A Negative Allosteric Modulator of PDE4D Enhances Learning after Traumatic Brain Injury

Abstract

Traumatic brain injury (TBI) significantly decreases cyclic AMP (cAMP) signaling which produces long-term synaptic plasticity deficits and chronic learning and memory impairments. Phosphodiesterase 4 (PDE4) is a major family of cAMP hydrolyzing enzymes in the brain and of the four PDE4 subtypes, PDE4D in particular has been found to be involved in memory formation. Although most PDE4 inhibitors target all PDE4 subtypes, PDE4D can be targeted with a selective, negative allosteric modulator, D159687. In this study, we hypothesized that treating animals with D159687 could reverse the cognitive deficits caused by TBI. To test this hypothesis, adult male Sprague Dawley rats received sham surgery or moderate parasagittal fluid-percussion brain injury. After 3 months of recovery, animals were treated with D159687 (0.3 mg/kg, intraperitoneally) at 30 min prior to cue and contextual fear conditioning, acquisition in the water maze or during a spatial working memory task. Treatment with D159687 had no significant effect on these behavioral tasks in non-injured, sham animals, but did reverse the learning and memory deficits in chronic TBI animals. Assessment of hippocampal slices at 3 months post-TBI revealed that D159687 reversed both the depression in basal synaptic transmission in area CA1 as well as the late-phase of long-term potentiation. These results demonstrate that a negative allosteric modulator of PDE4D may be a potential therapeutic to improve chronic cognitive dysfunction following TBI.

1. Introduction

Traumatic brain injury (TBI) is a significant public health problem affecting 1.74 million people per year in the United States (Faul, Xu, Wald, & Coronado, 2010). Over 40% of TBI patients discharged after hospitalization develop long-term disabilities resulting from the brain trauma (Faul, Xu, Wald, & Coronado, 2010). The number of people living with chronic disabilities resulting from TBI is estimated to range from 3.2 to 5.3 million in the United States (Selassie et al., 2008; Zaloshnja, Miller, Langlois, & Selassie, 2008). A common symptom of TBI is cognitive difficulty, with learning and memory impairments reported in nearly 80% of chronic TBI survivors (Frieden, Houry, & Baldwin, 2015). Therapeutics to improve cognition in the chronic recovery phase of TBI have had some success. Cognitive rehabilitation has the strongest empirical support to improve recovery after TBI, but pharmacotherapies targeting particular molecular pathways have yet to achieve clinical trial success (Cicerone et al., 2011).

Preclinical research studies of TBI have found that deficits in activating cAMP response-element binding protein (CREB) may underlie hippocampal learning deficits in the chronic recovery phase of TBI (Titus et al., 2013a). Basal levels of phospho-CREB are reduced at 3 months after TBI and treatment with rolipram, a pan-phosphodiesterase 4 (PDE4) inhibitor rescues deficits in activation of phospho-CREB after fear conditioning in TBI animals (Titus et al., 2013a; Titus et al., 2016). Inhibition of all PDE4 enzymes in the brain or PDE4B selectively to increase CREB phosphorylation improves TBI-induced cognitive dysfunction (Titus et al., 2013a; Titus et al., 2016). Several PDE4 subtypes are elevated in the hippocampus after TBI, suggesting that particular PDE4 subtypes may be viable therapeutic targets (Wilson, Titus, Oliva Jr., Furones, & Atkins, 2016). There currently are two PDE4 inhibitors approved by the FDA, roflumilast and apremilast. These are competitive inhibitors of all PDE4 subtypes (Chong, Leung, & Poole, 2013; Del Rosso & Kircik, 2016). However, both are limited by gastrointestinal and emetic side effects and have poor uptake in the brain. Thus, there is a need to develop subtype-selective PDE4 inhibitors with improved brain access and tolerability. Understanding whether selectively targeting particular PDE4 subtypes can improve cognitive outcome is of critical need to develop this highly promising therapeutic approach.

There are four subtypes of PDE4 which are encoded by different genes, Pde4a-d (Houslay, Schafer, & Zhang, 2005). Most PDE4 inhibitors currently available inhibit all four PDE4 subtypes (Press & Banner, 2009). Pan-PDE4 inhibitors that inhibit all four PDE4 subtypes improve learning and memory in non-injured rodents and monkeys (Barad, Bourtchouladze, Winder, Golan, & Kandel, 1998; Li et al., 2011; Rutten, Basile, Prickaerts, Blokland, & Vivian, 2008). Of the four PDE4 subtypes, several lines of evidence implicate PDE4D in particular for hippocampal memory formation. PDE4D isoforms have varying mRNA distribution in the hippocampus (Miro, Perez-Torres, Puigdomenech, Palacios, & Mengod, 2002; Richter, Jin, & Conti, 2005). Moreover, manipulating PDE4D expression alters hippocampal synaptic plasticity and memory formation. Knockout of Pde4d enhances hippocampal long-term potentiation (LTP; Rutten et al., 2008b). At the behavioral level, Pde4d knockout mice have enhanced water maze acquisition, radial arm maze performance and novel object recognition, but impaired long-term cue and contextual fear conditioning (Li et al., 2011; Rutten et al., 2008). These findings have been supported by knockdown studies of PDE4D using RNAi, which resulted in augmentation of novel object recognition (Li et al., 2011). Selective PDE4D inhibitors have been developed and found to enhance learning and memory in healthy and diseased rodents for some learning tasks (Brullo et al., 2016; Bruno et al., 2011; Burgin et al., 2010; Ricciarelli et al., 2017; Sierksma et al., 2014; Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017), consistent with the Pde4d knockout and RNAi studies. These studies suggest that targeting PDE4D is a rational approach for developing a selective and well-tolerated therapeutic to improve cognitive dysfunction in the chronic recovery phase of TBI.

