Optimizing CRBN-ligands for optical targeted protein degradation

Abstract

PROTACs are heterobifunctional molecules that modulate protein levels through targeted degradation, eliminating both physiological and pathophysiological functionality. Photoswitchable PROTACs (or PHOTACs) expand on this technology by enabling spatial and temporal control through embedded photoswitches. While these elements can be positioned in either the ligand or linker regions, the extent to which new photoswitches can be incorporated directly into CRBN-ligands without compromising degradation efficiency or light responsiveness remains difficult to predict and largely unexplored systematically. Here, we report new synthetic strategies for incorporating heteroarene and functionalized azobenzene photoswitches into lenalidomide ligands to evaluate their compatibility with CRBN-mediated degradation. Arylazopyrazole-based PHOTACs emerged as the most effective, with apPHOTAC-I-2C supporting robust, reversible, light-controlled BET protein degradation. This approach was extended to synthesize apPHOTAC-II-2C, the first nanomolar photoswitchable degrader of tumor-associated protein MTH1, establishing a promising synthetic approach for future PHOTAC development. Competitive binding studies showed that irradiation lowered CRBN affinity even as degradation increased, revealing a decoupling between light-dependent changes in affinity and efficacy.

A series of CRBN ligand-photoswitch conjugates were developed for photoswitchable PROTACs. Arylazopyrazole apPHOTACs exhibited light-dependent degradation of BRD2–4 and MTH1 with high bistability and quantitative isomerization.

Introduction

PROTACs (proteolysis targeting chimeras) are a broad class of synthetic compounds that suppress protein function through targeted protein degradation (TPD).¹ Generally, they consist of three modular components: a protein of interest (POI) ligand, an E3 ligase ligand, and a linker. This design facilitates the formation of a ternary complex between the POI, the PROTAC, and an E3 ligase (e.g. cereblon, CRBN), thereby resulting in the ubiquitylation and proteasomal degradation of the target protein (Fig. 1A).² Unlike conventional occupancy-driven inhibitors, this event-driven mechanism enables the sustained chemical knockdown of otherwise undruggable proteins at sub-stoichiometric doses.³ However, this distinction also raises unique risks: most notably, systemic degradation of a POI disrupts both its physiological and pathophysiological functions; conventional PROTACs also lack intrinsic spatial or temporal control, potentially exacerbating off-target activity. Thus, the potent and irreversible nature of PROTAC-mediated degradation motivates new strategies for external activity regulation.

Fig. 1. Summary of PROTACs. (A) Schematic depiction of a heterobifunctional PROTAC, ternary complex formation, ubiquitination, and protein of interest (POI) degradation. (B) Chemical structures of representative PROTACs dBET1 and ATAG 2139.

To address this, conditional PROTACs have been developed to enable user-defined control over degradation kinetics.⁴ These variants remain inert until triggered by an external stimulus (e.g., pH, enzymatic activity, hypoxia) to initiate degradative action.^5–7 In this respect, light-activated PROTACs offer unique advantages: light is non-invasive, readily tuneable, and spatially precise in nature, with established utility in medicine and chemical biology (e.g., photodynamic therapy, optogenetics).^8,9 This photopharmacological strategy can be achieved via photoswitches, molecules that undergo reversible structural changes in response to light (e.g., azobenzene cis–trans isomerization).¹⁰ When inserted into PROTACs, these elements can alter ligand engagement and ternary complex formation in a light-dependent fashion, thereby embedding reversible control over TPD and its functional consequences (Fig. 2A).

Fig. 2. Summary of photochemically-targeting chimeras (PHOTACs). (A) Schematic depiction of a PHOTAC. The molecule toggles between an inactive form (dark pentagon) and an active form (bright star) upon photochemical isomerization (hν₁ or hν₂) or thermal relaxation (k_BT). (B) Chemical structure and photoswitching of PHOTAC-I-3, a photoswitchable degrader of bromodomain and extra-terminal (BET) proteins. (C) Design of new photoswitch-conjugated CRBN ligands based on the azo-extension of lenalidomide by carbon excision, tetra-ortho-substitution, and heteroarene substitution. (D) Chemical structures of BET-targeting apPHOTAC-I-2C and MTH1-targeting apPHOTAC-II-2C.

