Non-coding enhancer RNA regulates NPAS4 in the mPFC to control chronic stress-induced anhedonia-like behavior

Background: Chronic stress can produce reward system deficits ( i.e. anhedonia) and other common symptoms associated with depressive disorders, as well as neural circuit hypofunction in the medial prefrontal cortex (mPFC). However, the molecular mechanisms by which chronic stress promotes depressive-like behavior and hypofrontality remain poorly understood. Methods: C57BL/6 adult male mice were subjected to chronic social defeat stress (CSDS). Neuronal PAS domain-containing protein 4 (NPAS4) or the Npas4 lnc-eRNA were reduced using a viral-mediated shRNA approach. CSDS-induced behaviors, including social avoidance, sucrose preference, natural reward motivation, and anxiety-like behavior, were then measured. CSDS-induced changes in mPFC dendritic spine density were assessed using confocal imaging, and the mPFC NPAS4-regulated transcriptome was assessed using RNA-seq analysis. Results: Social defeat stress induced transient expression of NPAS4 in the mPFC. Viral-mediated knockdown of mPFC NPAS4 blocked CSDS-induced reduction in sucrose preference and changes in natural reward motivation, but without influencing social avoidance. NPAS4 was also required for CSDS-induced reduction of pyramidal neuron dendritic spine density in mPFC. RNA-seq analysis from mPFC tissues revealed that NPAS4 influences expression of numerous genes linked to glutamatergic synapses and ribosomal function, and to genes dysregulated in multiple neuropsychiatric disorders, including depression. Finally, we found that stress-induced expression of NPAS4 in mPFC requires a novel, activity-regulated lnc-eRNA, and that this Npas4 lnc-eRNA in mPFC was required for CSDS-induced anhedonia-like behavior.


Introduction
Stress-related mental disorders continue to be a leading cause of disability and financial burden on society (1).The associated symptom domains of stress-related disorders are diverse and present with a high degree of comorbidity, thus treatment strategies for these disorders represent a major healthcare challenge.The rodent chronic social defeat stress (CSDS) paradigm produces multiple behavioral and neural phenotypes reminiscent of stress-related and depressive disorders in humans (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15), including anhedonia-like behaviors and social avoidance.CSDS produces social avoidance in a subset of mice (often termed the "susceptible" population); whereas the "resilient" subpopulation displays normal social behavior after CSDS (8,12,13).Anhedonia, a core symptom of major depressive disorder (MDD), is associated with deficits in hedonic capacity, reward evaluation, decision-making, and motivation to obtain rewards, as well as risk for suicide and treatment resistance (16,17).However, the neural mechanisms by which chronic stress produces anhedonia remain unclear.Multiple preclinical and clinical studies have revealed reduced function of the medial prefrontal cortex (mPFC), which is caused, at least in part, by stress-induced loss of structural and functional synaptic connections and neural circuits in this region (3,(18)(19)(20)(21).Furthermore, stress-induced hypofrontality is thought to underlie symptoms of MDD (17,(22)(23)(24)(25) and contribute to the neuropathology 4 of treatment-resistant depression (26), including the potential for anhedonia susceptibility (27,28).
In this study, we investigated the role of Neuronal PAS domain Protein 4 (NPAS4), and its eRNA transcribed from the Npas4 enhancer region (~3kb upstream of the TSS), in chronic stress-induced brain and behavior dysfunction.NPAS4 is an immediate early gene (IEG) transcription factor that modulates synaptic connections on excitatory (E) and inhibitory (I) neurons in response to synaptic activity -a proposed homeostatic mechanism to modulate E/I balance in strongly activated neural circuits (37)(38)(39)(40)(41)(42).In the brain, NPAS4 is required in the hippocampus and amygdala for contextual fear learning (43,44), in the visual cortex for social recognition (45), and in the nucleus accumbens (NAc) for cocaine reward-context learning and memory (46).Here, we discovered that acute social stress or CSDS induces NPAS4 expression in the mPFC and that NPAS4 in the mPFC is required for both CSDS-induced anhedonia and reduction of dendritic spine density.Similarly, we found that the Npas4 eRNA was required for CSDS-induced mPFC Npas4 mRNA expression and anhedonia-like behaviors, revealing an essential role for a non-coding eRNA in chronic stress-induced behavioral dysfunction.

