Abstract
The mouse olfactory bulb (OB) features continued, activity-dependent integration of adult-born neurons, providing a robust model with which to examine mechanisms of plasticity in the adult brain. We previously reported that local OB interneurons secrete the neuropeptide corticotropin-releasing hormone (CRH) in an activity-dependent manner onto adult-born granule neurons and that local CRH signaling promotes expression of synaptic machinery in the bulb. This effect is mediated via activation of the CRH receptor 1 (CRHR1), which is developmentally regulated during adult-born neuron maturation. CRHR1 is a GS-protein-coupled receptor that activates CREB-dependent transcription in the presence of CRH. Therefore, we hypothesized that locally secreted CRH activates CRHR1 to initiate circuit plasticity programs. To identify such programs, we profiled gene expression changes associated with CRHR1 activity in adult-born neurons of the OB. Here, we show that CRHR1 activity influences expression of the brain-specific Homeobox-containing transcription factor POU Class 6 Homeobox 1 (POU6f1). To elucidate the contributions of POU6f1 toward activity-dependent circuit remodeling, we targeted CRHR1+ neurons in male and female mice for cell-type-specific manipulation of POU6f1 expression. Whereas loss of POU6f1 in CRHR1+ neurons resulted in reduced dendritic complexity and decreased synaptic connectivity, overexpression of POU6f1 in CRHR1+ neurons promoted dendritic outgrowth and branching and influenced synaptic function. Together, these findings suggest that the transcriptional program directed by POU6f1 downstream of local CRH signaling in adult-born neurons influences circuit dynamics in response to activity-dependent peptide signaling in the adult brain.
SIGNIFICANCE STATEMENT Elucidating mechanisms of plasticity in the adult brain is helpful for devising strategies to understand and treat neurodegeneration. Circuit plasticity in the adult mouse olfactory bulb is exemplified by both continued cell integration and synaptogenesis. We previously reported that these processes are influenced by local neuropeptide signaling in an activity-dependent manner. Here, we show that local corticotropin-releasing hormone (CRH) signaling induces dynamic gene expression changes in CRH receptor expressing adult-born neurons, including altered expression of the transcription factor POU6f1. We further show that POU6f1 is necessary for proper dendrite specification and patterning, as well as synapse development and function in adult-born neurons. Together, these findings reveal a novel mechanism by which peptide signaling modulates adult brain circuit plasticity.
Introduction
The adult mammalian brain exhibits remarkable cellular and synaptic plasticity (Alvarez-Buylla and Temple, 1998; Ming and Song, 2005; Zhao et al., 2008). Although basic brain architecture is established at birth, sensory experience continually refines brain circuitry. In the adult rodent olfactory bulb (OB), this is accomplished in part by ongoing, activity-dependent integration of adult-born neurons (Corotto et al., 1994; Najbauer and Leon, 1995; Cummings et al., 1997; Fiske and Brunjes, 2001; Yamaguchi and Mori, 2005; Alonso et al., 2006; Mandairon et al., 2006), dynamic dendritogenesis (Saghatelyan et al., 2005; Yoshihara et al., 2012), new spine formation (Arenkiel et al., 2011), and sensory-dependent synaptic modulation (Gao and Strowbridge, 2009; Cauthron and Stripling, 2014). Therefore, the mouse OB represents a robust model system in which to investigate the mechanisms of neural circuit plasticity in adult brain tissue. Although it is well established that activity affects the integration and synaptogenesis of new neurons in the adult OB, the molecular signaling events that sculpt the bulb's circuitry in response to external stimuli are poorly understood.
We previously discovered that activity-dependent CRH signaling by local interneurons shapes continual refinement of adult OB circuitry. CRH-secreting interneurons reside within the external plexiform layer (EPL) of the bulb. These CRH interneurons are reciprocally connected to mitral cells (MCs) (Huang et al., 2013), which, together with tufted cells, comprise the principal excitatory cells of the OB. When excited by sensory input relayed via mitral/tufted (M/T) cells, EPL interneurons both provide inhibitory feedback onto MCs (Huang et al., 2013; Kato et al., 2013; Miyamichi et al., 2013) and locally secrete CRH (Huang et al., 2013; Garcia et al., 2014). Secreted CRH signals to CRHR1-expressing granule cells (GCs) (Garcia et al., 2014). Interestingly, much of the GC layer in adult OBs is populated by adult-born granule neurons, which migrate to the bulb from their birthplace in the subventricular zone (SVZ). As they undergo survival selection and integration, adult-born neurons upregulate expression of CRHR1. CRHR1 activation by locally secreted CRH drives expression of synaptic proteins in the bulb, promotes integration of adult-born neurons, and marks a developmental window in which sensory learning can broaden and reshape excitatory connections onto developing GCs (Quast et al., 2017). This suggests that signaling by neuropeptides such as CRH exert critical roles in adult brain plasticity. However, the transcriptional programs that govern the functional and morphological changes downstream of activity-dependent neuropeptide signaling remain unclear.
In the present study, we investigated gene expression changes downstream of CRHR1 activity in OB GCs. We found that CRHR1 activity modulated transcriptional cascades implicated in neuronal maturation and/or synaptic development. Of these, we found that CRH signaling dynamically regulates expression of the transcription factor POU Domain Class 6 Transcription Factor 1 (POU6f1). POU6f1 is a member of the POU factor family of transcription factors (Andersen et al., 1993; Okamoto et al., 1993; Bulleit et al., 1994) that have been shown previously to regulate neural development (Latchman, 1999). Despite previous characterization of POU6f1, contributions to peptide signaling and how POU6f1 influences ongoing circuit remodeling through adult neurogenesis remain unknown. Here, we generated two novel mouse models that harbor either an endogenous HA tag on POU6f1, or a conditional knock-out (cKO) allele of POU6f1, to investigate the expression patterns and functions of POU6f1 in CRHR1+ GCs of the OB. To examine the contributions of the POU6f1 transcriptional program toward peptide-dependent circuit refinement, we implemented both loss- and gain-of-function approaches. Using electrophysiological recordings and cell-tracing analysis, we found that POU6f1 modulates excitatory synaptic connections established onto CRHR1+ neurons and influences dendritic branch patterning of CRHR1+ neurons directly. Together, our studies reveal a novel pathway by which neuropeptide signaling activates transcriptional programs to govern cell and synaptic plasticity in the adult brain.
Materials and Methods
Experimental animals.
Experimental animals were treated in compliance with the Department of Health and Human Services and Baylor College of Medicine (BCM) Institutional Animal Care and Use Committee guidelines. Mice were maintained on a 12 h light/dark cycle with access to rodent chow and water ad libitum. CRHR1−/− animals (Smith et al., 1998) were obtained from JAX (https://www.jax.org/strain/004454). The generation and transgene expression patterns of CRHR1-Cre (Garcia et al., 2014) and CRHR1-EGFP (Justice et al., 2008) mice were described previously. Generation of POU6f1 cKO and POU6f1HA/HA mice are described below.
Generation of POU6f1HA/HA CRISPR knock-in mouse.
A DNA template for in vitro transcription of a 20 nt single guide RNA (sgRNA) targeted to the C terminal end of POU6f1 (5′- GAACGTCTTTCAGATCCCGT-3′) was produced by the BCM Mouse ES Cell Core via overlapping oligonucleotides and high-fidelity PCR (Bassett et al., 2013). In vitro transcription of the sgRNA was performed using the MEGA shortscript T7 kit (Thermo Fisher Scientific). The BCM Genetically Engineered Mouse Core microinjected Cas9 mRNA (100 ng/ml; Thermo Fisher Scientific), the sgRNA (20 ng/ml), and the single-stranded oligonucleotide donor DNA carrying the 3xHA peptide tag (5′-ACTGAAGAACACCAGCAAGCTGAACGTCTTTCAGATCCCGtCTAGAGCGGCCGTTTACCCATACGATGTTCCTGACTATGCGGGCTATCCCTATGACGTCCCGGACTATGCAGGATCCTATCCATATGACGTTCCAGATTACGCTCCGGCCGCCCTCGAgTAGGGCTCAGTGTCAGCGTGTGCCGGCGCACTGTACTCTT-3′) with 40 bp arms of homology to the target genomic region (Integrated DNA Technologies; 100 ng/ml), in RNase-free 1× PBS into the cytoplasm of 200 pronuclear (0.5 d postcoitum, dpc) stage C57BL/6J embryos. Injected embryos were then implanted into pseudopregnant ICR females. Offspring were genotyped (n = 43) using the primer set: 5′-GCAGAACCTGATGGAGTTCG-3′ (F) and 5′-AAACCCCCTCCTTCCACAT-3′ (R). One homozygous founder female tested positive for the POU6f1HA fusion-tagged allele and the POU6f1HA/HA homozygous genotype was verified by Sanger sequencing. The founder was bred with a wild-type C57BL/6 male mouse and progeny screened for transmission of POU6f1HA at P14.
Generation of POU6f1 cKO mouse.
