Partial or Complete Loss of Norepinephrine Differentially Alters Contextual Fear and Catecholamine Release Dynamics in Hippocampal CA1

Abstract

Background

Contextual fear learning is heavily dependent on the hippocampus. Despite evidence that catecholamines contribute to contextual encoding and memory retrieval, the precise temporal dynamics of their release in the hippocampus during behavior is unknown. In addition, new animal models are required to probe the effects of altered catecholamine synthesis on release dynamics and contextual learning.

Methods

We generated 2 new mouse models of altered locus coeruleus–norepinephrine (NE) synthesis and utilized them together with GRAB_NE and GRAB_DA sensors and in vivo fiber photometry to investigate NE and dopamine (DA) release dynamics in the dorsal hippocampal CA1 during contextual fear conditioning.

Results

Aversive foot shock increased both NE and DA release in the dorsal CA1, while freezing behavior associated with recall of fear memory was accompanied by decreased release. Moreover, we found that freezing at the recent time point was sensitive to both partial and complete loss of locus coeruleus–NE synthesis throughout prenatal and postnatal development, similar to previous observations of mice with global loss of NE synthesis beginning postnatally. In contrast, freezing at the remote time point was compromised only by complete loss of locus coeruleus–NE synthesis beginning prenatally.

Conclusions

Overall, these findings provide novel insights into the role of NE in contextual fear and the precise temporal dynamics of both NE and DA during freezing behavior and highlight complex relationships between genotype, sex, and NE signaling.

Contextual learning involves the integration of the multisensory features of a particular environment into a single representation that can support complex adaptive behaviors (1, 2, 3). Individual stimuli can have substantially different meaning and significance in different contexts, and contexts themselves can serve as powerful predictors of appetitive and aversive events. In contextual fear conditioning, the multisensory features that define the conditioning chamber are associated with an aversive unconditioned stimulus that subsequently drives fear expression (3, 4, 5, 6). Contextual fear is heavily dependent on the hippocampus (7,8), and numerous studies have demonstrated the importance of norepinephrine (NE) and dopamine (DA) signaling in different hippocampal subregions to contextual encoding and retrieval (9, 10, 11, 12, 13, 14, 15, 16).

With respect to NE, the hippocampus receives virtually all of its inputs from the locus coeruleus (LC) (17,18), and LC-NE signaling to dorsal hippocampal area CA1 (dCA1) has been specifically implicated in retrieval of contextual memories (19,20). LC-NE signaling through β-adrenergic receptors has been shown to modulate activation of dCA1 pyramidal neurons during memory retrieval (11), and optogenetic activation of LC-NE inputs to the dCA1 enhances subsequent retrieval of contextual memories 24 hours after training (21). Consistent with these observations, global loss of NE synthesis in dopamine β-hydroxylase knockout mice (Dbh^ko) (22) results in deficits in short-term memory retrieval (11). Despite these important discoveries, the precise temporal dynamics of NE and DA release in the dCA1 during fear conditioning remain unknown. In addition, it is unclear how catecholamine dynamics and freezing behavior during contextual fear are altered by more subtle disruptions of NE synthesis.

In the current study, we present 2 new mouse models of disrupted LC-NE synthesis beginning prenatally: a hypomorphic allele of Dbh that results in reduced NE in central noradrenergic neurons and an LC conditional knockout with total loss of NE in the LC and reduced NE in other central noradrenergic nuclei. Utilizing these mice in conjunction with a preexposure-dependent contextual fear assay (23,24) and in vivo fiber photometry in the dCA1, we sought to determine the effects of partial and complete embryonic loss of LC-NE synthesis on freezing behavior and assess NE and DA release dynamics during 4 phases of the contextual fear assay. Our data identify deficits in freezing behavior and unique patterns of catecholamine release dynamics in the dCA1 that are differentially sensitive to partial or complete embryonic loss of NE.

Methods and Materials

Animals

All procedures related to animal use were approved by the Animal Care and Use Committee of the National Institute of Environmental Health Sciences. Mice were group housed and maintained on a 12-hour light/dark cycle at 72 ± 2 °F with access to food and water ad libitum.

To establish the Dbh^tm2.2Pjen (Dbh^flx) mouse line, a targeting vector was electroporated into G4 embryonic stem cells (25), and chimeras were bred to C57BL/6J mice. All mice used in the study were backcrossed to C57BL/6J >10 generations. For details on targeting vector construction, genotyping, and crosses, see Supplemental Methods and Materials.

