Activity-dependent constraints on catecholamine signaling

Summary

Catecholamine signaling is thought to modulate cognition in an inverted-U relationship, but the mechanisms are unclear. We measured norepinephrine and dopamine release, postsynaptic calcium responses, and interactions between tonic and phasic firing modes under various stimuli and conditions. High tonic activity in vivo depleted catecholamine stores, desensitized postsynaptic responses, and decreased phasic transmission. Together this provides a clearer understanding of the inverted-U relationship, offering insights into psychiatric disorders and neurodegenerative diseases with impaired catecholamine signaling.

Graphical Abstract

Introduction

Norepinephrine (NE) and dopamine (DA) act as major catecholaminergic neuromodulators to regulate arousal states^1–3, attention^4–6, learning & memory ^5,7 and sensory-motor functions^8–10. Interestingly, these catecholaminergic systems exhibit an inverted-U response, between NE neuronal activity in the locus coeruleus (LC) and task performance⁸; between dopamine D1 receptor activation in the prefrontal cortex (PFC) and working memory⁷; and between DA neuronal activity in the ventral tegmental area (VTA) and short-term memory performance¹¹. The mechanisms for performance decline at higher neuronal firing remain incompletely understood. It has been hypothesized from observations at glutamatergic synapses that NE and DA may similarly deplete at higher firing of the respective catecholaminergic neurons, but this hypothesis has not been rigorously tested in vivo. It is also unclear whether the phasic transmission mode is affected by higher tonic firing. The recent development of genetically encoded, fluorescent NE¹⁵ and DA^16,17 sensors can address these questions by overcoming the sensitivity, speed, and spatial limitations of cyclic voltammetry¹⁸

Here, we report that high tonic LC^NE activity quickly depletes NE, decreases the postsynaptic excitability response, and reduces NE release to subsequent phasic firing in vivo. Similarly, high tonic VTA activity can also deplete DA, and reduce DA release to phasic firing in vivo. Moreover, acute stress affects primarily the postsynaptic response and likewise reduces phasic NE release. Together, these findings implicate a possible synaptic mechanism in which catecholamine signaling becomes progressively impaired at higher neuronal firing due to depleted catecholamine stores and desensitized postsynaptic responses, thus providing a more complete understanding of the nonlinear behavioral effects in catecholamine neuromodulation.

Results

To examine how noradrenergic signaling modulates arousal states via downstream effector regions, we chose two LC^NE projections to well-known wake-promoting structures, one to the posterior basal forebrain (BF) and the other to the medial thalamus (Thai)^19,20. As shown in Fig 1A, we virally expressed a cre-inducible yellow fluorescent protein (AAV5-EF1α-DIO-EYFP) in the LC^NE neurons of Dbh-cre mice, and visualized robust labeling of LC axons in the BF and Thai. Next, we measured axonal NE release in vivo by expressing a genetically encoded fluorescent NE sensor (GRAB_NE2m) in BF neurons, and a cre-inducible optogenetic actuator (ChrimsonR) in the LC^NE neurons of Dbh-cre mice; and implanted an optical fiber in the BF for both photometric recording and photostimulation (Fig 1B).

Figure 1. in vivo constraints on norepinephrine signaling.

A. Left, schematic shows cre-inducible EYFP virus injected into the LC^NE of Dbh-cre mice, accompanied by coronal sections showing EYFP expression in the LC. Right, schematic and coronal sections show the basal forebrain (top) and medial thalamus (bottom). Scale bars are 100μm. aca, anterior. CM, centromedian. HDB, horizontal limb of the diagonal band of Broca. IMD, intermediodorsal nucleus. MS, medial septum. MnPO, median preoptic nucleus. PVT, paraventricular thalamus. SI, substantia innominata.

B. Top, schematic and coronal sections show expression of ChrimsonR in LC^NE; and GRAB_NE2m and optical fiber in the basal forebrain (BF). Scale bars are 1mm. HDB, horizontal limb of the diagonal band. LPO, lateral preoptic area. SIB basal portion of substantia innominata. Bottom left, trace shows NE release in response to repeated axonal photostimulation at 10Hz, 50ms pw for 60s at 2min intervals. Bottom right, quantification of AUC normalized to S₀ response (N=8).

