Central position of histidine in the sequence of designed alternating polarity peptides enhances pH-responsive assembly with DNA

Abstract

Background

Self-assembling peptides hold great promise for medical applications, particularly as carriers for gene delivery, but their potential remains unrealized due to a limited understanding of how amino acid sequence positioning affects their properties. In this study, we designed and evaluated two alternating polarity peptides, RFH (RFRHRHRFR) and RHF (RHRFRFRHR), differing only in the position of their histidine and phenylalanine residues, to investigate the impact of sequence variation on pH-responsive DNA co-assembly and transfection efficiency.

Results

Both peptides formed stable co-assemblies with DNA at neutral pH. However, RHF retained its co-assembling ability at acidic pH, as confirmed by gel retardation studies. Coarse-grained molecular dynamics simulations further supported these findings, showing a reduced affinity of RFH for DNA and a sharper decrease in DNA binding when its histidine residues were protonated. Morphological analysis revealed that both co-assemblies underwent structural transitions with increasing N/P ratios, though their sizes and morphologies differed significantly. Biological studies demonstrated that both peptides achieved a higher transfection efficiency in 293T and HeLa cells compared to R₉, with a lower cytotoxicity than polyethyleneimine. Notably, RFH exhibited superior transfection performance at lower N/P ratios compared to RHF, likely due to its distinct histidine and phenylalanine arrangement and its pH-responsive co-assembly behavior with DNA.

Conclusions

These findings highlight the importance of histidine positioning within peptide sequences for tuning pH-responsiveness and optimizing DNA co-assembly and transfection efficiency. The results provide valuable insights for the rational design of efficient, safe, and pH-responsive peptide-based gene delivery systems.

Background

The inherently low transfection efficiency of naked DNA and RNA necessitates employing carrier systems in gene therapy, wherein viral vectors currently predominate owing to their evolutionary advantages as potent gene transfer agents. Despite their advantageous properties, viral vectors face significant limitations, including limited nucleic acid packing capacities, immunogenicity, risk of recombination, and random insertion into host cell genomes [1]. In pursuit of safe non-viral alternatives capable of attaining similar efficiencies, researchers have tested various biomaterials, including polymers, lipids, proteins, and peptides. The successful clinical application of lipid nanoparticles (LNPs) to deliver mRNA vaccine during the SARS-CoV-2 pandemic demonstrated the merits of this approach [2]. However, even LNPs have limitations, such as immunologic reactions to their surface poly(ethylene glycol), low stability at room temperature, and untargeted nature [3], in addition to lower transfection efficiency compared to viral vectors as a common disadvantage of non-viral gene carriers, underscoring the fact that the quest for safe and efficient non-viral vectors is yet to come to an end.

The high efficiency of viral particles in transducing eukaryotic cells can be attributed, among other factors, to their narrow size distribution and defined shape, which result from fine-tuned co-assembly of viral structural proteins and genomes [4]. Since the immunogenicity of unmodified viral structural proteins limits their application as gene delivery systems in the form of virus-like particles [5], researchers are focused on mimicking their co-assembling behavior with nucleic acids using synthetic materials, wherein peptides present themselves as valuable alternatives because they can harness the same self-assembling mechanisms [6]. Their lack of a tertiary structure enhances thermal and pH stability compared to proteins, while their well-established synthesis methods allow easy incorporation of targeting moieties without additional production steps [7]. The huge number of possible combinations in a peptide sequence, which makes testing every possible peptide sequence an impossible task, necessitates the discovery of applicable principles in the design of self-assembling peptides for their potential to be realized.

The general design principle of self-assembling peptides is inspired by block copolymers in which peptides consist of two or more amino acid blocks. Each block is incorporated into the sequence to enable specific functions such as nucleic acid binding, self-assembling, buffering, and cell-targeting/penetration. The self-assembling block of these peptides consists mainly of amino acids that can form $\beta$-sheet [8,9,10,11] or coiled-coil supramolecular structures [12, 13] via noncovalent interactions. Understanding how the sequence positioning of each amino acid influences DNA co-assembly and peptide transfection efficiency can advance the discovery of design principles that go beyond the incorporation of amino acid blocks, enabling more effective and safer self-assembling peptide gene delivery systems.