Recombinant plasmids and shRNA expression viral vectors
For knockdown of endogenous Npas4 mRNA and eRNA expression in medial prefrontal cortex (mPFC), validated anti-Npas4 shRNA or scramble (SC) shRNA control (Lin et al 2008 andTaniguchi 2017) and anti-Npas4 eRNA shRNA oligonucleotides were cloned into the pAAV-shRNA vector as previously described (46).See the Supplemental Table S3 for sequence of shRNAs.The adeno-associated virus serotype 2 (AAV2) vector consists of a CMV promoter driving eGFP with a SV40 polyadenylation signal, followed downstream by a U6 polymerase III promoter and Npas4 mRNA shRNA, Npas4 eRNA shRNA, or SC shRNA oligonucleotides, then a polymerase III termination signal -all flanked by AAV2 inverted terminal repeats.AAV2-Npas4 mRNA shRNA, eRNA shRNA, and SC shRNA were processed for packaging and purification (UNC Vector Core, NC, and USC Vector Core, SC).
Coordinates to target the mPFC (ventral portion of cingulate, prelimbic, and infralimbic cortices) were +1.85-1.95mm anterior, +0.75 mm lateral, and 2.65 to 2.25 mm ventral from bregma (relative to skull) at a 15 degrees angle in all mice (3).AAV2-scramble (SC) shRNA (2.9*10^9 and 1.1 *10 12 GC/mL), AAV2-Npas4 shRNA (4.3*10^9 and 3.1*10 12 GC/mL), and AAV-eRNA shRNA (2.13*10 13 and 1.49 *10 14 GC/mL) were delivered using 6 Hamilton syringes or nano injectors with pulled glass capillaries at a rate of 0.1 µL/min for 0.4 µL at the dorsovental sites, followed by raising the needle and an additional 0.4 µL delivery of virus.After waiting for an additional 5-10 min, needles were completely retracted.Viral placements were confirmed through immunohistochemistry for bicistronic expression of eGFP from the AAV2 viral vectors by experimenters blinded to the experimental conditions.Animals with off-target virus infection or no infection in either hemisphere were excluded from the analysis of behavioral phenotypes.

Chronic social defeat stress
Chronic social defeat stress (CSDS) was performed as previously described (6,8).CD1 retired male breeders (Charles River Laboratory, CA) were single-housed for 3-5 days before CSDS procedures to establish their territorial cage, then pre-screened for aggressive behavior.Experimental C57BL/6J male mice were introduced to the aggressor's territorial cage, physically contacted and attacked by the aggressor for 5-10 min, and then separated by a clear plastic board with multiple small holes for 24 hours.Experimental mice were introduced to a new CD1 aggressor each day.The no stress control mice were housed with another non-stressed C57BL/6J male mouse, separated by the same plastic board, and the cage partner was changed every day for 10 days of the CSDS experiment.

7
We first examined the expression of Npas4 mRNA in two key corticolimbic regions associated with stress and reward, the mPFC and the nucleus accumbens (NAc), following 11 days of CSDS.We compared CSDS responses to a single social defeat stress experience (acute stress; Figure 1A).In the mPFC, we observed a very rapid and transient induction of Npas4 mRNA in the mPFC (Figure 1B) and NAc (Supplemental Figure S1).We observed a similar response with cFos mRNA, albeit a slower induction and longer duration (Figure 1C).Interestingly, CSDS-induced expression of Npas4 and cFos was observed, but it was reduced compared to the acute stress response (Figure 1B-C), possibly as a consequence of CSDS-induced mPFC hypofunction.In contrast, and in support of this hypothesis, the CSDS-induced attenuation of Npas4 induction was not observed in the NAc (Supplemental Figure S1).Similar to the Npas4 mRNA induction, we observed a significant increase in cells expressing NPAS4 protein at 1 hr following CSDS or acute stress exposure in multiple mPFC regions, including the anterior cingulate and prelimbic cortex subregions (Figure 1D).Interestingly, the vast majority (~70%) of NPAS4+ neurons in the mPFC were CaMKIIa-expressing pyramidal neurons (Figure 1E).Similar to the mRNA, the relative NPAS4 protein expression per cell was highest following acute stress (Supplemental Figure S1B), suggesting that acute stress and CSDS activate a similar number of NPAS4+ mPFC neurons, but that the relative induction of NPAS4 within each cell becomes weaker following repeated psychosocial stress (Supplemental Figure S1B).

