Recent progress in nanophotosensitizers for advanced photodynamic therapy of cancer

Owing to their unique photophysical and physicochemical properties, nanoscale photosensitizers (nano-PSs) comprising nanocarriers and molecular photosensitizers (PSs) have emerged as the practical solutions to circumvent current limitations in photodynamic therapy (PDT) of cancer. Nanosized materials have demonstrated their superiority either as the delivery vehicles for PSs to enhance the therapeutic efficacy in selective PDT or as the active participants to improve the energy conversion under a near-infrared light for deep tumour treatment. In this mini-review, we provide an overview of recent progress on nano-PSs for advanced PDT by elaborating three key elements in the photodynamic reaction, i.e. PS, oxygen, and light. Specifically, we discuss the state-of-the-art design of nano-PSs via the following strategies: (a) intracellular PS delivery based on hierarchical modifications, (b) stimuli-responsive nano-PSs targeting the tumour microenvironment, and (c) improved photophysical characteristics of nano-PSs as the energy transducers under deep tissue-penetrating light irradiation. In addition, the utilities of nano-PSs for combinatory therapy or for theragnostic purposes were also discussed. In the end, the current challenges and future perspectives of nano-PSs towards clinical translation were also highlighted along with the concluding remarks.


Introduction
Photodynamic therapy (PDT) is a two-stage treatment that involves the administration of a photosensitizer (PS), followed by the activation with appropriate light irradiation to generate the cytotoxic reactive oxygen species (ROS) for cell killing. Figure 1 illustrates the established mechanism of photodynamic reaction to produce ROS or singlet oxygen ( 1 O 2 ) through both type I and type II pathways [1]. Owing to the intrinsic tumour selectivity of PSs and possible co-localization of illuminating light to the tumour lesions, PDT is considered as a promising tumour intervention modality with minimum long-term side effects [2]. To date, the Food and Drug Administration has approved the use of porfimer sodium (Photofrin ® ) and meta-tetrakis(3-hydroxyphenyl) chlorin (Foscan ® ) as PSs for PDT treatment of oesophageal cancer, non-small cell lung cancer, and precancerous changes of Barrett's oesophagus [3,4].
Typically, the PSs can be small organic molecules with tetrapyrrole structures, such as porphyrins, chlorins, derivatives of phthalocyanines with the incorporation of Zn or Al into the phthalocyanine macrocycle (e.g. zinc phthalocyanines, ZnPc4; chloro aluminium sulphonated phthalocyanine, AlPcS), natural products (e.g. hypericin, riboflavin), or indocyanine dyes (e.g. indocyanine green, ICG) [5,6]. However, the small organic PSs are often hydrophobic in nature with poor solubility, limited retention, rapid clearance, and a relative low molar absorption coefficient with the tissue-penetrating light. Meanwhile, these PSs tend to aggregate under the physiological conditions, significantly reducing the quantum yields of ROS production and photodynamic activity [7,8]. As a consequence, high concentrations of PSs and light dosage must be delivered to the tumours for a sufficient therapeutic efficacy, which may cause unwanted side effects. following one (hν) or two-photon (2hν) excitation and then relax to an excited triplet state ( 3 PS * ). This triplet PS can interact with molecular oxygen and generate ROS in the forms of superoxide anion radicals (O2•-), hydroxyl radicals (•OH), and hydrogen peroxide (H2O2) through type I mechanism or induce singlet oxygen ( 1 O2) generation via energy transfer through type II mechanism.
To this end, substantial efforts have recently been made to develop novel categories of PSs with optimal attributes, such as synthetic phenothiazinium, boron-dipyrromethene (BODIPY), and aggregation-induced emission dyes [5,6,9]. With their compelling advantages, nanoscale PSs (nano-PSs) have emerged as an effective alternative to partially if not fully incorporate three essential elements of photodynamic reaction, i.e. PS, oxygen, and light [10,11]. The research on nanotechnology-assisted PDT continues to evolve quickly, with frequent reports on novel materials and innovative techniques in recent years [10][11][12][13]. Chen et al overviewed various nanomaterial-based platforms for enhanced PDT from a material perspective and provided a clear framework of the up-to-date development of four generations of PSs [14]. One recent comprehensive review by Yang et al, on the other hand, emphasized on the employment of different physicochemical approaches, including light, magnetic fields, microwave, ultrasound, and x-rays, towards the transition of PDT into the medical domain [15].
In this short review, we discussed the recent progress in designing versatile nano-PSs for advanced PDT with a particular focus on the following state-of-the-art approaches: (a) enhanced PS delivery efficacy for selective PDT via intracellular targeting, (b) stimuli-responsive nano-PSs targeting the hypoxia tumour microenvironment, and (c) improved photophysical characteristics of nano-PSs under the deep tissue-penetrating light irradiation. Figure 2 shows some representative nanostructures employed in PDT with demonstrated functional advantages, such as efficient PS delivery, specific PS accumulation in tumours, and near-infrared (NIR) light-excited nano-PSs for deep PDT, etc.