In the current study, we designed two alternating polarity peptides, RFH (RFRHRHRFR) and RHF (RHRFRFRHR), with mirroring sequence symmetry with respect to the central arginine residue. This unique design enables phenylalanine and histidine residues from adjacent peptide molecules to face each other in both parallel and anti-parallel arrangements, facilitating self-assembly into ordered nanostructures via aromatic ring interactions, which enhances complex stability. Arginine residues were strategically placed at the center and symmetrically at both termini to exploit their strong nucleic acid binding affinity and promote DNA co-assembly. Histidine was incorporated to impart pH-responsiveness, as its protonation at acidic pH (pKa = 6) modulates assembly and facilitates DNA release, inspired by prior studies on histidine-containing peptides [14]. RFH and RHF, differing only in histidine and phenylalanine positioning, enabled us to investigate how histidine placement affects the peptide’s physicochemical and biological properties. At neutral pH, both peptides formed DNA co-assemblies that halted electrophoretic movement at N/P ratios comparable to R₉, though with distinct morphologies. Only RHF retained DNA-binding capability at acidic pH. Coarse-grained MD simulations showed that RFH formed more numerous, smaller clusters when histidine was ionized, reflecting a reduced self-assembling tendency. RFH also exhibited lower peptide density around DNA and a sharper reduction in DNA contacts upon histidine ionization. RFH also showed higher transfection efficiencies compared to RHF at lower N/P ratios. These results indicate that positioning histidine near the peptide’s center enhances the pH-responsiveness of self-assembly and DNA co-assembly.

Methods

Materials

Three peptides used in this research (Table 1), were synthesized by GenScript Company (each with a purity of $>$ 90% and were utilized in experiments without any further processing. A modified pGL4.17 vector (Promega, USA) containing a luciferase reporter gene and inserted Cytomegalovirus (CMV) promoter was utilized for the physicochemical and biological experiments (Supplementary Materials: Figure S1). The pDNA was propagated through Escherichia coli TOP10 transformation under antibiotic selection and purified using a plasmid extraction kit (ExpretTM Plasmid SV, GeneAll Biotechnology Company, South Korea). Branched polyethyleneimine (MW: 25 kDa, #408727) and poly L-lysine (MW: 30–70 kDa, #P2636) were procured from Sigma Chemical Co. (USA). Dulbecco’s Modified Eagle’s Medium (DMEM, #LM-D1107/500), and penicillin/streptomycin solution (#LM-A4118/20) were obtained from Biosera (France). Fetal Bovine Serum (#26-140-079) was obtained from Gibco (USA). Bicinchoninic acid (BCA) protein assay kit (#23235) and 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) powder (#L11939.03) were acquired from and Thermofisher Scientific (USA). The Luciferase assay kit (#E1500) was purchased from Promega (U.S.). Other solvents and chemicals were obtained from Merck (Germany).

Table 1 Characteristics of the synthesized peptides

Peptide-pDNA co-assembly

The co-assemblies were formed by mixing DNA and peptide solutions in neutral (pH $\approx$ 7) or acidic (pH $\approx$ 4) environments in different amine-to-phosphate (N/P) ratios. The N/P ratio was calculated based on the number of positively charged arginine (R) residues at pH = 7 (where histidine is predominantly neutral) relative to the phosphate groups in DNA. Histidine’s positive charge in acidic environment was excluded from N/P calculation to facilitate comparison. (Table 1). Peptide stock solutions were obtained by dissolving their lyophilized powders in pharmaceutical water for injection, adjusting the pH to $\approx$ 4 or $\approx$ 7, and diluting to final concentrations of 2.42 mM and 1.35 mM for the designed and R₉ peptides respectively. Peptide-pDNA co-assemblies were prepared by adding 15 $\mu$L of peptide solutions (diluted from the stock solution for N/P ratios in the range of 5–60) to 10 $\mu$L of pDNA solution (0.1 $\mu$g/$\mu$L) followed by repeated pipetting for 30 seconds and incubation at 4 ^oC for 30 minutes as described previously [15]. Co-assemblies of pDNA and equilibrated solution of branched polyethyleneimine with N/P of 10 [16], prepared with a similar method, were used as a control in the biological characterization of the peptide-pDNA co-assemblies.

Physicochemical characterization

The co-assembling tendency of peptides with pDNA in N/P ratios ranging from 5 to 40 in neutral and acidic pHs was evaluated by gel electrophoresis. The formed co-assemblies were electrophoresed in a 1% (W/V) agarose gel containing GelRed (Biotium, USA) and with Tris-Borate (TBE) running buffer at 90 V for 40 min, and gel pictures were obtained with Gel Doc XR System (Bio-Rad, Germany). To investigate the size and morphology of the co-assemblies, 5 $\mu$L of diluted (1 to 100) co-assembled particle mixtures in three N/P ratios (15, 30, and 45) were deposited onto freshly cleaved mica substrates and left to rest for 30 minutes near an open-flame burner to prevent the deposition of dust particles and dried overnight in a closed box. The prepared slides were coated with gold and their micrographs were obtained with Sigma VP FESEM (ZEISS, Germany) at a resolution of 1.3 nm at 1 kV.

