Biologic that disrupts PDE11A4 homodimerization in hippocampus CA1 reverses age-related cognitive decline of social memories in mice

Abstract

Age-related proteinopathies in phosphodiesterase 11A (PDE11A), which degrades 3’,5’-cAMP/cGMP and is enriched in the ventral hippocampus (VHIPP), drives age-related cognitive decline (ARCD) of social memories. Age-related PDE11A4 ectopically accumulates within the membrane compartment and in filamentous structures termed ghost axons. Previous studies show that expressing an isolated PDE11A4-GAF-B binding domain disrupts homodimerization and reverses aging-like PDE11A4 accumulations in vitro. Here, we show that in vivo lentiviral expression of the isolated PDE11A4-GAFB domain in hippocampal CA1 of aged mice reverses age-related PDE11A4 accumulations and ARCD of social transmission of food preference memory (STFP). It also improves 7-day remote long-term memory for social odor recognition without affecting non-social odor recognition. In vitro studies show that disrupting homodimerization does not alter the catalytic activity of PDE11A4 but may reverse age-related decreases in cGMP by relocating PDE11A4 from a cGMP-rich to a cAMP-rich pool independently of other intramolecular relocation signals (PDE11A4-pS162). Altogether, these data suggest that a biologic designed to disrupt PDE11A4 homodimerization may hold therapeutic potential for age-related PDE11A4 proteinopathies.

Graphical Abstract

1. INTRODUCTION

Decreases in 3’,5’-cyclic nucleotide signaling in the aged and demented hippocampus contribute to cognitive decline (Bonkale, Cowburn et al. 1999, Zhang, Cheng et al. 2014, Kelly 2018). These decreases in signaling are driven in part by altered function of 3’,5’-cyclic nucleotide phosphodiesterases (PDEs), which are the only known enzymes to break down 3’,5’-cyclic adenosine monophosphate (cAMP) and 3’,5’-cyclic guanosine monophosphate (cGMP) (Kelly 2018). Interestingly, expression patterns of the eleven PDE families differs across brain regions as does their enrichment among different subcellular compartments (Kelly, Adamowicz et al. 2014, Kelly 2018, Patel, Klett et al. 2018, Baillie, Tejeda et al. 2019). For example, some PDEs are expressed more in the cytosol than the membrane (e.g., PDE11A), while others are more highly expressed in the membrane versus cytosol (including PDE2A, PDE9A and PDE10A (Kelly 2018, Patel, Klett et al. 2018)). Due to the fact that PDEs are localized to specific subcellular domains, they are able to regulate individual pools or nanodomains of cyclic nucleotide signaling (Baillie, Tejeda et al. 2019). Such subcellular compartmentalization of cyclic nucleotide signaling allows a single cell to respond specifically to simultaneous intra- and/or extracellular signals. Therefore, the subcellular localization of any PDE is equally important as its catalytic activity when considering its function. During aging, the subcellular compartmentalization of select PDEs is known to change in the brain (Kelly, Adamowicz et al. 2014, Patel, Klett et al. 2018, Pilarzyk, Porcher et al. 2022), which likely compromises the physiological segregation of these signals within the cyclic nucleotide signaling cascades (Houslay 2010, Kokkonen and Kass 2017). Interestingly, age-related and neuropsychiatric diseases are accompanied by a loss of cyclic nucleotide signaling in one subcellular compartment but not another (Bonkale, Winblad et al. 1995, Rahman, Li et al. 1997, Bonkale, Cowburn et al. 1999, Fields, Li et al. 1999, Chang, Li et al. 2003, Kelly 2018), suggesting therapeutic strategies should optimally target enzymes in a compartment-specific manner.

PDE11A has garnered particular interest in the context of age-related cognitive decline and Alzheimer’s disease (Kelly, Adamowicz et al. 2014, Pilarzyk, Farmer et al. 2021, Qin, Zhou et al. 2021, Pilarzyk, Porcher et al. 2022). PDE11A is encoded by a single gene and has four isoforms, with only PDE11A4 being expressed in brain [9–12, 14](Kelly, Logue et al. 2010, Kelly 2015, Hegde, Capell et al. 2016). The regulatory N-terminus of PDE11A4 is longer than the other isoforms, uniquely containing two full GAF domains (i.e, cGMP binding PDE, Anabaena adenylyl cyclase and E. coli FhlA domains)(Kelly 2018). The GAF-A domain binds cGMP as a potential allosteric regulatory site, and the GAF-B domain regulates protein-protein interactions, including homodimerization (Weeks, Zoraghi et al. 2005, Gross-Langenhoff, Stenzl et al. 2008, Pathak, Agostino et al. 2017). PDE11A4 is the only PDE in the brain whose expression stems solely from the extended hippocampal formation, a brain region that is critical to learning and memory and vulnerable to age-related deficits in cyclic nucleotide signaling (Kelly, Adamowicz et al. 2014, Hegde, Capell et al. 2016, Kelly 2018). PDE11A4 expression in the hippocampus increases across the lifespan in mice, rats and humans (Kelly, Adamowicz et al. 2014, Hegde, Capell et al. 2016, Pilarzyk, Porcher et al. 2022). These conserved age-related increases in PDE11A4 protein expression are deleterious as 1) Pde11a KO mice are protected against age-related cognitive decline of distant/remote long-term social memories 7 days after training (Pilarzyk, Porcher et al. 2022) and 2) mimicking age-related overexpression of PDE11A4 in the CA1 field of hippocampus of either young or old Pde11a KO mice is sufficient to trigger decline of these social long-term memories (LTM) (Pilarzyk, Klett et al. 2019, Pilarzyk, Porcher et al. 2022).

While PDE11A4 protein expression in the young adult hippocampus is significantly higher in the cytosolic versus membrane compartment, age-related increases in PDE11A4 are found specifically in the membrane compartment of the ventral hippocampal formation (Pilarzyk, Porcher et al. 2022). Further, aging triggers a punctate accumulation of PDE11A4 protein in filamentous structures termed ghost axons (i.e., in homage to tau ghost tangles; (Pilarzyk, Porcher et al. 2022)) that are rarely seen in the young C57BL/6J mouse brain. Therefore, these age-related increases in membrane-associated PDE11A4 and PDE11A4 ghost axons may be considered ectopic (Pilarzyk, Porcher et al. 2022). Interestingly, disrupting PDE11A4 homodimerization in vitro by expressing an isolated GAF-B domain that acts as a negative sink for monomers disperses the punctate accumulation of PDE11A4 and selectively reduces expression of membrane-associated PDE11A4 (Pathak, Agostino et al. 2017). From a drug discovery perspective, GAF domains are highly interesting because they are found in no other mammalian proteins other than PDE2, PDE5, PDE6, PDE10, and PDE11 (Francis, Blount et al. 2011). Further, the amino acid sequences amongst the GAF domains of these 5 PDE families are not well conserved; thus, it is not possible for the GAF domain from one PDE family to bind that of another (Francis, Blount et al. 2011). Thus, a biologic targeting the PDE11A4 GAF-B domain would be highly specific and may represent a therapeutic option capable of reducing age-related PDE11A4 proteinopathies in a compartment-specific manner. Therefore, we seek to determine if a biologic that decreases PDE11A4 homodimerization will be sufficient to reverse age-related accumulations of PDE11A4 in ghost axons and rescue age-related cognitive decline of social memories.

