Mitochondria-targeting theranostics

Background Interest in subcellular organelle-targeting theranostics is substantially increasing due to the significance of subcellular organelle-targeting drug delivery for maximizing therapeutic effects and minimizing side effects, as well as the significance of theranostics for delivering therapeutics at the correct locations and doses for diseases throughout diagnosis. Among organelles, mitochondria have received substantial attention due to their significant controlling functions in cells. Main body With the necessity of subcellular organelle-targeting drug delivery and theranostics, examples of mitochondria-targeting moieties and types of mitochondria-targeting theranostics were introduced. In addition, the current studies of mitochondria-targeting theranostic chemicals, chemical conjugates, and nanosystems were summarized. Conclusion With the current issues of mitochondria-targeting theranostic chemicals, chemical conjugates, and nanosystems, their potentials and alternatives are discussed.


Background
The development of drug delivery systems (DDSs) has been initiated to solve the poor aqueous solubility of chemical drugs and the poor blood stability of protein and gene drugs. For this purpose, various micron-sized or nanosized DDSs, such as microspheres, micelles, and liposomes, have been investigated, and few clinically available products, such as Lupron Depot®, Genexol PM®, and Doxil®, respectively, are currently on the market [1]. However, their low functionalized characteristics have limited to improve the therapeutic effects and reduce the side effects of delivered drugs.
Thus, with certain terms, the roles of these DDSs have been expanded. First, the term "Targeting" has been introduced to maximize therapeutic effects and minimize unwanted effects of drugs because the places that require drugs (i.e., target sites) should be distinguished from other places (i.e., nontarget sites). Most targeted DDSs have been designed to selectively reach to the organs, tissues, and cells of interest by physiological, pathological, and anatomical differences and specific interactions between ligands on the targeted DDS and their counter receptors on the plasma membrane. Very recently, for deeper targeting than cellular levels, subcellular organelles have been attracted because the organelles are actual sites of the mode of action of drugs [2,3]. Second, the terms "Imaging" and "Diagnosis" of imaging or diagnostic molecules have been introduced instead of "Therapy" of drugs because the sites and the severity of diseases should be known for the effective therapeutic effects of prescribed drugs and the payloads in delivery carriers are not only drugs but also imaging or diagnostic molecules. In particular, beyond "Imaging", "Theranostics" (i.e., combined systems of "Therapy" and "Imaging/Diagnosis") has been interested in future DDSs. Thus, theranostic DDSs at subcellular organelles could be the future in the field of DDSs. Among various subcellular organelle-targeting theranostics, this review focuses on mitochondria-targeting theranostic DDSs.

Theranostics
Theranostics may be defined as a combined system of diagnostics and therapeutics in a formulation. Its significance is to reduce the "trial-and-error" process for identifying a correct medicine and then maximize the therapeutic effects of the medicine because diagnostics in theranostics can diagnose the locations and status of diseases in organs, tissues, or cells; moreover, therapeutics in theranostics can effectively treat the diseases. Thus, interest in theranostics with the dual functions of therapeutic efficacy and diagnosis/imaging is rapidly increasing.
The simultaneous delivery of both diagnostics and therapeutics can be performed using chemical conjugates or nanosized/micron-sized carriers. For chemical conjugates, diagnostics and therapeutics can be chemically linked to each other, at ends of certain linker molecules, or to water-soluble macromolecules. Moreover, in various nanosized or micron-sized carriers, diagnostics and therapeutics can be physically or chemically loaded. Currently, the wide variety of imaging molecules as diagnostics includes radioactive nuclides, optical probes, or metal chelates, which are detectable by positron emission tomography (PET)/single photon emission computed tomography (SPECT), fluorescence, or magnetic resonance imaging (MRI), respectively. In particular, when therapeutics can be detected by various imaging tools, one imaging-capable therapeutic molecule can be used instead of two separate molecules of diagnostics and therapeutics. As imaging-capable therapeutics, fluorescent therapeutics, such as photosensitizers (e.g., pheophorbide a) and fluorescent drugs (e.g., doxorubicin), are well established.

