Molecular photoswitches in aqueous environments

Abstract

Molecular photoswitches enable dynamic control of processes with high spatiotemporal precision, using light as external stimulus, and hence are ideal tools for different research areas spanning from chemical biology to smart materials. Photoswitches are typically organic molecules that feature extended aromatic systems to make them responsive to (visible) light. However, this renders them inherently lipophilic, while water-solubility is of crucial importance to apply photoswitchable organic molecules in biological systems, like in the rapidly emerging field of photopharmacology. Several strategies for solubilizing organic molecules in water are known, but there are not yet clear rules for applying them to photoswitchable molecules. Importantly, rendering photoswitches water-soluble has a serious impact on both their photophysical and biological properties, which must be taken into consideration when designing new systems. Altogether, these aspects pose considerable challenges for successfully applying molecular photoswitches in aqueous systems, and in particular in biologically relevant media. In this review, we focus on fully water-soluble photoswitches, such as those used in biological environments, in both in vitro and in vivo studies. We discuss the design principles and prospects for water-soluble photoswitches to inspire and enable their future applications.

Fig. 1. Diagram summarizing the required properties of photoswitches needed for successful applications in water.

Fig. 2. Descriptive representation of the dissolution process of a crystalline solid in water. Energy is needed to first remove the molecule from its crystal structure [from (A) to (B)] and subsequently the molecule is solvated by water. Once solvated, the molecule forms stabilizing interactions with water (C), represented by the blue halo, which results in a release of energy. (Figure adapted with permission from ref. . Copyright 2006, Elsevier.)

Fig. 3. Descriptive representation of solubility types: the compound is either in its most stable crystalline solid form (A) or in a less stable, amorphous, form (B). High amount of energy is needed to remove the molecule from the stable solid (C), while a lower amount of energy is needed to remove it from the less stable amorphous form (D, see ref. .

Fig. 4. Two possible processes can be influenced to increase solubility of a specific compound. One involves lowering the energy needed to remove a molecule from its solid (A). The second approach is to increase the energy that is released upon solvation (B). The chemical methods (C) to decrease the melting point thus lowering the energy needed for (A) are, e.g., by designing non-planar structures and by addition of charged or polar solubilizing groups increases the stabilizing interactions with the solvent. The physical methods (D) to increase solubility involve reduction of particle size thus creating a larger surface or by adding co-solvents or hydrotropes (see also ref. 60).

Fig. 5. The use of charged groups to increase water solubility. (A) Sulfonated anthraquinones and the solubility of the corresponding ion pairs in water at room temperature, (B) Akt-kinase inhibitors 2.2 with amine solubilizing group, their respective log D value and solubility in 0.1 M phosphate buffer at pH 7.4 at 25 °C.

Fig. 6. Examples of using non-ionic groups to enhance solubility (A) substitution of the naphthalene moiety (2.3a) of the STAT1 signal transduction inhibitor 2.3 with a quinoline moiety (2.3b) significantly increased the solubility, (B) the addition of a fluorine atom to 2.4 and its effect on solubility, clog P (calculated log P) and the melting point, (C) the oxetanyl sulfoxide group caused a 76-fold increase in solubility of 2.5.

Fig. 7. The difference in water-solubility, both kinetic and intrinsic, of the carboxylic acid and sodium salt form of diclofenac 2.6.

Fig. 8. Increased dihedral angle of 2.7 (A) upon addition of a methyl group, as well as the ortho substituents in 2.8b and c (B) cause a decrease of melting point and increased solubility in water. (C) Introduction of a stereocenter in 2.9, thus breaking the symmetry of the compound, significantly lowered the melting point and caused increased solubility in water.

Fig. 9. Illustrative examples of prodrugs containing solubilizing groups: phosphate (A), PEG chains (B) and amino acid moiety (C) and their respective effect on solubility in water.

Fig. 10. (A) The Jablonski diagram highlighting the key photophysical processes possible from excited state of a photoswitch (photochemical processes are described in C); (B) schematic representation of potential energy surfaces during isomerization of a photoswitch; (C) general scheme depicting isomerization of a photoswitch upon light irradiation. Most photoswitches reversibly isomerize from A to B with application of light and heat, while in some instances isomer B can further react to form a third isomer C; (D) graph (concentration of A vs. time) depicting the first order kinetics of photoswitching with highlighted half-life and lifetime values.

Fig. 11. Overview of various photoswitches based on double-bond isomerization (CC, NN, and CN) that are discussed in Section 3.2.

