Age-Related Increases in PDE11A4 Protein Expression Trigger Liquid-Liquid Phase Separation (LLPS) of the Enzyme That Can Be Reversed by PDE11A4 Small Molecule Inhibitors

Abstract

PDE11A is a little-studied phosphodiesterase sub-family that breaks down cAMP/cGMP, with the PDE11A4 isoform enriched in the memory-related hippocampal formation. Age-related increases in PDE11A expression occur in human and rodent hippocampus and cause age-related cognitive decline of social memories. Interestingly, age-related increases in PDE11A4 protein ectopically accumulate in spherical clusters that group together in the brain to form linear filamentous patterns termed "PDE11A4 ghost axons". The biophysical/physiochemical mechanisms underlying this age-related clustering are not known. Here, we determine if age-related clustering of PDE11A4 reflects liquid-liquid phase separation (LLPS; biomolecular condensation), and if PDE11A inhibitors can reverse this LLPS. We show human and mouse PDE11A4 exhibit several LLPS-promoting sequence features, including intrinsically disordered regions, non-covalent pi-pi interactions, and prion-like domains that were particularly enriched in the N-terminal regulatory region. Further, multiple bioinformatic tools predict PDE11A4 undergoes LLPS. Consistent with these predictions, aging-like PDE11A4 clusters in HT22 hippocampal neuronal cells were membraneless spherical droplets that progressively fuse over time in a concentration-dependent manner. Deletion of the N-terminal intrinsically disordered region prevented PDE11A4 LLPS despite equal protein expression between WT and mutant constructs. 1,6-hexanediol, along with tadalafil and BC11-38 that inhibit PDE11A4, reversed PDE11A4 LLPS in HT22 hippocampal neuronal cells. Interestingly, PDE11A4 inhibitors reverse PDE11A4 LLPS independently of increasing cAMP/cGMP levels via catalytic inhibition. Importantly, orally dosed tadalafil reduced PDE11A4 ghost axons in old mouse ventral hippocampus by 50%. Thus, PDE11A4 exhibits the four defining criteria of LLPS, and PDE11A inhibitors reverse this age-related phenotype both in vitro and in vivo.

Keywords: PDE11A; aging; biomolecular condensation; cognitive impairment; hippocampus; inclusion body; memory; phosphodiesterase; tadalafil.

Figure 1

Multiple computational analyses identify LLPS-promoting sequence features within PDE11A4. Graphical outputs from D2P2 [35] (IDRs and ANCHORS) and PhaSePred [41] (all other outputs) for (A) human PDE11A4 (hPDE11A4) and (B) mouse PDE11A4 (mPDE11A4). PhaSePred reported analyses from other primary computational tools as follows: pi–pi from Pscore [38], prion-like domain from PLAAC [39], low-complexity regions from SEG [40], and LLPS propensity from catGRANULE [43]. GAF—mammalian cGMP-regulated PDEs, Anabaena adenylyl cyclases, and the Escherichia coli transcription factor FhlA; IDRs—intrinsically disordered regions.