2. MATERIALS AND METHODS (see Supplement for Table of Key Resources)

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken polyclonal anti-PDE11A	Aves [9]	Custom
Rabbit polyclonal anti-PDE11A	Fabgennix	Cat#: PD11A-112; RRID: N/A
Rabbit polyclonal anti-GFP	Santa Cruz	Cat#: sc8334; RRID:AB_641123
Rabbit polyclonal anti-Creb	Cell Signaling	Cat#: 4820 ; RRID:AB_1903940
Rabbit polyclonal anti-Actin	Sigma- Aldrich	Cat#: A2066; RRID:AB_476693
Rat polyclonal anti-mCherry	ThermoFisher	Cat#: M11217; RRID:AB_2536611
Rabbit polyclonal anti-mCherry	Invitrogen	Cat#: PA534974; RRID:AB_2552323
Mouse polyclonal anti-mCherry	PhosphoSolutions	Cat#: 1203-mCherry; RRID:AB_2715492
Anti-Mouse FabFragments	Jackson Immunoresearch	Cat#: 715-007-003; RRID:AB_2307338
Rabbit polyclonal anti-PDE11A	Fabgennix	Cat #PPD11A-140AP; RRID: N/A
Rabbit polyclonal anti-PDE11A	Fabgennix	Cat #PPD11A-150AP; RRID: N/A
Alexafluor 488 AffiniPure Donkey Anti-Chicken	Jackson Immunoresearch	Cat#: 703-545-155; RRID:AB_2340375
Alexafluor 488 AffiniPure Donkey Anti-Rabbit	Jackson Immunoresearch	Cat#: 711-545-152; RRID:AB_2313584
Alexafluor 594 AffiniPure Donkey Anti-Chicken	Jackson Immunoresearch	Cat#: 703-585-155; RRID:AB_2340377
Alexafluor 594 AffiniPure Donkey Anti-Rabbit	Jackson Immunoresearch	Cat#: 711-585-152; RRID:AB_2340621
Alexafluor 594 AffiniPure Donkey Anti-Mouse	Jackson Immunoresearch	Cat#: 715-585-150; RRID:AB_2340854
Bacterial and Virus Strains
Lentivirus: SPW-mCherry	Viral Vector Core, University of South Carolina [90]	Custom
Lentivirus: SPW-mCherry-tagged PDE11A4-GAF-B domain	Viral Vector Core, University of South Carolina [90]	Custom
Plasmid: SPW-mCherry	Genscript	Custom
Plasmid: SPW-eGFP	Genscript	Custom
Chemicals, Peptides, and Recombinant Proteins
Supersignal West Pico Chemiluminescent Substrate	Thermoscientific;	Cat# 34078
Westernsure Premium Chemiluminescent Substrate	Licor	Cat# #926-95000
Dulbecco’s Modification of Eagle’s Medium (DMEM)	Fisher Scientific	Cat# MT15017CV
Penicillin/Streptomycin (P/S)	Mediatech, a Corning subsidary	Cat# MT30001CI
TrypLE Express	GIBCO	Cat# 12605010
Fetal bovine serum (FBS)	Atlantic Biologicals/R&D System	Cat#: S11150
Optimem	GIBCO	Cat# 51985091
Lipofectamine 2000	Invitrogen	Cat# 11668030
[3H]cGMP	Perkin Elmer	Cat# NET337
[3H]cAMP	Perkin Elmer	Cat# NET275
Snake venom	Crotalus atrox	Cat#: V-7000
DAPI fluoromount	Southern Biotech	Cat#: 0100-20
Thermo Pierce Scientific phosphatase tablet	Fisher Scientific	Cat #A32959
Thermo Pierce Scientific protease inhibitor 3	Fisher Scientific	Cat #P0044
Critical Commercial Assays
Cayman Chemical cAMP ELISA kit	Cayman Chemical Company	Cat# 581001
Cayman Chemical cGMP ELISA kit	Cayman Chemical Company	Cat# 581021
DC Protein Assay kit	Bio-Rad	Cat# 5000112
Experimental Models: Cell Lines
Cell Line: HEK293T	ATCC [90]	Cat#: CRL-11268
Cell Line: COS-1	ATCC [90]	Cat#: CRL-1650
Cell Line: HT-22	ATCC [90]	Discontinued
Experimental Models: Organisms/Strains
Mouse: Pde11a−/− (C57BL/6J-C57BL/6N-129S6/SvEvTac-Pde11a)	Deltagen [9, 10, 89, 90]	N/A
Mouse: C57BL/6J	The Jackson Laboratory/ In-house	Cat# 000664
Oligonucleotides
antisense probe: Pde11a: CCACCAGTTCCTGTTTTCCTTTTCGCATCAAGTAATC	IDT [10,69,91]	Custom
Software and Algorithms
ImageJ	National Institute of Health (NIH)	https://imagej.nih.gov/ij/
Sigmaplot 11.1	Systat Software	N/A
Statistica 9.0	Tibco	N/A
R-Project	R Foundation for Statistical Computing	https://www.r-project.org/
Brainspan database	Allen Institute for Brain Science	https://www.brainspan.org/rnaseq/search/index.html.
Neurolucida imaging software	MBF Bioscience	N/A
Other
Invitrogen NuPAGE Novex 4–12% Bis-Tris	LifeTechnologies	Cat# NP0322BOX
Superblock (PBS) Blocking Buffer	ThermoFisher	Cat# 37515
Nitrocellulose Blotting Membranes	GE Helathcare Life Sciences	Cat# 10600008
Neurostar Stereotax/Drill and Injection Robot	NeuroStar	https://robot-stereotaxic.com/drill-injection-robot/
Hamilton Syringe	Hamilton	Cat #7804-04 (26s gauge, 1” length, 25 degree bevel)
1” Wooden Beads	Woodworks Ltd.	Cat# RB1000
2-oz glass jar	Fisher Scientific	Cat# 02-911-773

2.1. Subjects.

C57BL6/J mice were originally obtained from Jackson Laboratory (Bar Harbor, ME) and the line was maintained at the University of South Carolina School of Medicine and then the University of Maryland School of Medicine. As previously described (Kelly, Logue et al. 2010, Pilarzyk, Klett et al. 2019, Smith, Farmer et al. 2021), the Pde11a mouse line obtained from Deltagen (San Mateo, CA) was maintained on a mixed C57BL6 background (99.8% multiple C57BL/6 substrains, 0.2% 129P2/OlaHsd) (Smith, Farmer et al. 2021). Pde11a mice were bred at the University of South Carolina School of Medicine or the University of Maryland School of Medicine in heterozygous (HT) × HT trio-matings. Multiple same-sex wild-type (WT) mice along with one HT mouse (used as the demonstrator in social transmission of food preference—see more below) were weaned and caged together. Typically, cages included ~4–5 mice/cage. We do not believe litter effects are driving findings here due to each cohort being composed of WT mice from multiple litters and, in the case of NSOR and SOR, the datasets are composed of multiple cohorts born and tested at different times (Porcher, Bruckmeier et al. 2021, Smith, Farmer et al. 2021). While both males and females were used in experiments, we are underpowered to analyze for sex effects (see figure legends for specific n’s/sex/group/experiment). In these studies, young mice were defined as 2–6 months and old mice were defined as 18–24 months. “Young” mice included both young Pde11a WT mice surgerized alongside old Pde11a WT mice receiving bilateral injections of the negative control mCherry lentivirus to the dorsal and ventral hippocampi as well as unsurgerized young C57BL6/J mice that were used as an internal control for the assays. Since no obvious differences were found between groups of young surgerized Pde11a WT mice and young unsurgerized C57BL6/J mice, the data from these 2 subgroups were subsequently combined into a singular “young” group. All mice used in experiments were generally healthy throughout the duration of testing. As previously described (Porcher, Bruckmeier et al. 2021), we do not conduct gross pathology but mice are routinely assessed by husbandry, veterinary, and laboratory staff. Mice with lethargy, altered gait, signs of malnutrition or dehydration, tumors >1 cm, are removed from study and euthanized. We aim to study the effects of healthy aging (Porcher, Bruckmeier et al. 2021), therefore if upon brain dissection we had found evidence of an anatomical abnormality (e.g., a pituitary tumor), the animals would have been excluded from the study (note: no occurrences in this study). A 12:12 light:dark cycle and ad lib access to food and water were provided. Experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Pub 85-23, revised 1996) and were fully approved by the Institutional Animal Care and Use Committee of the University of South Carolina and the University of Maryland, Baltimore.

2.2. Plasmid generation.

Plasmids were generated as previously described (Pathak, Agostino et al. 2017, Pilarzyk, Porcher et al. 2022). Briefly, constructs were generated by Genscript (Piscataway, NJ) that expressed either mCherry alone, an mCherry-tagged PDE11A4-GAF-B domain (aa388–558 of NP_001074502.1 that included aa 402–558 of the GAF-B domain along with the 14 upstream amino acids as a spacer), EmGFP alone containing an A206Y mutation to prevent EmGFP dimerization (Jackson, Howell et al. 1990), or an EmGFP-tagged mouse Pde11a4 (NM_001081033) with the BamHI and XhoI recognition sequences added to the N-terminal and C-terminal respectively (see supplementary material for full sequences of these inserts). These constructs were initially generated on a pUC57 backbone and then subcloned into a pcDNA3.1+ mammalian expression vector using the BamHI and XhoI sites (Life Technologies; Walthan, MA). The QuickChange procedure/products were used to generate Pde11a4 mutations as per manufacturer’s instructions (Agilent Technologies; Santa Clara, CA). Oligonucleotide primers were generated by Integrated DNA Technologies (Coralville, IA) and mutated DNA sequences were verified by Functional Biosciences (Madison, WI).