NPAS4 in the mPFC is required for CSDS-induced anhedonia-like behavior, but not social avoidance.
To examine the function of NPAS4 in CSDS-induced behaviors (Figure 2A), we employed a neurotropic AAV-mediated RNA-interference approach (AAV2-Npas4 shRNA) to reduce endogenous Npas4 in the mPFC (Figure 2B).Adult male mice (C57BL/6J) received a bilateral injection of AAV2-Npas4 shRNA (Npas4 shRNA PFC ) or AAV2scrambled shRNA control (SC shRNA PFC ).Subsequently, the mice were subjected to 10 days of CSDS or no stress and were then tested for sociability, natural reward preference and motivation, and anxiety-like behavior (Figure 2A).The CSDS-treated SC shRNA PFC and Npas4 shRNA PFC mice showed a significant reduction in the time spent interacting with a novel social target (Figure 2C-D), suggesting that mPFC NPAS4 is not required for CSDS-induced social avoidance.In addition, there was no significant difference between Npas4 shRNA PFC vs. SC shRNA PFC in the relative distribution of CSDS-treated mice that were "resilient" (i.e.>1.0 social interaction ratio) vs. "susceptible" (i.e.<1.0 social interaction ratio) (Figure 2D).However, unlike the CSDS-treated SC shRNA PFC mice, CSDS-treated Npas4 shRNA PFC mice did not develop anhedonia-like behavior measured in the sucrose preference test (Figure 2E), revealing an important role for NPAS4 in stress-induced anhedonia-like behavior.Interestingly, CSDS increased anxiety-like behavior, as measured in the elevated plus maze, in both SC shRNA PFC and Npas4 shRNA PFC mice (Figure 2F), indicating that mPFC NPAS4 function is required for some, but not all, of the behavioral sequelae of CSDS.Moreover, the presence of CSDSinduced social avoidance and anxiety-related behavior in Npas4 shRNA PFC mice argues against the possibility that they are simply less sensitive to stress and/or have deficits in threat-related learning and memory.9 Individuals who suffer from pathological stress often exhibit deficits in motivated, effort-based decision-making (47)(48)(49)(50)(51).However, the molecular mechanisms underlying chronic stress-induced deficits on the motivation behavior has remained unclear.To examine the role of NPAS4 in CSDS-induced changes in reward motivation, Npas4 shRNA PFC or SC shRNA PFC mice were allowed to self-administer sucrose (sucrose SA) under operant conditions.After stable sucrose SA was established, we examined motivation to work for a sucrose reward using the progressive ratio (PR) test (Figure 2A).
Reduction of mPFC Npas4 did not alter the normal acquisition of sucrose SA, including normal operant discrimination learning (nosepokes in the active vs. inactive port) (Supplemental Figure S2A and S2B), but we did detect significant differences in the PR breakpoint -the maximum number of nosepokes an animal was willing to perform to receive a reward.In SC shRNA PFC mice, CSDS produced an increase in the PR breakpoint in "resilient", but not "susceptible," mice (Figure 2G) suggesting that "resilience" was associated with increased reward motivation after chronic stress.In stark contrast, mice lacking Npas4 in the mPFC showed the opposite effect on PR breakpoint (Figure 2G), suggesting that mPFC NPAS4 regulates CSDS-induced changes in reward motivation.Together, these data suggest that NPAS4 in the mPFC controls anhedonialike behavior in chronic stress-susceptible animals.
Unlike SC shRNA control mice, CSDS failed to reduce mPFC dendritic spine density in Npas4 shRNA-expressing neurons (Figure 3A-B, right), suggesting that NPAS4, directly or indirectly, is required for this chronic stress-induced structural synaptic change.We did not observe any changes in mPFC dendritic spine density in Npas4 shRNA-expressing neurons under the "no stress" control condition (Figure 3A-B), indicating that in the absence of CSDS, NPAS4 is not required to maintain steady-state dendritic spine density in adult mPFC pyramidal neurons (Figure 3A-B).Of note, neither Npas4 shRNA nor CSDS produced any detectible changes in mean dendritic spine head diameter (Supplemental Figure S3A)
