Targeted protein degradation of PDE4 shortforms by a novel proteolysis targeting chimera

Abstract

Cyclic AMP (cAMP) has a crucial role in many vital cellular processes and there has been much effort expended in the discovery of inhibitors against the enzyme superfamily that degrades this second messenger, namely phosphodiesterases (PDEs). The journey of competitive PDE inhibitors to the clinic has been hampered by side effects profiles that have resulted from a lack of selectivity for subfamilies and individual isoforms because of high conservation of catalytic site sequences and structures. Here we introduce a proteolysis targeting chimera (PROTAC) that can specifically target a small subset of isoforms from the PDE4 family to send the enzyme for degradation at the proteasome by recruiting a ubiquitin E3 ligase into proximity with the PDE. We constructed our PDE4 PROTAC (KTX207) using a previously characterized PDE4 inhibitor, and we show that evolution of the compound into a PROTAC improves selectivity, potency and enables a long-lasting effect even after the compound is removed from cells after a short treatment duration. Functionally, KTX207 is more effective at increasing cAMP, is 100 times more anti-inflammatory, and is significantly better at reducing the growth in cancer cell models than the PDE4 inhibitor alone. Our study highlights the advantages of targeted degradation versus active-site occupancy for PDE4 inhibition and discusses the potential of this novel pharmacological approach to improve the safety profile of PDE4 inhibition in the future.

Keywords: PDE4; PROTAC; phosphodiesterases; protein degradation.

Fig. 1

Structure of PDE4‐PROTAC‐Cereblon. (A) The basic structure of KTX207 linking cereblon‐based E3 ligase recruiting moiety to PDE4. (B) Molecular dynamics simulations of KTX207‐mediated complexes using Molecular Operating Environment (MOE) software. Molecular degrader, KTX207‐induced interface contacts with PDE4D (left) and cereblon‐based E3 ligase (right). (C) Ternary complex of PDE4D and cereblon induced by KTX207. Blue cartoons represent PDE4D and orange cartoons represent cereblon E3 ligase. Magenta sticks represent KTX207.

Fig. 2

Examination of KTX207‐mediated degradation of PDE4. (A) Immunoblots showing degradation of PDE4D and PDE4B in A549 cells treated with different concentrations of KTX207 for 24 h. GAPDH was used as the loading control. The percentage ratio of (B) PDE4D and (C) PDE4B expression normalized to GAPDH are presented as mean ± SEM (n = 4–6). Values of IC₅₀ have been estimated. (D) cAMP reporter activation was assessed by signal generated from A549 cells stably expressing cAMP‐responsive luciferase construct. Luciferase activity was measured 24 h after initiation of the reaction by adding 1 nm of compound +100 μm d‐Luciferin with or without 3 h treatment of 10 μm forskolin (n = 4). (E) Representative laser scanning confocal micrographs demonstrating the distribution of Pan‐PDE4D (red) and PDE4D1 (green) in A549 cells treated with either 1 nm or 10 pm of compounds for 24 h. Scale bar: 10 μm. (F) Semi‐quantification of the PDE4D1 fluorescent staining compared with the BI control. The results are shown as means ± SEM of several images where at least 4 individual cells were analyzed for each field of view (n = 4). All statistical differences were examined by unpaired Student's t‐test compared with the BI control. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. IB, immunoblotting with the indicated antibody.

Fig. 3

KTX207 degrades PDE4D shortform with high selectivity. Representative immunoblot showing (A) degradation of PDE4D shortform in HEK293 cells overexpressing PDE4D2, (B) degradation of PDE4D longform in HEK293 cells overexpressing PDE4D5, (C) degradation of PDE4D longform in HEK293 cells overexpressing Quad‐PDE4D5 (K48R:K53R:K78R:K140R) mutant, (D) degradation of PDE4 dead‐short form in HEK293 cells overexpressing PDE4A7 following 24 h treatment with indicated concentrations of KTX207. The percentage ratio of PDE4D expression normalized to GAPDH are presented as mean ± SEM (n = 3). All statistical differences were examined by unpaired Student's t‐test compared with the BI control. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. IB, immunoblotting with the indicated antibody.