2.3. Cell culture and transfections.

COS-1 (male line), HEK293T (female line), and HT-22 (sex undefined) cell culture and transfection were performed as previously described (Pathak, Agostino et al. 2017). While kept in t-75 flasks, cells were grown in Dulbecco’s Modified Eagle Medium (DMEM) with (HT-22) or without (COS-1, HEK293T) sodium pyruvate (GIBCO; Gaithersburg, MD; or Corning, Manassas, VA,), 10% fetal bovine serum (FBS), and 1% Penicillin/Streptomycin (P/S) (GE Healthcare Life Sciences; Logan, UT, USA). Cells were incubated at 37°C/5% CO2 and passaged using TrypLE Express (GIBCO; Gaithersburg, MD, USA) as a dissociation agent once 65–90% confluent. One day before transfection began, cells were plated along with DMEM+FBS+P/S in either 24-well plates for imaging or 100 mm dishes for biochemistry. The day of transfection, Optimem (GIBCO) replaced the media. According to manufacturer protocol (ratio of 0.375 ug DNA plus 1uL lipofectamine per 1 mL of media), cells were transfected with Lipofectamine 2000 (Invitrogen; Carlsbad, CA). ~19 hours post-transfection, the Optimem/Lipofectamine solution was removed and replaced with DMEM+FBS+P/S. Cells then grew for five hours before sample processing, which meant they were harvested about 24 hours following transfection. Over the course of experiments, cells were sporadically tested for mycoplasma, with negative results always obtained. For assessment of subcellular trafficking, paraformaldehyde (4%) in PBS was used to fix the cells for fifteen minutes, after which they were kept in PBS until imaging. Images were captured using NIS-Elements BR-2.3 (Nikon,Tokyo Japan) on a CoolSNAP EZ CCD camera (Photometrics, Tuscon AZ) mounted on an inverted Leica (Wetzlar DE) DMIL microscope with a Fluotar 10X/0.3 ∞/ 1.2 objective.

All images pertaining to an experiment were quantified by an experimenter blind to treatment using the same computer within the same position in the room, the same lighting conditions, and the same percent zoom. Images were loaded onto a gridded template to facilitate keeping track of count locations within the image, and an experimenter scored each image box by box, with cells along the top and left edges of the entire image not included to follow stereological best practices. Images were quantified in a counterbalanced manner such that 1 picture from each condition was evaluated before moving onto a 2^nd image from that condition. The experimenter classified cells as exhibiting either cytosolic-only labeling or punctate labeling (with or without cytosolic labeling present), with data expressed as the % of the total number of labeled cells that exhibited punctate labeling.

2.4. Stereotaxic Surgeries.

Stereotaxic surgeries and injections were performed, as previously described (Pilarzyk, Klett et al. 2019, Smith, Farmer et al. 2021), using a NeuroStar motorized stereotaxic, drill, and injection robot (Tubingen, Germany). Mice were anesthetized with a steady flow of oxygen and isoflurane. The mice were induced at 3% isoflurane and maintained at 1–1.5%. Lack of reflexes was verified and the scalp was then shaved and cleaned with betadine. A small incision was made in the scalp and the skull was cleared with sterile saline. Cotton swabs were again used to visualize the skull and locate Bregma. Using the robotic drill, small holes were made above the dorsal and ventral hippocampi as per the following coordinates relative to Bregma: dCA1 AP, −1.7, dCA1 ML, +/− 1.6, vCA1 AP, −3.5, vCA1 ML, +/−3.0. Both dorsal and ventral CA1 were targeted even though we previously reported that manipulation of vCA1 PDE11A4 was sufficient to modulate social memories, because we also wished to study the effect of the biologic on social preferences, which are regulated by PDE11A4 in both dorsal and ventral CA1. The mice were tested for social preferences after SOR, then NSOR, then STFP memory tests, and those results are reported elsewhere (Smith, Farmer et al. 2021). At a speed of 10 mm/sec, a Hamilton syringe (custom needle #7804–04: 26s gauge, 1” length, 25 degree bevel) was then then placed to the following depths relative to Bregma: dCA1 DV, 1.3, vCA1 DV, 4.4. After a thirty second pause following the needle movement, the injection robot was used to inject 2 μl of lentivirus at 0.167 μl/minute. Following injection completion, the experimenter waited two minutes to allow the lentivirus to diffuse away from the needle and the needle was raised at the same speed. After all injections were complete, pronged tweezers were used to close the scalp and secured using GLUtures topical adhesive. Buprenorphine in sterile saline at a dose of 0.1mg/kg was injected IP for pain management. For recovery, the mouse was placed on a warm Deltaphase pad and allowed to recover until moving normally and posturing upright. Mice were returned to their home cages and allowed at least 2 weeks of recovery prior to behavioral testing.

As previously described (Pathak, Agostino et al. 2017, Smith, Farmer et al. 2021), a lentivirus carrying an mCherry-tagged PDE11A4-GAF-B domain served to disrupt PDE11A4 homodimerization, while an mCherry-only virus was used as a negative control. A lentiviral construct was used here in order to compare to previous studies examining the effects of overexpressing PDE11A4 in vivo (Pilarzyk, Klett et al. 2019, Pilarzyk, Porcher et al. 2022), which required the use of a lentiviral cassette due to the large size of PDE11A4. The Viral Vector Core of the University of South Carolina generated the lentiviruses. The viruses were made on an “SPW” backbone that drives expression using the phosphoglycerate kinase 1 (PGK) promoter. Though the PGK is a ubiquitous promoter, this backbone is preferentially taken up by neurons (Pilarzyk, Klett et al. 2019). We previously showed that the isolated GAF-B construct disrupts PDE11A4 homodimerization by binding to PDE11A4 monomers and acting as a negative sink that prevents monomers from binding to each other (Pathak, Agostino et al. 2017). For reasons that are not well understood, preventing homodimerization via this method triggers proteolytic degradation of membrane-associated PDE11A4 (Pathak, Agostino et al. 2017). The lentiviruses were prepared and diluted in 0.2M sucrose/42 mM NaCl/0.84 mM KCl/2.5 mM Na2HPO4/0.46 mM KH2PO4/0.35 mM EDTA and the original titers were as follows: mCherry, 7.37X10E10/mL; GAF-B, 1.82X10E10/ml. Pilot studies using wide-field fluorescent microscopy determined diluting the mCherry-only virus to one-third the original concentration yielded comparable mCherry expression between viruses, and so this concentration was used in experiments. The accuracy of the injection and expression of the viral construct was verified by direct visualization of raw florescence and/or following amplification of the signal via immunofluorescence. Images of the dorsal and ventral hippocampus were captured using the Leica Application Suite (LASX) software in conjunction with a Leica DM5000 B florescent microscope and scored with regard to the presence or absence of red fluorescence in CA1, CA3, DG and subiculum subfields by an experimenter blind to the treatment. We detected mCherry expression in ventral CA1 in one or both hemispheres of all but 2 mice. 1 of these mice was in the Young mCherry-only group and was kept in the study given we saw no separation between mCherry infused young WTs and young unsurgerized C57BL/6J mice. The second was in the GAF-B group and so was dropped from the study. A subset of mice exhibited additional viral expression in dentate gyrus, CA2, and/or CA3. The percent of subjects showing labeling in a given subregion, calculated as (# of subjects with expression/total # of subjects)*100, are shown in Figure 1G.We do not believe this ectopic viral expression confounds our results given that 1) it is found in both the mCherry-only and mCherry-GAF-B groups and 2) PDE11A4 expression does not emanate from the dentate, CA2 nor CA3. Of note, we found that cells in proximal CA1 (closer to CA3) relative to distal CA1 (closer to subiculum) more readily took up the virus as previously noted (Pilarzyk, Klett et al. 2019), which mirrors the preferential distribution pattern of endogenous PDE11A4 across CA1 (Hegde, Capell et al. 2016). We found no gross cellular toxicity or morphological damage with either virus and animals were healthy following surgery.

Figure 1. Disrupting PDE11A homodimerization in vivo decreases PDE11A4 expression in a compartment-specific manner and is sufficient to reverse age-related PDE11A4 accumulations in ghost axons.