Additionally, Npas4 shRNA DEGs were significantly enriched in two PsychENCODE modules (Figure 3E and Supplemental Table S2) (59,60).Npas4 shRNA-downregulated DEGs showed significant enrichment within gene module M15, an excitatory neuron module of genes that are associated with ribosome function and upregulated in Autism Spectrum Disorder (ASD) and Bipolar Disorder (BD).Interestingly, RNA-seq from human postmortem brains (BA8/9) of male MDD patients indicated significant enrichment of DEGs in ribosome-related pathways, identifying 69 upregulated and 8 downregulated genes in MDD patients (Figure 3F) (61).Moreover, the majority (64.1%) of the Npas4 shRNA DEGs from our analysis overlapped with upregulated genes in human MDD patients (Figure 3F).Finally, Npas4 shRNA upregulated DEGs showed significant PsychENCODE enrichment in gene module M1, an excitatory neuron module of genes that are associated with glutamate-driven excitability of neurons and downregulated in ASD (59).Together, these data suggest that NPAS4 in the mPFC regulates numerous genes related to glutamatergic synapses and ribosomal function that might contribute to stress-related anhedonia.Moreover, the enrichment of NPAS4-regulated genes in multiple neuropsychiatric disorders, including MDD (61), positions NPAS4 as an important factor in brain health and disease.
12 Npas4 lnc-eRNA regulates Npas4 mRNA expression and anhedonia-like behavior Next, we sought to examine the stress-dependent regulation of Npas4 mRNA in the mPFC.From total RNA-seq data from primary cultured neurons, we detected an eRNA mapping to an activity-regulated Npas4 enhancer region located ~3 kilobase pairs upstream of the Npas4 gene (Figure 4A) (46).This Npas4 enhancer associates with multiple transcription factors, epigenetic enzymes, and histone modifications (Figure 4A), some of which play critical roles in regulating basal and activity-dependent Npas4 gene expression (Supplemental Figure S4A) (46).Moreover, membrane depolarization of cultured primary cortical neurons significantly increased both Npas4 mRNA and eRNA expression (Figure 4B and 4C, respectively).
To examine the functional role of the Npas4 eRNA, we next designed and validated an AAV2-shRNA vector to reduce Npas4 eRNA levels in vitro and in vivo (Figure 4E-H).
Interestingly, the Npas4 eRNA shRNA also significantly decreased Npas4 mRNA expression (Figure 4F), revealing that the Npas4 eRNA is critical for Npas4 mRNA expression.Consistent with this idea, Npas4 eRNA overexpression was sufficient to increase Npas4 mRNA (Figure 4D).Following acute SDS, Npas4 mRNA was significantly increased in mPFC tissues, but this was significantly reduced by the AAV2-eRNA shRNA (Figure 4H), suggesting an important support role for Npas4 eRNA in regulation of Npas4 mRNA expression.In contrast, mPFC AAV2-eRNA shRNA had no effect on the expression of cFos mRNA (Supplemental Figure S4C), suggesting that the function of Npas4 eRNA is specific for Npas4 mRNA expression and not an indirect effect on neuronal activity or general IEG expression mechanisms.Finally, we assessed whether mPFC Npas4 eRNA was required for CSDS-induced anhedonia-like behavior.Following bilateral infusion of AAV2-eRNA shRNA or AAV2-SC shRNA into the mPFC, mice were allowed to recover and then subjected to 10 days of CSDS and depressive-like behavior testing.Similar to knockdown of Npas4 mRNA in mPFC (Figure 2), the AAV-eRNA shRNA blocked CSDS-induced reduction of sucrose preference, without altering sucrose preference in the "no stress" control mice (Figure 4I).These data suggest that Npas4 eRNA is required in mPFC for CSDS-induced expression of Npas4 mRNA, producing anhedonia-like behaviors.

