Intra-mitochondrial disulfide polymerization controls cancer cell necroptosis

In a biological system, energy-consuming reactions to attain a biological dissipative state are ubiquitous, and mimicking such reactions is a great challenge in synthetic chemistry. Herein, we report an intra-mitochondrial polymerization strategy for constructing macroscopic structures using a reactive oxygen species (ROS)-dissipative system. This is the first time that the occurrence of disulfide polymerization inside cancer mitochondria owing to the high ROS concentration of cancer mitochondria is reported. This polymerization hardly occurred inside cells owing to the intracellular reductive environment. The polymerization process of a thiol-containing monomer further increases the ROS level inside the mitochondria, thereby enabling the autocatalytic process to accelerate polymerization and induce mitochondrial dysfunction. This in-situ polymerization shows great potential for anticancer treatment against various cancer cell lines, including drug-resistant cancer cells.

Intracellular biomacromolecules, such as proteins, DNA, and polysaccharides, are synthesized by polymerization of small molecules, including amino acids and nucleotides.
Unlike small molecules, biomacromolecules show distinct properties and play important roles in cellular processes. [1][2][3][4][5] For instance, amino acids themselves display no enzymatic activities and exhibit trivial interactions with proteins. In contrast, proteins formed by the self-assembly of several polypeptides into well-ordered 3D or 4D structures often show enzymatic activities.
Proteins also play key roles in vital cellular signaling pathways involving protein-protein interactions. 6,7 Notably, the functions of these biomolecules strongly depend on their structures; therefore, disruptions of the protein assemblies significantly affect their properties, often leading to deactivation. 6 Research in the field of supramolecular chemistry has demonstrated the possibility of creating artificial intracellular assembly systems, which could imitate the biological and pharmaceutical functionalities of proteins. [8][9][10][11][12] It has been shown that small peptide-based amphiphilic molecules formed ordered structures inside cells. It is noteworthy that such systems exhibited intriguing interactions with biomacromolecules, enabling delivery of cargo into intracellular organelles, sequestering intracellular enzymes, and selectively causing cancer cell death. [13][14][15][16][17][18][19] Nevertheless, the formation of such ordered structures is dependent on the self-assembly equilibrium and requires high concentrations of small molecules, which is a result of a large entropic penalty associated with assembling ordered structures. Thus, this approach has significant limitations, including dynamic instability. 20 Polymerization-induced self-assembly (PISA) takes advantage of the chain-end reactivity of solvophilic macromolecules for the polymerization of a second monomer. Importantly, the growth of the second block, which is insoluble in the medium, results in the formation of block copolymers that self-assemble into nanoparticles via an entropy-driven chain collapse. 21,22 PISA has been demonstrated as a promising technique for the synthesis of polymeric materials used in drug delivery, medical imaging, and tissue culture. 23,24 It has been speculated that conducting PISA in living cells to construct artificial biological systems would enable intracellular formation of synthetic nanostructures with the ability to effectively interact with biomolecules. This would offer an opportunity to control cellular function, providing inspiration for the design of novel therapeutics. Several recent reports have described the potential of this approach. For example, in situ biosyntheses of inorganic nanomaterials by intracellular polymerization 25,26 and intracellular radical polymerization were successfully achieved. 27,28 Nevertheless, intracellular PISA leading to modification of the cellular fate remains a challenge. In this study, we report intramitochondrial polymerization to construct an artificial structure able to regulate cellular function for the first time.
Herein, a disulfide bond was used as a polymerizable functional group. Disulfide bonds are dynamic covalent bonds, which are in equilibrium between a disulfide and two thiols. This allowed for the construction of a dynamic combinatorial library of macrocycles, some of which formed fiber structures through autocatalytic reactions. 29-31 However, dissipative out-ofequilibrium state is common in nature. Consequently, instead of the formation of disulfide bonds, the presence of large amounts of a reducing agent (e.g., glutathione [GSH]) results in their destruction. Interestingly, in this study, we found that another type of dissipative reaction, namely disulfide polymerization, took place inside the mitochondria (Fig 1). The monomer was synthesized by conjugation of an aromatic molecule containing two thiol groups with a mitochondria-targeting unit, i.e., triphenylphosphonium (TPP). The resulting mitochondriatargeting monomers accumulated in the mitochondria of cancer cells, because compared with normal cells, the membranes of these organelles in cancer cells are more negatively charged. 32- 34 It is known that ROS are overproduced in the mitochondria of cancer cells; therefore, they could be utilized as chemical fuel. Oxidative agents catalyze the oxidation of thiol groups to form disulfide polymers. 35, 36 Furthermore, it has also been shown that the polymerization process induces oxidative stress in the mitochondria, leading to increased levels of ROS, which act as autocatalytic agents to catalyze further polymerization. Finally, the polymerization process induces self-assembly of the molecules into fibrous structures, resulting in mitochondrial dysfunction and activation of necroptosis.