Immunohistochemistry/Mass Spectrometry/Droplet Digital Polymerase Chain Reaction

For immunohistochemistry, tissue was collected, and immunofluorescent labeling was performed as previously described (26). For catecholamine analysis, 100 to 200 μL of whole blood, hippocampal samples, and adrenals were flash frozen and processed for mass spectrometry. For quantification of Dbh and tyrosine hydroxylase (Th) messenger RNA (mRNA), RNA was collected from the pons, medulla, adrenal, and stellate ganglion and analyzed by droplet digital polymerase chain reaction (PCR). For details, see Supplemental Methods and Materials.

In Vivo Fiber Photometry/Fear Conditioning

AAVs (adeno-associated viruses) expressing tdTomato (27) and NE (28) or DA (29) sensor were unilaterally injected in the dCA1. Mice were tested in a fear conditioning chamber with a grid floor (Med Associates), and freezing behavior was assessed using Ethovision XT software version 16 (Noldus Information Technology). Sensor and tdTomato fluorescence intensities were measured using a custom-built fiber photometry system (30). See Supplemental Methods and Materials for detailed procedures and analyses.

Results

Generation of a Mouse Model of Disrupted LC-NE Synthesis

In previous studies using Dbh^ko mice, the embryonic lethality believed to be caused by loss of NE synthesis in the peripheral nervous system was overcome by dosing pregnant dams with L-DOPS, thus permitting Dbh-independent NE synthesis during prenatal development (22,31). Consequently, Dbh^ko mice lack global NE postnatally after L-DOPS treatment is withdrawn. While the Dbh^ko is an excellent model to assess the requirement for NE throughout the postnatal period, the restoration of NE synthesis mediated by L-DOPS may obscure phenotypes associated with prenatal loss of NE. To eliminate LC-NE synthesis beginning in embryonic development and avoid the prenatal lethality associated with global loss of Dbh, we generated a Cre-dependent conditional knockout allele, Dbh^tm2.2Pjen (Dbh^flx), in which fluorescent tags allow recombination to be monitored in noradrenergic neurons (Figure 1A; Figure S1). In the absence of Cre-mediated recombination, transcription from the Dbh^flx allele will produce mRNA encoding separate DBH and tdTomato proteins. Cre-mediated recombination will delete most of the Dbh coding sequence, leaving a mutant allele (Dbh^null) that encodes a fusion between exon 1 and EGFP (enhanced green fluorescent protein). Thus, if Cre is expressed in a subset of noradrenergic neurons, the Cre+ noradrenergic neurons can be identified by EGFP fluorescence, while Cre-negative noradrenergic neurons that retain DBH activity are labeled with tdTomato. To confirm that recombination occurs as expected, we crossed Dbh^flx heterozygotes with Tmem163^{Tg(ACTB-cre)2Mrt} mice (32) which exhibit ubiquitous Cre activity. PCR analysis of genomic DNA demonstrated Cre-dependent deletion of the sequence between the 2 loxP sites (Figure 1A).

Figure 1.

Conditional knockout allele of the mouse Dbh gene. (A) Schematic diagram showing the wild-type and modified Dbh locus. Coding exons are indicated by black boxes, 5′ and 3′ untranslated regions by white boxes, and introns by black lines. In the conditional (Dbh^flx) allele, a single loxP site is inserted within intron 1, and a larger insertion in exon 12 consists of 2A peptide (56), tdTomato complementary DNA, SV40 polyadenylation signal, loxP, splice acceptor sequence, EGFP complementary DNA, and rabbit β-globin polyadenylation signal. After Cre recombination, the null allele consists of exon 1, a chimeric intron consisting of the 5′ portion of intron 1 and the exogenous splice acceptor, and EGFP. Arrows indicate polymerase chain reaction primers flanking the loxP sites. Box: Polymerase chain reaction analysis of genomic DNA from Dbh^wt/wt (wild-type), heterozygous Dbh^flx/wt, homozygous Dbh^flx/flx, and heterozygous Dbh^null/wt mice demonstrates excision of exons 2–12 and the tdTomato cassette following Cre recombination. (B) Sagittal schematic diagram of the embryonic neural tube showing overlap of noradrenergic progenitors and En1 expression domain in the anterior hindbrain and adult sagittal schematic showing location of noradrenergic nuclei and expected fluorescent labeling of noradrenergic neurons in Dbh^LC-null (En1^Cre/wtDbh^flx/flx). (C) Representative coronal sections from Cre-negative Dbh^flx/flx and Dbh^LC-null mice showing tdTomato or EGFP expression in neurons of the LC, SubCD, SubCV, C2/A2, and C1/A1 nuclei (top panel of each image pair). TH immunoreactivity confirms noradrenergic identity of labeled neurons (bottom panel of each image pair). Scale bar = 100 μm. EGFP, enhanced green fluorescent protein; LC, locus coeruleus; s.a., splice acceptor; SubCD, dorsal subcoeruleus; SubCV, ventral subcoeruleus; TH, tyrosine hydroxylase; wt, wild-type.

