Optimization of a Modular Nanotransporter Design for the Killing of Cancer Cells via Targeted Intracellular Delivery of a Photosensitizer

1. Introduction

One of the most essential factors in effective cancer treatment is its ability to act specifically on tumor cells [1]. Therefore, the development of targeted drug delivery systems represents a promising area to obtain more effective anticancer treatment modalities [2]. Polypeptide constructs as drug carriers for targeting intracellular sites are advantageous due to their high specificity and biocompatibility. However, they often face challenges such as rapid degradation and low bioavailability, mainly because of the necessity to penetrate lipid membranes[3,4,5,6]. One of these systems is modular nanotransporters (MNTs), which are artificial polypeptides composed of several transport modules [7]. MNTs are capable of increasing the efficiency of the delivery of cytotoxic agents into the nuclei of cancer target cells, which greatly enhances the effectiveness of photodynamic therapy (PDT) [8,9] and radionuclide therapy [10,11]. PDT is an emerging treatment method for skin neoplasms and a number of solid tumors, including esophagus, lung, and prostate cancers [12,13,14]. PDT employs substances that can mediated a transition to an excited state in response to light, producing reactive oxygen species (ROS) that cause biomolecule damage, activation of the immune response, and cell death [13,15,16]. ROS with a short lifetime lead to damage in the nanometer range from the place of their generation; therefore, intracellular delivery to the most damage-sensitive structures of cancer cells is necessary for maximum efficiency. Previous studies have demonstrated that delivering photosensitizers to cancer cell nuclei via MNT can increase their photodynamic activity by up to 2,500 times [17].

Each module of MNT provides a particular stage of transport when MNT enters the target cell and then relocates to a specified cell compartment. MNTs, unlike many other therapeutic polypeptides, can permeate membranes and reach the cytosol of target cells. The sequences of MNT modules derived from natural proteins are united in a single gene sequence that encodes a chimeric protein capable of delivering drugs to a specified location of the target cell. MNT modules are interchangeable, allowing for customization based on the target cell type and the desired intracellular location for drug delivery [7]. Depending on their composition, MNT can penetrate, for example, cell nuclei, lysosomes, or bind to the surface of mitochondria [7,11,17,18]. They contain a ligand module that binds to overexpressed cell surface receptors, allowing it to recognize a specified cell type. Receptor-mediated endocytosis leads to the internalization of MNT, and for subsequent release from endosomes, MNT contains an endosomolytic module capable of destroying lipid membranes in a slightly acidic environment within endosomes. MNT can also contain modules for transport to a specific cellular compartment, most commonly a nuclear localization signal (NLS) for the delivery of drugs into the nucleus [7]. The carrier module combines functionally active MNT modules and is also used to attach drugs that are transported by the transporter. Most of the MNT constructs developed to date use the hemoglobin-like protein (HMP, [19]) from Escherichia coli as the carrier module. Reducing the size of macromolecules facilitates diffusion in tissues, which contributes to a greater depth of penetration and, consequently, better delivery of the active principle [20]. HMP performs the function of a carrier and does not play a significant role in interaction with cellular structures during transport into the target cell compartments. Therefore, in order to reduce the size of MNT, it may be reasonable to consider truncating this module. HMP has two domains: the beta-folded C-terminal domain and the alpha-helical N-terminal globin domain with a pocket for porphyrin [20]. Keeping one of these HMP domains as a carrier module allows us to create a smaller construct. Because a crucial function of the carrier module is to ensure the availability of the MNT’s endosomolytic and ligand modules for interaction with cellular structures, it requires truncation without compromising their activity.

