F, L, and R) 0.5 μg/mL doxorubicin (positive control). Quadrants: Bottom left—normal or live cells; Top left—necrotic cells; Bottom right—early apoptotic cells; Top right—late apoptotic cells. These data from 3 trials underwent conversion to Annexin V and viability dot plots. Bar graph presenting apoptotic cells; (c) DY-8, and (d) IK-6 peptide, each bar shows the mean ± standard deviation while all tests are carried out in triplicate. ^a-cValues within a row with different letters are significantly different (p > 0.05).

https://doi.org/10.1371/journal.pone.0304701.g004

Apoptosis-related gene expression influenced by DY-8 and IK-6

There are two caspase-dependent pathways which serve to mediate the morphological and biochemical changes in cells due to apoptosis. These are the extrinsic pathway, known as the death receptor, in which caspase-8 is activated, and the intrinsic, or mitochondrial, pathway in which caspase-9 activation occurs. In either scenario, the caspase activation results in the further activation of caspase-3. The caspases are intracellular cysteine proteases which show specificity for aspartic acid residues, and when triggered by drugs they can induce apoptosis in many different kinds of cancer cells. The initiator and receptor caspases have specific roles to play in apoptosis. In a majority of apoptotic pathways, caspase-3 cleaves and breaks down the different parts of the cells which are involved in DNA repair and regulation. After activation, caspase-3 cleaves various cellular substrates, which in turn causes membrane blebbing, the fragmentation of DNA, cell structure disassembly, and eventually the death of the cell [54]. To learn more about the pathways for cell apoptosis which would be induced by DY-8 and IK-6, this study evaluated the caspase-3, -8, -9, p38-MAPK, and Bcl-2 activity in the context of treated Caco-2 cells. Fig 5 shows that when Caco-2 cells are introduced to DY-8 or IK-6, the caspase-3 and -9 activities are significantly increased in comparison to untreated cells, with the response significantly affected by both time and dosage (p < 0.05). Caspase-3 expression was upregulated 5-fold, while for caspase-9 the upregulation was 11-fold, after 48 hours of treatment with 1.25 μM of the DY-8 peptide. In contrast, caspase-8 showed no notable changes with various concentrations or incubation periods in comparison to the control, and therefore it is apparent that the induced apoptosis may involve the intrinsic pathway in this particular cell line. Furthermore, when the IK-6 peptide is activated, this leads to a notable increase in p38-MAPK gene expression when compared to the situation with the DY-8 peptide. Apoptosis can be induced by p38-MAPK as it alters the mitochondrial membrane properties, causing cytochrome C to be released, and in turn affecting the caspase-3 and -9 expression. Meanwhile, the DY-8 peptide has the activation effect of increasing the expression of caspase-3 and -9 to a greater extent than occurs with the IK-6 peptide. This may be because higher levels of p38-MAPK expression do not lead to apoptosis only via the upregulation of caspase-3 and -9, but instead can have an influence on various cellular mechanisms, such as proliferation and proinflammatory cytokine release, including the release of tumor necrosis factor. Research conducted by Fan et al. [55] revealed the influence of three peptides derived from quinoa, finding that they were able to induce p38-MAPK-related gene expression, which strongly upregulated caspase-3 and -9 in the context of Caco-2 cells. Meanwhile, Rasaratnam et al. [56] studied a VS-9, a new anticancer peptide obtained from garlic, which could inhibit MOLT-4 and K562 leukemic cell line proliferation while simultaneously having almost no effect upon normal PBMC. VS-9 was able to induce apoptosis while upregulating the mRNA levels of caspase-3, -8, -9, and Bax, and downregulating Bcl-2, Bcl-xL, and Bcl-w. In the case of peptides obtained from Laminaria japonica, there was a rise in p38-MAPK and caspase-3-related gene expression when the peptides were introduced to H22 tumor cells [57].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The influence of the DY-8 and IK-6 peptides upon apoptotic signaling pathways.