Fig. 1 | Intramitochondrial polymerization-induced self-assembly (Mito-PISA).
Schematic illustration of Mito-PISA for controlling cancer cell necroptosis. a, Chemical structure of the monomer and b, polymerization-induced assembly by a disulfide polymer. c, Intramitochondrial polymerization can induce necroptosis and lead to mitochondrial dysfunction in cancer cells. Generally, the formation of disulfide bonds is prohibited by the intracellular reductive environment. However, intramitochondrial accumulation of monomers facilitates polymerization, resulting in oxidative stress and generation of ROS. Increased levels of ROS provide effective conditions for the formation of large polymers, leading to mitochondrial dysfunction and activation of the necroptosis signaling pathway.

Results and discussion
PISA using a disulfide bond. The target monmer (Mito-1) for intracellular polymerization was synthesized by the conjugation of a an aromatic dithiol with a quaternary ammoniummodified triphenylphosphonium (TPP) species via 1-ethyl-3-(3dimethylaminopropyl)carbodiimide (EDC) coupling. The resulting product was purified by high-performance liquid chromatography (HPLC). To confirm the functionality of the polymerization, a control molecule (Mito-2) containing dihydroxy moieties instead of dithiol groups ( Fig. 1) was also prepared. The polymerization of Mito-1 in phosphate-buffered saline (PBS) at pH 8 was first analyzed by HPLC and gel permeation chromatography (GPC). As demonstrated in Fig. 2a, the intensity of the peak corresponding to Mito-1 (10 mM) sharply decreased within 4 h. It was determined that after 4 h, 80% of the monomer was consumed and the molecular weight of resulting polymer was established at 1.8 × 10 4 g/mol. In addition, the polydispersity index (PDI) was calculated at 2.3. Conversely, the intensity of the peak attributed to Mito-2 and the molecular weight of the monomer remained unchanged (Fig. 2a).
This indicated that the thiol groups in Mito-1 were oxidized to form disulfide bonds, generating a polymer.
In general, disulfide bonds easily decompose under reducing conditions, such as the cellular environment. 37,38 Thus, the occurrence of disulfide polymerization inside living cells is remarkably rare. Nonetheless, we speculated that if the concentration of the dithiol monomer was sufficiently high, disulfide polymerization could take place. Hence, the concentrationdependent polymerization of Mito-1 was investigated in PBS (pH 8) in the presence of 10 mM GSH to mimic the intramitochondrial environment. The intensity of the peak corresponding to the monomer was monitored at various concentrations of the compound (1 and 10 mM). When 10 mM of the monomer was used, the intensity of the peak considerably decreased after 6 h.
In contrast, no significant decrease was observed at 1 mM (Fig. 2b). This suggested that polymerization did not occur at low concentrations of the monomer due to high levels of GSH (10 mM), which resulted in the reduction of disulfide bonds. However, high concentrations of the monomer shifted the equilibrium of the dynamic covalent bond toward the bond formation reaction rather than decomposition, thus enabling the occurrence of polymerization in a reducing environment.
Intriguingly, we found that the self-assembling behavior depended on the concentration of the monomer during polymerization. Dynamic light scattering (DLS) analysis revealed that both 1 mM and 10 mM solutions initially exhibited small aggregates (20 nm), which resulted from the self-assembly of the monomers. After 12 h, a large aggregate (786 nm) was observed in the 10 mM solution of Mito-1, while no size change was detected in the case of the 1 mM solution (Fig. 2c). The formation of fibrous polymeric structures in the high-concentration solution (10 mM) was confirmed by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) (Fig. 2d). These outcomes indicated that the self-assembly during polymerization was induced by the increase in the hydrophobicity of the polymers by the disulfide bonds on the aromatic core, which resulted in a decrease in the entropic penalty.
