A mini-library system to investigate non-essential residues of lipid-phosphatidylserine (PS) binding peptide–peptoid hybrid PPS1

Tolerance of various physiochemical modifications on the 1st and 4th position residues of lipid-phosphatidylserine binding anti-cancer peptidomimetic PPS1D1.

Abstract

We recently identified a peptide–peptoid hybrid, PPS1, which specifically recognized lipid-phosphatidylserine (PS). PPS1 consists of distinct positively charged and hydrophobic residue-containing regions. The PPS1 monomer was inactive, but the dimeric form, PPS1D1, displayed strong cytotoxicity to lung cancer cells compared to normal cells in vitro, and reduced the tumor growth in vivo. The minimum pharmacophore of PPS1D1 showed that the first (methionine) and fourth (N-lysine) residues were not important for PPS1D1 cytotoxic activity. In this study, we further investigated these two residues, in particular the fourth residue that lies between the most important four-residue hydrophobic region and two positively charged residues, to determine whether replacements of these moieties could gain activity improvements or render PPS1D1 totally insensitive for binding recognition. The positively charged fourth residue N-lysine was replaced with substituents having varied physiochemical properties, such as aromatic-hydrophobic, aliphatic-alicyclic, heterocyclic, and negatively charged residues, developing a mini-library of 39 derivatives. The standard 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) colorimetric and/or the calcein AM cell viability assays performed on HCC4017 lung cancer cells indicated that the fourth position of PPS1D1 was insensitive to most changes, except that reversal of the negative charge significantly affected the activity. This observation may be due to the neutralization of the nearby positively charged residue that is essential for binding. In addition, shortening each monomeric sequence by eliminating the methionine at the first position did not affect the activity.

1. Introduction

Cancer therapeutic development has achieved significant progress, but the very heterogeneous expression of protein drug targets impedes the development of universally effective anticancer drugs.1–3 Therefore, there is a need to identify universally present cancer-specific biomarkers. One such biomarker is the anionic lipid-phosphatidylserine (PS).4–6 It is found in the inner layer of normal cell membranes, but in cancerous and tumor endothelial cells it is reported to be transferred (i.e., flipped) to the outer layer of the lipid membrane, thereby providing a promising cancer biomarker target with increased specificity over that of normal cells.4,7–12 More importantly, this PS flipping in cancer and tumor endothelial cells in the tumor microenvironment is universal, indicating that PS-targeting drugs may have efficacy as common anticancer drugs.7,8,13–17

We recently identified a peptide–peptoid hybrid, PPS1, which recognized PS on series of lung cancer cells.13,14 PPS1 displayed specific binding towards lipids that had an overall negative charge which were present on cancer cells, but not on normal cells.14,17 The simple dimeric form of PPS1, PPS1D1, displayed cancer cell killing activity, but had no effect on normal cells.13 Our recently reported alanine/sarcosine scan studies identified two positively charged residues at the second and third positions and four C-terminal hydrophobic residues that composed the minimum pharmacophore15 (Fig. 1). This suggested that the first (methionine) and fourth (N-lysine) residues were not essential for the binding and activity of PPS1D1.

Fig. 1. Structure of PPS1D1. Highlighted are the non-essential first and fourth residues of PPS1D1 in each monomeric unit.

Based on previous studies, we hypothesized that the additional negative charge on PS interacts with two positively charged residues that are present at the second and third positions of each of the PPS1 monomeric sequences of PPS1D1. In addition, the hydrophobic regions of each of the PPS1 monomeric sequences (each consists of four important residues) might interact with the hydrophobic tail regions of PS embedded within the cell membrane. In the current study, we further investigated the first and fourth positions that are the two non-essential residues of the PPS1 monomeric sequence. The N-lysine (Nlys) at the fourth position was replaced with substituents having varied physiochemical properties, because it was placed between the essential four-residue hydrophobic region and two positively charged residues of each PPS1 monomeric sequence of the PPS1D1 dimer. This provides a chance of identifying improved derivatives of PPS1D1 and also helps us to narrow down the PS recognition mechanism of PPS1D1 in the future. Assuming that the first methionine residue was also not essential (as previous minimum pharmacophore study revealed), we deleted it from each of the two PPS1 monomeric backbones to determine if the resulting PPS1D1 derivative retains its activity.