Another rationale for the development of constructs with a truncated carrier module is the recently discovered interaction of HMP, which is part of one of the MNTs, with Kelch-like ECH-associated protein 1 (Keap1) [18]. Keap1 is widely represented in the cytoplasm and is involved in the control of many cellular activities, primarily the regulation of the level of transcription factor NF-E2-related factor 2 (Nrf2) [21,22,23]. In addition, Keap1 interacts with the PGAM5 protein, which is localized on the outer membrane of mitochondria and is also involved in maintaining mitochondrial homeostasis [24,25]. Interaction with Keap1 can lead to the retention of MNT in the cytoplasm and the activation of cell signaling pathways, resulting in defense from the oxidative process. On the other hand, through the interaction of MNT with Keap1, it is possible to target cytotoxic agents to mitochondria. As demonstrated in our recent work, attaching a photosensitizer to MNT with the anti-Keap1 monobody sequence increases photocytotoxicity compared to a photosensitizer attached to a cytosol-targeted MNT [18]. Thus, the development and study of new MNT will shed light on the possibilities of creating functional MNT structures in smaller sizes. Furthermore, this allows us to examine the effect of an additional mitochondrial localization site on the photocytotoxicity of photosensitizers delivered to the nucleus of a cancer target cell.

2. Materials and Methods

2.1. Cell Lines

The human epidermoid carcinoma A431 cell line was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and maintained according to the ATCC specification.

2.2. Producing MNTs with a Truncated Carrier Module

To create MNT with a truncated carrier module, the previously described construct (DTox-HMP-NLS-EGF) was used. The MNT consists of the sequence of human epidermal growth factor (EGF) as a ligand module. The diphtheria toxin translocation domain (DTox) sequence was employed as an endosomolytic module. The sequence of the hemoglobin-like HMP protein was taken as a carrier module. The optimized nuclear localization signal (NLS) of the large T-antigen of the SV40 virus was used as a module for delivery into the cell nucleus. The gene encoding this MNT was deleted in the sections corresponding to two parts (C- or N-terminal) of the HMP sequence. Mutagenesis was performed using oligonucleotide primers by PCR amplification of the original gene without the corresponding fragments. The primers were synthesized by Evrogen, Moscow, Russia, and mutagenesis was performed using the QuickChange™ kit (Agilent Technologies, Santa Clara, CA, USA). Competent cells (XL1-Gold) were transformed with the obtained plasmids and grown in a SOC (Super Optimal Broth and Catabolic Repressor) medium overnight at 37 °C in the presence of ampicillin. Colonies with mutations were selected by PCR screening according to a standard protocol [26]. Furthermore, restriction analysis was used to determine whether a deletion happened in the areas of interest [27]. The obtained plasmids were sequenced using the Sanger technique [28].

2.3. MNT Expression and Purification

The production of DTox-HMP_C-NLS-EGF (MNT_C) with the C-terminal fragment of HMP, DTox-HMP_N-NLS-EGF (MNT_N) with the N-terminal fragment of HMP, and the original MNT (DTox-HMP-NLS-EGF, hereinafter referred to as the MNT_F) was carried out in the strain E. coli BCL-12. The cells were grown on LB Broth Miller medium (Amresco, Solon, OH, USA). The induction of MNT expression was performed with 200 µM of isopropyl-β-D-galactopyranoside (SybEnzyme, Russia) for 2.5 hours at 18 °C. All MNTs were isolated from the soluble fraction and purified by affinity chromatography on a HisTrap™ High Performance carrier (Cytiva, USA). Purified MNTs were refolded in 150 mM NaCl, 25 mM Na₂HPO₄, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, 2 mM oxidized glutathione, and 0.67 mM reduced glutathione. After refolding, the MNTs were re-purified by affinity chromatography and dialyzed against PBS (10 mM Na₂HPO₄, 150 mM NaCl, pH 8). The purity of MNTs produced using the given approach was approximately 85%, according to denaturing electrophoresis data. The concentration of MNT was assessed using the Bradford method [29].