To generate POU6f1 knock-out mice, AB2.2 embryonic stem cells (ESCs) derived from the 129 SvEv strain (BCM Mouse ES Cell Core) were electroporated with a custom-designed linearized targeting vector harboring codirectional LoxP sites that flank exons 2–11 of POU6f1. DNA from electroporated ESC lines was digested with EcoNI and analyzed by Southern blot using a radio-labeled 3′ probe (flanking the arms of homology and external to the targeting vector) and by genomic PCR with the primer set: 5′-CATATATACCCGTCGCTTTGGTA-3′ (F) and 5′-AGTTTTTAAGGCCCAGTCTCACT-3′ (R) to identify the 5′ end of the targeting vector. Three independently targeted cell lines were selected and microinjected into the C57BL/6 blastocysts to generate chimeric mice. Male chimeric mice were kept for breeding with female C57BL/6 mice to test germline transmission of the targeted POU6f1 locus. Of the three lines of mice generated from these chimeras, two (from ESC clones 1F11 and 1G11) displayed stable germline transmission. Progeny generated from clone 1G11 were selected for breeding and experimentation. To verify Cre-recombined excision of exons 2–11, we PCR genotyped to detect the recombined band using the primer set: 5′-CATATATACCCGTCGCTTTGGTA-3′ (F) and 5′-ACAAAACTTCTCTCATCCCAAGC-3′ (R).
Microarray experiments.
For control and overactive CRH signaling groups: CRHR1-Cre mice were stereotaxically injected with either AAV-FLEX-EGFP or AAV-FLEX(CA)CRHR1::EGFP (both serotype 2/9), respectively. The FLEX configuration renders expression of the payload dependent upon Cre recombinase (Atasoy et al., 2008). pAAV-(CA)CRHR1::EGFP was described previously (Garcia et al., 2014). Briefly, it is a constitutively active variant of the CRHR1 receptor (Nielsen et al., 2000) that encodes a moiety of the ligand fused to the active region of CRHR1. For the loss-of-CRH signaling group: CRHR1−/− animals were injected with AAV-FLEX-EGFP serotype 2/9. All animals (n = 5 per group) were adult males (6–8 weeks old) and were injected into the core OB with coordinates targeting the GC layer (GCL) (from bregma in mm: AP 3.92, ML 0.9, DV from surface of skull −2.88). Injection parameters were as follows: 20× 25 nl, 30 s apart, flow rate 23 nl/s. Animals were harvested 2 weeks after infection for RNA isolation. Whole OBs (∼24 mg tissue per animal) were dissected and total RNA extracted using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. RNA isolate was treated with DNase (Promega), followed by phenol–chloroform extraction. Purified RNA was quantified using a NanoDrop (NanoDrop Technologies).
Microarray hybridization and analysis.
RNA samples were shipped to Miltenyi Biotec on dry ice for hybridization. RNA samples were quality-checked via the Agilent 2100 Bioanalyzer platform (Agilent Technologies). Then, 100 ng of each total RNA sample was used for the linear T7-based amplification step. RNA samples were amplified and Cy3-labeled using the Agilent Low Input Quick Amp Labeling Kit (Agilent Technologies) according to the manufacturer's protocol. Yields of cRNA and the dye-incorporation rate were measured with the ND-1000 Spectrophotometer (NanoDrop Technologies). Hybridization was performed according to the Agilent Technologies 60-mer oligo microarray processing protocol using the Agilent Gene Expression Hybridization Kit (Agilent Technologies). Fluorescence signals of the hybridized Agilent Technologies microarrays were detected using their microarray scanner system. Agilent Feature Extraction Software was used to read out and process the microarray image files. Statistical analyses were performed by Dr. Zhandong Liu (BCM and Texas Children's Neurological Research Institute). Expression fold changes between groups were considered significant at a false discovery rate of q < 0.05 when unpaired Student's t tests were corrected for multiple comparisons (Storey and Tibshirani, 2003). The datasets have been deposited in NCBI's Gene Expression Omnibus (Edgar et al., 2002) and are accessible through GEO Series accession number GSE98799 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE98799).
RNA in situ hybridization (ISH).
ISH was performed on 25 μm coronal OB sections from fresh-frozen adult CRHR1+/+ or CRHR1−/− P90 male mouse brains. A digoxigenin-labeled mRNA antisense probe was designed against POU6f1 using reverse-transcribed mouse cDNA as a template. The sequence for the custom-designed POU6f1 probe was 5′-TCCTGTCTCNCCCCACCTCCCAAGCACTGGGTCTGCAGTCGCGTGCCACTGTGCCGCTGTGCCTGGCTCTTCTCCACTCCACAGCCTTACTGAGCTTGCTCTCTGGATGATGTTGACATCACGTGATCTACCCTGAAGCAATGACTCTCAAACATGGACATGATGCAGAGACACCAGGGCTCAGCTTTCTCTCTCGTACCCTGTTCCCTTTCCTGGTCCTCTTGGGTCCTGCAGTACTGGTGTGGAGGGACCTCCTAGTAGCTCTGACAATGAGAGGAGGGCAGTTGTGTGCTGATCACACACACACACACACACCACAAGTCTGTCAGGTATAAGGAGACAAGTAGGTTGCTTTCCTGATCCTGCCTCCCAAGAAAGAATAATTCCCAGAGCGATTCTAGGTCATGTGATTATCATCCTGCTTCTGTGAGATGGAAGTCAGGAGGGGTCCTGTAAGAGAGCTTTCAGTAGGGATGCTTAGAGCCCTGAAGCTTATCAAGATGATGCACCTGAGAAAACTATATGCAGATCTGCAGCTTTTCTTGCTCAAAGCCTGCACTCAGGAGGTACTACACAGTCAGCCATGCTGTGGTCAGCCATGCTGTGGTCAGCCATGCTGTGGACATGGGAGAGGATGCCATGCAGTCTATGCATGACAGGCGGGGATAATCATGGCCCACTAAATTAAGCCGACCACTCGTTCTCACCC-3′. ISH was performed and tissue imaged by the RNA In Situ Hybridization Core at BCM using an automated robotic platform as described previously (Yaylaoglu et al., 2005). Slides were imaged and post hoc pseudocoloring for cellular POU6f1 expression was performed using the Celldetekt protocol as described previously (Carson et al., 2005).
Luciferase reporter assay.
The 686 bp POU6f1 candidate promoter element was amplified from a template of purified mouse genomic DNA using a high-fidelity Taq (Roche 11732641001) and the primers 5′-gaaggtgggaaagcgagcta-3′ (F) and 5′-gcccgcagacaaagaagc-3′ (R) harboring KpnI and BamHI linkers, respectively. The amplicon was electrophoresed on a 1% agarose gel and purified using a kit (Qiagen 28704). The 66 bp element within the POU6f1 candidate promoter containing the predicted CREB-binding site (5′-tgacgtcc-3′) was generated by annealing the forward oligo 5′-cacagctgctccggagaccctgagccttagtgacgtccgagttccctctcctttacggcctagagg-3′ with the reverse oligo 5′-cctctaggccgtaaaggagagggaactcggacgtcactaaggctcagggtctccggagcagctgtg-3′, creating a double-stranded oligo, with 5′ KpnI and 3′ HindIII linkers. The 686 bp candidate promoter element and 66 bp CREB binding site-containing element were individually cloned into the pGL4.29-Luc2P reporter vector (Promega E8471) upstream of the firefly luciferase Luc2P gene. The promoter element vector and the CREB binding site-containing element vector were independently cotransfected with (CA)CRHR1, a constitutively active variant of CRHR1 (Nielsen et al., 2000), or an empty vector control into Neuro2A cells (ATCC CCL-131) or MCF-7 cells (ATCC HTB-22, kindly gifted by Dr. Rachel Schiff). Cells were plated to subconfluence in a multiwell plate and transfected with Lipofectamine 2000 reagent (Thermo Fisher Scientific 11668027; for Neuro2A cells) or Lipofectamine LTX reagent (Thermo Fisher Scientific 15338100; for MCF-7 cells) according to manufacturer's protocol. The pRL-TK vector carrying a copy of the Renilla luciferase under control of the weakly expressed HSV thymidine kinase promoter (Promega E2241) was cotransfected in every well as a control for transfection efficiency. A vector containing Luc2P downstream of a promoter sequence that was ablated of all potential CREB binding sites served as the negative control. Forty-eight to 120 h after transfection, cell lysates were obtained. Activity of firefly luciferase (Fluc) and Renilla luciferase (Rluc) were measured for each lysate using a Dual Luciferase Reporter Assay kit (Promega E1910) to generate the single measurement for reporter activity in relative light units (RLUs) in a TD-20/20 luminometer. Comparisons were made by Student's t test or Kruskal–Wallis test as indicated. All experiments were performed in replicate with three to six wells per transfection condition.
Western blot analysis.
Male POU6f1HA/HA mice were deeply anesthetized using isoflurane. Brain tissue was dissected and homogenized in ice-cold RIPA buffer with protease inhibitor (1:200; Roche 11873580001). Total protein from samples equalized by mass were separated on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane. The membrane was blocked in 1× TBS without Tween and with 5% nonfat dry milk for 1 h at room temperature, followed by overnight incubation with 1:1000 mouse anti-HA.11 (Covance MMS-101R-200) and 1:2500 rabbit anti-actin (Sigma-Aldrich A5060) in 0.1% TBST/5% milk. The next day, the membrane was washed with 0.1% TBST and incubated with 1:12500 goat anti-mouse (IRDye 800CW; Li-Cor 926-32210) and 1:12500 goat anti-rabbit secondary (IRDye 680RD, Li-Cor 926-68071) in 0.1% TBST/5% milk for 1 h at room temperature. The membrane was washed before visualization by near infrared imaging (Odyssey CLx; LiCor).