2. Experimental

2.1. Synthesis of PPS1D1

PPS1D1 was synthesized using NovaSyn® TGR resin (EMD Millipore, Hayward, CA, USA) and utilizing standard 9H-fluoren-9-ylmethoxycarbonyl (Fmoc)-peptide synthesis protocols and microwave-assisted peptoid synthesis.18,19 Two hundred mg of beads (0.25 mmol g^–1 loading capacity) in 5 mL of dimethylformamide (DMF) were swelled for 30 min. Fmoc-lys-(Fmoc)-OH [147.7 mg, 5.0 equivalents (eq.)], hydroxybenzotriazole (HOBt) (5.0 eq.), 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) (5.0 eq.), and N,N-diisopropylethylamine (DIPEA) (10.0 eq.) in 2 mL of anhydrous DMF were added to the beads, and the reaction vessel was incubated overnight with shaking. Both Fmoc groups were deprotected simultaneously by treating the beads twice with 20% piperidine in DMF (2 mL for 10 min each). Subsequent amino acids, Fmoc-Met-OH, Fmoc-d-Lys-(Boc)-OH, and Fmoc-Lys-(Boc)-OH, were coupled using similar reaction conditions for a reaction time of 2.0 h, and the Fmoc groups were removed each time using the method mentioned above. Subsequent moieties were then synthesized using microwave-assisted peptoid synthesis protocols. One mL each of bromoacetic acid (1.0 M) and N,N-diisopropylcarbodiimide (1.5 M) was added to the reaction vessel containing the beads, and gently shaken for 30 s. The reaction vessel was then placed in a microwave oven for 15 s (1000 W) and the power was set at 10%. The reaction vial was shaken again for 30 s and the microwave procedure was repeated for another 15 s. The solution was drained and washed 10 times with 2 mL of DMF. The reaction vessel was then treated with 2.0 M Nlys (Boc-1,4-diaminobutane), a primary amine, and microwaved two times for 15 s after gentle shaking as described above. These two-step peptoid coupling procedure was then used to couple four amines, 4-methoxybenzylamine (Npmba), (R)-(a)-methylbenzylamine (Nmba), piperonylamine (Npip), and (R)-(a)-methylbenzylamine (Nmba). The synthesized compound was cleaved from the beads with the simultaneous removal of acid-labile protecting groups by treating the beads with 2 mL of 95% trifluoroacetic acid (TFA), 2.5% water, and 2.5% triisopropylsilane (TIS) for 2 h. The compound was then purified using high-performance liquid chromatography (HPLC). Synthesis was subsequently confirmed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF) analyses.

2.2. Synthesis of compounds 1–39

Compounds 1–39 were synthesized using the same protocol as that for PPS1D1 (described in section 2.1), except that the fourth residue, Nlys, was replaced with various amines (in peptoid units) and amino acids (in peptide units) shown in Table 1. All syntheses were carried out using 50 mg of NovaSyn TGR resin (EMB Millipore). The completion of the reaction was analyzed by MALDI-TOF and the compounds were used without further purification.

Table 1. Mini-library derivatization of the fourth residue of PPS1D1. Different moieties were used for the synthesis of the 39 different compounds replacing the fourth position N-lysine. The structures of various amines used when this position was replaced as a peptoid residue are listed in the first panel. The different amino acids used to replace this position are listed in the bottom panel (full structures of all 39 compounds are given in ESI Fig. S1–S39).