2.4. Binding Assay

The effectiveness of the interaction of the ligand module with the epidermal growth factor receptor (EGFR) was evaluated on the cell line of human epidermal carcinoma A431, characterized by overexpression of EGFR. To do this, we determined the dissociation constant of the ligand-receptor complex in competition with a fluorescently labeled MNT₁ [18]. MNT₁ was labeled with Alexa-647 as described earlier [30]. Briefly, A freshly prepared solution of Alexa Fluor 647 succinimidyl ester (Molecular Probes, Eugene, OR, USA) (2 mg/mL) was added to the MNT₁ solution in carbonate buffer, pH 8.6 in molar excess of 5:1. Labeled MNT₁ was purified by ultrafiltration. Binding analysis was performed on A431 cells, which were dispersed onto 48-well plates and fixed with a 0.5% paraformaldehyde solution for 5 minutes on ice. Incubation with MNTs was performed in DMEM (Dulbecco’s Modified Eagle Medium) medium containing 1% bovine serum albumin (BSA), 100 μg/ml gentamicin, 0.4 g/l NaHCO₃, 15 mM HEPES, pH 7.3, overnight at +4 °C. For the study of competitive binding, 25 nM of labeled MNT1 and different concentrations of MNTC or MNTN were used. After incubation, the cells were washed five times with Hanks solution, and 100 µl of 0.25% trypsin-EDTA solution per well was added for 2 hours. The well contents were removed to a black-well plate, and the wells were rinsed with 150 µl of a 1% Triton X-100 solution. Sample fluorescence was measured in the 630–680 nm range using a ClarioSTAR Plus plate reader (BMG LABTECH, Germany) with an excitation wavelength of 625–630 nm.

2.4. Liposome Leakage Assay

The ability of MNT1 to provide liposome leakage was demonstrated on small unilamellar phosphatidylcholine (Sigma-Aldrich, Burlington, MA, USA) liposomes prepared by reverse-phase evaporation [31] loaded with the fluorescent dye calcein (Fluka, Seelze, Germany) to a concentration of 100 mM, which led to fluorescence self-quenching. Liposome leakage analysis was performed as described previously [18].

2.5. Thermophoresis

Binding between Keap1 and MNTs was assessed using a Monolith NT.115 instrument (NanoTemper Technologies, Munchen, Germany) in phosphate buffer containing sodium phosphate (25 mM) and sodium chloride (150 mM), pH 8.0, using Cy3-labeled Keap1, as described in our previous article [18].

The interaction of the MNTs with the α/β importin complex was evaluated in buffer containing 20 mM HEPES, 110 mM KCl, 5 mM NaHCO3, 5 mM MgCl₂, 0.1 mM CaCl₂, 1 mM EGTA, and 1 mM DTT, pH 7.4, using Cy3-labeled β-importin as described earlier [32].

2.7. Photocytotoxicity

Chlorin e₆ (Frontier Scientific, Philadelphia, PA, USA) was coupled to MNT using 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide and N-hydroxysuccinimide (Sigma-Aldrich) as described previously [18]. Briefly, a 3 mM aqueous solution of sodium chlorin e₆ was diluted in buffer containing 10 mM MES (Sigma-Aldrich), pH 6.5, and 1 mM sodium dodecyl sulfate (SDS-Na) to a final concentration of 150 μM. The resulting solution was mixed with freshly prepared 0.1 M aqueous solutions of 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (Sigma-Aldrich) and N-hydroxysuccinimide (Sigma-Aldrich) up to the molar ratio clorin: carbodiimide:N-hydroxysuccinimide = 1:4.5:15. The solution of the activated chlorin e₆ was added to 10 μM of MNTs in buffer containing 30 mM sodium phosphate, 20 mM sodium borate, 150 mM sodium chloride, pH 8.0, and SDS-Na in the molar ratio MNT:SDS-Na = 1:5. The reaction was stopped the next day with hydroxylamine (Reakhim, Moscow, Russia) at a final concentration of 10 mM. The conjugates were purified using affinity chromatography on Ni-NTA Sepharose.

A431 cells were seeded in 96-well plates (1000 cells per well) for the cytotoxicity experiment. The next day, the media was changed to DMEM with 10 mg/mL BSA, and the cells were treated with various doses of MNT-chlorin conjugates or free chlorin e₆. After 20 hours of incubation, the cells were washed twice with Hank’s solution and irradiated for 10 minutes (0.0276 W/cm2) with a slide projector in Hank’s solution. Hank’s solution was then replaced with fresh medium containing 10% fetal bovine serum and incubated for 4-5 days. Cell viability was assessed using 3-(4,5-dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide (MTT). The experiments were repeated six times, and each of them was performed in six replicates. The concentration causing 50% cell death (EC₅₀) was estimated using data interpolation along a four-parameter logistic curve in the GraphPad Prism 9 software package (GraphPad Software Inc., La Jolla, CA, USA). The significance of the differences was assessed using ANOVA with the Prism software.

