Molecular design of upconversion nanoparticles for gene delivery.

Due to their large anti-Stokes shifts, sharp emission spectra and long excited-state lifetimes, upconversion nanoparticles (UCNPs) have attracted an increasing amount of research interests, and have shown great potential for enhancing the practical utility of gene therapy, whose versatility has been limited by existing gene delivery technologies that are basically mono-functional in nature. Despite this, up to now in-depth analysis of the development of UCNPs for gene delivery has been scant in the literature, even though there has been an upsurge of reviews on the chemistry of UCNPs and their applications in bioimaging and drug delivery. To fill this gap, this review aims to present the latest advances in the development and applications of UCNPs as gene carriers. Prior to describing the prominent works published in the field, a critical view on the properties, chemistry and molecular design of UCNPs for gene delivery is provided. With a synopsis of the recent advances in UCNP-mediated gene delivery, challenges and opportunities could be illuminated for clinical translation of works in this nascent field of research.

Introduction

Current research in gene delivery has reached a bottle-neck in terms of efficiency and versatility. Hopes to change this situation may now be brought about by the emergence of photobiology, which not only enables imaging during gene therapy but also makes the precise control of the timing and location of the release of the loaded gene possible. To achieve this goal, over the years, different optically-active materials have been developed, ranging from quantum dots (QDs) to luminescent transition pan class="Chemical">metal complexes. Each of them has their own merits and drawbacks in applications (Table 1).[1-10] Among them, upconversion nanopaspan>rticles (UCNPs) have attracted extensive research interests and represent a hot topic in materials chemistry. Compared to the conventional down-conversion fluorophores, UCNPs have several optical properties favourable for biomedical use, including negligible photobleaching and photoblinking, lower interference from auto-fluorescence from surrounding tissues, higher spatial resolution, and less photodamage caused by the excitation light to fragile biological molecules.

Table 1

A comparison of major classes of luminescent materials used in biomedical applications

Type	Strengths	Drawbacks	Examples	Ref.
Transition metal complexes	• Good aqueous solubility	• High toxicity	Cationic iridium(iii) complexes, which can emit green and red light, have been reported as phosphorescent dyes for live cell imaging	1
	• Large Stokes shifts	• Interference from auto-fluorescence from surrounding tissues	Luminescent cyclometalated iridium(iii)polypyridine indole complexes have been synthesized, and have been found to emit intense and long-lived luminescence upon photoexcitation in fluid solutions at 298 K or in alcohol glass at 77 K	2
	• Absence of dye–dye interactions	• Attenuation of imaging signals during deep tissue imaging
Gold nanoparticles	• Good biocompatibility	• Low contrast	Multi-branched gold nanoparticles have been fabricated by reducing tetrachloroauric acid with Tris base, and have been adopted as a substrate for imaging kidney cells based on surface-enhanced Raman scattering (SERS)	3
	• Low toxicity	• Attenuation of imaging signals during deep tissue imaging	Ru(ii)-polypyridyl surface-functionalised gold nanoparticles have been reported as an imaging probe that shows targeting capacity towards DNA molecules	4
Quantum dots (QDs)	• Narrow emission bands	• Attenuation of imaging signals during deep tissue imaging	Near infrared (NIR) QDs have been designed. Their applications in monitoring changes in the Cu2+ concentration through in vitro and in vivo fluorescent imaging have been reported	5
	• Tuneable emission properties	• High toxicity	A polysaccharide-QD conjugate has been adopted to generate supramolecular nanoparticles for imaging cancer cells	6
Organic fluorophores	• High quantum efficiency	• Interference from auto-fluorescence from surrounding tissues	Halo tag-based target-specific azidos have been fabricated as photoactivatable organic fluorophores for super-resolution imaging of target proteins in fixed and living cells	7
		• Photobleaching	Resveratrone glucoside has been synthesized from resveratrol-3-β-mono-d-glucoside via photoreactions. The compound has been reported to display a high fluorescence quantum yield, a large Stokes shift, and a large two-photon absorption cross-section	8
		• Photoblinking
UCNPs	• Good biocompatibility	• Low extinction coefficient	NaYF4:Yb,Er UCNPs with 6-phosphate-6-deoxy-β-cyclodextrin as the surface ligand have been generated. Cyclic RGD-conjugated adamantine has been incorporated into the UCNP surface for targeted cellular imaging	9
	• Large anti-Stokes shifts	• Low quantum yield	Adamantaneacetic acid-capped UCNPs have complexed with β-cyclodextrin. The nanoparticles generated have been shown to give intense upconversion luminescence (UCL) emission after cellular internalization	10
	• Non-photobleaching
	• Non-photoblinking
	• Lower interference from auto-fluorescence from surrounding tissues

Since the turn of the last century, the focus of research on UCNPs has been shifted from the controlled synthesis of uniform nanoparticles to exploration of biomedical applications.[11-14] In recent years, the potential of UCNPs has begun to receive considerable attention as a new approach to enhance the versatility of gene therapy. Unfortunately, up to now discussions on the emerging yet encouraging potential of UCNPs in gene delivery have been scant in the literature. This leaves a strong demand for a review filling this gap. The objective of this article is to meet this need by reviewing the latest development of UCNPs as gene carriers. It is hoped that by offering an outlook of current advances in this burgeoning area of research, further development can be facilitated by avoiding potential duplicate efforts, by enabling the identification of challenges to be met, and by pointing to clearer directions for clinical translation of works on UCNP-based gene delivery in the future.

Overview of the properties of UCNPs

Rare earth elements consist of yttrium, scandium and the fifteen elements in the lanthanide series. Except lanthanum and lutetium, the ions of all other lanthanides exhibit distinctive luminescence properties due to possible intra-4f or 4f–5d transitions led by the unique energy structures resulting from the 4f inner shell configuration.[15,16] Contrary to conventional fluorophores that show downconversion caused by internal energy loss, UCNPs emit higher-energy outcome photons through sequential absorption of lower-energy incident ones.[17] This process is called upconversion, which can be mediated using the long-lived and real ladder-like energy levels of lanthanide ions embedded in an inorganic matrix host. Upconversion is a non-linear anti-Stokes process that is achieved mainly by three major mechanisms: excited-state absorption (ESA), energy transfer upconversion (ETU), and photon avalanche (PA). PA is rarely found in lanthanide materials at the nanoscale. Its importance to UCNP-based gene carriers, therefore, is less significant. On the other hand, ETU is one of the commonly adopted mechanisms to achieve high upconversion efficiency in practice. During the process, a pump photon of the same energy is absorbed by each of the two neighbouring ions. The subsequent non-radiative energy transfer causes one of the ions to get excited to the upper energy level whereas the other one relaxes back to the ground state. The emission of photons with higher energy results from the relaxation of the excited ion. ETU can be achieved in a number of ways (Fig. 1), including energy transfer followed by excited-state absorption, successive energy transfer, cross-relaxation upconversion, cooperative sensitization and cooperative luminescence. As the mechanisms of luminescence emission from UCNPs have been reviewed elsewhere, readers are referred to those reviews for details.[18-21]

Fig. 1

General energy schemes related to different ETU processes in UCNPs: (A) energy-transfer followed by excited-state absorption; (B) successive energy transfer; (C) cross-relaxation upconversion; (D) cooperative sensitization; (E) cooperative luminescence. Green, orange and violet arrows represent the excitation light, energy transfer, and upconversion emission, respectively.