Next, we selectively eliminated NE synthesis in LC neurons by taking advantage of the fact that the anatomically defined LC (within the central gray of the pons, adjacent to the fourth ventricle), together with a rostral portion of the adjacent dorsal subcoeruleus, is uniquely defined by embryonic expression of the transcription factor En1 (engrailed 1) and later expression of Dbh (18). Because En1 is expressed before Dbh, the LC neurons of mice heterozygous for En1^Cre (33) and homozygous for Dbh^flx (Dbh^LC-null mice) (Figure 1B) will never produce DBH protein and therefore never synthesize NE. As expected, Dbh^LC-null mice were viable, and LC noradrenergic neurons of Dbh^LC-null mice were labeled with EGFP, indicating loss of Dbh expression (Figure 1C). In non-LC noradrenergic neurons, we observed tdTomato, indicating the unrecombined allele (Figure 1C). EGFP was fainter than tdTomato, and detection of both fluorophores required immunofluorescent labeling; nevertheless, these results indicate that the fluorescent tags can be used to monitor Cre-dependent recombination of Dbh^flx.

Mutation of a Putative Transcriptional Enhancer Is Associated With Reduced Dbh mRNA Expression

To confirm that Dbh^LC-null mice lack Dbh expression in the LC, we performed co-immunofluorescent labeling of DBH and TH. Consistent with the recombination data, DBH was not detected in the LC of Dbh^LC-null mice (Figure 2A). We observed no differences in DBH labeling between Cre+ and Cre-negative Dbh^wt/wt controls and no genotype-specific differences in TH labeling. However, in Cre-negative Dbh^flx/flx mice, DBH labeling appeared qualitatively weaker than in Dbh^wt/wt controls (Figure 2A).

Figure 2.

Disruption of norepinephrine synthesis in Dbh^LC-null and Dbh^flx/flx (Dbh^hypo) mice. (A) Coronal sections from Dbh^wt/wt, Dbh^flx/flx, and Dbh^LC-null mice showing DBH (green) and/or TH (magenta) expression in the LC (representative images from n = 4 female and 4 male mice per genotype). Scale bar = 100 μm. (B) Levels of norepinephrine and dopamine or DOPAC in the hippocampus, adrenal, and plasma of Dbh^wt/wt, Dbh^flx/flx, and Dbh^LC-null mice. Data are mean ± SEM. For hippocampus samples, n = 7 female (closed circles) and 8 male (open circles) Dbh^wt/wt (includes 8 En1^Cre/wt and 7 En1^wt/wt), 5 female and 5 male Dbh^flx/flx, and 11 female and 13 male Dbh^LC-null mice. For adrenal, n = 8 female and 8 male Dbh^wt/wt (8 En1^Cre/wt and 7 En1^wt/wt), 4 female and 4 male Dbh^flx/flx, and 4 female and 4 male Dbh^LC-null mice. For plasma samples, n = 14 female and 18 male Dbh^wt/wt (17 En1^Cre/wt and 15 En1^wt/wt), 10 female and 9 male Dbh^flx/flx, and 10 female and 10 male Dbh^LC-null mice. ∗∗∗∗p < .0001, ^###p = .0004 (main effect of sex). LC, locus coeruleus; TH, tyrosine hydroxylase; wt, wild-type.