A) An mCherry-tagged isolated GAF-B domain disrupts wild-type (WT) PDE11A4 homodimerization; whereas, mCherry alone does not (Pathak, Agostino et al. 2017). B) In vitro, addition of the GAF-B construct to either WT PDE11A4 (WG) or the hyperaccumulating PDE11A-S117D/S124D mutant (D/D-G; HT-22 cells shown) reduced PDE11A4 accumulation relative to that seen when mCherry-alone was added to either construct (WC and D/D-C, respectively). C) In neuronal HT-22 cells, the GAF-B construct reversed the aging-like effects of the D/D mutations (D/D-G), normalizing levels of accumulation back to that observed in the WT + mCherry group (WC; n=12 biological replicates/group; effect of S117D/S124D: F(1, 43)=8.80, P=0.005; effect of GAF-B: F(1,43)=16.74, P<0.001). D) A reduction in accumulation was also observed in non-neuronal HEK293T cells (n=12 biological replicates/group; effect of S117D/S124D: F(1, 44)=46.54, P<0.001; effect of GAF-B: F(1,44)=10.31, P=0.002). E) The mCherry-tagged isolated GAF-B domain construct and the mCherry-alone construct (i.e., the negative control) were then subcloned into the SPW lentiviral transfer vector. F) Subsequently generated lentiviruses were injected bilaterally into dorsal and ventral CA1 of hippocampus, G) with post hoc mapping of viral expression in surgerized mice showing limited spread to other subfields. H) As previously reported (Smith, Farmer et al. 2021), expression of mCherry-GAF-B in vCA1 of old wild-type (WT) mice decreased PDE11A4 expression in distal dendrites (dotted-arrows) and axons (solid arrows) relative to the cell body layer of the ventral hippocampal CA1 field; whereas, infusion of mCherry alone resulted in a uniform expression pattern across layers. I) Relative to the mCherry construct (n=3M,4F), the GAF-B construct (n=7M,9F) also reduced the number of age-related PDE11A4 ghost axons when infused into vCA1 of old WT mice ((Pilarzyk, Porcher et al. 2022). J) A ~60% reduction in PDE11A4 ghost axons was detected in old vCA1 using an antibody that recognizes all PDE11A4 (i.e., “#1 PDE11A”; fails normality, Rank Sum Test for effect of group: T(6,16) = 99.00, P=0.022) or an antibody cocktail specifically recognizing the hyperaccumulating PDE11A4-pS117/pS124 (i.e., “140/150 antibodies”; fails normality, Rank Sum Test for effect of group: T(7,15) = 109.00, P=0.043). Brightness, histogram stretch, and/or contrast of images adjusted for graphical clarity. #vs mCherry, P=0.005 to <0.001; @vs WT, P=0.002 to <0.001; *vs WT+mCherry, P=0.043–0.022.

2.5. Tissue/slide Collection:

As previously described (e.g., (Pilarzyk, Klett et al. 2019)), mice were euthanized (during light cycle) via rapid cervical dislocation and brains were immediately collected and flash frozen in 2-methylbutane sitting on dry ice. Brain tissue was then stored at −80 °C until cryosectioning at −18 °C. 20 μm sections were collected and thaw mounted onto glass +/+ slides. Slides were briefly dried at room temperature and then temporarily stored in the cryostat until all sections were collected. Once all sections were collected for a given brain, slides were boxed and stored at −80 °C.

2.6. Immunofluorescence.

Immunofluorescence was chosen to assess protein expression instead of Western blots so that injection coordinates could be verified. As previously described (Hegde, Capell et al. 2016), slides were fixed in room temperature 4% paraformaldehyde in phosphate-buffered saline (PBS) for 20 minutes. After fixation, 3X10 minute washes with PBS and 3×10 minute washes with PBT (phosphate- buffered saline/0.4% BSA/0.3% Triton-X 100) were performed to reduce background. Primary antibodies were combined in a tube with PBT at validated and optimized concentrations: PDE11A4 (Aves custom PDE11A4 #1 at 1:10,000; Fabgennix PPD11A-140AP at 1:1000; Fabgennix PPD11A-150AP at 1:500) and mCherry (ThermoFisher #PA534974 at 1:1000; Invitrogen #M11217 at 1:500; PhosphoSolutions #1203-mCherry at 1:10,000). Multiple PDE11A antibodies were utilized to discern the ectopic accumulation of PDE11A4 in ghost axons. While the PDE11A4#1 antibody labels all PDE11A4 and, thus, detects both diffusely localized and accumulated PDE11A, the PDE11A4–140 and 150 antibodies specifically labels PDE11A4 phosphorylated at serines 117 and/or 124 and, thus, only detects PDE11A4 accumulated in ghost axons (Pilarzyk, Porcher et al. 2022). To limit non-specific labeling of the mCherry antibodies hosted in a rodent species (see above Invitrogen and PhosphoSolutions), tissue was pretreated with anti-Mouse FabFragments (0.15mg/ml; Jackson Immunoresearch # 715-007-003) in PBS for 2 hours, followed by 3×10 minute washes in PBT, prior to adding primary antibody. Primary antibody solution was added over brain sections and the slides were kept level at 4°C overnight. Primary antibodies were removed using 4X10 minute washes in PBT. For optimal labeling, PDE11A4 secondary antibody (Alexafluor 488 AffiniPure Donkey Anti-Chicken, 1:1000, Jackson Immunoresearch #703-545-155) was applied first to the slides for 90 min at room temperature. The secondary was washed off using 3X10 minute washes with PBT. Next, the mCherry secondary (Alexafluor 594 AffiniPure species-specific, 1:1000, Jackson Immunoresearch) was repeated for the same time and conditions. Finally, 3×10 minute washes with PBT were used to clear the slides of any remaining secondary and the slides were briefly dip-rinsed in PBS to remove Triton. The slides were wiped dry along back, sides, and edges and mounted using DAPI Fluoromount-G (Southern Biotech, #0100-20). Ghost axons (i.e., intensely stained PDE11A4-filled filamentous-shaped structures) were counted by an experimenter blind to treatment. Images were captured using the Leica Application Suite (LASX) software in conjunction with a Leica DM5000 B florescent microscope and analyzed in Image J using the multi-point tool

2.7. Social Transmission of Food Preference.

As previously described (Hegde, Capell et al. 2016, Pilarzyk, Klett et al. 2019), subjects’ access to food was restricted the two days prior to testing to one hour per day. The day prior to testing, all mice were placed in a clean home cage and given access to plain powdered chow packed into a glass jar. The following day, the designated “demonstrator mouse” was individually placed in a clean home cage and fed powdered chow flavored with a household spice (e.g. 5% orange vs 0.5% anise for the 24-hour test or 2% basil vs 1.5% thyme for the 7 day test). After one hour, the “demonstrator mouse” was returned to the original home cage where the “observer” cage mates were allowed unrestricted access to the demonstrator for 15 minutes. It is during this time that the observers make an association between the social pheromones in the breath of the demonstrator and the non-social odor (household spice). Recent and remote long-term memory were assessed 24 hours or 7 days after training, respectively. At that time, the observer mice were individually placed in clean home cages and given access to two flavored/powdered chows for 1 hour. One flavored chow contained a novel spice and the other contained the spice that their demonstrator was given. The amount of food eaten was measured by an experimenter blind to treatment. All mice met the minimum inclusion criteria of eating at least 0.25 grams of food. Note, the same cohort of mice was used to test both recent and remote memory, to reduce the total number of mice used since we have shown that this does not confound interpretation of the data (Pilarzyk, Klett et al. 2019). Mice were allowed to recover from food restriction for ~3 weeks prior to training for the second retrieval time point. Observer mice eating more food containing the familiar spice (i.e., the spice on their demonstrator’s breath) versus the novel spiced food constituted memory (preference ratio: familiar-novel/familiar+novel).

2.8. Odor Recognition.

As previously described (Hegde, Capell et al. 2016, Pilarzyk, Klett et al. 2019), subjects were allowed to habituate to 1” round wooden beads (Woodworks) for at least seven days prior to testing by placing several beads in the subjects’ home cages. For social odor recognition (SOR), the wooden beads are placed in home cages of other mice so the beads can take up the scent of different mouse strains (e.g., C57BL/6J Jax #000664, BALB/cJ Jax #000651, 129S6/ SvEv Taconic #129SVE). For non-social odor recognition (NSOR), the wooden beads are placed in a bag containing bedding saturated with a household spice (e.g., marjoram, cumin, etc.) for at least 7 days. Training for SOR and NSOR consisted of a habituation trial with 3 beads from the subject’s home cage, followed by two training trials that included 2 home-cage beads and 1 novel-scented bead. Recent and remote long-term memory were assessed 24 hours or 7 days after training, respectively. During SOR, mice were tested with one home cage bead, one bead from the trained donor strain (familiar), and one bead from a second donor strain (novel). During NSOR, mice were tested with only two beads, one scented with the training spice and one a novel spice. The designation of which scent was “novel” within a given testing trial and the location of the novel scent (i.e., left versus right) was counterbalanced across subjects. Mice were given two minutes to investigate the beads and the amount of time spent on each was manually scored by an experimenter blind to treatment and bead. For reasons not yet understood, we previously determined that infusion of even a negative control lentivirus into CA1 of hippocampus reverses the recent long-term memory impairment observed in Pde11a KO mice 24 hours after training (Pilarzyk, Klett et al. 2019). Therefore, we do not report here 24-hour SOR recent long-term memory following injection of the isolated GAF-B domain as the results would not be interpretable. All mice met minimum inclusion criteria of spending a minimum of 3 seconds in total sniffing the beads. Spending more time investigating the novel vs familiar scent constituted memory (preference ratio: novelfamiliar/novel+familiar).