Cells underwent incubation at different concentrations for 24 hours (A, C) with DY-8 peptide, and 48 hours (B, D) with IK-6 peptide. Following treatment, apoptotic mRNA transcripts were evaluated via quantitative reverse transcription—polymerase chain reaction (RT-qPCR). (mRNA expression normalization was performed against β-actin, to the control). Experiments were carried out in triplicate with 3 biological replicates. Each point represents the mean ± S.D. *p < 0.05 indicates a significant difference in comparison to the non-treated control via one-way ANOVA.

https://doi.org/10.1371/journal.pone.0304701.g005

Influence on protein expression of DY-8 in the context of apoptosis

It was finally necessary to examine how DY-8 affects protein expression, and in turn how it can induce intrinsic apoptosis. The protein expression levels can be determined via Western blot analysis, investigating caspase-3, cleaved caspase-3, caspase-9, Bax, and Bcl-2, while β-actin was used as the loading control, as down in Fig 6 (S1 Raw images). Various concentrations were used, and the examination was carried out after 48 and 72 hours. Caspase-3 can indicate the induction of the mitochondrial apoptotic pathway, and its expression levels were shown to rise. For cleaved caspase-3, the induction of mitochondrial dysfunction was observed at a higher rate than the control, whereas Bcl-2, in contrast, is an anti-apoptotic protein whose expression levels dropped, suggesting that Bcl-2 played no role in disrupting mitochondrial functions such as cytochrome C release. Cell apoptosis is also supported by increased Bax expression. Bax is able to create a permeable membrane as it binds to the mitochondrial membrane, resulting in the formation of a mitochondrial membrane channel. Meanwhile, cytochrome C, which is a mitochondrial protein that serves to activate caspase-3, will signal intrinsic apoptosis [54]. There was an additional increase in caspase-9 expression along with the rise in caspase-3 expression, since it undergoes activation within the apoptosome complex in order to sustain catalytic status and also to activate caspase-3, leading to the induction of apoptosis at 48 hours (Fig 6(a) and 6(b)) and also 72 hours (Fig 6(c) and 6(d)) in the case of Caco-2 cells undergoing treatment with the DY-8 peptide. These findings show that the DY-8 peptide is able to induce apoptotic gene expression in Caco-2 cells, and that Bcl-2 and Bax proteins can regulate the apoptotic process by maintaining control over the mitochondrial function. The Bcl-2 monomers or homodimers play a role in promoting survival whereas Bax homodimers promote apoptosis. Furthermore, the activity of the ion channel relating to Bcl-2 and Bax might manage apoptosis through the effect upon mitochondrial membrane permeability. Bcl-2 and Bax both serve as transcriptional targets of p53, and when DNA damage occurs, this will result in the induction either of cell cycle arrest or apoptosis [54, 58]. This study suggests that p53 induces the up-regulation of Bax and down-regulation of Bcl-2 in Caco-2 cells treated with the DY-8 peptide.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Western blot analyses indicate changes in the associated protein levels for the apoptotic Caco-2 cells following treatment with the DY-8 peptide.

Whole-cell lysates prepared from Caco-2 cells underwent treatment using the DY-8 peptide at different concentrations for 48 hours (a, b) and 72 hours (c, d), followed by immunoblotting with anti-caspase-3, anti-cleaved caspase-3, anti-caspase-9, anti-Bax, anti-Bcl-2, and anti-β-actin antibodies. One representative blot drawn from 3 independent experiments is presented. The findings from independent experiments performed in triplicate can also be seen. Data take the form of mean ± S.D. of the fold changes (relative to β-actin expression). *p < 0.05 indicates a significant difference in comparison to the non-treated and DMSO-treated control via one-way ANOVA.

https://doi.org/10.1371/journal.pone.0304701.g006

From the molecular perspective, the intrinsic or the extrinsic apoptotic pathways activate the caspases. Mitochondrial membrane permeabilization is an important component of the intrinsic pathway, along with cytochrome C release into the cytosol, apoptosome formation, and the subsequent activation of caspase-9 and other downstream caspases, which can cause the fragmentation of DNA. The activation of executioner caspases such as caspase-3 and -7 can be modulated by caspase-9 via proteolysis, which transmits the apoptotic signal to the execution phase. Meanwhile, in the extrinsic pathway, procaspase-8 is an initiator caspase which undergoes proximity-induced auto-activation as it is recruited by the adaptor protein FADD (Fas-associated death domain) to the death domain-containing receptors [59]. The treatment of Caco-2 cells with the DY-8 peptide leads to the up-regulation of caspase-9 and -3, but there was no change to the level of caspase-8 mRNA. The findings from this current research indicate clearly that the DY-8 peptide is able to induce apoptosis effectively through the intrinsic mitochondrial pathway when treating Caco-2 cells. Earlier research has also found that bioactive peptides offer the possibility of inducing apoptosis-related protein expression. One bioactive peptide derived from the protein of chickpea seeds (Cicer arietinum L.) was able to induce apoptosis-related protein expression when used to treat the Ishikawa cell lines of human endometrial cancer. The outcome suggested lowered Bcl-2 expression and notably enhanced expression of caspase-3 and Bax in the Ishikawa cells according to the Western blot analysis results [60]. Meanwhile, the work of Fan et al. [55] showed that novel peptides obtained via the in vitro digestion of quinoa protein demonstrated anti-colon cancer activity when tested in Caco-2 cells. The Western blot findings revealed lower Bcl-2 expression but significantly higher caspase-3 expression. The progression of cancer might be arrested by quinoa peptides since they are able to regulate certain proteins related to cancer, so they may find a role in functional foods as a treatment for colon cancer. Finally, the work of Liang et al. [61] described the ability of rice-derived bran bioactive peptides (RBAP) to induce apoptosis-related proteins in human umbilical vein endothelial cells (HUVECs). The findings from the Western blot assay indicated the significant up-regulation of pro-caspase-3 and cleaved caspase-3 protein levels in the H₂O₂ injury group. However, the levels of pro-caspase-3 protein under H₂O₂ treatment underwent notable down-regulation in comparison to the control and to the group undergoing RBAP treatment. This suggests that the pro-caspase-3 and cleaved caspase-3 proteins rely upon the cytochrome C process and play important roles in cell apoptosis.