Type of substituent at the 4th position of PPS1D1	Hydrophobic		Aliphatic		Heterocyclic	Negatively charged
Amines	1	3	2	4	12
	5	6	8	10	18
	7	9	13	15	21
	11	14	20	24	26
	17	19	25	27
	22	23	28
	29
Amino acids	33		30	31	39	34
	37		32	36		35
			38

2.3. Synthesis of PC462D1

PC462D1 was synthesized using similar protocols as described above (section 2.1). The sequence of amino acid residues for PC462D1 was Fmoc-Lys-(Fmoc)-OH, Fmoc-Met-OH, Fmoc-d-Lys-(Boc)-OH and Fmoc-Gly-OH, and the removal of the Fmoc group each time was performed by using 20% piperidine as described above (section 2.1). Next, the five peptoid residues were coupled using the microwave-assisted synthesis protocol mentioned in section 2.1, and the pentameric peptoid sequence was allylamine (N-all), 2-methoxyethylamine (Nmea), allylamine, 2-methoxyethylamine, and allylamine.

2.4. MTS cell viability assay

The lung cancer cell line HCC4017 and normal HBEC-3KT cell line were obtained from the cell collection of Dr. John Minna's research group at UT Southwestern Medical Center. HCC4017 was grown in RPMI supplemented with 5% FBS. The normal lung cell line HBEC-3KT was grown with keratinocyte serum-free media (KSFM) supplemented with human recombinant epidermal growth factor and bovine pituitary extract. HCC4017 lung cancer cells (10 000 cells per well) were plated in triplicate in a clear-bottom 96-well plate and allowed to adhere overnight. The cells were treated with 20 μM crude compounds 1–39 for 24 h, then 20 μL of the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) reagent (CellTiter 96® Aqueous One Solution Cell Proliferation Assay; Promega, Fitchburg, WI, USA) was added and the samples were incubated for 2 h. The absorbance was read at 490 nm using a plate reader (Spectramax i3; Molecular Devices, Sunnyvale, CA, USA). In the MTS studies with purified compounds: (I) in section 3.2, the cells were treated with increasing concentrations of test compounds from 100 nM to 50 μM for 24 h, and (II) in section 3.4, HCC4017 and HBEC-3KT cells were treated at 1 and 20 μM concentrations of compounds 3, 6, 11, 12, 24, 37, PPS1D1 and PC462D1 for 24 h.

2.5. Calcein AM cell viability assay

HCC4017 lung cancer cells (10 000 cells per well) were plated in triplicate in a clear-bottom 96-well plate and allowed to adhere overnight. The cells were treated with compounds at 100 nM to 50 μM concentrations for 24 h. After the treatments, the culture media were replaced with 1× sterile phosphate-buffered saline containing 1 mM calcein AM (Molecular Probes, Eugene, OR, USA) and the samples were allowed to incubate for 15 min at room temperature. Fluorescence emission was measured at 520 nm using the plate reader (Spectramax i3; Molecular Devices).

2.6. Statistical analysis

Statistical analysis (two-tailed t-test) was used to verify the significance between the mean for each compound and crude PPS1D1. Differences with P < 0.05 were considered statistically significant and marked as ‘*’. All statistical analyses were performed with GraphPad Prism 7 software.

3. Results and discussion

3.1. The effects of derivatization at the fourth residue

We first wanted to determine the effects of various residue substitutions by replacing the non-essential fourth residue with residues having different physiochemical properties, such as different hydrophobicity, aliphatic-alicyclic or heterocyclic structures, and negative charges. A total of 39 new derivatives were synthesized with different amino acid residues and with N-substituted functionality (e.g. peptoid residues) (Table 1). The hydrophobic residues consisted of benzyl aryl and ethyl aryl amines with meta and para substituents having electron donating and withdrawing properties. Heterocyclic substituents were five-membered cyclic rings with N, O and/or S as heteroatoms. Negatively charged moieties had carboxyl groups (–COOH) as their functional groups. We kept the remaining residues of PPS1D1 (i.e., methionine, d-lysine, and lysine) and the hydrophobic regions unchanged. Each of these compounds was synthesized on 50 mg of resin and, after completion of the synthesis, cleaved from the resin and the structure was confirmed using MALDI-TOF mass spectrometry (ESI† Fig. S1–S39). The MALDI-TOF mass spectrometry data indicated that the compounds were successfully synthesized. Although MALDI-TOF mass spectrometry does not provide the purity level of the compounds, we tested these crude compounds for their cytotoxic activity as a library screen to get an initial indication of their activity levels. More importantly, we included purified PPS1D1 to compare the levels of crude compound activity and also included the non-active control crude PC462D1 compound.

