RuPHOTACs Provide Photocontrol Over Protein Degradation with Optimized Properties for Biological Applications

TL;DR

This paper introduces RuPHOTACs, a new type of light-controlled molecule that improves the precision and effectiveness of protein degradation in cells.

Contribution

The novel use of Ru(II) photocages in PROTACs enables precise light-controlled protein degradation with enhanced selectivity and in vivo efficacy.

Findings

RuPHOTACs showed high selectivity for light-triggered activation and improved potency against target proteins.

The system effectively reduced levels of c-MYC and PIM1 in cells using low energy red light.

A new reporter system using Dendra2 fusion proteins accurately measures protein degradation rates in live cells.

Abstract

PROteolysis TArgeting Chimeras (PROTACs) are bifunctional molecules that catalyze degradation of selected proteins by inducing protein:protein interactions (PPIs) between E3 ubiquitin ligases and the protein of interest. A critical limitation is undesired effects in untargeted tissues, necessitating approaches to impose spatiotemporal control over PROTAC function. Here we present Ru(II) photocages that can be released with low energy light, providing triggered PROTAC activity on demand. The systems, termed Ruthenium-based PHOToActivated Chimeras (RuPHOTACs), were validated by targeting bromodomain-containing proteins, which act as crucial epigenetic regulators, and also strongly reduced levels of c-MYC and PIM1. The novel RuPHOTACs demonstrate that the incorporation of metal components within organic PROTACs confers multiple advantages for light-controlled systems for chemical biology applications, including the highest selectivity for activation in the light vs. the dark, improved potency against the target proteins of interest, and increased efficacy in vivo using low energy red light. To efficiently monitor protein degradation via optical means, a new strategy was implemented by creating a fusion of a photoconvertible protein, Dendra2, with the proteins of interest. This bifunctional reporter system for live cell analysis decouples protein degradation efficiency and rates from signals arising from new protein production, and is superior to prior reporter systems described for PROTACs because it directly measures the true degradation rate of the target protein in intact, viable cells while simultaneously tracking newly synthesized protein without perturbing translation, proteostasis, or cell health.

Figures

b](#F2)–[d](#F2); see Supplementary Figure S18 for additional expansions of the mass spectra and theoretical isotope patterns). The use of orthogonal techniques confirmed the selective photochemistry of 8, with release of 3 and producing a single Ru(II) product 9. The clean and quantitative photoreaction (see [Figure 2d](#F2) inset) is an important advantage over photocages such as the 2-nitrosobenzyl group, where the cage release is often incomplete, and produces reactive 2-nitrosoarenes, which undergo multiple reactions in biological systems, including N-nitroso aldol reactions, nitroso-ene

e](#F3)–[j](#F3) and Supplementary Figure S26, the impact of the PROTACs, RuPHOTAC 8, and BRD inhibitor JQ1 were readily monitored by observing changes in the emission of BRD-D2 fusion proteins. The loss of the emissive signals was abrogated by treatment with JQ1, which competes with the PROTACs for binding to the BRD2, or with the proteasome inhibitor Bortezomib (BTZ), which prevents protein degradation. Notable differences were observed in the apparent potency of the compounds based on monitoring the green vs. red forms of the protein. As the photoconverted red population of protein reports

d](#F4). We note that this is qualitative, rather than a precise and quantitative evaluation, as different experimental conditions were used for the three systems, but the relative comparison of features is instructive. Due to the photocleavage mechanism, 8 was activated with 660 nm light while pcPROTAC1^[9](#R9)^ was uncaged with 365 nm light and PHOTAC-I-3^[12](#R12)^ required irradiation with 390 nm light to produce the active *cis*-isomer. Both RuPHOTAC and PHOTAC-I-3 exhibited higher percentages of light activation (100 and 90% respectively) in contrast to pc-PROTAC1, which produced only

Full text

Introduction

Selective degradation of proteins, rather than direct enzyme inhibition, has emerged as a powerful strategy to modulate biological functions in vitro and in vivo.^1–3^ PROteolysis TArgeting Chimeras (PROTACs) are bifunctional small molecules that simultaneously bind a target protein of interest (POI) and an E3-ubiquitin ligase, promoting the ubiquitination and degradation of the target protein by the proteasome.^4^ Like other small molecules, PROTACs generally possess good tissue distribution and the ability to target intracellular proteins.^1^ PROTACs have been developed to degrade medically important targets, such as BRD4,^2, 5–6^ FKBP12,^7^ and BTK,^8^ and the clinical utility of PROTACs has been validated with the entry into Phase III trials of ARV-110 and ARV-471, which target the androgen receptor and estrogen receptor. However, a persistent concern for PROTACs is undesired and potentially dangerous on-target side-effects, where systemic application can impair essential functions in healthy tissue. Notably, both ARV-110 and ARV-471 degrade proteins for which systemic inhibitors exhibit excellent safety profiles, with multiple anti-androgens and selective estrogen receptor modulators (SERMs) in clinical use. For PROTACs to gain broad utility against diverse and important medical targets, it is essential to gain a means of control to limit activity to desired tissues or physiological regions. This would potentially translate to an acceptable safety profile for compounds that target the many proteins which play aberrant roles in disease but perform essential functions in healthy tissues and organs.

Control over the site and timing of activating or inactivating biological regulators is a challenge that has been described as “the Holy Grail of chemical biology and drug discovery”,^9^ and requires sophisticated chemical design. Recently, “conditional” PROTACs have been developed, where chemical components are introduced so that different stimuli can control activity.^10^ Such molecular systems include organic moieties as photoswitchable^11–12^ and photocaged^9, 13–14^ protecting groups to enable control over protein degradation with light. These are innovative and promising models, but they suffer many restrictions. Key drawbacks include the following: (1) The photo-index, PI (the ratio between activity without and with irradiation) is generally very low, with values rarely exceed 10; this does not allow for significant control over biological processes (vide infra). (2) The active isomers of photoswitchable PROTACs possess low thermal stability, which results in progressive inactivation over time. (3) Organic photocages produce multiple potentially toxic side products upon photoactivation. (4) The reported photocages or photoswitches for PROTACs require high energy light that is not biologically compatible due to the cytotoxicity, mutagenicity, and genotoxicity of these photons. In addition, the penetration depth of short wavelengths of light is extremely limited, and significant efforts to extend the absorption profile of known chromophores have been accompanied by only modest changes in the activation wavelengths, and thus small improvements in penetration depth. For example, a recent synthetic tour de force for photocaged PROTACs shifted the activation wavelength from 365 to 405 nm light.^15^ The authors achieved the highest PI ever reported for a PROTAC, with a 100-fold improvement following photoactivation. However, 365 nm light penetrates ~0.75 mm in skin, and shifting to 405 nm increases the light penetration depth only to 1 mm. In contrast, 660 nm light penetrates ~ 5 mm, while the near-IR region of 700–900 nm improves this to ~6 mm, making red- and near-IR activated systems strongly preferred for biological and in vivo applications. However, this requires a different type of chromophore.