The above observations suggest that the monomer concentration is important in the processes of disulfide polymerization and polymer self-assembly, particularly in an intracellular reducing environment. To estimate the optimal concentration of the monomer for intracellular PISA at 10 mM GSH, we performed the polymerization at varying concentrations of Mito-1 (0.1-10 mM) in the presence of a hydrophobic probe, namely Nile red (300 nM).
Nile red exhibits no fluorescence in water; however, undergoes fluorescence enhancement in hydrophobic environments. As demonstrated in Fig. 2e, a sharp increase was detected at a Mito-1 concentration of ~2 mM, indicating that PISA can take place above this concentration even in a reducing environment. The progress of the polymerization was further monitored at different monomer concentrations using a solution mimicking the mitochondrial environment.
HPLC analysis revealed that no polymer formation was observed in a 1 mM monomer solution; however, construction of large polymers was noted as the concentration was increased to >2 mM ( Fig. 2f, g). This suggested that in the intramitochondrial environment, a monomer concentration of at least 2 mM was required for the formation of disulfide macromolecules.
As mentioned above, oxidizing agents can acts as fuels for the formation of disulfide bonds. 39 Thus, it was envisioned that the presence of such compounds would accelerate PISA of Mito-1 in a reducing environment even at low concentrations of the monomer. In this study, we monitored the consumption of Mito-1 in the presence of 100 µM hydrogen peroxide (H2O2) as the oxidizing agent using HPLC. No significant changes in the intensity of the peak corresponding to the monomer (1 mM) were observed in the absence of the oxidizer. In contrast, a decrease in the peak intensity was noted following treatment with 100 µM H2O2 ( Fig. S1a). This implied that the employed oxidizing agent catalyzed the polymerization and PISA (Fig. S1a, c). The sizes of the assemblies in 1 and 10 mM solutions of Mito-1 incubated in the presence of 100 µM H2O2 were determined at 706 and 2308 nm, respectively (Fig. S1b).
The addition of 100 µM H2O2 lowered the critical concentration of PISA to 1.1 mM (Fig. S1d,   e), which was in accordance with the microscopic observations (Fig. S1c). Disulfide polymerization and the self-assembly process were further investigated employing a computational method (see section titled Simulation details). The possibility of disulfide polymerization was predicted by constructing reaction coordinates using density functional theory (DFT) calculations (Fig. S2a). It was found that the formation of a disulfide bond was possible when H2O2 was located between two 1,3-benzenedithiol groups. The distance between the hydrogen atom of the thiol group (-SH) and the oxygen atom of H2O2 was determined at 2.3 Å. Analysis of the exothermic formation of the disulfide bond showed that the dissociation of H2O2 into OH − and hydration reaction occurred when the energy barrier was ~20 kcal/mol. Conversely, the scission of the disulfide bond required a large amount of energy (i.e., ~91 kcal/mol). Hence, it was anticipated that the disulfide bond formation would be thermodynamically favorable and stable in the ROS environment. In addition, the promotion of polymerization in the presence of H2O2 was evaluated using a molecular dynamics (MD) simulation. Unlike in pure water, at higher concentrations of H2O2, the thiol moieties were located in a range of distances (i.e., 3.5-4.7 Å) ( Fig. S2b) due to the formation of hydrogen bonds (HB) with the oxidizer (Fig. S2c). A representative configuration, in which H2O2 is located between two thiol groups, is shown in Fig. S2d. Thus, based on the theoretical analysis, polymerization through disulfide bonding could be promoted in the ROS environment.
Moreover, higher degree of polymerization could be expected at higher H2O2 concentrations.