2.9. PDE and cyclic nucleotide assays.

cAMP- and cGMP-PDE catalytic activity were measured as previously described (Smith, Farmer et al. 2021). The assay was validated in vitro using HT-22 cells (a mouse hippocampal cell line; see (Smith, Farmer et al. 2021)). Buffer containing 20 mM Tris-HCl and 10 mM MgCl₂ was used to harvest cells and kept on ice until ready to use. PDE activity was measured using 50 μl of sample and 50 uL of of [³H] cAMP (Perkin Elmer, NET275) or cGMP (Perkin Elmer, NET337) and incubated for 10 minutes. After incubation, 0.1M HCL was added to quench the reaction, followed by 0.1M Tris to neutralize the reaction. 3.75 mg/mL snake venom (Crotalus atrox, Sigma V-7000) was then added to complete the reaction and the mixture was incubated at 37 °C for 10 minutes. Samples were put into 5’polystyrene chromatography columns with coarse filters (Evergreen, 208-3383-060) containing DEAE Sephadex A-25 resin (VWR, 95055–928). The columns were equilibrated in high salt buffer (20mM Tris-HCL, 0.1% sodium azide, and 0.5M NaCl) and low salt buffer (20mM Tris-HCL and 0.1% sodium azide). The reactions were then run down the equilibrated columns. Following four washes with 0.5 ml of low salt buffer, 4ml of Ultima Gold XR scintillation fluid (Fisher, 50-905-0519) was added to the eluate and mixed thoroughly. A Beckman-Coulter liquid scintillation counter Beckman LS 6000) was used to read counts per minute (CPM). As an assay control, two reactions free of sample lysate were ran in parallel to account for background activity and could then be subtracted from the sample CPMs. Total protein levels were quantified using the DC Protein Assay Kit (Bio-Rad, Hercules, CA) as described above, and CPMs were then normalized to the total amount of protein in each sample. To measure cAMP and cGMP levels directly, Cayman Chemical ELISA kits were used as per manufacturer’s instructions, as previously described (Pilarzyk, Klett et al. 2019).

2.10. Biochemical Fractionation and Western Blotting.

As previously described (Pathak, Agostino et al. 2017, Patel, Klett et al. 2018, Porcher, Bruckmeier et al. 2021), biochemical fractionation was performed to obtain cytosolic and soluble membrane fractions. Cells were mixed with ice cold fractionation buffer (FB: 20 mM Tris-HCl, pH 7.5; 2 mM MgCl2; Thermo Pierce Scientific phosphatase tablet #A32959 and protease inhibitor 3 #P0044) and sonicated. First, a low-speed spin (1000 × g) removed cellular debris and the supernatant from this spin was transferred to a new tube. Next, a high-speed spin (89,000 × g) was performed to obtain the membrane (pellet) and cytosolic (supernatant) proteins. The pellet was rinsed and resuspended by sonication in fractionation buffer with 0.5% Triton-X 100. To solubilize the protein, samples were nutated for at 4°C for 30 minutes. A second high-speed spin (60,000 × g) was done for 30 minutes to separate the soluble membrane (supernatant) from the insoluble membrane (pellet). The soluble membrane sample was then transferred to a clean tube and used for western blot (see below). A DC Protein Assay kit (Bio-Rad; Hercules, CA, USA) was used to determine protein concentrations by which total protein was equalized across samples at 1 μg/μl. Samples were stored at −80°C until used in Western blotting. For Western blotting, 10ug of protein was loaded onto 12% NuPAGE Bis-Tris gels (Invitrogen, Waltham MA) and run at 180 volts. After about one hour, proteinwas transferred onto 0.45um nitrocellulose membrane using 100 mA for two hours. Following transfer, tris-buffered saline with 0.1% tween20 (TBS-T) was used to wash membranes. Membranes were cut into multiple strips to probe for multiple antibodies if needed. The membranes were blocked using either 5% milk or Superblock (PBS) Blocking Buffer (ThermoFisher, Cat#37515). Primary antibody was applied overnight at 4°C for PDE11A (Fabgennix PD11-101 at 1:500). The following day, membranes were washed with TBS-T (4X10 minutes). Secondary antibody (Jackson Immunoresearch Anti-Rabbit, 111-035-144; 1:10 000), was applied at room temperature for one hour. Finally, the membrane was washed in TBS-T (3X15 minutes). Chemiluminescence (SuperSignal West Pico Chemilumiscent Substrate; ThermoScientific, Waltham MA) was captured using film and multiple exposures were taken to ensure densities were within the linear range of the film. ImageJ was used to quantify optical densities. As previously described (e.g., (Kelly, Adamowicz et al. 2014, Patel, Klett et al. 2018, Porcher, Bruckmeier et al. 2021)), each blot was normalized to a control condition (e.g., WT) to account for any technical variables (film exposure, antibody signal-noise, variance in chemiluminescence, etc.).

2.11. Data Analysis.

Data was collected by investigators blind to treatment and experiments were designed to counterbalance technical/biological variables. Outliers more than 2 standard deviations away from the mean were removed prior to analyses, as previously described (e.g., (Kelly, Logue et al. 2009, Pathak, Ibrahim et al. 2015, Patel, Klett et al. 2018)). Outliers removed/total n: Figure 1C, 1/48; Figure 1J, 2/46; Figure 2A, 2/18; Figure 2B, 1/22; Figure 2D, 5/47; Figure 2E, 4/48; Figure 3AB, 1/18; Figure 3C, 3/45; Figure 3E, 2/20; Figure 3F, 2/35; Figure 3G, 3/48; and Figure 3G, 9/112;. Data were analyzed for effect of genotype, behavioral measure (e.g., bead or food), and treatment. Parametric statistical analyses were run on SigmaPlot 14.5 (San Jose, CA, USA) including ANOVA (F), Student’s t-test (t), and one-tailed one-sample t-test (t) when datasets met assumptions of normality (Shapiro-Wilk test) and equal variance (Brown-Forsythe test). To offset the possibility of a Type I error associated with multiple comparisons, a false-rate discovery (FDR) correction was applied to P-values from one-sample t-tests within an experiment, as previously described (Pilarzyk, Klett et al. 2019, Smith, Farmer et al. 2021). If analyses failed normality and/or equal variance, nonparametric Kruskal-Wallis ANOVA (H) or Mann-Whitney rank sum test (T) were used instead. Student-Newman-Keuls or Dunn’s test were performed for Post hoc analyses. Significance was defined as P<0.05.

Figure 2. Disrupting PDE11A4 homodimerization in the hippocampus of old WT mice is sufficient to reverse age-related decline of social memory.

Into CA1 of Old WT mice, we injected either the mCherry-tagged GAF-B domain (O-WG) or mCherry alone (O-WC). We then tested recent long-term memory (LTM) 24 hours after training as well as remote LTM 7 days after training. A) In the social transmission of food preference assay (STFP), O-WG mice exhibited impaired recent LTM relative to O-WC and young mice (O-WG; n=3M/3F, O-WC; n=3M/2F, Young; n= 5F; F(2,13) = 16.52, P<0.001; Post hoc: GAF-B vs each group, P<0.001). B) In contrast, O-WG mice (n= 3M/3F) and young mice (n=8F) demonstrated significant remote LTM for STFP, but O-WC mice did not (n= 4M/3F; Table 1). C) O-WG and O-WC mice learned non-social odor recognition (NSOR) equally well relative to young mice (O-WG, n=5M/10F; O-WC, n=5M/11F; Young, n= 2M/15F; effect of trial for novel odor: F(1,44) = 72.73, P<0.001). D) All groups showed equivalent NSOR recent LTM (O-WG, n=3M/9F; O-WC, n=4M/11F; Young, n= 2M/13F; F(2,39) = 1.85, P=0.17) and E) NSOR remote LTM (O-WG, n=4M/10F; O-WC, n=5M/10F; Young, n= 2M/13F; failed equal variance, ANOVA on Ranks: H(2)= 1.22, P=0.543). F) O-WG and O-WC mice also learned social odor recognition (SOR) equally well relative to young mice (O-WG, n=5M/10F; O-WC, n=5M/11F; Young, n=2M/15F; effect of trial for novel odor: F(1,45) = 110.57, P<0.001). G) In contrast to NSOR, O-WG mice showed improved remote SOR LTM relative to O-WC mice (O-WG, n=5M/9F; O-WC, n=5M/11F; Young, n= 2M/15F; F(2,45) = 3.42, P=0.041; Post hoc: O-WC vs. O-WG P= 0.026). Data plotted as individual points (females as circles, males as squares) and expressed as mean ±SEM. #vs Young, P=0.001; ^vs Trial 1, P<0.001; *has memory (i.e., significantly >0), P=0.023 to <0.001 (see Table 1 for one-sample t-tests); @vs O-WC, P=0.026.