Fig. 4

Efficacy of KTX207‐mediated degradation of PDE4D shortform. (A) Proteasome‐dependent degradation of PDE4D shortform by KTX207. A549 cells were treated with compounds with or without 100 nm of proteasome inhibitor, bortezomib (BTZ) which caused accumulation of poly‐ubiquitinated proteins in A549 cells. GAPDH was used as the loading control. (B) PDE4D protein levels normalized to GAPDH (n = 4). (C) Onset of action of KTX207‐mediated degradation of PDE4. Immunoblot analysis of (D) PDE4D and (E) PDE4B expression after treatment with 1 nm of KTX207 at indicated timepoints. PDE4D protein levels normalized to GAPDH (n = 3). (F) Long‐lasting effect of KTX207. A549 cells were treated with 1 nm of KTX207 for 24 h. Then, the medium was removed and fresh medium without compounds was added. Protein lysates were collected after indicated time and analyzed by western blotting. (G) The percentage ratio of PDE4D expression normalized to GAPDH are presented as mean ± SEM (n = 4). All statistical differences were examined by unpaired Student's t‐test compared with the BI control. *P < 0.05, **P < 0.01, ***P < 0.001. IB, immunoblotting with the indicated antibody.

Fig. 5

Cytosolic FRET response of A549 cells expressing Epac1‐camps FRET sensor. (A) Representative FRET curve, (B) Log(agonist) vs. response line chart, and (C) bar chart normalized to isoprenaline (0.1, 0.5, 1, 5, and 10 μm) stimulation followed by saturator (10 μm forskolin and 100 μm IBMX). Data expressed as mean ± SEM; n = 11 cells; *P < 0.01, **P < 0.01, ***P < 0.001, ****P < 0.0001 based on one‐way ANOVA with Tukey's multiple comparisons test. (D) Representative FRET curve, (E) bar chart of FRET ratio shift, and (F) bar chart of normalized FRET response to isoprenaline stimulation of 1 nm BI (cyan), 10 nm BI (violet), 1 nm KTX207 (green), 10 nm KTX207 (purple), and DMSO‐treated control (red). Data expressed as mean ± SEM; n = 59 cells from three independent experiments; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 based on one‐way ANOVA with Holm‐Šídák's multiple comparisons test.

Fig. 6

Functional readout for KTX207‐mediated PDE4 degradation. Representative immunoblots showing (A) PDE4D and (B) PDE4B expressions in T cells and PHA‐activated T cells (left panel). Quantification of PDE4 expression normalized to GAPDH and present as mean values ± SEM (n = 4) (right panel). (C) LPS‐induced production of TNF‐α by PBMC with or without removal of KTX207. Data are presented as mean ± SEM (n = 3). All statistical differences were examined by unpaired Student's t‐test compared with the BI control. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. IB, immunoblotting with the indicated antibody.

Fig. 7

KTX207 suppresses cancer cell growth in A549 cells. (A) Real‐time cell proliferation of A549 cells monitoring by xCELLigence system (left panel). Continuous impedance measurement displayed as normalized cell index. Data were normalized to the time of compound addition (T = 0 h). The rate of A549 cell proliferation was determined by measuring the slope of the line between the 25‐ and 65‐h interval (right panel). (B) xCELLigence real‐time cell analysis of HEK293 cells (left panel). The rate of HEK293 cell proliferation was determined by measuring the slope of the line between the 10‐ and 35‐h interval (right panel). (C) Representative immunoblots showing phospho‐PKA substrate, phospho‐ERK, total ERK, PARP1, and Ki67 in A549 cells after 24 h of treatment as indicated. GAPDH was the loading control. The percentage ratio of (D) phospho‐PKA substrate, (E) phospho‐ERK, (F) PARP1, and (G) Ki67 expressions normalized to GAPDH and are presented as mean ± SEM (n = 4). All statistical differences were examined by unpaired Student's t‐test compared with the BI control. ***P < 0.001, ****P < 0.0001. IB, immunoblotting with the indicated antibody.

Fig. 8

Growth kinetics of A549‐derived tumor spheroids. (A) Immunoblot showing degradation of PDE4D in A549 spheroids treated with different concentrations of KTX207 for 24 h. GAPDH was used as the loading control. The percentage ratio of (B) PDE4D longforms and (C) PDE4D shortform expressions normalized to GAPDH are presented as mean ± SEM (n = 4). (D) Representative brightfield images showing morphology of A549 cell line cultured as 3D spheroids in ultra‐low attachment 96‐well plates. Scale bar: 200 μm. The images were captured by a phase‐contrast microscopy at day 0, 3, 7, and 10 in culture treated with 100 pm of compounds (n = 5). (E) Time‐course monitoring of spheroid growth expressed in perimeter (microns) and (F) the roundness of spheroids evaluated over a period of 10 days (n = 5). All statistical differences were examined by unpaired Student's t‐test compared with the BI control. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. IB, immunoblotting with the indicated antibody.