We subsequently investigated whether the degree of polymerization affected the selfassembled structure by varying the polymer chain length using the coarse-grained MD (CGMD) simulation. All long (i.e., 34 constituent monomers) and short (i.e., 5 constituent monomers) polymer chains (Fig. S3a) in a high and low ROS environment self-assembled to fibrous ( Fig.   3a) and spherical structures (Fig. S3b), respectively. In both fibrous and spherical structures, the main interactions during the self-assembly process were non-polar interactions between the backbone (BB) groups ( Fig. 3b and Fig. S3c). During the formation of the fibrous structure, the BB groups were entangled along the long chain axis and formed a stacked structure with the benzene rings (Fig. 3c). It was found that long polymerized chains, which hardly bent after a 30% decrease in the end-to-end distance during the initial 20 ns of the self-assembly ( Fig.   S4a), formed net-like stems that turned into a fibrous structure (Fig. S4b). During the construction of the net-like stems, the side chain (SC) and TPP groups were positioned at the surface of the fibrous structure (Fig. 3d). It is noteworthy that both SC and TPP groups are positively charged; therefore, they are able to target tumor cell membranes. In contrast, in the spherical structure, the BB groups could not interact in the parallel direction due to the insufficient length of the chains (Fig. S5a). Thus, the entanglement of BBs along the axis was less likely. Consequently, with the exception of the core, the contents of the three Mito-1 components (BB, SC, and TPP) were similar throughout the structure (Fig. S5b). Accordingly, we anticipated that compared to the spherical assembly, the membrane-interaction ability of the fibrous structure, which was self-assembled by a high degree of polymerization, would be higher due to the exposed TPP and SC groups on the surface. c, Binding states of two polymer chains at the surface of the fibrous structure. Stacking of the BB groups is shown in the red box. The distance and angle of the stacking interaction are also indicated. d, Number density of TPP, SC, and BB in the fibrous polymeric structure between the principal axis of the cylindrical shape to its surface. Surface region in the fibrous structure is indicated in light blue.
Intramitochondrial PISA. The above results indicated that PISA of Mito-1 required a certain concentration (~2 mM) of the monomer and could be accelerated in an oxidative environment.
It was hypothesized that mitochondria of cancer cells would be suitable locations for PISA of Mito-1, because the TPP moiety could drive the selective localization of the monomers into the organelles with high local concentration. [40][41][42] In addition, ROS are overexpressed in most cancerous mitochondria. Nitrobenzoxadiazole (NBD)-labeled Mito-1 (Mito-1-NBD) and Mito-2 (Mito-2-NBD) were synthesized to investigate the intramitochondrial accumulation of the monomer as well as the subsequent polymerization process (Fig. 4a). It was confirmed that Mito-1-NBD exhibited green fluorescence with a maximum intensity at 545 nm (Fig. 4b).
Consequently, PISA could be confirmed by the appearance of bright fluorescence inside the cell. NBD displays strong fluorescence in hydrophobic environments; therefore, it could be effectively employed as a reporter for the intracellular assembly. 43  Pearson's correlation coefficient (0.806) (Fig. 4d).
Moreover, the mitochondrial accumulation of Mito-1-NBD was evaluated to establish whether an effective concentration of the monomer inside the organelle could be achieved.

Mito-1-NBD was incubated with HeLa and HEK293 cells for 4 h, and the emission spectrum
was monitored using a mitochondria-isolated solution. The conducted calculations showed that the accumulation of Mito-1-NBD inside cancerous mitochondria was consistent with the incubation concentration, i.e., 7.5 ± 1.3 mM and 18.9 ± 2.4 mM for incubation concentrations of 10 and 30 µM, respectively. Furthermore, the Mito-1-NBD accumulation level was >7 times greater in the mitochondria of cancer cells than that in normal cells (Fig. 4e). lysate. The molecular weight of the polymer was determined at 8 kDa and the DPI was calculated at 1.2 (Fig. 4f). To verify the importance of mitochondrial localization, we subsequently synthesized TPP-free compounds (Cyto-1) as controls (Fig. S10a). In the cytosol, abundant reducing agents, such as GSH, promote the destruction of disulfide bonds rather than their generation, preventing polymerization. Expectedly, no polymer formation was observed in the cell lysate treated with 30 µM Cyto-1 (Fig. 4g), suggesting that mitochondria targeting is crucial for the occurrence of disulfide polymerization. To assess the intramitochondrial polymerization in more detail, a TEM experiment was conducted using HeLa cells treated with 30 µM Mito-1 for 12 h. The obtained TEM images showed disruption of the cellular membrane and the lack of normal mitochondria. Instead, the presence of a fibrous polymeric structure, which penetrated the mitochondrial membrane, was noted (Fig 4h). However, the mitochondria in the HEK293 cells treated with Mito-1 remained unaffected. Thus, it is expected that the presence of a polymeric structure inside the mitochondria of cancer cells could adversely affect mitochondrial function. in HeLa cells treated with Mito-2 (Fig 5b, S7b). This indicated that polymerization-induced oxidative stress in the mitochondria, which led to the generation of ROS.