Figure 3

The dispersing effects of PDE11A4 inhibitors do not diminish with a 24 h treatment and the effects can be reversible upon washout. (A) A graphical outline of the experimental time course. (B) Scaffolds for PDE11is tadalafil and BC11-38. (C) Exemplar image of aging-like LLPS of EmGFP-mPDE11A4 following vehicle treatment (left) and dispersal following a 1 h treatment with 100 µM BC11-38. The extent of mPDE11A4 LLPS measured following a 1 h treatment, a 1 h treatment followed by a 5 h washout period, a 24 h treatment or a 24 h treatment followed by a 5 h washout period for (D–G) tadalafil and (H–K) BC11-38 (n = 4 biological replicates/group; each data set is representative of 2+ independent experiments). The extent of mPDE11A4 LLPS measured following a 1 h treatment of (L) the PDE4 inhibitor rolipram or (M) the PDE10 inhibitor papaverine (10 µM tadalafil was included as a positive control; data is representative of 2 independent experiments). (N) Representative Western blot (top: EmGFP-PDE11A4; bottom: Ponceau Stain (PS)) and (O) quantification showing aging-like accumulation of mPDE11A4 in the membrane fraction in vehicle-treated (Veh, n = 3 biological replicates) cells that is reversed by a 1 h treatment with BC11-38 (BC, n = 4) and tadalafil (Tad, n = 4), thereby eliciting an “anti-aging” effect by increasing the mPDE11A4 ratio in the cytosolic fraction (cyto) versus the membrane fraction (memb). The ability of PDE11Ais to reverse PDE11A4 LLPS appears to be independent of their ability to inhibit PDE11A4 catalytic activity because neither (P) cAMP nor (Q) cGMP analogs mimic the effect of PDE11Ais and neither (R) cAMP nor (S) cGMP antagonists block the effect of PDE11Ais on LLPS. * vs. 0 µM/vehicle p < 0.05–0.0001. Brightness, contrast, and histogram stretch of images are adjusted for graphical clarity.

Figure 2

PDE11A4 demonstrates physical features of LLPS in that its punctate accumulation starts as small membraneless spherical droplets that progressively fuse over time in a reversible and concentration-dependent nature. (A) Membrane staining using BODIPY-568 in HT22 cells transfected with EmGFP-mPDE11A4 shows no colocalization of the signals (representative image of 16 cells across four coverslips, see Results text for quantification). (B) Qualitative time-lapsed imaging of EmGFP-mPDE11A4 in 4 HT22 cells reveal GFP-mPDE11A4 droplets form and fuse over time. Images shown capture fusion events of droplets, numbered in the order of appearance,that occured over 105 minutes. (C) Seven COS1 cells across five experiments confirm GFP-mPDE11A4 droplets form and fuse over time. Images shown capture fusion events that occurred over 120 minutes. (D) Qualitative experiments in which HT22 cells formed GFP-mPDE11A4 droplets (indicated by arrow heads) and were then imaged over time immediately prior to and following treatment with 5% 1,6-hexanediol show droplets disperse within minutes of the treatment. (E) Quantitative study in which HT22 cells formed GFP-mPDE11A4 droplets and then were treated with either vehicle or 5% 1,6-hexanediol confirms that GFP-mPDE11A4 droplets are reversed by 5% 1,6-hexanediol and not vehicle treatment (n = 4 biological replicates/group). (F) Quantification and representative images of HT22 cells expressing low, moderate, or high levels of GFP-mPDE11A4 protein show mPDE11A4 droplets form in a concentration-dependent manner (n = 4 biological replicates/group). Note, GFP alone shows no punctate clustering despite high expression throughout the cell, particularly the nucleus. (G) Deletion of the N-terminal intrinsically disordered region (xN-IDR) prevents PDE11A4 spherical droplets from forming (n = 5 biological replicates/group; representative of 3 experiments). * vs. veh/low/WT, p = 0.0146 to <0.0001, # vs. mod, p = 0.0158. oof—out of focus; mod—moderate, GFP—emerald-green fluorescent protein. Brightness/contrast/histogram stretch and/or sharpness adjusted for graphical clarity of images.

Figure 4

Tadalafil reverses age-related clustering of endogenous mPDE11A4 in the ventral hippocampus of old female and male mice. (A) Quantification of endogenous mPDE11A4 ghost axons in ventral subiculum that were selectively labeled using antibodies that recognize PDE11A4-pS1117/pS124 [9] (n = 2/sex/dose). Exemplar images of mPDE11A4 immunofluorescence in an (B) old vehicle-treated and (C) old tadalafil-treated mouse. (D) Upon higher magnification (i.e., of the area in panel A outlined by the white box), many mPDE11A4 ghost axons reveal themselves to be a linear trail of spherical dots as opposed to a continuous filament, consistent with LLPS. * Post hoc vs. 0 mg/kg tadalafil (i.e., vehicle), p = 0.045. Brightness, contrast and histogram stretch of images adjusted for graphical clarity.