The cytotoxicity of the compounds was determined by MTS assays in HCC4017 lung cancer cells using a 96-well array format. MTS is a tetrazolium-based salt that is reduced by cellular mitochondrial dehydrogenases to form formazan dye, which is used in the colorimetric quantitative assay of cell viability. Cytotoxicity was investigated using 20 μM of each compound. The activity results were analyzed by comparing the compounds with 20 μM crude PPS1D1, purified PPS1D1, and non-active control crude PC462D1 (Fig. 2). In this primary investigation, compounds with hydrophobic residues retained their activity or showed somewhat better results than crude PPS1D1. However, the activity was dependent on their substituted functional group. Compounds having 3,4-(methylenedioxy)benzylamine (piperonylamine), 3,4-dimethoxyphenylamine, 4-aminobenzylamine, (R)-(+)-α-methylbenzylamine, and tyrosine at the fourth position by replacing Nlys showed reduced cytotoxic activity compared to crude PPS1D1. Compounds 5, 19, 22, 23 and 29, having 2-phenoxyethylamine, (4-methoxyphenyl)methanamine, (4-chlorophenyl)methanamine, 2-(4-chlorophenyl)ethanamine and phenethylamine at the fourth position residues, respectively, showed cytotoxicity comparable to that of crude PPS1D1. However, when the fourth position N-lysine was replaced by 1,4-benzodioxan-6-amine, 3-methoxybenzylamine, β-methylphenethylamine, 4-methylbenzylamine and phenylalanine (compounds 3, 6, 11, 14 and 37 respectively), the resulting compounds showed somewhat improved activity compared to crude PPS1D1. Compounds 2, 4, 31 and 38 which have aliphatic substituents (2-methylbutyl)amine, sec-butylamine, alanine and valine as the fourth position residue, respectively, showed comparable activity to crude PPS1D1, whereas compounds with other aliphatic and alicyclic substituents had lower cytotoxic activity than crude PPS1D1, except for compounds 24 and 36, which had 3-isopropoxypropan-1-amine and leucine as the fourth position residue, respectively, showing better activity than crude PPS1D1. All heterocyclic substitutions showed similar cytotoxic effects when compared to crude PPS1D1 (compounds 12, 18, 21 and 26).

Fig. 2. MTS cell viability assay results for HCC4017 cells treated with the 39 compounds, crude PPS1D1, purified PPS1D1, and crude PC462D1 at 20 μM. The green bars are for compounds that display cell killing activity ±10% in comparison to crude PPS1D1. The blue and red bars are for compounds that display increased and decreased activity compared to crude PPS1D1, respectively. *P < 0.05.

Compound 39, which had proline as the fourth residue, showed lower cytotoxicity than PPS1D1. This is probably because it might have altered the backbone orientation of the two PPS1 monomeric peptoid chains in the PPS1D1 dimer, which could have affected lipid binding. When the fourth position Nlys was replaced with negatively charged glutamic or aspartic acids (compounds 34 and 35, respectively), loss of cytotoxic activity was observed. Compound 16 was crude PPS1D1 synthesized separately and tested along with other derivatives to establish the repeatability of the results. Similar activity levels were observed for both compound 16 and crude PPS1D1 samples, establishing the reproducibility of the synthesis and MTS assay results.