Molecular characterization

Coarse-grained molecular dynamics simulations were conducted to understand the molecular mechanisms behind self- and DNA co-assembly of the designed peptides. Molecular models of peptide sequences were generated using the PEP-FOLD server [17, 18]. The best model for each peptide was converted to coarse-grained representation utilizing the provided Python script for MARTINI force field version 2.2 [19]. The secondary structure for both peptides in both histidine ionization states was set as CCCCCCCCC (C = Coil). Histidine residues were both set as neutral (denoted as RHF and RFH) or positively charged (denoted as RHFP and RFHP), and C and N terminals were kept charged. A single peptide molecule was put in a dodecahedron box with at least a 2 nm distance from the box edges. Simulations were carried out by GROMACS simulation package version 2021.3 [20, 21]. Peptides were energy minimized in the vacuum, solvated with MARTINI polarizable water, neutralized with Cl^- ions, and equilibrated. Then, the production run was carried out for 100 ns with an integration step of 20 fs, standard settings for non-bonded interactions in the NPT ensemble, and periodic boundary conditions. The temperature was kept at 300 K with the V-rescale method with a time constant of 1 ps. The pressure was kept at 1 bar with the Parrinello-Rahman method with a time constant of 12 ps. After the simulation of the single-peptide systems, multi-peptide systems were generated for each peptide (RFH, RHF, RFHP, and RHFP) by randomly distributing 150 copies of the peptide structures from the last frame of the simulation in a 15 nm cubic box with or without a 30 bp model of DNA with a charge of  −60 in its canonical B form that was modeled using a stiff elastic network in the MARTINI force field (total of 8 systems each with 3 independent replicates). The generated systems were minimized, solvated, neutralized, and equilibrated similar to the single-peptide systems and finally, production runs were carried out for 500 ns. Trajectory analysis was performed using the GROMACS commands. Root Mean Squared Deviation (RMSD), Root Mean Squared Fluctuation (RMSF), Solvent Accessible Surface Area (SASA), and the number of clusters were calculated for all the simulated systems. Moreover, the Radial Distribution Function (RDF) of different peptide residues around the DNA surface and phenylalanine center of mass, and the number of contacts between the peptide residues and the DNA backbone or side chain at the end of the simulation were obtained for the systems with the DNA fragment. The obtained data was extracted from the generated xvg files and written to pandas [22] dataframes, processed by NumPy [23, 24] package, and visualized by Matplotlib [25] package in python 3.9 [26].

Biological characterization

The in vitro gene delivery performance of the co-assembled particles was investigated by measuring luciferase activity after transfection. The 293T and HeLa cells were cultured in DMEM supplemented with 10% FBS, and 1% penicillin/streptomycin and were incubated at 37 ^oC and 5% CO₂. Cells were used in transfection assays prior to five passages. The day before transfection, 15000–18000 cells were seeded in each well of a 96-well plate, which was treated with 10% poly-L-lysine (PLL) solution 2 hours before seeding to promote cell adhesion, and incubated with 100 $\mu$L of DMEM containing 10% FBS for approximately 18 hours. The day after, the medium was removed and the cells were gently washed once with PBS and 75 $\mu$L serum-free medium (without antibiotics) plus 25 $\mu$L of the peptide-pDNA co-assembled particle mixture (equal to 1$\mu$g DNA) was added to each well and incubated for 4–5 hours in 37 ^oC and 5% CO₂. After the incubation time, 100 $\mu$L of DMEM supplemented with 20% FBS was added to reach a medium containing 10% FBS in each well and then incubated again for approximately 48 hours. Then the medium was removed, cells were washed with PBS, and reporter gene expression was measured using a luciferase assay kit as instructed by the kit manual. Light emission was measured by a Cytation 3 cell imaging multi-mode reader (BioTek, U.S.), normalized by total cellular protein, which was measured with a Bicinchoninic acid (BCA) protein assay kit, and presented in Relative Light Units (RLUs) per mg cellular protein.

The cytotoxicity of peptide-pDNA co-assembled particles and PEI-pDNA complexes on HeLa and 293T cells was measured by MTT assay. Briefly, cells were treated with the mixtures as explained for the transfection assay. After 48 hours, the medium was replaced with 100 $\mu$L of fresh medium. Then, 10 $\mu$L of MTT solution (5 mg MTT powder in 1 mL of PBS buffer) was added to each well and further incubated for 4 hours. Thereafter, 75 $\mu$L of the medium was removed and 75 $\mu$L DMSO was added. The plate was then covered with aluminum foil and mixed at 100 rpm for 15 minutes. Finally, the absorbance was measured at 570 nm using an Epoch microplate reader (BioTek, U.S.). The relative cell viability was calculated as follows:

$$Cell\;viability\;(\% ) = \frac{{O{D_{570}}\;(samples)}}{{O{D_{570}}\;(control)}} \times 100$$

Data analysis

Data analysis and plotting for biological characterization were carried out using GraphPad Prism version 7 for Windows (GraphPad Software, San Diego, California USA, http://www.graphpad.com). All quantitative data are expressed as means $\pm$ SD of three independent replicates. Two-way ANOVA followed by multiple comparisons was performed to analyze the effect of peptide type and N/P ratio on transfection and cytotoxicity in HeLa and 293 T cell lines.