Discussion
Here we find that social defeat stress (acute or chronic) induces rapid and transient expression of NPAS4 in mPFC neurons and that NPAS4 in the mPFC is required for CSDS-induced anhedonia-like behavior, changes in effort-based reward seekingmotivated behavior in CSDS-susceptible animals, and CSDS-induced dendritic spine loss on mPFC pyramidal neurons.However, mPFC NPAS4 was not required for CSDSinduced social avoidance or anxiety-like behavior, suggesting that CSDS produces different depressive-like phenotypes through distinct molecular and/or circuit mechanisms.In addition, we discovered that Npas4 shRNA dysregulated ~700 mPFC genes, including upregulation of genes linked to glutamatergic synapses, suggesting that, following CSDS, NPAS4 might directly or indirectly downregulate these synapse-related genes to enable structural dendritic spine reduction.We also detected strong enrichment of downregulated ribosomal genes, and many of these are also dysregulated in human MDD.Finally, we discovered a novel lnc-RNA, Npas4 eRNA, that is transcribed from an 14 activity-sensitive Npas4 enhancer, is required for normal social defeat stress-induced Npas4 mRNA expression in the mPFC and subsequent CSDS-induced anhedonia-like behavior.Together, our findings reveal a novel and essential role for mPFC NPAS4 in the induction of anhedonia-related behavior and structural synaptic changes produced by CSDS.
NPAS4 is a neuronal-specific, synaptic activity-regulated and experiencedependent transcription factor that regulates excitatory/inhibitory synapse balance and synaptic transmission (37,38,40,41,62), promoting cell type-specific gene programs and cellular responses.Synaptic activity-dependent induction of NPAS4 in pyramidal neurons reduces excitatory synaptic transmission onto these neurons (38) and decreases excitatory synaptic inputs (39), consistent with our finding that NPAS4 is required for CSDS-induced loss of mPFC pyramidal neuron dendritic spine density.While one report indicated that CSDS-induced reduction of dendritic spine density is associated with social avoidance phenotypes (63), we observed that mPFC NPAS4 reduction selectively blocked CSDS-induced spine loss and anhedonia, but not social avoidance and anxietylike behavior, suggesting that deep-layer mPFC pyramidal cell spine loss, per se, is not required for the expression of social and anxiety-related phenotypes in the CSDS model.Interestingly, chronic exposure to stress hormones, chronic mild unpredictable stress, and chronic restraint stress also induce anhedonia-like behavior and dendritic spine loss (20,54,57,(64)(65)(66)(67)(68)(69)(70)(71), and PFC pyramidal cell dendritic spine density is also reduced in human postmortem brains of individuals diagnosed with anhedonia-associated neuropsychiatric disorders, such as SCZ, BD, and MDD (57,64,(72)(73)(74)(75)(76)(77)(78)(79).These data support the possible functional relationship between PFC spine density and anhedonia.NPAS4 in cultured neurons regulates a large, cell type-specific program of gene expression, including key targets like brain-derived neurotrophic factor (BDNF), to alter E/I synapse balance (14,(37)(38)(39)(40)(41)44).However, social defeat stress failed to induce Bdnf mRNA in mPFC and Npas4 knockdown did not alter basal mPFC Bdnf expression (Figure 3C and Supplemental Table S2), suggesting that Bdnf is not a key downstream target of mPFC Npas4 in the context of CSDS.Our RNA-seq analysis of mPFC tissues, with or without Npas4 shRNA, revealed an abundance of significant DEGs (Figure 3C).Of the upregulated DEGs, gene ontology (GO) pathway analysis revealed enrichment of genes linked to glutamatergic synaptic transmission and excitability, and PsychENCODE analysis identified a neuronal module of genes linked to glutamatergic excitability that are downregulated in Autism Spectrum Disorders (59), suggesting the possibility that CSDSinduced mPFC dendritic spine density loss is produced, in part, by one or more of these synapse-linked genes that are downregulated following stress-induced mPFC NPAS4 expression.Interestingly, 22% of these upregulated DEGs overlapped with NPAS4 target genes identified by ChIP-seq analysis from cultured pyramidal neurons (33), suggesting that some of the upregulated, synapse-related DEGs could be direct NPAS4 gene targets.