General advantages of nano-PSs for PDT
Nanomaterials have been generally regarded as drug delivery vehicles to provide enhanced aqueous stability and controlled release attributes. Incorporation of hydrophobic PSs into nanomaterials has been extensively explored to increase PS payloads and prevent aggregation. By selecting appropriate nanomaterials, PSs can be protected from direct exposure to external interferences without compromising their photophysical properties [12,16,17]. Various nanocarrier platforms, including liposomes [18,19], micelles [20,21], dendrimers [22,23], polymeric nanoparticles [24,25], mesoporous silica [26,27], gold nanostructures [28][29][30], and other metallic nanomaterials [31], have been used to encapsulate or conjugate PSs. PS loading in the vehicular nanocarriers can be achieved by noncovalent methodologies using self-assembly, interfacial deposition, core-shell entrapment, electrostatic interaction, physical adsorption, or covalent processes via chemical immobilization [13]. The nano-PSs can be further modified using hydrophilic functional groups, such as polyethylene glycol (PEG) [32], hyaluronic acid [33], or cell membrane [34], to achieve a steric exclusion activity ('stealth effect') and to increase circulation time in the body after administration [35].

Conventional targeting strategies of nano-PSs for PDT
Selective delivery of PS molecules to malignant cells has involved the particle design using versatile approaches to enhance tissue penetration, tumour specific accumulation, and cellular internalization of PSs. By virtue of their nanosize, traditional tumour targeting strategies of nanomedicine include enhanced permeability and retention (EPR) effect-based passive targeting and active targeting by grafting ligands onto the nanocarrier surfaces. For passive targeting, nano-PSs can diffuse through the permeable tumour vasculature and impaired lymphatic drainage system, and retain preferentially in the solid tumours [36,37]. Nano-sized carriers with large surface-to-volume ratio can also serve as superior vehicles for long-sustained and selective delivery of PSs to tumour cells by conjugating versatile targeting moieties, such as antibodies [38], proteins [39], peptides [40], folate [41], biotin [42], aptamers [43], and other small molecules, which bind to specific receptors over-expressed on the tumour cells, including cancer-associated receptors, folic acid receptors, epidermal growth factor receptor, and transferrin receptor [44][45][46].

Intracellular targeting nano-PSs for advanced PDT
In contrast to chemotherapy, the key cytotoxic molecule during PDT, i.e. 1 O 2 , has a very short lifetime, which leads to a very limited diffusion distance (radius of action of 1 O 2 is <20 nm) [1,47]. Therefore, not only the distribution and accumulation of PSs in the tumour tissues is crucial, but the localization and penetration of PSs into subcellular compartments is also of great essence for a satisfactory efficacy of PDT. Subcellular organelles, such as mitochondria [48], lysosomes [49], plasma membrane [50], endoplasmic reticulum [51], and nuclei [52], have been regarded as the potential therapeutic targets of PDT.
In our previous review [53], we summarized the recent advances in developing subcellular targeting gold nanomaterials with the capabilities of cellular penetration and endosomal escape via various surface modifications or physical injection approaches. Similarly, hierarchical targeting strategies have been proposed for an efficient nano-PS intracellular delivery system to achieve an idealized scheme, including cell-specific targeting with an internalizable ligand and subcellular targeting signal with endosomolytic activity [54,55]. In table 1, we summarized the representative targeting strategies to specifically deliver nano-PSs to subcellular compartments for enhanced PDT.

Mitochondrial targeting nano-PSs
Mitochondria are crucial regulators of stress responses, apoptosis, and diverse cellular signal transduction pathways. In the case of PDT, mitochondria are known as the sensitive repository for ROS production. It is noted that mitochondria are polarized as the negative electric potential differences between outer and inner membranes with the lipid bilayer structure. Nano-PSs with positive charge may have a high tendency to localize in the mitochondria by potential-driven translocation upon their entry into the cells. Some of clinically approved PSs do show their capability of partial localization to the mitochondria [64], further endeavours are pursued to further enhance mitochondria-accumulation efficiency of PSs, including conjugation of positively charged lipophilic moieties or mitochondrial targeting sequences onto the nanocarriers [65][66][67]. Self-assembled Ru-Pt metallacage Ruthenium Lysosome Amphiphilic polymer [63] For instance, the triphenylphosphonium (TPP)-based mitochondrial targeting strategy was employed by Li group to deliver TiO 2 -coated upconversion nanoparticles (UCNPs) [56] and chlorin e6 (Ce6)-conjugated-mesoporous silica nanoparticles [57] to the mitochondria, where activated TiO 2 and Ce6 could boost local ROS generation, leading to mitochondrial collapse and irreversible cell apoptosis.
In our previous study [58], we also conjugated the mitochondria-targeting TPP groups with quaternary ammonium cations onto the gold nanoparticle surface for selective delivery of 5-aminolevulinic acid, the precursor of natural PS of protoporphyrin IX (PpIX), to the mitochondria of human breast cancer cells. The modified nano-PSs released from the endocytic vesicles, trafficked to the mitochondria, and elevated ROS formation upon light irradiation, leading to significantly enhanced therapeutic efficacy of PDT. Following a hierarchical targeting strategy, Song et al designed the mesoporous Au@Pt-PEG-Ce6 NPs structures for selective delivery of Ce6 to MCF-7 cells in combination of PDT with photothermal therapy (PTT) [59]. The Au@Pt nanoparticles were labelled with folic acid for cancer cell internalization, and modified with TPP for mitochondria-targeting. Significant improvement of the cell killing efficacy was seen as a result of enhanced cellular uptake, effective mitochondrial ROS burst, and thermal destruction. The cell viability was 46.2% after 150 s irradiation using an 808 nm laser (1.2 W cm −2 ), and then significantly decreased to 13% upon a combination of 660/808 nm laser irradiation.