Results

Two alternating polarity peptides were rationally designed to be similar in length to R9 and include two phenylalanine and two histidine residues in symmetric positions in the peptide sequence. The symmetric positions were chosen to facilitate aromatic ring interaction between assembling peptide molecules in both parallel and anti-parallel arrangements. The two peptides differ only in the positions of their phenylalanine and histidine residues. In RFH, the two phenylalanine residues are in number 2 and number 8 positions and the two histidine residues are in number 4 and number 6 positions. For RHF, the place of histidine and phenylalanine residues was switched (Table 1). After the formation of the co-assemblies between the designed peptides and pDNA, DNA entrapment (in two different pHs) and morphology of the co-assembled particles were evaluated by gel retardation and Field Emission Scanning Electron Microscopy (FESEM). Self-assembly and co-assembly with DNA for the two designed peptides were studied by coarse-grained MD simulations in different histidine ionization states. The toxicity and transfection efficiency of these co-assemblies in two human cell lines (293 T and HeLa) were also evaluated. In physicochemical and biological evaluations, complexes of pDNA formed with R9 peptide and branched polyethyleneimine (PEI) were also included as controls.

RFH showed a reduced tendency to co-assemble with pDNA in acidic pH

The tendency of the designed peptides to form co-assemblies with pDNA in acidic (pH $\approx$ 4) and neutral (pH $\approx$ 7) conditions was compared to R₉ in gel retardation experiments, where naked pDNA was used as a control (Fig. 1). Both peptides were able to stop the electrophoretic movement of pDNA at neutral pH at the N/P ratio of 10 similar to R9. However, as there are 4 fewer arginine residues on each designed peptide molecule, it can be concluded that a higher number of peptide molecules were necessary to stop pDNA movement compared to R9. Unlike RHF, RFH lost its ability to arrest pDNA movement in acidic pH, where their histidine residues were positively charged, which indicates a reduced tendency for RFH to co-assemble with pDNA compared to RHF in acidic environments.

Fig. 1

RFH showed a reduced tendency to co-assemble with pDNA in acidic pH. Electrophoretic movement of pDNA was studied after particle formation with R9, RHF, and RFH at N/P ratios ranging from 5 to 40 in acidic (pH $\approx$ 4) and neutral (pH $\approx$ 7) environments. All three peptides were able to retard DNA movement at N/P ratios equal to or higher than 10 in a neutral environment. Unlike RHF, RFH did not retain its ability to stop pDNA movement in acidic pH. Raw images obtained directly from the gel imaging instrument are presented here. Moderately and highly brightened gel images from the same experiment are presented in the Supplementary Material (Figure S2)

RFH and RHF formed morphologically different co-assemblies with pDNA

Morphology of the stable co-assemblies (formed in pH $\approx$ 7 and at N/P ratios greater than 10) with pDNA was studied with FESEM (Fig. 2). Particle size distribution and zeta potential graphs obtained from Dynamic Light scattering (DLS) measurements were inconclusive (results not shown), probably because of the high aspect ratio of the assembled particles. R₉ was able to form spherical particles with diameters lower than 100 nm with pDNA in all three studied N/P ratios. The number of these spherical particles increased with an increase in the N/P ratio. At lower N/P ratios (e.g., N/P 10), RHF formed small, indistinct aggregates with pDNA, typically less than 200 nm in diameter, lacking a defined shape. As the N/P ratio increased to 30, RHF transitioned to forming distinct fusiform (spindle-shaped) and X-shaped co-assemblies with pDNA, measuring approximately 600 nm in length and 200 nm in width. The X-shaped co-assemblies with greater sizes were the dominant morphology at the N/P ratio of 45. It seems that these morphologies were aggregates of needle-shaped peptide assemblies with a length of more than 1 $\mu$m and a width of less than 100 nm. In contrast, RFH consistently formed fusiform co-assemblies with pDNA across all tested N/P ratios. However, the size of these fusiform structures decreased with increasing N/P ratio, from several micrometers in length at N/P 15 to less than 1 $\mu$m at N/P 45. Needle-shaped peptide assemblies with a length of several $\mu$m and a width of less than 100 nm were also observable for RFH in higher N/P ratios. Therefore, it can be concluded that both peptides primarily tend to form needle-shaped morphologies during self-assembly, i.e. in higher N/P ratios when fewer pDNA molecules are available for co-assembly, while their co-assembly with pDNA resulted in particles with different morphologies. However, co-assemblies of both peptides with pDNA went through a morphological transformation as the N/P ratio increased, highlighting the concentration-dependent nature of their assembling process.