We anticipated that the utilization of inorganic photoprotecting groups, specifically Ru(II) polypyridyl complexes, can overcome the current challenges for conditional PROTACs (Figure 1). Different “caged” functional groups have been protected by coordination to Ru(II) scaffolds, including imidazole, pyridine, diazines, amines, nitriles, and thioethers.^16^ Each of these small functionalities can be incorporated in more sophisticated organic molecules for light triggered release, as reflected by pioneering work in this area.^17–20^ Recently we demonstrated unprecedent control over enzyme inhibition, providing a PI of 6,300 using deeply penetrating red light.^21^ Here we expand the toolkit of Ru(II) photocages with molecular optimization in order to regulate PROTACs. These systems provide the first example of employing a metal complex to control post-translational modifications, and regulation of protein:protein interactions (PPIs). The utility we demonstrate highlights multiple key advantages of the Ru(II) photocages designed for PPI control, which we term Ruthenium-based PHOToActivated Chimeras (RuPHOTACs). RuPHOTACs provide an enabling technology that give functional selectivity based on irradiation without the drawbacks detailed above, serving as a general method for controlling heterobifunctional molecules that regulate PPIs.

Results

Design and synthesis of PROTACs for photocaging and evaluation of targeted protein degradation

To demonstrate proof of principle, PROTACs were designed to target bromodomain and extraterminal domain (BET) proteins. As shown in Figure 1, the newly synthesized PROTACs are based on dBET1, and incorporate JQ1^22^ to bind the POI, linked to thalidomide, a ligand of the E3 ligase cereblon (CRBN). The key feature of the design strategy was the incorporation of different coordinating groups (thioether, pyridine, pyrimidine, bipyridine) to form bonds to the Ru(II) center; these bonds could later be broken with light activation. The use of the dBET core allowed for a rigorous comparison of the novel RuPHOTACs to previously investigated systems containing organic caging and switching groups.

As incorporation of the metal center was designed to provide steric and electronic features to impede PPIs, the effect of both the position and the nature of the coordinating group on the free PROTACs was first evaluated. The thioether (1) was investigated as it provides a bioisostere replacement of one methylene in the middle of the linker, resulting in a minimal structural change from dBET1. Alternatively, pendant pyrimidine and pyridine were appended via coupling with the amide nitrogen of the thalidomide moiety in 2 and 3. Compound 4 incorporated the pyridine through the amide nitrogen of the JQ1 moiety. Bidentate ligands were also incorporated, including 6-substituted 2,2’-bipyridine in compound 5 and a 6,6’-methylated bipyridine analogue in 6. These were connected to the PROTAC via coupling with the amide nitrogen, as in 2 and 3.

The cytotoxicity of the coordinating PROTACs was assessed in the Ramos cell line, as it is dependent on BET activity, and the effective half-maximal response (EC_50_) for cell death was determined. Chemical modifications of the linker and appended groups resulted in variability of biological responses (Table 1, Figs S19, 22, and 23). Compound 4 was inactive, as incorporation of even small substituents at the amide next to JQ1 inhibits binding, consistent with previous findings with pc-PROTAC1.^9^ Compound 2 exhibited the same potency as dBET1, and compounds 1 and 3 were slightly more potent. Notably, both 5 and 6 were 3–4 fold more potent than dBET1, with EC_50_ values of 0.09 and 0.13 μM respectively.

Given the known promiscuity of JQ1, which is a pan-BET inhibitor, the impact of the new PROTACs on degradation of BRD2, BRD3, and BRD4 were assessed by immunoblot. The progressive reduction in cellular levels of the POIs was evaluated in dose response, and the blots were quantified and ratioed to the no compound control to give the half-maximal concentration for degradation (DC_50_). While the approximate DC_50_ values for dBET1 for BRD2, BRD3, and BRD4 were 78, 5, and 313 nM, respectively, this was shifted to ~3–30-fold greater potency for 3, with DC_50_ values of 2.6, 1, and 133 nM (Supplementary Figure S23, S30, S31).

The hook effect is a common characteristic of PROTACs where activity diminishes at high concentrations as the PROTAC individually saturates the binding sites of the POI and E3 ligase, preventing formation of the functional ternary complex. A hook effect was observed with 3, which occurred at lower concentrations than seen for dBET1 (for example, at 10 M for BRD4 and BRD2, but surprisingly, no hook effect was observed for BRD3). Compounds 5 and 6 were even more potent, with DC_50_ values for BRD2 ~0.4 nM, and for BRD3 of 0.3 and 0.2 nM, respectively (Supplementary Figure S24, S32, S33). The DC_50_ value for BRD2 and BRD3 degradation represents a 195- and 25-fold improvement over dBET1.

Next, the efficacy of the new PROTACs to reduce levels of two oncogenes, c-MYC and proviral integration for the Moloney murine leukemia virus isoform 1 (PIM1), were probed. c-MYC is a transcription factor with extensive activities in malignant progression across various cancer types, and also exerts deleterious effects on host immunity and the tumor microenvironment (TME).^23^ BRD4 controls expression of c-MYC, but paradoxically, BRD4 also has a direct destabilization effect on cMYC, and PROTAC MZ1 has been shown to actually increase c-MYC levels,^24^ highlighting the subtle and often unpredictable impacts of PROTACs on these key targets. In contrast to c-MYC, PIM1 is under the transcriptional control of the JAK/STAT pathway rather than BET proteins, but prior reports indicated that JQ1 and dBET1 treatment modestly reduce both PIM1 protein and mRNA levels.^2^ PIM1 is a constitutively active serine/threonine kinase oncoprotein implicated in a number of human cancers, and drives the tumor phenotype and resistance to chemotherapy,^25^ and c-MYC collaborates with PIM1 to reduce sensitivity for apoptosis induction. Accordingly, both oncoproteins are key targets, and c-MYC was probed as an on-pathway downstream product of BRD activity, while PIM1 was probed as an off-pathway reporter.