3.2. Verification of the crude library screen observations using representative purified compounds

The MTS assay results shown in Fig. 2 indicated that there were variable cytotoxic effects when the fourth residue was replaced with substituents that differed in their physiochemical characteristics. However, our previous minimum pharmacophore studies showed that the fourth position was not essential for the binding and activity of PPS1D1.15 Also, in our recent study, we used this position to connect two PPS1 monomeric units in a very unique ‘mid-linker’ multimerization process, which produced even better derivatives than PPS1D1.20 Therefore, we wanted to further verify the above noted observations by repeating the MTS assay on selected compounds that were resynthesized and purified. When verifying these results, we used representative compounds that had different types of physiochemical properties at the fourth position having hydrophobic residues (compounds 1, 3, 6, 11 and 37), heterocyclic residues (compound 12), aliphatic residues (compounds 24 and 32), and negatively charged residues (compound 34) which belong to all three categories of improved, equal and reduced activity in comparison to crude PPS1D1. The cytotoxic activity of these compounds was determined and compared with those of pure PPS1D1 and non-active control PC462D1 that were treated at a concentration gradient on HCC4017 cells (Fig. 3A and B). When Nlys was replaced by 3-methoxybenzylamine (6), β-methylphenethylamine (11), 2-thiophenemethylamine (12), 3-isopropoxypropylamine (24), or phenylalanine (37), the IC₅₀ was 7–10 μM, as compared to the IC₅₀ of PPS1D1 (13 μM) (Fig. 3A). When all these compounds were previously tested at crude levels as shown in Fig. 2, all of them displayed better cytotoxic effects than PPS1D1 indicating that our crude screen was mostly reliable. Compound 32, which had a hydroxyl group with a probable slightly negative charge, showed an IC₅₀ of 14 μM, whereas compound 34, which had prominent negative charge characteristics, showed nearly complete loss of activity (Fig. 3B). The previous crude screen data (Fig. 2) also depicted the same pattern. The result for compound 34 is prominent, as it was the only derivative that displayed a huge variation of cytotoxicity when compared to all derivatives that have been re-tested. A possible explanation for this result may be the neutralization of the positive charge (through a salt bridge) on the adjacent lysine (at the third position), which is an essential residue for the PS binding and activity. In addition, the replacement with 3,4-(methylenedioxy)benzylamine (1) also caused a decrease in IC₅₀ to 33 μM (Fig. 3B), which still followed the results of crude screen (Fig. 2). The only outlier to the crude results shown in Fig. 2 was the replacement of Nlys of PPS1D1 with 1,4-benzodioxan-6-amine (3), which gave an IC₅₀ of 17 μM (Fig. 3A) in comparison to the IC₅₀ of PPS1D1 (13 μM).

Fig. 3. MTS and calcein AM cell viability assay results on HCC4017 cells treated with selected PPS1D1 derivatives. (A) MTS assay results for compounds 3, 6, 11, 12, 24 and 37 that showed activity which was equal or slightly improved to that of PPS1D1. PC462D1 was used as a control compound. (B) MTS cell viability assay results for compounds 1, 32 and 34 that showed reduced activity in comparison to PPS1D1. PC462D1 was used as a control compound. (C) Calcein AM viability assay results on HCC4017 cells treated with selected compounds 6, 11, 24, 34, PPS1D1 and PC462D1.

As already mentioned, the verification of the MTS results that are shown in Fig. 3A and B follows the same pattern as that observed in Fig. 2 for the crude MTS assay. However, the actual differences of IC₅₀ values as compared to PPS1D1 were small. This further verified our previous findings that the fourth residue is non-essential for the activity of PPS1D1. In the case of compound 34, the loss of activity may be another indication of the importance of the positively charged third residue, as having a negatively charged group right next to it would have neutralized it. In order to further confirm the activity differences observed with the MTS assay, compounds 6, 11, 24, 34, PPS1D1 and PC462D1 were tested using another cell viability assay, the calcein AM assay, on HCC4017 cells. Calcein acetoxymethyl ester (calcein AM) is a non-fluorescent dye, which is lipophilic and therefore allows easy passage through the cell membrane. After it enters the cells, intracellular esterases cleave the ester bonds of the acetomethoxy group to form a fluorescent anionic and hydrophilic calcein dye that becomes trapped inside the cell. Nonviable cells do not contain active esterases, allowing this assay to be used as a measure of viability. The results are shown in Fig. 3C. These cytotoxicity results confirmed those of the MTS assay, thus validating our results. These data confirmed that even though the fourth residue was flanked by two important positively charged residues and a four-residue hydrophobic region, the changes made at this position had no effect on the binding and activity of PPS1D1.