C. Left, trace shows NE release with repeated photostimulations at 10Hz, 50ms pw for 60s at 2min and 30min intervals. Right, graph shows corresponding AUC normalized to So response (N=8).

D. Top left, diagram shows the drugs used to alter presynaptic NE metabolism. Top right, graphs show normalized AUC S₀ after multiple days of reserpine treatment (SFig 1C , N=7) and 2h after intraperitoneal (i.p.) administration of 50mg/kg nepicastat (N=3). Bottom left, traces of NE release 20min after i.p. saline (gray) or atomoxetine (red) with repeated 10Hz photostimulations. Bottom right, graph shows changes to AUC normalized to S₀ response (N=3).

E. Top left, schematic shows expression of ChrimsonR in LC^NE, and GRAB_NE2h and optical fibers in BF and Thal. Top right, histological coronal brain slices show viral expressions and optical fiber placements in Thal and BF. Scale bars are 1mm. CM, centromedian nucleus. aca, anterior commissure. HDB, horizontal limb of diagonal band. Bottom left, traces show NE release in response to 10Hz photostimulation in the BF and Thal. Bottom right, plot shows corresponding AUC normalized to S₀ response in BF (N=5) and Thal (N=7), respectively.

F. Top left, schematic shows expression of ChrimsonR in LC^NE, and GCaMP6f and optical fibers in BF and Thal. Top middle and right, histological coronal brain slices show viral expressions and optical fiber placement in Thal and BF. Scale bars are 1mm. CM, centromedian nucleus. aca, anterior commissure. HDB, horizontal limb of diagonal band. Bottom left, traces show postsynaptic activity in response to 10Hz photostimulation in BF and Thal. Bottom right, plot shows AUC normalized to So response in BF (N=6) and Thal (N=6), respectively.

G. Left, average traces show the postsynaptic activity in BF and Thal in response to repeated 10Hz photostimulations 20min after i.p. 10mg/kg prazosin injection. Right, plot shows AUC normalized to S₀ response in BF (N=6) and Thal (N=6), respectively.

*p<0.05, **p<0.01, ***p<0.001. Graphs: mean±SEM; photometry traces: mean±std.

LC^NE tonic frequency typically ranges from 1-6Hz^12,21, but can increase to ~8-10Flz when severely stressed^22,23 Given tonic 10Flz somatic LC^NE activation for 10min has been previously shown to decrease PFC NE², we reasoned that 10Flz may provide a suitable starting frequency to examine NE depletion. When LC^BF axons in unanesthetized mice were repeatedly photostimulated at 10Hz, 50ms pulse width (pw) for 60s at 2min intervals, the initial stimulation (S₀) elicited a robust increase in NE release, but the following stimulations (S₁ and S₂) elicited only 27±2% and 22±2%, respectively, of the initial evoked release (Fig 1B, Supplemental Table 1), consistent with NE depletion from axonal release sites. Repeating stimulation 30min after the depleting stimulation induced 75±3% of the initial NE release (Fig 1C), consistent with NE repletion. Moreover, the observation that S₁ and S₂ NE release was similar suggests the existence of a fast- and a slow-repleting NE pool.

Next, we determined whether pharmacologically altering NE metabolism changes the proportions of these fast- and slow-repleting NE pools. Using reserpine to inhibit vesicular monoamine transporter or nepicastat to inhibit dopamine β-hydroxylase, we found that both treatments significantly reduced NE release (Fig 1D), but did not alter the extent of NE depletion and repletion (SFig 1B,C), indicating that the reduced NE is likely distributed equally amongst the pools. Blocking NE reuptake transporter (NET) with atomoxetine (ATOM) when compared to saline control (SAL), produced a significantly slower decay (T_ATOM 0.006±0.002s versus T_SAL 0.022±0.004s), but not rise, of NE fluorescence after stimulation (SFig 1D). Interestingly, we also found a significantly reduced NE release to S₁ (SAL: 31 ±7% versus ATOM: 9±3%) and S₂ (SAL: 33±9% versus ATOM: 9±4%), indicating that the fast-repleting NE pools rely on NE reuptake; and that the reduction in NE fluorescence to S₁ and S₂ is not due to sensor internalization, as seen in vitro?⁵. Furthermore, when compared to SAL, selective activation of α2-adrenergic receptors (α2-ARs) with dexmedetomidine (DEXMED) and inhibition with atipamezole (ATI) significantly altered S₁ NE release (SAL: 48±7% DEXMED: 22±4%; SAL 26±3%, ATI: 39±2%) and S₂ (SAL: 37±5%, DEXMED: 21 ±3%; SAL 22±3%, ATI: 34±2%, SFig 1E). This result suggests that α2-mediated signaling partially modulates the sizes of fast- and slow-repleting NE pools. As an added control, we showed that 1 mg/kg ATI could antagonize the sedating effect of 50mcg/kg DEXMED (SFig 1F). Combined, these experiments show that although NE depletion cannot be wholly prevented, its extent can be modified by NE metabolism.