Development of selective PDE4D inhibitors has been hindered by the conservation of the catalytic domain sequence across all PDE4 subtypes (Omori & Kotera, 2007). Each Pde4 gene encodes several isoforms characterized by the presence or absence of regulatory regions termed upstream conserved regions (UCR) 1 and 2. Long isoforms of PDE4D dimerize and UCR2 closes over the catalytic site to inhibit cAMP hydrolysis. Within UCR2, PDE4D has an amino acid difference at phenylalanine 196, which is a tyrosine at 274 in PDE4A, 4B and 4C. Recently, x-ray crystallography studies have identified a PDE4D negative allosteric modulator D159687 that exploits this unique amino acid in PDE4D (Burgin et al., 2010). D159687 bridges Phe196 in the UCR2 domain over the catalytic domain in a “trans” position. This conformational change completely inhibits one active site of a PDE4D dimer and also reduces turnover rate at the other catalytic site, resulting in negative allosteric regulation of the enzyme. This negative allosteric modulation explains why D159687 partially, but not completely inhibits PDE4D with I_max reaching only 80–90%. Due to this amino acid difference across PDE4 subtypes, D159687 is 10-fold more selective towards mouse PDE4D7 (IC₅₀ 203 nM) versus PDE4B3 (IC₅₀ 2028 nM), 20-fold more selective towards human PDE4D7 (IC₅₀ 28 nM) versus human PDE4B1 (IC₅₀ 562 nM) and does not appreciably inhibit any other PDE (Burgin et al., 2010). All PDE4D isoforms that contain UCR2 can be negatively allosterically modulated by D159687; this includes all of the PDE4Ds 1–11 (Zhang, 2009). At an intravenous dose of 2 mg/kg, D159687 has a half-life of 2.3 hr in the mouse brain, 1.7 hr in plasma and a brain/plasma AUC ratio of 1.2. The T_max in the brain is 1 hr and C_max is 1478 ng/gm (Burgin et al., 2010). D159687 improves novel object recognition and Y-maze alternation in scopolamine-impaired mice and enhances novel object recognition in naïve mice (Burgin et al., 2010; Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). D159687 does not appear to have antidepressant- or anxiolytic-like effects when assessed in mice using the forced swim test, tail suspension test or elevated plus maze (Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). D159687 also has significantly improved tolerability for emetic side effects as compared to other PDE4 inhibitors (Burgin et al., 2010; Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). Given the advantages of this selective negative allosteric modulator of PDE4D, we tested the hypothesis that D159687 could improve learning and memory deficits in the chronic recovery phase of TBI.

2. Materials and methods

2.1. Subjects

All animal procedures were in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council of the National Academies, 8th ed.) and approved by the University of Miami Animal Care and Use Committee. Adult male Sprague-Dawley rats (behavior n=35, electrophysiology n=44, cAMP ELISAs n=36, flow cytometry n=6, 2–4 months old) were obtained from Charles Rivers Laboratories, housed singly and maintained on a 12 hr light/dark cycle with food and water ad libitum. Rats were prospectively, randomly assigned to receive moderate parasagittal fluid-percussion injury (FPI) or sham surgery, and treatment with vehicle or D159687. A prospective power analysis was performed to determine the minimum number of animals needed for the behavior experiments (Titus et al., 2013a). A sample size of 10 animals/group was obtained. Every effort was made to minimize the number of animals used and their suffering.

2.2. Fluid-percussion injury surgery

The FPI surgery was performed as previously described (Titus et al., 2013a). In brief, animals were anesthetized (3% isoflurane for induction, 0.5–2% isoflurane for maintenance with 70% N₂O, 30% O₂), received a 4.8 mm craniotomy over the right parietal cortex (−3.8 mm bregma, 2.5 mm lateral) and a beveled 18 gauge syringe hub was secured to the craniotomy site. Animals recovered for 12–16 hr while fasting with water ad libitum, and were re-anesthetized and mechanically ventilated. A fluid-percussion pulse (14–16 ms duration) was delivered to the right parietal cortex. Sham-operated rats received all surgical manipulations. Thermistors placed in the rectum and temporalis muscle and a tail artery catheter were used to maintain temperature and blood gases (pO₂ and pCO₂), blood pH and mean arterial blood pressure (MABP). No significant differences in these physiological parameters were observed in TBI animals pre-assigned to receive vehicle versus D159687 (Table 1). To minimize pain, distress and infection, buprenorphine (0.01 mg/kg, subcutaneous) and penicillin G potassium (20,000 IU/kg, intramuscular) were administered. Exclusion criteria were: mortality, >15% body weight loss, non-resolving infection at the surgical site, inability to feed or drink, motor paralysis, listlessness, self-mutilation, excessive grooming leading to loss of dermal layers, spontaneous vocalization when touched or poor grooming habits. Attrition for sham surgery was 0% and for TBI surgery was 2% (2 animals) which died at the time of surgery due to lung edema. Investigators were blind to the surgery assignment and drug treatment for all behavior, electrophysiology and histology analyses.

Table 1.

Physiological parameters

	Treatment group	Weight (gm)	ATM (at injury)	MABP (mmHg)	Blood pO2	Blood pCO2	Blood pH	Head temperature (°C)	Rectal temperature (°C)
Start of surgery	Sham n=14	378.0±13.9	N/A	123.8±3.5	143.1±5.7	39.0±0.6	7.45±0.01	36.7±0.1	37.1±0.1
	TBI+Vehicle n=10	383.9±10.6	1.98±0.03	125.1±3.9	151.0±7.6	39.5±0.9	7.44±0.01	36.7±0.0	36.9±0.1
	TBI+D15687 n=10	367.2±15.9	1.97±0.02	130.3±2.9	152.2±8.9	40.3±1.7	7.46±0.01	36.7±0.1	37.2±0.1
End of surgery	Sham			124.7±3.1	144.6±6.3	38.0±0.6	7.46±0.01	36.6±0.0	37.2±0.1
	TBI+Vehicle			122.9±3.5	139.7±5.4	36.7±0.5	7.47±0.01	36.7±0.0	36.9±0.0
	TBI+D15687			114.7±4.2	137.9±7.3	38.2±0.9	7.46±0.01	36.7±0.0	37.1±0.1
At perfusion	Sham	663.9±19.8	ATM: atmospheres of pressure, MABP: mean arterial blood pressure pO2: partial arterial oxygen pressure, pCO2: partial arterial carbon dioxide pressure
	TBI+Vehicle	642.0±14.0
	TBI+D159687	653.0±20.6

2.3. Drug administration

D159687 (0.3 mg/kg or 3 mg/kg, 6 ml/kg) or vehicle (10% PEG 400, 1% polysorbate 80, 5% ethanol, 84% saline) were administered intraperitoneally (i.p.). For electrophysiology experiments, D159687 was dissolved in DMSO at 10 mM and diluted in artificial cerebrospinal fluid (aCSF) to 100 nM.

2.4. Behavior assessments

Animals were tested in a series of tasks beginning at 3 months post-surgery (Fig. 2A): fear conditioning (week 12), water maze (week 13), spatial working memory (week 14), fear conditioning retention and shock threshold (week 16) and perfused at 18–20 weeks post-surgery. Animals received vehicle or D159687 at 30 min prior to training on fear conditioning, water maze acquisition, spatial working memory testing and shock threshold evaluation for a total of 8 doses in a 1 month period.

Fig. 2.