In contrast to the upregulated genes, downregulated Npas4 shRNA DEGs showed strong enrichment for ribosomal function and a PsychENCODE module (M15) of excitatory neuron genes associated with ribosome function that is upregulated in ASD and BD.
While the functional relevance of ribosome gene enrichment is unclear, the marked enrichment of ribosome-related DEGs is very striking.Microarray analysis of blood samples from stress-vulnerable vs. stress-resistant adults found that DEGs were most markedly enriched in ribosome-related pathways and were upregulated based on stress 16 vulnerability (80).Additionally, RNA-seq analyses from orbitofrontal cortex of postmortem human brains with SCZ, BD, and MDD also identified DEGs enriched for the ribosomal pathway, most of which were upregulated in patient samples (81).Together, these data suggest that NPAS4 regulates expression of numerous mPFC genes, including those linked to ribosome function and glutamatergic synapses, as well as genes dysregulated in patients with anhedonia-associated neuropsychiatric disorders, and that NPAS4 might promote anhedonia and hypofrontality through one or more of these gene targets.
We previously observed that HDAC5 associates with the Npas4 activity-sensitive enhancer region to limit the Npas4 mRNA expression in the NAc (46).Here, we found that a novel lnc-eRNA transcribed from the Npas4 enhancer region in mPFC is required for normal Npas4 mRNA expression and CSDS-induced anhedonia-like behavior.
In the future, it will be important to determine the precise cellular mechanisms by which Npas4 eRNA regulates Npas4 mRNA expression.
Overall, our findings reveal a novel role for mPFC NPAS4 in CSDS-induced anhedonia and pyramidal neuron dendritic spine loss, but not social avoidance or anxietylike behavior.NPAS4 in the mPFC regulates hundreds of genes, including clusters of genes linked to glutamatergic synapse function and ribosomal function, both of which are well-positioned to alter neuronal function.We find that social defeat stress induces expression of mPFC NPAS4 through a process requiring a long non-coding eRNA that is produced from a nearby activity-regulated enhancer, and targeted knockdown of the Npas4 eRNA blocks CSDS-induced anhedonia-like behavior.Therapeutic strategies targeting Npas4 or its regulatory eRNA could be a useful approach for treating PFC hypofrontality and anhedonia-related symptoms in patients struggling with depression, bipolar disorder, and other stress-related neuropsychiatric disorders.S1 for detailed statistical analyses.S1 for detailed statistical analyses.

Figure 1 .
Figure 1.Social defeat stress induces NPAS4 expression in the mPFC

Figure 2 .
Figure 2. NPAS4 in the mPFC is required for CSDS-induced anhedonia-like

Figure 3 .
Figure 3. NPAS4 regulates CSDS-induced reductions in mPFC dendritic spine for detailed statistical analyses.(C) List of top differentially expressed genes in mPFC of Npas4 shRNA PFC mice.(D) Gene ontology analysis of down-and up-regulated DEGs in Npas4 shRNA PFC mice.(E) Npas4 DEG enrichment in gene modules that are dysregulated in neuropsychiatric disorders; Modules M1 and M15.(F) Overlap of DEGs in Npas4 shRNA PFC mice (left; blue) and DEGs in BA8/9 of human MDD patients (right; pink).

Figure 4 .
Figure 4. Npas4 lnc-eRNA regulates Npas4 mRNA and CSDS-induced anhedonia- t r o l A c u t e C h r o n i cStress C o n t r o l A c u t e C h r o n i