To determine if the NE depletion-repletion properties observed in the BF are similar in other brain regions, we next examined NE release in LC^Thal projections using the GRAB_NE2h sensor (which has higher NE affinity than, but comparable on-off kinetics to GRAB_NE2m) and LC^NE-expressed ChrimsonR (Fig 1E). Using repeated photostimulation in BF elicited significantly decreased NE release (S₁ 31 ±6% and S₂ 30±6% of S₀), suggesting that GRAB_NE2h also consistently detect this NE depletion. Interestingly, NE depletion in the Thal (S₁ 53±11% and S₂ 43±9% of S₀) was significantly less than in the BF, suggesting brain region-specific variations in fast- and slow-repleting NE pools.

We next determined whether local NE depletion affects local postsynaptic activity in vivo. Similar to the above strategy, ChrimsonR was expressed in the LC^NE neurons of Dbh-cre mice, but the calcium sensor GCaMP6f was expressed in the BF and Thal (Fig 1F). Repeatedly photostimulating LC^NE axons at 10Hz (50ms pw) for 60s either in the BF or Thal significantly reduced the postsynaptic calcium rise, as shown from S₁ (BF: 53±11%, Thal: 31±6% of S₀) and S₂ (BF: 43±9%, Thal: 30±6% of S₀) stimulations, mirroring the axonal NE depletion. This increased postsynaptic calcium rise was blocked by prazosin (10mg/kg), a selective α1-AR antagonist, but not by propranolol (10mg/kg), a β-AR antagonist (Fig 1G, SFig 1G,H), consistent with the α₁/G_q-mediated signaling cascade that leads to increased intracellular calcium²⁵. It also demonstrates that the postsynaptic calcium rise is indeed primarily mediated by NE release and not by other co-released transmitters.

We next examined whether NE depletes at the higher frequencies of tonic LC^NE firing. Using unanesthetized mice with LC^NE expressing ChrimsonR and BF expressing GRAB_NE2m, we measured NE release during 30min of no stimulation versus axonal photostimulations at 3, 5, and 10Hz (10ms pw). As shown in Fig 2A, the stimulations elicited an initial, frequency-dependent increase in NE release, followed by a slow decline, consistent with NE depletion. When GRAB_NE2m was expressed in the Thal, a similar pattern of frequency-dependent NE release kinetics was observed (Fig 2B). In a similar manner, we also investigated postsynaptic activity in the BF and Thal during the 30min axonal stimulations using GCaMP6f to measure calcium dynamics. Despite the expected higher variability, we observed a transient increase in postsynaptic activity in a frequency-dependent manner in both the BF and Thal (SFig 2A, Fig 2C,D), again mirroring the corresponding patterns of axonal NE release. These experiments, therefore, indicate that NE depletion occurs with prolonged, elevated tonic activity.

Figure 2. Norepinephrine depletion and postsynaptic response to varying stimulation frequency and stress.

A. Left top, schematic shows ChrimsonR in LC^NE, and GRAB_NE and optical fiber in BF. Left bottom, average traces show NE release with 30min photostimulation at 0, 3, 5, and 10Hz 10ms pw. Right top, plot shows maximum NE release in relation to stimulation frequency (N=8). Right bottom, plot shows NE release at baseline (1min before stimulation), early stimulation (1-5min), and late stimulation (25-30min) at different stimulation frequencies (N=8).

B. Left top, schematic shows ChrimsonR in LC^NE, and GRAB_NE and optical fibers in Thal. Left bottom, average traces show NE release with 30min photostimulation at 0, 5, and 10Hz 10ms pw. Right top, plot shows maximum NE release in relation to stimulation frequency (N=7). Right bottom, plot shows NE release at the baseline (1min before stimulation), early stimulation (1-5min), and late stimulation (25-30min) at different stimulation frequencies (N=7).