Fig. 2

RFH and RHF formed morphologically different co-assemblies with pDNA. FESEM micrographs obtained from peptide-pDNA co-assemblies at three N/P ratios of 15 (left), 30 (middle), and 45 (right). Spherical, X-shaped, and fusiform morphologies were dominant in particles formed by R₉, RHF, and RFH with pDNA respectively. The co-assembled particles formed by RFH and RHF went through a morphological transformation as the N/P ratio increased from 15 to 45

RFH had lower assembling tendency than RHF

Since RFH and RHF showed different pDNA co-assembling tendencies in acidic pH and also co-assembled with pDNA into morphologically different particles, we sought to better understand the molecular mechanisms behind the assembly of their molecules with and without DNA in two different histidine ionization states using coarse-grained Molecular Dynamics (MD) simulations. The simulated peptide systems with uncharged histidine residues are labeled RFH and RHF, while the systems with charged histidine residues are labeled RFHP and RHFP (Snapshots of the simulations with the presence of the DNA fragment are presented in Supplementary Material Figure S3). RMSD was plotted during simulations to assess their convergence (Supplementary Material: Figure S4). RMSF was calculated for each residue position to study changes in residue fluctuation in the context of peptide-peptide or peptide-DNA assembly (Fig. 3). Changes in SASA (Fig. 4: a and b and Supplementary Material: Figure S5) and the number of clusters (Fig. 4: c and d) were monitored during the simulation as a decrease in SASA and the number of clusters is an indication of increased aggregation. RDF of peptides (Fig. 4: e) and their residues (Fig. 4: e and Supplementary Material: Figure S6) around DNA fragment surface was used to determine their affinity to bind to the DNA fragment. Additionally, this affinity was also investigated through the measurement of the number of contacts between different peptide residues and DNA (Fig. 5). On the other hand, the affinity of the peptide molecules to self-assemble was assessed by measuring the RDF of other peptide residues around the center of mass of phenylalanines (Supplementary Material: Figure S7) as it is hypothesized that aromatic ring interaction played a major role in the self-assembly of RFH and RHF.

Fig. 3

The number of outliers in RMSF distribution increased with an increased self- and co-assembling tendency. The distribution of RMSF values for each peptide residue in each sequence position was illustrated in a box-plot. The sample size for each box-plot is 450 (3 replicates for each simulated system with 150 peptide molecules.)

Fig. 4

(a, b): The peptides had a higher solvent-accessible surface area in the presence of the DNA fragment and when their histidine residue was ionized. SASA for all peptide molecules in the systems without a DNA fragment (a) and with a DNA fragment (b) were obtained using the “gmx sasa” command of the GROMACS package. (c, d): Peptides form a higher number of clusters when their histidines are protonated both with and without a DNA fragment. The number of clusters formed in systems without (c) or with (d) a DNA fragment was determined by the “gmx clustsize” command of the GROMACS package. (e, f): The density of RHF around the DNA surface is higher than RFH. This difference is more distinct for their histidine residues. The Radial Distribution Function (RDF) of whole peptides (e) and their histidine (f) residues around the DNA surface were calculated using the “gmx rdf” command of the GROMACS package. Each point in the graph is the mean $\pm$ SD calculated from three independent simulation runs

Fig. 5

The number of contacts between Different arginine residues and the DNA backbone was higher for RHF in both ionization states of its histidine residues. The number of contacts between terminal (R1/9), second (R3/7), and the middle (R5) arginine and also histidine (HIS) and phenylalanine (PHE) residues for RHF (a), RHFP (b), RFH (c), and RFHP (f) was determined from the backbone, third side chain bead (side chain 3) and fourth side chain bead (Sidechain 4) of the DNA fragment by the “gmx mindist” command from GROMACS package. Each number is the mean number of contacts calculated from three independent simulation runs

When their histidines were neutral, whether in the presence or absence of the DNA fragment, RFH and RHF did not show any distinct differences in their RMSF (Fig. 3), SASA (Fig. 4: a and b), and number of clusters (Fig. 4: c and d) profiles. However, in the same histidine ionization state, RFH residues showed a lower density around phenylalanine COMs, regardless of the DNA presence, compared to RHF (Supplementary Material: Figure S7). Moreover, RFH showed a lower binding tendency to the DNA fragment when its histidine was not ionized, which is reflected in the RDF of the peptides (Fig. 4: e) and their phenylalanine (Supplementary Material Figure S6:d) and terminal arginine (Supplementary Material: Figure S6:a) residues around DNA surface. This lower tendency was also supported by the sum of contacts between arginine residues (R1/9, R3/7, and R5) and DNA backbone (RFH: 113.7, RHF: 119.5, refer to Fig. 5: a and c). These observations collectively suggest that RFH had a lower tendency for self-assembly and DNA co-assembly compared to RHF when their histidine residues were neutralized.

DNA had a disruptive effect on peptide assembling tendency

Only a fraction of peptide molecules bound to the DNA fragment as is observable in the snapshots from the simulations, where the majority of the peptide molecules were spread in the simulation box and not concentrated around the DNA fragment (Supplementary Material: Figure S3). Presence of the DNA fragment led to the increase of outliers with a low fluctuation in the RMSF graphs (Fig. 3) in both histidine ionization states, probably due to the binding of the peptide molecules to the surface of the DNA fragment. The presence of the DNA fragment also increased peptide SASA (Fig. 4: a and b) and the number of supramolecular clusters formed (Figure 4: c and d) in plateau regardless of histidine ionization state. Notably, the presence of the DNA fragment reduced the densities of arginines, histidines, and phenylalanines around COM of phenylalanine residues, irrespective of the histidine ionization state (Supplementary Material: Figure S7). Lower density of these residues, combined with increased SASA and a higher number of clusters, suggests a weaker assembling tendency, it can be concluded that the presence of the DNA fragment diminished the self-assembly propensity of both peptides across all histidine ionization states.