Immunohistochemistry.
Animals were deeply anesthetized with isoflurane, followed by intracardial perfusion of ice-cold 1× PBS, pH 7.35, and 4% PFA/1× PBS. Brains were dissected and postfixed in 4% PFA for 2 h at room temperature before cryoprotection in 30% sucrose at 4°C for 2 d. Brains were then frozen in optimal cutting temperature compound (Thermo Fisher Scientific 4585) on dry ice and OBs coronally sectioned into 1× PBS at 25–30 μm using a cryostat (Leica CM1860). Free-floating OB sections were blocked in 0.3% Triton X-100/10% normal goat serum/1× PBS for 30 min at room temperature on a shaker and stained with rabbit anti-HA (1:250; Cell Signaling Technology C29F4), mouse anti-NeuN (1:500; Millipore MAB377), rabbit anti-GFAP (1:1000; DAKO Z0334), rabbit anti-DCX (1:500; Cell Signaling Technology 4604), or rabbit anti-TBX21 (1:500; kindly gifted by Dr. Sachiko Mitsui, RIKEN Brain Science Institute) in the same blocking solution as above overnight at 4°C. The following day, sections were washed in 1× PBS twice for 15 min at room temperature on a shaker and then stained with (1:500) anti-rabbit Alexa Fluor-546 (Thermo Fisher Scientific) and (1:500) anti-mouse Alexa Fluor-633 (Thermo Fisher Scientific) in blocking solution for 1 h at room temperature on a shaker. Sections were again washed twice in 1× PBS and then mounted in Fluoromount-G (Southern Biotech) on glass slides for imaging on a Leica DM6000B microscope. For quantification of coexpression, serial epifluorescent images from 4 to 7 OB slices per mouse (n = 3 mice) were obtained and examined for coexpression of CRHR1-GFP, POU6f1::HA, or NeuN in the total GCL or the superficial GCL, as specified. For colocalization of CRHR-GFP with POU6f1::HA, CRHR-GFP cells were selected first in a single image visualized in the green fluorescence channel, followed by determination of expression of POU6f1::HA in a superimposed image visualized in the red fluorescence channel. For each analysis, 180–400 cells per animal were randomly selected and counted. Statistical comparison for coexpression of CRHR1 and POU6f1::HA in the total versus superficial GCL was made by Student's t test.
5-Ethynyl-2′-deoxyuridine (EdU) pulse labeling and POU6f1 coexpression or survival analysis.
EdU injection was performed as described previously (Garcia et al., 2014). Briefly, adult male and female POU6f1HA/HA mice received four doses of EdU intraperitoneally (Invitrogen, 50 mg/kg) with each dose injected 2 h apart. Then, 8, 14, 21, 31, and 45 d postpulse (dpp), animals were killed and OBs sliced coronally at 30 μm for staining against EdU (Click-iT EdU Alexa Fluor 647 imaging; Thermo Fisher Scientific C10340) and POU6f1::HA (using the procedure described above). Analysis of adult-born neuron survival in the OBs of CRHR1-Cre; POU6f1ΔExon2-11/ΔExon2-11 cKO animals was performed by pulse labeling adult-born neurons in > 4 wk old male and female mice, and sacrificing animals at 31–32 dpp for OB sectioning and staining against EdU. Comparisons for survival rates were made between cKO tissue to that obtained from POU6f1floxed Exon2-11/floxed Exon2-11 littermates that did not inherit the CRHR1-Cre transgene.
Electrophysiology.
For all groups, both male and female mice >4 wks of age were used. For the POU6f1 loss-of-function group (n = 12): CRHR1-Cre; POU6f1ΔExon2-11/ΔExon2-11 cKO mice were stereotaxically injected with AAV-FLEX-EGFP or AAV-FLEX-mRuby2 (both serotype DJ/8) to label CRHR1+ cells lacking POU6f1 for recording. For the control (n = 7 mice) and gain-of-function (n = 8 mice) groups: CRHR1-Cre mice were injected, respectively, with either AAV-FLEX-EGFP (or AAV-FLEX-mRuby2) or with AAV-FLEX-GFP::POU6f1 (all serotype DJ/8). All mice were injected into the OB core to target the GCL (coordinates from bregma in mm: AP 3.7–3.92, ML 0.65–0.9, DV from pia −0.8–1.1). Injection parameters per OB were as follows: 10 × 69 nl, 10 s apart, flow rate at 69 nl/s. Animals were killed >14 d after infection and before animals reached 12 weeks of age.
Slice preparation and recordings.
Animals were deeply anesthetized with isoflurane and perfused intracardially with ice-cold artificial CSF (ACSF) containing the following (in mm): 122 NaCl, 3 KCl, 1.2 NaH2PO4, 26 NaHCO3, 20 glucose, 2 CaCl2, and 1 MgCl2 at 305–310 mOsm, pH 7.3. Brains were dissected, embedded in low-melting point agarose, sectioned to 300 μm on a microtome (VT1200; Leica) and placed in ice-cold oxygenated (5% CO2, 95% O2) dissection buffer containing the following (in mm): 87 NaCl, 2.5 KCl, 1.6 NaH2PO4, 25 NaHCO3, 75 sucrose, 10 glucose, 1.3 ascorbic acid, 0.5 CaCl2, and 7 MgCl2. Sections were recovered (at least 30 min at 37°C) in oxygenated ACSF and then acclimated at room temperature for 10 min before recordings. After recovery, slices were placed in a recording chamber on a fixed stage of an upright microscope (SliceScope Pro 6000 platform; Scientifica) and perfused with oxygenated ACSF at room temperature. Labeled cells were identified by transmitted light (DIC) and fluorescent imaging with a CoolLED pE-100 470 nm (EGFP) or 565 nm (RFP) excitation light source, 49002- ET-EGFP (FITC/Cy2) or 41043-HcRed1 emission filters (Chroma Technology) and an optiMOS camera (QImaging) controlled by Micro-Manager version 1.4.22 software (University of California–San Francisco). Whole-cell voltage-clamp recordings were performed using a Multiclamp 700B amplifier (Axon CNS; Molecular Devices) at room temperature (25–27°C). Electrodes were prepared from borosilicate glass electrodes (outer diameter 1.5 mm) by a micropipette puller (Sutter Instruments), pulled to tip resistances between 4 and 7 MΩ, and filled with internal solution containing the following (in mm): 110 CsMeSO3, 0.2 EGTA, 4 NaCl, 30 HEPES, 2 Mg-ATP, 0.3 Na-GTP, and 14 creatine phosphate, 300–310 mOsm, pH 7.3. Miniature excitatory postsynaptic events were recorded in oxygenated ASCF with the addition of 1 μm TTX and 20 μm gabazine. Neurons were held at −80 mV throughout the experiment. The recorded current was digitized at 20 kHz using a Digidata 1440A (Molecular Devices) and Clampex version 10.6.2.2 software (Molecular Devices) and stored on a PC hard disk. Offline analysis included filtering at 1 kHz and was performed using MiniAnalysis (Synaptosoft) and Clampfit version 10.0.0.3 software (Molecular Devices). Statistical comparisons were made by one-way ANOVA or Kruskal–Wallis test or by a two-sample Kolmogorov–Smirnov test, as indicated. Post hoc tests for ANOVA analysis included examinations for linear trends among the loss-of-function, control, and gain-of-function groups, Dunnett's multiple-comparisons test for ANOVA tests, and Dunn's multiple-comparisons test for Kruskal–Wallis tests. To test whether the proportion of total observed mEPSC decay time constants <2.5 ms or, separately, >4 ms changed for POU6f1-overexpressing neurons, we performed a χ2 test (without Yates' correction) for deviations of these proportions from expected frequencies determined by recordings from control neurons.
Cell morphology analyses.
Animal groups representing loss- and gain-of-function and control groups were established as above in electrophysiology experiments. Adult male and female mice (n = 3 per group) were injected using the same parameters as outlined above into the core of OBs with low-titer viruses to achieve single-cell labeling (for loss-of-function: AAV-FLEX-tdTomato, for gain-of-function: AAV-FLEX-GFP-2A-POU6f1, for control: AAV-FLEX-tdTomato). Animals were killed >14 d after infection.
Brain slice preparation and imaging.