3.3. The effect of removing the first residue and shortening the length of each PPS1 monomeric unit to seven residues in PPS1D1

The other residue of the PPS1 monomeric sequence that was not required for binding activity was the C-terminal methionine, according to our previous study.15 To further confirm this, methionine was removed from each of the PPS1 monomeric sequences of PPS1D1 to produce PPS1D1-M (Fig. 4A). The cytotoxicity of this compound was evaluated on HCC4017 lung cancer cells and compared with those of PPS1D1 and the non-active control (PC462D1). Both MTS (Fig. 4B) and calcein AM (Fig. 4C) cell viability assays were used to determine the cytotoxicity. The results showed that removing methionine had no effect on the activity of PPS1D1-M. This further confirmed that the first residue of PPS1D1 was not required for PPS1D1 activity, and this residue can be removed without affecting the activity of the compound.

Fig. 4. (A) Chemical structure of PPS1D1-M. (B) MTS cell viability assay results on HCC4017 cells treated with PPS1D1-M, PPS1D1 and PC462D1. (C) Calcein AM viability assay results on HCC4017 cells treated with PPS1D1-M, PPS1D1 and PC462D1.

3.4. New derivatives have no effect on normal cells

To test whether the substitution at the 4th position with various physiochemical properties would affect the specificity, the cytotoxic effects of compounds 3, 6, 11, 12, 24 and 37 were tested on HBEC-3KT normal cells. The standard MTS assay was performed by treatment of both HCC4017 lung cancer and HBEC-3KT normal cells with 1 and 20 μM concentrations of those derivatives. Compounds at 1 μM did not show any cell killing activity on both HCC4017 and HBEC-3KT cells (Fig. 5). But as expected, these compounds showed cell killing activity on HCC4017 cancer cells at 20 μM concentration. No activity was observed on HBEC-3KT normal cells at 20 μM concentration. These results showed that even after replacing the 4th residue, the compounds displayed specificity towards cancer cells and do not have any effect on normal cells.

Fig. 5. MTS cell viability assay results for compounds 3, 6, 11, 12, 24, 37, PPS1D1 and PC462D1 on HCC4017 cancer cells and HBEC-3KT normal cells treated at 1 μM and 20 μM concentrations (for each compound, the first bar shows 1 μM and the adjacent second bar shows 20 μM treatment data). The compounds did not show any cell killing activity on both HCC4017 and HBEC-3KT cells at 1 μM. However, strong cytotoxicity was observed at 20 μM concentration on HCC4017 cancer cells, but not on normal HBEC-3KT cells (the % was calculated as compared to the absorbance of untreated control wells. Some of the data showed a little above 100% because of the edge effect of the plate readings).

Conclusions

We have synthesized 39 derivatives of the lipid-PS-targeting antagonist, PPS1D1, by replacing the non-essential residue at the fourth position with hydrophobic, aliphatic-alicyclic, heterocyclic, and negatively charged residues. While this position is flanked by two important positively charged residues and a four-residue hydrophobic region, introduction of these various physiochemical properties did not affect the PS binding and activity, except the introduction of a negatively charged residue, which reduced the activity. Removal of the other non-essential residue at the first position also did not have any effect on the activity. These data further confirmed the minimum pharmacophore of PPS1D1 and the insensitivity of the non-essential residues for binding recognition, while providing new insight for future studies elucidating the binding mechanism of PPS1D1 to lipid-PS.

Conflicts of interest

The authors declare no competing interests.