Next, we examined NE tissue content in the hippocampus, which receives virtually all of its noradrenergic inputs from the LC, using mass spectrometry. As expected, NE was absent in the hippocampus of Dbh^LC-null mice (Figure 2B). Consistent with DBH immunolabeling, we observed an approximately 60% reduction of hippocampal NE in Cre-negative Dbh^flx/flx mice relative to Dbh^wt/wt mice (Figure 2B), demonstrating that the Dbh^flx allele is hypomorphic (exhibits less than wild-type activity). We also found that NE content was significantly higher in Dbh^wt/wt and Dbh^flx/flx male than in female mice (p = .003, sex × genotype interaction) (Figure 2B; Figure S2A). Because DBH converts DA to NE, we also examined DA content in the hippocampus and found that it was significantly increased in Dbh^flx/flx and Dbh^LC-null mice relative to Dbh^wt/wt controls (p < .0001, main effect of genotype, two-way analysis of variance [ANOVA] with Tukey’s test) (Figure 2B). Furthermore, DA in Dbh^LC-null hippocampus was significantly higher than in Dbh^flx/flx (p < .0001, two-way ANOVA with Tukey’s test) (Figure 2B), demonstrating that larger disruptions in NE synthesis corresponded to greater hippocampal DA levels. In contrast, we observed no genotype-specific differences in NE content in adrenal or release in plasma and no difference in levels of the DA metabolite DOPAC in plasma. However, DA was significantly increased in adrenal of both Dbh^flx/flx and Dbh^LC-null mice (p < .0001, two-way ANOVA with Tukey’s test) (Figure 2B). Taken together with the results of the analysis of recombination (Figure 1), our findings confirm that Dbh^LC-null mice lack NE synthesis in the LC and that the Dbh^flx allele is hypomorphic rather than functionally wild-type.

To explore possible causes of the unexpected hypomorphy of Dbh^flx/flx mice (hereafter designated Dbh^hypo), we examined regulatory annotations for the mouse genome (GRCm39) in the Ensembl Genome Browser (Release 108) (34). We found that a 1399-bp putative enhancer region in mouse Dbh intron 1 (Ensembl ENSMUSR00000824281) is interrupted by the upstream loxP at an EcoRI site beginning at nucleotide 291. Disruption of an enhancer is expected to alter mRNA levels; therefore, we performed droplet digital PCR to assess Dbh mRNA in both the central and peripheral noradrenergic systems (Figure 3). We examined mRNA levels in the pons (which includes the LC, subcoeruleus, and A5 nuclei), medulla (A1 and A2 nuclei), adrenal, and stellate ganglion. In Dbh^hypo mice, we found reduced Dbh mRNA relative to wild-type in all tissues examined (reduction of approximately 73% in the pons, 64% in the medulla, 57% in adrenal, and 78% in the stellate ganglion). We also found a sex-specific difference in adrenal, with males having higher Dbh expression than females (p = .0010, main effect of sex) (Figure 3; Figure S2B). Next, we tested whether reduced Dbh expression affects Th and found no difference in Th mRNA expression between Dbh^hypo and Dbh^wt/wt samples (Figure 3). Therefore, these data indicate that Dbh^hypo and Dbh^LC-null mice represent 2 distinct models of disrupted Dbh expression.

Figure 3.

Dbh and Th mRNA levels in Dbh^wt/wt and Dbh^hypo mice. Data are mRNA levels estimated by droplet digital polymerase chain reaction, mean ± SEM. For Dbh^hypo pons, n = 4 female (closed circles) and 3 male (open circles) mice. For all other samples, n = 4 female and 4 male mice. ∗∗∗∗p < .0001, ∗∗p = .0020, ^###p = .0010 (main effect of sex). mRNA, messenger RNA; Th, tyrosine hydroxylase; wt, wild-type.

Impaired Contextual Fear Response in Dbh^hypo and Dbh^LC-null Mice

It has previously been shown that global loss of NE synthesis beginning in the early postnatal period results in deficits in hippocampus-dependent memory retrieval (11). To assess the effects of embryonic loss of LC-NE—both partial and complete—on behavior and NE and DA release dynamics, we used spectrally resolved fiber photometry (30) combined with contextual fear conditioning. Male and female Dbh^LC-null, Dbh^hypo, and Dbh^wt/wt mice were injected with AAVs expressing tdTomato and either an NE sensor (GRAB_NE) (28) or a DA sensor (GRAB_DA) (29) and implanted with fiber optical probes in the hippocampal dCA1 (Figure 4A). Next, we utilized the preexposure-dependent contextual fear assay (Figure 4B), which consists of 4 phases. A preexposure phase allows the mice to explore the novel context and acquire a hippocampus-dependent contextual representation of the conditioning chamber. During the immediate-shock day, this contextual representation is rapidly retrieved and associated with the shock to form a contextual fear memory. Recent (24 hours after shock) and remote (2 weeks after shock) tests assess short- and long-term retrieval of the contextual fear memory, as measured by the amount of time that the mice spend freezing (i.e., motionless, a species-specific fear response) when placed back in the conditioning chamber.

Figure 4.