Figure 3. Disrupting PDE11A4 homodimerization reverses biochemical phenotypes associated with aging independently of PDE11A4-pS162.

Relative to hippocampal HT-22 cells transfected with green fluorescent protein (GFP) alone + mCherry alone (GC, n = 6 biological replicates), HT-22 cells transfected with either GFP-PDE11A-WT + mCherry (WC; n=6 biological replicates) or GFP-PDE11A-WT + mCherry-GAF-B (WG; n=5 biological replicates) showed equivalent increases in both A) cGMP-PDE activity (F(2,14)=168.01, P<0.001; Post hoc: GFP vs. all groups, P<0.001) and B) cAMP-PDE activity (F(2,14)=12.85, P<0.001; Post hoc: GFP vs. WC P<0.001, GFP vs. WG P=0.005). C) In contrast, disrupting homodimerization attenuated PDE11A4-induced decreases in cGMP levels (n=14 biological replicates/group; F(2,39)=4.33, P=0.020; Post hoc: GFP vs. WC P=0.020, WG vs. WC P=0.041) while D) exacerbating PDE11A4-induced decreases in cAMP in COS1 cells (n=10 biological replicates/group; failed equal variance; ANOVA on Ranks: H(2)=16.34, P<0.001; Post Hoc: GFP vs. all groups P<0.001, WC vs. WG P=0.034). E) As previously reported for the GAF-B construct (Pathak, Agostino et al. 2017), S162D shifted PDE11A4 from the membrane to the cytosol relative to WT (n=9 biological replicates/group; t(16)=−3.21, P=0.005). F) That said, S162D differentially regulated cyclic nucleotide levels, with no effect on PDE11A4-induced decreases in cGMP levels (n=10–12 biological replicates/group; F(2,30)=6.99, P=0.003; Post hoc GFP vs. WT P=0.004, GFP vs. 162D P=0.007, WT vs 162D P=0.924) and G) an attenuation of PDE11A4-induced decreases in cAMP levels (n=15 biological replicates/group; F(2,42)=8.76, P<0.001; Post hoc GFP vs. WT P<0.001, WT vs. 162D P=0.020, GFP vs 162D P=0.089). H) Importantly, the GAF-B construct was able to reduce the punctate accumulation of the phosphoresistant mutant PDE11A4-S162A (S162A-G) relative to S162A plus mCherry (S162A-C) in HEK293T cells (n=12 biological replicates/group; effect of GAF-B: F(1,26)=135.65, P<0.001) and I) COS-1 cells (n=25–26 biological replicates/group; effect of GAF-B: F(1,99)=137.52, P<0.001). *vs. GC, P=0.02-<0.001; #vs. mCherry P=0.041-<0.001; @vs. WT, P=0.02-<0.001. Data plotted as individual data points and mean ±SEM.

3. RESULTS

3.1. Disrupting PDE11A4 homodimerization in the aged hippocampus selectively decreases PDE11A4 expression and accumulation in a compartment-specific manner.

We previously found that ventral hippocampal PDE11A4 expression and phosphorylation of PDE11A4 at S117 and S124 increases with age in mice, rats, and humans and that these age-related increases accumulate ectopically in the membrane compartment and filamentous structures termed “ghost axons” (Kelly, Adamowicz et al. 2014, Pilarzyk, Porcher et al. 2022). Additionally, we found in vitro that disrupting PDE11A4 homodimerization by expressing an isolated PDE11A4 GAF-B binding domain (Figure 1A) leads to proteolytic degradation specifically of membrane-associated PDE11A4 and reduces the punctate accumulation of PDE11A4 (Pathak, Agostino et al. 2017). Therefore, we sought to determine if disrupting PDE11A4 homodimerization would be sufficient to reverse age-related PDE11A4 proteinopathies in a compartment-specific manner. To do this, we first determined if disrupting PDE11A4 homodimerization would normalize the increased protein expression observed with PDE11A4-S117D/S124D, which mimics the age-related increase in PDE11A4-pS117/pS124 (Pilarzyk, Porcher et al. 2022). Consistent with our previous report using Western blots to measure increased protein expression of PDE11A4-S117D/S124D versus PDE11A4-WT, here we observed across multiple experiments and cell lines (i.e., HT-22 and HEK293T) that the total number of cells expressing enough protein to be considered labelled was significantly greater with PDE11A4-S117D/S124D than PDE11A4-WT when co-expressed with the negative control mCherry (Table 1). However, when co-expressed with the isolated GAF-B domain, the total number of cells labelled with PDE11A4-S117D/S124D was normalized to that observed in the PDE11A4-WT + mCherry and PDE11A4-WT + GAF-B groups (HT22 effect of interaction: F(1,43)=15.73, P<0.001; HEK293T effect of interaction: (F(1,44)=9.74, P=0.003; see Table 1 for means and Post hoc analyses). In addition, expression of the isolated GAF-B domain reduced the percent of labelled cells that demonstrated a punctate accumulation of PDE11A4, normalizing the level of accumulation observed withPDE11A4-S117D/S124D back to that of PDE11A4-WT (Figure 1B–D). Given the effectiveness of the isolated GAF-B domain against the aging-like mimic PDE11A4-S117D/S124D in vitro, we next used lentiviruses (Figure 1E) containing either mCherry alone (i.e., negative control) or an mCherry-tagged isolated GAF-B domain that disrupts PDE11A4 homodimerization (Figure 1E; (Pathak, Agostino et al. 2017)).

Table 1.

Total number of HT22 and HEK293T cells expressing enough protein to be considered labelled following transient transfection with with PDE11A4-S117D/S124D (D/D) or with PDE11A4-WT (W) in combination with the negative control mCherry (C) or an mCherry-tagged isolated GAF-B domain (G; Data expressed as mean ±SEM).

cell line	WC	D/D-C	WG	D/D-G	Significant post hoc tests:
HT22	96.8 ±4.4	131.5 ±5.2	99.1 ±4.4	95.2±5.3	D/D-C vs WC, P<0.001; D/D-C vs D/D-G, P<0.001
HEK293T	161.2 ±9.3	202.1 ±14.3	184.3 ±8.7	168.8±5.7	D/D-C vs WC, P=0.006; D/D-C vs. D/D-G, P=0.008

^*Data analyzed for effects of Plasmid 1 (WT vs D/D) vs Plasmid 2 (C vs G) by Two-Way ANOVA (F) having passed assumptions of normality and equal variance. Post hoc analyses by Student-Newman-Keuls.

These lentiviruses were stereotaxically injected bilaterally into the CA1 field of dorsal and ventral hippocampus of old Pde11a WT mice (Figure 1F–G), since this is the field where PDE11A4 regulates social learning and memory (Pilarzyk, Klett et al. 2019). Some limited spread of the viruses was observed outside of CA1 into dentate gyrus, CA3, CA2 (Figure 1G), but should not be of concern since PDE11A4 is not expressed in those areas. While mCherry-treated mice exhibited a uniform pattern of PDE11A4 expression across stratum radiatum (i.e., dendritic layer), stratum pyramidale (i.e., cell body layer) and stratum oriens (i.e., axonal layer) of ventral CA1, GAF-B treated mice exhibited a significant compartment-specific decrease in PDE11A expression, with reduced expression in the distal segment of stratum radiatum and stratum oriens relative to stratum pyramidale (Figure1H; see (Smith, Farmer et al. 2021) for quantification of effect). Consistent with the fact that the isolated GAF-B domain reduces PDE11A4 protein expression in the membrane compartment (Pathak, Agostino et al. 2017), we found that disrupting homodimerization of PDE11A4 in vivo reduced age-related increases in so-called “PDE11A4 ghost axons” (i.e., filamentous structures harboring age-related accumulations of PDE11A4 (Pilarzyk, Porcher et al. 2022); Figure 1I–J). The ability of the GAF-B construct to reduce age-related accumulations of PDE11A4 in ghost axons was confirmed using two different PDE11A4 antibodies (#1 that detects all PDE11A4 and a “140/150” cocktail that detects PDE11A4-pS117/pS124 specifically; Figure 1J). All together, these data suggest disruption of PDE11A4 homodimerization in vivo is sufficient to reverse age-related increases in PDE11A4 protein expression and ectopic accumulation.

3.2. Disrupting PDE11A4 homodimerization in the hippocampus of old mice is sufficient to reverse age-related cognitive decline of remote long-term social memory.