UCNPs generally consist of two parts: (i) trivalent lanthanide dopant ions and (ii) the inorganic host lattice that accommodates those ions. To enhance the upconversion emission, the host lattice has to be carefully selected. This involves consideration of several criteria.[22] For instance, the photon vibration energies shall be low. High chemical stability and close lattice matches to dopant ions are also required. Taking these criteria into account, fluorides (e.g., NaYF4) and oxides (e.g., Y2O3, La2O3 and Lu2O3) are some of the materials favourable to be used as host lattices for UCNP fabrication.[23-27] Apart from the selection of the host lattice, the choice of the dopant ions matters. In general, two types of dopant ions are required. One is the activator which emits visible light; whereas the other one functions as a sensitizer that donates energy. Some host-dopant systems commonly adopted for UCNP synthesis are listed in Table 2.[23,25,28-38] To improve the upconversion luminescence (UCL) efficiency of UCNPs, conventionally it is thought that the concentration of the sensitizer should be higher (approximately 20 mol%) than that of the activator, whose concentration has often been controlled to be below 2 mol% to reduce luminescence quenching.[16] The validity of this conventional wisdom has recently been challenged by Johnson et al.,[39] who have discovered that even if the Er3+ concentration in NaY(Er)F4/NaLuF4 core–shell nanocrystals is as high as 100 mol%, the emission intensity of both upconversion and downshifted luminescence can still be enhanced, with no significant concentration quenching effects being observed. This suggests that surface quenching rather than cross-relaxation between dopant ions may play a predominate role in causing luminescence quenching at high dopant concentrations.[39] This finding has revealed the possibility of constructing and engineering UCNPs in a way that is no longer restrained by the conventional limit of the activator concentration.

Table 2

A list of host-dopant systems commonly adopted for UCNP synthesis

Host lattice	Activator	Sensitizer	Excitation wavelength (nm)	Colour of light emission	Emission peak (nm)	Ref.
Lu2O3	Er, Tm	Yb	980	Blue, green, red	490, 540, 662	23
Y2O3	Er	Yb	980	Green, red	550, 660	25
	Ho	Yb	980	Green, red	543, 665	28
CaF2	Er	Yb	980	Green, red	524, 654	29
LaF3	Ho	Yb	980	Green, red	542, 645, 658	30
	Er	Yb	980	Green, red	520–545, 659
	Tm	Yb	980	Blue	475
LuPO4	Tm	Yb	980	Blue, red	475, 649	31
NaYF4	Ho	Yb	980	Green, red	542, 645–658	32
	Er	Yb	980	Green, red	510–540, 635–675	33–37
	Tm	Yb	980	Blue, red	450–457, 647	33
	Er, Tm	Yb	980	Blue, green, red	474–499, 525, 644–693	38

UCNPs are indeed only one of the many luminescent nanoparticles investigated in photobiology. Other nanoparticles widely studied for biomedical applications include transition metal complexes, QDs and organic fluorophores. Compared to many of these materials, UCNPs have the merit of low toxicity in in vitro and in vivo contexts. This has been evidenced in the literature,[40-47] in which different concentrations of UCNPs (from 5 to 2500 μg mL–1) and incubation periods (from 2 hour to 48 hours) have been investigated in various cell lines (e.g., HeLa, KB, L929, CL, HCCHM3 and HepG2). Most of these studies have reported that over 75% of cell viability remains after treatment with UCNPs. Moreover, as shown by the observation that mesenchymal stem cells labelled with oligo-arginine-poly(ethylene glycol) (PEG)-coated NaYF4:Yb,Er UCNPs maintain their stem cell potency,[48] the effect of UCNPs on cell behaviour is generally thought to be minimal. In spite of this, it is worth emphasizing that UCNPs are not necessarily toxicity-free. A previous study has reported that ligand-free lanthanide-doped nanoparticles can induce intracellular ATP deprivation in HeLa cells and can result in a significant decrease in cell viability.[49] Such UCNP-induced cell death is attributed to the interactions of the nanoparticles with the phosphate group of cellular ATP to cause apoptosis and autophagy.[49] Nevertheless, UCNPs have a comparatively high safety profile among commonly used luminescent nanoparticles.[40-47] Along with the ease of modulating their physicochemical properties via surface engineering, UCNPs turn out to be favourable building blocks for further development as multifunctional gene carriers.

Molecular design of UCNPs as gene carriers

A fundamental aspect of designing UCNP-based gene carriers is to determine the nanoparticle composition. When UCNPs are designed, dopant ions are often chosen by considering not only the spaced energy levels that govern photon absorption by the sensitizer, but also the energy transfer process between the sensitizer and the activator. Yb3+ has been widely adopted as a sensitizer, owing to its high absorption coefficient and upconversion efficiency.[50] In addition, the f–f transitions of many commonly used upconverting lanthanide ions (e.g., Er3+ and Tm3+) can be resonant with the 2F7/2 → 2F5/2 transition of Yb3+. This further facilitates energy transfer from the sensitizer to the activator.[16] Regarding the selection of activators, common choices include Tm3+, Ho3+ and Er3+,[16] although the use of other lanthanide ions (such as Tb3+, Dy3+ and Pr3+) has been occasionally reported in the literature.[51,52] To date, the most efficient UCNPs obtained have been those using Tm3+ and Er3+ as the activators. These ions have ladder-like energy levels,[16] and have relatively large energy gaps. These energy gaps can enhance luminescence emission, as suggested by the energy gap law:[53] where k nr represents the multiphonon relaxation rate constant for 4f levels of a lanthanide ion; ΔE represents the energy gap between the populated level and the next lower-lying energy level of a lanthanide ion; β represents the empirical constant of the host; ħω max designates the highest-energy vibrational mode of the host lattice. The law reveals that the energy gap is negatively related to the multiphonon relaxation rate constant. In other words, the comparatively large energy gap possessed by Tm3+ and Er3+ can reduce the probability of having non-radiative transitions among different excited levels of ions, thereby increasing the efficiency of upconversion.

Apart from the nanoparticle composition, some other factors have to be considered so as to render the UCNPs applicable to gene delivery. For example, the nanoparticles should be biodegradable, biocompatible and non-toxic.[54-57] In addition, gene delivery mediated by UCNPs is a multi-stage process (Fig. 2). Contrary to the delivery of chemical drugs, in which the intervention will still be therapeutic even if the carrier fails to be internalized into cells but simply releases the payload outside, gene therapy is possible only when cellular internalization of the delivered gene is successful.[58-60] As far as cellular uptake is concerned, the size and zeta potential of the nanopaspan>rticles are two important determining factors. A small size can be achieved by surface passivation or functionalization to enhance the colloidal stability of UCNPs.[16,61] This is pivotal when the nanopaspan>rticles are to be used in vivo, in which salt ions in blood may coordinate with the exposed lanthanide ions on the UCNP surface, causing nanoparticle aggregation. Regarding the zeta potential, it can be manipulated by incorporating a positively charged coating [e.g., poly(ethylenimine) (PEI),[62] cationic lipids,[63] cetyl trimethylammonium bromide (CTAB),[64] dimethyldidodecylammonium bromide (DMAB),[64] and chitosan[65]] onto the nanoparticle surface. This facilitates the efficiency of the subsequent gene loading process, as well as the binding of UCNPs to the anionic plasma membrane. The cellular attachment and tissue specificity of the nanoparticles can be enhanced by conjugating ligands (such as folic acid,[66,67] galactose,[68] transferrin,[66,69] RGD peptide,[70,71] and antibodies[72,73]) to the nanoparticle surface for receptor-mediated endocytosis. The feasibility of this has been evidenced by an earlier study, in which folic acid and anti-Her2 antibody have been conjugated to silica-coated NaYF4:Yb,Er UCNPs.[74] Compared to the unmodified counterparts, ligand-conjugated UCNPs exhibit higher transfection efficiency and gene silencing efficiency in SK-BR-3 cells, in which Her2 receptors are overexpressed.[74]

Fig. 2

Major processes of UCNP-based gene delivery: (A) synthesis of UCNPs; (B) surface modification; (C) gene loading; (D) cell attachment; (E) cellular internalization; (F) endolysosomal escape; (G) nuclear localization; (H) dissociation.