Our group was among the first to report photoswitchable PROTACs, or photochemically targeting chimeras (PHOTACs), for optical targeted protein degradation.¹¹ Uniquely, we leveraged the diazotization of lenalidomide, a well-established cereblon (CRBN) E3 ligase ligand, to extend azobenzenes from its arene core. This azo-extension approach yields a compact, modular scaffold complementary to established PROTAC linkers and targeting ligands, with no detrimental impact on photoswitching efficiency. To validate this concept, we created PHOTAC-I-3 (derived from dBET1) as an optical degrader of epigenetic reader proteins BRD2–4 (Fig. 1B and 2B). PHOTAC-I-3 exhibits no activity in the dark but can be activated with visible light to achieve robust degradation in a reversible fashion. Since then, this strategy has been adapted to construct PHOTACs for a variety of protein targets (e.g., FKPB12, HDAC, neuronal CaMKIIα).^11–13

Many alternative approaches for constructing photoswitchable PROTACs have since emerged, placing the photoswitch in different regions of the molecule. The most common mode positions a switch within the linker, a design that decouples photoswitch and ligand construction to facilitate a variety of photoswitch designs with assorted photophysical profiles (e.g., efficient isomer enrichment, red-shifted activation, thermal bistability).^14–16 Photoswitches have recently been embedded within the targeting ligand itself, offering new entry points for optical TPD.¹⁷ Despite these advances, the impact of photoswitch placement and geometry on ligand-protein interactions and ternary complex formation remains poorly understood and difficult to accurately predict. Moreover, while our azo-extension strategy has reliably supported the development of light-active and dark-inactive PHOTACs, the extent to which new advantageous photoswitches can be incorporated directly into CRBN-ligands without compromising degradation efficiency and light-dependency has not yet been rigorously explored.

In this study, we developed new synthetic routes for embedding photoswitches into CRBN-binding lenalidomide derivatives. Using these approaches, we generated new ligand-photoswitch conjugates with arylazopyrazole, azobenzene-benzamide, and tetra-chloro-azobenzene-benzamide motifs as modular building blocks for PHOTAC development. In doing so, we report apPHOTAC-I-2C and apPHOTAC-II-2C as arylazopyrazole-based photoswitchable degraders of BET-family proteins and MTH1, respectively (Fig. 2D). These synthetic advances provide new entry points for embedding photoswitches into PROTAC building blocks for achieving optical targeted protein degradation.

Results and discussion

Building on our established azo-extension strategy, we aimed to broaden the scope of our PHOTAC design platform by developing 1) arylazopyrazole-lenalidomide and 2) azobenzene-benzamide derivatives (Fig. 2C). Arylazopyrazole photoswitches offer exceptional bistability and quantitative photoisomerization while retaining the compact structural footprint of azobenzenes. These heteroarenes have already been adapted to photoswitchable PROTAC design, but only as linker elements.^16,18 Alternatively, benzamide-type CRBN ligands are well-established for their favourable pharmacological profiles (e.g., reduced neosubstrate bias, enhanced enzymatic stability) and synthetic accessibility, yet their application in photoswitchable PROTACs has not yet been explored.¹⁹ The resulting architecture would eliminate the phthalimidine core inherited from lenalidomide, facilitating synthetic accessibility and further derivatization via tetra-ortho-chlorination.²⁰ These photoswitches can be actuated with >550 nm light, aligning with broader efforts in photopharmacology to overcome the limited tissue penetration of shorter wavelengths. Collectively, these structural modifications provide practical photophysical optimization and introduce controlled variations around the glutarimide pharmacophore for activity profiling.

Guided by these considerations, we synthesized a small library of PHOTACs to evaluate their efficacy for light-controlled protein degradation (Fig. 3). This collection spans three distinct series: 1) arylazopyrazole apPHOTACs, 2) azobenzene-benzamide bzPHOTACs, and 3) chloro-azobenzene-benzamide redPHOTACs. As our initial POI ligand, we selected (+)-JQ1, a well-established, high-affinity BET inhibitor widely used in classical PROTAC development.^21,22 To account for linker-dependent effects on degradation, we generated matched derivatives with varying alkyl linkers across PHOTAC types.