As anticipated, PROTACs that induced BRD4 degradation also reduced c-MYC levels. Gratifyingly, 3 reduced c-MYC levels more potently than dBET1 (29 vs. 130 nM; Supplementary Figures S23, S30, S31), and 5 and 6 were much more effective, as they were able to essentially eliminate c-MYC at 10 nM (Supplementary Figure S24, S32, S33). These PROTACs also exhibited a remarkable enhanced efficacy in reducing PIM1 levels. While dBET1 and 3 only had subtle effects on this protein, both 5 and 6 caused protein elimination at 100 nM concentrations. The potency against PIM1 is particularly noteworthy as it is equivalent or superior to recently reported PROTACs designed specifically for this protein,^26^ and indicates these new PROTACs have potentially synergistic activities against multiple targets.

Development of RuPHOTACs

Efforts to coordinate the bidentate ligands in 5 and 6 to Ru(II) centers resulted in poor yields even under forcing conditions, which motivated advancing the PROTACs containing sterically unencumbered monodentate ligands. As the pyridyl group is easier to coordinate to Ru(II) than a thioether, compound 3 was prioritized for caging. Two Ru(II) scaffolds were synthesized based on a building block approach, utilizing different co-ligands to independently adjust key physiochemical and photophysical features that were anticipated to regulate the behavior of the RuPHOTACs. A 2,2′-biquinoline (biq) core was used for the ligands in both cases to provide two key features. First, the extended conjugation of this system alters the photophysics and photochemistry of the metal complex by causing a shift in the metal to ligand charge transfer (MLCT) to longer wavelengths, providing for activation with 660 nm light.^27–28^ Second, the presence of the additional rings in the biq ligand creates steric clash within octahedral metal complexes, which lowers the energy of a dissociative metal centered (MC) excited state. These sterically encumbered systems can populate ^3^MLCT excited states with low energy light, similar to many coordination complexes, but unlike prototypical photoactive metal complexes, they are capable of populating the ^3^MC state, which ejects a ligand.^27^ When there is a monodentate ligand present in the metal complex, it is the preferred leaving group,^21, 28^ allowing for selective delivery of caged molecules.

Two different modifications were investigated to diminish the affinity of the intact photocaged PROTAC for the protein targets. First, a derivative of biq, [2,2′-biquinoline]-4,4′-dicarboxylic acid ligand (bicinchoninic acid, bca)^28^ was incorporated in compound 7, as negatively charged groups have previously helped to reduce protein binding of Ru(II) complexes.^21^ The bca was combined with a tridentate ligand, 2,2’;6’,2”-terpyridine (tpy), producing an overall neutral scaffold for addition of the PROTAC. Alternatively, in compound 8 a long PEG4 group was appended to the tridentate tpy ligand to impede the interaction of the intact Ru(II) caged PROTAC with the POI and E3 ligase. As coordination of the hydrophobic PROTAC 3 significantly diminished the solubility of the prodrug in the neutral state in compound 7, the scaffold in 8 was redesigned using the neutral biq ligand to give +2 charged Ru(II) compound to address the solubility issues.

Photo-activation with 660 nm light in water was monitored by UV/Vis, HPLC, and mass spectrometry (Figure 2b–d; see Supplementary Figure S18 for additional expansions of the mass spectra and theoretical isotope patterns). The use of orthogonal techniques confirmed the selective photochemistry of 8, with release of 3 and producing a single Ru(II) product 9. The clean and quantitative photoreaction (see Figure 2d inset) is an important advantage over photocages such as the 2-nitrosobenzyl group, where the cage release is often incomplete, and produces reactive 2-nitrosoarenes, which undergo multiple reactions in biological systems, including N-nitroso aldol reactions, nitroso-ene reactions, and rearrangements.^29^

The intact RuPHOTAC exhibited high thermal stability, with no degradation observed over 72 hours (Supplementary Figure S19). This was reflected in the cytotoxicity studies of 8 in Ramos cells, which demonstrated minimal effects of the prodrug at concentrations of up to 10 M (Dark; Figure 2e, Table 1). Irradiation with red light activated the PROTAC, resulting in an EC_50_ of 0.33 M and giving a PI of 87. A hook effect was observed at 30 M, consistent with the expectation that cell death is driven by on-target action of the PROTAC rather than an unrelated photochemical product. The dose response was relatively steep (Hill coefficient = 3); this had the effect that a concentration of 3.33 M afforded full viability in the dark and 100% cytotoxicity with red light activation.

To ensure the cytotoxicity was associated with BET protein degradation, immunoblots were performed in dose response. As shown in Figure 3a–d, irradiation of RuPHOTAC 8 with 660 nm light induced degradation of BRD4, with a DC_50_ value of 140 nM, providing a PI value of 111. Light activated DC_50_ values for BRD2 and BRD3 degradation were 8 and 0.4 nM, which reflects a 10–30-fold improvement in potency over 3; for BRD3, the PI was 500, which is the largest PI value for any photo-PROTACs reported (Figure 3c, d). Interestingly, while dBET1 and 3 did not induce complete BRD4 degradation, the light activated system 8 was able to achieve this. Notably, photoactivated 8 also induced significant reductions in both c-MYC and PIM1 levels (DC_50_ of 22 nM and of 161 nM, respectively). The PI values for BRD3, BRD4, c-MYC, and PIM1 all exceed the target 100-fold cut-off (Figure 3c).

Attaining high PI values has been a long-standing challenge for photo-controlled systems, and values of 10 or less have been reported in most cases. Unfortunately, this is insufficient for achieving control over biological processes, which becomes evident when comparing dose response curves. Simulated data in Figure 3d and Supplementary Figure S22 highlights this key feature. Assuming a Hill coefficient of 1 (the standard value under conditions with no cooperativity) and a PI of 10, an undesired response of 50% in the dark occurs at a concentration where a 90% response is produced with the “light” sample. Unfortunately, this ratio is insufficient for achieving a therapeutic window, and is unsuitable for chemical probes used to study biological systems. In contrast, a 100-fold PI provides a 10% response in the dark at the EC_90_ for light activation, which is potentially acceptable; this value falls further to 3% at a 300-fold ratio. By considering the full dose response, it is clear that the therapeutic window is not determined solely by the separation of the EC_50_ values, which are usually the only values reported. Rather, the true window for photocontrol depends critically on the relative slopes of the dose responses and decreases with lower Hill coefficients (Supplementary Figure S22). Unless there is significant cooperativity which increases the Hill coefficient – a rare occurrence – the data show a minimum ratio of 100 is needed in real world applications for selective activation is to be realized. With rational chemical optimization, this was obtained with the RuPHOTAC, and the outcome highlights what can be achieved if potent PROTACs are protected with an optimized photocaging moiety.