Mitochondria-targeting drug delivery
Active ingredients can target various subcellular organelles depending on their modes of action and their physico-chemical properties. For example, in general, cisplatin and pDNA are delivered into the nucleus to alkylate DNA and express mRNA, respectively. For paclitaxel and siRNA, they need to reach microtubules and target mRNA in the cytosol, respectively. Moreover, some drugs are required to be accumulated in lysosomes, mitochondria, and endoplasmic reticulum (ER). Among these subcellular organelles, this review focuses on mitochondria-targeting theranostics because mitochondria modulate significant physiological functions. In particular, mitochondria control the homeostasis of intracellular Ca 2+ levels and oxidative stress and rule cell viability/death and signaling through bioenergy production, cellular differentiation/ growth, the cell cycle, and cell necrosis/apoptosis/autophagy [4,5]. Thus, their malfunction and dysfunction frequently cause unwanted growth or death of cells and lead to many neurodegenerative, neuromuscular, cardiac, and metabolic diseases and cancers [5,6].
Despite the significance of delivering therapeutic molecules into mitochondria, as well as monitoring mitochondrial functions and morphology and therapeutic molecules localized in the mitochondria, the number of research articles on mitochondria-specific delivery was approximately 200 by 2014 [7]. Janus Green B was reported as the first mitochondria-staining dye in 1899 [8]. Moreover, the mitochondrial targeting and accumulating characteristics of the well-studied mitochondria-targeting triphenylphosphonium (TPP) and dequalinium (DQA) as molecules were first reported in 1969 [9] and 1987 [10], respectively. However, the first mitochondria-targeting drug conjugates or complexes using TPP and DQA were reported in 1995 [11] and 1998 [12], respectively. Studies on mitochondria-targeting drugs/ diagnostics, drug/diagnostic conjugates, and drug/diagnostic delivery systems have been silently conducted by a few researchers by 2014. However, the interest in the topics has been substantially increasing since 2015 [7]. With the critical developments in biological and imaging analyses, chemical synthesis, and nanotechnology, various mitochondria-targeting moieties and their chemical conjugates and delivery systems have been extensively investigated. Nevertheless, the total number of research articles on mitochondria-specific delivery of therapeutics and imaging molecules remains less than 1000 [7]. Thus, mitochondrial drug delivery and particularly mitochondrial theranostics would be very hot research topics in the biomedical and pharmaceutical fields.
To date, many researchers have investigated various mitochondria-targeting moieties, including chemicals and peptides. As shown in Fig. 1, well-known mitochondria-targeting chemicals, such as TPP, DQA, (E)-4-(1H-Indol-3-ylvinyl)-N-methylpyridinium iodide (F16), rhodamine, and guanidine, have lipophilic cations or delocalized cations because these characteristics enable them to cross the mitochondrial membrane using the difference in the membrane potentials between the outer mitochondrial membrane and the inner mitochondrial membrane [7]. Furthermore, when introducing mitochondrial targetability into theranostics, it is possible to design three functionalities (i.e., treatment, imaging, and mitochondria-targeting abilities) in one conjugate and delivery system. Thus, this review will provide an overview of recent mitochondrial theranostic chemicals, chemical conjugates, and delivery systems.

Mitochondria-targeting theranostics
As shown in Fig. 2, mitochondria-targeting theranostics have been designed by combining three different functional components, such as mitochondria-targeting components (M), therapeutic components (T), and imaging/diagnostic components (D) and have been prepared by mixing or conjugating single functional components with dual or triple functional components [13][14][15]. These components include chemical molecules, macromolecular structures, or nanosized systems. Thus, if each component has mono-functionality, three different components should be included in one mitochondria-targeting theranostic chemical conjugate or delivery system. However, if one component has dual functionalities (i.e.
