Spatiotemporal Control of Biology: Synthetic Photochemistry Toolbox with Far-Red and Near-Infrared Light

Abstract

The complex network of naturally occurring biological pathways motivates the development of new synthetic molecules to perturb and/or detect these processes for fundamental research and clinical applications. In this context, photochemical tools have emerged as an approach to control the activity of drug or probe molecules at high temporal and spatial resolutions. Traditional photochemical tools, particularly photo-labile protecting groups (photocages) and photoswitches, rely on high-energy UV light that is only applicable to cells or transparent model animals. More recently, such designs have evolved into the visible and near-infrared regions with deeper tissue penetration, enabling photocontrol to study biology in tissue and model animal contexts. This Review highlights recent developments in synthetic far-red and near-infrared photocages and photoswitches and their current and potential applications at the interface of chemistry and biology.

Graphical Abstract

1. INTRODUCTION

Light provides a noninvasive tool to perturb or probe biological systems by incorporating light-active triggers in endogenous biomolecules or synthetic compounds. By accurately turning on or off the activity of such molecules, researchers are able to analyze downstream molecular biology processes,¹ study participating molecules,² and ultimately alter the cellular dynamics toward therapeutics.³ Over the past two decades, light-sensitive chemical functionalities have been widely incorporated into small molecules, peptides, proteins, and oligonucleotides to control their activities.^4–6 In addition to synthetic approaches, light-responsive proteins inspired genetically encoded protein domains with reversible light-controlled domain–domain interactions, giving rise to optogenetics in molecular biology.^7,8 Optogenetics provides protein-level accuracy without exogeneous chemical pretreatment; however, the large size of proteins and difficulty of photophysical tuning are limitations of these technologies. Synthetic light-responsive triggers, on the other hand, can overcome these drawbacks due to their small sizes and easily tunable properties.

Classic photochemistry relies on photons in the ultraviolet (UV) region of the electromagnetic spectrum due to their higher energy and wider choices of functionalities that absorb UV light (Figure 1a). To date, much success has been reported with photochemical methods to affect and analyze biological activities.^1–8 These triggers have advantages of being small in size, readily accessible synthetically, and stable under ambient light. Despite these benefits, a major limitation is that short-wavelength photons exhibit shallow penetrations through tissue due to the strong scattering and absorption of biomolecules.⁹ Furthermore, UV light can cause damage to biological samples, rendering this strategy less ideal for biological applications.^10,11 As such, more recent efforts have been put into developing photochemistry employing visible light (400–700 nm) for improved biocompatibility due to the lower photon energy of longer wavelength light.^12–15 While photodamage from visible light is much less, tissue scattering and absorption of visible light are still prominent, limiting biological applications to primarily cells and transparent animals.

Figure 1.

Regions of the electromagnetic spectrum, chromophore scaffolds, and photochemical events. (a) Wavelengths of the optical window containing UV, visible, NIR, and SWIR regions. (b) Some common chromophore scaffolds that have been modified to absorb in the NIR region. R = alkyl substituents; R′ = sp² substituents; M = Fe, Zn, or other metals; M′ = Si, Zn, or other metals. (c–e) Schematics of photochemistry modalities applied in biology: (c) photocage, (d) photoswitch, and (e) photolabeling. Grey shapes represent protein of interest.

Following the same trend, photochemistries taking place with near-infrared (NIR, 700–1000 nm) light and even lower energy shortwave infrared (SWIR, 1000–2000 nm, Figure 1a) light hold promise in controlling biological activities in live animal models. While being minimally damaging to biological samples due to the reduced energy, photons in the NIR region can provide greater penetration and/or enhanced resolution in animal experiments as a result of their reduced interaction with tissue.^16–18 This advantage can be even more pronounced with SWIR light.^19–21 As such, manipulating biological processes with photochemistry within the NIR and SWIR windows will greatly enrich scientists’ toolbox in biological research and pharmacology development.

To build a molecule that can respond to NIR or SWIR light, the first step is to absorb these long wavelength photons with a chromophore. Chromophores that have red-shifted absorption can be derived from canonical visible dyes such as Cy3 (trimethine dye), boron-dipyrromethenes (BODIPYs), and porphyrins by introducing extended conjugation systems or other structural modifications that lower the HOMO–LUMO gap, resulting in NIR chromophores such as heptamethine dyes,²² extended BODIPYs or aza-BODIPYs,²³ and phthalocyanines²⁴ (Figure 1b). In many cases, the harvested energy is then released as heat (nonradiative decay) or light (fluorescence or phosphorescence), but with careful design, chemical events can also be triggered on certain excited chromophores, ranging from cis–trans isomerization of double bonds to complete bond scission/formation to form a new structure. Depending on the type of reaction, such chromophores can be categorized as photocages (deprotection upon light activation, Figure 1c), photoswitches (reversible isomerization, Figure 1d), and photolabeling (producing reactive intermediates to form new covalent linkages, Figure 1e). These photochemical processes have been harnessed to alter or probe biological pathways. Photocages can be incorporated either in biomolecules for their light-triggered activation or in synthetic prodrug molecules for their photocontrolled release.^12,25–27 Similarly, photoswitches can be installed in biomolecules or synthetic compounds for reversible, light-induced turn-on or turn-off of activity.^8,28–31 Photolabeling, which occurs when a highly reactive intermediate is released and then captured by a nearby molecule, is frequently utilized in photocrosslinking assays to reveal biomolecule interactions.^2,32

This Review highlights recent developments of photochemistry in the far-red (650–700 nm) and NIR windows that holds potential for regulating in vivo biological processes. We focus on synthetic photocages (section 2) and photoswitches (section 3) using one-photon absorption of far-red and NIR light. To date, no NIR or SWIR photolabeling reagents have been reported, leaving this third application of photochemistry ripe for investigation.^2,32–34 Some of the red-shifted photoprobes are obtained from optimization of their shorter-wavelength predecessors, while others are derived from newly designed scaffolds developed specifically for low-energy photons. We conclude by highlighting potential future avenues to explore in the area of long-wavelength photochemistry and their applications in biology and clinical systems (section 4). Other approaches to far-red and NIR photocages and photoswitches not covered within this Review include those based on two-photon absorption or the use of upconverting nanoparticles. These approaches often require higher intensity light and/or larger probes and have been reviewed previously.^15,35–38 Finally, numerous reviews for UV/visible photoprobes have been published already and can act as comparisons for the results presented herein.^{8,12–14,25–31}

