Fixing a Photosensitizer Unlocks and Localizes Its Lethality.

The combination of genetic encoding with sophisticated photochemistry can yield powerful ways to elucidate biological function as it enables the perturbation of complex biological systems with very high temporal and spatial precision. In living organisms, a few cells, or even a single one, can play a crucial role, for instance, as progenitor cells in developing tissue or as pacemaker cells in neural networks. Therefore, the ability to control the function of a single cell or a small subset of cells in tissue is an important goal in biology. Most radically, this can be done through their controlled destruction, also known as ablation. Urano and co-workers now report a method that allows for highly effective and precise ablation of cells in vivo.(1) The ablation of cells with single-cell resolution can provide important insights into biological networks.

Cell ablation has a storied history in biology and medicine. It can be achieved through targeted expression of a lethal factor, such as diphtheria toxin, which is a comparatively slow process.[2] Irradiation with light affords much faster action but has its own challenges. Direct ablation with high-powered lasers is straightforward, but it results in the destruction of virtually every cell along the light beam and is more applicable to larger tissues. Two-photon scanning microscopes, which focus high-intensity laser light of long wavelength onto very small volumes, can be used for accurate 3D ablation of cells. However, this is technically challenging and requires expensive instrumentation. To make it effective, the cells of interest need to be identified with some labeling technique, such as genetically targeted fluorescent chromophores.

Genetic targeting can also be used to place chromophores that promote photoelectron transfer or function as photosensitizers, providing local phototoxicity, which results in ablation. Such chromophores can effectively mediate the production of reactive oxygen species through a variety of pathways.[3] Three distinct modes have emerged to produce oxidative stress in specific cells (Figure ). The first one uses genetically encoded photoactive proteins, such as the GFP-variants KillerRed and SuperNova, which primarily produce superoxide radical anions (Figure A).[4] They are formed through oxidative maturation of a polypeptide and do not require the addition of an external dye. Genetically encoded chromophores, however, have the disadvantage that they cannot be quickly replenished, undergo bleaching through photoisomerism, and are generally not very efficient.

Figure 1

Genetic encoding of singlet oxygen producing photosensitizers: (A) KillerRed, SuperNova, and their chromophore. (B) Attachment of a photosensitizer to a genetically encoded bioconjugation tag (HaloTag). (C) Enzymatic activation of a diffusible photosensitizer. (D) Enzymatic activation of a diffusible photosensitizer with concomitant covalent attachment. The genetically encoded component is indicated in boldface.

A second approach places synthetic small-molecule photosensitizers into specific cells using genetically encoded bioconjugation motifs. These were initially developed for chromophore-assisted light inactivation (CALI) of specific proteins, but they could also be used to damage and inactivate entire cells. Early examples include biarsenicals, such as ReAsH-EDT2, which bind to tetracysteine motives.[5] Their use in biology was limited, however, due to cytotoxicity caused by nonspecific binding to endogenous proteins. More recently, genetically encoded SNAP-tags[6] and HaloTags[7] have been employed to mount photosensitizers to specific target cells. An example of such dyes, diAc-eosin-AM, is shown in Figure B. The alkyl chloride end of this molecule selectively reacts with the HaloTag to provide a covalent linkage. To make the molecule membrane permeable and to suppress photochemical background activity, the phenolic hydroxy groups were masked as acetates, which were presumably cleaved by esterases inside of the cell. Since this takes place more or less in every cell, and surplus reagent cannot be easily removed by washing, significant background activity can be expected, at least in vivo. In addition, all of these approaches require stoichiometric targeting of a synthetic chromophore to the genetically encoded bioconjugation motif limiting their efficiency for cell ablation.

A third mode involves activation with an enzyme that is genetically encoded in selected cells. This has the advantage that a large number of photoactive dyes (or other toxic principles) can be generated from inactive precursors. This amplification, in addition to the amplification the photosensitizer itself provides, can deliver high concentrations of singlet oxygen and effective ablation. In 2014, Urano disclosed a molecule termed HMDESeR-βGal, that could be activated by bacterial β-galactosidase (gene name: lacZ, Figure C).[8] This enzyme is foreign to mammalian cells but can be easily heterologously expressed. Following enzymatic hydrolysis, the probe shifts to the open xanthene form since its phenolic hydroxy group (pKa = 5.2) can be easily deprotonated at physiological pH. HMDESeR-βGal features a heavy selenium atom which promotes fast intersystem crossing and triplet oxygen sensitization following irradiation. It could be used to ablate lacZ-positive cells upon wide field irradiation with green 550 nm light. However, its resolution was limited due to the ability of the cleaved sensitizer (presumably in its closed spiro form) to diffuse into neighboring cells.