Fig. 3. Structures and synthesis of the PHOTACs. (A) Synthesis of arylazopyrazole-lenalidomide apPHOTAC-I series. (B) Synthesis of the azobenzene-benzamide bzPHOTAC-I series. (C) Synthesis of the chloro-azobenzene-benzamide redPHOTAC-I series.

Beginning with the synthesis of the apPHOTACs, lenalidomide was readily diazotized and coupled by enolate addition with acetylacetone to give hydrazone 1. Condensation with hydrazine furnished arylazopyrazole 2 in excellent yield. Alkylation and deprotection afforded the key acid 3, which was subjected to amide coupling with various aminoalkanes to yield amines 5a–c after deprotection. A final coupling with the free acid of (+)-JQ1 afforded apPHOTAC-2C, -4C, and -6C (Fig. 3A).

To synthesize the bzPHOTACs, 3-nitrobenzoic acid and 3-amino-glutarimide were subjected to amide coupling and nitro reduction to generate benzamide amine 7, which was diazotized and coupled with 2,6-methoxy phenol to afford azobenzene 8. Subsequent alkylation, deprotection, and amide coupling steps yielded bzPHOTAC-I-2C and -4C (Fig. 3B). While we attempted to subject an unsubstituted derivative of 8 to Pd-catalysed C–H halogenation to quickly access tetra-ortho-chloro derivatives, extensive testing yielded only tri-chlorinated products.²⁰ We attribute this to steric interference from the meta-benzamide arm impeding catalyst activity. To circumvent this, we instead pre-functionalized this hindered ortho site at the outset with chloro-benzamide 12, and mono-chloro 13 was converted to tetra-chloro 14 in good yield to subsequently deliver redPHOTAC-I-2C and -4C (Fig. 3C). To our knowledge, this is the first reported instance of a meta-substituted tetra-ortho-chloro azobenzene. Thus, this pre-installation strategy may offer a generalizable approach towards these densely substituted, red-shifted photoswitches.

Next, we evaluated the photophysical properties of our new PHOTAC designs. Among them, apPHOTAC-I-2C emerged as the most effective for optical TPD (Fig. 4). In DMSO-d₆, photostationary states (PSSs) exceeding 98% cis under 360 nm light and 95% trans under 525 nm light were obtained (Fig. 4D). In the absence of light, cis-apPHOTAC-I-2C exhibited a thermal half-life >48 hours at 37 °C in a DMSO : PBS (8 : 2) mixture (Fig. 4C). High isomeric purity and prolonged thermal stability are attractive features for the photopharmacological control of biological events on timescale of TPD. Analogue bzPHOTAC-I-2C exhibited high cis-enriched PSS with violet light (74%) and moderate bistability (t_1/2 = 15 h) (Fig. S1 and S2); redPHOTAC-I-2C exhibited moderate cis-enriched PSS with yellow-green light (58%) and high bistability (t_1/2 > 24 h) (Fig. S3 and S4). Structurally related PHOTACs exhibited similar photophysical and thermal properties, and all compounds were stable to over 20 rounds of repeated photochemical isomerization (Fig. S1E–S10E).

Fig. 4. Photochemical switching, bistability and photostationary states of apPHOTAC-I-2C. (A) Switching of apPHOTAC-I-2C between the trans isomer (left) and cis isomer (right). (B) Ultraviolet-visible (UV-vis) absorption spectra of apPHOTAC-I-2C following irradiation with various wavelengths (left) and single-wavelength absorption at 350 nm (normalized to dark) visualizing relative isomer content by intensity of the π → π* absorption band (right). (C) Thermal relaxation of cis-apPHOTAC-I-2C at 37 °C in DMSO : PBS after irradiation with UV light (D) ¹H-NMR spectra of apPHOTAC-I-2C at different photostationary states (PSS) upon irradiation in d₆-DMSO. Arrows indicate signals and corresponding integrations used for PSS determination.