D159687 treatment rescued contextual and cue fear conditioning deficits in TBI animals at 3 months post-surgery. (A) Contextual and cue recent recall at 24 hr after training. TBI animals treated with vehicle froze significantly less than sham animals or TBI animals treated with D159687. No significant differences were observed in baseline freezing on the training day. ***p<0.001 vs Training, ^bp<0.01, ^cp<0.001 vs TBI+Vehicle, repeated measures two-way ANOVA with Tukey’s HSD correction. (B) Contextual and cue remote recall at 1 month after training. The deficits in contextual and cue recall were persistent in TBI+Vehicle animals as compared to sham or TBI+D159687 animals. ^ap<0.05, ^bp<0.01 vs TBI+Vehicle, one-way ANOVA with Tukey’s HSD correction. Mean±SEM, Sham n=14, TBI+Vehicle n=10, TBI+D159687 n=10.

2.4.1. Fear conditioning

Fear conditioning was performed as described (Titus et al., 2013a). Animals were habituated to the chamber on day 1 (10 min). On day 2, animals received vehicle or D159687 and 30 min later were fear conditioned by pairing a tone (30 s, 75 dB, 2.8 kHz) that co-terminated with a modest foot shock (1 mA, 1 s) occurring at 120 s. Total trial duration was 210 s. Recent recall was assessed at 24 hr after training and remote recall was evaluated at 1 month without drug treatment. Freezing was quantified by video analysis (Freeze Frame 3.32, Coulbourn Instruments, Holliston, MA, USA). For shock threshold evaluation, animals received vehicle or D159687 at 30 min prior to testing. The minimum foot shock intensity to elicit a flinch, jump or vocalization was measured beginning at 0.1 mA (1 s duration) and with 0.02 mA step increments (30 s intervals).

2.4.2. Water maze

Water maze training and testing were conducted as previously described (Titus et al., 2013a). Animals were trained with 4 acquisition days to find a hidden platform in a constant location (4 trials/day, 4 min inter-trial interval, 60 s maximum trial duration). Animals received vehicle or D159687 at 30 min prior to training for each acquisition day. Retention was assessed with a probe trial (60 s duration) at 24 hr after the last acquisition day in the absence of drug treatment. Water maze performance was measured with video tracking and analysis (EthoVision XT 11.5, Noldus Information Technology, Leesburg, VA, USA). Path length to reach the platform, thigmotaxis, time spent floating and swim velocity were measured during acquisition. Time spent in each quadrant was measured during the probe trial.

2.4.3. Spatial working memory

Spatial working memory was assessed for 2 days using a delayed match-to-place task in the water maze as previously described (Titus et al., 2013a). Animals received vehicle or D159687 at 30 min prior to each testing day. The spatial working memory task consisted of four location-match paired trials with the hidden platform remaining invariant between each location-match pair trial (trial duration 60 s). Location-match pair trials were separated by 5 s and trial pairs were separated by 4 min. Animals were video tracked and escape latency to find the hidden platform was measured (EthoVision XT 11.5). Data shown are from day 2.

2.5. Electrophysiology

At 3 months post-surgery, animals were decapitated under deep anesthesia (3% isoflurane, 70% N₂O, 30% O₂, 5 min). Transverse hippocampal slices (400 μm) were prepared as previously described (Titus et al., 2013a). CA1 field excitatory postsynaptic potentials (fEPSPs) were recorded in aCSF (125 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 25 mM NaHCO₃, 10 mM D-glucose, 2 mM CaCl₂, 1 mM MgCl₂, saturated with 95% O₂/5% CO₂) at 31°C. Recordings were made in stratum radiatum in response to Schaffer collateral pathway stimulation. The initial slope of the fEPSP was measured. After an input-output (I-O) curve was obtained, stimulation intensity was set at 40–50% of the maximum fEPSP. LTP was induced (4 trains of 1×100 Hz for 1 sec, separated by 5 min) at test stimulation intensity. Paired-pulse facilitation was assessed at test stimulation intensity with 12.5–250 msec inter-stimulus intervals.

2.6. Histology

Animals were anesthetized (3% isoflurane, 70% N₂O, 30% O₂, 5 min) and transcardially perfused with saline and 4% paraformaldehyde in 0.1M phosphate buffer, pH 7.4. Brains were paraffin-embedded and sectioned (10 μm thick) in a stereological series (150 μm apart) using a microtome (Leica RM2125 RTS, Buffalo Grove, IL, USA). Sections were stained with hematoxylin and eosin (H&E) plus Luxol fast blue. Microscope slides were scanned using Quick Scan PathScan Enabler IV 3.60.0.12 (Meyer Instruments, Inc., Houston, TX, USA) to obtain images at 7500 dpi (3.5 μm/pixel). The cortex and hippocampus were contoured at bregma levels −3.3 to −6.8 mm using Neurolucida 11.11.2 (MicroBrightField, Williston, VT, USA). Atrophy was quantified by subtracting the contralateral volume from the ipsilateral volume and normalizing to the contralateral volume. Images were obtained on an Olympus BX51TRF microscope (Olympus America, Waltham, MA, USA) with a 20× objective and stitched with StereoInvestigator 5.65 software (MicroBrightField).

2.7. Flow cytometry

At 3 months after surgery, animals were deeply anesthetized (3% isoflurane, 70% N₂O, 30% O₂, 5 min) and transcardially perfused with phosphate-buffered saline (PBS, pH 7.4, 4°C) for 6 min. The ipsilateral parietal cortex and hippocampus were dissected in isotonic saline at 4°C. Tissues were mechanically dissociated into single cell suspension in Hank’s balanced salt solution (HBSS, pH 7.4, 4°C, minus Ca²⁺ and Mg²⁺) with a 70 μm filter. Dead cells were excluded using LIVE/DEAD Fixable Near-IR dead cell stain (1 μl/ml, L10119, Life Technologies, Waltham, MA, USA). Cells were labeled with CD45 Alexa 647 (1.25 μg/ml, 202212, BioLegend, San Diego, CA, USA) and CD11b v450 (1 μg/ml, 53-4321-80, eBioscience, Waltham, MA, USA). The surface markers CD45 and CD11b were used to distinguish microglia (CD45^low, CD11b⁺) from other immune cell populations. Cells were fixed and permeabilized with BD Cytofix/Cytoperm Fixation/Permeabilization kit (554714, BD Biosciences, San Jose, CA, USA). Cells were intracellularly labeled with PDE4D (2 μg/ml, sc-25814, Santa Cruz Biotechnology). PDE4D was detected with PE-conjugated secondary antibody (10 μg/ml, 12-4739-81, eBioscience). Flow cytometry data was acquired on a BD LSR II flow cytometer with 4 emission lasers at 407, 488, 532 and 640 nm (BD Biosciences). Data were collected using BD FACSDiva 8.0.1 (BD Biosciences) and analyzed with Kaluza 1.2 software (Beckman Coulter, Brea, CA, USA).