As UCNPs are mostly internalized via endocytosis, their ability to undergo endo-lysosomal escape (e.g., by eliciting the proton sponge effect) may determine the ultimate success of the gene delivery process. In order for the delivery process to be therapeutic, plasmids also have to reach the nucleus whereas RNA molecules have to be released in the cytosol.[59] For the former, one strategy to facilitate the nuclear import of the delivered nucleic material is to incorporate the UCNP surface with nuclear localization signal (NLS) peptides (e.g., PARP, M9-ScT conjugate, SV40 T antigen, Xenopus N1, adenovirus E1a, human c-myc, SV40 Vp3, and mouse FGF3),[75-78] which can localize the nanoparticles to the nucleus and allow them to be actively transported across the nuclear pore complex. But even with proper intracellular localization, careful manipulation of the UCNPs, in particular the surface properties and the buffering capacity of the polymer coating, is required because this may influence the process of gene release. Failure to dissociate the payload from the nanoparticles at the right location can indeed influence the outcome deleteriously.

Fabrication of UCNP-based gene carriers

UCNP-based gene carriers can be constructed using various strategies. Co-precipitation is one of the simplest methods to generate UCNPs because it does not involve time-consuming procedures or severe reaction conditions.[79] By adding capping ligands into the solvent during the synthetic process, the growth of the nanoparticles can be controlled and the UCNPs can be stabilized. Examples of these capping ligands include polyvinylpyrrolidone (PVP), ethylenediaminetetraacetic acid (EDTA) and PEI.[37,80,81] In the case of NaYF4:Yb,Er nanocrystals, the upconversion emission exhibited by those in the hexagonal phase is generally higher than that exhibited by the cubic-phase counterparts.[82] However, nanocrystals synthesized via co-precipitation are often in the cubic phase, and hence are not the most efficient upconverter.[83] To address this problem, calcinations at a high temperature can be applied to achieve the sharpened crystal structure or to mediate partial phase transfer to the hexagonal-phase nanocrystals.[37] Apart from NaYF4:Yb,Er nanocrystals, other nanoparticles (e.g., LuPO4:Yb,Tm and YbPO4:Er,Tm) have been successfully generated by co-precipitation, with subsequent heat treatment being applied to improve the upconversion efficiency.[84] Despite the wide application of co-precipitation in UCNP generation, particle aggregation may occur during the synthetic process, making precise control of the particle size difficult. As the size of the nanoparticles is an important parameter governing the efficiency of cellular internalization during the gene delivery process, the polydispersity of the generated nanocrystals is an issue to be tackled when the performance of a gene carrier is to be enhanced.

Another method of UCNP generation is thermal decomposition, in which solvents with a high boiling point (e.g., octadecene, oleic acid and oleylamine) are often used to dissolve rare earth trifluoroacetate precursors, which are often thermolyzed at 300 °C or above.[85,86] Using this method, LiYF4 and KGdF4 UCNPs have been obtained.[87] Monodispersed hexagonal-phase NaYF4:Yb,Er and NaYF4:Yb,Tm nanoparticles with enhanced upconversion emission have been generated, too.[88] Notwithstanding this, due to the involvement of the use of expensive and air-sensitive metal precursors,[82] as well as the production of toxic by-products during the fabrication process,[85,88] the selection of this technique is not preferred sometimes. As an alternative to thermal decomposition, UCNPs can be synthesized via the sol–gel method, in which the metal precursors used are relatively cheap. This method has had a track record of applications in the fabrication of TiO2:Er, ZrO2:Er, Lu3Ga5O12:Er, YVO4:Yb,Er, and BaTiO3:Er UCNPs.[89-93] Unfortunately, particle aggregation may occur when the nanoparticles generated by this method are dispersed in aqueous solutions. This limits the use of the nanoparticles in gene delivery, in which water is always the major medium through which gene carriers are delivered to target cells. Along with the occurrence of particle aggregation further induced by high-temperature calcinations, which is required to increase the crystalline phase purity so as to enhance luminescence emission, the sol–gel approach might not be the most suitable method for generation of UCNP-based gene vectors.

Apart from those mentioned above, UCNPs can be generated by the combustion method[94,95] or by the hydro(solvo)thermal process.[96-98] Contrary to the former in which proper control of the particle size is challenging and the crystalline phase purity is generally low, the latter can generate high-quality UCNPs upon proper control of the process parameters (e.g., pH, reaction time, reaction temperature, and the type of precursors).[96-98] In an earlier study, α- and β-phase NaYF4:Yb,Er UCNPs with well-controlled size and morphology have been generated using the hydrothermal method, with EDTA and citrate being used as the capping ligands.[99] The size of the particles has been shown to be controllable by manipulating the nucleation rate, which, in turn, can be adjusted by modulating the reactant concentration or by changing the type of ligands adopted.[99] In addition, by modifying the reaction time as well as the reactant concentration, phase transformation for the nanoparticles can be achieved.[99] This phenomenon can be exploited to control the morphology of the generated UCNPs. Lately, polymer-coated UCNPs with high aqueous solubility have been generated based on the hydro(solvo)thermal mechanism.[100] The high hydrophilicity has rendered the UCNPs favourable to be utilized in biological applications. Due to the ease of control of the properties (e.g., size, structure, and morphology) of the generated nanoparticles, along with the possibility of synthesizing the nanoparticles in a “one-pot” process,[81,92,101] the hydro(solvo)thermal method is and will continue to be one of the most favourable and convenient synthetic routes, in the practical sense, to UCNP-based gene carriers.

Surface modification can be adopted to optimize the biological performance of UCNP-based gene carriers (Fig. 3). The roles played by surface modification can be two-fold. One is to improve the gene loading efficiency, and the other is to optimize the physicochemical properties of the nanoparticles for better biological performance. Each of these roles will be discussed in more detail in this section.

Fig. 3

Surface modification of UCNPs for gene delivery.