Nucleus targeting nano-PSs
Cell nucleus is the central regulator of cell proliferation, metabolism, gene activation, and cell cycle management; therefore, it is also a desirable target of various therapeutics, including PDT. It is anticipated that intra-nuclear transport of PSs can increase the photocytotoxicity due to ROS generation in close vicinity to the nucleus for enhanced DNA damage [52]. Cell-penetrating peptides (CPPs) or nuclear localization sequence (NLS) with nucleus homing capabilities are often used to modify the drug delivery system for translocation into the nucleus [68]. Typical CPPs include trans-activating transcriptional activator (TAT) and arginylglycylaspartic acid (RGD) peptide. NLS consisting of a chain of positively charged amino acids can interact with the integral membrane protein family receptors on the nucleus and trigger the receptor-mediated nuclear importing [69][70][71].
Shi group has reported a series of studies designing cancer cell nucleus-targeting nanocomposites for advanced tumour therapeutics [52]. In one study, mesoporous silica nanoparticles co-conjugated with nucleus-targeting sequence TAT and RGD were used as carriers to deliver PS Ce6 to the nuclei of HeLa cancer xenografts, resulting in oxidative damage of the DNA helix under an extremely low light dose [60]. Less than 40% cancerous cells survived after PDT treatment under as low as 5 mW cm −2 irradiation for 5 min. For the hierarchy targeting, Han et al constructed the nanocomposites consisted of pH-responsive 2,3-dimethylmaleic anhydride (DMA) conjugated with NLS peptide and loaded with alkylated PpIX as the PS. The nano-PSs underwent the charge reversal in response to the acidic tumour microenvironment and penetrated the tumour cells based on the electrostatic interactions. The NLS sequence further guided PSs to the cell nucleus, thus achieving improved and selective PDT [37].

Lysosome targeting nano-PSs
Lysosomes are involved in important cellular processes via mediation of the macromolecule degradation. Lysosomes also play an essential role in intracellular trafficking of nanomaterials via endocytosis, and are considered as the major compartments for nanomaterial accumulation [72]. Although lysosomes are much less preferred as the intracellular targets of photooxidation in comparison to mitochondria and nucleus, the significant acid interior (pH 4.5-5.0) of lysosome could serve as a potential site for pH-responsive PSs [73]. In one example, Hu et al incorporated the acid-sensitive dimethylaminophenyl group in the NIR-absorptive BODIPY PS core and then encapsulated within the amphiphilic DSPE-mPEG5000 via precipitation, affording the water-soluble nano-PSs. Coupling with the high-selectivity enabled by acidic lysosomes, the viability of A549 cancer cells incubated with lysosome-targeting BODIPY NPs was dramatically reduced after acid-activatable PDT under the NIR light [61]. As shown in figure 3, Xiang et al developed a nanoplatform composed of a ruthenium nitrosyl donor (Lyso-Ru-NO) and the carbon-doped titanium dioxide nanoparticles for cancer cell and lysosome dual-targeting. Incorporation with folic acid and morpholine-directing groups rendered its capability of targeting folate-receptor overexpressing-cancer cells and specific accumulation in the subcellular lysosomal organelles, where NO and ROS were simultaneously released upon 808 nm light irradiation. Such a lysosome-targeting nanoplatform showed the highest anticancer efficacy compared to the nontargeted counterparts under NIR light sensitization [62].
Due to the close relationship between lysosomes and apoptosis/necrosis, lysosome-localized PSs are also more effective in directly inducing photodamage on lysosomes and lead to subsequential cell death. Zhou et al found that the self-assembled Ru-Pt metallacage encapsulated within an amphiphilic polymer selectively accumulated in the lysosomes and the production of 1 O 2 within lysosomes induced lysosomal disruption under two-photon NIR light irradiation, which resulted in a high phototoxicity to tumour cells [63].

Tumour-microenvironment targeting nano-PSs
Compared to healthy tissues, it is well established that the tumour microenvironment has unique physiological characteristics such as hypoxia, acidosis, vascular abnormalities, and up-regulation of certain enzymes [74].
The strategy of targeting the corresponding aspects of tumour microenvironment would offer much broader application potentials to better address the tumour heterogeneity than merely targeting tumour-specific receptors. Increasing attention has been given to exploit the stimuli-triggered activation of nano-PSs in response to endogenous hypoxia and acidic pH once they extravasate into the tumour microenvironment [75]. We listed several representative nano-PSs (table 2) that achieved the enhanced PDT antitumor efficiency by targeting the hypoxia or/and acid tumour microenvironment.