Previously, we found that deletion of PDE11A in mice prevented age-related cognitive decline of long-term social memories 7 days after training (i.e., remote LTM; (Pilarzyk, Porcher et al. 2022)). Interestingly, this protection of remote LTM 7 days after training comes at the expense of not being able to access more recent LTM 24 hours after training (Pilarzyk, Porcher et al. 2022). Therefore, we determined if disrupting homodimerization of PDE11A4 using the isolated GAF-B domain would be sufficient to induce a transient amnesia in old Pde11a WT mice that ultimately reverses age-related cognitive decline of remote long-term social associative memories. Here we measured social associative memory using social transmission of food preference, an assay where mice form an association between a non-social odor (i.e., a household spice) and a social odor (i.e., pheromones in their cage mate’s breath). The memory of that association then indicates a food with that scent is safe to eat (Galef, Mason et al. 1988, Munger, Leinders-Zufall et al. 2010). Indeed, mice treated with the GAF-B domain showed no recent LTM for social transmission of food preference (Figure 2A; Table 2) but did show remote LTM for social transmission of food preference on par with that of young adult mice (Figure 2B; Table 2). In contrast, mice treated with mCherry alone showed age-related cognitive decline of remote LTM for social transmission of food preference (Figure 2B; Table 2). Importantly, we found no significant differences between groups in terms of the total amount of food eaten (Table 3). To disentangle effects of the GAF-B construct on the non-social versus social components of the social transmission of food preference assay, we turned to memory tests for odor recognition. Notably, recognition memories do not exhibit age-related cognitive decline (Pilarzyk, Porcher et al. 2022). Old Pde11a WT mice treated with mCherry alone or mCherry-GAF-B spent the same amount of time sniffing the beads (Table 3) and learned equally well during training for non-social odor recognition (Figure 2C). Old mice also demonstrated equally strong recent and remote LTM for non-social odor recognition as young mice (Figure 2D–E; Table 2). Thus, disrupting PDE11A4 homodimerization—like genetically deleting PDE11A (Hegde, Capell et al. 2016, Pilarzyk, Klett et al. 2019)—does not alter the ability to detect, learn about, or retrieve memories for recognizing non-social odors (Hegde, Capell et al. 2016). In contrast, disrupting PDE11A4 homodimerization—again, like genetically deleting PDE11A (Hegde, Capell et al. 2016, Pilarzyk, Klett et al. 2019)—did significantly improve remote LTM for social odor recognition memory (Figure 2G; Table 2), despite having no effect on social odor recognition learning (Figure 2F) or total time sniffing (Table 3). These data suggest that disrupting PDE11A4 homodimerization is sufficient to reverse age-related cognitive decline of remote social memory.

Table 2.

Preference ratios for each group were calculated and subsequently analyzed by one-sample t-test to determine if the group exhibited a significant memory (i.e., ratio significantly differed from chance = 0). The resulting P-values within an experiment were then corrected for multiple comparisons using false discovery rate correction (FDR).

Datasets	Young	O-WC	O-WG
Fig2A STFP 24h LTM	t(4)=9.51, FDR-P<0.001	t(4)=12.60, FDR-P<0.001	t(6)=−0.043, FDR-P=0.344
Fig2B STFP 7d LTM	t(7)=2.90, FDR-P=0.017	t(6)=0.14, FDR-P=0.896	t(5)=3.84, FDR-P=0.018
Fig2D NSOR 24h LTM	t(14)=8.19, FDR-P<0.001	t(14)=11.9, FDR-P<0.001	t(11)=7.80, P<0.001
Fig2E NSOR 7d LTM	t(16)=7.90, FDR-P<0.001	t(14)=9.28, FDR-P<0.001	t(13)=5.70, FDR-P<0.001
Fig2G SOR 7d LTM	t(16)=4.99, FDR-P<0.001	t(15)=2.32, FDR-P=0.023	t(14)=7.16, FDR-P<0.001

Table 3.

Amount of food eaten (grams) during social transmission of food preference long-term memory (LTM) tests or amount of time spent investigating (seconds) during social/non-social odor recognition LTM tests by young controls and old WT mice treated with either mCherry (O-WC) or GAF-B (O-WG; Data expressed as mean ±SEM).

	Young	O-WC	O-WG	effect of group:
Fig2A STFP 24h LTM	0.96 ±0.08	1.15 ±0.11	1.23±0.13	F(2,15)=1.63, P=0.23
Fig2B STFP 7d LTM	0.94 ±0.12	0.93 ±0.10	0.95±0.05	F(2,19)=0.01, P=0.99
Fig2D NSOR 24h LTM	39.80 ±2.0	31.13 ±2.5	33.5 ±3.4	F(2,39)=3.15, P=0.054
Fig2E NSOR 7d LTM	30.94 ±2.7	28.33 ±3.2	31.84 ±3.5	F(2, 41)=0.33, P=0.72
Fig2G SOR 7d LTM	43.29 ±2.5	26.44 ±2.5	31.47 ±3.0	F(2,45)=10.92, P<0.001

There were no effects of group on food eaten during social transmission of food preference (STFP) or the time spent investigating during non-social odor recognition (NSOR). Old mice did, however, spend less time sniffing during social odor recognition (SOR) than young mice (post hoc: Young vs O-WC, P<0.001; Young vs O-WG, P=0.003).

3.3. PDE11A4 homodimerization is an independent intramolecular mechanism that regulates PDE11A4 trafficking and functioning.

As noted above, disrupting PDE11A4 homodimerization significantly changes the subcellular compartmentalization of the enzyme. To better understand the functional consequences of disrupting PDE11A4 homodimerization, we measured PDE catalytic activity and cyclic nucleotide levels in cells transfected with GFP + mCherry (i.e., negative control), GFP-PDE11A4 + mCherry, or GFP-PDE11A4 + mCherry-GAF-B. Compared to the negative control, expression of PDE11A4 significantly increased cGMP- and cAMP-PDE activity in HT-22 cells as expected. This increase in PDE activity was not altered by disruption of PDE11A4 homodimerization (Figure 3A–B), which is consistent with previous studies using purified PDE11A4 enzyme (Weeks, Zoraghi et al. 2005). Also as expected, expression of PDE11A4 + mCherry decreased both cGMP and cAMP levels in COS1 cells relative to the negative control (Figure 3C–D). Interestingly, disrupting homodimerization of PDE11A4 decreased PDE11A4-mediated degradation of cGMP (Figure 3C) while exacerbating PDE11A4-mediated degradation of cAMP (Figure 3D). Together, these data suggest disruption of PDE11A4 homodimerization alters cyclic nucleotide signaling not by altering PDE11A4 catalytic activity directly, but rather by changing the subcellular localization of PDE11A4 protein (e.g., shifting from a cGMP-rich pool to a cAMP-rich pool).

3.4. Effects of disrupting PDE11A homodimerization do not require phosphorylation of PDE11A4-S162.

As described above, disrupting PDE11A4 homodimerization reduces the punctate accumulation of PDE11A4 both in vitro and in vivo ((Pathak, Agostino et al. 2017); Figure 1). Such a dispersing phenotype is also triggered by a phosphomimic mutation at PDE11A4-S162 (i.e., S162D) (Pilarzyk, Porcher et al. 2022). As such, we sought to determine if disrupting PDE11A4 homodimerization may work by promoting phosphorylation of S162. Indeed, S162D changed the subcellular compartmentalization of PDE11A4 in a manner similar to that observed when disrupting homodimerization with the isolated GAF-B domain (Pathak, Agostino et al. 2017)—namely, shifting PDE11A4 from the membrane to the cytosolic fraction (Figure 3E). That said, we found that the S162D mutation elicited quite different effects on cyclic nucleotide levels than were described above for disrupting PDE11A4 homodimerization. Specifically, the S162D mutation did not alter PDE11A4 hydrolysis of cGMP (Figure 3F) and reduced PDE11A4 hydrolysis of cAMP (Figure 3G). Importantly, the isolated GAF-B domain was able to effectively reduce the punctate accumulation of both PDE11A4-WT as well as the phosphoresistant PDE11A4-S162A (Figure 3H–I), suggesting phosphorylation of PDE11A4-S162 is not needed for the dispersing effect of the isolated GAF-B domain. Together, these data suggest that homodimerization and pS162 are independent intramolecular mechanism that regulate PDE11A4 function.