Experimental conditions and results of context-dependent fear conditioning. (A) Coronal schematic of mouse brain showing injection site of AAV9-GRAB_NE or -GRAB_DA and AAV5-tdTomato viruses in the dorsal CA1 region of the hippocampus. Fiber optic probes were implanted 3 weeks later above the dorsal CA1. In vivo fiber photometry recordings were performed using a spectrometer-based system with a 488-nm emission laser. Software allowed for simultaneous visualization of photon counts from fluorescent signals at various wavelengths. (B) Schematic of context-dependent fear conditioning paradigm performed in parallel with fiber photometry. (C) Distance the mice moved during the preexposure recording. (D) Distance moved during shock. (E) Percentage of time spent freezing during the recent context test (∗∗p = .0021, ∗p = .031). (F) Freezing during the remote context test (∗∗p = .0042). Data are mean ± SEM. Two-way analysis of variance with Tukey’s multiple comparisons test, n = 31 Dbh^wt/wt control (15 female, 16 male), n = 26 Dbh^hypo (13 female, 13 male), and n = 23 Dbh^LC-null (12 female, 11 male) mice. DA, dopamine; dist., distance; LC, locus coeruleus; NE, norepinephrine; wt, wild-type.

During the preexposure phase of the test, we observed no genotype- or sex-specific differences in the distance moved, indicating no baseline differences in motor exploration (Figure 4C). Similarly, no genotype-specific differences in distance moved were observed during the activity burst associated with the 2-second aversive foot shock (Figure 4D). However, males were observed to move more than females (p = .0090, main effect of sex) (Figure S2C). These results indicate that NE deficits associated with the Dbh^hypo and Dbh^LC-null mouse models do not affect overall activity or sensitivity to aversive stimuli. To assess retrieval of contextual fear memory, we measured the percentage of time mice spent freezing when placed back in the chamber in which they received the shock. During the recent retrieval test, both Dbh^hypo and Dbh^LC-null mice exhibited reduced freezing compared with controls (Dbh^hypo, p = .0305; Dbh^LC-null, p = .0021) (Figure 4E; Figure S3). However, during the remote retrieval test, only Dbh^LC-null mice still exhibited significantly reduced freezing (p = .0042) (Figure 4F).

Dynamic Release of NE and DA in the dCA1 During Contextual Fear Conditioning

Next, we analyzed the fiber photometry data to assess NE and DA dynamics in the hippocampal dCA1 during the 4 phases of the fear conditioning paradigm. The photometry signal was aligned to aversive shock and to onset of immobility or freezing during the preexposure phase and tests of memory retrieval. During the preexposure, immobility was relatively rare and of shorter duration (average immobility bout length = 3.851 ± 0.51 seconds vs. freeze bout length for recent and remote context tests = 7.655 ± 0.78 seconds and 8.505 ± 0.65 seconds, respectively) but sufficient to permit analysis of event-triggered averages. We observed no genotype- or sex-specific differences in NE or DA release during immobility in the preexposure phase (Figure S4), consistent with our observation that Dbh^LC-null, Dbh^hypo, and littermate control mice have similar locomotion at baseline. However, during the aversive foot shock, we observed an immediate increase in NE release in both Dbh^hypo and control mice. As expected, the evoked increase in NE was absent in Dbh^LC-null mice (Figure 5A, B), consistent with our measure of tissue content (Figure 2B). This result also serves as a negative control and confirms the specificity of the GRAB_NE sensor. During the 10 seconds following the foot shock, NE release was significantly increased in both wild-type controls (p = .0009) and Dbh^hypo mice (p = .0384) compared with Dbh^LC-null mice (Figure 5C). We also observed increased DA release in response to the shock (Figure 5A, D), with release in Dbh^LC-null mice significantly greater than in Dbh^hypo mice (p = .0067) and wild-type controls (p < .0001) (Figure 5E). Despite the significant difference in NE and DA tissue content between Dbh^hypo and wild-type mice (Figure 2B), we found no significant differences in NE or DA release between those 2 genotypes (Figure 5C, E). However, we did note a trend for a genotype × sex interaction for NE release (p = .059, two-way ANOVA) in which fluorescence ratios from male Dbh^hypo mice resembled those from wild-type males and female Dbh^hypo mice resembled female Dbh^LC-null mice (Figure 5B, C; Figure S2D).

Figure 5.