Hypoxia targeting nano-PSs
Scant oxygen supply has been recognized as one of the hostile hallmarks of solid tumours from the increased oxygen consumption during the metabolic activities of proliferating carcinoma cells [78]. In the meantime, PDT is an oxygen-dependent process where ROS production greatly relies on the availability of oxygen. Thus, hypoxia in the tumour microenvironment is closely related to the increased metastasis, poor prognosis, and resistance to chemotherapy, and insufficient oxygen in the tumour tissues would also impair the PDT efficacy. Moreover, along with the progress of PDT, the hypoxic tumour microenvironment is further exacerbated by continuous oxygen depletion and vice versa. Extensive attempts have been made to address the tumour hypoxia for better PDT, including oxygenation of tumour tissues with various oxygen-replenishing approaches or amplifying the hypoxia conditions to achieve antitumor outcomes through cancer starvation [82]. For the first approach, nanocarriers have demonstrated their capability of alleviating tumour hypoxia and boosting PDT via delivery of a high concentration of molecular O 2 or in situ O 2 generation by catalysing the decomposition of endogenous hydrogen peroxide (H 2 O 2 ) [83]. Qian group reviewed the recent trend in modulation and utilization of tumour hypoxia via nanomedicine-based strategies to improve PDT [82]. Yang et al also summarized the design and application guideline of various H 2 O 2 -responsive nanomaterials with oxygen-self-generation capacity to target the hypoxia tumour microenvironment for enhanced PDT [84]. For instance, Liu et al designed a nanocomposite (O 2 @DANPCe6+PFOB) to alleviate tumour hypoxia by incorporating perfluorooctyl moieties (perfluorooctyl bromide, PFOB) as O 2 carriers and to enhance the tumour cell accumulation by modifying with the TAT peptide (figure 4). The surface corona group of 2,3-dimethylmaleic anhydride (DMA) protected the CPP via an acid-labile amide bond to prolong the circulation time while achieving acid-activated tumour penetration. This O 2 self-supplemented nanoplatform co-loaded with Ce6 exhibited a potent inhibition of tumour growth compared to traditional PDT as observed in both in vitro cytotoxicity and in vivo anti-tumour study on 4T1 tumour-bearing mice after intravenous injection and 10 min 660 nm laser (0.5 W cm −2 ) irradiation [76]. Ping et al reported a design of nano-PSs composed of a hybrid perfluorosiloxane-polystyrene particle core doped with a fluorinated PS and a biocompatible poly-L-lysine shell [77].
Because of the O 2 -carrying capability of intra-particle 'F-C' bonds, the fluorinated nano-PSs saturated with O 2 exhibited approximately 3.5 folds more 1 O 2 production yield and a higher in vitro PDT efficiency. Manganese dioxide (MnO 2 ) and Mn 3 O 4 nanoparticles could decompose H 2 O 2 and sustainably generate oxygen under H 2 O 2 -rich physiological conditions. In a recent study reported by Yang et al, a Mn 3 O 4 @ MSNs@IR780 nanocomposite was prepared to scavenge the hypoxic tumour microenvironment by self-generating oxygen [78]. The PS IR780 was absorbed into the mesoporous silica nanoparticles (∼90 nm in diameter) and then the surface pores were capped with 5 nm Mn 3 O 4 nanoparticles, which could catalyse H 2 O 2 into O 2 and thus ameliorate the hypoxic microenvironment in an MKN-45P tumour xenograft model. The release of IR780 specifically destructed the respiration of the mitochondria due to its natural mitochondrial affinity and efficiently killed MKN-45P cancer cells upon 808-nm laser irradiation (1 W cm −2 , 5 min) in both in vitro and in vivo conditions ( figure 5).
Recently, we also developed a bimetallic and biphasic rhodium (Rh) and Au porous core-shell nanosystem (Au@Rh-ICG-CM) to address tumour hypoxia while achieving bimodal imaging-guided high PDT efficacy (figure 6) [31]. Such porous Au@Rh core-shell nanostructures exhibited catalase-like activity to . Schematic illustration of O2@DANPCe6+ PFOB mediated oxygen-replenishing PDT. PFOB as O2 carriers to oxygenate hypoxia tumour tissue during PDT. The protective DA corona will detach from the particle surface at the tumoural acidic microenvironment and the TAT peptide will facilitate its tumour-penetrating ability, leading to an enhanced therapeutic efficacy in both 4T1 cells in vitro and 4T1 tumour-bearing mice in vivo. Adapted with permission from [76]. Copyright 2019 The Royal Society of Chemistry.  Instead of harnessing tumour hypoxia, the combination of PDT with other hypoxia-responsive chemical moieties or cancer starvation therapeutic agents can collaboratively achieve the synergistic anticancer efficiency. For example, PDT-induced hypoxia would fully activate the hypoxia-responsive prodrug and severely damage the tumour blood vessels in a PDT-aggravated hypoxic condition. By incorporating the hypoxia-sensitive nitroimidazole molecules, Qian et al encapsulated the widely-used chemotherapeutic drug doxorubicin (DOX) in 2-nitroimidazole-grafted conjugated polymer to achieve a synergistic treatment with PDT and chemotherapy [85]. The nanocarrier complex could generate ROS after light-triggered stimulus and subsequently induce hypoxia to release DOX upon nanocomposite disassembly, which provided an innovative design guideline for drug delivery using programmed stimuli-responsive triggers. Noted that tirapazamine (TPZ), an aromatic N-oxide, has 300-fold higher toxicity under anoxic conditions than aerobic conditions, Liu et al designed an Hf-porphyrin metal-organic framework platform with high porphyrin and TPZ loading capacity [79]. Depletion of oxygen during PDT aggravated the hypoxic environment of tumours and further activated TPZ to enhance the treatment efficacy.