Animals were deeply anesthetized and killed using isofluorane, followed by intracardial perfusion of ice-cold 1× PBS, pH 7.35, and 4% PFA/1× PBS. Brains were dissected, postfixed in 4% PFA for 2 h at room temperature, and then transferred to ice-cold 1× PBS. OBs were mounted on a sectioning post in 2–3% low-melting point agarose and sectioned horizontally into 120–150 μm slices on a vibratome. If bulbs were injected with a GFP-containing virus, then the slices obtained were blocked with 0.3% Triton X-100/10% normal goat serum/1× PBS and then stained overnight at 4°C with rabbit anti-GFP (1:500; Synaptic Systems) and subsequently with (1:500) anti-rabbit Alexa Fluor-488 (Thermo Fisher Scientific). Images of labeled cells were obtained with a Leica TCS SPE confocal microscope with a 20× objective using a digital zoom of 1.25×. Z-stacks were collected at 1.01 μm intervals, covering the entire cell and all dendritic arbors, and images were rendered to Neurolucida 360 (MBF Bioscience) for tracing. Analysis was performed using Neurolucida Explorer (MBF Bioscience) by an experimenter blinded to genotype. Measurements for dendritic lengths, branch order, and branch number were compared by ordinary one-way ANOVA with post hoc tests for linear trend and Dunnett's multiple-comparisons test. Sholl analyses were compared using two-way ANOVA with post hoc Bonferroni's multiple-comparisons test.
Olfactory behavior.
Adult male and female mice cKO (CRHR1-Cre; POU6f1ΔExon2-11/ΔExon2-11) or control (POU6f1floxed Exon2-11/floxed Exon2-11) were tested for Go/No-Go olfactory associative task learning (n = 9 control and n = 12 cKO mice) and short-term olfactory memory (n = 8 per group) deficits using described previously approaches (Garcia et al., 2014). Briefly, for Go/No-Go olfactory task learning, mice were water restricted 1 d before training. Mice were trained in an odorant delivery chamber equipped with infrared beam-break nose-poke ports (Med Associates). Using a stepwise paradigm to shape behavior, mice were trained to respond to the Go odor (hexanol) stimulus by seeking a water reward and to respond to the No-Go odor (butyric acid) stimulus by refraining from seeking the water reward. Rates of correct responses were tracked by blocks of 200 completed trials during the learning acquisition phase and performance between groups assessed by two-way ANOVA.
For short-term olfactory memory assays, animals were acclimated to a mock test chamber for 30 min before being individually placed inside the experimental odor chamber by an experimenter blinded to genotype. The chamber featured two odorant ports on opposite sides, each loaded with either a cotton tip applicator soaked in 100 μl of a test odorant diluted to 1% volume/volume in mineral oil or in 100 μl of pure mineral oil. The following odors comprised the odor test panel: iso-amyl acetate, carvone, eugenol, and anisole. At each port, IR beam break sensors detected and recorded the number and length of nose pokes to quantify olfactory investigation of novel odors or mineral oil. Each mouse was tested at time 0 (initial exposure to novel odor), 30, 60, 90, and 120 min for 10 min per trial. One odorant per day was tested for each mouse, with one time interval (30, 60, 90, or 120 min after initial presentation) observed, for a total of 4 odorants tested per mouse over the course of 4 d. Data were transmitted to an Arduino Uno board and a custom Arduino script was written to quantify the number of nose pokes per animal per port. Negative pressure was applied on both sides of the box to prevent passive diffusion of odors throughout the testing phase.
Experimental design and statistical analysis.
All means are reported as means ± SE unless otherwise specified. Details including statistical analyses specific to each experiment are listed in appropriate subsections in Materials and Methods. A minimum number of animals to be used toward each experiment was determined by previous findings on the electrophysiological and morphological effects of altered CRH signaling on adult-born neurons in the OB (Garcia et al., 2014) and power analysis based on α = 0.05 at 80% statistical power. Data were analyzed using Prism 6 (GraphPad Software) or Excel (Microsoft). For relevant datasets, before analysis, statistical outliers were identified by either Grubb's test for a single outlier (two tailed at p < 0.05) for n < 25 or Rosner's extreme studentized deviate test for multiple outliers (two tailed at p < 0.05) with a maximum number of outliers of 2 for n ≥ 25 datasets, using internet-based software (http://contchart.com/outliers.aspx) and removed before comparisons. This procedure identified outliers each in the following datasets and groups for electrophysiology experiments: mEPSC frequency, control (×2) and overexpression (×2); mEPSC amplitude, cKO (×2); membrane resistance, cKO (×2), and control (×1). Removal of outliers did not change conclusions for mEPSC frequency or membrane resistance but did change the conclusion for mEPSC amplitudes (with two outliers in cKO group retained: Kruskal–Wallis test, p = 0.1888 vs with two outliers removed: one-way ANOVA, p = 0.0481). No outliers were detected for morphological datasets. Statistical comparisons between control and loss- or gain-of-function groups were made by ordinary one-way ANOVA or Kruskal–Wallis test as appropriate, followed by Dunnett's or Dunn's multiple-comparisons tests, respectively, as specified in the Results. For normally distributed electrophysiological and morphological data, we performed an additional post hoc test for linear trend among the groups for each dependent variable, when the groups were ordered by increasing levels of the independent variable (POU6f1 expression in CRHR1+ GCs): CRHR1-Cre;POU6f1ΔExon2-11/ΔExon2-11;AAV-FLEX-GFP (cKO) representing low expression of POU6f1, CRHR1-Cre;AAV-FLEX-GFP (Control) representing endogenous expression of POU6f1, and CRHR1-Cre;AAV-FLEX-GFP::POU6f1 (POU6f1) representing high expression of POU6f1.
Results
CRH signaling in the bulb is associated with gene expression changes
We showed previously that local CRH signaling functions as an activity-dependent cue for morphological and functional development of adult-born GCs in the OB (Garcia et al., 2014). As adult-born neurons mature into CRHR1-expressing GCs, olfactory learning influences their connectivity (Quast et al., 2017). Therefore, CRHR1 expression and synaptic input mark a developmental window for sensory-dependent refinement in the olfactory circuit. To determine how CRH signaling regulates the maturation of adult-born granule neurons mechanistically in the bulb, we queried the downstream transcriptional targets of altered levels of CRH signaling. Due to technical limitations of controlling expression of the small, secreted neuropeptide CRH, we established animal models of loss- and gain-of-function using genetically modified variants of the CRHR1 receptor (Fig. 1A, top).
Figure 1-1
We modeled loss of local CRH signaling by using a knock-out CRHR1−/− mouse line in which GCs of the OB lack functional CRHR1 expression (Smith et al., 1998). To model endogenous or increased levels of CRH signaling, we used CRHR1-Cre transgenic animals (Garcia et al., 2014). CRHR1-Cre animals were stereotaxically injected with conditional adeno-associated viruses (FLEX) AAV (Atasoy et al., 2008) to target CRHR1+ neurons selectively with either AAV-FLEX-EGFP (modeling endogenous levels of CRH signaling) or a FLEXed constitutively active (CA) version of CRHR1 fused to EGFP [AAV-FLEX-(CA)CRHR1::EGFP; Nielsen et al., 2000; modeling overactive CRH signaling]. To control for the effects of viral infection on gene expression analysis, we also injected AAV-FLEX-EGFP into the bulbs of CRHR1−/− animals. All AAVs were injected into OBs using coordinates to target the GCL. Two weeks after injection, total bulb RNA was extracted from all 3 animal groups (n = 5 per group) and microarray hybridization was performed using Agilent Whole Mouse Genome Oligo Microarrays (8x60K, Design ID 028005).
To query the potential transcriptional targets of local CRH-dependent circuit refinement, we examined differential transcript expression in conditions of altered CRHR1 activity. Changes in transcript expression were considered significant when unpaired Student's t tests, corrected for multiple comparisons, had a false discovery rate (q) < 0.05 (Storey and Tibshirani, 2003). Of the transcripts for which expression patterns changed significantly in response to altered CRH signaling, we narrowed candidates for further study to those with reciprocal expression changes between loss- and gain-of-function animal models. This filter criterion focused our study on changes that were more likely to be in direct response to changes in CRH signaling and yielded a total of 57 genes with expression that changed dynamically and reciprocally when CRHR1 activity was altered. Interestingly, one of the most common cellular processes affected by reciprocally expressed transcripts was gene regulation (Fig. 1-1). This suggested that one major contribution of local CRH signaling in the bulb is to induce action of an array of gene regulators to direct adult-born neuron maturation and integration systematically.
Because we were interested in identifying the most upstream genetic changes related to CRH signaling in the bulb, we further focused our study toward genes implicated in direct transcriptional regulation. Of the transcription factors identified from our filter criteria, we observed the largest reciprocal expression changes for POU6f1. Relative to expression in CRHR1−/− bulbs, POU6f1 expression was 5.28-fold higher in CRHR1-Cre;AAV-FLEX-EGFP control bulbs (n = 5 animals per group, q = 0.04) and 6.82-fold higher in (CA)CRHR1 overactivation bulbs (q = 0.0076; Fig. 1B,C). This suggested that POU6f1 transcription mirrors CRHR1 expression and activity in the adult OB. POU6f1 is a member of the POU factor family of transcription factors (Andersen et al., 1993; Okamoto et al., 1993), which regulate expression of developmental genes in the nervous system (Latchman, 1999). Expression of POU6f1 is noted widely throughout the embryonic CNS (Cui and Bulleit, 1998) and adult CNS (Lein et al., 2007), yet its functions in both peptide signaling mediation and synaptic and circuit development in the adult brain have not been elucidated.