3. Results

For the study, plasmids encoding MNT with C-terminal and N-terminal parts of the carrier module were obtained. Both truncated MNTs were successfully produced and purified, as described in the "Materials and Methods" section.

A study of the binding of fluorescently labeled MNT1 showed that it binds to EGFR on the surface of A431 cells with a dissociation constant, K_d 46±15 nM (mean±SEM) (Figure 1a). This value is consistent with the data obtained by radioligand analysis (62±10 nM) [18]. Competitive analysis yielded K_d values of 10 ± 2 nM for MNT_C and 20 ± 4 nM for MNT_N (Figure 1b and 1c). This demonstrates that the ligand module in both MNTs retains functionality and can target MNT cells with overexpressed EGFR.

Figure 2 shows the leakage of the fluorescent dye calcein from phosphatidylcholine liposomes at different pH levels. The experiments have shown that the new MNTs have membranolytic activity in the pH 5–6 region corresponding to the pH of the endosomes, which makes it possible to release MNTs from early endosomes after internalization.

Presumably, HMP participates in the membranolytic activity of MNT, and its truncation could lead to a decrease in this activity in the area of slightly acidic pH. As shown in Figure 2, MNT containing the N-terminal HMP domain exhibits somewhat less activity in this range. At pH 6, MNT_N releases approximately 45% of calcein, while MNT with full-size HMP releases nearly 60%. Despite this, it appears that the loss of one of the HMP domains has no significant impact on MNT-membrane interaction.

The thermophoresis assay makes it possible to observe a change in the diffusion rate of molecules during the formation of complexes due to the registration of fluorescence by one of the interacting substances.

Thermophoretic curves for MNT_N and MNT_C are shown in Figure 3a,b, respectively. The results of the experiments demonstrated that MNT_N with N-terminal domain of HMP interacts with Keap1 with K_d of 428 ± 193 nM. At the same time, MNT_C, has a much higher dissociation constant of the complex with Keap1 (K_d = 1800 ± 500 nM). The difference between the measured values of the dissociation constants is significant (p < 0.05, n = 12). In addition, the values of dissociation constants for MNT_C and MNT_F with full-size HMP (205 ± 22 nM) differ significantly (Figure 3). At the same time, the difference in dissociation constants for MNT_F and MNT_C is nonsignificant. Thus, it can be concluded that HMP interacts with Keap1 due to the N-terminal (globin) part.

The thermophoretic curves of the interaction of MNTs with α/β-importins are shown in Figure 4. The dissociation constants of the new constructs with the complex of imports were 160 ± 31 nM and 179 ± 22 nM for MNT_N and MNT_C, respectively. This demonstrates that modules for delivery to the nucleus of these MNTs are functional, allowing these MNTs to be delivered into the cell nucleus.

The effectiveness of the photosensitizer was assessed by the concentration of MNT, which causes the death of 50% of cells. Figure 5 shows the results of a comparison of the effects on cells of two new MNTs and the original MNT_F with the attached chlorin e₆. For MNT_C, the EC₅₀ was 6 ± 3.4 nM, and for MNT_N, it was 1.3 ± 0.5 nM. The EC₅₀ of free chlorin e₆ was about 550 ± 110 nM, which shows that truncated structures significantly enhance the photodynamic effect of chlorin e6 on A431 cells. For the original MNT_F with a full-size carrier module, the EC₅₀ was 13.7 ± 3.6 nM; that is, the new constructs destroy cells more efficiently. At the same time, the significance of differences from the effect of the original MNT_F was revealed only for MNT_N with the N-terminal part of HMP (p = 0.015).