Molecular docking

To rationalize the molecular features of the new PROTAC 3 that regulate its binding to its targets, and to predict the efficacy of installing the Ru(II) cage as a blocking group in 8, docking studies were performed using Autodock Vina.^30^ The crystal structure for BRD4_BD1_, CRBN, and DDBI was used, as shown in Figure 4a and d. PROTAC 3 docked with both the JQ1 and the pomalidomide occupying the anticipated binding sites. Interactions included an expected edge to face π interaction between the pomalidomide moiety with Trp81 of BRD4. This residue also contacts His353 of CRBN, which then interacts with Met149 of BRD4. The oxygen of the carbonyl amide in the pomalidomide moiety forms a strong H-bond with His378. The JQ1 moiety of 3 is nestled in a hydrophobic pocket of BRD4, and the oxygen atom in the carbonyl amide proximal to the JQ1 moiety forms an H-bond with Gln85.

It is known that the interprotein contacts mediated by PROTACs are plastic, with diverse binding conformations of both the small molecules and proteins, as illustrated by protein crystal structures.^31^ As anticipated, the conformations of the linker and the orientation of the JQ1 moiety were variable in the docked structures for both 3 and 8, but in all cases the pendant Ru(II) protecting group of 8 was presented at the interprotein surface. Due to the size of the complex (155 Å^2^) relative to the protein:protein interface between BRD4 and CRBN (550 Å^2^)^31^ and the presence of the long pendent polyethylene glycol chain, the Ru(II) group produces significant steric and hydrophilic interference, as shown Figure 4e, reducing the affinity of the RuPHOTAC for the ternary complex.

The structures of the JQ1 binding sites of BRD2, BRD3, and BRD4 are similar,^32^ as shown in Figure 4c. However, it is known that PROTACs with different linker lengths and linkage positions can result in distinct binding conformations in the ternary complex, complicating structure-based design of caged degraders. Moreover, the identity of the linked E3 ligase bait can promote selective BET protein removal, as is the case with MZ1,^6^ highlighting the sensitivity of the protein:protein interface to subtle differences between the BRD proteins. The differences we observed in PROTAC efficacy in inducing degradation of BDR3 and BRD4 compared to BRD2 are likely driven by the different features of the interface between the target proteins and the CRBN. However, these cannot currently be predicted based on the analysis of currently available crystal structures, reflecting a limitation of PROTAC design, which commonly requires empirical evaluation to pick promising candidates.

Optical detection of protein degradation using the photoconvertible protein Dendra2

The evaluation of protein degraders commonly relies on end point data provided by immunoblot, which is a tedious approach that does not allow for longitudinal studies or higher throughput screening of multiple agents. To overcome this limitation, an alternative method was envisioned that would rely on optical detection of POI protein levels in live cells for rapid screening of compound efficacy. To achieve this, the photoconvertible protein Dendra2 (D2)^33^ was fused to the target BRD proteins, and Flp-In HEK reporter cell lines was generated. D2 is expressed as a green fluorescent protein, with emission at 507 nm, but exposure to 405 nm light induces a chemical reaction of the chromophore that results in extended conjugation, producing emission centered at 573 nm. This allows for the creation of two pools of protein: the post-light exposure form, emissive in the red region of the spectrum, and the green emissive pool, which continues to increase as transcription and translation continues. This feature has been applied to monitor protein trafficking using D2 fusions with POIs, and D2 can also be used as a reporter of protein synthesis and degradation rates.^34–35^ Monitoring the disappearance of the red emissive form of the fusion protein gives an optical readout of protein degradation rates, while the green form reports on the combined effects of protein degradation and new protein synthesis (Figure 3g). Unlike previously utilized ratiometric two-reporter systems, which require co-expression of a second, non-degradable protein that can itself perturb proteostasis or compete for degradation machinery, this strategy uses a single fusion and the intrinsic photophysics of the tag to achieve “ratiometric” information with minimal added experimental complexity.

As shown in Figure 3e–j and Supplementary Figure S26, the impact of the PROTACs, RuPHOTAC 8, and BRD inhibitor JQ1 were readily monitored by observing changes in the emission of BRD-D2 fusion proteins. The loss of the emissive signals was abrogated by treatment with JQ1, which competes with the PROTACs for binding to the BRD2, or with the proteasome inhibitor Bortezomib (BTZ), which prevents protein degradation. Notable differences were observed in the apparent potency of the compounds based on monitoring the green vs. red forms of the protein. As the photoconverted red population of protein reports on the process of protein degradation mediated by the PROTACs without any signal arising from new protein synthesis, full protein elimination was achieved. This contrasts to data from monitoring the green emissive population, which provides more discrimination between the efficacy of individual PROTACs. Notably, the new PROTACs 3, 5, and 6, and RuPHOTAC 8 are all more effective than both dBET1 and MZ1 when the green signal is observed, allowing for rank ordering of the agents without use of immunoblot. Mechanistic questions could also be probed. BET inhibitor JQ1 competed with the PROTACs, blocking protein degradation more effectively than inhibiting the proteasome, indicating that JQ1stabilizes the protein differently than simply blocking its degradation. This finding was further corroborated by a dose response (Supplementary Figure S27) showing that JQ1 prevents ~50% of the degradation of the red population of BRD3-D2, but this action is concentration independent. In contrast, JQ1 stabilizes the newly formed BRD3-D2 by >50%, and in a concentration dependent manner. The results from monitoring the two populations of the BRD proteins highlights the disparate impacts of small molecule regulators on the protein lifecycle, which may have co-translational effects.

Biological variability of protein degradation

PROTAC activity is known to be cell line and condition dependent, and similar PROTACs exhibit varying efficacy depending on the experimental system used. To evaluate the PROTACs in a different cell type, the Diffuse Intrinsic Pontine Glioma (DIPG) cell line SF8628 was investigated. DIPG is an invariably fatal childhood brainstem tumor, and BET proteins have been associated with DIPG pathogensis,^36^ with BET inhibitors and degraders under consideration as potential treatment strategies.^37^ Both dBET1 and the new PROTACs were effective in degrading BRD2 and BRD3 and reducing c-MYC levels, but not BRD4 or PIM1in the SF8628 cell line (Supplementary Figure S25). This cell line is not as sensitive to loss of BRD function, and the cytotoxic effects of the compounds were muted compared to the Ramos cell line (Table S1). These results highlight the importance of BRD2 and BRD3 in c-MYC transcription and the variability of PROTAC activity by cell lines.