To verify the relationship of CRHR1 activity to POU6f1 in the OB, we performed RNA in situ hybridization using a custom-designed probe against POU6f1 in OB sections of CRHR1+/+ or CRHR1−/− animals (Fig. 1D). In CRHR1−/− bulbs, lower POU6f1 signal intensities, as determined by post hoc pseudocolor cell expression analysis (Fig. 1E), suggested that POU6f1 expression declines in the absence of functional CRHR1 expression. Therefore, we next sought to investigate whether CRHR1 activity promotes transcriptional activation of POU6f1 directly.
CRH signaling promotes transcriptional activation of POU6f1
CRH-to-CRHR1 binding drives GS-coupled signaling (Blank et al., 2003; Berger et al., 2006) and leads to cAMP-dependent gene transcription via phospho-activated cAMP-responsive element binding (CREB) protein (Giguère et al., 1982; Aguilera et al., 1983; Gonzalez and Montminy, 1989; Thiel and Cibelli, 1999) in the receptor-expressing cell. Interestingly, cAMP signaling has been shown to activate transcription of POU family factors related to POU6f1 (Wang et al., 2009; Monuki et al., 1989) and the promoter region of POU6f1 [Ensembl release 87] (Zerbino et al., 2015) harbors several potential CREB-binding sites, suggesting that POU6f1 may be a direct target of cAMP signaling. The sequence 5′-TGACGTCC- 3′, located proximally upstream of the transcription start site of POU6f1, differs from the consensus CREB-binding site 5′-TGACGTCA-3′ (Montminy et al., 1986) by only the terminal base pair and thus functions as a potential CREB binding site.
To test whether CRHR1 activity regulates POU6f1 transcription via a CREB-dependent mechanism, we cloned two separate vectors that contained elements from the regulatory region of POU6f1 upstream of a luciferase enzyme. One vector comprised a 686 bp region at the 5′ end of the POU6f1 gene, including the 564 bp immediately upstream of exon 1 and 122 bp of the 5′ segment of exon 1 (Fig. 2A, “promoter”). The second vector contained only the 66 bp region that immediately flanks the predicted CREB-binding site within the promoter region of POU6f1 (Fig. 2A, “CREB site”). The “CREB site” reporter was comprised of the predicted CREB-binding site (8 bp) flanked by a total of 58 bp of genomic sequence unique to POU6f1 (Kent, 2002). Both reporter vectors were independently cotransfected into Neuro2A or MCF-7 cells, each with either a construct encoding the constitutively active variant of CRHR1 [(CA)CRHR1; Nielsen et al., 2000] or an empty control construct to test whether CRHR1-mediated cAMP signaling in cells altered activity of the regulatory components of POU6f1. To verify that observed changes in reporter activity in the presence of (CA)CRHR1 were mediated via CREB binding, we also tested a negative control vector harboring the luciferase reporter downstream of a minimal promoter, in which all predicted CREB-binding sites had been mutated. Indeed, when this negative control vector was cotransfected with (CA)CRHR1 into cells, no significant change in the level of reporter activity was observed relative to cotransfection of the negative control with an empty plasmid (unpaired Student's t test, t(4) = −1.74, p = 0.157; Fig. 2B, left two solid gray bars). These data suggest that activity of POU6f1-regulatory elements induced by (CA)CRHR1 was dependent on CREB binding.
Interestingly, transfection of the “promoter” vector into cells resulted in an increase in reporter activity versus the negative control vector (Fig. 2B; negative control + empty plasmid: 4.33 ± 0.33 relative light units or RLU versus promoter vector + empty plasmid: 23.67 ± 5.67 RLU, unpaired Student's t test, t(4) = −3.41, p = 0.027, measurements on lysates obtained from 3 independent wells per transfection condition, means reported as means ± SE). This indicated that the POU6f1 promoter showed basal activity in a neural cell line, consistent with the predicted role of POU6f1 in developing brain cells (Bulleit et al., 1994). We further observed increased reporter activity when (CA)CRHR1 was cotransfected with the “promoter” vector, but not with the negative control [Fig. 2B; negative control vector + (CA)CRHR1: 8 ± 2.08 RLU vs promoter vector + (CA)CRHR1: 37 ± 6.11 RLU, unpaired Student's t test, t(4) = −4.49, p = 0.011], providing evidence of a synergistic interaction between CRHR1 activity and the POU6f1 candidate promoter sequence in a neural cell line.
Unexpectedly, when the “promoter” vector was cotransfected with (CA)CRHR1, reporter activity was not significantly increased over basal activity of the “promoter” alone [Fig. 2B, right two bars; promoter vector + empty plasmid: 23.67 ± 5.67 RLUs vs promoter vector + (CA)CRHR1: 37 ± 6.11 RLU, unpaired Student's t test, t(4) = −1.6, p = 0.18]. We reasoned that this resulted from either constitutively high basal promoter activity and/or the presence of other regulatory components within the promoter element that could mute the effect of (CA)CRHR1-associated activation.
To address whether reducing basal promoter activity in cells could clarify the relationship of (CA)CRHR1 to activity of the candidate POU6f1 promoter sequence, we identified the MCF-7 cell line (Soule et al., 1973), which has been reported to have low POU6f1 expression as determined by RNA sequencing (Uhlén et al., 2015). MCF-7 cells were cotransfected with either the “promoter” vector plus (CA)CRHR1 or empty vector control. Data from these experiments revealed a direct activation of the POU6f1 promoter by constitutive CRHR1 activity in a cell line with low baseline POU6f1 expression [Fig. 2C; promoter vector + empty plasmid: 10.1 ± 0.64 RLU versus promoter vector + (CA)CRHR1: 15.4 ± 0.45 RLU, Kruskal–Wallis test, H(49) = 14.593, p = 0.00013].
To address the possible presence of inhibitory components within the candidate POU6f1 promoter that could lessen the effects of (CA)CRHR1, we cotransfected (CA)CRHR1 or an empty vector control with the “CREB site” reporter vector (Fig. 2A) containing only 66 bp of the predicted POU6f1 promoter (including the 8 bp CREB binding site) into the Neuro2A cell line. Because the “CREB site” reporter vector contained only a minimal segment of the predicted POU6f1 promoter, we anticipated that when the “CREB site” vector was cotransfected with (CA)CRHR1, reporter activity would rise. Indeed, reporter activity increased significantly when the “CREB site” reporter plasmid was cotransfected with (CA)CRHR1 [CREB site reporter vector + empty plasmid: 3.33 ± 0.33 RLU vs CREB site reporter vector cotransfected with (CA)CRHR1: 12.33 ± 2.33 RLU, unpaired Student's t test, t(4) = −3.82, p = 0.019; Fig. 2D]. Together, these data suggest that activation of CRHR1 can initiate CREB-dependent transcriptional activation of POU6f1 and that POU6f1 is a CRHR1/cAMP-inducible transcription factor the expression of which in a developing neural cell can be mediated by external stimuli.
POU6f1 is expressed in CRHR1+ GCs in vivo
By microarray and in situ hybridization analyses (Fig. 1), we inferred a relationship between CRHR1 activity and POU6f1 expression in the adult OB. Consistent with these data, in vitro luciferase reporter assays showed regulation of POU6f1 expression by CRHR1 activation (Fig. 2). We next sought to reveal the relationship between POU6f1 and CRHR1 expression in intact brain tissue. To avoid nonspecific labeling of POU6f1 from commercial antibodies (data not shown), we pursued a strategy of endogenously labeling POU6f1 by epitope tagging. We used a CRISPR-based gene-targeting strategy (Cong et al., 2013) to generate a novel mouse line that harbors a multimeric hemagglutinin tag (3xHA) at the C-terminal end of POU6f1. For this, a 20 nt sgRNA and 200 nt single-stranded oligo donor template were designed to target endogenous POU6f1 upstream of the native stop codon, creating a cut site for the insertion of the 3xHA sequence (Fig. 3A). Two hundred mouse embryos (0.5 dpc) were used for pronuclear injection of these components, along with wild-type Cas9 mRNA. Injected embryos were then implanted into pseudopregnant female mice. PCR-based genotyping and sequencing revealed that 1 of 43 (2.3%) live-born pups carried the correctly targeted, full sequence of 3xHA tagged POU6f1 (Fig. 3B,C). This mouse was used as a founder to generate a colony of POU6f1HA/HA homozygote animals. Translation of the tagged-variant of POU6f1 in the brains of POU6f1HA/HA animals was verified by Western blot, which detected the 3xHA peptide at the predicted molecular weight for 3xHA-tagged POU6f1 protein in the OB, cortex, and cerebellum of POU6f1HA/HA adult mouse brains (Fig. 3D).
To determine whether POU6f1 is expressed in CRHR1+ GCs of the adult OB, we crossed the POU6f1HA/HA mouse line with a CRHR1-EGFP reporter mouse line (Justice et al., 2008) and stained OB sections obtained from CRHR1-EGFP; POU6f1HA/HA mice for HA antigen. Colocalization analysis of CRHR1+ cells selected randomly from the total GCL showed that 50.25 ± 7.4% of CRHR1+ cells are POU6f1::HA+ (n = 3 mice, 4–7 brain slices per mouse, n = 400 cells total counted, means reported as ± 95% confidence interval), suggesting that CRHR1 activation in vivo may drive expression of POU6f1 in only a subset of CRHR1-expressing GCs.