pH-responsive nano-PSs
As tumour cells rely heavily on the glycolysis for energy consumption rather than the oxidative phosphorylation, leading to a higher level of extracellular lactic acid, thus, the pH of the tumour microenvironment is more acidic (6.5-6.8) than that of healthy tissues. pH-responsive PS delivery systems have also been investigated for PDT. These nanostructures are usually grafted with reactive linkages, which are cleavable upon pH stimuli and then undergo PS release. Li et al fabricated a tumour-pH-responsive supramolecular PS structure based on the electrostatic interaction between negatively charged octasulfonate-modified zinc(II) phthalocyanine (ZnPcS8) and cationic layers of double hydroxide, which could be efficiently activated in the acidic tumour microenvironment (pH 6.5), resulting in 95.3% tumour growth inhibition with minimal skin phototoxicity [80]. Hu et al synthesized the pH-sensitive nanoparticles of acetylated β-cyclodextrin (Ac-β-CD) using oil-in-water emulsion technique and then modified with gelation-folic acid ester for cancer targeting [81]. This nanodevice was further loaded with an anticancer drug camptothecin and a PS phthalocyanine (PcZn) to improve the synergistic outcomes of chemo-PDT.

Nano-PSs for deep PDT
Current PS molecules (e.g. chlorin, ZnPc4, porphyrin, and texaphyrin) generally require ultraviolet (UV) or visible light for efficient activation upon one-photon excitation. Due to light scattering and absorbance by biological tissues, UV or visible light of high energy with limited tissue penetration hardly reaches deep tumour sites or inner cores of large solid tumours, which remains the major obstacle for current PDT. Figure 7(a) shows the tissue penetration depth of light with varying wavelengths, where NIR (650-1350 nm) is a favourable range for in vivo applications with efficient luminous flux yet minimal auto-fluorescence in its spectral region. Some PSs such as ICG (810 nm) and aluminium sulfophthalocyanine (790 nm) can absorb NIR light with deeper tissue penetration but are considered less effective than traditional PSs considering the low yield of the triplet state upon irradiation. In this regard, upconversion (UC), two-photon activation, and self-illumination-based strategies have been developed with the advances of nanotechnology to circumvent the limitations of current PDT and expand its applicability for deep tumour treatment [86].

Upconversion PDT
Photon UC is known as a nonlinear optical phenomenon that can convert a lower energy excitation to a higher energy emission through an anti-Stokes process. Nanoparticles with controllable upconverted emission properties can absorb more than two low-energy photons of the NIR region in a cascade manner and then efficiently convert them into a higher-energy photon in the visible or UV region. Figure 7(b) illustrated the simplified process during photon UC for PS activation based on the energy transfer mechanism. UCNPs can be used as energy donors to activate PS for cytotoxic ROS production through electronic excitation based on either radiative (luminescence) or non-radiative (Föster resonance energy transfer, FRET) energy transfer. The primary mechanism of FRET-based energy transfer for PS activation was shown in figure 7(c).
As photon UC typically occurs between two different types of rare-earth ions, UCNPs are usually comprised of an appropriate host lattice, such as fluorides NaYF4, NaGdF4, NaLuF4, and BaYF5, doped with transition metal, actinide or lanthanide ions (Yb3+/Er3+/Tm3+) [87]. So far, many PSs, such as porphyrin derivatives, ZnPc4, Ce6, methylene blue, and rhodamine B, have been used in a synergistic combination with UCNPs to augment efficient 1 O 2 generation under NIR light irradiation. Hamblin recently summarized up-to-date research works of UCNPs for the use in PDT with a wide range of different PSs [88]. For efficient energy transfer, the UCNP emission wavelength should match the absorption of entrapped PS, and the PS should be in close proximity to the luminous inner core of UCNPs. The entrapment of PSs in UCNPs can be achieved through covalent conjugation, physical adsorption, or silica encapsulation, dependent on the hydrophobicity of PSs. Similar to the above-mentioned nano-PSs, UCNPs can be modified with hydrophilic supports and further functionalized with targeting moieties to enhance their selective accumulation at the tumour site.
Gnanasammandhan et al employed the mesoporous-silica-coated UCNPs (NaYF4:Yb3+/Er3+) to convert deeply penetrating NIR light at 980 nm to activate two PSs (MC-540 and ZnPc) simultaneously for enhanced PDT [89]. In vivo studies showed the promising tumour growth inhibition in PDT-treated mice upon intratumor injection or intravenous injection. However, while the porous structures are beneficial for oxygen and ROS diffusion, leakage of PSs during systemic circulation may lead to inadequate dosing of PSs in the targeted site. Yuan et al conjugated DOX with PEGylated polyelectrolyte through a UV-cleavable ortho-nitrobenzyl linker and used this structure as the matrix for UCNP (NaYF4:Yb3+/Tm3+) encapsulation [90]. This NIR light-regulated UCNP@CPE-DOX structure could photocleave the linker for controlled release of DOX upon 980 nm laser irradiation and activate Ce6 to produce ROS at a high efficiency.
Ongoing researches have focused on further enhancing the UC quantum yield of UCNPs and optimizing FRET efficiency between PSs and UCNPs based on intensive investigations on the underlying mechanism. In particular, the amount of PS loading and the distance control between UCNP core and attached PS molecules should be optimized when designing the UCNP-PS system for maximum PDT effect.