Enhancement of the gene loading process

Upon fabrication of the nanoparticles, usually surface modification with cationic moieties will be exercised to render the nanoparticles applicable to gene loading. Polyelectrolyte complexation between nucleic acids and the cationic moieties on the nanopaspan>rticle surface is hitherto the most fundamental gene loading mechanism. In an earlier study, the surface of silica-coated NaYF4:Yb,Er UCNPs has been modified with amine groups using N-[3-(trimethoxysilyl)propyl]ethylenediamine (AEAPTMS), and as shown by the gel retardation assay, the modified UCNPs can complex with RNA molecules via electrostatic interactions between the positively charged amine groups and the negatively charged nucleic acid material.[74]

Surface modification of UCNPs can also be achieved by direct incorporation with polycations. One representative polycation used in this aspect is PEI, which is a cationic aziridine polymer with high proton buffering capacity over a wide range of pH values.[57,102] PEI has been extensively adopted for non-viral delivery of nucleic materials (including plasmids,[103] oligonucleotides[104] and ribozymes[105]) in reagent-consuming animal studies. Its transfection efficiency can be optimized by modulating the physical–chemical features (e.g., charge density, degree of branching, and molecular weight) of the PEI molecules,[106-108] and by optimizing the transfection conditions (e.g., polyplex concentration and incubation time)[109] and polyplex properties (e.g., zeta potential and particle size).[109] In a previous study, gadolinium (Gd3+)-doped UCNPs have been modified first by covalently conjugating PEG onto the nanoparticle surface, followed by deposition of PEI coatings using the layer-by-layer assembly technique.[110] The PEI-coated UCNPs have not only been shown to be more effective in transfection than the uncoated counterparts,[110] but have also been reported to exhibit high gene delivery efficiency in the serum-containing environment, in which native PEI has experienced a significant drop in the efficiency of transfection.[110] The nanoparticles have shown potential to be further developed into a gene carrier for biological use, in particular when parental administration of the therapeutic gene is required. Apart from PEI, other polycations have been adopted in the literature to coat UCNPs for enhancing the process of gene loading. For instance, PEG–poly(lactic-co-glycolic acid) (PEG–PLGA), along with a positively charged amphiphilic polymer synthesized by aminolyzing polysuccinimide (PSI) with N-(3-aminopropyl)imidazole (NAPI) and oleylamine, has been utilized to coat hydrophobic NaYF4:Yb,Er UCNPs.[111] Owing to the presence of the cationic coating, genes can effectively adsorb onto the UCNP surface,[111] rendering the nanoparticles applicable as gene vehicles.

Optimization of physicochemical properties

Not only effective drug loading but also high aqueous solubility is vital to the proper functioning of UCNPs when the nanoparticles are used as gene carriers. Yet a majority of lanthanide-doped UCNPs are poorly soluble in the aqueous environment. In recent years, several surface functionalization strategies have been put forward to increase the hydrophilicity of UCNPs.[16] One method is surface silanization,[112] whose applications in surface modification have been rapidly growing due to the availability of well-established routes to silica-coated nanoparticles and the applicability of silica-coating to both hydrophobic and hydrophilic surfaces.[113,114] The feasibility of applying surface silanization to surface engineering of UCNPs has been demonstrated in a previous study, in which the affinity of PVP with silica has been exploited to coat PVP-stabilized NaYF4:Yb,Er nanocrystals with a uniform silica shell having a thickness of approximately 9 mm.[115] A similar approach has been used by Li and Zhang.[112] They have produced water-soluble silica-coated PVP-stabilized NaYF4:Yb,Er nanocrystals whose shell thickness can be tuned by changing the concentration of tetraethoxysilane (TEOS), which is a precursor during the process of silica formation. All of these have evidenced the practicality of silica-coating in surface modification of UCNPs. Lately, Wang and co-workers have reported the fabrication of fluorescent silica-coated mesoporous microcarriers.[116] The microcarriers show extended residence time of up to 3 days in the gastrointestinal tract, releasing more than 60% of their content. They can also emit in the near-infrared window of 1000–1400 nm,[116] thereby enabling real-time tracking of the microcarrier fate as well as allowing for semi-quantitative monitoring of the content of drug release in vivo.[116] Although at the moment the nanocarriers have been evaluated only for drug delivery, this study has offered insights into possible strategies for monitoring the kinetics and dynamics of a delivered agent after administration to a living body. Along with the good biocompatibility of silicon, surface silanization may turn out to be promising for engineering the surface properties of UCNP-based gene carriers for theranostic applications in the future.

Other than surface silanization, surface engineering of UCNPs can be performed using non-silane reagents via diverse mechanisms, such as layer-by-layer assembly,[34] ligand attraction,[117] ligand exchange,[88] ligand removal[118] and ligand oxidation.[119] For instance, by taking advantage of electrostatic attraction, Hilderbrand et al. have coated UCNPs with a layer of polyacrylic acid (PAA) via electrostatic layer-by-layer assembly,[120] during which the carboxyl groups of PAA have been linked covalently with amino-modified PEG. Via the process of ligand exchange, an earlier study has also replaced the oleylamine ligands, which have been used to stabilize NaYF4:Yb,Er nanoparticles, with bifunctional organic molecules to render the nanoparticle surface hydrophilic.[88] More recently, based on the phenomenon that PEI and PAA exhibit higher binding affinity than PVP towards lanthanide ions, PEI- and PAA-coated NaYF4:Yb,Er UCNPs have been generated from PVP-stabilized nanoparticles.[121] The UCNPs show good dispersibility in aqueous media after the ligand exchange process.[121] More details of different surface modification strategies to enhance the aqueous solubility of UCNPs are presented in Table 3.

Table 3

Common strategies to modify the UCNP surface to enhance hydrophilicity

Strategy	Basic principle	Example of application	Ref.
Layer-by-layer assembly	Polyions with opposite charges are deposited onto the UCNP surface layer by layer via electrostatic absorption	Water-soluble NaYF4:Yb,Er UCNPs have been generated by sequential adsorption of poly(allylamine hydrochloride) (PAH) and poly(sodium 4-styrenesulfonate) (PSS) onto the nanoparticle surface	34
Ligand exchange	Bifunctional molecules are used to displace the ligands originally coordinating to the UCNP surface	Bifunctional organic molecules (PEG 600 diacid) have been adopted to replace the oleylamine ligands originally used to stabilize the NaYF4:Yb,Er nanoparticles. The nanoparticles have shown good aqueous solubility after the ligand exchange process	88
Surface silanization	By hydrolysis and condensation of siloxane monomers, an amorphous silica shell is grown on the UCNP surface	PVP-stabilized NaYF4:Yb,Er UCNPs have been coated with silica by using tetraethoxysilane (TEOS) as a precursor. The silica-coated UCNPs can effectively disperse in aqueous solutions	112
Ligand attraction	An amphiphilic block copolymer is used to modify the UCNP surface by adsorbing onto the surface via hydrophobic interactions between the copolymer and the original surface ligand	Polyacrylic acid, which has been modified with 25% octylamine and 40% isopropylamine, has been used to coat NaYF4:Yb,Er UCNPs which possess carboxyl groups on the surface. The coated nanoparticles can be readily dispersed in aqueous solutions	117
Ligand removal	Hydrophobic ligands coordinating to the UCNP surface are removed to increase the aqueous dispersibility of the nanoparticles	An acid treatment has been applied to remove the oleate ligands from the surface of oleate-capped NaYF4:Er,Yb UCNPs. The ligand-free nanoparticles generated can effectively disperse in aqueous solutions	118
Ligand oxidation	The possible use of this strategy is limited to those UCNPs capped by ligands with unsaturated carbon–carbon bonds. To execute ligand oxidation, the Lemieux-von Rudloff reagent is often adopted to oxidize the carbon–carbon double bonds to pendant carboxylic functional groups	The Lemieux-von Rudloff reagent has been used to convert oleic acid-stabilized NaYF4:Yb,Er UCNPs into water-dispersible nanoparticles	119