In vivo validation of RuPHOTACs

Given the promising results with red light in vitro, targeted protein degradation was performed with zebrafish as an in vivo xenograft model. GFP-labeled Ramos cells were injected into the yolk sac before compound treatment and irradiation (Figure 5 a–c), followed by assessment of the number of viable cells per fish. Photocaged 8 had no significant impact in the dark, while photoactivation at 1 and 10 M resulted in the elimination of GFP-labeled cells. This efficacy in vivo paralleled the in vitro findings and further confirmed that the addition of the large Ru(II) protecting group does not impede uptake. In fact, 8 proved to be superior to dBET1, with a ~3-fold larger antitumor effect at the same concentration. The RuPHOTAC was also well tolerated as determined by studies in zebrafish embryos. The effects of 8 in the dark matched the no compound control, and the toxicity of 8 following irradiation was similar to irradiation alone, demonstrating the biocompatibility of the Ru(II) caging group.

To provide a direct comparison of the RuPHOTAC system with previously reported organic photocaged PROTACs and photoswitching PROTACs, a radar plot of key features is shown in Figure 4d. We note that this is qualitative, rather than a precise and quantitative evaluation, as different experimental conditions were used for the three systems, but the relative comparison of features is instructive. Due to the photocleavage mechanism, 8 was activated with 660 nm light while pcPROTAC1^9^ was uncaged with 365 nm light and PHOTAC-I-3^12^ required irradiation with 390 nm light to produce the active cis-isomer. Both RuPHOTAC and PHOTAC-I-3 exhibited higher percentages of light activation (100 and 90% respectively) in contrast to pc-PROTAC1, which produced only 50% of the active PROTAC. While cis-PHOTAC-I-3 exhibited good activity, it can isomerize back to an inactive trans form in the dark. Alternatively, continuous irradiation would be required for full efficacy, as noted by others^11^ who developed bistable azobenzene PROTACs, which exhibited reduced potency when activated, but could be deactivated with light. BRD4 degradation was observed in the dark at a higher concentration for PHOTAC-I-3 and the RuPHOTAC compared to pc-PROTAC1, likely due to the larger sizes of the photoactive groups causing more steric clash. In combination, these features are responsible for the optimized features of 8 and its improved biological utility.

Discussion

Extensive efforts have been expended to develop systems that regulate the activity of pharmacological agents using light. These include compounds used for photodynamic therapy (PDT; generating ROS), photocaging strategies, and photoswitching approaches. Each has different strengths and limitations. As the only catalytic approach, PDT outperforms all other systems in terms of achieving high photo-potencies at low compound concentrations, and PI values of >100 are common. However, the ROS species produced by PDT are non-specific cytotoxins, inducing apoptosis and other cell death pathways, depending on the extent of cellular damage. Moreover, this mechanism of action relies on the presence of molecular oxygen, which is depleted during the treatment process. As a result, traditional PDT agents have the limitation that they are inactive under hypoxia, which is a common feature of most solid tumors. Photocages and photoswitches are advantageous as they do not rely on ROS and can trigger very specific biological effects, similar to most drugs, but like these drugs, the disadvantage is that they are stoichiometric agents. Given the residual activity or thermal instability of most photocaged and photoswitches in the dark, photoactivation with a suitable therapeutic window for selectivity for biological applications, which requires a PI that is ≥100 vide supra, has remained an elusive goal for nearly all reported systems. Overcoming this limitation for activation of PROTACs was the goal of this work.

In fact, PI values for conditional PROTACs based on JQ1 are not impressive, regardless of the activation mechanism. This is likely due to the known polypharmacology of the ligand, and the difficulty of evading the multiple protein targets of this molecule when relying only on the steric clash or altered geometries provided by appending small moieties. The PI for cytotoxicity of PHOTAC-I-3 and pc-PROTAC1 did not exceed 8, and similarly modest therapeutic windows are apparent in other conditionally active systems, where the function is turned on by ionizing radiation,^38^ or environmental conditions, such as hypoxia,^39–40^ the presence of redox mediators,^41–42^ or the activity of specific enzymes.^43–44^ Indeed, this has prompted some groups recently to apply “tandem”^45^ or “dual-action”^46–48^ approaches, where two orthogonal stimuli are combined to induce PROTAC action. These efforts highlight both the current intense focus on producing triggerable PROTACs and the limited progress that has been achieved, and frames the significant improvement gained by the RuPHOTAC photocage detailed here. Another key limitation has been translation, and for critical evaluation of chemical leads with triggered activity, it may be beneficial for researchers to report PI values along with the Hill coefficient for each dose response, as this is key information to allow for determining the therapeutic window.

The use of the large metal complex as a protecting group in the RuPHOTAC prompted concerns about cellular permeability and tissue penetration. The in vitro and in vivo activity of 8 demonstrates that the incorporation of the Ru(II) center does not prevent uptake; rather, it appears to enhance biological activity. How the Ru(II) center achieves this outcome is the subject of ongoing mechanistic studies. In addition, the potency of the new PROTACs and RuPHOTAC against PIM1 was a fortuitous finding. The PIM family of kinases are implicated in pro-tumorigenic roles in a variety of solid tumors, and inhibitors of PIM1 has been explored in several clinical trials as potential standalone treatments over the past decade.^49^ However, PIM inhibitors cause target stabilization and resistance,^26^ which motivates approaches to PIM1 targeted degradation. The new PROTACs 5 and 6 induced maximal PIM1 degradation at 100 nM concentrations, comparing favorably to a recently reported PROTAC, PIM447-VHL-01 (750 nM).^26^ These new systems are only the second example of small molecules that induce PIM1 degradation. It is interesting that the POI “bait” bears no structural relationship to that used in PIM447-VHL-01, and this finding should open the door for more chemically diverse PIM1 degraders.

In addition to innovative chemical agents, robust biological assays are needed for high-throughput screening to evaluate PROTAC activity and optimize designs. Several complementary assay formats are used to monitor PROTAC-induced degradation, ranging from traditional immunoblot to sophisticated live-cell optical reporters. Immunoblot is reliable and can be performed on cells, tissues, and even whole organisms, and does not require genetic manipulation. However, disadvantages include the fact that it is low throughput, can only provide end point data, and is semi-quantitative unless carefully standardized. Proximity and binding assays such as the amplified luminescent proximity homogenous assay (AlphaLISA)^50^ and time-resolved fluorescence energy transfer (TR-FRET)^51^ have been developed to detect formation of ternary complexes and ubiquitination, rather than direct loss of endogenous protein. The use of optical methods for detection makes the assay suitable for plate readers for high throughput and parallel screening, but these approaches are not compatible with live cell assays and don’t always translate to degradation efficiency. Protein loss has been monitored in cells by engineering fusion proteins with optical reporters, including GFP, split GFP, and other systems such as NanoLuc^®^ and HaloTag^®^ technologies.^52^ None of these approaches, however, actually measure real-time protein degradation kinetics, as the reporter signal includes the new protein formed as translation continues. Under conditions where the rate of protein synthesis is high or a protein is being monitored that is produced in large quantities, a PROTAC can appear less effective. Alternatively, if transcription or translation is slowed, or the target protein is intrinsically unstable or produced only in trace quantities, the PROTAC will appear artificially active. This is a confounding variable that can cause significant variability between systems.