POU6f1 has been hypothesized to be involved in maturation of developing neurons in culture (Bulleit et al., 1994). Therefore, we reasoned that CRHR1+POU6f1::HA+ cells may represent olfactory GCs in later stages of development than CRHR1+POU6f1::HA− cells. To determine whether expression of POU6f1 is associated with a specific developmental stage, we costained GCs for POU6f1::HA and NeuN, a marker for mature neurons (Mullen et al., 1992). We found that 96.25 ± 3.4% of POU6f1::HA+ cells are NeuN+ and 94.38 ± 3.6% of NeuN+ cells are POU6f1::HA+, suggesting that POU6f1 is correlated with maturity in neurons. Further, to examine how POU6f1::HA expression changed through development of adult-born neurons, we pulse-labeled adult-born neurons with the thymidine analog EdU and costained for EdU and POU6f1::HA from birth through integration (Fig. 3E,F). The proportion of EdU-labeled adult-born neurons that expressed POU6f1::HA was very low on the day of pulsing (1.67 ± 1.67% at 0 dpp) and increased progressively over time (16.79 ± 4.39% at 8 dpp, 31.19 ± 4.52% at 14 dpp, 58.33 ± 1.67% at 21 dpp, 68.33 ± 1.67% at 31 dpp, and 72.81 ± 2.19% at 45 dpp). This steady increase in POU6f1::HA expression by adult-born GCs over time as GCs migrate functionally mature inside the bulb supported the notion that POU6f1::HA expression is correlated with neuronal maturity.
Interestingly, the localization patterns of CRHR1+POU6f1::HA+ versus CRHR1+POU6f1::HA− GCs within the GCL suggested that POU6f1 may alter other aspects of neuronal fate and/or maturation in CRHR1-expressing neurons. We observed that the majority of double-labeled GCs localized to the superficial regions of the GCL (Fig. 3G). Others have shown that the superficial GCL is enriched for long-surviving (Lemasson et al., 2005; Imayoshi et al., 2008) granule neurons with abundant synaptic connections (Orona et al., 1983, 1984). Superficial GCs also display distinct electrophysiological profiles relative to GCs deeper within the core of the bulb (Carleton et al., 2003). We therefore reasoned that CRHR1+POU6f1::HA+ cells may be earmarked for a functional fate distinct from that of CRHR1+POU6f1::HA− cells. In consideration of this, we investigated whether POU6f1::HA expression in CRHR1+ neurons was increased in the superficial GCL relative to CRHR1+ neurons selected at random (from the “total” GCL). Whereas approximately half of randomly selected CRHR1+ cells in the total GCL coexpressed POU6f1::HA, that ratio increased significantly in the superficial GCL, where abundant synapses are found (50.25 ± 7.4% colocalization in the total GCL vs 85.56 ± 7.3% in the superficial GCL, Fig. 3H, right, I; unpaired Student's t test, t(36) = 7.106, p < 0.0001, n = 3 mice, 5–7 brain slices per mouse, n = 180 cells total counted). Enrichment of CRHR1+POU6f1::HA+ colabeling in synapse-rich regions of the GCL supported the possibility that CRHR1+POU6f1::HA+ cells may be functionally distinct from those that express CRHR1 alone. We note that the above results do not distinguish whether POU6f1 could broadly promote neuronal maturation or specifically promote of a specific functional fate in CRHR1+ GCs. In either of these cases, however, we predicted that POU6f1 expression would affect the synaptic connectivity of adult-born CRHR1+ GCs.
POU6f1 influences synaptic formation and connectivity in CRHR1+ neurons
Having established that CRHR1+POU6f1::HA+ neurons selectively populate OB regions enriched for the formation (Orona et al., 1983, 1984) and long-term maintenance of synapses (Lemasson et al., 2005; Imayoshi et al., 2008), we next sought to determine the effects of altered POU6f1 expression on synaptic connectivity. For this, we generated models of both loss- and gain-of-POU6f1-function, again using CRHR1-Cre transgenic mice (Garcia et al., 2014) to target CRHR1+ neurons for POU6f1 knock-out or overexpression, respectively. Target cells in the bulb were verified independently by crossing CRHR1-Cre mice to a ROSALSL-tdTomato reporter line, which revealed the vast majority of fluorescent reporter expression in the bulb to be within GCs (Fig. 4A, right). Scattered periglomerular cells were also labeled, consistent with the small population of adult-born neurons that adopts this cell fate in the OB (Luskin, 1993; Ming and Song, 2005). To determine further the specificity of CRHR1 expression by GCs, we examined coexpression of CRHR1 and GFAP, a marker of glial cells (Fig. 4B, top), or TBX21, a marker of M/T cells (Fig. 4B, bottom), in OB sections prepared from CRHR1-EGFP reporter mice (Justice et al., 2008). These data affirmed that the primary cell population targeted by our manipulations in CRHR1-Cre mice was granule neurons of the OB.
For loss-of-function studies, we crossed CRHR1-Cre transgenic mice to a novel POU6f1 cKO-ready (POU6f1floxed Exon2-11) mouse line (Fig. 4C) to generate CRHR1-Cre; POU6f1ΔExon2-11/ΔExon2-11 cKO animals. cKO mice harbored a recombined floxed POU6f1 allele excised of exons 2–11, including the DNA-binding domain (Wey et al., 1994), in CRHR1+ cells (Fig. 4D).
For gain-of-function studies, CRHR1-Cre animal OBs were stereotaxically injected with AAV carrying a Cre-dependent copy of POU6f1 cDNA fused to a GFP marker (AAV-FLEX-GFP::POU6f1) or GFP alone (AAV-FLEX-GFP) as a control (Fig. 5A). CRHR1 expression in adult-born GCs is detected at approximately day 14 after birth in the SVZ (Garcia et al., 2014), which coincides with activity-dependent selection (Yamaguchi and Mori, 2005; Mouret et al., 2008; Kelsch et al., 2009). Therefore, our manipulations specified either cell-type-specific cKO or overexpression of POU6f1 at this period in development for adult-born neurons.
To determine whether and how POU6f1 expression altered synaptic inputs onto CRHR1+ neurons, we performed whole-cell voltage-clamp recordings from labeled CRHR1+ GCs in the presence of the sodium channel blocker tetrodotoxin (TTX). TTX blocks action potential-driven circuit activity, so its application isolates small subthreshold currents that are transmitted via synaptic inputs onto patched cells. We analyzed the frequency and amplitudes of currents recorded in TTX from patched CRHR1+ cells to reveal quantal features of their synaptic contacts. Because a major contribution of GCs to odor processing is to shape bulb output via reciprocal, dendrodendritic connections with excitatory M/T cells (Jahr and Nicoll, 1980; Isaacson and Strowbridge, 1998), we focused on how altered POU6f1 expression within GCs influenced excitatory currents back onto GCs. mEPSC recordings were performed from CRHR1+ GCs labeled with a Cre-dependent vector, AAV-FLEX-GFP for cKO and control animals, AAV-FLEX-GFP::POU6f1 for gain-of-function animals, to restrict our cross-group comparisons to granule neurons of a specific (CRHR1+) subpopulation in the OB.
With altered POU6f1 expression in CRHR1+ neurons, we detected a significant change in the mEPSC frequency among loss-of-function, control, and gain-of-function groups (Kruskal–Wallis test, H(2) = 8.382, p = 0.0151). Neurons that lacked POU6f1 showed a decreased frequency of mESPCs (0.41 ± 0.055 Hz) compared with controls (0.811 ± 0.15 Hz; Fig. 5B,C; p < 0.05, Dunn's multiple-comparisons posttest, cKO group: n = 12 animals, n = 29 cells, control: n = 7 animals, n = 24 cells, means reported as means ± SE), suggesting that POU6f1 expression is required for CRHR1+ neurons to establish the normal repertoire of excitatory connections. Although the frequency of mEPSCs was higher for POU6f1-overexpressing neurons (0.838 ± 0.13 Hz for POU6f1-overexpressing cells vs 0.811 ± 0.18 Hz for controls), this difference did not reach significance (p > 0.05, Dunn's multiple-comparisons posttest, POU6f1 overexpression group: n = 8 animals, n = 25 cells).
We next investigated whether POU6f1 influences other properties of excitatory synapses that are made onto GCs. GCs receive excitatory input from M/T cells via dendrodendritic connections in the EPL and via axodendritic connections from M/T cell collaterals in the GCL (Shepherd et al., 2004). The disparate distances of these synapses from GC somas influence recorded electrophysiological properties, including excitatory current decay time constants (τdecay; Schoppa, 2006). Dendrodendritic M/T cell synapses onto GCs exhibit currents with longer τdecay values relative to more proximal axodendritic M/T cell synapses. Therefore we reasoned that significant changes in τdecay distributions for mEPSCs recorded from POU6f1-overexpressing GCs could suggest a difference in the type of M/T connections established onto these cells. Because the majority of GCs display a mix of excitatory inputs (Schoppa, 2006), we examined the distributions of τdecay from randomly sampled populations of mEPSC events across experimental groups.