Fig. 12. Photochemical side-reactions of stilbenes. Top: E–Z photoisomerization of stilbene and the competing photo-oxidation and electrocyclic ring-closure reaction followed by irreversible oxidation. Bottom: E–Z photoisomerization in 1,1′-bis-indanyliden (stiff stilbene) and the competing photo-oxidation.

Fig. 13. A stiff stilbene-based first generation Feringa molecular motor was equipped with alkylammonium group to promote water-solubility. The unidirectional rotation of the motor relies on photochemical E–Z isomerization (irradiation with 312 nm light) and thermal helix inversion (THI, here, five days at ambient temperature in water).

Fig. 14. Overview over photochemical and thermal isomerization pathways in azobenzenes.

Fig. 15. Thermally stable Z and photochemically generated E isomer in diazocines with various substituents X in the bridging unit tuning their properties.

Fig. 16. (Photo)switching in indigo photoswitches. (A) Thermally stable E and metastable Z isomer in (thio)indigo derivatives. (B) Different photochemical and thermodynamic properties of (thio)indigo switches depending on the substituent X.

Fig. 17. (A) Chemical structures of hemiindigo (HI), hemithioindigo (HTI), and aurone. (B) Photochemical E–Z isomerization of HTI 3.10.

Fig. 18. Thermally stable Z and metastable E isomer of Iminothioindoxyl (ITI).

Fig. 19. Overview of various (acyl)hydrazone derivatives and their photochemically and thermodynamic properties.

Fig. 20. Thermodynamically stable E and metastable Z-isomer of an azo-BF₂ switch.

Fig. 21. Overview of molecules discussed in Section 3.3.

Fig. 22. (Photo)switching in diarylethenes (DAEs). (A) Schematic representation of different photochemical mechanisms accessible to DAEs. (B) The influence of different substituents R on the thermal lifetime of the closed isomer. (C) DAE 3.15 shows close to quantitative photoisomerization quantum yields. (D) DTE 3.16 show close to quantitative PSDs in both directions and high, solvent independent ring-closure quantum yields.

Fig. 23. Tuning the photochemical stability of DAEs. (A) Photochemical side-product formation upon prolonged irradiation times with UV-light. (B) Scaffolds that prevent side-product formation: benzothiophene-based DAEs, e.g.3.19, DTEs with electron-poor substituents such as 3.20, and DTEs isomerizing upon excitation of an intramolecular triplet sensitizer like 3.21.

Fig. 24. Schematic representation of the population of a twisted intramolecular charge transfer (TICT) state depending on the solvent polarity. After excitation to the S₁ the compound relaxes preferably into a state where the thiophene rings have a planar (p) or twisted (t) conformation, depending on the solvent polarity.

Fig. 25. Photoswitching electronic properties in DTEs. (A) Switching between electronically insulated and connected isomers, where the closed form is able to serve as catalyst after imine formation. (B) A photoswitchable redox cofactor bearing a benzoquinone bridging unit that can be toggled between an oxidative (open) and non-oxidative form (closed).

Fig. 26. The dihydropyrene/cyclophanediene photochromic couple. (A) (Photo)switching between the two isomers. (B) Photoswitching in the presence of oxygen can lead to endoperoxide derivative 3.27, which then thermally releases ¹O₂.

Fig. 27. (A) The norbornadiene/quadricyclane photochromic couple, (B) the effect of substituents on the photochemical properties, including the position of the absorption band and the half-life of the metastable quadricyclane.

Fig. 28. Overview of the molecules discussed in Section 3.4: fulgides, fulgimides, chromenes, naphthopyrans, spiropyrans, DASAs, and dihydroazolenes.

Fig. 29. General switching of fulgides and fulgimides and the various isomers involved.

Fig. 30. Stability of differently functionalized fulgimides. (A) Hydrolysis of a CF₃-substitutent in the bridging unit under slightly basic conditions and subsequent loss of CO₂H upon acid conditions. (B) Photoswitching of fulgimide 3.30 with CO₂H group in the bridge.

Fig. 31. General photoswitching mechanism in chromene and naphthopyran.

Fig. 32. Overview of various characteristics in the spiropyran/merocyanine photochromic couple, using a representative structure (3.32 and 3.35) bearing an NO₂ substituents in para-position to the pyran oxygen. Central box (C): SP-3.32 can be photoisomerized via the Z-MC intermediate displayed its two resonance structures in (A) into the corresponding MC-3.32, which is sensitive to hydrolysis (B). Both the SP and the MC form can be protonated, with the open isomer being the weaker acid making the system a photo acid generator (PAG) upon ring-closure (D). The MC-form can undergo selective E–Z photoisomerization when the phenolic oxygen is protonated or alkylated (E).