The production of ROS driven by the initial polymerization process promoted further autocatalytic polymerization, leading to the construction of bulky polymeric structures and severe damage to the mitochondria. To evaluate the mitochondrial damage, the morphology of these organelles was observed using fluorescence microscopy. A fluorescent Mito-tracker was used to exhibit elongated healthy mitochondria in the HeLa cells after 1 h of incubation with 40 µM Mito-1. Following incubation for 4 h, however, severe mitochondrial fragmentation and swelling were observed (Fig. 5c). In contrast, mitochondrial morphology remained unchanged after incubation with Mito-2 (Fig. S7c). It is known that mitochondria play crucial roles in the production of ATP; hence, to investigate the mitochondrial dysfunction in more detail, we analyzed the cellular ATP levels. It was found that the ATP levels in the HeLa cells treated with Mito-1 for 24 h decreased by 85% (Fig. S8a). Undoubtedly, ATP depletion significantly contributed to cellular dysfunction, causing membrane disruption, protein damage, and cellular death. In this study, the cellular damage was analyzed by monitoring the release of lactate dehydrogenase (LDH), an enzyme released from the cytosol as a result of damage to the plasma membrane. The release of LDH was studied in HeLa and HEK293 cells treated with Mito-1 for 24 h. A release of LDH of ~80% was confirmed in HeLa cells treated with 50 µM Mito-1, whereas a release only 20% was observed in HEK293 cells (Fig. S8b). These outcomes suggested that the fibrous polymers inside mitochondria destroyed the mitochondrial membrane and led to the dysfunction of the organelle.
The penetration of fibrous polymeric structures into the cancer cell membranes was also confirmed by CGMD simulations. It was shown that the fiber reached the center of the lipid bilayer after 1 µs (Fig. 5d). Initially, i.e., within 0-5 ns, the TPP functionality strongly interacted with the surface of the cell membrane, which was composed of negatively-charged hydrophilic phosphate groups (Fig. 5e). Due to the interaction of TPP with the membrane surface, the side chain of the molecule was stretched and the angle between BB and TPP increased (Fig. 5f). In addition, we found that the TPP group penetrated the cell membrane earlier and deeper than the BB and SC moieties. This was evidenced by the distance of the Mito-1 components from the center plane of the membrane determined during the simulation (Fig. 5g). When the fibrous structure reached the center of the membrane, the TPP moiety mainly interacted with the hydrophobic groups of the membrane. In addition, the interactions between the SC group and the hydrophilic functionalities of the membrane became strong (Fig.   S9). This indicated that the TPP groups penetrated the membrane first, followed by the SC moieties. Hence, we theoretically predicted that the fibrous polymeric structure allowed the TPP groups to effectively interact with the cancer cell membrane, which was advantageous for targeting and inducing its destruction. Based on these results, we expected that intramitochondrial polymerization could be a potential anticancer strategy, as the dysfunction of mitochondria is associated with cellular death. We first examined the toxicity of intramitochondrial polymerization against HeLa and HEK293 cells for 24 h. Mito-1 exhibited high cytotoxicity against HeLa cells (Fig. 6a), while very low toxicity was observed against HEK293 cells, suggesting that Mito-1 selectively induced cancer cell death (Fig. 6b). In contrast, at the same concentration, Mito-2 displayed almost no toxicity toward both cancer and normal cells (Fig. 6a, b). Mitochondria are key organelles responsible for the production of energy. They are also associated with the cellular death mechanisms; therefore, the dysfunction of these organelles significantly affects the intrinsic cell death signaling pathway. Hence, targeting mitochondria would enable cancer therapy with reduced side effects. Notably, such an approach would also prevent drug resistance. 23 We subsequently analyzed the toxicity of Mito-1 toward various other cancer cell lines, including MDA-MB-468, MDA-MB-231, P-132, and SKBR3. As shown in Fig. 6c, Mito-1 exhibited toxicity toward the investigated cancer cell lines, with IC50 values of 18-28 µM. Importantly, lower toxicity was observed in the case of normal cell lines (i.e., IMR90 and HEK293) (Fig. S11).