To evaluate the biological activity of our library, we analysed the light dependence of degradation of BET proteins in RS4;11 cells, a well-established model for PROTAC development with reliable high expression of BRD2–4. Cells were treated with each PHOTAC at a range of concentrations and incubated in the dark or with a wavelength optimized for cis-isomer enrichment (380 nm for bzPHOTACs, 565 nm for redPHOTACs, and 360 nm for apPHOTACs). Irradiation was performed by applying light for a continuous 60 seconds to quickly drive isomerization, followed by pulsed irradiation to maintain PSS (100 ms per 10 s); redPHOTACs were pre-irradiated to their PSS before treatment, to account for lower molar absorptivity. Upon cell lysis and western blot analysis, protein bands were quantified and normalized to PCNA loading control to calculate DC₅₀ (the concentration at which half maximal degradation is observed) and D_max values (the maximal degradation achieved), if applicable. These data are summarized in Fig. 5.

Fig. 5. Summarized BRD4 and BRD3 degradation of PHOTACs in RS4;11 cells after 4 hours. Values were quantified from Western blot analysis of three independent experiments. Irradiation wavelength was optimized for cis-isomer enrichment depending on PHOTAC type (380 nm, 565 nm, and 360 nm for bzPHOTACs, redPHOTACs, and apPHOTACs, respectively). DC₅₀ = the concentration at which half maximal degradation is observed; D_max = the maximal degradation achieved; nd = not determined.

Across all chemotypes, the arylazopyrazole series displayed the most pronounced light-dependent enhancement of degradation. Among them, apPHOTAC-I-2C showed no detectable degradation in the absence of light across all concentrations after 4 hours (Fig. 6A). Upon 360 nm irradiation, however, apPHOTAC-I-2C induced robust degradation of BRD4 (DC₅₀ = 1.05 μM, D_max = 67%), BRD3 (DC₅₀ = 164 nM, D_max = 90%), and BRD2 (DC₅₀ = 484 nM, D_max = 81%) (Fig. 6B). Under light, no hook effect was observed up to 30 μM. Control experiments with parent dBET1 under analogous conditions showed an emerging hook effect at 30 μM, though activity was not fully abolished (Fig. S13). Our structural modifications may enhance productive PROTAC-POI interactions within the tested range, though a hook effect may still arise >30 μM. While longer linker lengths were detrimental to dark inactivity, apPHOTAC-I-4C and -6C both exhibited hook effects that emerged earlier in the dark compared to light conditions, indicating isomer-dependent differences in productive PROTAC-POI interactions (Fig. S11 and S12). Co-treatment of apPHOTAC-I-2C with inhibitors MG132 and MLN-4924 confirmed proteasome and neddylation blockade impaired degradation, consistent with a CRBN-dependent mechanism (Fig. S14).²³

Fig. 6. Optical control of apPHOTAC-I-2C protein degradation. (A) Western blot after treatment of RS4;11 cells with varying concentrations of apPHOTAC-I-2C for 4 hours. Cells were either irradiated with 360 nm light (60 s continuously, followed by 100 ms pulses every 10 s) or kept in the dark. BET-targeting PROTAC dBET1 (1 μM) and co-incubation with 1 μM neddylation inhibitor MLN 4924 (MLN) acted as positive and negative controls. (B) Quantified results of three independent experiments shown as the mean ± standard deviation (normalized to vehicle control). DC₅₀ values were obtained by nonlinear regression using a four-parameter dose–response model, shown as a dotted best-fit curve (if possible). (C) Time-dependence of light-dependent BRD3 degradation with 1 μM aPHOTAC-I-2C upon 360 nm irradiation (60 s continuously, followed 100 ms every 10 s) or darkness, and (D) 360 nm irradiation (60 s continuously) or 525 nm irradiation (100 ms every 10 s) by Western blot, with quantification of three independent experiments. (E) BRD3 post-elimination with 1 μM aPHOTAC-I-2C (4 h, pulsed 360 nm) demonstrating (left) bistability in the dark and (right) reversibility promoted by 525 nm irradiation (60 s continuously, then 100 ms per 10 s) by Western blot, with quantification of three independent experiments.