Two-photon PDT
Two-photon excitation takes place with simultaneous absorption of two photons. Similar to the UC process, two-photon PDT is characterized by nonlinear absorption of two low-energy photons of NIR light with the resulting emission of high-energy visible light to activate PSs [91]. The main difference between UC process and two-photon-induced PS excitation is that the light frequency conversion efficiency during UC is orders of magnitude higher than that of a nonlinear two-photon absorption mechanism. The two-photon absorption process requires excitation at a high density of ∼10 6 W cm −2 by an ultra-short pulsed (e.g. femtosecond) laser, while NIR light UC can be achieved with an excitation density of 10-100 W cm −2 provided by a low energy continuous-wave diode laser [92].
To increase the efficiency of two-photon PDT, large two-photon cross-sections are desirable, which have been obtained with symmetrically or dissymmetrically π-extended porphyrins, and supramolecular assemblies [93]. Different nanosystems have been designed to expand the two-photon cross-sections of PSs based on the FRET mechanism [94].
Duan et al conjugated two-photon light-harvesting hydrophobic polymer to tetraphenylporphyrin (TPP) and a red-emitting imaging dye (TPD) and found that the two-photon emission of tetraphenylporphyrin and TPD was enhanced up to ∼161 and ∼23 times, respectively [95]. These nanostructures were further modified with folic acid groups and remarkably improved the PDT efficiency up to about 149 times with selective targeting of cancer. In a recent study, Karges et al reported the design of Ru(II) polypyridine complexes with (E,E ′ )-4,4 ′ -bisstyryl-2,2 ′ -bipyridine ligands and a red-shifted one photon and strong two-photon absorption using an in silico optimization [96]. In vivo studies confirmed that the compounds could eradicate a multi-resistant tumour on a mouse model upon clinically relevant one-photon (500 nm) and two-photon (800 nm) excitation.
The relatively low two-photon absorption efficiency and low 1 O 2 quantum yield remain as the major concerns of conventional PSs for two-photon PDT. Clearly, materials exhibiting a long triplet-exciton emission are more favourable to extend the energy transfer to surrounding O 2 for enhanced generation of cytotoxic 1 O 2 . Recently, various strategies have been explored to achieve a longer excited triplet state lifetime and to increase the efficiency of light energy transfer in the design of photosensitive materials [15,97]. For example, the long-lived room temperature phosphorescence materials have attracted increasing attention for PDT applications [98]. Yan group confirmed that space-confined and interface-confined microenvironments of 2D matrix would facilitate ultralong-lived triplet exciton with second-scale lifetimes, which was significantly longer than that of conventional PS in microsecond to millisecond range [99]. As shown in figure 8, they also developed a nanohybrid system through the assembly of different aromatic acids within the layered double hydroxide nanosheets for two-photon-induced 1 O 2 generation. The quantum yield of 1 O 2 was greater than most as-reported PSs to date and led to dramatical tumour ablation in vivo under the 808 nm laser irradiation [100].

Self-illuminated PDT
Luminescent materials (or phosphors) or luminescent proteins (such as luciferase and horseradish peroxidase) can emit chemiluminescence or bioluminescence after an external excitation source is removed. The self-illuminating luminescent materials can serve as an internal light source for flourishing PDT in deep tissue treatment, and their afterglow properties are particularly attractive for continuous PDT, since they can continuously excite the nearby PSs even after the stoppage of excitation. Persistent luminescence nanoparticles are a newly emerging class of functional optical biomaterials in PDT. Fritzen et al recently reviewed the up-to-date literatures on their applications in luminescence imaging and PDT [101].  . Schematic illustration of self-assembly of the luminol and Ce6 conjugate into a core-shell nanoparticle for in vivo luminescence imaging of the colitis development in mice and for in vivo antitumor PDT. Through the bioluminescence resonance energy transfer, the luminescence originating from luminol upon triggering can excite Ce6 to produce fluorescence and 1 O2. Adapted from [105]. CC BY 4.0.
In one example reported by Wang et al, they fabricated the phosphorescent Cr3+, Yb3+, Er3+ triple-doped zinc gallogermanate nanostructures, which were coated with mesoporous silica for the loading of AlPcS. The as-prepared nanoparticles exhibited long-lasting persistent luminescence and remarkable afterglow properties under fractionated light irradiation which triggered the generation of a large amount of 1 O 2 and enhanced tumour destruction [102].
In another recent review, major types of persistent luminescence nanoparticles and their multiple biomedical applications were summarized, in which controllable synthesis methods were considered as the current technical challenges to achieve the optimal optical properties [103]. In recognition of the decrease of luminescence performance and low utilization efficiency of traditional persistent luminescence materials, Sun et al recently tried to prevent the mass loss and maintained the intact crystal lattices of persistent luminescence materials by homogeneously dispersing high temperature-sintered persistent luminescence materials into the alginate-Ca 2+ hydrogel. Owing to the retained NIR luminescence and red-light renewability, the hydrogel composites could sensitize continuous generation of 1 O 2 after injection into the tumour site, favouring repeated PDT [104].
Luminol is another extensively studied luminescent donor. As shown in figure 9, Xu et al developed an amphiphilic polymer conjugated with luminol and Ce6 that could self-assemble into a nanoparticle for in vivo luminescence imaging and PDT in deep tissues. Higher level of ROS and myeloperoxidase generated  [120] in the inflammatory sites or the tumour microenvironment can trigger the bioluminescence resonance energy transfer and the production of 1 O 2 from the nanoparticle, enabling in vivo imaging and antitumor activity [105]. Jiang et al constructed an in-situ luminescing and O 2 -supplying system for efficient PDT based on luminol-H 2 O 2 -Hemoglobin (Hb). Hb conjugated polymer nanoparticles simultaneously catalysed the chemiluminescence and internal activation of luminol/H 2 O 2 and served as the carrier for delivering sufficient molecular oxygen, which led to more ROS generation and enhanced cytotoxicity in PDT under a hypoxia condition [106].