Furthermore, considering that the mitochondria-targeting strategy might be a promising approach to overcome drug resistance, we tested the cytotoxicity of Mito-1 toward doxorubicin-resistant cancer cells (MCF7/ADR) after 24 h of incubation. Notably, Mito-1 displayed higher cytotoxicity than doxorubicin, with an IC50 of 41 µM (Fig. S11). This implied that employing the mitochondria-targeting polymerization system would be an effective strategy to overcome multidrug resistance in cancer treatment. To investigate the mechanism of cellular death under these conditions, fluorescenceactivated cell sorting analysis was performed utilizing the FITC-annexin V/PI apoptosis detection kit. HeLa cells were treated with 40 µM Mito-1 for 12 h. It was found that the cells lapsed into the necrosis stage within 12 h (Fig. 6d, S12). Moreover, TNF-α, a necrosis related protein, was overexpressed in the cells treated with Mito-1 (Fig. S13a). To examine the cell death mode in more detail, we evaluated the effect of necrostatin-1 (Nec-1), a necroptosis inhibitor, on the cells treated with Mito-1. 45,46 Notably, treatment of these cells with Nec-1 prevented them from cellular death. This suggested that Mito-1 induced necroptosis, a programmed form of necrosis, in HeLa cells (Fig. S13b). Using western blot analysis, we further explored the mechanism of cell death by focusing on the necroptotic signaling pathways of HeLa cells subjected to intramitochondrial polymerization. The cellular stress resulting from Mito-1 exposure led to elevated expression of TNFR2 and decreased expression of IKKα, IKKβ, and Akt (Fig. 6e, f). In contrast, incubation of the HeLa cells with Mito-2 did not have a significant effect on the expression levels of the necroptotic proteins. These outcomes indicated that intramitochondrial polymerization activated cellular death and contributed to the anticancer activity of the monomer. However, an experiment involving treatment of the HeLa cells with Cyto-1 revealed lack of toxicity after 24 h of incubation, suggesting that mitochondria targeting is essential for disulfide polymerization as well as for effective regulation of the cellular function (Fig. S10b).
Intriguingly, the cytotoxicity of Mito-1 was enhanced after treatment with H2O2, which increased the intracellular ROS levels, indicating that polymerization was accelerated in an oxidative environment. As demonstrated in Fig. S14a, higher cytotoxic activity was observed in HeLa cells pretreated with 30 µM H2O2 for 24 h (IC50 = 17.7 µM) than in cells without any oxidizer pretreatment (IC50 = 22.3 µM) (Fig. S14a). However, an opposite trend was observed in the case of cells treated with glutathione monoester (GSH-OMe). It is known that due to its anionic nature, GSH cannot be internalized into cellular membranes. Conversely, neutral GSH-OMe can be incorporated into cells, where it is hydrolyzed to GSH. We found that Mito-1 exhibited lower toxicity toward cells treated with GSH-OMe (IC50 > 50 µM) (Fig.   S14b). Inspired by these results, we assumed that a combination of Mito-1 and β-lapachone (Lapa), a therapeutic agent resulting in the generation of ROS catalyzed by NQO1, would lead to synergistic effects, thus representing an attractive strategy for boosting the efficacy of intramitochondrial polymerization. 47 As the expression of the NQO1 enzyme in cancer cells is 100 times higher than in normal cells, the use of Lapa could improve the selectivity of the ROS-responsive system. 47 The ROS assay demonstrated high levels of ROS in HeLa cells treated with 5 µM Lapa. In contrast, the ROS levels in HEK293 cells remained unchanged after incubation with 5 µM Lapa (Fig. 6g). Lapa significantly enhanced ROS levels in cancer cells, and the corresponding cytotoxicity of Mito-1 was evidently higher. Combination of Mito-1 (20 µM) and 5 µM Lapa led to a decrease in the cell viability to 35%. In the case of Mito-1 alone, the cell viability decreased to 78%. It is noteworthy that 80% of normal cells treated with 20 µM Mito-1 and 5 µM Lapa remained unaffected (Fig. 6h), which was attributed to the low level of ROS in these cells (Fig. 6g). To examine the improved selectivity in more detail, we calculated the selectivity index by dividing the IC50 value of the normal cells by the IC50 value of the cancer cells. Mito-1 alone exhibited a low selectivity index of 2.79, whereas treatment with both Mito-1 and 5 µM Lapa yielded a higher selective index of 3.75 (Fig. 6i).