2. FAR-RED AND NIR PHOTOCAGES

2.1. Photocages Cleaved from Excited Dye Molecules.

Direct bond cleavage on excited dye molecules, capitalizing on changes in bond strength upon chromophore excitation, is the most straightforward way to design photocages. This method has been widely successful using UV and visible light on nitrobenzyl, coumarin, and xanthene chromophores.³⁹ The first approach to extend photocages to a more red-shifted BODIPY chromophore was presented by Urano and co-workers. In this work, phenol or coumarin groups were attached to the boron and were shown to be effective leaving groups upon excitation with green (500 nm) light.⁴⁰ While this approach has not yet been expanded to the far-red/NIR, recent improvements have yielded a BODIPY photocage that is cleaved with yellow light and elegantly used to control neural activity in C. elegans.⁴¹

In an alternative approach to photocages based on BODIPY fluorophores, Winter and Weinstain independently reported meso-methyl BODIPY as photocages that can be cleaved with green light (Figure 2a).^42,43 The meso-methyl BODIPY photocages were successfully tuned to improve the quantum yield of the photoreaction to up to 95% (Φ_rel ≤ 0.95).⁴⁴ The underlying mechanism was explained by Klán and co-workers, who discovered that the meso position on BODIPY exhibits a substantial increase in electron density upon excitation, which is analogous to the well-established UV/vis coumarin photocages (Figure 2a).^45,46 The electronic changes upon excitation can be visualized in the calculated HOMO and LUMO distributions. The increase of negative charge in the excited state promotes the dissociation of the leaving group on the meso sp²-bond carbon through mixing with its σ* orbital (Figure 2a).⁴⁵ Using the approach of leaving group installation at the meso-methyl position, BODIPY photocages have been successfully extended to the NIR.

Figure 2.

BODIPY-based photocages. (a) Comparison of photorelease of a leaving group (LG) on coumarin and BODIPY chromophores with Hückel highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) analysis. Adapted with permission from ref 45. Copyright 2015 American Chemical Society. (b) NIR light-triggered release of CO on carboxyl-substituted BODIPY 1.⁴⁶ (c) NIR light-triggered release of H₂S on thiocarbamate-substituted BODIPY 2.⁴⁷ (d) Light-induced release of carboxylic acids from substituted BODIPY 3 and 4.⁴⁸ (e) Far-red light-induced release of amino group on dopamine from its BODIPY photocage 5 and (f) effect of light-activation of 5 on beating frequency of human cardiomyocytes. Reprinted with permission from ref 49. Copyright 2020 The Authors. (b–d) Φ_rel stands for quantum yield of photorelease.

The first NIR-responsive BODIPY photocage, 1, was reported by Klán and co-workers as a carbon monoxide-releasing molecule, containing a NIR-absorbing BODIPY chromophore with a carboxylic acid on the meso position (Figure 2b). Upon irradiation with light as red as 732 nm, the carboxyl on 1 underwent diradical reorganization to release carbon monoxide (CO), a gaseous signaling molecule, both in vitro and in a mouse model.⁴⁶ Thereafter, the same group reported chromophore 2 that contains a thiocarbamate moiety as cargo. This functionality released carbonyl sulfide (COS) upon photodeprotection, which subsequently hydrolyzed into hydrogen sulfide (H₂S), another gaseous signaling molecule (Figure 2c).⁴⁷

To expand the scope of NIR BODIPY cages toward the release of more complex payloads, Winter and co-workers reported a palette of BODIPY-photocages for carboxylic acids spanning absorption maxima from yellow to far-red (e.g., 3 and 4, Figure 2d), where the conjugated arylvinyl substituents carrying electron-donating groups can lower the HOMO–LUMO gap facilitating red-shifted absorption.^48,50 Further modification of the structure with conformational restraint inhibits unproductive relaxation and enhances the quantum yield of photorelease from 1.1 × 10^–3 (8, λ_max 689 nm) to 3.8 × 10^–2 (λ_max 681 nm).⁵¹ More recently, the BODIPY photocage was adapted to far-red photorelease of caged amines by Feringa and co-workers, where the amino group was connected to the meso-methyl position via a carbamate linkage as seen in 5 (Figure 2e). In this report, dopamine was appended to the BODIPY photocage and light-dependent release of dopamine was achieved in human cardiomyocytes. Upon 650 nm illumination, cells treated with 5 had increased beating frequency, similar to those treated with dopamine alone, indicating successful photorelease (Figure 2f).⁴⁹ However, the low photorelease quantum yield under NIR light, originating from competing relaxation pathways associated with the lower-energy excited state, proved to be a major challenge for the use of BODIPY photocages in animals. The hydrophobic nature of this class of molecules also poses a challenge, where water-solubilizing groups need to be introduced at specific sites for their use in a biological context.⁵² As such, further tuning of the red-shifted BODIPY structure to improve its quantum efficiency and bioavailability appears necessary for release of other functionalities and in vivo applications.