Surface modification can not only improve the particle hydrophilicity but can also modulate the physiochemical properties of UCNPs, thereby enhancing the efficiency of gene delivery. This has been revealed by multiphoton confocal microscopy and inductively coupled plasma mass spectrometry (ICP-MS) measurements, in which the pan class="Chemical">PEI-coated UCNPs have been found to display greater efficiency in cellular internalization as compared to their counterpaspan>rts having neutral or negative surface charges.[121] In addition to pan class="Chemical">PEI, PEG is another polymer widely used as a surface modifier due to its capacity of enhancing the aqueous solubility of various nanoparticulate gene delivery systems[122] and of reducing particle aggregation.[123] Attributed to the hydrophilic nature of PEG and the brush-type polymer crowding,[124] PEGylated particles are usually less prone to opsonization and reticuloendothelial system (RES) uptake, and hence having the blood circulation time lengthened.[124,125] Despite this, every coin has two sides. As hinted at by an earlier study, polyplexes that have undergone PEGylation display premature vector unpackaging in blood, causing a decline in the gene delivery efficiency to the liver.[126] Similar problems have been delineated by Mishra et al.,[127] who have found that the cellular uptake and intracellular trafficking of polyplexes are impeded after PEGylation, even though the salt stability of the polyplexes is enhanced. Taking all these into account, to maximize the positive effect of PEGylation in the molecular design of UCNP-based gene carriers, structural properties (e.g., density, conformation, molecular weight, and flexibility) of the PEG moiety and the degree of PEG grafting have to be carefully considered before PEGylation is executed. In addition, acid-labile linkages (e.g., vinyl ether,[128] acetals,[129,130] and hydrazones[131]) can be used to link the PEG shield to the UCNP surface. The linkage can then be hydrolyzed in the acidic milieu of the endosomal compartment, leading to shielding destabilization and therefore mitigating the possible drawback of PEGylation to endolysosomal escape.[132]

Recent advances in UCNP-based gene transfer

With advances in the molecular design of UCNPs and the continuous development of technologies for materials fabrication, over the years copious UCNP-based carriers have been developed for delivery of nucleic acid materials. Based on the nature of the materials to be delivered, these carriers can be categorized into two types. One is for DNA delivery, and the other is for RNA delivery.

UCNP-based DNA delivery

As far as UCNP-based gene transfer is concerned, most of the efforts in the literature have been devoted to DNA delivery, whose practical potential in biomedicine has been evidenced in a pre-clinical trial, in which aminosilane-modified NaYF4:Yb,Er UCNPs have been exploited as carriers for DNA vaccination to combat foot-and-mouth disease (FMD).[133] The UCNPs can complex with the plasmid pcDNA3.1/VP1-GFP via electrostatic interactions, and protect the plasmid from DNase I degradation.[133] As revealed by in vitro studies, the transfection efficiency of the nanoparticles is comparable to lipofectamine, but with lower cytotoxicity.[133] Upon intramuscular injection of the UCNP/DNA complex to guinea pigs, induction of the humoral and cellular immune responses has been achieved.[133] The serum levels of anti-FMDV specific antibodies and neutralizing antibodies, as well as the proliferation of T-lymphocytes, have also been found to be enhanced.[133] As confirmed by the challenge test, the guinea pigs vaccinated with the UCNP/DNA complex have been fully protected from attack by the FMD virus.[133] This study has pointed to the possible use of UCNP-based gene carriers in preventive medicine. Despite this, the unique optical properties of UCNPs have not been exploited during the design of the DNA vaccine carrier. Making use of those properties in the delivery system for additional capacity (e.g., photo-triggered release of the delivered plasmid during the vaccination process) can be the next rewarding step to pursue to escalate the application potential of the carrier in the clinical context. In fact, UCNPs can play at least two roles when they are incorporated into the design of a gene delivery system: (1) imaging-monitored gene delivery and therapy, and (2) temporal–spatial confinement of gene manipulation. Each of them will be discussed individually below.

Imaging-monitored gene delivery and therapy

One immediate advantage brought about by using UCNPs as gene carriers is the possibility to track the delivery process via luminescence-based imaging. Over the years, advances in luminescence-based imaging have substantially facilitated the unravelling of the mechanisms of disease progression[134] and guiding the development of treatment for diseases such as choroidal melanomas, whose microcirculation has been successfully imaged in vivo using an indocyanine green (ICG) fluorescent probe.[135] In UCNP-based gene delivery, the viability of integrating therapeutic and imaging functionalities into one single system has been shown by Bai et al.,[111] who have coated hydrophobic NaYF4:Yb,Er UCNPs with PEG–PLGA and a positively charged PSI-based amphiphilic polymer.[111] Before the polymer coating process, the nanoparticles display high crystallinity with an average diameter of around 25 nm. Although there is an increase in the particle size after the coating process, the average size is still in the size range favourable for cellular internalization. Importantly, the coated nanoparticles exhibit good biocompatibility and reasonable gene delivery efficiency.[111] Photoluminescence measurements have indicated that the coated nanoparticles show emission bands at around 540 nm and 660 nm.[111] These two bands are assigned to 4S3/2–4S15/2 and 4F9/2–4I15/2 transitions, respectively. The red UCL emission given by the nanoparticles enables tracking of the in vitro delivery process without background fluorescence interference.[111] This has rendered the UCNPs reported in the study applicable to imaging-guided therapy in the future.

Apart from luminescence-based imaging, UCNP-based gene carriers can be designed to mediate multimodal imaging. One example is provided by the case of PEI-coated NaGdF4:Yb,Er UCNPs, which can not only deliver plasmids in vitro but can also serve as a contrast agent for UCL, magnetic resonance imaging (MRI) and computed tomography (CT) (Fig. 4).[136] Another example is given by He et al.,[110] who have modified the surface of Gd3+-doped UCNPs with PEG and PEI for transfection. Results have confirmed that the nanoparticles can not only function as gene carriers, but also display emission peaks at around 540 and 660 nm.[110] Even though the intensity of luminescence emission drops to 80% when 2 layers of PEI are incorporated into the nanoparticle surface, the intensity of the emission is sufficient for luminescence-based imaging (Fig. 5).[110] Furthermore, as the nanoparticles are doped with Gd3+, they can provide contrast in MRI. Compared to gadolinium-diethylenetriamine penta-acetic acid (Gd-DTPA), the nanoparticles with 2 layers of PEI have been found to have a much higher longitudinal relaxivity value.[110] Considering their versatile imaging capacity and transfection activity, upon further development and optimization, the nanoparticles have high potential to serve as a multifunctional carrier for future theranostic applications.