2.8. cAMP ELISA

Rats received D159687 (0.3 or 3 mg/kg, i.p.) or vehicle and after 1 hr were decapitated while anesthetized (3% isoflurane, 70% N₂O, 30% O₂, 5 min). The parietal cortex and hippocampus were dissected at 4°C and frozen in liquid nitrogen. cAMP levels were analyzed using the Direct cAMP ELISA kit (ENZO Life Sciences, Farmingdale, NY, USA) as previously described (Titus et al., 2016). cAMP levels were normalized to total protein.

2.9. Statistical analysis

Data are expressed as group mean ± standard error of the mean (SEM). Significance was designated at p<0.05. Statistical analyses were performed in GraphPad Prism 6.05 (La Jolla, CA, USA) and SigmaPlot 12.0 (Systat, San Jose, CA, USA). Normality of the data was assessed prior to using parametric tests. Flow cytometry data were analyzed using an unpaired Student’s t-test. Remote recall of cue and contextual fear conditioning, shock threshold, path length during day 4 of water maze acquisition, LTP data from 0–5, 45–60 and 165–180 min post-tetanization, atrophy and cAMP ELISA data were analyzed with a one-way ANOVA and Tukey’s HSD correction for multiple comparisons. Probe trial data (animal treatment × quadrant), swim velocity (animal treatment × time) and depolarization during tetanization (animal treatment × time) were analyzed with a two-way ANOVA and Tukey’s HSD correction for multiple comparisons. All other data were analyzed by repeated measures two-way ANOVA (animal treatment × trial, time or current) and Tukey’s HSD correction for multiple comparisons.

3. Results

3.1. A PDE4D negative allosteric modulator did not alter cognitive outcome or synaptic plasticity in non-injured animals

Beginning at 3 months post-surgery, animals were evaluated using a series of behavioral tasks to determine the effects of D159687 on learning and memory ability and hippocampal synaptic plasticity (Fig. 1). For each behavioral task, animals were treated with vehicle or D159687 (0.3 mg/kg) at 30 min prior to training, but not prior to retention assessment (Fig. 1A). This dose was selected based on a previous study investigating a dose-response curve of D159687 to enhance novel object recognition in naïve rats and mice (Burgin et al., 2010). In rats, the minimum effective dose was 1 μg/kg and D159687 was effective up to 1 mg/kg with intravenous administration. In mice, D159687 improved novel object recognition at a minimum effective dose of 0.03 mg/kg and was maximally effective at 0.3 mg/kg when administered orally. We first assessed whether the PDE4D negative allosteric modulator D159687 would enhance learning and memory ability in non-injured, sham animals (Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). D159687 treatment had no significant effect on contextual fear conditioning in sham animals. Baseline freezing during training and both recent (24 hr post-training) and remote (1 month post-training) contextual fear memory were unaltered with D159687 treatment (Fig. 1B). D159687 did not alter shock threshold (data not shown). Similarly, water maze acquisition and retention in a probe trial and spatial working memory were unaltered by D159687 (Fig. 1C–E). Thigmotaxis during water maze acquisition was not significantly different in D159687-treated animals (data not shown). Hippocampal synaptic transmission and long-term potentiation (LTP) in area CA1 were also unaffected by administration of D159687 (100 nM) to hippocampal slices (Fig. 1F–H). These results indicate that at this particular dose of the PDE4D inhibitor D159687, baseline behavior, learning and memory ability in these behavioral tasks and hippocampal LTP were unaltered in sham, non-injured animals. Thus, the vehicle and D159687-treated sham animals were pooled and analyzed as one group (Sham) in subsequent comparisons to TBI animals.

Fig. 1.

D159687 treatment did not alter cognitive functioning or hippocampal synaptic plasticity in the non-injured, sham animals. (A) Timeline of behavioral testing. Animals received vehicle or D159687 (0.3 mg/kg, i.p., arrows) at 30 min prior to training for fear conditioning (Fear cond), water maze acquisition, spatial working memory (Work mem) and shock threshold assessment (Shock thresh). Retention in the water maze and fear conditioning (FC ret) were assessed without drug treatment. D159687 treatment had no significant effects in non-injured, sham animals on (B) contextual and cued fear conditioning, (C) water maze acquisition, (D) quadrant preference during the water maze probe trial, (E) escape latency to match the location in a spatial working memory task, (F) hippocampal fEPSP I-O curves in area CA1, (G) hippocampal LTP and (H) fEPSP responses between 170–180 min after LTP induction. Mean±SEM, Sham+Vehicle n=6, Sham+D159687 n=8 panels B–F, Sham+Vehicle n=8, Sham+D159687 n=6 panel F, Sham+Vehicle n=4, Sham+D159687 n=3 panels G and H.

3.3. Effects of the PDE4D negative allosteric modulator D159687 on cognitive outcome after TBI

To evaluate whether the PDE4D negative allosteric modulator D159687 would improve cognitive functioning in chronic TBI animals, cue and contextual fear conditioning were assessed at 3 months post-surgery. Animals were treated with vehicle or D159687 (0.3 mg/kg) at 30 min prior to training, and then were assessed for recent (24 hr post-training) and remote (1 month post-training) fear memory in the absence of drug treatment (Fig. 2). There were no significant differences between treatment groups in baseline freezing during training. In addition, all three treatment groups demonstrated contextual and cue fear conditioning by freezing significantly more when exposed to the context or cue at 24 hr after training as compared to baseline freezing at training (Context trial F_(1,31)=199.17, p<0.001; Cue trial F_(1,31)=94.05, p<0.001). However, TBI+Vehicle animals froze significantly less than sham animals or TBI+D159687 animals during recent recall of both the context and cue (Context Animal treatment F_(2,31)=7.57, p=0.002; Context Animal treatment × Trial F_(2,31)=7.52, p=0.002; Cue Animal treatment F_(2,31)=5.59, p=0.008; Cue Animal treatment × Trial F_(2,31)=4.38, p=0.021). Remote recall was assessed at 1 month after training. The improvement in contextual and cue fear conditioning persisted in TBI animals treated with D159687 as compared to vehicle-treated TBI animals (Fig. 2B). Shock thresholds were not significantly different between animal treatment groups (data not shown). These results indicate that D159687 improves both recent and remote recall of contextual and cue fear memory in chronic TBI animals.