Among the previously described mitochondria-targeting photosensitizers, Tan et al. designed IR-DBI as an alternative of theranostic indocyanine green (ICG) [29]. ICG is an FDA-approved NIR contrast agent with photodynamic therapy (PDT) and photothermal therapy (PTT). However, the poor cellular uptake and poor tumor-specific accumulation of ICG have limited its clinical applications. In contrast to the chemical structure of ICG, the chemical modification of a rigid cyclohexenyl group and the lipophilic cationic N-alkyl side chains in IR-DBI ( Fig. 3(a)) enabled mitochondria-targeting and upgraded PDT and PTT, consequently leading to effective and selective tumor growth inhibition [29]. As shown in Fig. 5(a), when an NIR of 770 nm as an excitation wavelength was irradiated to IR-DBI or ICG-treated A549 cells, IR-DBI-treated A549 cells emitted substantially stronger fluorescence at 830 nm than ICG-treated A549 cells. In particular, in two breast cancer cell lines (i.e., 4T1 and MCF-7 cells), the intracellular localization of IR-DBI was strongly overlapped with the intracellular distribution of MitoTracker™ Green localized in the mitochondria, which indicates the strong mitochondria-targeting activity of IR-DBI ( Fig.  5(b)). Interestingly, NIR fluorescence imaging of IR-DBI confirmed that IR-DBI accumulated selectively in tumor cells (e.g., A549 and MCF-7 cells); however, it was not detected in normal cells (e.g., HBE and MCF-10A cells) (Fig.  5(c)). The results suggest that IR-DBI specifically targets tumor cells and then further reaches their mitochondria and that IR-DBI could reduce the nonspecific cytotoxicity to normal cells. Beyond in vitro cell imaging, IR-DBI imaged solid tumors in A549, QBC-939, HeLa, or 4T1 subcutaneous tumor xenograft-bearing mice (Fig. 5(d)). Moreover, NIR-irradiated IR-DBI produced more singlet oxygen than control ICG (Fig. 5(e)) and heated body temperatures to approximately 54°C in 4T1 tumor-bearing mice (Fig. 5(f)). As a result, PDT and PTT of NIR-irradiated IR-DBI almost completely inhibited tumor growth in 4T1 tumor-bearing mice (Fig. 5(g)). The research showed that M-T-D 3-in-1 typed IR-DBI as a small mitochondria-selective theranostic chemical could diagnose tumor cells and synergistically kill the cells with the multimodal therapeutic activities of chemotherapy, PDT, and PTT. In addition, Noh et al. designed a similar chemical structure of IR-DBI, but introduced TPP into indole containing photosensitizers, which ultimately led to the formation of MitDt compounds [34].
For the M-D 2-in-1 case, Xu et al. designed a fluorescent oridonin probe 17d [15]. The probe was synthesized by chemically linking an anticancer drug, oridonin, with an M-D 2-in-1 typed N,N-dialkyl-7-aminocoumarin derivative. A quaternary ammonium and a coumarin in the aminocoumarin derivative (Fig. 3(a)) endowed mitochondria-targeting activity and fluorescence imaging at an excitation/emission of 425 nm/454 nm, respectively. In particular, oridonin in the conjugate destabilized mitochondrial membrane potentials, triggered to release cytochrome c, and induced apoptotic cell death. As a result, the fluorescent oridonin probe 17d showed approximately 2.04-fold, 5.85-fold, and 8.9-fold higher killing activities than oridonin in A549, HepG2, and HeLa cells, respectively.
Interestingly, Hu et al. designed mitochondria-targeting pro-theranostic chemical conjugates [35]. In Fig. 3(a), the indole derivative in HMX-1 delivered nonfluorescent HMX-1 (i.e., a pro-theranostic) into mitochondria. A hypoxic condition triggered the break of an azo bond in an azobenzene spacer, which functions to quench a fluorescent probe and deactivate an aniline nitrogen mustard, and then released the fluorophore and the active drug for imaging mitochondria and killing cells, respectively.