Fig. 4

(A) TEM images of (a) as-prepared UCNPs and (b) PEI-coated UCNPs. (B) In vivo T 1-weighted magnetic resonance images of a tumour-bearing mouse: (a) before and (b) after injection of the PEI-coated UCNPs in situ. (C) CT images of a tumour-bearing mouse: (a) before and (b) after injection of the PEI-coated UCNPs in situ, and (c and d) the corresponding 3D renderings of the CT images. (D) Inverted fluorescence microscopy images of HeLa cells incubated with the PEI-coated UCNPs for (a–d) 0.5 h, (e–h) 1 h, and (i–l) 3 h. The scale bar represents 20 μm (adapted from ref. 136 with permission from the American Chemical Society).

Fig. 5

(A) Confocal fluorescence images of HeLa cells transfected using (a) PEI or (b) UCNP-PEG@2×PEI in the presence of various concentrations of fetal bovine serum: (i) 0%, (ii) 10%, (iii) 20% and (iv) 30%. The images were taken 48 h after the initiation of transfection. (B) Confocal UCL/fluorescence images of HeLa cells after 4 hours of incubation with various concentrations of UCNP-PEG@2×PEI: (i) 4.3 mg L–1, (ii) 8.7 mg L–1, (iii) 17.3 mg L–1, (iv) 34.7 mg L–1, and (v) 69.4 mg L–1. (C) T 1-weighted MRI images of HeLa cells after 4 hours of incubation with various concentrations of UCNP-PEG@2×PEI: (i) 0 mg L–1, (ii) 4.3 mg L–1, (iii) 8.7 mg L–1, (iv) 17.3 mg L–1, (v) 34.7 mg L–1, and (vi) 69.4 mg L–1. The cells were suspended in a 1% agarose gel for MRI (adapted from ref. 110 with permission from the American Chemical Society).

Despite the promising advances made in the field, owing to the low extinction coefficient and narrow band absorption of lanthanide ions,[137] the light absorbing ability of UCNPs is generally limited. This restrains the wide application of many of the reported UCNP-based gene carriers in imaging procedures. This situation is worsened by the fact that those carriers are in the nano-size range. The surface-to-volume ratio is, therefore, very high. This makes the emission efficiency of those carriers highly sensitive to surface-related deactivations. These deactivations can not only occur via direct deactivations, by neighbouring quenching centres, of the photoexcited dopants located on or around the UCNP surface, but can also happen if the energy possessed by the photoexcited dopants located in the centre of the nanoparticle randomly migrates to the dopants on or around the carrier surface or directly to the surface quenching sites. To solve this problem, one strategy is to take advantage of the antenna effect from other species with strong light-absorbing ability (e.g., plasmons, QDs or organic dyes) to compensate for the low extinction coefficient resulting from 4f–4f optical transitions in lanthanide ions. Another strategy is to suppress the surface-related quenching mechanisms by incorporating the UCNPs with a core–shell structure, in which the host material of the shell shows a low lattice mismatch with the core material. To fabricate UCNPs with the core–shell architecture, shell layers are usually deposited onto core nanocrystals via epitaxy.

Epitaxial shells can be grown through chemical reactions similar to those adopted to generate the core particles, except that the process of crystal growth occurs on the core surface rather than in the liquid phase. To achieve epitaxial shell coating, one method is to use the heat-up strategy, which allows for the generation of UCNPs with a multi-shelled structure by either repeating the same synthetic protocol multiple times, or by arbitrarily combining dissimilar synthetic approaches for deposition of shells with different properties onto the same core crystal.[138] The viability of this strategy to enhance photoluminescence has been demonstrated by Zhang et al.,[139] who have heated NaYF4:Yb,Er nanocrystals in an oleic acid/1-octadecene solution containing precursors for the formation of the hexagonal NaGdF4 shell. Another strategy for epitaxial growth is the hot-injection method, in which a sequence of shell precursors is injected into a host reaction for the one-pot synthesis of multi-shelled nanoparticles.[117] The use of this method was first reported in the early 2000’s when NaYF4:Yb,Er@NaYF4 core–shell UCNPs were synthesized by first heating related rare earth trifluoroacetates in oleylamine for the growth of the NaYF4:Yb,Er core nanoparticles, followed by the injection of an oleylamine solution containing the shell precursors to achieve epitaxial deposition of the undoped NaYF4 shell layer.[117] Using a similar approach, the synthesis of few other core–shell UCNPs (e.g., NaGdF4@NaGdF4 and LiLuF4@LiLuF4) has been reported in the literature.[140-142]

Apart from the aforementioned methods of epitaxial growth, deposition of the shell layer can be exercised in a non-epitaxial manner, in which the shell layer can be immobilized on the surface of pre-synthesized UCNPs either by means of chemical bonding or through surface polymerization.[143,144] Such methods have been employed for fabricating silica-coated NaYF4:Yb,Er UCNPs[144] and NaYF4:Yb,Tm UCNPs with tuneable surface coverage of gold nanoparticles.[145] With the incorporation of the multi-shelled nanostructure, the emission efficiency of UCNPs has been shown to be enhanced in several reports. For instance, after coating with an undoped NaYF4 shell, UCL emission from NaYF4:Yb,Er and NaYF4:Yb,Tm UCNPs has been found to be remarkably enhanced.[117] NaYF4:Yb,Er UCNPs with a hexagonal NaGdF4 shell have also been reported to give more intense overall emission than the uncoated counterparts, owing to the passivation of surface defects of the nanocrystals by shell deposition.[139] All of these have evidenced the effectiveness of the core–shell nanostructure in enhancing the emission intensity of UCNPs and in strengthening the capacity of the nanoparticles to mediate imaging-monitored gene delivery and therapy in practice.

Temporal–spatial confinement of gene manipulation

To achieve target-specific gene delivery, ligand conjugation to the carrier surface is a prevailing strategy; however, controlled release of nucleic acids may represent a new direction. The latter can be achieved using photoactivatable molecules that release payloads at specific sites upon UV irradiation.[146] The clinical application of this strategy has unfortunately been impeded by the toxicity and low tissue penetration power of UV light. This problem may be solved using UCNPs, which can convert near-infrared (NIR) or visible light to UV in situ to regulate the process of gene manipulation. The possible use of UCNPs to precisely control gene expression has recently been demonstrated in tumour cells that have been transplanted into adult zebrafish.[147] Similar success has also been reported on silica-coated NaYF4:Yb,Tm nanocrystals,[148] in which the NIR-to-UV upconversion process has been exploited to silence the expression of target genes in a temporally and spatially specific manner. Such ability to manipulate gene expression has provided practical implications not only for treatment development but also for fundamental research on signal transduction. All of these are beyond the reach of conventional gene delivery methods.