Norepinephrine and dopamine dynamics in the dorsal CA1 during and immediately following a noxious foot shock. (A) GRAB_NE/tdTomato (norepinephrine) and GRAB_DA/tdTomato (dopamine) fluorescence ratios expressed as a normalized response aligned to foot shock (dotted line). Each row in the heat map corresponds to an individual mouse, with the largest ratio for that mouse set to 100 and the smallest ratio set to 0. (B) Average GRAB_NE/tdTomato fluorescence ratios aligned to shock in female (left) and male (right) Dbh^wt/wt control, Dbh^hypo, and Dbh^LC-null mice. (C) Average norepinephrine response during the first 10 seconds after the foot shock for the Dbh^wt/wt control (95% CI of mean, 1.31 to 4.76), Dbh^hypo (1.03 to 2.83), and Dbh^LC-null mice (−0.64 to 0.56; encompasses 0). ∗∗∗p = .0009, ∗p = .0384. (D) Average GRAB_DA/tdTomato fluorescence ratios aligned to shock. (E) Average dopamine response during the first 10 seconds after shock for the Dbh^wt/wt control (95% CI of mean, 0.98 to 2.30), Dbh^hypo (1.10 to 4.99), and Dbh^LC-null mice (4.41 to 8.44). ∗∗p = .0067, ∗∗∗∗p < .0001. GRAB_NE cohort: n = 13 Dbh^wt/wt control (6 female, 7 male), n = 14 Dbh^hypo (7 female, 7 male), and n = 11 Dbh^LC-null (6 female, 5 male) mice. GRAB_DA cohort: n = 17 Dbh^wt/wt control (8 female, 9 male), n = 12 Dbh^hypo (6 female, 6 male), and n = 12 Dbh^LC-null (6 female, 6 male) mice. LC, locus coeruleus; wt, wild-type.

Next, we assessed NE and DA release in the dCA1 during freezing in the recent (24 hours after shock) and remote (2 weeks after shock) tests, which assess short- and long-term retrieval of the contextual fear memory. Both Dbh^hypo mice (p = .0384) and wild-type controls (p = .0003) exhibited significantly decreased NE release during freezing in the recent test (Figure 6A, B). As expected, we observed no change in NE release in Dbh^LC-null mutants (Figure 6A, B). To examine the relationship between the decrease in NE signal and the length of a freezing bout, we used a general linear model. We found that longer freezing bouts predicted a larger freezing-related drop in NE signal in males (main effect of bout length in males and females, p < .001 and p = .225, respectively) (Figure S5). DA release was also decreased during freezing in the recent test; however, this decrease occurred in all genotypes including Dbh^LC-null mice, and there was no significant difference between the groups (Figure 6C, D). We found no relationship between DA signal and length of freezing (main effect of bout length in males and females, p = .676 and p = .585, respectively) (Figure S5).

Figure 6.

Norepinephrine and dopamine dynamics in the dorsal CA1 during freezing in the recent context test. (A) Average GRAB_NE/tdTomato fluorescence ratios aligned to freezing in female (left) and male (right) Dbh^wt/wt, Dbh^hypo, and Dbh^LC-null mice. (B) Averages of norepinephrine response during the first 6 seconds of freezing episodes in Dbh^wt/wt control (95% CI of mean, −1.3 to −0.36), Dbh^hypo (−0.70 to −0.23), and Dbh^LC-null mice (−0.12 to 0.29; encompasses 0). ∗∗p = .0003, ∗p = .0384. (C) Average GRAB_DA/tdTomato fluorescence ratios aligned to freezing. (D) Average dopamine responses during the first 6 seconds of freezing episodes in Dbh^wt/wt control (95% CI of mean, −0.61 to −0.18), Dbh^hypo (−0.78 to −0.35), and Dbh^LC-null mice (−0.77 to −0.14). Data are mean ± SEM. Two-way analysis of variance with Tukey’s test. GRAB_NE cohort: n = 13 Dbh^wt/wt control (6 female, 7 male), n = 14 Dbh^hypo (7 female, 7 male), and n = 11 Dbh^LC-null mice (6 female, 5 male). GRAB_DA cohort: n = 18 Dbh^wt/wt control (9 female, 9 male), n = 12 Dbh^hypo (6 female, 6 male), and n = 12 Dbh^LC-null mice (6 female, 6 male). LC, locus coeruleus; wt, wild-type.

We observed no genotype-specific decrease in NE release during freezing in the remote test of long-term retrieval (Figure 7A, B), although male mice exhibited a significant decrease relative to females (p = .0484, main effect of sex, two-way ANOVA with Tukey’s test). Consistent with the recent test, DA release was decreased in all mice during freezing in the remote test, with no significant difference found between the groups (Figure 7C, D). We found no relationship between length of freezing and either NE signal or DA signal (Figure S5). Next, we examined whether NE or DA release in response to the shock predicted subsequent freezing behavior in the recent and remote context tests. DA, but not NE, showed a significant predictive relationship in all genotypes examined, and this was driven by the female mice (sustained shock response coefficient in females, p = .003 and p = .001 for recent and remote, respectively) (Figure S6).