While BODIPY is an excellent chromophore for NIR photocages due to its similar molecular orbital changes on excitation compared to chromophores employed in well-established photocages, other NIR chromophores have also been transformed into photocages. Lawrence and co-workers have leveraged the corrin chromophore, related to the porphyrin scaffold, in vitamin B12 as far-red photocages with the help of red fluorophores (e.g., 6, Figure 3a). The axial Co–C bond within vitamin B12 undergoes ligand-to-metal charge transfer upon light activation that can lead to its homolytic cleavage.^53,54 However, vitamin B12 naturally absorbs at wavelengths below 550 nm, preventing its usage as a NIR-responsive probe. Lawrence and co-workers bathochromically shifted the vitamin B12 photoprobes by appending far-red and NIR fluorophores which can undergo energy transfer with the corrin chromophore. These fluorophore “antennae” can be either covalently attached to the vitamin B12 cobalamin ring as in 6 (Figure 3a)⁵⁵ or placed nearby the Co complex and the energy transferred intermolecularly.⁵⁶ As an example, BODIPY-substituted cobalamin 6a (Cbl-Bod), which localized in the endosome, was shown to release a mitochondria-localizing BODIPY fluorophore upon cleavage of the Co–C bond with 650 nm irradiation (Figure 3b).⁵⁵ Such a platform has been expanded to light-controlled change of membrane permeability for drug release⁵⁷ and also light-removal of a blocking peptide for protein activation.⁵⁸

Figure 3.

Photocages from far-red and NIR chromophores. (a) Structure of vitamin B12–fluorophore conjugate 6 for Co–C bond photocleavage and far-red BODIPY-containing counterpart 6a (Cbl-Bod) and (b) translocation of BODIPY fluorescence from endosome to mitochondria upon light irradiation. Adapted with permission from ref 55. Copyright 2014 WILEY-VCH Verlag GmbH & Co. KGaA. (c) Silicon phthalocyanin-CA4 conjugate 7 as a NIR photocage (without O₂) or photosensitizer (with O₂) and (d) its cytotoxicity in HeLa cells under half-well illumination with (top) or without (bottom) O₂. Dark region indicates live cells. Adapted with permission from ref 59. Copyright 2016 The Authors. (e) Photoinduced cleavage of Pt–O bond on cyanine-platinum complex 8 (IR797-Platin) and (f) its cytotoxicity on C-33 A cells with or without light irradiation. Reprinted with permission from ref 60. Copyright 2018 Wiley-VCH Verlag GmbH & Co. KGaA. (g) Quinone photocage 9 with trimethyl lock ring closure forming lactone to release an alcohol.⁶¹ (h) Quinone photocage 10 forming o-quinone methide to release benzoic acid.^62,63

Similarly, Schnermann and co-workers have developed NIR photocages based on silicon phthalocyanine chromophores, leveraging a weakened axial Si–O bond upon excitation (Figure 3c). While silicon phthalocyanines primarily function as a photosensitizer in the presence of O₂ (see section 2.2.1), without O₂ the absorbed energy leads to the cleavage of the axial Si–O bond, releasing its axial substituent as an alcohol (Figure 3c).⁵⁹ Schnermann and co-workers prepared silicon phthalocyanine 7 and demonstrated the release of combrestastatin-A4 (CA4), a cytotoxin, under hypoxia conditions with 690 nm illumination. Compared to normoxia conditions where photosensitized singlet oxygen killed cells locally, in the absence of O₂ photoreleased CA4 diffused intercellularly, resulting in a wider area of cytotoxicity (Figure 3d).⁵⁹

Cyanine scaffolds have also been employed as photocages. In an example by Hartman and co-workers, a Cy7 chromophore was used to harvest NIR photons and direct the energy to the displacement of Pt–O bonds on the fluorophore in complex 8 (IR797-Platin, Figure 3e). The coordination bonds between the Cy7 bidentate ligand and Pt(II) were disrupted upon NIR light illumination, releasing Pt(II) species that showed enhanced cytotoxicity in cells compared to dark conditions (Figure 3f).⁶⁰ The Cy7 chromophore has also been fused with the CO-releasing flavonol by Klán and co-workers to facilitate CO release upon NIR excitation rather than the traditional visible excitation necessary to release CO from free flavonol chromophroes.⁶⁴

Last, it has long been known that quinones are able to undergo hydrogen abstraction from nearby C–H bonds; the resulting phenol can be further engineered to release a leaving group to act as a photocage (Figure 3g,h). The quinone absorption can be pushed into the visible range by amine or sulfide substitution, with its broad peak shape enabling photocages under red light. Dougherty and co-workers developed a trimethyl o-hydroxydihydrocinnamic derivative (9) as a photocage within visible range. Its favorable conformational restraint after photoreduction results in efficient nucleophilic replacement to liberate alcohol or amine cargo (Figure 3g).⁶¹ Similarly, incorporating an alkene conformation lock also leads to efficient visible photocages.⁶⁵ In a different approach, Steinmetz and co-workers designed 10a, which releases benzoic acid and o-quinone methide after photocyclization under 458 nm (Figure 3h).⁶³ The structure was further tuned by Kalow and co-workers to afford 10b with an absorption maximum at 534 nm and efficient cleavage under 636 nm illumination (Figure 3h).⁶² Although currently the quinone photocages still operate in the visible range and in a water/organic solvent mixture, they hold promise to be extended to the NIR region as redox-active photocages that do not rely on canonical dye structures.

Taken together, multiple platforms have been developed as far-red and NIR photocages as opposed to a few privileged chromophores for their UV or near-UV counterparts. The application of these red-shifted photocages in animals is still scarce due to their early development. Aside from further optimization to provide improved decaging efficiency, these far-red and NIR protection groups, especially for those incorporating ester functionality or labile coordination bonds, require thorough evaluation and/or improvement of their dark stability in conditions mimicking cellular and tissue contexts to obtain precise light-mediated control in complex in vivo environments. Nonetheless, the wider choices in NIR photocages can offer valuable tools for far-red and NIR photoactivation of bioactive molecules in vivo.

2.2. ¹O₂-Dependent Photocages.

2.2.1. Far-Red and NIR Photosensitizers.

Visible and NIR chromophores that are able to access the triplet state can sensitize oxygen and produce singlet oxygen (¹O₂) upon excitation.^66,67 Being 94.3 kJ/mol (22.5 kcal/mol) above ground state, this highly reactive species readily undergoes [2 + 2] or [4 + 2] cycloadditions and ene reactions that lead to the oxidation of nearby molecules.⁶⁸ Depending on the oxidized substrate, the consequence can be (1) photobleaching of the chromophore; (2) cellular stress coming from oxidation of the nucleobase in DNA and/or aromatic side-chains in proteins [i.e., photodynamic therapy (PDT)]; or (3) cleavage of an engineered ¹O₂-sensitive linker leading to opportunities for controlled photocleavage. It is the latter application which is relevant to this Review.