Multifunctional nano-PSs for combinational therapies and theragnostic PDT
Taking advantage of the unique physiochemical properties of nanomaterials, additional functionalities can be incorporated in the design of nano-PSs for either multidrug delivery involved in combination therapy or synergetic imaging for diagnostic analysis and imaging-guided therapy. Some representative multifunctional nano-PSs used in combinational or theragnostic PDT were summarized in table 3. With the ease of co-encapsulating of multiple agents, more than one PS can be loaded in the nanocarriers to achieve significantly improved synergistic anticancer effects. For example, Lee et al loaded two PSs, Ce6 and Rose Bengal, to UCNPs with a core@shell structure (NaYF4:Yb,Er,Nd@NaYF4:Yb,Nd) and found the dual PS system showed a synergistic ROS generation, which is significantly higher than that of a single PS system [107]. As shown in figure 10, Chang et al constructed the core/shell UCNPs-PS modified with poly(allylamine) for intracellular targeting and dual-loaded with two PSs, i.e. Rose Bengal and Zinc(II) phthalocyanine, to achieve better therapeutic PDT effects on A549 cells [108]. Besides the PS molecules, other anticancer therapeutics can also be co-delivered to the targeted tumour sites, and noted synergistic therapeutic outcomes have been achieved upon the combination with chemotherapy, PTT [109,110], ionizing radiation [111], or gene therapy [112].
In the case of combining PDT with chemotherapy, PDT-induced ROS can suppress the active efflux translocator and consequently inhibit the efflux of chemotherapeutics. As such, PDT and chemotherapy drugs can mutually promote the therapeutic efficacy on a synergy basis. DOX [113,121], paclitaxel [114], mitomycin [122], and cisplatin [123] have been often combined with PSs to develop the integrated Figure 10. Construction of core/shell UCNP nanoplatform co-loaded with Rose Bengal-ZnPc, which led to NIR-triggered PDT to induce mitochondrial damage as reflected with the loss of mitochondrial membrane potential and the decrease of cell viability in A549 cells. Adapted from [108]. CC BY 4.0. nanosystems. In one example, Tian et al designed α-CD modified red-emitting UCNPs to co-deliver both Ce6 and DOX for combined deep PDT/chemotherapy under 980-nm laser irradiation, where a much higher efficiency was found as compared to individual ones [115]. Liu et al recently reviewed the UC-based PDT where additive benefits and profound therapeutic effects were highlighted as a result of the integration of UCNPs with other traditional therapies [124].
The in vivo tumoricidal effect of PDT is not only capable of eliminating primary tumours but also has demonstrated its potency of sensitizing the immune system to destroy metastasis and prevent tumour relapse. We have recently reviewed the PDT facilitated-antitumor immunity based on various immunotherapeutic approaches in concert with PDT, among which nano-PSs became a powerful means for PDT-induced immunity via tuneable delivery of multiple immunostimulatory agents [125]. Xu et al designed a UCNP-based platform by co-loading Ce6 and imiquimod (R837), a Toll-like-receptor-7 agonist as an immune adjuvant, onto the polymer-coated UCNPs for NIR-assisted deep PDT of colorectal cancer [116]. In addition to photodynamic destruction of primary tumours upon 980-nm light irradiation, the nano-PSs were able to trigger the maturation of DCs and subsequent secretion of cytokines which significantly strengthened the antitumor immune responses.
The cytotoxic T-lymphocyte-associated protein 4 blockade could further inhibit the activities of T reg cells and destroy distant tumours that is normally hard to reach with the light for PDT.
In addition, as part of the inflammatory responses, localized damage of tumour vasculature is another therapeutic effect of PDT. The extent of PDT-triggered vascular damage can be further exacerbated by synergistically combining with inhibitors against angiogenesis factors such as vascular endothelial growth factor (VEGF), tumour necrosis factor (TNF-α), and interleukin-1β. Lecaros et al reported their study of combining PDT with targeted VEGF-A gene therapy to achieve an enhanced therapeutic outcome for human head and neck squamous cell carcinoma (HNSCC) [117]. Lipid-calcium-phosphate nanoparticles were used to deliver VEGF-A small interfering RNA to significantly decrease the expression of VEGF-A, which is overexpressed under the hypoxic conditions during PDT. The combination of VEGF-A siRNA gene therapy and photosan-mediated PDT did result in a significant antiangiogenic effect by silencing the angiogenic markers and led to the tumour growth rate decrease of ∼70 and ∼120% in comparison to untreated group in subcutaneous human HNSCC xenograft models.
Aside from being as transporting vehicles of PSs or as energy transducers in photodynamic reaction, nanomaterials have also endowed additional capabilities in imaging for diagnoses. Owing to the intrinsic fluorescence of most PSs, the biodistribution of nano-PSs is traceable using the in vivo whole-body fluorescence imaging system [36]. Other nanomaterials such as magnetic nanoparticles and quantum dots by themselves are well-established nanoplatforms used for optical imaging and magnetic resonance imaging (MRI) as contrast agents, and have been continuously adopted for PS delivery in PDT with concurrent diagnosis abilities [126,127]. Lamch et al scrutinized the recently engineered multifunctional colloidal nanoparticles for enhanced PDT and bioimaging, with an emphasis on the design principles and the comparison of bio-performance of various nano-PS platforms [35].
In one example, Li et al developed PEGylated iron oxide nanoclusters (IONCs) to load the PS molecule of Ce6 for PDT with demonstrated efficacy in vitro and in vivo [32]. Meanwhile, taking advantage of the strong magnetic resonance signals attributed to IONCs and intra-tumoural fluorescence of Ce6, the IONCs-Ce6 could also be used for fluorescence-MRI dual-mode imaging of cancer while achieving the magnetic-guided PDT by an external magnetic field. Kim et al designed Ce6-loaded and manganese ferrite nanoparticle (MnFe 2 O 4 )-anchored mesoporous silica nanoparticles (MFMSNs) for enhancing the therapeutic efficiency of PDT by relieving the hypoxic conditions [118]. Besides continuous supply of oxygen through the catalytic Fenton reaction to improve the PDT effects, MnFe 2 O 4 nanoparticles also served as the T2-contrast agent for in vivo MRI tracking. After intravenous injection, Ce6-loaded MFMSNs were selectively retained at the tumour sites owing to the EPR effect. Lin et al reported the synthesis of two-dimensional tellurium nanosheets, which could generate ROS and exhibit high imaging performance for multispectral optoacoustic tomography due to the strong NIR absorbance [119]. Clearly, this material can be further engineered as a nanoplatform for photoacoustic imaging-guided PDT. Wu et al fabricated Ce6-loaded Au/Ag-MnO 2 hollow nanospheres with multifunction of endogenous oxygen generation for enhanced PDT, remarkable photothermal conversion in the NIR window for deep PTT, and triple-modal (fluorescence/photoacoustic/magnetic resonance) imaging for diagnosis [120].