Wang et al. designed a mitochondria-targeting, dual-mode imaging guided multifunctional theranosome (TNS) (Fig. 6) [38]. The vesicle-typed TNS was mainly constructed by a self-assembly of TPP-PEG 2000 -DSPE polymer-lipid conjugates, and mitochondria-targeting TPP was located on the surface of TNS due to the hydrophilicity of TPP-PEG 2000 . During the preparation of TNS, two photosensitizers (i.e., IR780 and Ce6) were physically encapsulated in the TNS, which resulted in the formation of TPP/IR780/Ce6 TNS (TICT). The TICT was designed to produce PTT and PDT for effective tumor killing and both NIR fluorescence and photoacoustic imaging. In particular, the sequence of laser irradiation is very significant. If a 660 nm laser for activating Ce6 is first irradiated to TICT, the blocked status of Ce6 in TICT cannot produce singlet oxygen. Thus, as shown in Fig. 6(a), 808 nm NIR light is first irradiated to TICT to produce heat and PTT activity and disrupt lysosomes. TICT escaped from the lysosomes targets mitochondria using TPP in TICT and is then disrupted under hyperthermia. The integrity loss of TICT releases Ce6, and a 660 nm laser is irradiated to the released Ce6 to generate singlet oxygen and activate PDT. The sequential activation of PTT and PDT by two different photosensitizers in TICT effectively kills tumor cells and monitors body temperatures, NIR fluorescence, and photoacoustic signals. In contrast to the poor mitochondrial localization of IR780/Ce6 TNS (ICT), the red fluorescence of IR780 in TICT was perfectly merged with the green fluorescence of MitoTracker™ Green-stained mitochondria ( Fig. 6(b)). For singlet oxygen generation, when only 660 nm light was irradiated, TICT and ICT produced low ROS levels. However, when sequentially irradiating 808 nm and then 660 nm light, both TICT and ICT produced more intracellular ROS levels than the cases of only 660 nm irradiation. The singlet oxygen levels produced by TICT were substantially higher than those produced by ICT (Fig. 6(c)). Interestingly, the designed theranostic nanosystems generated heat to increase body temperature for effective PTT. When an 808 nm NIR laser was on for 2 min, IR780 generated heat, and TICT and ICT in a test tube made 59.4°C and 58.2°C of T max , respectively. The heat generation of TICT was substantially more effective than that of free IR780. Moreover, when giving the irradiation cycle through the switch "ON" and "OFF" of the 808 nm NIR laser, IR780 in TICT slowly lost the heat generation activity in contrast to the rapid loss of free IR780. NIR fluorescence imaging generated by IR780 in TICT monitored the biodistribution of TICT in tumor-bearing mice. Compared to free IR780/Ce6, TICT exhibited a stronger NIR fluorescence, and its fluorescence was maintained for a longer time. Interestingly, the photoacoustic imaging produced by TICT was continuously increased by 24 h, and its signal intensity was approximately 2.5-fold higher than that of free IR780/Ce6 at 24 h. Moreover, although TICT did not contain contrast agents, the photoacoustic imaging of TICT was better than its ultrasound imaging. Although mice were exposed to the 808 nm NIR laser for 5 min, the 808 nm-irradiated TICT produced substantially higher body temperatures in tumor sites than the 808-irradiated free IR780/Ce6. Combining PTT and PDT resulted in TICT-induced complete growth inhibition of HeLa tumors in mice (Fig. 6(d)). Thus, TICT has multiple potentials for mitochondria-targeting, synergistic tumor killing via the dual modes of PTT and PDT, as well as multiple tumor imaging by NIR fluorescence, photoacoustic signals, and temperatures.