C. Left top, schematic shows ChrimsonR in LC^NE, GCaMP6f in BF, and implanted optical fiber in BF. Left bottom, average traces show postsynaptic activity with 30min photostimulation at 0, 3, 5, and 10Hz 10ms pw. Right top, plot shows maximum postsynaptic activity in relation to stimulation frequency (N=6). Right bottom, plot shows postsynaptic activity at baseline (1min before stimulation), early stimulation (1-5min), and late stimulation (25-30min) at different frequencies (N=6).

D. Left top, schematic shows ChrimsonR in LC^NE, GCaMP6f in Thal, and implanted optical fiber in Thal. Left bottom, average traces show postsynaptic activity with 30min photostimulation at 0, 5, and 10Hz 10ms pw. Right top, plot shows maximum postsynaptic activity in relation to stimulation frequency (N=6). Right bottom, plot shows postsynaptic activity at the baseline (1min before stimulation), early stimulation (15min), and late stimulation (25-30min) at different frequencies (N=6).

E. Left, schematic shows GCaMP6s in LC^NE, GRAB_NE or GCaMP6f in BF, and tail suspension. Right, plots show corresponding traces of the LC^NE activity (N=6), BF NE release (N=8), and BF postsynaptic activity (N=6), and corresponding AUC in 1-min binning normalized to 1min before 10min tail suspension.

F. Left, schematic shows forced swim. Right, plots show traces of LC^NE activity (N=7), BF NE release (N=6), and BF postsynaptic activity (N=7); and corresponding AUC in 1min binning normalized to 1min before forced swim.

*p<0.05. n.s. not significant. Graphs: mean±SEM; photometry traces: mean±std.

Given the observed NE depletion and postsynaptic desensitization in BF with photostimulation, we then asked whether these effects occur under naturalistic stimulations such as during acute stress. Cohorts expressing either GRAB_NE2m/h or GCaMP6f in the BF were used to determine axonal release of NE and postsynaptic activity, respectively, while another cohort expressing GCaMP6s in LC^NE neurons of Dbh-cre mice was used to determine LC^NE activity. We first tested tail suspension; and observed that the LC^NE activity remained continuously elevated during the suspension, with the NE release gradually plateaued, and the postsynaptic activity initially increased then steadily decreased (Fig 2E). Pre-treating with 10mg/kg prazosin, but not 10mg/kg propranolol or saline, caused a faster decay in postsynaptic activity (SFig 2B). Together, these results indicate that although tail suspension does not deplete axonal NE, the postsynaptic activity diminishes over time, likely due to well-established AR desensitization²⁶. We then tested forced swim, another acute stressor in rodents²⁷, and observed that LC^NE activity initially increased, then decreased sharply after 1min. Interestingly, NE release again steadily increased, while the postsynaptic activity reflected LC^NE activity with an initial increase, followed by a decrease after 1min (Fig 2F). These results show that NE release does not acutely deplete; rather, a decreased postsynaptic response occurred on the order of minutes.

We then sought to determine whether NE depletion from increased tonic LC^NE activity affects the phasic mode of LC^NE transmission. Using ChrimsonR expressed in the LC^NE and GRAB_NE2m/h in the BF, we simulated a range of tonic LC^NE activity by continuously photostimulating LC^BF terminals at 0, 3, 5, and 10Hz for 30min. Concurrently, we simulated phasic activity (typically ~10-20Hz, 200-500ms for LC¹²) by photostimulating at 20Hz, 10ms pw for 200ms every 20s. The progressive increased NE release from higher tonic stimulation frequencies was associated with a progressive decreased release from phasic stimulations (Fig 3A). This inverse relationship suggests that both transmission modes draw from the same NE pools. Additionally, the decreased phasic response is not due to α2-mediated inhibition of release, since ATI pre-treatment did not prevent this decrease (Fig 3B). In a complementary experiment, we further examined simulated phasic NE release during changes in tonic activity from acute stressors. During a 10min tail suspension, for instance, tonic NE release remained continually elevated, resulting in a significant decrease in phasic NE release as compared to baseline (Fig 1C). However, during restraint stress, NE release was only transiently elevated, and the simulated phasic NE release during restraint did not significantly differ from before (SFig 2C). We also stressed the animals with repeated footshocks to elicit strong phasic NE release as shown in Fig 3D. After the initial footshock, the basal NE release increased and remained elevated for the duration of the session. The phasic NE release from repeated footshocks, on the other hand, diminished over the course of the session, indicating that increased NE release from tonic activity limits the release from phasic activity. This decreased phasic NE release was not due to decreasing LC^NE activity over the session (Fig 3D), nor due α2-mediated inhibition of LC^NE activity or NE release (SFig 2D,E). Together, these experiments demonstrate that phasic NE release can be directly influenced by the extent of tonic NE release depending on stressor.