4. Discussion

When creating or modifying complex transport structures for intracellular delivery to target cells, it is necessary to ensure that all components of these structures retain their functionality. Due to the ligand module, which is able to bind to EGFR, MNT is able to recognize cancer cells and penetrate them via receptor-mediated endocytosis. As previous work has shown, the K_d for MNT_F with a full-size carrier module is 29 nM [17]. In experiments on A431 cells carrying an increased number of these receptors, it was shown that truncated constructs retain the ability to bind to EGFR receptors on the cell surface, forming MNT-receptor complexes with dissociation constants of about 10 and 20 nM, respectively. This binding of new MNT with EGFR suggests that they are able to successfully penetrate target cells by receptor-mediated endocytosis.

The endosomal activity of the new transporters was assessed by their ability to disrupt the integrity of lipid membranes in a model system of liposomes loaded with calcein. The membranolytic properties of MNT reflect the release of calcein from liposomes during incubation with MNT at different pHs. The main role in the ability of MNT to destroy membranes is played by the translocation T domain of the diphtheria toxin, which, with a decrease in pH, is able to change its conformation and integrate into the lipid bilayer [33,34,35]. Previous research has also demonstrated HMP’s potential to attach to lipid membranes and its membranolytic activity in the pH range of 3–4. The occurrence of two peaks on the pH curve of the dependency of MNT’s endosomolytic activity may be attributed to the joint action of two modules: one emerges as a result of HMP action, while the other is DTox [17]. According to atomic force microscopy studies of lipid membranes incubated with MNT, the combined action of HMP and DTox most likely results in the formation of ring-shaped structures on the surface of lipid membranes that contain MNT embedded in the membrane and ensure MNT release from endosomes [17]. MNT_N containing only the N-terminal part of HMP exhibits decreased membranolytic action. In this regard, it can be assumed that the C-terminal fragment of HMP also plays an important role in this module’s endosomolytic activity, providing endosomolytic activity in the pH 3–4 region while also causing MNT to be incorporated into the membrane and pores to form in the pH 5–6 region. Membranolytic activity of the globin N-terminal domain of HMP is more expected [36]. Nevertheless, the effect of MNT_N on lipid membranes is sufficient to lead to their destruction at a slightly acidic pH. Thus, when the contents of the endosomes are acidified, both MNTs are able to disturb the lipid bilayer.

The nucleus is an important target for many drugs as it contains DNA, damage to which, as a rule, leads to cell death [37,38,39,40]. The ability to bind to importins that allow proteins with NLS to penetrate into the nucleus through the nuclear pore, as shown by thermophoresis, indicates the possibility of successful penetration of MNT with N- and C-terminal parts of the carrier module into the nucleus.

In this article, we also investigated how the previously discovered interaction of the MNT carrier module with Keap1 can affect the effectiveness of the photodynamic action of a photosensitizer delivered to the cell nucleus. Kear1 is an Nrf2 inhibitor that controls the expression of most proteins involved in the metabolism of reactive oxygen species [41,42,43].

The displacement of Keap1 from the complex with Nrf2 can be used to activate the Nrf2 system of protection against oxidative stress [36], for example, with the help of antibodies to Keap1 [37]. For MNT targeted at Keap1 using the AntiKeap1 monobody antibody-like molecule, it has been shown that the interaction occurs both through the AntiKeap1 monobody and through HMP [7]. Now, by thermophoresis, it has been shown that MNT with N-NMR interacts with Keap1 in contrast to MNT with C-HMP, which suggests that the binding site of NMR to Keap1 is located in the N-terminal globin domain.

The displacement of Keap1 from the complex with Nrf2 can be used to activate the Nrf2 system of protection against oxidative stress [36], for example, with the help of peptides or antibodies against Keap1 [37]. For MNT targeted at Keap1 using the anti-Keap1monobody antibody-like molecule, it has been shown that the interaction occurs both through Anti-Keap1 monobody and through HMP [7]. Now, by thermophoresis, it has been shown that MNT with N-NMR interacts with Kear1 in contrast to MNT with C-end of HMP, which suggests that the binding site of HMP to Keap1 is located in its N-terminal globin domain.