In the present paper, Urano addresses this shortcoming by adding an additional functional feature: a conditional covalent linkage (Figure D). Urano’s new probe, termed SPiDER-killer-βGal, is also based on a spirocyclic selenorhodol aglycone that forms a β-linked glycoside with galactose. In addition, it contains a fluoromethyl group as a “pro-electrophile”. In its glycosylated form, the probe is not colored and not photoactive as it features a spirocyclic ether, and it will not undergo covalent attachment. Enzymatic hydrolysis by β-galactosidase yields a phenol that quickly eliminates hydrogen fluoride to form an ortho-quinine methide. This highly reactive intermediate now reacts with cellular nucleophiles (e.g., lysine side chains) to form covalent adducts. Importantly, these adducts predominately reside in the open, colored, and phototoxic xanthene form of the chromophore at physiological pH. As such, enzymatic hydrolysis of SPiDER-killer-βGal simultaneously activates both its photosensitizing ability and its reactivity to nucleophiles. Hence, the phototoxic products generated by light irradiation are confined to lacZ-positive cells. The very chemistry that fixes the probe to a genetically engineered cell also unlocks its phototoxicity as it shifts the equilibrium from the closed, photochemically inactive form to the open active form.

To demonstrate the usefulness of SPiDER-killer-βGal and its advantages over existing methods, Urano and colleagues applied it to progressively more complicated biological settings. First, they demonstrated selective ablation of lacZ-positive mammalian cells through coculturing HEK293 and HEK/lacZ(+) cells. Irradiation with 550 nm light for 3 min induced selective ablation of the lacZ-positive cells. These results confirmed that SPiDER-killer-βGal allows for ablation with single-cell resolution, which was not possible with previously reported enzymatically activated small-molecular photosensitizers. Next, the new probe was used in cultured Drosophila larvae wing discs in which β-galactosidase expression was restricted to the posterior regions. This experiment enabled selective ablation of cells in this region showing that SPiDER-killer-βGal works in cultured tissue. Finally, the probe was used in vivo in Drosophila where lacZ expression was induced in a subpopulation of cells in the pupal notum using a promotor that could be induced by heat shock. A fluorescent apoptosis marker was used to detect ablation. Indeed, cells expressing lacZ were found to selectively undergo apoptosis after irradiation with 561 nm laser light demonstrating the usefulness of this new tool in vivo. This series of experiments also demonstrated compatibility of SPiDER-killer-βGal with other fluorophores such as Calcein-AM (viability marker in tissue), H2B-ECFP (expression marker), VC3Ai (apoptosis reporter), and Hoechst-33342 (nuclear stain). With its unprecedented precision and effectiveness in vivo, SPiDER-killer-βGal has clear advantages over other methods for cell ablation, and it is likely to be embraced by the biology community. However, a redesign of its synthesis, which is lengthy and contains several low-yielding steps, might be necessary to make it widely available.

Urano’s approach to the enzymatic activation and concomitant fixation of photosensitizers could prove to be fairly general. Other activating enzymes, such as engineered esterases, phosphatases, nitroreductases, or azoreductases, could be used to unleash the electrophilicity and the phototoxicity of the probe. Enzymatic activation has already been broadly explored to generate fluorophores in a genetically targeted fashion. For instance, Urano himself introduced an azoreductase to activate a rhodamine fluorophore.[9] It would be straightforward to extend some of these methods to the genetically encoded activation of photosensitizers. Nitroreductases, which convert an electron-withdrawing nitro group to an electron-donating amino group, have also been used to unleash the toxicity of prodrugs, such as metronidazole, in genetically tagged cells, resulting in cell ablation.[10] The method worked well but suffered from limited resolution due to diffusion of the toxic principle out of the target cells. As demonstrated by Urano, such limitations can be overcome with smart chemistry that enables the simultaneous activation and localization of a lethal factor.