Figure 3. Phasic norepinephrine release constrained by tonic activity.

A. Top left, schematic shows ChrimsonR expressed in LC^NE, GRAB_NE2m in BF, and an implanted optical fiber in BF. Bottom, traces of NE release for 30min of 0, 3, 5, and 10Hz tonic photostimulation (left) with 20Hz, 10ms pw, 200ms phasic stimulation every 20s (right). Top right, plot shows the relation between AUC of phasic NE release and tonic stimulation frequency (N=8).

B. Top, trace shows NE release in response to phasic photostimulations (20Hz, 10ms pw for 200ms every 20s) after 1mg/kg atipamezole treatment, with or without 5Hz tonic photostimulation. Bottom left, averaged traces are shown in response to phasic stimulation at baseline and during 5Hz tonic stimulation. Bottom right, corresponding AUC of phasic NE release is shown (N=8).

C. Top, trace shows NE release in response to 10min tail suspension with 20Hz, 10ms pw, 200ms phasic stimulation. Bottom left, averaged traces are shown in response to phasic stimulation at baseline and during tail suspension. Bottom right, corresponding AUC of phasic NE release is shown (N=8).

D. Top left, schematic shows GCaMP6s in LC^NE, GRAB_NE in BF, and mouse experiencing 10 footshocks. Middle, average traces are shown of LC^NE activity (black) and GRAB_NE in the BF (red) in response to footshocks. Top right, plot shows the AUC of LC^NE activity in response to footshocks (N=7). Bottom, plots show the basal changes to NE release (left, red) and AUC of NE release (right, blue) in response to footshocks (N=8), along with exponential fitting curves and equations.

*p<0.05, **p<0.01, ***p<0.001. Graphs: mean±SEM; photometry traces: mean±std.

Next, we asked whether DA, another catecholamine, has similar constraints on the release and modes of transmission (tonic ~4-5Hz, phasic ~20Hz, <200ms)^13,14 as observed in the LC^NE projections. ChrimsonR was expressed in DA neurons in the VTA of DAT-cre mice; a fluorescent DA sensor GRAB_DA2m was expressed in the nucleus accumbens core (NAc), with an optical fiber placed there (SFig 3A). When repeated axonal photostimulations in unanesthetized mice were applied for 1min at 10Hz, 50ms pw, DA release was robustly elicited with S₀, but subsequently decreased with S₁ (72±7%) and S₂ (69±6%). Increasing stimulation frequency to 20Hz, 10ms pw, showed similar decrease in S₁ (77±6%) and S₂ (75±4%). Photostimulating for 30min at 0, 5, and 10Hz 5ms pw (SFig 3B) also produced increased DA release with increased stimulation frequency, without significant decay over the course of stimulation. DA release using 10ms pw instead as in the LC^NE stimulation did not differ significantly from using 5ms pw at 10Hz (SFig 3C). Furthermore, when we simulated phasic stimulations (20Hz 5ms pw for 200ms every 20s) during various tonic stimulation frequencies, we found that phasic DA release decreased as tonic frequencies increased (SFig 3D), similar to the inverse relation observed for NE release. These experiments also show that although NAc DA depletes at the high tonic activity, the DA pools replete more rapidly than the NE pools we examined. Taken together, these results demonstrate an optimal activity window for catecholamine signaling.

Discussion

In this study, we used the latest catecholamine biosensors to unveil two distinct pools of the respective NE and DA in vivo based on fast and slow repletion kinetics (Fig 1C). The fast-repleting pool constitutes ~25-50% of NE transmission in the brain regions examined from our axonal photoactivation, but constitutes a much larger proportion, ~70-75%, of DA transmission. For NE, the slow repletion takes place on the order of tens of minutes, suggesting that these different pools may translate to differing time scales on which these neuromodulators can act to regulate behavior.