Fig. 33. Chemical structure of a Py-SP (A) and spirooxacine (C). (B) Photoisomerization of 3.38, with SP-3.38 being more hydrophobic than MC-3.38, allowing it to diffuse through lipid bilayers.

Fig. 34. Overview of the (photo)isomerization in DASA (A) and general structures of generation 1, 2, and 3 derivatives (B).

Fig. 35. Simplified (photo)isomerization mechanism of DASA photoswitches.

Fig. 36. Stable isomers of DASA. (A) Solvent-dependent thermal equilibrium between the open and the closed isomer in DASA. (B) Non-photochromic derivatives.

Fig. 37. General mechanism of the (photo)isomerization in the dihydroazulene/vinylheptafulvene photochromic couple.

Fig. 38. Azobenzene photoswitches furnished with permanently charged (A and B) (4.1b, 4.2a, 4.2b) ammonium groups or with non-permanently charged alkylamines (4.1a and 4.1b). (C) The charged elongated E isomer can easily fit into the channel cavity thus blocking it, while the Z isomer does not fit into the cavity leaving the channel open.

Fig. 39. (A) The spiropyran photoswitch SP undergoes a ring opening reaction upon irradiation with light thus forming the open zwitterionic MC isomer. Under acidic conditions, MC is protonated, forming the positively charged MCH+ isomer which can be ring-closed upon irradiation with visible light. (B) Both charged isomers MC and MCH+ bind to double stranded DNA, while the SP does not interact with DNA. (C) Spiropyrans 4.3a and b, carrying a nitro and cyano groups, were further modified to carry one positive charge (4.4a) or two positive charges (4.4b) to explore the effect of additional charge on their DNA binding properties. (D) Visible light-controlled azobenzenes 4.5a, carrying two positive charges, and azobenzene 4.5b, carrying one positive charge, were designed to modulate DNA.

Fig. 40. Photoswitches solubilized with positively charged cyclic amines and heteroatoms containing nitrogen atom. (A) Azobenzene 4.6, carrying two methylated morpholine moieties, and stilbene 4.7 with two pyridinium groups were used to modulate G quadruplex DNA structures. (B) Irreversible photooxidation of stilbene 4.7 into 4.7bvia4.7a. (C) Stilbene 4.8 functionalized with two piperazine groups exhibited affinity for G4 DNA over duplex DNA, facilitating toxicity towards cancer cells while the structurally related 4.9 was bound too strongly to DNA prohibiting photoswitching. (D) The photoisomerization of a DTE photoswitch carrying two pyridinium groups of which the open form 4.10o showed higher toxicity towards cells opposed to the closed form 4.10c.

Fig. 41. The chromene photoswitch 4.11 forms the open 4.11TC isomer upon UV irradiation, which interconverts to the TT isomer (4.11TT).

Fig. 42. Sulfonated azobenzene dyes, carmoisine and Congo red.

Fig. 43. Introduction of a methyphosphonate group (4.12) to the AMPA antagonist NBQX ensured water-solubility. Molecule 4.12 was modified to incorporate a light-responsive azobenzene switch (4.13) as photocontrolled AMPA receptor antagonist.

Fig. 44. DTE-base mtPriA inhibitors with phosphate and phosphonate groups on different positions of the phenyl ring.

Fig. 45. The prodrug based on a DAE forms the ring-open isomer upon irradiation with 400 nm light which can undergo reverse Diels–Alder reaction releasing maleimide and the open furyl-substituted DAE.

Fig. 46. The conjugate acid of azobenzene sulfonylurea 4.17 with a pK_a of 4.76 is deprotonated and thus negatively charged in aqueous solutions.

Fig. 47. (A) Selected examples of photoswitchable hydroxamic acid-based HDAC inhibitors. (B) Arylazopyrazoles carrying a hydroxamic acid moiety functionalized with different substituents and their effect on solubility.

Fig. 48. The water-soluble dihydropyrene photoswitch, carrying a PEG chain, undergoes a ring opening reaction upon irradiation with visible light and the ring-closing process under UV irradiation.

Fig. 49. Azobenzene 4.21 carries multiple PEG linkers as well as guanidinium groups which enable water-solubility as well as enable binding to the target protein via salt bridges.