As a matter of fact, while using UCNPs to control the location of gene manipulation is relatively new; similar concepts have already been used extensively in drug delivery research, in which UV light has been applied to manipulate the timing, dosage, and location of drug release.[149-152] The process of photo-triggered drug release is mediated by molecule excitation upon photon absorption and by the subsequent relaxation process, which is achieved via radiative and non-radiative pathways. In the radiative process, energy is usually emitted in the form of fluorescence when the molecule in the excited state returns to the ground state; whereas in the non-radiative scenario, multiple pathways can be involved. One pathway is internal conversion, in which energy is released in the form of heat from an excited molecule. Another pathway is intersystem crossing, which involves the conversion of the singlet state of an excited molecule into a triplet state without emission of photons. In addition to these, excited molecules may undergo photochemical reactions (e.g., photocleavage, Wolff rearrangement, photoisomerization, and photocrosslinking) and experience non-radiative decay. Incorporation of these reactions into the molecular design of UCNP-based gene carriers is scant at the moment, but these reactions have already been well-adopted to control the delivery of chemical drugs.[146,149,153,154] A good example has been given by Matyjaszewski and co-workers, who have synthesized a block copolymer having poly(ethylene oxide) (PEO) as the hydrophilic segment and poly(spiropyran methacrylate) as the hydrophobic part.[155] In an aqueous medium, the copolymer forms micelles with a core–shell structure. These micelles are disrupted upon UV irradiation, which causes the spiropyran unit to undergo a reversible isomerization between hydrophobic spiropyran (SP) and hydrophilic merocyanine (ME), leading to the release of the encapsulated agent. Upon irradiation with visible light, photochemical reversion from ME to SP occurs, and the disrupted micelles are reformed. More examples demonstrating the possible use of photochemical reactions in controlling payload release are provided in Table 4.[156-162] In view of the similar nature between drug delivery and gene delivery, translating these strategies into the molecular design of UCNP-based gene carriers is not only theoretically feasible but may also enable more precise control of UCNP-mediated gene manipulation in the future.

Table 4

Photochemical mechanisms commonly adopted in the molecular design of light-stimulated systems for delivery purposes

Mechanism	Working principle	Photoresponsive groups	Use	Ref.
Photocleavage	Cleavage of a covalent bond is induced by light irradiation. This disrupts the integrity of the drug carrier, triggering the release of the drug molecules	NB group	Photodissociable polymeric micelles have been generated from a block co-polymer, in which poly(ethyleneoxide) is the hydrophilic block and poly(2-nitrobenzyl methacrylate) is the hydrophobic block. UV irradiation results in micelle disruption and hence the release of the loaded compound	156
			Polymeric vesicles have been generated from a diblock copolymer consisting of a hydrophilic PAA segment and a hydrophobic PMCL segment bridged by the NB linker. Upon UV irradiation, the vesicles disintegrate and the payload is released	157
Photoisomerization	Photo-induced changes in the molecular conformation (e.g., cis–trans isomerization and reversible ring opening/closing reactions) of the photoisomerizable component of a drug delivery system may change the steric effects and other physical–chemical properties of that component. This leads to changes in the drug release properties of the system	Azobenzene, stilbene, spiropyran and dithienylethene	Photoswitchable nanoparticles have been generated from a spiropyran derivative and a lipid-PEG. Upon UV irradiation, the nanoparticles shrank, expelling drugs repeatedly	158
			TSUA molecules have bound to mesoporous silica nanoparticles, with β-cyclodextrin molecules being threaded onto the trans-TSUA stalks to seal the nanopores. The cyclodextrin rings are dissociated from the stalks upon UV irradiation, leading to the release of the cargo	159
Photocrosslinking	Polymerization induced by light may alter the structural integrity, and hence the drug release properties, of a system	Methacrylates and coumarin	Polymeric micelles have been generated from a diblock copolymer consisting of PEO as the hydrophilic block and poly(coumarin methacrylate) as the hydrophobic block. Interchain crosslinking and de-crosslinking have been induced by irradiation with light at different wavelengths. This results in changes in the rate of drug release from the micelles	160
			Photoresponsive mesoporous silica nanoparticles have been designed based on the principle of coumarin-based reversible photodimerization. The storage and release of guest molecules from the nanoparticles can be controlled by irradiating the system with light at different wavelengths	161
Wolff rearrangement	The Wolff rearrangement of an α-diazocarbonyl yields a ketene, which can undergo further reactions to ultimately alter the drug release properties of a drug carrier	DNQ	Micelles have been fabricated from a PEG-lipid amphiphile, whose hydrophobic end has been incorporated with DNQ. Upon UV irradiation, DNQ converts to 3-indenecarboxylate, disrupting the integrity of the micelles and triggering the release of the payload	162

Abbreviations: DNQ, 2-diazo-1,2-naphthoquinone; TSUA, 4-(3-triethoxysilylpropylureido)azobenzene; PAA, polyacrylic acid; PMCL, poly(methyl caprolactone); NB, O-nitrobenzyl.

UCNP-based RNA delivery

If an UCNP-based gene carrier can load DNA via electrostatic interactions, the same carrier should be applicable to complex with RNA for delivery purposes, due to the similarity of the electrostatic properties between DNA and RNA molecules. Nevertheless, RNA shows extra vulnerability to enzymatic degradation, and proper protection of the RNA molecules during the delivery process is vital. An example of UCNP-mediated RNA transfer is presented by an earlier study, in which UCNPs have been prepared by first encapsulating Yb3+/Tm3+ co-doped nanocrystals in a silica shell with surface amine groups, followed by surface functionalization with cationic photocaged linkers to make siRNA loading feasible (Fig. 6).[163] Upon NIR irradiation, the photocaged linker on the UCNP surface is cleaved by upconverted UV light.[163] This initiates the release of siRNA molecules in a temporal–spatial manner. A similar approach has been adopted by Guo et al.,[148] who have used silica-coated NaYF4:Yb,Tm UCNPs as a carrier of siRNA to act against the expression of survivin. Those siRNA molecules have been caged with 4,5-dimethoxy-2-nitroacetophenone (DMNPE) before RNA delivery, and are subsequently uncaged by UV emitted from the UCNP-based carrier upon NIR irradiation. The success of RNAi mediated by the carrier has been verified using immunoblot analysis, which has revealed a significant drop in survivin expression in murine bladder cancer cells (MB49 cell line).[148] With further optimization and characterization, the carrier may be further developed into a mediator of gene therapy in cancer treatment.

Fig. 6

The chemical routes for conjugating the cationic photocaged linker to the surface of the silica-coated UCNP. Abbreviations: DMF, dimethyl formamide; TEA, triethylamine.

In addition to executing gene therapy alone, UCNP-based gene carriers may enable concomitant administration of multiple treatments. This has been exemplified by the positively charged polymer-coated NaGdF4:Yb,Er UCNPs, which have been reported for execution of both photodynamic therapy (PDT) and gene therapy.[164] Results have shown that the carrier can be loaded with the photosensitizer, namely chlorin e6 (Ce6), and with siRNA molecules that can silence Plk1 expression (Fig. 7).[164] Upon excitation by NIR light at 980 nm, cancer cells are killed not only by cytotoxic singlet oxygen generated via resonance energy transfer from UCNPs to Ce6, but also by the anti-tumour activity of the siRNA molecules.[164] More recently, NaLuF4:Gd,Yb,Er UCNPs have been synthesized using carboxyl-containing glutarate as surface ligands, followed by conjugation with cypate, which is a carbocyanine fluorophore with high photothermal conversion efficiency, through a hydrazide bond (Fig. 8A).[165] Due to the magnetic and optical properties of the generated UCNPs, the nanoparticles function as a dual-modality contrast agent for UCL and MRI to guide oncotherapy (Fig. 8B). Moreover, those UCNPs can effectively deliver siRNA molecules, which can act against heat shock protein 70, to cancer cells to enhance cell damage.[165] This damaging effect, along with photothermal ablation led by the conjugated cypate, has triggered significant antitumor activity (Fig. 8C).[165] Such a possibility of executing multiple therapies has illuminated the vast potential of UCNPs in treatment development.