In general, photosensitizers have two functionalities, including therapeutic effects and imaging activities. Their theranostic activities enable the design of multifunctional and multipotent organelle-targeting theranostics with their intrinsic subcellular localization. In particular, as previously discussed, some photosensitizers showed mitochondria-targeting activities. Choi et al. attempted to confirm whether one photosensitizer, PhA, can target specifically to mitochondria and evaluated whether the mitochondria-targeting activity of PhA is higher than that of the well-known mitochondria-targeting TPP [37]. To answer these two questions, two different amphiphilic polymers (i.e., TPP-b-poly(ε-caprolactone) (PCL)-b-TPP (TPCL) polymers and PPA polymers) were used to make their mixed micellar structures (i.e., PPA n -TPCL 4-n nanoparticles (NPs)) ( Fig. 7(a)). The fluorescence of free PhA was perfectly overlapped with the green fluorescence of MitoTracker™ Green-stained mitochondria ( Fig. 7(b)). Although the fluorescent intensities of PhA delivered by PPA n -TPCL 4-n NPs were lower than those of free PhA, the mitochondrial distribution of PPA n -TPCL 4-n NPs was confirmed by comparing with the intracellular distribution of MitoTracker™ Green (Fig. 7(b)). However, using confocal images, a colocalization analysis of PhA  [38]; Copyright © 2018 American Chemical Society delivered by PPA n -TPCL 4-n NPs indicated that many PPA n -TPCL 4-n NPs remained entrapped in endolysosomal compartments due to their poor endosomal escaping activities (Fig. 7(c)). Nevertheless, the mitochondria-to-nucleus preferences (MNPs) of PPA n -TPCL 4-n NPs were approximately 3~3.5 and were similar or slightly lower than that of free PhA (Fig. 7(d)). The research showed that PhA can be used as a mitochondria-targeting theranostic and its mitochondria-targeting activity is almost equivalent to that of TPP.
TPP-C NPs were also formed by a self-assembly of TPP-coumarin chemical conjugates in aqueous solutions [36]. The formed mitochondria-targeting, fluorescent TPP-C NPs can load various hydrophobic therapeutic molecules. Interestingly, although approximately 80% of TPP-C NPs were localized in endolysosomes at 30 min of incubation, approximately 80% of TPP-C NPs escaped from the endolysosomes and then reached the mitochondria at 2 h of incubation. In the case of carboxylated PEG-b-polyMTOS, the polymer was chemically linked with an NIR photosensitizer, IR780 with PTT and PDT at the end carboxylic acid of the polymer [39]. A self-assembly of IR780-PEG-b-polyMTOS polymers in aqueous solutions formed phototherapeutic IR-NP. Furthermore, when IR780-PEG-b-polyMTOS polymers were self-assembled in aqueous solutions, additional IR780 molecules were added into the formed nanostructure, IR-NP-eIR. It is unclear whether the phototherapeutic NPs targeted to the mitochondria because the location of mitochondria-targeting polyMTOS in the NPs could be at their hydrophobic core. Nevertheless, IR-NP-eIR generated more heat by NIR irradiation than IR-NP and killed more tumor cells than IR-NP. In addition, after constructing self-assembled nanosystems from amphiphilic materials, organelle-specific targeting moiety and components for therapeutics, diagnostics, or theranostics are introduced into the nanosystems to produce mitochondria-targeting theranostic NPs. Zhang et al. first made upconversion NPs (UCNPs) using amphiphilic poly(maleic anhydride-alt-1-octadecene) (C18PMH)-b-PEG-NH 2 , Reproduced with permission from reference [37]; Copyright © 2017 Elsevier B.V and the mitochondria-targeting TAT peptide was then chemically linked on the end functional amine of UCNPs [40]. The formed TAT-UCNPs physically encapsulated a theranostic PPhA derivative (i.e., Ppa) for PDT-mediated tumor killing and NIR fluorescence imaging.
Second, organic NPs can be prepared by in situ synthesis. For example, a one-step hydrothermal method enabled the three materials of chitosan, ethylenediamine, and mercaptosuccinic acid as carbon sources to form nanosized carbon quantum dots (CDs) [41]. The formed CDs possess intrinsic green fluorescence, which may be used for intracellular tracking, such as a mitochondrial tracker. In particular, when chemically conjugating a photosensitizer, rose bengal (RB) on CDs, the formed CDs-RB can be considered mitochondria-targeting theranostic nanosystems ( Fig. 8(a)). Colocalization studies of CDs with various organelle-trackers (e.g., MitoTracker™, ER-Tracker™, Golgi-Tracker™, and LysoTracker™) showed that most CDs were accumulated in mitochondria; however, many CDs were also distributed in the ER. This less organelle-specificity could be caused by complicate functional groups, such as amine, hydroxyl, thiol, and carboxylic acid, on the surface of CDs. Nevertheless, the mitochondria tracking activity of CDs was activated at a very early incubation time (e.g., 5 min) and remained for 24 h (Fig. 8(c)). For mitochondria-targeting theranostic CDs-RB, intracellular RB delivery with CDs-RB was substantially more effective than free RB (Fig. 8(d)) and caused substantially more tumor killing activities than free RB (Fig.  8(e)). In addition, Xu et al. constructed CDs using citric acid and ethylenediamine (i.e., Cdot) [42]. The Cdot also showed intrinsic fluorescence. However, for mitochondria-targeting activity, TPP was chemically introduced on the end functional group of Cdot. Moreover, nitric oxide (NO)-releasing therapeutic molecules were chemically modified on the surface of Cdot. The resultant Cdot-TPP-SNO was selectively accumulated in mitochondria and then light-induced NO release from the nanosystems killed tumor cells.