RFH assembling tendency was more profoundly affected by the histidine ionization state

The number of outliers with lower fluctuation in RMSF plots (Fig. 3) reduced for both peptides when their histidine residues were ionized. This reduction was greater in the systems with a DNA fragment. Also, RFH showed a sharper drop in low-fluctuation outliers than RHF as a result of histidine ionization, indicating a greater reduction in DNA-bound peptides for RFH compared to RHF. The histidine ionization state did not affect SASA for the DNA fragment (Supplementary Material: Figure S5), suggesting that regardless of the fraction of peptide molecules that bind to DNA in all simulated systems, an equal amount of surface becomes inaccessible to solvent molecules. On the other hand, both SASA for the peptide molecules (Fig. 4: a and b) and the number of peptide supramolecular clusters (Fig. 4: c and d) in plateau increased by the ionization of histidine, regardless of DNA presence. Also, SASA (Fig. 4: a) and the number of supramolecular clusters in the absence of the DNA fragment (Fig. 4: c) increased to a higher extent for RFH compared to RHF. The density of histidine around phenylalanine COMs (Supplementary Material: Figure S7) reduced upon its ionization regardless of the DNA presence for both peptides. This reduction was greater for RFH compared to RHF. However, the density of phenylalanine increased from a maximum at 0.5 nm of $\approx$ 9 to $\approx$ 12 after histidine ionization for RFH in the absence of the DNA fragment, while for RHF this quantity showed a reduction from $\approx$ 19 to $\approx$ 18 parallel to the histidine density. Despite the overall reduction in assembly, this suggests a distinct structural response in RFH with enhanced local Phe-Phe interactions in the expense of His-Phe interactions. Furthermore, the radial distribution function of the peptides and their histidine residues around the surface of the DNA fragment (Fig. 4: e and f) showed a reduction in their density when histidine was ionized. Also, the number of contacts between the arginine residues of the peptide (R1/9, R3/7, and R5) and DNA backbone (Fig. 5) reduced with histidine ionization. The number of contacts reduced to a lower quantity for RFH (91.7) compared to RHF (96.9) as a consequence of histidine ionization (Fig. 5: b and d). This observation is in agreement with the result of gel retardation in acidic pH (Fig. 1). These results collectively suggest a reduced assembling tendency for both peptides when their histidine residues were ionized, which can be attributed to the increased repulsive force between the same charges of imidazole rings that limits the aromatic ring interactions driving the self-assembling process. Moreover, histidine ionization affects RFH’s self- and DNA co-assembling tendency to a higher extent compared to RHF.

RFH showed higher in vitro transfection efficiency compared to RHF

The in vitro transfection efficiency of the designed peptides was assessed on two cell lines (293T and HeLa) in comparison to R9 and polyethylenimine using luciferase assay (Fig. 6: a and c). The level of luciferase expression as measured by Relative Light Units (RLUs) per mg cell protein (log(RLU/mg protein)) was generally higher in 293T cells (Fig. 6: a) compared to HeLa cells (Fig. 6: c) for all examined N/P ratios. Luciferase expression in cells transfected with PEI was higher than those transfected with all peptides at N/P ratios of 20 and 30 in both cell lines. The difference between PEI and the designed peptides became statistically insignificant (p $>$ 0.05) at N/P ratios between 40 and 60 in 293T cells. The same phenomena were observed in HeLa cells except for RHF at an N/P ratio of 40. In 293T cells, both RFH and RHF significantly outperformed R9 at N/P ratios higher than 30. This gap grew as the N/P ratio was increased from 40 to 60. In HeLa cells, RFH, unlike RHF, also outperformed R9 at N/P ratios of 20 and 30 (p $<$ 0.01). The difference between the transfection efficiency of RFH and RHF was insignificant at all N/P ratios in both cell lines. However, the two-way ANOVA test between RFH and RHF at all N/P ratios showed a significant (p $<$ 0.05) effect for RFH versus RHF. Our observations indicate that generally, the designed peptides had transfection efficiencies lower than PEI and higher than R₉. We also observed that RFH marginally outperformed RHF. The observed differences between these peptides in co-assembly with DNA in acidic pH as reflected in gel retardation experiments and coarse-grained MD simulations might play a role in the higher transfection efficiency of RFH, which can only be attributed to the different position of its histidines and phenylalanines.