Multiple dye structures have been utilized to build photosensitizers, including porphyrins, phthalocyanines, and halogenated dyes such as xanthenes, cyanines, and BODIPYs.^69,70 Zinc- and silicon-phthalocyanines, in particular, show appreciable ¹O₂ quantum yields (>0.3) with far red absorption (600–700 nm), making them notable red-shifted photosensitizers in PDT applications.^24,71,72 Moving beyond 700 nm to achieve deeper penetration in the NIR region, multiple phthalocyanine, cyanine, (aza-)BODIPY and nanoparticles have been put forward as NIR photosensitizers and have been reviewed elsewhere.⁷³ These long-wavelength photosensitizers, in addition to facilitating deeper PDT, can be modified with therapeutics connected by an ¹O₂-cleavable linker to enable a dual therapeutic effect (Figure 4d).

Figure 4.

Photocages based on unsaturated ¹O₂-cleavable linkers. (a) Electron-rich ethylenes as ¹O₂-cleavable linkers. (b) Photosensitizer–drug conjugate containing aminoacrylate and its ¹O₂-mediated cleavage. (c) Structure of aminoacrylate-linked photosensitizer–prodrug conjugate 11 [Pc-(L-CA4)₂].⁸⁶ (d) Scheme showing the extended damage radius of releasable photosensitizer–drug conjugate in PDT. (e) Image of cells treated with 11 [Pc-(L-CA4)₂] and its noncleavable analog, Pc-(NCL-CA4)₂, illuminated from one side at 690 nm. Reprinted with permission from ref 86. Copyright 2016 Elsevier Ltd. (b,d) PS stands for photosensitizer.

2.2.2. ¹O₂–Cleavable Linker as Photocages.

With the wide selection of photosensitizers, building a red-shifted photocage can be simplified to finding a robust, selective ¹O₂-cleavable linker to combine with a photosensitizer and cargo moiety. Since there is a correlation between electron density and ¹O₂ reactivity, heteroatom-substituted, electron-rich ethylene moieties are common ¹O₂ sensitive linkers. These functionalities undergo a [2 + 2] cycloaddition with ¹O₂, followed by a retrocycloaddition to release two aldehyde products (Figure 4a).^74–76

Early studies with singlet oxygen cleavable linkers employed vinyl ether^77,78 and vinyl dithioether^79,80 functionality. More recently, You and co-workers designed aminoacrylate as an efficient, easy-to-prepare ¹O₂-cleavable linker for the photosensitized release of a drug molecule protected on its hydroxyl group. The electron-rich aminoacrylate can be synthesized via yne-amine reaction and cleaved by ¹O₂ oxidation (Figure 4b).⁸¹ Other reports have incorporated aminoacrylate into a variety of drug-linker-photosensitizer molecules for photocontrolled drug release during PDT.^82–89 Compared with simple PDT that relies on short-lived ¹O₂ with an effective radius of <1 μm,⁹⁰ the released cell-permeable drug molecule can diffuse intracellularly and provide extra potency (Figure 4d). In particular, when the sensitizer in the photoreleasing composite is a silicon phthalocyanine, the aminoacrylate linkage can be effectively cleaved under 690 nm illumination, facilitating drug release.^83–88 As an example, silicon phthalocyanine was conjugated to CA4 on its axial positions via the aminoacrylate linker to form photoactivable prodrug 11 [Pc-(L-CA4)₂] (Figure 4c).⁸⁶ When MCF-7 cells treated with 11 were illuminated on one side, the liberated CA4 toxin was able to diffuse and kill most cells in the dish, whereas the noncleavable control only resulted in cell death near the most intensely illuminated side of the culture (Figure 4e).⁸⁶ Similarly, the CA4 moiety was also conjugated via an aminoacrylate linker to both an intracellular glutathione-activable photosensitizer and a biotin targeting group to form a multifunctional therapeutic agent.⁹¹

Alternatively, other oxidation-labile functional groups can be used for ¹O₂-dependent photorelease. One such example is a thioketal that releases an aldehyde or ketone and two thiols upon ¹O₂ oxidation, which can be further engineered in cascade reactions to release alcohols and amines (Figure 5a).^92–96 Thioketals are also responsive to cellular reactive oxygen species (ROS), leading to lower selectivity upon photoactivation as compared to the electron-rich ethylene trigger.^97,98 Nonetheless, such a linker has been leveraged for red and NIR light-activated drug release for cancer treatment,^92–96 where light-induced ¹O₂ can boost cleavage in tumor cells that usually exhibit higher ROS levels due to factors such as higher metabolic activity, increased signaling events, and inflammation. Among these reports, Liu and co-workers combined a far-red BODIPY chromophore and a drug moiety, camptothecin (CPT), via a thioketal linker to make 12 (BDP-TK-CPT, Figure 5b), which releases free CPT upon far red illumination (Figure 5c).⁹⁶ In mice bearing H22 tumors, free CPT only showed slight tumor inhibition due to limited tumor accumulation, whereas 12 and its noncleavable analog were not effective in the dark. In contrast, upon 670 nm light radiation, a mouse treated with 12 or its noncleavable analog both showed tumor reduction, with 12 being the most effective due to its synergistic effect of ¹O₂ generation and CPT photorelease (Figure 5d).⁹⁶

Figure 5.

Photocages based on thioketal ¹O₂-cleavable linkers. (a) Thioketal as ¹O₂-cleavable linker and its triggered release of amino and hydroxyl groups. (b) Structure of a thioketal-linked photosensitizer–drug conjugate, 12 (BDP-TK-CPT). (c) Scheme showing the release mechanism of 12. (d) Photographs of excised H22 tumors in mice after treatment with control, CPT, 12, or its noncleavable analog BDP-CC-CPT, with or without 670 nm light irradiation. Adapted with permission from ref 96. Copyright 2021 Elsevier Masson SAS.