Analysis showed a right shift in the distribution of τdecay values for mEPSCs recorded from POU6f1-overexpressing neurons versus controls (Fig. 5D; two-sample Kolmogorov–Smirnov test, p < 0.0001, control group: n = 7 animals, n = 26 cells, 50 events per cell, POU6f1 overexpression group: n = 8 animals, n = 27 cells, ∼50 events per cell). Therefore, more mEPSCs onto POU6f1-overexpressing CRHR1+ GCs displayed larger τdecay values versus mEPSCs onto control CRHR1+ GCs. No significant change in the distribution of τdecay values was found between cKO and control mEPSCs (two-sample Kolmogorov–Smirnov test, p = 0.1542, control: n = 7 animals, n = 26 cells, 50 events per cell, cKO: n = 12 animals, n = 27 cells, ∼50 events per cell). This suggested that POU6f1-overexpressing CRHR1+ GCs gain excitatory input from synapses with decay kinetics similar to those of dendrodendritic M/T cell synapses (and/or those further away from the soma). Together with the above findings, these data suggest the possibility that, although loss of POU6f1 may indiscriminately reduce the quantal content of excitatory connections onto developing GCs, increased POU6f1 levels may enhance the quantal content of particular types of excitatory connections. However, this idea would require further experimentation and analysis to substantiate.
In our experiments, levels of POU6f1 expression in CRHR1+ cells also linearly affected the amplitudes of mEPSCs across cKO (10.23 ± 0.42 pA), control (10.17 ± 0.57 pA), and POU6f1-overexpressing (11.93 ± 0.67 pA) neurons (Fig. 5E; one-way ANOVA, F(2,77) = 3.156, p = 0.0481, animals per group as above, cKO: n = 27 cells, control: n = 26 cells, POU6f1 overexpression: 27 cells; posttest for linear trend, R2 = 0.05532, p = 0.035). Although differences between the groups did not reach statistical significance in pairwise comparisons against controls (p > 0.05, Dunnett's multiple-comparisons test for both cKO vs control and POU6f1-overexpression vs control), the observed subtle but linear increase in mEPSC amplitudes in response to increasing levels of POU6f1 could reflect a modest role for POU6f1 in influencing the quantal size of excitatory synapses onto CRHR1+ neurons. The above results suggest that POU6f1 could influence multiple properties of excitatory inputs onto CRHR1+ GCs.
We next sought to determine whether POU6f1 influenced passive membrane properties of CRHR1+ GCs. No differences were observed in membrane resistance among groups (Fig. 5F; cKO mean: 751.6 ± 79.9 MΩ, control mean: 729.5 ± 46.6 MΩ, POU6f1-overexpression mean: 685.3 ± 47.5 MΩ; Kruskal–Wallis test, H(2) = 0.636, p = 0.7256, cKO: n = 12 animals, n = 32 cells, control: n = 7 animals, n = 39 cells, POU6f1-overexpression: n = 8 animals, n = 34 cells). Similarly, no difference in membrane electrical capacitance (Cm) was observed between neurons that lacked POU6f1 expression and controls (cKO mean: 13.41 ± 0.85 pF vs 13.03 ± 0.63 pF in controls; one-way ANOVA, F(2,105) = 6.236, p = 0.0028, p > 0.05 post hoc Dunnett's multiple-comparisons test between cKO and control). Overexpression of POU6f1, however, resulted in higher Cm of CRHR1+ GCs (Fig. 5G; 16.53 ± 0.83 pF in POU6f1-overexpressing cells vs 13.03 ± 0.63 pF in controls; p < 0.05, post hoc Dunnett's multiple-comparisons test between POU6f1-overexpressing cells and controls), with a significant linear trend identified for Cm dependent on levels of POU6f1 (R2 = 0.06733, p = 0.0059). This observation suggested a role for POU6f1 in membrane outgrowth or branching of CRHR1+ GCs.
POU6f1 influences dendritic branching in CRHR1+ neurons
Having established that levels of POU6f1 in CRHR1+ cells are associated with a linear increase in Cm, we next examined the effects of POU6f1 expression on dendritic outgrowth. Loss- and gain-of-function groups were again established using CRHR1-Cre mice to conditionally excise or overexpress POU6f1 in CRHR1+ cells, respectively. Cells were targeted with sparse viral labeling to enable single-cell analysis for the effects of POU6f1 on morphological development of CRHR1+ GCs. Using this approach, we observed increased total dendrite length associated with higher levels of POU6f1 expression (Fig. 6A,B; 1469.5 ± 94 μm for controls vs 1891.2 ± 124.8 μm for overexpression; one-way ANOVA, F(2,34) = 8.254, p = 0.0012, p < 0.05, Dunnett's multiple-comparisons test, n = 3 animals per group, n = 12–13 cells per group, means reported as means ± SE; posttest for linear trend R2 = 0.3198, p = 0.0003). This increase in total dendritic length was primarily due to an increase in apical dendrite length (Fig. 6C; one-way ANOVA, F(2,34) = 7.280, p = 0.0023, p < 0.05 Dunnett's multiple-comparisons test; 1620.3 ± 112.2 μm vs 1262.8 ± 90.4 μm for controls). Moreover, we again observed a linear association between the lengths of apical dendrites with respect to levels of POU6f1 expression (posttest for linear trend, R2 = 0.2949, p = 0.0006). This contrasted with unchanged lengths of basal dendrites for CRHR1+ granule neurons regardless of POU6f1 expression (Fig. 6D; cKO: 182.9 ± 41.4 μm, control: 206.7 ± 57.9 μm, POU6f1 overexpression: 270.9 ± 40.4 μm; one-way ANOVA, F(2,34) = 0.9214, p = 0.4077). Increased apical dendrite lengths associated with increased POU6f1 expression provided further support that POU6f1 influences the functional development of dendrodendritic synapses with M/T cells because apical dendrites of mature GCs project into the EPL to establish such contacts (Shepherd et al., 2004; Schoppa, 2006).
To determine whether changes in apical dendrite length associated with POU6f1 expression were also accompanied by changes in branch architecture, we measured both the total dendritic branch number and the maximum branch order for neurons in each group. We found that POU6f1 overexpression in CRHR1+ cells resulted in increased formation of dendritic branches (Fig. 6E; 12.46 ± 1.27 branches for control vs 28.75 ± 2.85 branches for POU6f1 overexpression; one-way ANOVA, F(2,34) = 20.57, p < 0.0001; post hoc Dunnett's multiple-comparisons test, p < 0.05 for control vs POU6f1 overexpression). Moreover, the maximum branch order for neurons overexpressing POU6f1 was greater than that of control (Fig. 6F; 8 ± 0.60 for POU6f1-overexpressing neurons vs 4.23 ± 0.26 for controls; one-way ANOVA, F(2,34) = 22.01, p < 0.0001; post hoc Dunnett's multiple-comparisons test, p < 0.05 for control vs POU6f1 overexpression). These results suggested that POU6f1 influences dendritic branch formation and architecture in CRHR1+ neurons. To examine the consequence of changed POU6f1 expression on dendrite patterning, we further performed Sholl analysis (Fig. 6G) on traced dendritic branches. For this, we quantified branch intersections with concentric circles drawn starting from the somata of traced neurons, moving in 10 μm steps distally to cover the entirety of dendritic expanse. Both POU6f1 cKO and POU6f1-overexpressing neurons showed changes in neuronal arborization compared with controls (two-way ANOVA, F(2,2890) = 68.72, p < 0.0001 for levels of POU6f1 as a source of variation). Loss of POU6f1 expression resulted in reduced branching complexity relative to controls (p < 0.05, at radii of 520–650 μm distal to somas, Bonferroni multiple comparisons post hoc test), whereas POU6f1 overexpression enhanced branching complexity (p < 0.05, at radii of 290–410 μm distal to somas). Together, the above results suggest that the POU6f1 transcription program influences dendritic outgrowth and branch patterning in CRHR1+ neurons.
Discussion
Plasticity in the adult brain is typified by activity-dependent structural alterations (Volkmar and Greenough, 1972; Magariños et al., 1996). The mouse OB circuit features ongoing, activity-dependent structural changes via integration of adult-born GCs. Interestingly, enhanced activity by odor enrichment or learning does not affect dendritic arborization of M/T cells (Mizrahi and Katz, 2003), the principal excitatory cells of the OB that relay multiple streams of output from the bulb (Fukunaga et al., 2012). The structural rigidity of M/T cells contrasts sharply with remodeling observed in newborn inhibitory neurons in the bulbs of animals subjected to odor task learning and enrichment (Saghatelyan et al., 2005; Arenkiel et al., 2011; Breton-Provencher et al., 2016; Huang et al., 2016; Quast et al., 2017). This suggests that, in the OB, activity-dependent circuit sculpting is primarily the domain of interneuron dynamics. However, the molecular pathways by which interneurons initiate and maintain structural changes in response to sensory input remain largely unresolved.