The displacement of Keap1 from the complex with Nrf2 can be used to activate the Nrf2 system of protection against oxidative stress [44], for example, with the help of peptides or antibodies against Keap1 [45,46,47]. For MNT targeted at Keap1 using the anti-Keap1 monobody antibody-like molecule, it has been shown that the interaction occurs both through the anti-Keap1 monobody and through HMP [18]. Now, by thermophoresis, it has been shown that MNT with N-NMR interacts with Kear1 in contrast to MNT with the C-end of HMP, which suggests that the binding site of HMP to Keap1 is located in its N-terminal globin domain.

Our study shows that truncation of the HMP carrier module does not affect the functional activity of the MNT. Moreover, transporters with a truncated carrier module demonstrated a more pronounced photodynamic effect on A431 cells compared to MNT with full-size HMP. This may indicate that they carry out a more efficient delivery of the photosensitizer to vulnerable cell compartments. It is possible that the size of MNT plays a significant role in penetration into cells and cellular compartments; therefore, both MNT and truncated HMP turned out to be more cytotoxic as carriers of photosensitizers. The hemoglobin-like E. coli protein (HMP), used as a carrier module in MNT, belongs to the flavohemoglobin group. Although it is believed that the main role of HMP in bacterial cells is to neutralize NO radicals under aerobic and anaerobic conditions [48], the processes occurring in this case are still not fully understood. HMP consists of two domains: N-terminal globin and C-terminal reductase, both of which perform their function in the process of neutralizing NO [49]. HMP has been shown to exhibit enzymatic activity in human cells; for example, it has been used to study the functions of NO signaling in mammalian cells. The expression of flavohemoglobins led to increased resistance to NO-induced cell death and weakened various NO-signaling pathways [50]. According to some reports, HMP proteins are able to protect bacteria not only from nitrosative stress but also from oxidative stress caused by ROS. A number of structural features of HMP indicate the similarity of this protein to peroxidases [50,51]. It has also been shown that mutant bacteria lacking the gene encoding HMP are less resistant to oxidative stress [52], although its overexpression leads to the accumulation of peroxide and superoxide [53]. Purified MNT contains only a small fraction of heme, making it unlikely that HMP has enzymatic action in cells. Nonetheless, it is possible that these features of full-sized HMP can be observed in eukaryotic cells, potentially weakening the oxidative processes induced by photodynamic treatment. As both HMP domains operate together to achieve reductase activity, MNTF can be less effective than truncated MNTs.

Interaction with Keap1 at the same time, as it turned out, does not interfere with the delivery process but, on the contrary, gives MNT_N an advantage. It is known that Keap1 is localized, in particular, on the outer membrane of mitochondria [24,25], in connection with which the alleged interaction of MNT with Kear1 may lead to the accumulation of a photosensitizer on the outer membrane of mitochondria. Damage to mitochondria is one of the important pathways leading to cell death under the action of photosensitizers. Thus, mitochondria-associated photosensitizers can cause photodamage of the membrane-bound protein Bcl-227, which can lead to the release of caspase activators such as cytochrome c and Smac/DIABLO, or other pro-apoptotic molecules, including apoptosis-inducing factor (AIF) [54].

Photosensitizers can cause cell death through three mechanisms: apoptosis, necrosis, and autophagy. Some photosensitizers target mitochondria specifically [15]. MNT targeting Keap1 with an anti-Keap1 monobody has been shown to boost photosensitizer efficiency by localizing on the mitochondrial membrane [18]. Thus, an increase in the cytotoxic effect of the N-terminal HMP transporter capable of binding to Keap1 may indicate that this MNT is being targeted in addition to the nucleus and mitochondria.

5. Conclusions

New MNTs obtained as a result of the truncation of the carrier module are able to recognize target cells, penetrate them, and have a cytotoxic effect when photosensitizers are attached to them. This work shows that the HMP carrier module in the structure can be significantly reduced, and the effectiveness of MNT as carriers of therapeutic drugs into cells does not decrease. In this case, it is preferable to leave the N-terminal part of the HMP capable of binding to Keap1 in the MNT.