In contrast, the bzPHOTAC series induced robust target degradation in both the presence and absence of light after 4 hours (Fig. S15). Notably, bzPHOTAC-I-2C DC₅₀decreased upon irradiation, inverting the optical control observed with parent PHOTAC-I-3 (structurally differing by a single carbon bridging the amide and arene).¹¹ Conversely, the redPHOTAC series failed to elicit any degradation upon extended 24 hour incubation across all tested concentrations and irradiation conditions (Fig. S16). We hypothesize that replacing the rigid phthalimidine core of lenalidomide with a more flexible benzamide scaffold reduces the impact of photoisomerization on protein binding and ternary complex formation. In turn, steric interference from tetra-ortho-chloro substitution may impair these events altogether. The dramatic polarity changes of this substitution pattern also raise the possibility that limited cellular uptake or intracellular sequestration contributes to the general inactivity of this series. While furthers studies are needed to rationalize these trends, these patterns offer direction for future CRBN ligand and PHOTAC design.

To assess whether the light-dependent phenotype of apPHOTAC-I-2C extends beyond RS4;11, we evaluated its activity in Jurkat cells, a distinct T-cell leukemia line with a different proteostasis network (Fig. S17). Under identical experimental conditions, we again observed clear light-dependent degradation, although with a shifted window of efficacy: apPHOTAC-I-2C was less potent at lower concentration yet remained active up to 30 μM. As in RS4;11, no hook effect was detected; however, this was consistent with the behaviour of dBET1 in this line and may be indicative of cell-specific differences in target and E3 ligase stoichiometry (Fig. S13).

Having demonstrated consistent light-dependent TPD with apPHOTAC-I-2C, we next evaluated the extent of its optical control in RS4;11. To assess long-term activity trends, BRD3 levels were monitored over 24 hours at 1 μM with a selection of irradiation conditions. In the absence of light, BRD3 levels remained indistinguishable from vehicle controls, demonstrating sustained inactivity in the absence of an optical trigger (Fig. 6C). Likewise, pulsed 525 nm irradiation produced no detectable degradation, consistent with both quantitative trans-enrichment and negligible trans-activity (Fig. 6D). In contrast, a brief 60 second light dose was sufficient to clear BRD3 after 12 hours without further irradiation (Fig. 6D).

To validate the photochemical reversibility of degradation, samples were treated with apPHOTAC-I-2C and irradiated for 4 hours (100 ms per 10 s), then either incubated in darkness or 525 nm irradiation to promote optical deactivation. Under green light, BRD3 levels began to recover by 12 hours, nearing baseline levels after 24 hours (Fig. 6E). Again, extended incubation without additional light input did not halt degradation, and BRD3 levels remained suppressed throughout the experiment. This behaviour contrasts with parent azobenzene-based PHOTAC-I-3, which undergoes thermal relaxation and self-deactivates in hours without continuous activation.¹¹ Combined, these observations illustrate the attractive photophysical profile of arylazopyrazoles, enabling fully light-defined ON/OFF control.

Next, we examined how photoisomerization influences the mechanistic steps and outcomes of PROTAC-mediated TPD. Prior studies have demonstrated that cis-isomerization enhances ternary complex formation in a manner consistent with light-activated degradation.^13,16 However, the extent to which photoisomerization also modulates E3 ligase binding, particularly when the switch is positioned near the binding site, has remained unclear. Motivated by this uncertainty, we evaluated isomer-dependent CRBN engagement for apPHOTAC-I-2C using a TR-FRET competitive binding assay (Fig. 7A). Samples were either maintained in the dark (trans-enriched) or pre-irradiated to their PSS (cis-enriched) and incubated with recombinant CRBN/DDB1 and Thalidomide-Red tracer ligand, after which IC₅₀ values were determined. Both conditions retained sub-micromolar IC₅₀ values comparable to thalidomide control (37.9 nM), with modest differences between dark (15.9 nM) and irradiated (33.5 nM) states. Notably, irradiation produced a ∼2-fold decrease in affinity. This divergence between CRBN binding affinity and degradative efficacy suggests isomer-dependent degradation may be predominantly driven by productive geometries for ternary complex formation in the cis state, even when isomerization perturbs CRBN engagement.²⁴ This finding addresses a key mechanistic uncertainty in our azo-extension strategy and refines our broader understanding of how photoswitches influence PROTAC behaviour.