Figure 7.

Norepinephrine and dopamine dynamics in the dorsal CA1 during freezing in the remote context test. (A) Average GRAB_NE/tdTomato fluorescence ratios aligned to freezing in female (left) and male (right) control, Dbh^hypo, and Dbh^LC-null mice. (B) Average norepinephrine response during the first 6 seconds of freezing in Dbh^wt/wt control (95% CI of mean, −0.64 to 0.057), Dbh^hypo (−0.43 to −0.06), and Dbh^LC-null mice (−0.25 to 0.13). (C) Average GRAB_DA/tdTomato fluorescence ratios aligned to freezing. (D) Average dopamine responses during the first 6 seconds of freezing episodes in Dbh^wt/wt control (95% CI of mean, −0.78 to −0.29), Dbh^hypo (−0.53 to −0.063), and Dbh^LC-null mice (−0.64 to −0.087). Data are mean ± SEM. Two-way analysis of variance with Tukey’s test. GRAB_NE cohort: n = 12 Dbh^wt/wt control (5 female, 7 male), n = 14 Dbh^hypo (7 female, 7 male), and n = 11 Dbh^LC-null (6 female, 5 male) mice. GRAB_DA cohort: n = 18 Dbh^wt/wt control (9 female, 9 male), n = 12 Dbh^hypo (6 female, 6 male), and n = 12 Dbh^LC-null (6 female, 6 male) mice. Avg., average; fluoresce., fluorescence; LC, locus coeruleus; wt, wild-type.

Finally, we performed additional analyses to clarify the specific relationship between freezing behavior and catecholamine release. A repeated measures analysis of freezing behavior and DA and NE release confirmed that the decrease in NE during freezing was specific to the recent test and showed that there were no differences in freezing between the recent and remote tests (Figure S7). We also found no relationship between locomotor activity and NE or DA release (Figure S8). Together with the lack of change in release during immobility in the preexposure phase (Figure S4), this analysis suggests that the decreased release is specific to freezing and not a general consequence of reduced locomotion.

Overall, these data identify unique patterns of NE and DA release dynamics in the dCA1 during the contextual fear assay that are differentially sensitive to partial or complete embryonic loss of NE.

Discussion

Here, we describe 2 new mouse models of disrupted Dbh expression beginning prenatally: a fortuitous hypomorphic allele, Dbh^hypo, with reduced Dbh expression in noradrenergic neurons and an LC conditional knockout, Dbh^LC-null, with total loss of Dbh expression in the LC and reduced expression in other noradrenergic nuclei. Despite reduced Dbh mRNA expression in the adrenal and stellate ganglion of both mutants, NE content in the adrenal and release in the plasma was unchanged. Surprisingly, we did find an accumulation of DA in the adrenal which was not accompanied by any change in Th mRNA expression. Our new mouse lines will be useful tools to uncover the biochemical mechanisms behind these interesting differences and reveal important details of catecholamine biosynthesis in the central and peripheral noradrenergic systems. Regardless, our current findings indicate that monitoring of plasma NE as a proxy for brain NE levels should be approached with caution.

Using our new models of disrupted Dbh expression to link catecholamine dynamics with specific outcomes in the context-dependent fear conditioning paradigm, we found deficits in freezing and unique patterns of catecholamine release dynamics in the dCA1 that are differentially sensitive to partial or complete embryonic loss of NE. During acquisition of the contextual fear memory and the 10 seconds immediately following the shock, NE and DA release was increased in the dCA1 of both wild-type and Dbh^hypo mice. Although we found ∼61% reduction in NE and ∼113% increase in DA tissue content in the hippocampus of Dbh^hypo mice, we observed no significant difference in either NE or DA release compared with wild-type controls. DA release was only significantly increased in the total absence of NE release in the dCA1 of Dbh^LC-null mutants.

The absence of increased DA release in the dCA1 of Dbh^hypo mice, despite the significant increase in tissue content, highlights the tight regulation of release dynamics. The increase in DA tissue content was found in the hippocampus of both Dbh mutants and was inversely proportional to deficits in NE content, suggesting that the increase is a direct result of decreased conversion of DA to NE in the LC. Several studies have shown that LC neurons are capable of releasing DA when artificially stimulated (35,36), but other experiments suggest that midbrain dopaminergic neurons are the most likely natural source of DA in the dorsal CA1 (14). Our data do not reveal the origin of the increased DA release observed in Dbh^LC-null mice. Although it is plausibly due to an accumulation of DA in LC neurons, it could also be an indirect response of midbrain dopaminergic neurons to decreased LC-NE signaling.

