Development of compounds for targeted degradation of mammalian cryptochrome proteins

Abstract

The mammalian cryptochrome proteins (CRY1 and CRY2) are transcriptional repressors most notable for their role in circadian transcriptional feedback. Not all circadian rhythms depend on CRY proteins, however, and the CRY proteins are promiscuous interactors that also regulate many other processes. In cells with chronic CRY deficiency, protein homeostasis is highly perturbed, with a basal increase in cellular stress and activation of key inflammatory signalling pathways. Here, we developed tools to delineate the specific effects of CRY reduction, rather than chronic deficiency, to better understand the direct functions of CRY proteins. Performing a bioluminescence screen and immunoblot validation, we identified compounds that resulted in CRY reduction. Using these compounds, we found that circadian PERIOD2 (PER2) protein rhythms persisted under CRY-depleted conditions. By quantitative mass spectrometry, we found that CRY-depleted cells partially phenocopied the proteomic dysregulation of CRY-deficient cells, but showed minimal circadian phenotypes. We did, however, also observe substantial off-target effects of these compounds on luciferase activity and could not ascertain a specific mechanism of action. This work therefore highlights both the utility and the challenges of targeted protein degradation and bioluminescence reporter approaches in disentangling the contribution of CRY proteins to circadian rhythmicity, homeostasis and innate immune regulation.This article is part of the Theo Murphy meeting issue 'Circadian rhythms in infection and immunity'.

Keywords: PROTAC; cellular physiology; circadian rhythms; cryptochromes; mass spectrometry; protein degradation.

Figure 1.

Development of compounds that result in effective depletion of CRY1 and CRY2. (a) PROTAC concept: a small molecule CRY ligand is conjugated with a ligand for the E3 ligase, CRBN or VHL, via a PEG linker. Upon binding by an effective PROTAC, CRY would be ubiquitinated by CRBN and consequently targeted to the 26S proteasome for degradation. (b) Screening approach: putative compounds were screened using a U2OS reporter cell line expressing a CRY1::LUC fusion protein, generated by knocking firefly luciferase into the endogenous CRY1 locus, followed by bioluminescence imaging in live cells. (c) Chemical structures of the KS15-comprising PROTACs designed to harness the VHL (left) and CRBN (right) E3 ligases. (d) Percentage of vehicle CRY1::LUC bioluminescence (in relative light units, RLU) following treatment with respective putative CRY degrader compounds, C1−10, at 1 μM. E3 ligase ligand used and relative length of PEG linker are indicated. Bars show mean ± s.e.m. of six replicates, and show the median luminescence value of each replicate on the third day (48–72 h) following compound addition, expressed as a percentage of the vehicle (0.01% DMSO) mean. Statistics: Kruskal Wallis with Dunn’s multiple comparisons test (C7: p = 0.0167; C8: p = 0.0006; C9: p = 0.0104); n = 6 replicate populations of cells. (e) Dose–response for dilution series after addition of C8. Data are presented as percentage of mean vehicle (0.1% DMSO) CRY1::LUC bioluminescence signal (RLU) after 48 h treatment. Points show mean ± s.e.m.; n = 6 replicate wells. Solid red line indicates a variable sigmoidal fit; with associated IC₅₀ and R² values also reported. (f) Immunoblots for CRY1 and CRY2, respectively, following 60 h incubation with vehicle (0.01% DMSO), compound C7 or C8, normalized to β-actin. Bars show percentage of mean vehicle intensity (mean ± s.e.m.). Statistics: one-way ANOVA with Holm–Šídák multiple comparisons test (CRY1 adjusted p-values: C7: p = 0.0007; C8: p = 0.0007; CRY2: C7: p = 0.008; C8: p = 0.0049); n = 3 independently treated populations of U2OS PER2::LUC cells.

Figure 2.

Effective CRY degraders have on- and off-target effects and do not ablate circadian rhythms in bioluminescence in human and mouse cells. (a) Immunoblot for CRY2 following 6 h pre-incubation with MLN4924 followed by 48 h incubation with 1 μM C8 and respective controls, normalized to β-actin. Bars show percentage of vehicle (0.03% DMSO final) mean intensity ± s.e.m. Statistics: one-way ANOVA with Holm–Šídák multiple comparisons test (adjusted p-values: MLN4924: p = 0.0008; MLN4924+C8: p = 0.608); n = 3 independently treated populations of U2OS cells. (b) CRY1::LUC in cells treated with 1 μM C8 following 24 h preincubation with 30 μM pomalidomide. Bars show mean ± s.e.m. of 3–5 replicates and show the median luminescence value of each replicate in the 24–48 h following C8 addition expressed as a percentage of the vehicle (0.01% DMSO) mean. Statistics: Kruskal Wallis with Dunn’s multiple comparisons test; n = 3–5 populations of cells. (c-e) Longitudinal bioluminescence recordings and circadian parameters in luciferase reporter cell lines treated with putative CRY degraders and associated compounds. Period estimated by fitting a damped cosine wave to individual replicate data (see methods). Total luminescence is calculated as the sum of luminescence (in relative light units, RLU) over the first 24 h window following treatment. All bar graphs show mean ± s.e.m. Statistics: respective one-way ANOVAs with Holm–Šídák multiple comparisons tests (p-value threshold = 0.05, *p < 0.05; **p < 0.01; ***p < 0.001; *p < 0.0001). (c) PER2::LUC luminescence in U2OS cells following treatment with 1 μM C7, C8 or KS15. Luminescence presented as a percentage of vehicle (0.01% DMSO) mean at its pre-treatment peak. (d) Constitutive SV40:LUC luminescence in U2OS cells following treatment with 1 μM C8 (treatment applied at time 0). Luminescence presented as a percentage of vehicle (0.02% DMSO) mean at its post-treatment peak. (e, f) PER2::LUC luminescence in WT (e) and CKO (f) mouse lung fibroblasts before and following treatment with 1 μM C8. Luminescence presented as a percentage of vehicle (0.03% DMSO) means at respective post-treatment peaks.

Figure 3.

Mass spectrometry reveals CRY degraders replicate CRY knockout proteome dysfunction. (a) Numbers of proteins showing significant abundance increases or decreases in C8-treated WT and vehicle-treated (0.01% DMSO) CKO mouse fibroblasts, both with respect to vehicle-treated WT cells (two-way t‐test, BHQ < 0.05). Significance of overlapping proteins determined by Fisher’s exact tests (four asterisks, BHQ value < 0.0001). (b) Probability density plot showing absolute-value fold change in all 7559 proteins in C8-treated WT cells and CKO cells, both with respect to vehicle-treated WT cells (full list of proteins in electronic supplementary material, table S6). Distributions were compared using a Kolmogorov–Smirnov test; p‐value < 0.0001. (c,d) Bar charts showing selected enriched GO Process terms associated with individual lists of significantly downregulated (c) and upregulated (d) proteins from (a). Full lists of terms given in electronic supplementary material, tables S1,2. Shown are terms that co-occurred in C8-treated WT and vehicle-treated CKO ontology lists (p-value threshold 0.01). Omitted are terms enriched C8-treated CKO relative to vehicle-treated CKO.