Third, noncarbon-based nanostructures have also been applied for mitochondria-targeting theranostic systems. Guo [44]. After forming 2D graphitic phase C 3 N 4 NSs, Fe 3+ was doped on the NSs. The mitochondria-targeting TPP was then chemically introduced into Fe 3+ -doped C 3 N 4 NF, and a photosensitizing methylene blue (MB) was also physically loaded into the NF. The resultant C 3 N 4 -Fe-TPP NF/MB showed substantially more mitochondria-targeting activities and light-irradiated higher cell-killing activities, as well as more tumor growth inhibition than C 3 N 4 -Fe NF/MB and C 3 N 4 NS/MB. Furthermore, the doped Fe 3+ on C 3 N 4 -Fe-TPP NF/MB enabled imaging of tumor areas in mice by T 1 -weighted MRI.

Conclusions and future perspectives
In summary, the current research on mitochondria-targeting theranostic chemicals, chemical conjugates, or nanosystems is substantially increasing with the significance of subcellular organelle specificity and simultaneous modes of therapy and diagnosis. In the case of mitochondria-targeting theranostic chemicals and chemical conjugates, mitochondria-targeting lipophilic cations have been frequently introduced into the chemicals and chemical conjugates, resulting in the synthesis of M-T-D 3-in-1 typed materials. These reasons caused many photosensitizers to be applied to design mitochondria-targeting theranostics, and the use of well-known mitochondria-targeting moieties (Fig. 1) has been limited. In particular, although chemical synthesis using three different M, T, and D components enables various combinations and many new mitochondria-targeting theranostic chemical conjugates, the combination or design-based chemical syntheses have been limited due to the complexity, difficulty, and inconvenience of their syntheses. In addition, although lipophilic cations of mitochondria-targeting moieties could distinguish differences in the mitochondrial membrane potentials between tumor cells and normal cells, in some cases, the resistance of the designed chemicals (e.g., F16) could reduce their applications.
As alternatives of chemicals or chemical conjugates, mitochondria-targeting nanosystems composed of various materials have been considered. The nanosystems could detour the resistance mechanisms of chemicals. Physical loading of therapeutic molecules and diagnostic molecules in nanosystems could extend various combinations of T and D in mitochondria-targeting nanosystems. These platform-based nanosystems could minimize the complexity, difficulty, and inconvenience of chemical synthesis. Nevertheless, the current studies have often employed TPP as a mitochondria-targeting moiety and photosensitizers due to their intrinsic theranostic characteristics. Although TPP has good mitochondria-targeting activity, its positive charge could limit its use because many negatively charged blood components could cause TPP-based theranostic nanosystems to be useless via the formation of nonspecific aggregates. Thus, alternatives to TPP, such as negatively charged or hydrophilic, neutral charged mitochondria-targeting moieties, should be investigated. Furthermore, the lack of an endosomal escaping ability of the nanosystems could reduce their mitochondria-targeting effects because nanosystems, in general, follow endocytic pathways for their cellular entry. Thus, to produce effective mitochondria-targeting nanosystems, the nanosystems should equip endosomolytic functions. In addition, although NIR photosensitizers enable imaging of NIR fluorescence, photoacoustic signals, and photothermal signals, their corresponding diagnostic materials should be considered for other imaging tools, such as MRI, PET, and CT.