Fig. 6

(a, c): RFH showed higher in vitro transfection efficiencies compared to RHF. Luciferase assay was carried out on 293T (b) and HeLa (d) cell lines to compare the transfection efficiency of PEI in N/P ratio of 10 with peptides in N/P ratios from 20 to 60. All toxicity and transfection are reported as mean $\pm$ SD of three replicated tests. A two-way ANOVA test with multiple comparisons was utilized to analyze the results. (b, d): The peptides in all N/P ratios generally showed lower toxicity than PEI. MTT cell viability assay was performed on 293 T (a) and HeLa (c) cell lines to compare the toxicity of PEI with the peptides in N/P ratios from 20 to 60

RFH and RHF showed less in vitro toxicity than PEI

Toxicity of the peptides in N/P ratios of 20 to 60 was assessed by MTT cell proliferation assay in 293T and HeLa cell lines (Fig. 6: b and d). All three peptides were significantly less toxic than PEI in all N/P ratios. The difference between the viability of the peptide-treated and untreated cells was insignificant at N/P ratios 20, 30, and 50 on both cell lines, which indicates negligible toxicity by these peptides at these N/P ratios. Moreover, there was no statistically significant difference between the toxicity of the peptides on both cell lines. Therefore, one can conclude that the designed peptides had a similar in vitro toxicity to R₉ with a higher transfection efficiency.

Discussion

Two novel rationally designed alternating polarity self-assembling peptides RFH (RFRHRHRFR) and RHF (RHRFRFRHR) with mirroring symmetry were assessed for their ability to co-assemble with and deliver DNA to eukaryotic cells along with R9 as a frequently reported peptide in transfection studies. The designed peptides were able to form morphologically distinct stable co-assemblies with pDNA at pH $\approx$ 7. This ability was only retained by RHF at pH $\approx$ 4. The analysis of the trajectories driven from coarse-grained MD simulations revealed that histidine ionization, which is a function of pH, had a greater effect on the assembling tendency of RFH with and without DNA compared to RHF. Both peptides could transfect 293T and HeLa cells with higher efficiencies and similar toxicities compared to R₉. RFH showed a better performance in gene delivery compared to RHF, which can only be attributed to the difference in histidine and phenylalanine position.

Each of the consisting amino acids contributed to the unique properties of the designed peptides. Our designed peptides are based on the nona-arginine peptide. Arginine is known to have the highest affinity among other coded amino acids to nucleotides [27, 28]. This amino acid also has demonstrated high membrane penetration activity [29]. Here, arginine residues showed the highest number of contacts with the DNA backbone as demonstrated in the coarse-grained MD simulations. Peptides containing aromatic hydrophobic amino acids are prone to the formation of nano-fibers in aqueous solutions [30]. Phenylalanine and histidine residues in RFH and RHF had probably facilitated the formation of peptide-DNA co-assemblies with a defined shape and a narrow size distribution by forming elongated supramolecular structures via aromatic ring interactions. Histidine-containing peptides have been applied for gene delivery to enhance endosomal escape through the proton sponge mechanism by Wiradharma et. al [31]. They reported the dependence of transfection efficiency on the pH at which peptide and DNA form complexes. Moreover, alternating histidine/aromatic residue sequences in amphiphilic peptides have shown pH-dependent self-assembling behavior [32]. Here, both gel retardation studies and coarse-grained MD simulations demonstrated that the self-assembling and DNA co-assembling tendency of RFH and RHF depends on the histidine ionization state, which in turn is pH dependent.

The symmetric design of RFH and RHF might facilitate aromatic ring interaction between phenylalanine and histidine residues from two peptide molecules, which was demonstrated by high density of these residues around phenylalanine COMs (Supplementary Material: Figure S7). As histidines and phenylalanines are not positioned next to each other in peptide sequences, this high density is probably because of the parallel or antiparallel arrangement of assembling peptide molecules. Peptides with alternating arginine/phenylalanine residues have been previously reported to form amyloid fibers by forming $\beta$-sheets that are positioned perpendicular to the long axis of the fiber [30], which became more ordered from RF to [RF]₅. The sequence of our designed peptides is similar to [RF]₄ except for the inclusion of histidine residues in symmetric positions (4 and 6 for RFH or 2 and 8 for RHF). Therefore, it is plausible that RFH and RHF may adopt the same arrangements. Mello et al. proposed a model for the co-assembly of KIW7 (KIWFQNR) peptide with a linear DNA fragment [11]. In their model, the DNA acts as a template to bring individual peptide molecules close to each other and facilitate nano-fiber formation via $\pi$-stacking. In this model, only the terminal positively charged residues of the peptide molecules directly bind to the DNA molecule. Our designed peptides were also able to form nano-fibers, probably via aromatic ring interactions, as demonstrated in the SEM micrographs (Fig. 2). In addition, coarse-grained MD simulations showed that the majority of the contacts were between terminal arginine residues and the DNA backbone (Fig. 5), which is in agreement with the proposed model by Mello et al., further enhancing the possibility of $\beta$-sheet formation by RFH and RHF. Moreover, histidine sequence position had a profound effect on the assembling behavior of the designed peptides, where central positioning of histidine residues resulted in a lower density of aromatic rings around phenylalanine COMs and a greater decrease in self-assembling and DNA co-assembling tendency of RFH when its histidines were ionized, leading to a lower tendency of RFH to stop the electrophoretic movement of pDNA in an acidic environment. More studies are needed to better understand structural differences between RFH and RHF assembled structures and the mechanism by which positioning histidine residues near the middle of the peptides affects their assembling behavior.