We recently discovered an interneuron-to-interneuron signaling mechanism mediated via activity-dependent, locally secreted CRH in the bulb. This neuropeptidergic signaling cascade influences morphology and functional maturation of CRHR1+ GCs (Garcia et al., 2014). Both CRH (Swanson et al., 1983) and its main CNS receptor CRHR1 (Chalmers et al., 1995) are expressed in other brain areas and the local release of CRH in these areas is also influenced by environmental cues (Imaki et al., 1991). Therefore, we hypothesize that CRH and/or other neuropeptide signaling networks are broadly embedded to convey network-shaping effects of sensory activity. Indeed, others have reported that CRH signaling within the hippocampus remodels pyramidal cell dendrites in response to stress (Chen et al., 2004). The role of CRH in activity-dependent circuit restructuring in the bulb may therefore exemplify a more generalizable principle of peptide-mediated plasticity.
Here, we examined a mechanism by which local CRH signaling reshapes circuitry in the adult OB. We found that CRHR1 activity influences expression of the transcription factor POU6f1 in neural cells and that expression of this factor is required for normal synaptic integration and dendritic elaboration.
Local CRH signaling promotes activation of the transcriptional regulator POU6f1
CRH-to-CRHR1 signaling has been implicated previously in altered dendritic morphology in the hippocampus (Chen et al., 2004) and bulk-loaded CRH is reported to induce expression of immediate early genes such as the transcription factor c-fos in other brain regions (Clark et al., 1991; Parkes et al., 1993). However, to date, no link between local CRH signaling and a specific transcriptional program that regulates dendritic plasticity has been established. Here, we performed a screen for gene expression changes downstream of CRH signaling in the bulb and found that CRHR1 activation was associated with expression of the transcription factor POU6f1. Our screen was performed on whole bulb tissue and thus did not discern cell-type specificity of expression changes. However, in situ hybridization against POU6f1 in the OBs of adult CRHR1−/− mice showed that CRHR1−/− cells in the GCL downregulated expression of POU6f1.
In vitro data further suggest that POU6f1 is modulated directly by CREB-dependent transcription in response to CRHR1 activity. This is consistent with the known CREB-activating mechanism of CRHR1, a GS-protein-coupled receptor (Berger et al., 2006; Blank et al., 2003). However, it is possible that an alternative or additional CREB-dependent mechanism of POU6f1 activation operates in vivo. Bulleit et al., 1994 showed that POU6f1 mRNA increased in cultured neurons after NMDA receptor activation. NMDAR-mediated calcium influx could therefore also contribute to CREB-dependent transcriptional activation (Sheng et al., 1990) of POU6f1 in vivo.
POU6f1 expression correlates with neuronal maturity
We found that POU6f1::HA is expressed progressively as adult-born neurons develop. Moreover, most mature (NeuN+) GCs also expressed POU6f1::HA, and reciprocally most POU6f1::HA+ GCs expressed NeuN, suggesting that POU6f1 expression correlates with maturation of GCs. Interestingly, NeuN expression, which appears after developing neurons exit the cell cycle (Kim et al., 2009), reportedly increases rapidly in adult-born neurons of the OB during the first 2 weeks after birth before plateauing (Petreanu and Alvarez-Buylla, 2002). Therefore, NeuN is predicted to appear before expression of both CRHR1 (Garcia et al., 2014) and POU6f1::HA. We found previously that CRHR1 expression was required for normal rates of survival (Garcia et al., 2014) and report here that POU6f1 influences both structural and functional aspects of development in CRHR1-expressing cells. Therefore, early NeuN expression relative to expression of either CRHR1 or POU6f1 in adult-born neurons suggests that they may be marked for generalized maturation (i.e., cell cycle exit) before subsequent expression of genes involved in neuropeptide signaling. Interestingly, we did not observe a direct effect of POU6f1 on survival of CRHR1+ adult-born (data not shown). This is consistent with the notion that POU6f1 influences later aspects of maturation.
POU6f1 mediates synapse and dendrite branch formation
We found that loss of POU6f1 reduced the quantal content, encompassing the number of release sites and/or probability of vesicle release, of excitatory synaptic connections formed onto CRHR1+ GCs. Increased POU6f1 expression resulted in delayed excitatory current kinetics in a manner consistent with increased input by dendrodendritic synapses with M/T cells. This was further consistent with analysis suggesting that POU6f1 promotes formation of apical dendrites, onto which dendrodendritic synapses between CRHR1+ GCs and M/T form. Together, these results indicate that POU6f1 contributes to the content and function of excitatory synaptic inputs onto CRHR1+ cells.
We observed that low levels of POU6f1 were associated with reduced total dendrite length in GCs and high levels of POU6f1 resulted in increased dendrite length. Interestingly, these differences were almost entirely explained by changes in apical dendrite lengths because basal dendrites were unaffected by POU6f1 expression. This has intriguing implications for whether POU6f1 influences preferentially dendrodendritic synapses of GCs with M/T cells, which tend to make contacts onto distal apical dendrites of GCs (Schoppa, 2006), over inputs by centrifugal fibers, which tend to form basal and proximal connections onto GCs (Shipley and Adamek, 1984; Laaris et al., 2007). It is tempting to speculate that local CRH secreted from within the EPL promotes the formation of dendrodendritic synapses in this same layer to enhance olfaction at proximal sites of sensory processing. Others have reported that functionally correlated GCs appear to cluster (Wienisch and Murthy, 2016) in alignment with M/T cells and along the borders of odor-specific glomeruli (Guthrie et al., 1993; Wilhite et al., 2006). A localized peptidergic cue that is secreted with odor activity to guide synapse formation in such a functional vicinity provides a potential mechanism by which these topographic relationships are established and/or maintained across the surface of the bulb. However, more direct testing would be required to investigate such a possibility.
The notion that POU6f1 reshapes GC connectivity was further supported by observations of increased dendritic branch number, maximum branch order, and branch pattern complexity in POU6f1-overexpressing neurons. Such intricate outgrowths of GC arborizations could give rise to more and/or varied granule-to-M/T cell contacts, which tend to form on lateral secondary dendrites of M/T cells distal to their somas (Bartel et al., 2015). Functionally, more numerous or expanded GC connections may contribute to broadened GC tuning and contrast enhancement (Cazakoff et al., 2014) or serve as a means to synchronize M/T firing, such that the bulb relays a tight representation of the collective attributes of complex odors (Davison and Katz, 2007). Such outcomes would be promising paths for future inquiry.
One caveat to our studies arises from the expression pattern of CRHR1 in adult-born neurons, thus affecting our maniuplations in CRHR1-Cre mice. CRHR1 is expressed by adult-born neurons both during and continually after integration, when dendrite extension has ceased (Mizrahi, 2007). Our assays examined aspects of plasticity normally restricted to the preintegration period. Therefore, because our experiments targeted both younger and older CRHR1+ GCs, our interpretations likely underestimate the effects of POU6f1 in shaping these processes in younger cells. This may explain why we observed no differences in odor learning or short-term olfactory behavior in cKO versus control mice (data not shown). More selective tests of developmental contributions of POU6f1 would require a CRHR1-Cre inducible mouse model.
Nevertheless, our data suggest that, within CRHR1+ GCs in the adult OB, POU6f1 is a transcriptional regulator that can influence dendrite specification and is required to establish a normal complement of excitatory synapses onto CRHR1+ GCs. The effects of loss and gain of POU6f1 in the present studies did not perfectly mirror the effects of loss and gain of CRHR1 in our previous studies (Garcia et al., 2014). This implies that CRH signaling in the OB likely results in other changes beyond those mediated by POU6f1 and that POU6f1 may have roles outside of conveying the effects of local CRH signaling. However, the data herein present the first evidence that a specific transcription program can be initiated downstream of activity-dependent local peptide signaling to the end effect of functional circuit refinement in the adult brain (Fig. 7).
Note Added in Proof: The tenth author's last name was accidentally misspelled on the Early Release version published January 5, 2018. The name has now been corrected.
Footnotes
This work was supported by the McNair Medical Institute, the National Institute of Neurological Disorders and Stroke (NINDS)–National Institutes of Health (NIH; Grant 1F31NS092435–01A1 to C.K.M. and Grant R01NS078294 to B.R.A.), the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)–NIH (Grant 5U54HD083092 to B.R.A.), the RNA In Situ Hybridization Core facility at BCM (with the expert assistance of Dr. Cecilia Ljungberg, and funding from NIH Grant 1S10 OD016167 and NICHD Grant 1U54 HD083092 to the Intellectual and Developmental Disabilities Research Center); and the National Cancer Institute–NIH (Grant CA125123 to the Dan L. Duncan Comprehensive Cancer Center). We thank Dr. Andrew Groves, Dr. Roy Sillitoe, and Kevin Ung for helpful discussion and critical input on this manuscript; the Baylor College of Medicine (BCM) Medical Scientist Training Program for training support; and the BCM Mouse ES Cell and Genetically Engineered Mouse Cores for assistance with CRISPR/Cas9 reagent production, mouse ES cell culture, and mouse production.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Benjamin R. Arenkiel, Dept. of Molecular and Human Genetics, Jan and Dan Duncan Neurological Research Institute, Baylor College of Medicine, Houston, TX 77030. arenkiel{at}bcm.edu