Fig. 50. Photoswitchable CENP-E inhibitors containing an azobenzene were poorly water-soluble (4.22a and b) while the heteroaromatic arylazopyrazole photoswitch 4.22c exhibited much better aqueous solubility.

Fig. 51. Inhibitors of RET transmembrane tyrosine kinase receptor, illustrating the effect of a different switch system (stilbene 4.23avs. azobenzene 4.23c) and an additional heteroatom (4.23b and c) on the (photo)chemical properties in aqueous solutions.

Fig. 52. Different substituents in azopyridine-based sirtuin inhibitors and their effect on their water-solubility, (photo)chemical properties and activity.

Fig. 53. (A) The potent sirtuin inhibitor 4.26 was redesigned into the photoswitchable system 4.27. (B) Since 4.27 could not undergo ring-closing in aqueous solution, the system was further redesigned into photoswitchable fulgimides 4.28.

Fig. 54. Light-controlled inhibitors of zinc dependent HDACs. (A) DTE 4.29a was unstable upon irradiation in water, quickly forming the byproduct, while DTE 4.29b formed 80% if the closed isomer with minimal byproduct formed. (B) Fulgides 4.30a and b showed excellent photochemical properties in water however the difference in activity was much smaller compared to the DTE switches.

Fig. 55. Breaking planarity to increase water-solubility. (A) Azobenzene-based photoswitches where the Z isomer was 20-times (4.31) or 1.4-times (4.32) more soluble in phosphate buffer then the E form. (B) Introduction of three methoxy groups increases water-solubility and disables aggregate formation.

Fig. 56. Photoswitchable glutamate receptors carrying anionic amino acid moieties. Azobenzene 4.34, featuring an aspartate moiety, could be solubilized at concentrations of 0.1 mM in phosphate buffer (A), while the glutamate azo-derivative 4.35 was poorly soluble in water and required heating and addition of DMSO in the stock solutions despite carrying a glutamate moiety (B).

Fig. 57. Azobenzene photoswitches decorated with multipole sugar moieties with increased water-solubility.

Fig. 58. Photoswitchable antibiotics. (A) Ciprofloxacin was modified to incorporate an azobenzene (4.37) or a spiropyran photoswitch. (4.38). Azobenzenes 4.39a–c (B), hydantoin 4.40a and phytochrome 4.40b (C) contain the nalidixic acid motif ensuring antimicrobial activity.

Fig. 59. (A) Spiropyran photoswitch with a fluorescent linker and a sugar targeting group. (B) Spiropyran based photoswitch carrying a fluorescent moiety and a galactose group locking it in the merocyanine form. Upon cleavage of the sugar, the photoswitch can reversibly switch from MC to SP form in water.

Fig. 60. Azobenzene-based amino acids with a covalent attachment point on the opposite side of the photoswitch, both controlled by UV light (A) and by visible light (B). (C) After the light responsive amino acid is incorporated into the protein, the anchoring group reacts with a free thiol moiety nearby thus crosslinking the protein.

Fig. 61. Azobenzene-based amino acids for SPPS with free (amino) acid groups highlighted in green and protected ones highlighted in orange.

Fig. 62. The open isomer of a DTE-based amino acid was used to mimic a β-turn (A) in a cell penetrating peptide by incorporation into the cyclic peptide backbone and modified it with a fluorescein group for visualization (B). The fluorescent molecule was attached via a long flexible linker to not compromise the cellular permeability.

Fig. 63. (A) Gramicidin S, a cell penetrating cyclic peptide with antibiotic and anticancer properties contains two β-turns (in orange), of which one (in blue) was replaced with a photoswitchable DTE amino acid. (B) Two modified photoswitchable Gramicidin S analogues, where the amino acids shown in orange, are substituted in the original structure CPP 4.52 was the less water-soluble design due to too high hydrophobicity while 4.53 was the most successful and subsequently was tested in animal studies.

Fig. 64. (A) Azobenzene-based nucleosides 4.54p and 4.54m used to photocontrol RNA duplex formation and their location upon incorporation into the RNA strand (B). Visible light-responsive tetra-ortho-chloro substituted azobenzene 4.55 incorporated into RNA (C).

Fig. 65. Photoswitchable nucleotides with an incorporated DEA switch. DEA 4.56 (A) with an adenosine did not undergo efficient cyclization reaction in water, while the uracil (4.57-U) and the cytosine (4.57-C)-based DEA nucleotides had favorable properties in aqueous solutions.