4. DISCUSSION

Previously, we found that age-related increases in PDE11A4 occur specifically within the membrane compartment of the ventral hippocampus and ectopically accumulate in filamentous structures we term “ghost axons” (see (Pilarzyk, Porcher et al. 2022)) due to age-related increases in the phosphorylation of PDE11A4-S117/S124. Here, we show that disrupting PDE11A4 homodimerization within hippocampal CA1 using a biologic encoding an isolated PDE11A4 GAF-B domain reverses the accumulating effect of S117D/S124D in vitro (Figure 1B–D), reduces PDE11A4 expression in ventral CA1 in a compartment-specific manner (i.e., in distal dendrites and stratum oriens; Figure 1H), and reverses the age-related accumulation of PDE11A4 in ghost axons in vivo (Figure 1I–J). Further, we show that disrupting PDE11A homodimerization is sufficient to reverse age-related cognitive decline of remote social LTMs—albeit at the expense of an inability to retrieve recent social LTMs (Figure 2A–B). Such a transient amnesia that ultimately produces stronger remote social LTMs is also observed with genetic deletion of Pde11a in young and old adult mice (Pilarzyk, Klett et al. 2019, Pilarzyk, Porcher et al. 2022). Although disrupting PDE11A homodimerization does not alter catalytic activity of the enzyme, it ultimately leads to higher cGMP levels and slightly lower cAMP levels (Figure 3C–D) by virtue of changing the subcellular localization of PDE11A4 (Figure 1B–D). This suggests that disrupting PDE11A4 homodimerization shifts PDE11A from a cGMP-rich pool to a cAMP-rich pool, thereby alleviating age-related decreases in cGMP that are widely reported to occur in the aging hippocampus (Kelly 2018).

4.1. Homodimerization at the GAF-B domain regulates the subcellular compartmentalization in vivo.

Our work here (and elsewhere (Pathak, Agostino et al. 2017, Smith, Farmer et al. 2021)) suggests that homodimerization is a key type of protein-protein interaction that regulates the compartmentalization of PDE11A4 both in vitro and in vivo. For example, the PDE11A4 sequence in BALB/cJ mice differs from that of C57BL/6J mice at a single amino acid that falls within the GAF-B domain (Pathak, Agostino et al. 2017). Whereas BALB/cJ mice encode a threonine at amino acid 499, C57BL/6J mice encode an alanine. This A499T BALB/cJ mutation strengthens homodimerization, elevates PDE11A4 protein expression, and increases the punctate accumulation of PDE11A4 both in vitro and in vivo (Pathak, Agostino et al. 2017, Smith, Farmer et al. 2021). In contrast, disrupting homodimerization using the isolated GAF-B domain has the opposite effect, lowering protein expression of PDE11A4 due to increased proteolysis and reducing the punctate accumulation of PDE11A4 both in vitro and in vivo ((Pathak, Agostino et al. 2017, Smith, Farmer et al. 2021); Figure 1B–D and Figure 1H–I). Homodimerization/oligomerization likely controls the subcellular compartmentalization of other GAF domain-containing PDE families, given GAF-B mutations change the subcellular compartmentalization of PDE6C as well (Cheguru, Majumder et al. 2015). The fact that disrupting PDE11A4 homodimerization lowers PDE11A4 expression and its punctate accumulation suggests that age-related PDE11A4 proteinopathies may be prevented or reversed by reducing levels of PDE11A4 homodimerization. Indeed, we show here that disrupting PDE11A4 homodimerization in old WT mice was sufficient to reverse age-related accumulation of PDE11A4 in ghost axons in the ventral hippocampus (Figure 1I–J) and age-related cognitive decline of a remote social LTM (Figure 2B). It will be of interest to future studies to determine how a reduction in PDE11A4 function might affect the morphology of hippocampal dendrites and axons. From a drug discovery perspective, GAF domains are highly interesting because they are found in no other mammalian proteins other than PDE2, PDE5, PDE6, PDE10, and PDE11, with poor conservation of sequence among these families leading to an inability to bind each other (Francis, Blount et al. 2011). Thus, a biologic targeting the PDE11A4 GAF-B domain would be highly specific in its binding.

4.2. Disrupting PDE11A4 homodimerization elicits effects independently of PDE11A4-pS162.

Interestingly, phosphorylation of PDE11A4 at serine 162 (pS162) reduces the accumulation of PDE11A4 in punctate structures (Pilarzyk, Porcher et al. 2022) and shifts PDE11A4 from the membrane to the cytosol (Figure 3E) as does disruption of PDE11A4 homodimerization (Pathak, Agostino et al. 2017). That said, we show here that these two regulatory mechanisms clearly operate independently. While disrupting homodimerization decreases cAMP levels (Figure 3D), mimicking phosphorylation of S162 increases cAMP levels (Figure 3G). These opposing effects suggest PDE11A4 is being redistributed to differing cytosolic compartments in response to disrupted homodimerization versus phosphorylation of S162. Further, disrupting PDE11A4 homodimerization reduces the accumulation of PDE11A4 even when phosphorylation of S162 is prevented (Figure 3H–I). Together, these data suggest that disrupting PDE11A4 homodimerization in vivo via expression of its isolated GAF-B domain is sufficient to reverse the age-related ectopic accumulation of PDE11A4 within ghost axons and rescue age-related cognitive decline of remote social LTMs independently of promoting the phosphorylation of S162.

4.3. Disrupting PDE11A4 homodimerization as a potential therapeutic approach for treating age-related PDE11A4 proteinopathies.

A hallmark pathology of the aging brain includes ectopic protein expression and accumulation in the brain (Yanar, Atayik et al. 2020), as well as the loss of cGMP signaling in the hippocampus (Vallebuona and Raiteri 1995, Chalimoniuk and Strosznajder 1998, Kelly 2018). Much like hyperphosphorylated tau causes neurofibrillary tangles that lead to impaired cell communication, function, and death (Moloney, Lowe et al. 2021), we find that hyperphosphorylation of PDE11A4 at serines S117 and S124 increases PDE11A4 protein expression and ectopic accumulation in filamentous structures termed “ghost axons” (Pilarzyk, Porcher et al. 2022). These age-related increases in PDE11A protein appear to cause age-related cognitive decline of social memories via the cGMP-PKG, as opposed to cAMP-PKA, pathway (Pilarzyk, Porcher et al. 2022). This is consistent with reports that aging and age-related cognitive decline are associated with decreases in cGMP, but not cAMP, in the hippocampus (Vallebuona and Raiteri 1995, Chalimoniuk and Strosznajder 1998, Kelly 2018). It is also consistent with the fact that elevating cGMP levels elicits nootropic effects in the context of age-related cognitive decline and neurodegenerative diseases (Kleppisch and Feil 2009, Puzzo, Staniszewski et al. 2009, Fiorito, Saeed et al. 2013, Palmeri, Privitera et al. 2013, Cossenza, Socodato et al. 2014, Fiorito, Vendome et al. 2017). Therefore, it is particularly noteworthy that disrupting PDE11A4 homodimerization increases cGMP levels (Figure 3C). The fact that cGMP levels are increased (Figure 3C) while cAMP levels are decreased (Figure 3D), coupled with the fact that PDE11A4 catalytic activity remains intact (Figure 3A–B) while its subcellular localization dramatically changes (Figure 1B–D), suggests that disrupting PDE11A4 homodimerization causes PDE11A4 to relocate from a cGMP-rich subcellular compartment to a cAMP-rich subcellular compartment. Indeed, the subcellular localization of a PDE is just as important to its overall function as is its catalytic activity (Baillie, Tejeda et al. 2019).

Although it is clear that age-related increases in PDE11A4 protein expression are deleterious to remote social memory, it is important to consider the possibility that the ectopic accumulation of PDE11A4 in ghost axons may actually serve as a protective mechanism to sequester and/or neutralize excess PDE11A4 (Hill, Hanzen et al. 2017, Pilarzyk, Porcher et al. 2022). An example of this type of sequestration is found with some PDE4A isoforms when they are bound by conformationally-altering catalytic inhibitors (e.g., (Christian, Anthony et al. 2010)). Therefore, therapeutic targeting of PDE11A4 intended to reduce its expression and ectopic accumulation may also need to promote clearance of PDE11A4 from the system. As noted above, disrupting PDE11A4 homodimerization significantly decreases the punctate accumulation of PDE11A4 and specifically promotes degradation of membrane-associated PDE11A4 (Pathak, Agostino et al. 2017). Our results here suggest that disruption of PDE11A4 homodimerization may be a sophisticated mechanism for therapeutic targeting of age-related PDE11A4 proteinopathies, by decreasing both its accumulation and clearing it from select compartments.

Highlights:

• Disrupting PDE11A4 homodimerization reduces aged PDE11A4 expression and ghost axons

• Disrupting PDE11A4 homodimerization moves it from a cGMP-rich to a cAMP-rich pool

• Disrupting PDE11A4 homodimerization reverses age-related decline of social memory