Here, the designed histidine-containing peptides unlike RH9 reported by Alhakamy et al. were able to effectively bind to and transfer DNA without the addition of 100 mM CaCl2 [33]. They were able to stop pDNA electrophoretic movement at N/P ratios similar to R₉ and demonstrated higher transfection efficiencies compared to this peptide at each tested N/P ratio. However, as there are fewer arginine residues in each designed peptide molecule compared to R9, the number of the designed peptide molecules per pDNA was higher than R9. Moreover, it is plausible that a higher number of peptide molecules are needed for the formation of peptide-DNA co-assemblies as DNA is trapped by the self-assembled peptide nano-fibers rather than individual peptide molecules.

The toxicity of [RF]_n peptides has been reported to be correlated with increased order in the assembled supramolecular structures as n increased from 1 to 5 [34]. The toxicity of our designed peptides was insignificant from the control group in N/P ratios lower than 50. The inclusion of histidine residues in symmetric positions may have contributed to the lower toxicity observed for RFH and RHF. However, its mechanism for lowering toxicity compared to [RF]n peptides needs more investigation to be fully understood. One possible mechanism might be pH-induced ionization of these residues that generate repulsive force and reduce the order of the self-assembled structures after endocytosis, likely through a reduction in the number of peptide-DNA (Fig. 5) and peptide-peptide (Fig. 4: a and c, Supplementary Material: Figure S7) interactions.

We obtained valuable insights about peptide-DNA interactions in different histidine ionization states from coarse-grained MD simulation using the MARTINI force field. However, this method has its limitations. Tuttle and co-workers used the same simulation technique to screen sequence space of di [35], tri [36], and up to hepta-peptides [37] for self-assembling peptides. Aggregation propensity (AP), which was calculated by dividing the initial SASA by the same value at the end of the simulation, was applied as screening quantity. The higher AP values were interpreted as a higher ability to self-assemble. However, they observed that in the case of oligo-phenylalanines, this value decreased from 2.9 for tri-phenylalanine to 2.1 for hepta-phenylalanine, while one expects that the tendency to form aggregates increase parallel to peptide chain length. They hypothesized that this might be due to the branched structure of these peptide series and also the formation of weak elastic bonds in peptides longer than 4 amino acids in the MARTINI force field. Pranami et. al. reported that the diffusion coefficient of an aggregation cluster in a medium at infinite dilution is dependent on the number of original particles that formed the cluster (Np) and its fractal dimension (df) [38]:

$$D \propto N_p^{\frac{{ - 1}}{{{d_f}}}}$$

According to the equation, dense clusters with higher fractal dimension (d_f = 2.5) have higher diffusion constant than the clusters with the same number of consisting particles but lower density (d_f = 1.8). Therefore it is conceivable that initial clusters formed from peptides with bulkier residues have a lower translational diffusivity and therefore have a lower probability of colliding with other clusters in the simulation box. This might lead to a false reduction in aggregation propensity. We think this lower transnational diffusivity is more severe in the case of peptide-DNA clusters. However, we could not determine whether this was the major mechanism for the reduction of peptide assembling tendency in the presence of the DNA fragment. Consequently, alternative simulation methods and also energy analysis are needed for a more robust understanding of peptide-DNA complex formation.

Conclusions

We engineered two novel peptides, RFH and RHF, featuring alternating polarity and mirroring symmetry in the arrangement of arginine, phenylalanine, and histidine residues relative to the peptide’s center. This strategic design enabled us to explore how the positioning of histidine and phenylalanine influences DNA co-assembly and gene delivery efficiency. Physicochemical analyses revealed distinct behaviors at lower pH, where ionized histidine residues destabilize RFH–pDNA co-assemblies, triggering DNA release, while RHF maintained stable DNA binding. Coarse-grained simulations corroborated these findings, showing that histidine ionization disrupted RFH’s DNA assembly more significantly than RHF’s, consistent with experimental data. Biological assessments demonstrated that both peptides outperformed R₉ in in vitro transfection efficiency, with comparable toxicity profiles, and RFH exhibited a slight edge over RHF. The centrally positioned histidine residues in RFH enhance its pH-responsiveness, facilitating efficient DNA release in acidic conditions—crucial for endosomal escape in gene delivery. Conversely, RHF’s sustained stability across a broader pH range makes it better suited for environments where premature DNA release risks degradation, such as the acidic tumor microenvironment. Beyond introducing promising delivery vectors, these results reveal a broader design principle: positioning histidine closer to the peptide center, relative to charge-stable hydrophobic residues like phenylalanine, amplifies pH-responsive DNA co-assembly and boosts transfection efficiency. This principle offers a blueprint for developing advanced peptide-based gene delivery systems with superior in vitro and in vivo performance.