Future perspectives and concluding remarks
The emerging nano-PSs will not only address the current impediments in PDT by significantly improving the overall PS pharmacokinetics and guiding tumour-specific accumulation but also provide alternatives for deep PDT as energy donors under NIR to increase the treatment depth. However, with similar obstacles to other nanomedicine, nano-PSs also face great challenges in clinical translations. Prior to successful clinical utility, the in vivo performance of nano-PSs via different administration routes, including the blood retention, tissue penetration capability, and their possible interactions with serum proteins and immune system, must be thoroughly investigated. Despite the increasing efforts in unravelling the systemic response, biodistribution and pharmacokinetics of various particulates in nano scale, limited information has been gathered regarding the long-term effects of nanoparticles inside the human body.
Many of current researches on nano-PSs still primarily focus on the innovative design of nanostructures. During the fabrication process of multifunctional nano-PSs, nanomaterials normally need to undergo a complicated synthesis route that requires multiple reagents, followed by a series of surface modification, purification, chemical extraction, and centrifugation steps. This could be further complexed with the involvement of more than one PSs, other therapeutic agents, targeting ligands or functional motifs for combinational therapy, intracellular delivery, and/or theragnostic functions. The large variation in experimental setup and the lack of standardization of methodologies would bring in another dimension of challenges in reproducibility, scalability, and quality control, which hamper the translation of research findings to bedside.
Moreover, during a typical fabrication process of nano-NPs, covalent conjugation tends to lower the quantum yields of PSs, and the involvement of hydrophilic coatings may induce PS aggregation and diminish the ROS generation efficiency. In this regard, it is always noteworthy that rational functionalization of nanocarriers should not compromise the photophysical properties of PSs during the encapsulation of PSs in various nanostructures. It remains elusive to determine the optimal cocktails of material composition and ratios of each functional motifs across or/and within various modalities. Collectively, the delicate balance of multiple factors in the design of nano-PSs has thus been a major bottleneck, resulting in the current nano-PSs being stuck in the research phase.
Furthermore, more in-depth mechanistic understanding of the photophysical properties of nano-PSs upon different physicochemical stimulations becomes essential, accompanying with substantial improvement in analytical technologies for localized molecular imaging, and intracellular 1 O 2 tracing without additional probes.
In addition, in order to bring nano-PSs closer to clinical PDT application, it is always essential to perform preclinical tests to confirm the biosafety and the in vivo therapeutic index, and to verify the cancer-targeting capability based on experimental models that can closely mimic human responses. Tumour xenograft models are generally considered as necessary approaches to study the systemic effects of nano-PSs; however, due to the inherent genetic and immunological differences and large individual variations, murine models can barely represent the performance of nano-PSs in the human physiological systems. Moreover, as the administration route is also closely related to the systemic outcomes, it will be necessary to monitor the dynamic interactions between tumour cells and nano-PSs in real time and to determine the fate of nanoparticles through extended observation of their biodistribution upon different administrations. In this regard, the recently established tumour-on-a-chip platforms could be beneficial for high throughput evaluation of PDT efficacy by providing physiologically relevant microenvironments with controlled mechanical and chemical cues [128,129]. Despite the need of a consensus on in vivo performance of nano-PSs, we still believe that engineered nano-PSs offer an innovative horizon toward future development of cancer therapeutics, and endow current PDT with new opportunities for broader clinical implications.