The effects of D159687 on cognition were assessed in another hippocampal-dependent learning paradigm, the hidden platform version of the water maze at 13 weeks post-surgery (Fig. 3). Animals were treated with vehicle or D159687 (0.3 mg/kg) at 30 min prior to each acquisition trial, and were then evaluated for recall in a probe trial at 24 hr after the last acquisition trial without drug treatment. Although all animal groups showed an improvement in finding the hidden platform during acquisition, TBI+Vehicle animals were significantly impaired during acquisition as compared to sham animals or TBI+D159687 animals (Fig. 3A; Trial F_(3,93)=52.58, p<0.001; Animal treatment F_(2,93)=13.72, p<0.001). On acquisition day 4, TBI+Vehicle animals took a longer path to find the hidden platform as compared to sham animals or TBI+D159687 animals (Fig. 3B). The impaired acquisition was reflected in the probe trial (Fig. 3C). Sham animals and D159687-treated TBI animals, but not vehicle-treated TBI animals, spent significantly more time searching in the target quadrant as compared to the other quadrants (Quadrant F_(3,124)=58.75, p<0.001; Animal treatment × Quadrant F_(6,124)=8.41, p<0.001). To evaluate the effects of surgery or D159687 treatment on mediating anxiety or altering motor activity, thigmotaxis, swim speed and time spent floating during acquisition were measured. There were no significant differences between treatment groups for these parameters (data not shown). At 14 weeks post-surgery, spatial working memory was assessed (Fig. 3D). Animals received vehicle or D159687 treatment at 30 min prior to the task. TBI+Vehicle animals were significantly impaired in working memory as compared to TBI+D159687 animals (Animal treatment F_(2,31)=5.95, p=0.006). Comparison of TBI+Vehicle animals with sham animals did not reveal a statistically significant difference, although there was a strong trend (p=0.052). These results demonstrate that treatment of TBI animals with D159687 at 3 months post-injury significantly improves acquisition and retention in the water maze task and may improve spatial working memory.

Fig. 3.

D159687 improved spatial reference memory in TBI animals. (A) Path length to reach the hidden platform during water maze acquisition. TBI+Vehicle animals took a longer path length to find the platform as compared to sham or TBI+D159687 animals. ***p< 0.001, repeated measures two-way ANOVA with Tukey’s HSD correction. (B) Path length to find the platform on the fourth day of water maze acquisition. **p<0.01, ***p<0.001 vs TBI+Vehicle, one-way ANOVA with Tukey’s HSD correction. (C) Time spent in each quadrant during the probe trial. TBI+Vehicle animals spent significantly less time in the target quadrant as compared to sham or TBI animals treated with D159687. ***p<0.001 vs TBI+Vehicle in the target quadrant, ^cp<0.001 vs other quadrants, two-way ANOVA with Tukey’s HSD correction. (D) Spatial working memory was significantly improved in TBI+D159687 animals as compared to TBI+Vehicle animals. **p<0.01, repeated measures two-way ANOVA with Tukey’s HSD correction. Mean±SEM, Sham n=14, TBI+Vehicle n=10, TBI+D159687 n=10.

3.4. D159687 improved basal synaptic transmission and LTP in TBI animals

Pde4d knockout mice have enhanced hippocampal LTP following multiple tetanizations (Rutten et al., 2008). To determine whether the reversal of cognitive deficits in the TBI animals with D159687 was associated with improvements in hippocampal LTP, acute hippocampal slices were prepared at 3 months after surgery to assess basal synaptic transmission and synaptic plasticity (Fig. 4). As previously reported, TBI resulted in a significant reduction in the I-O curve (Current F_(11,249)=302.02, p<0.001; Animal treatment F_(2,249)=19.63, p<0.001; Animal treatment × Current, F_(22,249)=5.71, p<0.001; Titus et al., 2013a; Titus et al., 2016). Bath application of 100 nM D159687 to hippocampal slices from TBI animals significantly reversed the reduction in the I-O curve (Fig. 4A). Expression of hippocampal LTP was also significantly reduced in slices from TBI animals, and the late phase was improved with D159687 (Fig. 4B; Time F_(33,495)=39.51, p<0.001; Animal treatment F_(2,495)=4.10, p=0.038; Animal treatment × Time, F_(66,495)=1.98, p<0.001). Analysis of the fEPSP slope at 0–5 min and 50–60 min after tetanization indicated that D159687 treatment did not significantly alter posttetanic potentiation or early phase LTP. However, the fEPSP slope between 170–180 min after tetanization was significantly improved with D159687 treatment of TBI slices, indicating an improvement in late phase LTP (Fig. 4C). The rate of synaptic fatigue and total depolarization during tetanization were not significantly different between animal treatment groups (data not shown). Paired-pulse facilitation was also not affected by D159687 treatment in TBI slices (Fig. 4D). These results indicate that D159687 treatment improves synaptic transmission and late phase LTP in the hippocampus during the chronic recovery phase of TBI.

Fig. 4.

The PDE4D inhibitor D159687 reversed the deficits in basal synaptic transmission and late phase of LTP in area CA1 of the hippocampus at 3 months after TBI. (A) I-O responses were significantly reduced in hippocampal slices from TBI animals treated with vehicle as compared to sham animals or TBI animals treated with D159687 (100 nM). *p<0.05, ***p<0.001 vs TBI+Vehicle, ^ap<0.05 vs Sham, repeated measures two-way ANOVA with Tukey’s HSD correction. Sham n=14 slices/13 animals, TBI+Vehicle n=7 slices/6 animals, TBI+D159687 n=6 slices/6 animals. (B) Expression of LTP induced by high frequency tetanization (arrows, 4 trains of 1×100 Hz for 1 sec, separated by 5 min). Deficits in the late phase of LTP after TBI were rescued by treatment with D159687 (bar). *p<0.05 vs Sham, repeated measures two-way ANOVA with Tukey’s HSD correction. Sham n=7 slices/7 animals, TBI+Vehicle n=6 slices/6 animals, TBI+D159687 n=5 slices/5 animals. (C) Average of fEPSP slopes from 0–5, 50–60 and 170–180 min post-tetanization. *p<0.05, **p<0.01 vs TBI+Vehicle, one-way ANOVA with Tukey’s HSD correction. (D) Paired-pulse facilitation ratio was not significantly altered after TBI or D159687 treatment. Mean±SEM. Sham n=10 slices/10 animals, TBI+Vehicle n=6 slices/6 animals, TBI+D159687 n=6 slices/6 animals.

3.5. The PDE4D negative allosteric modulator did not reduce brain atrophy

At the completion of behavioral testing, cortical and hippocampal atrophy were measured at 5 months after surgery (Fig. 5). Significant cortical and hippocampal atrophy were observed in TBI animals treated with vehicle. There was no significant effect of D159687 treatment on cortical or hippocampal atrophy. These results suggest that treatment with D159687 did not improve cognition in chronic TBI animals by reducing atrophy.