The large release of DA and NE in response to the foot shock observed in wild-type mice argues that the dorsal hippocampus is powerfully modulated by the aversive unconditioned stimulus. While the increase in NE release is consistent with previous studies that have examined the LC response to noxious stimuli (37, 38, 39, 40, 41, 42), the increase in both catecholamines has important implications for theories of dorsal hippocampal involvement in contextual fear. Modulation of the hippocampus by shock-induced DA and NE release may serve to strengthen and/or modify the contextual representation, consistent with previous work showing that shock presentation causes remapping of place cells (43,44). In addition, there is some evidence that the shock itself may become integrated into the contextual representation directly, which is thought to play an important role in fear extinction (45), a process that is strongly modulated by LC-NE signaling (41,46).

In Dbh^LC-null mice, the observed increase in DA release during foot shock could contribute to behavioral changes, potentially compensating for the loss of LC-NE in some aspects of behavior and/or exacerbating others. These effects may be complicated by hypersensitivity to DA signaling, a previously reported consequence of Dbh loss (47,48). Methods to selectively manipulate DA and NE during initial formation of the contextual representation, the context-shock association, consolidation, and retrieval will be required to determine which of these phases of learning are most impacted by DA and NE neuromodulation.

The observed reduction in NE release during freezing is interesting considering that increasing LC activity leads to expression of anxiety-related behaviors and is aversive (49,50). A possible unifying explanation comes from findings that lower levels of tonic LC activity enhance signal detection and behavioral performance in attentionally demanding tasks (51). In contrast, higher tonic LC activity may facilitate evaluation of the sensory environment that is more appropriate in new and uncertain conditions (51). Thus, what we may be observing during expression of freezing is a shift from higher tonic activity during exploration of the environment to a lower level of tonic activity that enhances detection of threat (52).

Comparisons of the Dbh mouse models used in the current study with a full-body Dbh^ko (22) indicate that freezing is sensitive to levels of NE and the timing of the insult. The reduced freezing in the recent but not the remote test exhibited by Dbh^hypo mice is similar to the phenotype of the full-body Dbh^ko mice, which also have a deficit in recent but not remote retrieval (53), suggesting that NE is not required at remote time points when contextual fear memories have already gone through a systems consolidation process. However, the deficits in freezing at both recent and remote time points found in Dbh^LC-null mutants may suggest a more critical role for LC-NE in remote retrieval. Alternatively, the data may indicate a more complex role for LC-NE that involves other aspects of the fear response or general sensory information processing.

The phenotypic differences between the mouse models may be at least partially due to differences in NE levels during prenatal development. Dbh^LC-null mice lack LC-NE throughout prenatal and postnatal development, while Dbh^hypo mice have at least partial NE synthesis during embryonic development. Similarly, Dbh^ko mice have low levels of prenatal NE synthesis resulting from prenatal L-DOPS treatment to avoid the embryonic lethal effect of total NE loss (22). It is possible that these early development differences in NE levels differentially affect other important brain regions such as the amygdala and prefrontal cortex. Behavior in the mutants may also be modulated by differences in Dbh expression, and subsequent NE and epinephrine production, in the periphery (54). Peripheral Dbh expression is eliminated postnatally in Dbh^ko mice but present, although at reduced levels, in both of the new models. Alternatively, the behavioral differences among Dbh mutants may result from differences in background strain, the specific molecular defect, or the fear conditioning protocol. Future work will be necessary to untangle these competing hypotheses.

Several sex differences observed in this study (Figure S3) are worth highlighting given the well-established sex differences in NE signaling (55). Wild-type males showed higher NE levels in the hippocampus and higher Dbh expression in adrenal. We also found that Dbh^hypo and wild-type males showed a more robust drop in NE during freezing that was predicted by the length of the freezing bout. In addition, we observed a sex interaction in Dbh^hypo mice wherein males showed similar levels of shock-induced NE release to controls, and females more closely resembled Dbh^LC-null mutants (Figure S2D). Lastly, the predictive correlation between shock-induced DA release and subsequent freezing behavior was driven by the females (Figure S6). These findings suggest a complex relationship between sex, NE, and DA dynamics and partial versus complete loss of NE production. Our findings also highlight the critical importance of conducting sufficiently powered, sex-balanced photometry studies so that these sex interactions can be uncovered. Determining whether the sex differences have a single underlying cause or multiple causes—possibly including sex hormones, epigenetics, differing expression levels of receptors or other genes involved in catecholamine metabolism—will require a focused analysis.