Interestingly, the NE fast repleting pool strongly depends on NET (Fig 1D), shown previously to colocalize with Rab11, a marker of recycling endosomes²⁸, suggesting a potential source for this NE pool. Flowever, it is important to consider several other possibilities for these different pools. NE and DA are reported to reside in both small synaptic vesicles and large dense core vesicles^29–31, and the differing molecular compositions of these vesicles that are thought to explain exocytic differences³², may explain endocytic differences as well in repletion. These pools may also differ in their synaptic localization as readily releasable versus reserve pools^33–35. They may also represent differing release from different axons, as seen before with NE release³⁶, or represent (synaptic) point-to-point versus (non-synaptic) volumetric transmission^37,38. Thus, the molecular identities of these pools remain to be determined. NE depletion from the same stimulus also differed depending on the brain region, likely due to variations in NET and α2-AR levels as observed across LC^NE subpopulations from single-nucleus RNA sequencing²⁴, and may explain the non-homogenous effect that increased LC^NE activity has across the brain³⁹. Previous modeling of striatal DA release also showed three short-term processes regulating release⁴⁰, suggesting possible additional regulations of catecholamine pools. Moreover, given increased evidence of NE and DA co-release from the LC^41–43, together with our observation of more pronounced NE depletion in BF and Thal (Fig 1E) as compared to DA depletion in NAc (SFig 3B), it would be important to determine whether co-released NE depletes more rapidly than DA within the same brain regions.

Furthermore, our observation that NE and DA gradually exhaust their release pools at higher stimulation frequencies, mirrored by a gradual decrease in postsynaptic calcium, provides an overdue plausible mechanistic explanation for the inverted-U relation. Specifically, as phasic activity of LC^NE is associated with focused task performance⁸, our finding that higher tonic NE release from increased tonic activity causes a proportional reduction in phasic-mediated NE release, provides a natural explanation for the decreased behavioral performance at higher LC^NE activity. Interestingly, tail suspension and forced swim, two highly stressful, acute interventions known to increase LC^NE firing³⁴ did not deplete NE, but exhibited instead a gradual decrease in postsynaptic calcium following an acute rise, suggesting that acute stressors do not exhaust brain NE, likely due to negative feedback or asynchronous release kinetics. The postsynaptic calcium response, on the other hand, diminishes in minutes, likely due to desensitization of α₁-ARs^44,45, though downstream signaling may still persist⁴⁶. The temporal pattern of β-AR-mediated signaling, however, remains to be explored. Thus, postsynaptic AR-desensitization appears to be the predominant mechanism in acute stress, and catecholamine depletion may play a more important role in prolonged, elevated neuronal firing.

Moreover, we show that the tonic and phasic LC and DA firing, as two modes of transmission, exhibit an inverse relationship (Fig 3A, SFig 3D), suggesting they draw from the same respective stores. Specifically, in situations where LC tonic activity is high (e.g. acutely stressful events), the NE signal-to-noise from phasic firing is greatly diminished, and thus, contribute to increased exploratory behaviors and decreased task performance^8,47. NE levels from prolonged, elevated LC activity from chronic stress⁴⁸ remains to be explored, and may have implications in psychiatric disorders and treatments. Additionally, the study corroborates the hypothesized impaired phasic activity from increased tonic activity that may help explain neurodegeneration in Alzheimer’s and Parkinson’s diseases⁴⁹, especially as LC likely modulates brain’s waste clearance^50,51.

Limitations of the Study

There are several study limitations to consider. First, although photostimulation was performed at axonal terminals to specify LC projections, the response at terminals may differ from soma-generated action potentials. Second, tonic photostimulation generates synchronized release, whereas tonic LC^NE activity is asynchronous, which may buffer against catecholamine depletion. Third, we used ChrimsonR for photostimulation, an optimized version of the red-shifted excitatory opsin Chrimson that has not been tested for opsin fatigue during prolonged stimulations. However, unlike Chrimson, its faster recovery and its reported ability to support 40Hz photostimulation⁵² suggest that ChrimsonR likely does not suffer from the same degree of opsin fatigue as Chrimson.