To observe protein degradation in the absence of protein synthesis, the current approach is treatment with cycloheximide to prevent any new protein production. However, cycloheximide, as a translation inhibitor, is intrinsically cytotoxic, so experimental conditions are limited to the time- and concentration-range where cell health is not adversely impacted. Additional complications arise from the activity of cycloheximide that is more specific to and deleterious for studies of protein degradation. For example, the interruption of protein synthesis is known to trigger stress pathways, such as the activation of protein kinase B (AKT), which can lead to changes in the degradation rate of other proteins.^53^ In addition, cycloheximide prevents the production of ubiquitin,^54^ which would have obvious implications with the perturbation of PROTAC efficacy. As a result, biologically benign approaches to monitor protein loss are needed.

The D2 assay presented here is the only approach, to the best of our knowledge, that allows for the decoupling of protein degradation activity and rates from the complication of signals arising from new protein production. Collectively, the features of the D2 assay we present—direct, inhibitor-free measurement of degradation of a defined protein cohort; separation from synthesis; superior kinetic and spatial resolution; and compatibility with cell health readouts—make the photoconvertible D2 target-fusion reporter system a conceptually and practically superior platform for studying PROTAC-induced degradation compared with existing fluorescent, luminescent, and biochemical assays. The differences in the end point assessments based on the D2 assay monitoring the red vs. the green emissive signal is instructive. Both dBET1 and MZ1 appear far less effective than the new PROTACs 3, 5, and 6, and RuPHOTAC 8 when the green signal is observed (Figure 3i), while all PROTACs appear equally effective when following the red signal, as all PROTACs induce complete protein removal at the observed time point (Figure 3j). Whether these differences arise from an improved ability of our novel PROTACs to induce ubiquitination co-translationally is a challenging question that remains to be determined; such studies of detailed protein lifecycle are complex and beyond the scope of this study.

Another strength of the D2 assay is its potential for providing enhanced single-cell and spatial information. Photoconvertible proteins were originally developed for tracking protein dynamics at single-cell and subcellular resolution, and this strength translates directly to PROTAC studies, enabling visualization of degradation in live cells as a function of various features, including local pools or cell-cycle states. Heterogeneity in PROTAC response, based on subpopulations of cells with slower or incomplete degradation, can be resolved by following the decay of the converted population in individual cells, which bulk biochemical or plate-based luminescence assays average out. Finally, while not demonstrated here, spatially restricted photoconversion in specific cells or subcellular regions would allow testing to define how PROTACs degrade particular localized pools more or less efficiently. This technology is at the forefront of microscopy methods, but would be difficult or impossible with nonconvertible reporters or global biochemical assays.

The RuPHOTAC presented here provides true functional control over protein degrader technologies with multiple advantages over organic photo-controlled systems. To the best of our knowledge this is also the first description of a metal-containing PROTAC where the metal is used to turn on or off biological activity. Other creative systems that contain inorganic components and systems like PROTACs have been reported, but are conceptually and fundamentally different. A metal-containing photocatalyst that generates ROS was added to a BRD ligand,^55^ akin to the pioneering work of Kodadek to selectively photo-inactivate proteins;^56^ however, the creation of indiscriminately cytotoxic species directly contravenes the fundamental premise of PROTACs and is considered mechanistically distinct. A creative approach used Pt(II) derivatives of carboplatin conjugated to pomalidomide to induce the degradation of Pt-binding proteins, using the metal center as the POI binding group.^57^ Another report investigated the incorporation of ferrocene into the linker of PROTACs, creating systems termed FerroTACs.^58^ In this case, the organometallic sandwich compound serves as a flexible hinge in the linker via the free rotations of the two cyclopentadiene rings relative to each other, and elegantly addresses key issues in PROTAC “linkerology”. The RuPHOTACs reported here provide a similarly robust approach to improved light-control over PROTAC activity. These hybrid systems showcase the successful marriage of inorganic chemistry and chemical biology with the inclusion of metal components within organic PROTACs.

Methods

Full and more detailed experimental information can be found in the Supplementary Information.

Synthesis and chemical analysis

General.

All materials were purchased from commercial sources and used without any further purification. All ^1^H NMR spectra were obtained on a Bruker Avance NEO (400 MHz) and Bruker Avance III 700 MHz spectrometers with chemical shifts referenced to residual solvent peaks. ^13^C NMR spectra were obtained on a Bruker Avance NEO (101 MHz) spectrometer with chemical shifts referenced to residual solvent peak of CDCl_3_ (δ 77.16). HRAM Mass spectra was obtained on a Thermo Scientific UltiMate 3000 UHPLC System interfaced with the Thermo Scientific Exactive Plus Mass Spectrometer at Molecular Education, Technology, and Research Innovation Center (METRIC) NC State University. Electrospray ionization (ESI) mass spectra were obtained on a Thermo Fisher QExactive mass spectrometer at the University of Kentucky Mass Spectrometry Facility. UV/Vis absorption spectra were obtained on a BMG Labtech FLUOstar Omega microplate reader or a Cary 60 spectrometer. Light activation experiments were performed using a 660 nm LED with cooling fan (Lumidox II) with an irradiance of 16 mW/cm^2^. For this LED, peak emission λ_P_ = 660 nm, and spectral line full width at half-maximum Δλ =20 nm. The Prism software package was used to analyze data.

HPLC analysis for purity.

The purity of each Ru(II) complex was analyzed using an Agilent 1100 Series HPLC equipped with a model G1311A quaternary pump, G1315B UV diode array detector and Chemstation software version B.01.03. Chromatographic conditions were optimized on a Phonomenex Luna 5 μm C18(2) 100 Å column fitted with a Phenomenex C18 guard column. Mobile phases of 0.1% formic acid in dH_2_O and 0.1% formic acid in HPLC grade CH_3_CN were used. The UV diode array detector was set to a 280 nm wavelength.

Photochemical studies.