Fig. 66. (A) Spiropyran 4.58 was attached to the 2′-position on a nucleotide in the minor groove of the DNA strand, placing it too close to the DNA bases thus diminishing its photoswitching properties. The design was improved by attaching the spiropyran 4.59(B), which could form 40% of the open merocyanine isomer upon irradiation which was prone to hydrolysis. By utilizing the reversible aldol reaction, the aldehyde part of the photoswitch 4.59 was replaced with 4.60ald after it was incorporated into DNA, resulting in 4.60 with improved thermal stability.

Fig. 67. Schematic representation of the role of the photoswitch when interfering with the target biosystem itself, or its interaction with another entity (neurotransmitter molecule, small biomolecule interacting, bacterial cell, etc.).

Fig. 68. (A) Azobenzene photoswitches with water-solubilizing groups (4.61c and d) were successfully used to photocontrol nanopore assembly of the FraC toxin. (B) Azobenzene 4.62, carrying a mannoside group, was incorporated onto the surface of human cells; the Z isomer did not allow bacterial adhesion, while the elongated E isomer enabled it.

Fig. 69. (A) Five spiropyran 4.63 units covalently attached to the MscL channel. Upon photoisomerization the charged merocyanine isomers force the channel to open, while the uncharged isomer 4.63 allows the channel to remain closed. (B) Azobenzene carrying a homocholine group was attached to the ligand-gated channel nAChR in the Z orientation preventing the binding of the natural ligand acethylcholine and leaving the channel open. Upon irradiation with 380 nm light the E isomer does not block the binding site of acethylcholine resulting in closure of the channel.

Fig. 70. The maleimide-azobenzene-glutamate probe which can undergo two-photon (2p) excitation thus enabling photoswitching with near-infrared light, 4.65, had a limited response to two-photon excitation (A). Therefore, 2p absorbing antennae were introduced to the design by incorporating naphthalene (4.66a) or pyrene (4.66b) groups (B). However, the large aromatic antennae drastically reduced the water-solubility of the constructs leading to the design of 4.67 (C). Azobenzene 4.67b contains an electron-donating group at the para position and fluoro substituents in ortho position which significantly enhance the 2p absorption.

Fig. 71. Fluorescence photoswitching for super-resolution microscopy. (A) Fluorescence switching of cyanine dyes in the presence of primary thiol-based reducing agents or TCEP; (B) fluorescence switching in functionalized rhodamine dyes which undergo a reversible transition from open to closed form (and vice versa) from the fluorescent zwitterionic form to the dark lactone.

Fig. 72. DTE fluorescent probes are lipophilic molecules that require significant amounts of co-solvents during attachment to the target protein (30% EtOH in water for 4.70). DTE derivatives containing benzothiophene dioxide groups with perfluorocyclopentene as the bridging unit were decorated with carboxylic acid moieties to enable sufficient water-solubility (B). Multiple carboxylated probes were sufficiently soluble for application in in celullo STORM fluorescence imaging.

Fig. 73. While for many applications the hydrophobic azobenzene crosslinker 4.75 could be successfully applied as the target itself ensured water-solubility of the crosslinker-protein construct (A), for other examples, such as the collagen peptides with photoswitch 4.76, the resulting crosslinked biomolecule aggregated due to insufficient aqueous solubility (B). The Z isomer of azobenzene 4.77 allowed retention of the secondary structure of the α helical cell-penetrating peptide upon crosslinking, while the E isomer disrupted the secondary structure (C). The Z isomer of the diazocine 4.79 allowed formation of the Trp fold, while the extended E isomer destabilized it.

Fig. 74. Water-soluble crosslinkers. (A) Sulfonated azobenzene 4.80 has been successfully applied to crosslink numerous targets and was subsequently multiplied and elongated in designs 4.81 and 4.82 to achieve larger end-to-end distance change upon switching (B). The azonium-based azobenzene 4.83 with tetra-ortho-methoxy substituents can be switched with 633 nm light and has a protonated nitrogen atom which forms a stabilizing hydrogen bond with the adjacent methoxy group only in the E isomer.

Fig. 75. Azobenzene photoswitches modified with carbohydrate moieties to ensure water-solubility. The sugar groups have one or two reactive handles at various positions for attachment to a target.

Fig. 76. Decision tree guideline for choosing the water-solubilizing group for a specific molecule based on the type and location of the photoswitchable entity (in orange). Grey boxes describe general considerations to keep in mind while designing a construct, mint-green circles ask key questions for the design and purple areas possible roles of the solubilizing group with illustrative examples.

Jana Volarić

Wiktor Szymanski

Nadja A. Simeth

Ben L. Feringa