Fig. 5.

Cortical and hippocampal atrophy at 5 months post-surgery. (A) Representative sections stained with H&E plus Luxol fast blue at bregma level −3.0 mm. Scale bar 1 mm. (B) Cortical atrophy was significantly increased in TBI animals treated with vehicle or D159687. (C) Significant hippocampal atrophy was observed in vehicle-treated TBI animals. A trend for hippocampal atrophy was observed in D159687-treated TBI animals (p=0.0551). *p<0.05, ***p<0.001 vs Sham, one-way ANOVA with post-hoc Tukey’s HSD correction. Mean±SEM, Sham n=14, TBI+Vehicle n=10, TBI+D159687 n=10.

3.6. PDE4D expression in microglia

After TBI, chronic microglia activation has been associated with impairments in synaptic plasticity and learning and memory deficits (Kumar & Loane, 2012; Muccigrosso et al., 2016; Ramlackhansingh et al., 2011). To determine whether D159687 may be targeting PDE4D in microglia chronically after injury, we used flow cytometry to evaluate PDE4D expression in microglia at 3 months after sham or TBI surgery. Using the surface markers CD45 and CD11b, we found that microglia (CD45^low, CD11b⁺) cell numbers were significantly increased in the ipsilateral parietal cortex of TBI animals as compared to sham animals (Fig. 6A–C). A fraction of microglia expressed PDE4D and the number of microglia expressing PDE4D was decreased after TBI at 3 months post-injury (Fig. 6B, D). In the ipsilateral hippocampus, there were no significant changes in microglia cell numbers after TBI and PDE4D expression levels were unaltered in the hippocampal microglia cell population (Fig. 6E–H). Infiltrating myeloid-lineage cells (CD45^high, CD11b⁺) were not detected in either the ipsilateral parietal cortex or hippocampus at this time point after TBI. These results suggest that the improvements in hippocampal-dependent learning and LTP with the PDE4D negative allosteric modulator D159687 were likely not due to targeting microglia expressing PDE4D.

Fig. 6.

PDE4D expression in Cd11b⁺/CD45⁺ microglia at 3 months after TBI. Representative flow cytometry scatter plots from the ipsilateral (A, B) parietal cortex and (E, F) hippocampus of sham and TBI animals. Total microglia cell numbers significantly increased in the (C) parietal cortex, but not in the (G) hippocampus. PDE4D expression was significantly decreased in the microglia cell population in the (D) parietal cortex, but not in microglia within the (H) hippocampus. *p<0.05, **p<0.01 vs Sham, Student’s unpaired t-test. Mean ± SEM, n=3/group.

3.7. cAMP levels in naïve animals and at 3 months after TBI

To shed light on the potential molecular mechanisms involved in the effects of D159687 on learning and memory, cAMP levels were evaluated in the parietal cortex and hippocampus at 1 hr after administration of D159687 (0.3 or 3 mg/kg, i.p.). In naïve animals, D159687 (0.3 or 3 mg/kg) significantly increased cAMP levels in the hippocampus when assayed at 1 hr after treatment. cAMP levels in the parietal cortex were significantly increased with 3 mg/kg, but not 0.3 mg/kg D159687 (Figure 7A). At 3 months after sham surgery or TBI, animals received D159687 (0.3 mg/kg, i.p.) and were analyzed 1 hr later for cAMP levels in the parietal cortex. D159687 treatment restored cAMP levels in TBI animals as compared to vehicle-treated TBI animals (Figure 7B).

Fig. 7.

cAMP levels were increased with D159687. (A) Treatment of naïve animals with D159687 at 3 mg/kg (i.p.) significantly increased cAMP levels in the parietal cortex and hippocampus. At the dose used in the behavior studies, 0.3 mg/kg D159687 also significantly increased cAMP levels in the hippocampus, but not in the parietal cortex. *p<0.05, **p<0.01 vs Vehicle, one-way ANOVA with Tukey’s HSD correction. Mean ± SEM, n=8/group. (B) Treatment of TBI animals with D159687 at 0.3 mg/kg (i.p.) significantly increased cAMP levels in the parietal cortex as compared to vehicle-treated TBI animals. *p<0.05 vs Vehicle, one-way ANOVA with Tukey’s HSD correction. Mean ± SEM, Sham+Vehicle n=4, TBI+Vehicle n=3, TBI+D159687 n=4.

4. Discussion

An ultra-rare, missense mutation in PDE4D causes a neurodevelopmental disorder known as acrodysostosis without hormone resistance (ACRDYS2) that is associated with intellectual disability, speech delay and psychomotor retardation (Butler, Rames, & Wadlington, 1988; Linglart et al., 2012). All of the ACRDYS2 mutations described to date are missense mutations that alter amino acids on the surface of the protein such as the contact residues between the PDE4D catalytic domain and the UCR2 regulatory domain (Gurney, D’Amato, & Burgin, 2015; Kaname et al., 2014; Lee et al., 2012; Lindstrand et al., 2014; Linglart et al., 2012; Lynch et al., 2013; Michot et al., 2012). These human genetic findings directly link PDE4D to cognitive processing. Furthermore, inhibitors of PDE4D enhance cognition in naïve rodents, Pde4d knockout mice have enhanced learning and memory ability as well as hippocampal synaptic plasticity and knockdown of PDE4D with RNAi improves cognitive performance (Brullo et al., 2016; Bruno et al., 2011; Burgin et al., 2010; Li et al., 2011; Ricciarelli et al., 2017; Schaefer et al., 2012; Sierksma et al., 2014; Wang et al., 2015; Wang et al., 2013; Zhang et al., 2014; Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). The cognitive-enhancing effects of rolipram, a compound that inhibits all four subtypes of PDE4, appears to be mediated primarily through PDE4D since the positive effects on learning and memory were not observed in Pde4d knockout mice (Li et al., 2011). Given the results from our previous studies demonstrating that cAMP signaling is depressed in the chronic recovery phase of TBI and that rolipram improves cognition in TBI animals, we hypothesized that a negative allosteric modulator of PDE4D would reduce learning and memory deficits in animals after TBI (Titus et al., 2013a). In this study, we report that sham, non-injured animals treated with a low dose of the PDE4D negative allosteric modulator D159687 had no significant baseline behavioral changes on contextual and cue fear conditioning, water maze acquisition and retrieval, and spatial working memory. Likewise, hippocampal basal synaptic transmission and LTP in area CA1 were unaltered in sham animals treated with D159687. In contrast, in TBI animals D159687 treatment beginning at 3 months after injury reduced deficits contextual and cue fear conditioning, water maze acquisition and recall, and spatial working memory when administered before training on each task. These improvements in hippocampal-dependent learning and memory were associated with improvements in basal synaptic transmission in area CA1, a rescue of the late-phase of hippocampal LTP and an increase in cAMP levels. However, D159687 treatment did not improve cortical or hippocampal atrophy. Although microglia numbers were increased in the injured cortex at 3 months post-TBI, PDE4D levels were not increased in this cell population. These results suggest that a PDE4D negative allosteric modulator reduces learning and memory deficits induced by TBI and these improvements are associated with a rescue of hippocampal LTP expression.