In summary, ¹O₂-cleavable linkers represent a straightforward method to build up photocages from photosensitizers. With their longer wavelengths, it becomes necessary to evaluate the efficacy of these ¹O₂-dependent photocages in whole animal models to push toward their clinical applications. These photocage designs will also benefit from more accessible and selective ¹O₂-cleavable linkers and reliable NIR (or SWIR) photosensitizers. While the dark stability of these carboxylate and carbonate photocages in the presence of esterases requires evaluation during biological applications, the synergistic effects of released drug and ¹O₂ production shows promise for enhanced efficacy of this set of modular NIR-responsive prodrug molecules.

2.2.3. Cyanine Photooxidation-Based Photocages.

More often than not, photobleaching of a dye molecule by its sensitized ¹O₂ is harmful, and much work has been done to prevent these processes for applications in imaging.⁹⁹ However, chromophore photobleaching can be tailored to direct the oxidized intermediate toward the release of a certain functionality, which constitutes a ¹O₂-dependent photocage. Schnermann and co-workers have elegantly applied this approach to NIR cyanine dyes.

Cyanine dyes are a class of strongly absorbing dyes with a polymethine bridge between two heterocycles. Schnermann and co-workers have developed scaffold 13, a 4′-substituted heptamethine cyanine (Cy7), as a ¹O₂-dependent NIR photocage (Figure 6a).¹⁰⁰ In this design, sensitized ¹O₂ photobleaches the Cy7 dye to give aldehyde oxidation products 13a and 13b, which are susceptible to hydrolysis to expel 12c that undergoes cyclization to form urea 13d and release a phenolic hydroxyl group in 13e (Figure 4a).¹⁰⁰ In followup work, the photocaged functionality was also expanded to aryl amines.¹⁰¹

Figure 6.

Heptamethine cyanine dye photocages. (a) Mechanism of photorelease of phenol from Cy7 derivative 13.¹⁰⁰ (b) Scheme of photocontrolled release of 4-hydroxycyclofen from 14 toward activation of EGFP expression by Cre/LoxP recombination^100,102 and (c) fluorescence microscopy showing EGFP expression in cells illuminated under 633 nm light (circled in yellow). Reprinted with permission from ref 102. Copyright 2018 Wiley-VCH Verlag GmbH & Co. KGaA. (d) A cyanine-based photoactivatable antibody–drug conjugate 15 and (e) its tumor-inhibitory effect under different radiation intensities. Reprinted with permission from ref 103. Copyright 2015 WILEY-VCH Verlag GmbH & Co. KGaA. (f) A cyanine-based photoactivatable antibody–drug conjugate 16 and (g) its effect on tumor volume reduction under different doses. Reprinted with permission from ref 104. Copyright 2017 American Chemical Society.

This oxidation-based photochemistry has found utility in diverse applications. Initially, Schnermann and co-workers used the Cy7 photocage to protect 4-hydroxycyclofen in 14 to enable NIR light controlled gene expression (Figure 6b). Upon 633 nm irradiation, photoreleased 4-hydroxycyclofen activated Cre-ER/LoxP recombination and resulted in eGFP expression (Figure 6c).^100,102 This photorelease system has also been applied to spatiotemporal chemotherapeutic delivery. CA4 antibody-prodrug conjugate 15 was prepared (Figure 6d) and introduced to mice bearing A431 tumors. Upon tumor targeting, CA4 release was induced by 690 nm irradiation, leading to strong tumor reduction (Figure 6e).¹⁰³ In another report, NIR light-activated antibody–drug conjugate 16 (Figure 6f) was designed by introducing two fused benzene rings on the heterocycles to extend the conjugated system and further red-shift the absorption, two N-ethyl groups to reduce background hydrolysis, and sulfonate groups for decreased dye aggregation. The resulting compound targeted EGFR⁺ tumors with effective duocarmycin release under 780 nm illumination and showed high antitumor efficacy (Figure 6g).¹⁰⁴ Multiple successes were also reported based on similar Cy7-photocage designs, utilizing other prodrugs^105–107 or targeting motifs¹⁰⁸ for photocontrolled in vivo drug release. With these photocage examples, century-old cyanine dyes find even more biological applications beyond their wide use in imaging experiments. It is foreseeable that the tunability of cyanine dyes will expand the utilities of photocages with varying wavelengths, brightness, cleavage efficiency, and biodistribution properties such that they can be tailored toward specific biological studies.

3. FAR-RED AND NIR PHOTOSWITCHES

Distinct from photocages, photoswitches achieve control of intracellular processes by reversible conformational or structural changes to the chromophore. Small molecule photoswitches that operate with UV light have already been widely explored to control protein activity; in contrast, their visible-light counterparts have only been developed in recent years. Following these new designs, photoswitches that can operate with far-red and NIR light have emerged. Due to the early age of this photoswitch subcategory, most of the reported long-wavelength photoswitches are still chemical concepts with limited biological uses. Nonetheless, we expect increased utility of NIR photoswitches in biological settings as this technology matures to enable noninvasive reversible control in animal models. Here, we focus only on photoswitches that are directly excited by far-red or NIR light. Notably, azobenzene switches excited by absorbing two photons^109,110 and spiropyran switches on upconverting nanoparticles^111,112 have also enabled NIR-excitable photoswitches. These approaches generally require higher intensity light and are beyond the scope of this Review.

3.1. Azobenzene Photoswitches.

Azobenzene is by far the mostly studied photoswitch moiety with compatibility in many environments, which has been extensively explored in biological settings.¹¹³ Its photocontrolled conformational change upon cis/trans-isomerization results in both a 3.5 Å difference in para- positions and an orientation change (Figure 7a),¹¹⁴ which has been utilized to modulate protein/nucleic acid interactions, peptide or small molecule/protein interactions, ion channel activity, and many others.^{29–31,115–119} As a result of the existing success of azobenzenes in biology, modification of this photoswitch toward far-red and NIR wavelengths holds great promise for the photocontrol of biomolecules in vivo.