Fig. 7. Photoisomerization of apPHOTAC-I-2C impacts binary CRBN binding and neosubstrate degradation bias. (B) TR-FRET competitive binding of apPHOTAC-I-2C to CRBN–DDB1. Compounds were tested for displacement of thalidomide-red from CRBN–DDB1 using a TR-FRET assay. HTRF signal (%) reports the fraction of fluorescent tracer remaining bound to CRBN–DDB1, normalized to DMSO controls. Plotted values represent mean ± SD from three independent experiments. Thalidomide served as a positive control for probe displacement. IC₅₀ values were determined using a 4-parameter logistic (4PL) sigmoidal dose–response model. (A, Left) Neosubstrate degradation (Ikaros, Aiolos, and GSPT1) of apPHOTAC-I-2C evaluated by Western blot after incubation for 24 h in the dark or with 360 nm irradiation (100 ms per 10 s). 1 μM lenalidomide (LEN) and 1 μM dBET1 served as positive controls for comparison. (A, Right) Quantified results of degradation at 10 μM from three independent experiments, shown as the mean ± standard deviation. Statistical analysis was performed using one-way ANOVA. Significance levels: ns = not significant, * = p < 0.05, ** = p < 0.01.

While IMiD-based PROTACs like apPHOTAC-I-2C are widespread, a known concern is their tendency to induce off-target degradation via neosubstrate recruitment to CRBN.²⁵ Given our structural modification near the CRBN-binding domain, we therefore evaluated apPHOTAC-I-2C neosubstrate degradation by Western blot analysis (Fig. 7B). Specifically, we monitored IKZF1 (Ikaros), IKZF3 (Aiolos), and GSPT1, three well-established CRBN neosubstrates, after an extended 24 hour treatment in dark and light conditions (Fig. 7A and S19). Among the concentrations tested, 10 μM showed the strongest light-dependent changes among neosubstrates (Fig. 7A). As anticipated, BRD3 levels decreased 4.1-fold upon irradiation compared to dark conditions (p = 0.029). Some dark degradation was observed, though this was abolished ≤1 μM. Among the neosubstrates, Ikaros exhibited a modest but significant 1.6-fold reduction in abundance (p = 0.041). Inversely, GSPT1 levels increased upon irradiation, corresponding to a significant 1.8-fold increase relative to dark (p = 0.0057). Aiolos levels were not impacted across any light condition or concentration (Fig. S18). While the underlying mechanism is unclear, these changes indicate distinct differences in protein-CRBN engagement profiles between isomers. To our knowledge, this represents the first reported screening of light-modulated neosubstrate degradation by photoswitchable PROTACs. While limited in scope, these findings underscore the value of off-target profiling when developing photoswitch-embedded PROTACs.

Last, to evaluate the generalizability of the apPHOTAC scaffold, we synthesized the apPHOTAC-II series as optical degraders of purine nucleoside hydrolase MTH1 (or NUDT1), based on ATAG 2139 (Fig. 1B).²⁶ MTH1 plays a critical role in neutralizing oxidized nucleotides to mitigate DNA replication stress, a function found to be hijacked in unregulated cancer cells.²⁷ MTH1-targeting PROTACs have already been developed for degrading endogenous MTH1 for functional interrogation and MTH1-tagged proteins for chemogenetic protein control (via the AchillesTag degradation system), both functions that may benefit from an optical trigger.^26–28 These compounds feature an arylazopyrazole-lenalidomide ligand, an aminoalkane linker, and a ligand targeting MTH1; synthetic details and characterization are provided in the Supplementary Materials (Fig. S8–S10 and S23).