System preparation and the molecular docking of antiapoptotic proteins

The HDOCK webserver was used to examine the molecular docking interactions of the DY-8 peptide with Bcl-2, Bcl-xL, and Mcl-1. Fig 7 shows that each of the antiapoptotic proteins under investigation had four hydrophobic groove pockets designated as P1-P4, whose role is to bind to the BH3 hydrophobic residues [62]. The findings confirm that F4 and L6 of the DY-8 peptide could be enclosed respectively by the P1 and P2 pockets. The stability of these two positions of the peptide was due to hydrophobic interactions with various protein residues: Met115, Leu119, and Val133 of Bcl-2, Val126 and Phe146 of Bcl-xL, as well as Met231, Leu235, Val249, and Phe270 of Mcl-1 toward F4, and also Phe104, Met115, and Ala149 of Bcl-2, Phe97, Leu108, and Ala142 of Bcl-xL, as well as Phe228, Met231, and Phe270 of Mcl-1 toward L6. This result concurs with the observation of the x-ray structure of the Bcl-2 antagonist, S55746, which binds to Bcl-2 to show that the phenyl moiety of S55746 was well occupied in the Bcl-2 P1 pocket [63, 64]. Notably, the hydrophobic residues (Leu94, Iso90, Iso97, and Phe101) of the Bim BH3 peptide exhibit intimacy with the hydrophobic residues in Bcl-xL, such as Leu, Val, Phe, and Tyr [65]. Bax BH3 peptide residues including Glu9, Ala13, Gly20, and Leu23 will bind to the sockets of Bcl-2 and Bax BH3 interactions [66]. Furthermore, Noxa and Puma form bonds to a hydrophobic groove in the Mcl-1 residues, such as Met212, Val230, Val234, Thr247, and Phe251 [67]. Similarly, the binding of the crystal structures of Bcl-xL and Mcl-1 to the BH3 peptide of BIM, and the binding of Bcl-2 to the Bax BH3 peptide indicate that these antiapoptotic proteins have a P2 pocket which is easily recognized by the amino acid leucine [25, 26, 68], just as the L6 of DY-8 peptide exhibited in our own docking results. In addition, it was possible for the amino acid tyrosine (Y8) at the peptide C-terminal to bind at the P4 pocket for each of the antiapoptotic proteins, creating van der Waals interactions with the adjacent residues as shown in Fig 7. Hydrophobic interactions were also observed between the Y8 residue and Val148 of Bcl-1, Val141 of Bcl-xL, and V220 of Mcl-1 via Pi-alkyl and Pi-sigma interactions. This result was appropriate based upon the structure of Bcl-xL and Mcl-1 in complex with the BIM BH3 peptide, suggesting that phenylalanine would be the ideal amino acid for binding to this pocket [24, 26, 68]. This result can be considered similar to the case in our own study of the aromatic amino acid tyrosine (Y8) of the DY-8 peptide. It is therefore the case that in silico results can provide valuable guidance in the synthesis of peptides which are better able to target the Bcl-2 family of antiapoptotic proteins.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. 3D presentation of the molecular docking results (top) and 2D interactions (bottom) for the binding of the DY-8 peptide to three antiapoptotic proteins Bcl-2, Bcl-xL, and Mcl-1.

https://doi.org/10.1371/journal.pone.0304701.g007