Figure 7.

Azobenzene photoswitches. (a) Photoisomerization of azobenzene photoswitches. hν1 and hν2 stand for light to switch to and from cis-azobenzene; k_bT stands for thermal relaxation. (b) Absorption spectra of cis (dotted) and trans (solid) isomers of unsubstituted azobenzene in ethanol. Reprinted with permission from ref 115. Copyright 2011 Royal Society of Chemistry. (c) Structure of visible and far-red azobenzene photoswitches 16a–16d.^120–122 (d) Absorption spectra of cis- (dotted) and trans- (solid) 16a in DMSO. Reprinted with permission from ref 120. Copyright 2011 American Chemical Society. (e) Photoswitching of a peptide containing 16b (left), its change of absorption spectra in phosphate buffer (middle) and change of CD spectra (right) upon irradiation. Reprinted with permission from ref 121. Copyright 2013 American Chemical Society. (f) Switching of 16d in siRNA targeting luciferase and its inhibition effect on luciferase expression in HeLa cells over 24 h under constant light (gray), the initial 2 h of light (yellow), or a dark environment (black). Reprinted with permission from ref 122. Copyright 2020 Wiley-VCH Verlag GmbH & Co. KGaA. (g) Equilibrium of azobenzene–azonium cation species. (h) Hydrogen bond in ortho-oxygen substituted azonium cation and structures of azonium photoswitches 17 and 18.^123,124 (i) Absorption spectra of 17 at different pH. Reprinted with permission from ref 123. Copyright 2013 Wiley-VCH Verlag GmbH & Co. KGaA. (j) Absorption spectra of 18 at different pH.¹²⁴ Reprinted with permission from ref 124. Copyright 2017 American Chemical Society.

The trans- and cis-isomers of azobenzene show overlapping yet distinct enough absorption profiles that photocontrol can be achieved. For unsubstituted azobenzene, both conformations show strong π → π* absorption in the UV region and weak n → π* absorption centered at 440 nm (Figure 7b). The thermodynamically stable trans state can be switched into a photostationary cis state upon excitation of n → π* or π → π* bands, the former usually being less efficient due to the overlap of absorption between the isomers. The cis isomer can be switched back to the lower-energy trans state by irradiation at 450 nm or by thermal relaxation. It is often desirable if the relaxation kinetics of the cis isomer, described by the half-life in the absence of light, are slow to give sufficient time for altering biological processes without the need for intense and/or continuous illumination.¹²⁵

One way to red-shift the operating wavelength of azobenzenes is to separate the n → π* bands of trans- and cis- isomers so that the trans- isomer can be selectively excited to convert to the cis-conformation. This can be achieved by appending substituents on the four ortho-positions to affect the steric twisting of the cis (and/or trans) isomer. Woolley and co-workers reported that incorporation of methoxy groups on the aryl rings of amdio-azobenzene in 16a (Figure 7c) leads to a ~35 nm red-shift of the n → π* absorption band of the trans isomer centered at 480 nm. With its cis isomer exhibiting an absorption peak with a 36 nm blue shift (Figure 7d), the trans isomer can be selectively excited and switched to the cis isomer using longer-wavelength green light, showing stability over multiple rounds of switching with minimal bleaching.¹²⁰ Such design, when incorporated into peptides as in 16b (Figure 7c), showed another 20 nm bathochromic shift with a small tail absorption in the far-red, enabling switching under 635 or 660 nm of irradiation (Figure 7e). Upon trans- to cis-isomerization of the modified peptide with far-red illumination, a loss of helicity was observed by circular dichroism (CD) spectroscopy (Figure 7e). This process can be reversed under blue light irradiation.¹²¹ Alternatively, chlorine or fluorine substituents can also be introduced onto the azobenzene to elicit a similar band separation with a visible to far-red switching window (16c, Figure 7c).^126–131 Desaulniers and co-workers employed chlorinated azobenzene 16d to create photoswitchable siRNA (Figure 7c).¹²² Due to its similar conformation to RNA, the trans isomer of 16d in the sense strand is able to form an siRNA duplex with an antisense strand, which can be recognized by the RNAi pathway to effectively inhibit firefly luciferase expression, whereas when excited under 660 nm, the cis isomer disrupts the double-stranded siRNA, preventing its uptake by the RNAi pathway and thereby showing a reduced inhibition effect (Figure 7f). While an exciting application of photoswitches, 16d is limited by the relatively fast thermal isomerization of the cis isomer to the trans isomer (estimated t_1/2 = 2 min), which required continuous red-light exposure during the assay to sustain the cis state. The red-shifted ortho-substituted azobenzenes have also found other applications in the creation of photoresponsive polymers^132–135 and bioactive molecules.^127,128,136

Another approach to red-shifting azobenzenes is to alter their pK_a to favor the protonated azonium ion. The azonium ion also exhibits photoswitching activity, and its isomers exist in equilibrium with azobenezene (Figure 7g). Most azonium ions have a pK_a below 3, rendering them unfavorable in physiological conditions; however, upon the addition of ortho-methoxy groups to 4-amino azobenzenes (such as 17, Figure 7h) the pK_a is increased to 7.2 due to a combination of the electron-rich aryl rings and stabilizing hydrogen bonds between the azonium and ortho-methoxy group (Figure 7h). As a result, an appreciable amount of azonium species of 17 are present at neutral pH (Figure 7i), enabling effective photoswitching over many cycles using less intense red light.¹²³ Following these approaches, Woolley and co-workers further designed 18 with substituted ethers in a fused dioxane ring for improved conjugation on diaminoazobenzene (Figure 7h). This molecule (18) had a pK_a of 6.7 and displayed an absorption maximum around 600 nm at pH 7.4 with a considerable tail in the NIR range, making it capable of being photoswitched upon 720 nm irradiation under physiological conditions (Figure 7j).¹²⁴