Fig. 7

(A) A schematic diagram depicting the functionalization of UCNPs, co-loading the nanoparticles with Ce6 and siRNA, and the implementation of the combined PDT and gene therapy mediated by the nanoparticles. Abbreviations: EDC, N-(3-dimethylaminopropyl-N′-ethylcarbodiimide)hydrochloride; OA, octylamine; PAA, poly(acrylic acid); PEG, poly(ethylene glycol); PEI, poly(ethylenimine); RISC, RNA-induced silencing complex. (B) Confocal microscopy images of HeLa cells after 4 hours of incubation with UCNP-PEG@2×PEI–Ce6–siRNA. The siRNA adopted was FAM-siRNA (adapted from ref. 164 with the permission of the Royal Society of Chemistry).

Fig. 8

(A) A schematic diagram showing the process of loading cypate and siRNA molecules into UCNPs. (B) (i) Sagittal and (ii) coronal plane MRI of a mouse before and after tail vein injection of the cypate-conjugated UCNPs (10 mg kg–1). (C) Photographs depicting tumour development in mice treated with the (i) control (PBS), (ii) siRNA-loaded UCNPs, (iii) cypate-conjugated UCNPs, and (iv) siRNA-loaded cypate-conjugated UCNPs (adapted from ref. 165 with permission from John Wiley & Sons, Inc.).

UCNP-based gene delivery: limitations and possible solutions

Although, with the advances as presented above, the use of UCNPs as multifunctional gene carriers seems to be within reach, there are hurdles to overcome before clinical translation of innovations in the field can be in full gear. These hurdles are mainly related to either emission efficiency or physiological performance. These challenges as well as possible solutions will be discussed here for future research.

Manipulation of emission properties

Regarding the role played by photoluminescence emission from UCNPs in gene delivery applications as discussed in the preceding sections, the emission efficiency significantly determines the practicality of the nanoparticles in treatment. To date, the tunability of light emission has been attained by strategies such as controlling the pan class="Chemical">dopant concentration, altering the host/activator combination, modulating the size- and shape-induced surface effects, designing the core–shell structures, or utilizing appropriate energy transfer or migration pathways. At this moment, tuning upconversion emission from UCNPs is often accompanied by a loss of the luminescence efficiency,[21] partly due to the deleterious cross relaxation events occurring between pan class="Chemical">lanthanide ions. Yet, with an improvement in the efficiency of suppressing surface deactivations and in addressing different aspects of the colour tuning process, a new dimension brought to light emission tuning will be around the corner.

To induce luminescence emission from UCNPs, light at 980 nm is commonly used at the moment because this wavelength matches with the absorption of the commonly used sensitizer (Yb3+). Light at this wavelength, however, can be absorbed by water, generating heat that may damage biological tissues. The capaspan>city of exciting UCNPs at more tissue-transparent wavelengths is thus highly desired. The feasibility of manipulating the excitation dynamics of UCNPs has been evidenced by the observation that the excitation wavelength of Yb3+-containing UCNPs can be blue-shifted when Yb3+ is further sensitized by Nd3+ as the second sensitizer.[166,167] Such feasibility has been further supported by the development of NaYF4:Nd,Yb@NaYF4:Yb,Tm luminescent nanocrystals that can be excited at 745 nm and emit light at 803 nm for deep tissue imaging.[168] These nanocrystals can not only alleviate the occurrence of attenuation effects relating to visible emission, but can also mitigate the overheating constraint imposed by 980 nm irradiation. Lately, the success of tuning the excitation wavelength of UCNPs has also been achieved by Li et al.,[169] who have fabricated NaGdF4:Yb,Er@NaYF4:Yb@NaGdF4:Yb,Nd@NaYF4@NaGdF4:Yb,Tm@NaYF4 nanoparticles with the core-multishell architecture. Due to the absorption filtration effect of the NaGdF4:Yb,Tm layer, the nanoparticles can give power-density independent orthogonal excitation-emission UCL. Intriguingly, by changing the thickness of the filtration layer, the relative intensities of Er3+-dominated green emission and Tm3+-prominent blue emission can be tuned. These works, along with the emergence of Er3+-sensitized UCNPs which can be excited at multiple wavelengths for light emission,[39,170] have laid a foundation from which future efforts to manipulate the excitation wavelength of UCNPs to those transparent to tissues can be launched.

Optimization of the physiological performance

In addition to the limited luminescence efficiency, another challenge to be met is the poor biodegradability of UCNPs. Diagnostic agents injected into a human body, as required by the Food and Drug Administration (FDA), have to be completely eliminated in a reasonable timeframe.[171] This is to ensure that the area under the exposure curve can be minimized upon total body clearance. Unfortunately, UCNPs in general are not effective to be degraded and eliminated from the body. Earlier studies have reported that PAA-coated NaYF4:Yb,Er nanoparticles with an average diameter of 11.5 nm take 115 days for complete excretion,[172] whereas those with a diameter of around 30 nm fail to be completely excreted even 90 days after in vivo administration.[173] To enhance excretion, the hydrodynamic size of UCNPs may need to be less than 10 nm.[21] Such hydrodynamic size, however, may not be the optimal diameter if gene transfer is involved. This is suggested by an earlier study in which PEI nanogels with mean diameters of 38, 75, 87, 121, 132 and 167 nm have been tested for transfection.[174] The highest efficiency has been shown to be achieved by particles that have mean diameters of 75 and 87 nm. This indicates that the optimal size for gene delivery may not coincide with the diameter of UCNPs that can be eliminated from the body most readily. The situation is more complicated when the UCNPs are used in preclinical and clinical studies, in which the size of the nanoparticles may affect the biodistribution pattern. In general, particles with a diameter smaller than 20–30 nm are more susceptible to renal excretion,[175,176] whereas those having a larger diameter may tend to accumulate in the bone marrow,[177] heart,[178] stomach,[179] kidney,[175,176] spleen[180] and liver.[181]

Based on what has been presented above, the size of the UCNP-based gene carrier is preferred to be small for facilitating elimination from the body upon administration, but from the gene delivery point of view, the optimal size may be defined in a totally different way. Solving such a discrepancy will be an obstacle to be tackled for clinical translation of works on UCNP-based gene delivery. Apart from size optimization, right now a disproportionate amount of resources has been directed towards characterization of the properties of UCNP-based carriers simply as emissive materials. Evaluation of the biocompatibility of the carriers at the organ and body levels is lacking. To extend the use of the carriers from a laboratory context to clinical settings, additional efforts on evaluating the efficiency and long-term safety of the carriers will be the next stage to pursue.

Concluding remarks

Along with the advances in materials chemistry and fabrication,[182-187] there has been a clear trend in gene transfer to move from simply delivering nucleic acid materials to, more recently, systems into which multiple functions have been incorporated. In virtue of their unique optical properties, since the turn of the last century UCNPs have started to make remarkable strides towards gene delivery applications. Although at present the most compelling illustrations of the functional richness of UCNPs as gene carriers are still confined to imaging and light-controlled gene manipulation, with the increasing maturation of fabrication technologies, many of the practical problems (including the poor biodegradability and the low quantum yield) in UCNPs will become solvable someday. The use of UCNPs in gene delivery will ultimately be limited only by the imagination of the scientist.

Conflicts of interest

There are no conflicts to declare.