The Ru(II) complex 8 (200 μL) was analyzed in a 96 well plate at a final concentration of 100 μM in water and a path length of 0.5 cm. UV/Vis absorption spectra were obtained on a BMG Labtech FLUOstar Omega microplate reader. Scans were taken in the dark and after 1 h irradiation with 660 nm LED array from Elixa. Then, irradiated samples were injected for HPLC analysis (Figure 2) to identify reaction products. To study the kinetic of photoconversion, 8 (200 μL) was analyzed in a 96 well plate at a final concentration of 30 μM in 10% MeOH in water and a path length of 0.5 cm. Methanol was added to improve the solubility of released PROTAC 3. The plate was irradiated with 660 nm Lumidox^®^ II 96-Well LED Arrays using 170 mW per well settings at an irradiation distance of 3 inches (56 mW/cm^2^). The samples for HPLC were collected at intervals of 0, 2, 4, 7, 10, 15, 20 and 30 min. The final sample (30 min, 100 J/cm^2^) was tested by HRMS (Figure 2, Supplementary Figure 18).

Molecular docking

Docking was initially performed using different DDB1-CRBN-BRD4_BD1_ crystal structures, including 6BNB, 6BN7, and 6BOY. The structure for 6BNB did not allow for observation of the bound PROTAC (DBET57), with a resolution of 6.34 Å, so this was not used. The 6BN7 structure contained DBET23, which including the linker for the PROTAC attached on the thiophene ring of the JQ1, opposite the site of modification used in the new PROTACs. This resulted in twisting of the JQ1 in the docked structures. In contrast, 6BOY contained the DBET6 PROTAC, substituted on the same position of the JQ1 moiety with a similar linker length, allowing for a better model for docking. Accordingly, the PDB structure 6BOY was used in all molecular docking experiments.

The structure was prepared for docking analysis in Chimera (1.19) through the addition of hydrogen atoms, charge assignment, and removal of existing ligand. The AutoDock Vina (1.1.2) built in tool in Chimera was utilized for the generation of pdbqt files for protein and ligands preparation, as well as the receptor grid box generation. The grid size was set to 25 × 30 × 25 with a grid center set to the dimensions (x, y, z): 71.5304, 40.4484, 52.2995, and the default exhaustiveness value of 8 was used. The top 10 docking poses were visually inspected and evaluated based on proximity of the JQ1 moiety and the pomalidomide to their known binding sites. The installation of the bipyridine moieties in 6 resulted in additional hydrophobic contacts, even replacing the pomalidomide in several of the docked structures, stacking with Trp386.

Biological studies

In vitro cytotoxicity studies.

Ramos cells were obtained from ATCC and maintained in RPMI 1640 media supplemented with 10% FBS and 100 U of Penicillin and 100 g/mL units of Streptomycin at 37 °C with 5% CO_2_. For cytotoxicity studies, cells were plated in 96 well plates at 100,000 cells per well in Opti-MEM (supplemented with 2% FBS and 100 U Penicillin and 100 g/mL Streptomycin). The compounds were serially diluted in Opti-MEM then added to the cells. The cells were incubated with the compounds at 37 °C with 5% CO_2_ for 1 hour followed by irradiation with red light for 60 minutes (58.7 J/cm^2^). Following a further 16 hours of incubation, resazurin was added to the cells. The cells were allowed to incubate with 73 M resazurin for 3 hours. The plates were then read on a SpectraFluor Plus plate reader with an excitation filter of 535 nm and emission filter of 595 nm.

Immunoblot.

Ramos cells were plated in Opti-MEM at 1×10^6^ cells/mL in 6 well plates with 5 mL of cells per well. Compounds were serially diluted in DMSO and added to the cells. Cells were incubated for 16 hours then transferred to 15 mL tubes and pelleted at 270 xg for 5 min at 4 °C. The media was aspirated, the cell pellets resuspended in 1 mL of ice cold PBS, and transferred to 1.5 mL tubes on ice, then pelleted at 1300 xg for 5 min at 4 °C, aspirated, followed by the addition of lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% NP40, 0.5% Sodium Deoxycholate, 5 mM NaF, 2 mM sodium orthovanadate, 5 mM sodium pyrophosphate, 5 mM EDTA, 1x G Biosciences protease inhibitor cocktail). The samples were incubated for 15 min on ice followed by centrifugation at 19,500 xg for 10 min at 4 °C. Following detergent lysis the protein concentration for each of the samples was determined by BCA. Protein lysates of 10 μg’s were loaded per lane on a 4–12% bis-tris gel and separated by electrophoresis followed by a 75 minute transfer to nitrocellulose at 100 V. The membrane was blocked for 1 hour with PBST (PBS with 0.1% Tween 20) containing 5% non-fat milk. Primary antibodies for BRD2 (Cat #ab245436) and BRD4 (Cat #ab128874) were purchased from Abcam and used at 1:10000 dilution. GAPDH and BRD3 were purchased from Santa Cruz Biotechnology and used at a 1:1000 dilution. Following an overnight incubation of the membrane with primary antibody at 4 °C, the membrane was washed with PBST and incubated with HRP conjugated secondary antibody (Jackson Labs) at a 1:5000 dilution in PBST containing 5% nonfat milk. After a 1 hour incubation at room temperature, the membrane was washed with PBST and developed with Clarity ECL (Bio-Rad laboratories). The immunoblots were imaged with a ChemiDoc MP Imaging System (Bio-Rad Laboratories).

Quantification of the blots was performed in Image Lab 6.1 (Bio-Rad), where images below the point of pixel saturation were analyzed. In brief, the bands in the immunoblots were quantified and ratioed to the no compound control. The intensity values were plotted vs. concentration to provide a DC_50_ value. NOTE: the quantification of immunoblots is qualitative. The fits to the date are intended to reflect on possible cooperativity.

Creation of Dendra2 fusion and microscopy.

Plasmids containing human BRD2, BRD3, and BRD4 were obtained from Genscript. The gene for BRD4 was amplified by PCR with primers to incorporate HindIII and EcoRV restriction sites for cloning into the plasmid pcDNA5 FRT-TO. Addition of a 6x glycine linker followed by Dendra2 (D2) at the 3’ end was carried out using the EcoRV and NotI restriction sites. For BRD2, PCR was accomplished with primers for BamHI and EcoRV restriction sites, while BRD3 utilized the EcoRV and NotI restriction sites. For BRD3 the linker consists of 3 alanine followed by 3 glycine residues and was incorporated with D2 with the NotI and XhoI restriction sites. The resulting plasmids were confirmed by sequencing.