Alternatively, red-shifted azobenzene photoswitches can be built from diazocines, the bridged form of azobenezene (Figure 8a). As a complement to linear azobenzene switches, diazocines have a thermodynamically stable cis configuration rather than trans configuration due to the ring strain. Consequently, longer wavelength light switches these diazocines from the photostationary trans state back into the thermodynamically stable cis state. During repeated switching cycles, the unsubstituted diazocine 19 can undergo trans to cis isomerization at 520 nm after it is switched to the trans isomer under 385 nm illumination, as opposed to 365 nm for simple linear azobenzenes to switch to the cis isomer (Figure 8a,b).¹³⁷ To further red-shift the absorption, Herges and co-workers introduced oxygen (20) or sulfur (21) into the diazocine (Figure 8a), yielding trans isomers that exhibit broad absorption into the far-red region, allowing for efficient switching under 660 nm light with good photostability over many switching cycles.¹³⁸ In their later reports, Herges and co-workers incorporated nitrogen as the heteroatom, either as amines, amides or carbamates, affording heterodiazocines all showing appreciable reversibility. Among these photoswitches, the methylamine version is the most red-shifted, displaying an absorption maximum around 570 nm for the trans isomer, and can be switched into its cis isomer with 740 nm irradiation due to its strong tail absorption in the NIR window (22, Figure 7a,c).¹³⁹

Figure 8.

Other azo-containing photoswitches. (a) Structures of diazocine 19 and heterodiazocines 20–22 as red-shifted photoswitches.^137–139 (b) Absorption spectra of cis (blue)- and extrapolated trans (red)-19 in n-hexane. Black lines show the thermal relaxation from the 385 nm photostationary state (green) after 2–15 h. Reprinted with permission from ref 137. Copyright 2009 American Chemical Society. (c) Absorption spectra of cis (black)- and extrapolated trans (blue)-22 in CH₂Cl₂. Photostationary state under 405 nm is shown as dotted red line.¹³⁹ Reprinted with permission from ref 139. Copyright 2019 American Chemical Society. (d) Synthesis and structures of azo-BF₂ photoswitches with switching wavelength in visible (24)¹⁴⁰ and NIR (25)¹⁴¹ regions. (e) Absorption spectra of dark (black) and 570 nm-irradiated (blue) 24 in CH₂Cl₂. Reprinted with permission from ref 140. Copyright 2012 American Chemical Society. (f) Absorption spectra of dark (black) and 710 nm irradiated (red) 25 in CH₂Cl₂. Reprinted with permission from ref 141. Copyright 2014 American Chemical Society.

Other methods toward red-shifted azo-containing photoswitches include work by Aprahamian and co-workers, who have elegantly employed Lewis acid coordination to not only induce a bathochromic shift but further spectrally separate the trans and cis isomers and slow relaxation rates. Treatment of hydrazone 23 with BF₃·OEt₂ led to BF₂-azo compounds 24 (Figure 7i). Compound 24 proved to be an effective photoswitch whose trans isomer absorbed at λ_max,abs = 530 nm, 50 nm red of the cis isomer. The well separated cis-/trans-absorption profile enabled photoswitching cycles with 450/570 nm light. Unfortunately, 24 was prone to hydrolysis, preventing its use in aqueous conditions.¹⁴⁰ Leveraging the successful strategy of appending electron-donating groups, Aprahamian and co-workers were able to prepare NIR switchable azo compounds.¹⁴¹ Notably, with an electron-donating dimethylamino group at the para position, the azo-derivative showed a 680 nm absorption maximum, allowing for effective switching cycles with 710 nm irradiation and thermal relaxation (25, Figure 7i).¹⁴¹ Fortunately, the electron-donating substituents also reduced the hydrolysis rate of the heterocycle, enabling photoswitching in a PBS/acetonitrile mixture.

In summary, through chemical modification, the operating wavelengths of azobenzenes and related photoswitches have been shifted dramatically from the UV region to the far-red/NIR window. Despite such improvements, full switching to the active state is still limited by the ratio of trans to cis products achieved in the photostationary state. Moreover, the usually fast thermal relaxation of these red-shifted azobenzene switches necessitates constant red-light illumination to keep the azobenzene in the switched-on state. This can often be tolerated in cell experiments, whereas in in vivo assays constant illumination can be a significant limitation, especially for slower downstream biological processes. Other potential concerns regarding use in vivo include the photostability of the chromophores and chemical stability toward glutathione and complex protein environments. Nonetheless, we anticipate that future optimization of red-shifted and NIR azobenzenes will eventually lead to more robust photoswitchable biomolecules that can facilitate spatiotemporal control in tissue and animals.

3.2. Photoswitches Based on Other Structures.

Adapting canonical UV photoswitches to shift their wavelengths to the far-red and NIR region is challenging since considerable changes in energy are necessary. Additionally, as discussed in the previous section, red-shifting azobenzenes often leads to fast thermal relaxation, making it more difficult to retain the switched-on state. Against this backdrop, several new scaffolds of photoswitches are being explored to create alternative far-red to NIR photoswitches with potentials for future development toward photoswitching in physiological conditions.

Read de Alaniz and co-workers designed a photoswitch structure named the donor–acceptor Stenhouse adduct (DASA; 26, Figure 9a).¹⁴² This molecule undergoes isomerization from a linear, colored state (26a, λ_max = 545 nm) to a cyclized, colorless form, 26b, under visible light irradiation, which thermally relaxes back to its colored form in nonpolar solvents. This switching cycle in toluene proceeds with negligible material degradation. By replacing the terminal amino group (donor side) with a more electron rich fused aniline structure, the absorption maxima of the resulting molecules 27a–27e can be tuned from 582 to 669 nm (Figure 9b).¹⁴³ Changing the acceptor side of the molecule into other carbon acids provides photoswitches with a tunable dark-state on/off species ratio, with the most red-shifted version capable of photoswitching upon irradiation with 710 nm light.¹⁴⁴ These newer-generation DASAs also expand the bidirectional switching compatible solvents from nonpolar to polar solvents including THF and acetonitrile. DASA photoswitches have also been incorporated in polymers, but their solvent-dependent photochemistry is still under investigation, and photoswitching in water for biological applications still remains a challenge.