The lack of an effect of D159687 in sham rats is in contrast to several lines of evidence that PDE4D regulates cognition in naïve rodents during some, but not all learning tasks (Brullo et al., 2016; Bruno et al., 2011; Burgin et al., 2010; Li et al., 2011; Ricciarelli et al., 2017; Schaefer et al., 2012; Sierksma et al., 2014; Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). Differences in the dose, animal species and cognitive tasks may account for the cognitive-enhancing effects observed with some PDE4D inhibitors in naïve mice that were not recapitulated in our study with sham rats (Brullo et al., 2016; Bruno et al., 2011; Burgin et al., 2010; Ricciarelli et al., 2017; Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). In addition, use of the acute treatment with D159687 is not entirely comparable to studies of Pde4d knockout mice, since the mechanism of action is to negatively allosterically inhibit PDE4D transiently versus a complete loss of the enzyme throughout development (Li et al., 2011; Rutten et al., 2008; Schaefer et al., 2012). In naïve rats, D159687 enhanced novel object recognition at doses ranging from 1–1000 μg/kg with oral administration (Burgin et al., 2010). The lack of effects in our sham animals may be due to differences in strain or in the tasks used in these studies. Another possibility is that our tasks were not optimized to observe further enhancement in non-injured animals due to potential ceiling effects.

A caveat of the current study is the lack of a direct assessment that the PDE4D negative allosteric modulator actually reduced PDE4D activity in vivo in TBI animals. D159687 does not bind PDE4D irreversibly, thus D159687 would not remain bound to the enzyme during methods to purify PDE4D from other PDE4 isoforms in brain homogenates when assaying specifically for PDE4D activity in vivo (Clapcote et al., 2007). Therefore, we cannot conclude that D159687 improved cognition by directly acting on PDE4D. Another caveat of this study is the possibility that D159687 may have targeted other PDE4 subtypes. D159687 exhibits selectivity of 200-fold towards PDE4D versus PDE4B in primates. This selectivity is achieved due to a phenylalanine at position in 196 in PDE4D which is a tyrosine in PDE4A, PDE4B and PDE4C (Burgin et al., 2010). The Phe/Tyr polymorphism is not present in rodents, resulting in a lowered selectivity of 10-fold towards PDE4D as compared to PDE4B in mice (Gurney, Burgin, Magnusson, & Stewart, 2011). In a study with rats, D159687 (1 mg/kg, intravenous) reached a C_max of 1 μg/gm in brain (M. Gurney, unpublished data). Extrapolation based on dose linearity suggests that at the dose used in our study, the C_max in the brain may be approximately 333 ng/gm, at an estimated 750 nM. This is 3.7-fold higher than the IC₅₀ for mouse PDE4D7 and 2.7-fold lower than the IC₅₀ for mouse PDE4B3 (Burgin et al., 2010). D159687 was profiled in the standard CEREP panel of 64 common off-targets including receptors, ion channels and transporters and found to have no cross-reactivity with these potential off-targets (M. Gurney, personal communication). D159687 also lacks any inhibitory effects on any of the other PDEs 1–11 (Burgin et al., 2010). Although we can conclude this PDE4D negative allosteric modulator improved cognitive outcome after TBI, whether this effect was through direct actions on PDE4D selectively in TBI animals remains to be established.

Another limitation of this study is that we cannot definitively conclude whether the observed pro-cognitive effects of D159687 were due to effects on neurons or other cell types within the central nervous system. In our previous studies, we found that PDE4D expression was upregulated in the injured parietal cortex and hippocampus acutely after TBI, and in a subset of PDE4D-positive microglia (Oliva et al., 2012; Wilson, Titus, Oliva Jr., Furones, & Atkins, 2016). In the current study, while we found that microglia were significantly increased in the chronically injured brain, only a subset of microglia expressed PDE4D. Therefore, the cognitive benefits of D159687 are unlikely to be due to targeting microglia expressing PDE4D. Previous studies have demonstrated that PDE4 inhibitors such as rolipram increase cerebral blood flow (CBF) through vasodilatory effects (Rutten et al., 2009). After TBI, CBF is reduced which may contribute to chronic learning and memory deficits (Jullienne et al., 2016). Given that PDE4D is expressed in endothelial cells and vascular smooth muscle, D159687 may have reduced cognitive deficits after TBI by increasing CBF (He, Cui, Patterson, & Paule, 2011; Rampersad et al., 2010; Tilley & Maurice, 2005). Future studies investigating the expression of PDE4D isoforms in the cerebral vasculature after TBI, as well as the vasoactive effects of PDE4D-selective inhibitors could provide valuable information on PDE4D as a target to improve CBF and cognition in the injured brain.

Based on our electrophysiology results, one potential mechanism of action of D159687 may have been through CREB signaling within neurons. Consistent with this hypothesis, we found that paired-pulse facilitation and posttetanic potentiation were unaffected by D159687 treatment, whereas expression of late phase LTP was rescued. Similarly, mice with targeted disruption of CREB have intact paired-pulse facilitation and posttetanic potentiation, but the expression phase of LTP decays to baseline 90 min after tetanic stimulation (Bourtchuladze et al., 1994). In further support of this potential mechanism, Pde4d knockout mice have increased levels of phosphorylated CREB in dentate granule cells (Li et al., 2011). Additionally, a recent study by Zhang et al. demonstrated that D159687 treatment increased levels of phosphorylated CREB in a cultured hippocampal neuronal cell line (Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). Our results demonstrated that in naïve animals and in TBI animals at 3 months after injury, D159687 significantly increased cAMP levels. Further studies are needed to definitively identify whether D159687 acts through PDE4D to boost CREB signaling within neurons to improve learning and memory after TBI. Use of Pde4d knockout mice may facilitate this important mechanistic question.

TBI results in a transient increase in neurogenesis followed by a prolonged depression (Atkins et al., 2010; Osier, Carlson, DeSana, & Dixon, 2015; Urrea et al., 2007). Pde4d knockout mice have increased hippocampal neurogenesis (Li et al., 2011). One potential mechanism of action of D159687 in our study may have been increased neurogenesis. However, the learning paradigms used in this study are not sensitive to impairments in neurogenesis (Clelland et al., 2009; Deng, Saxe, Gallina, & Gage, 2009; Jessberger et al., 2009). Refinements of the behavioral testing are needed to clarify how the PDE4D negative allosteric modulator may improve chronic cognitive deficits after TBI.