Among the molecules tested, apPHOTAC-II-2C exhibited superior light-dependent control (Fig. 8A and S19). Interestingly, the yellow-pigmented MTH1-targeting ligand appeared to attenuate UV/violet light absorption and reduced cis-isomer enrichment relative to the apPHOTAC-I series (58% cis under 360 nm light). Regardless, apPHOTAC-II-2C induced substantial degradation upon pulsed irradiation, with MTH1 largely eliminated after 4 hours (DC₅₀ = 8.5 nM, D_max = 70%). At this interval, dark treatments showed no significant degradation relative to vehicle controls (Fig. 8C and D). This light-dependent phenotype was recapitulated in Jurkat cells, indicating that the optical control of apPHOTAC-II-2C extends across distinct cellular contexts (Fig. S21). While extended 24 hour treatment in RS4;11 induced degradation in both dark and light conditions, dark activity was abolished at concentrations >1 μM due to a pronounced trans-specific hook effect (consistent in all apPHOTAC-II compounds) (Fig. S22). Unlike apPHOTAC-I-2C, no significant light-dependent trends in neosubstrate degradation were observed at this interval, indicating an altered target engagement profile with a different warhead (Fig. S20A and B). Even so, the variant displayed equivalent isomer-dependent differences in CRBN binding under dark (IC₅₀ = 14.6 nM) and light (IC₅₀ = 33.0 nM) conditions, consistent with preserved CRBN engagement. Finally, apPHOTAC-II-2C was inactivated in the presence of proteasome or neddylation inhibitors, consistent with a CRBN-dependent mechanism (Fig. S20). In sum, apPHOTAC-II-2C is a potent, light-dependant nanomolar degrader of MTH1, and its compatibility with degron-tagging (via the AchillesTag) offers a new foothold for extending optical control to genetically encoded protein targets.²⁶

Fig. 8. Optical control of MTH1 levels by apPHOTAC-II-2C. (A) Switching of apPHOTAC-II-2C between the trans isomer (left) and the cis isomer (right). (B) ¹H-NMR spectra of apPHOTAC-II-2C at different photostationary states (PSS) upon irradiation in d₆-DMSO. Arrows indicate signals and corresponding integrations used for PSS determination. (C) Western blot after treatment of RS4;11 cells with apPHOTAC-II-2C for 4 hours at different concentrations. Cells were either irradiated with 360 nm light (60 s continuously, followed by 100 ms pulses every 10 s) or kept in the dark. MTH1-targeting ATAG 2139 (1 μM) and co-incubation with 1 μM neddylation inhibitor MLN 4924 (MLN) acted as positive and negative controls. (D) Quantified results of three independent experiments shown as the mean ± standard deviation.

Conclusions

In this study, we expand photoswitchable PROTAC design by developing new synthetic strategies for embedding photoswitches into CRBN-binding lenalidomide and benzamide-based derivatives. Arylazopyrazole-based PHOTACs emerged as the most effective, with apPHOTAC-I-2C providing robust, reversible, persistent, and fully light-controlled degradation of BET proteins. The arylazopyrazole-lenalidomide scaffold refines the previous azobenzene-based design, and its quantitative isomerization and high bistability support sustained degradation without the need for repeated light exposure. The successful development of an MTH1-targeting apPHOTAC-II-2C further demonstrates the modularity of this scaffold.

The observed differences across PHOTAC chemotypes reinforces that the azo-extension strategy can substantially reshape the mechanism and outcomes of targeted protein degradation. While we primarily examined the photophysical and functional consequences of photoswitch derivatization, future studies may benefit from evaluating its impact on physicochemical properties (e.g. uptake, lipophilicity), especially for more substantial modifications (e.g. tetra-ortho-chlorination). Furthermore, evidence of light-modulated neosubstrate degradation underscores an important consideration for future PHOTAC development and could impact the development of photoswitchable IMiDs. Overall, these results establish new entry points for constructing targeted protein degraders and offer guiding principles for future photoswitchable PROTAC design.

Author contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI).

Supplementary information is available. See DOI: https://doi.org/10.1039/d6md00266h.