Using the HEK Flp-In cell line (Life Technologies), plasmids were transfected and a stable pool of cells selected. For transfections, cells were seeded in Matrigel coated 6 well plates at a density of 5.5×10^5^ cells per well and allowed to adhere overnight in DMEM media (DMEM supplemented with 10% FBS and 100 U/mL penicillin, 100 g/mL streptomycin). Lipofectamine 2000 was prepared with 2 L of lipofectamine, 32 L Opti-MEM. The plasmid was prepared at 1 g with 32 mL of Opti-MEM. After 30 min, the lipofectamine:plasmid mixture was added to the cells in 1.2 mL Opti-MEM for 5 hrs. Media was removed and replaced with DMEM media and incubated for 48 hrs. The cells were dissociated and diluted 1:10 in media with 15 g/mL blasticidin and 300 g/mL hygromycin B. A stable pool was isolated after 7 to 10 days and maintained with the addition of blasticidin and hygromycin B.

Expression of BRD-D2 cell lines was determined by tetracycline induction followed by western blot to verify inducible BRD expression and imaging for dendra2 confirmation. Cells were plated in glass bottom, Matrigel coated, 35 mm dishes at 7.5×10^4^ cells per dish in DMEM media with 1 g/mL tetracycline, and incubated overnight. The media was changed to extracellular solution and the cells imaged on a Nikon Ti2. To confirm D2 photoswitching from green to red emission, the cells were irradiated with 405 nm light (LED flood array, Loctite) for 1 min and imaged.

Efficacy of BRD PROTAC compounds were evaluated by the loss of D2 green and red emission. Cells were plated at 7.5×10^4^ per 35 mm glass bottom dish in DMEM media with 1 μg/mL tetracycline and incubated overnight. The next day the media was removed, extracellular solution added, followed by D2 photoswitching. The extracellular solution was aspirated and replaced with compounds dissolved in OptiMEM supplemented with 2% FBS with 1 μg/mL tetracycline and incubated for 16 hrs before imaging. For studies involving red light activation of the PROTAC, compound was added and the cells incubated for 30 min. The cells were then placed at a distance of 3 inches from a 660 nm LED with cooling fan (Lumidox II) having a radiance of mW/cm^2^ and exposed for 15 min followed by incubation for 16 hrs. PROTAC compounds were incubated with cells at 1 M for BRD2 and BRD4, while 0.1 M was used for BRD3. Mechanism based verification studies were done with 0.1 M Velcade and 10 M JQ1.

Imaging was carried out on a Nikon Eclipse Ti2 with a 60x oil objective (1.42 numerical aperture). The LEDs were set to 10% power and images taken with an ORCA-Fusion C14440 digital camera (Hamamatsu) where the exposure time was 1s. The intensity for each nuclei was measured for a minimum of 10 cells to quantify the average intensity and distribution with the background emission subtracted from each image. The data were normalized to the untreated control and cells imaged without tetracycline induction.

In vivo studies

Zebrafish care and use.

Use and handling of zebrafish was approved by the University of Kentucky’s Institutional Animal Care and Use Committee (IACUC), protocol 2019–3399. Adult Casper (mitfa^w2/w2^; mpv17^a9/a9^) strain zebrafish were maintained at a temperature of 28 °C with a light:dark cycle of 14:10 h in compliance with IACUC animal care regulations. Eggs were collected into 1× E3 media (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl_2_, and 0.33 mM MgSO_4_) with 200 μL/L of methylene blue (Kordon brand).

Zebrafish toxicity.

Zebrafish larvae at 2-day post-fertilization (dpf) were dechorionated with Pronase (Fisher Scientific, 74–332) at a final concentration of 1 mg/mL in 1× E3 media and placed individually into a well of a 96-well plate. Compounds were added at final concentrations of: 0.01, 0.03, 0.1, 0.3, 1, 3 and 10 μM. Ten larvae were tested per dose and either kept in the dark or exposed to red light. Light activation was performed with larvae irradiated with a 660 nm red light LED for 1 hr (58.7 J/cm^2^). Embryos were screened at 0, 6, 24, 48 and 72 h post treatment for survival and deformities.

Xenograft of human cells into zebrafish.

Ramos cells (ATCC, CRL-1596) were cultured in RPMI 1640 media supplemented with 10% heat-inactivated fetal bovine serum at 37 °C with 5% CO_2_. GFP-labeled cells were generated using a pLEN-TI-PGK:GFP expression construct as previously described.^59^ Following transduction, the cells were filtered through a sterile 40 μm cell strainer to ensure a single cell suspension, counted using a Countess^®^ Automated Cell Counter system (Invitrogen, C10227), and centrifuged at 1200 rpm (200 xg) for 5 min at room temperature. The cell pellet was resuspended in pre-warmed RPMI 1640 media to a concentration that would deliver 500 cells in a 2 nL injection volume.

Zebrafish larvae at 2-day post-fertilization (dpf) were dechorionated with Pronase (Fisher Scientific, 74–332) at a final concentration of 1 mg/mL in 1× E3 media, immediately prior to xenograft. Transplantation of human cells into zebrafish were performed using non-filament borosilicate glass capillaries (Sutter Instrument Company, B100–50-10). Capillaries were prepared by heating and pulling into needlepoints using a Flaming/Brown micropipette puller (Sutter Instrument, P-87). Needlepoints were cut to a bevel using a sterile razor. Droplet size was measured to 2 nL (~0.15 mm diameter) using a micrometer and kept at a constant volume throughout injection. Two-day post-fertilization (dpf) larvae were anesthetized prior to microinjection using 4 mg/mL Tricaine-S (Pentair Aquatic, NC0342409), with xenograft injections performed using an air-pressure driven system (MPPI-3 Pressure Injector and Microinjector Pipette Holder, ASI). Post-injection, embryos were kept in 1× E3 media at 28 °C for a 1 h recovery time then incubated at 34 °C.

Drug treatment of xenografted zebrafish and counting of GFP positive cells.

Xenografted larvae were screened for consistent cell engraftment before treatment with JQ1, dBET1, and 8 at 1 and 10 μM concentrations in E3 media. Light activation was done as previously described. After 48 h zebrafish were collected, digested with Liberase TL (Roche, 05401119001) at 500 μg/mL, 37 °C for 10 min and blocked with 2% FBS. The suspension was then filtered through a 40 μm cell strainer and centrifuged at 250 xg for 5 min. The cells were resuspended in 100 μL PBS and GFP positive cells were manually counted under a fluorescent microscope.

Supplementary Material

Supporting Information. Detailed compound characterization, chemical structures and synthetic methods, full biological procedures, and Additional Figures.

Supplementary Files

This is a list of supplementary files associated with this preprint. Click to download.

PhotoPROTACSI12292025.pdf