Figure 9.

Non-azo far-red and NIR photoswitches. (a) Photoswitching of donor–acceptor Stenhouse adduct 26 with visible light.¹⁴² (b) Structures of Stenhouse adducts 27a–27e with varying absorption covering 582–669 nm.¹⁴³ (c) Structures of indigo 28a and thioinidgo 28b as photoswitches, and a far-red photoswitch example 28c.¹⁵⁰ (d) Structures of hemiindigo 29a and hemithioindigo 29b as photoswitches, and far-red photoswitch example 29c.¹⁵² (e) Photoswitching of dihydropyrene 30 with NIR light.¹⁵⁴

Another notable family of photoswitches include indigo and its derivatives. Aside from being a blue textile dye that absorbs in the red region, selected indigos interconvert between stable trans and photostationary cis isomers, providing a flip of the conjugated plane on one side during photoswitching (28a, Figure 9c). The absorption of the trans isomer is slightly red-shifted compared to the cis isomer, allowing for its selective excitation.¹⁴⁵ Such photoswitching is reported in a handful of N-substituted indigos, such as diacyl,¹⁴⁶ dimethyl,¹⁴⁷ di(t-butoxycarbonyl),^148,149 and diaryl indigos.¹⁵⁰ Notably, the introduction of electron-withdrawing N-aryl substituents extends the thermal half-life of the trans isomer to a few hours as compared to a few seconds for the alkyl counterparts, enabling photoswitching in acetonitrile from the trans to cis isomer under 660 nm illumination.¹⁵⁰ Alternatively, the thermal half-life of the photostationary isomer can be improved in indigo derivatives, such as thioindigo (28b, Figure 9c),¹⁵¹ hemiindigo (29a, Figure 9d),¹⁵² and hemithioindigo (29b, Figure 9d).¹⁵³ The Z- and E-isomers of hemiindigo 29c showed a half-life of days to months depending on the solvent, and its thermally equilibrated mixture can be converted to 99% Z isomer at wavelengths up to 680 nm in DMSO (Figure 9d).¹⁵² Although the low solubility of these planar chromophores in water and their strong solvent dependency limit their wider application, these long-wavelength absorbing chromophores hold promise for the future tuning toward NIR photoswitches in physiological conditions.

In a different approach, Hecht and co-workers prepared a dihydropyrene derivative 30 with both electron withdrawing and donating substituents on the 2 and 7 positions of the ring, which can be switched from a colored, closed form 30a to a colorless, open form 30b under NIR light (Figure 9e). This push–pull molecule exhibits a red-shifted absorption in the NIR region, with operating wavelengths from 718 nm in n-heptane to 815 nm in DMSO.¹⁵⁴ By incorporating different electron withdrawing and donating substituents at either 2 and 7 or 4 and 9 positions, Hecht and co-workers were able to achieve a photoswitch that operates with a wavelength as long as 900 nm.¹⁵⁵ The geometry change in this photoswitch is subtle compared to other photoswitches, making it potentially useful for switching inside protein pockets.

4. CONCLUDING REMARKS AND FUTURE PROSPECTS

Harnessing photochemistry in the far-red to NIR and potentially SWIR regions is a rapidly growing area in chemical biology. While fluorescence imaging has already made the migration from the visible to NIR and SWIR, photochemistry is just beginning a similar transition. In this Review, we have highlighted several recent examples of photocages and photoswitches in the far-red and NIR regions. While some of these probes originate from existing UV/visible photoprobe design, other new platforms are inspired by the joint knowledge of photo-, bio-, and organic chemistry. The growing number of recent reports on new far-red and NIR photochemistry tools marks an excellent start for this field of chemical biology.

However, many aspects of NIR photochemistry still deserve further development. From a chemist’s perspective, new compounds that harvest long-wavelength photons are un-avoidably larger, often leading to decreased solubility and synthetic accessibility, both of which limit their applications in biological studies. Modular, aqueous-soluble building blocks will greatly facilitate the translation of the design of NIR photochemical tools to wider biological applications. Biological applications of these photoreactions are ripe for exploration. Currently, many NIR photocages are designed to achieve light-induced cell death; however, other uses such as controlling gene expression or protein activity will enable a more precise study of molecular biology and signaling events in vivo. NIR photoswitches hold potential for remote control of cellular activities in contrast to invasive UV/visible optogenetics with light fiber; however, their applications are still scarce, largely owing to the young age of this new category of photoswitches.

Beyond operating in NIR light, future development of photocages and photoswitches can expand into the SWIR region where photons experience even less scattering, deeper penetration depth, and higher spatial resolution inside tissues or whole animals as showcased by SWIR imaging.^19–21 Although facing the challenge of reduced photon energy, controlling biological activities with these benefits will allow for a higher signal-to-noise ratio for biological research and less side-effects for therapeutics. Given the ever increasing demand in controlling the activity of biomolecules and drugs in biology and pharmacology encompassing cell, tissue, and whole animal systems, we envision that longer-wavelength photochemistry tools will continue to be reported to help decipher and manipulate biological processes in living systems.

KEYWORDS

Photocage

a chemical protecting group that can be activated and cleaved upon light illumination

Photoswitch

a molecule that reversibly changes its chemical structure and/or geometry upon light illumination

Photoactivation

a process to activate an inert substrate, including biomolecules and pharmaceutical compounds, to its active form by light illumination

Photosensitizer

a molecule that absorbs light followed by transferring the energy into another nearby molecule from its excited triplet state; in many cases, photosensitizers refer to photosensitizers of diatomic oxygen

Singlet oxygen

excited states of diatomic oxygen with all electrons spin-paired, which includes the lowest excited state of diatomic oxygen

Drug delivery

methods toward transporting a pharmaceutical compound to its target site to achieve desired therapeutic outcome with less off-target side effects

Far-red

electromagnetic irradiation at the red end of the visible spectrum, often defined as light with wavelengths between 650–700 nm

Near-infrared

infrared electromagnetic irradiation with wavelengths closer to visible light, often defined as 700–1000 nm