Future studies are needed to evaluate the optimal dosing administration paradigm. We chose to treat animals 30 min prior to training on each behavior task to test a therapeutic that is given acutely during a learning task and not administered chronically. This is due to previously published studies demonstrating that chronic treatment with the pan-PDE4 inhibitor rolipram for 2 weeks upregulated several PDE4 isoforms (Dlaboga, Hajjhussein, & O’Donnell, 2006). In addition, daily, high doses of rolipram for 3 weeks actually impaired hippocampal-dependent memory formation (Giralt et al., 2011). This dosing paradigm did indicate that a single administration of D159687 was able to rescue the remote memory deficit at 1 month after training in TBI animals despite having no effects on hippocampal or cortical atrophy. Remote (28 days) contextual fear memory remains intact even when the hippocampus is lesioned (Maren & Fanselow, 1997). Previous studies have demonstrated that elevating cAMP-PKA signaling during memory consolidation augments encoding of long-term memories (Barad, Bourtchouladze, Winder, Golan, & Kandel, 1998). Our working model is that D159687 improved encoding by elevating cAMP levels at the time of learning, and this was sufficient to result in long-term memory formation. For clinical translation, determining whether chronic or sub-chronic treatment with D159687 has further beneficial effects on memory or atrophy will be important.

PDE4 inhibitors have been limited in clinical development as cognitive enhancers due to their side effects of nausea and emesis (Robichaud, Savoie, Stamatiou, Tattersall, & Chan, 2001). PDE4D in particular has been implicated in these deleterious side effects since PDE4D is expressed in noradrenergic pathways of area postrema and nucleus of the solitary tract (Cherry & Davis, 1999; Mori et al., 2010). Some, but not all PDE4D inhibitors are emetic in multiple species (Brullo et al., 2016; Burgin et al., 2010; Kalgutkar, Choo, Taylor, & Marfat, 2004; Ricciarelli et al., 2017). However, since there is an intrinsic ceiling on the inhibition of PDE4D, D159687 does not cause emesis and nausea until 30 mg/kg (Burgin et al., 2010; Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). This dose is 100-fold higher than the dose used in this study, indicating that the cognitive-enhancing effects of D159687 are well below the range that results in emetic effects.

Previously, we have found that a PDE4B selective inhibitor, A33, improved cognition during the chronic recovery phase of TBI (Titus et al., 2016). These results together with our current study suggest that selective targeting a particular PDE4 subtype may not be necessary to improve cognition after TBI. These results are in contrast to studies of Pde4b and Pde4d knockout mice where knockout of Pde4d, but not Pde4b, has effects on learning and memory (Li et al., 2011; Siuciak, McCarthy, Chapin, & Martin, 2008). Likewise, a recent evaluation of the PDE4B inhibitor A33 versus the PDE4D inhibitor D159687 indicates that these two PDE4 subtypes have very different roles in modulating behavior (Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). The selective PDE4B inhibitor A33 did not improve novel object recognition memory in naïve mice whereas D159687 was pro-cognitive in this task (Zhang, Xu, Zhang, Gurney, & O’Donnell, 2017). PDE4B is considered to be involved in anxiety and inflammation, whereas PDE4D is involved in memory formation, adult neurogenesis and depression (Li et al., 2011; Schaefer et al., 2012; Siuciak, McCarthy, Chapin, & Martin, 2008; Zhang et al., 2008). In our TBI model, both PDE4B and PDE4D are acutely upregulated and then return to non-injured levels or even decrease slightly (Oliva et al., 2012; Titus et al., 2016; Wilson, Titus, Oliva Jr., Furones, & Atkins, 2016). Given the proposed roles of PDE4B and PDE4D in mediating differential behaviors, it is intriguing that both PDE4B and PDE4D selective inhibitors achieved similar cognitive-enhancing effects in the chronic TBI animals (Titus et al., 2016). One potential explanation for these results may be that the PDE4B inhibitor and the PDE4D negative allosteric modulator target the same molecular pathomechanism(s). TBI results in a dysregulation of cAMP-PKA-CREB signaling (Titus et al., 2013a; Titus, Furones, Kang, & Atkins, 2013b). Perhaps elevation of cAMP levels with any PDE4 subtype-specific inhibitor may be sufficient to rescue this signaling deficit in TBI animals. In accordance with this, Pde4d knockout mice have elevated basal levels of CREB phosphorylation and similarly, we have found that PDE4B inhibition increases phospho-CREB levels in chronic TBI animals (Titus et al., 2016). In support of this interpretation, we also found that the PDE4D negative allosteric modulator increased cAMP levels. Alternatively, although both drugs worked in TBI to improve cognitive deficits, their mechanism of action may have some distinctions. A33 has the additional benefit by being anti-inflammatory and reducing TNF-α levels, but D159687 likely does not have this effect given that TNF-α levels are not altered in Pde4d knockout mice (Jin & Conti, 2002; Jin, Lan, Zoudilova, & Conti, 2005).

4.1. Conclusions

In summary, we report that negative allosteric modulation of PDE4D did not alter cognitive functioning or hippocampal synaptic transmission in non-injured animals, but did significantly reverse the cognitive and synaptic plasticity deficits in TBI animals at 3 months post-injury. PDE4 inhibitors are being actively investigated for clinical usage in the treatment of Alzheimer’s disease, mild cognitive impairment and cognitive impairments after stroke (Hansen & Zhang, 2015). Inhibitors that target all PDE4 inhibitors subtypes result in significant side effects. This study supports the therapeutic potential of targeting PDE4D selectively with a negative allosteric modulator to improve cognitive impairments after TBI.

Highlights.

• A PDE4D negative allosteric modulator did not alter cognition in non-injured rats

• Cognition after TBI is improved with a PDE4D negative allosteric modulator

• A PDE4D negative allosteric modulator improves LTP in TBI animals

Abbreviations

aCSF

artificial cerebrospinal fluid

cAMP

cyclic AMP

CREB

cAMP response-element binding protein

CBF

cerebral blood flow

fEPSP

field excitatory postsynaptic potential

FPI

fluid-percussion injury

HBSS

Hank’s balanced salt solution

H&E

hematoxylin and eosin

I-O

input-output

i.p

intraperitoneally

LTP

long-term potentiation

MABP

mean arterial blood pressure

PBS

phosphate-buffered saline

PDE

phosphodiesterase

PKA

protein kinase A

SEM

standard error of the mean

TBI

traumatic brain injury

UCR

upstream conserved region