These outcomes demonstrated that such a synergistic strategy could improve the available anticancer treatments. 1 was measured at 4.1, 3.9, 2.1, and 1.8 g, respectively (Fig. 7b, c). In the control group, 60%

In vivo
of the mice died within 45 days. An obvious improvement of the mice survival was observed for the Mito-1 and Lapa+Mito-1 groups (Fig 7d). This suggested that the intramitochondrial polymerization resulted in significant anticancer effects. Administration of Mito-1 and Lapa+Mito-1 did not induce any noticeable changes in the mice body weight, indicating that the intramitochondrial polymerization was tolerated by the animals (Fig. 7e). In addition, we evaluated the cell morphology by staining the cells with hematoxylin and eosin using the TUNEL assay to investigate the mode of cell death. Both control-and Lapa-treated groups exhibited no damage in the tumor section. Necrosis was observed in Mito-1-and Lapa+Mito-1-treated groups (Fig 7f, g). Furthermore, histopathology of other organs revealed negligible toxicity resulting from intramitochondrial polymerization in vivo (Fig. S15). Overall, we successfully demonstrated that intramitochondrial PISA led to anticancer effects. Notably, combination of the synthesized monomer with Lapa resulted in promising synergistic effects.

Conclusion
In the present study, intramitochondrial polymerization using disulfide bonds was achieved for to elucidate the mechanism of the disulfide bond formation and dissociation in the 1,3benzenedithiol moiety in Mito-1 using the DMol 3 software. 48 included 214 water and 12 Mito-1 molecules. All MD systems contained the same number of atoms. The NPT MD (i.e., isothermal-isobaric) simulation was performed for 500 ps and the temperature was increased from 300 to 400 K in 10 K intervals. The cycle was repeated 5 times.
Subsequently, the NPT simulation was performed for 1 ns at 300 K. COMPASS II force field 55 was used and the Mulliken charge was applied to 1,3-benzenedithiol and H2O2. To maintain the isothermal state, a velocity rescaling thermostat was employed with a time step of 1 fs. The radial distribution function (RDF) was analyzed using the trajectories of the final 500 ps and the number of hydrogen bonds was counted based on 5 trajectories of the final 100 ps.
Coarse-grained molecular dynamics (CGMD). The coarse-grained polymer model was composed of 5 or 34 constituent monomers. The molecular weight was ~2.9 and ~20 kDa, respectively. The MARTINI force field was applied, and the bead types for each CG bead are summarized in Table SX1. 56 For the simulation of the self-assembly of the fibrous polymeric structure, 40 polymer chains were used in a box of 40 × 40 × 40 nm 3 filled with water and chlorine ions. To consider the self-assembly following polymerization, in the initial state, the monomers in each chain were connected by distances of >1.5 nm. The number of water beads was 11,200 and the concentration of the Mito-1 polymer was 8.99 wt%. Sequential NVT and NPT simulations were conducted at 298 K for 300 ps and 100 ps, respectively, and the polymer was aggregated to a fibrous structure. To confirm the maintenance of the fibrous structure, water beads were added around the structure and the concentration of the Mito-1 polymer was 2.97 wt%. The NPT simulation was performed for 15 ns and the temperature was increased from 300 to 400 K in 20 K intervals. The cycle was repeated 3 times. The NPT simulation was then performed for 300 ns at 300 K. The process of self-assembly of the spherical structure and the concentration of Mito-1 were same as those used for the fibrous structure. The number of short polymer chains consisting of 5 monomers was 272 and a box with the dimensions of 23×23×23 nm 3 filled with water and chlorine ions was used for the simulation. For polymer aggregation, the NPT simulation was performed for 150 ns at 300 K. To confirm the preservation of the spherical structure, water beads were added around the macromolecule and the same sequential simulation as that for the fibrous structure was performed. The number density of the three Mito-1 components was analyzed using 5 trajectories of the final 25 ns. In addition, the number of binding sites in the fibrous structure was determined using a distance between the particles of <0.