In summary, this study demonstrates in vivo constraints on catecholamine signaling at the higher end of neuronal activity, showing axonal catecholamine depletion, postsynaptic desensitization, and overall decreased signal-to-noise in the phasic mode of transmission. These constraints introduce a form of non-linearity in neuromodulation occurring at level of synaptic transmission that may explain the upper end of the inverted-U relationships between catecholaminergic neural activity and cognitive performance^7,8,11, and provide insights into catecholamine-associated psychiatric disorders^48,53–56 and neurodegenerative diseases^57,58.

STAR Methods

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Michael Bruchas (mbruchas@uw.edu).

Materials availability

There are restrictions to the availability of the novel genetically encoded norepinephrine sensors, GRAB_NE2m and GRAB_NE2h, due to Material Transfer Agreement.

Data and code availability

• All photometry data reported in this paper will be shared by the lead contact upon reasonable request.

• All original code is available in the paper’s supplemental information.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and subject details

Mice.

All animal experiments were conducted in accordance with the guidelines and regulation of the National Institutes of Health and the University of Washington under IACUC protocol 4452-01. Adult male and female Dbh-cre C57BL/6J mice (Dbh^{tm3.2(cre)Pjen}, Jackson Laboratory #033951) were group housed on a 12:12 h reverse light-dark cycle and given food and water ad libitum.

Method details

Surgery

Surgeries were performed on mice ages >8 weeks under 1.5-2.0% isoflurane. Mice were injected in the LC (AP −5.45mm, ML +1.1 mm, DV −4.2 to −3.7mm DV to bregma) using a medial-facing beveled 1 μl Hamilton syringe. A blunt tip 1 μl Hamilton syringe was used for injections to the posterior basal forebrain (BF) (AP +0.1 mm, ML +1.3mm,. DV −5.3 to −5.15mm), the medial Thal (AP −1.7mm, ML +1.0mm at 15° angle, DV −3.95 to −3.8mm), the ventral tegmental area (VTA) (AP −3.4mm , ML +0.6mm, DV −4.7 to −4.4mm), and nucleus accumbens core (NAc) (AP +1.4mm, ML +1.5mm at 10° angle, DV −4.1 to −3.8mm). Respective 400 μm optical fibers (Doric Inc., MFC_400/430-0.48_MF2.5_FLT) were placed along the same AP and ML coordinates except DV is −3.8mm for LC, −5.2mm for BF, −3.8mm for Thal, −3.85mm for NAc and secured using Metabond (Parkell #S380). Mice recovered for at least 5 weeks before experiments to allow optimal viral expression.

Viruses

AAV5-EF1α-DIO-EYFP	Addgene #20756
AAV5-Syn-FLEX-rc[ChrimsonR-tdTomato]	Addgene #62723
AAV-DJ-EF1α-DIO-GCaMP6s	Stanford Gene Vector and Viral Core
AAV-DJ-EF1α-DIO-GCaMP6f	Stanford Gene Vector and Viral Core
AAV-hSyn-GRAB_NE2m	Yulong Li (Peking University)
AAV-hSyn-GRAB_NE2h	Yulong Li (Peking University)
AAV9-hSyn-GRAB_DA2m	Addgene #140553

Histology

Mice were euthanized with sodium pentobarbital and transcardially perfused with 4% paraformaldehyde (PFA), post-fixed for 1-3 days in 4% PFA, and then cryo-protected in 30% sucrose. Brain sections (30-100μm) were collected and kept in 0.1M phosphate buffer at 4°C. Then, the sections were mounted with VectaShield Vibrance Hardset mounting medium (Vector Laboratories) with DAPI, and coverslips placed. Images were acquired on an epifluorescent microscope (Leica DFC700T).

Fiber photometry and photostimulation

Fiber photometry recordings of the genetically encoded fluorescent sensors were performed as previously described⁵⁹. Briefly, the implanted fiber was connected to the patch cable via a ferrule sleeve. A real-time processor (TDT RZ5P, sampling rate of 1017.25Hz) recorded the filtered emission (Doric FMC4 filter cube) from the fluorescent sensor at upon excitation at 470nm and 405nm using the accompanied TDT Synapse software. For terminal stimulations, the same photometry fiber served as conduit for 625nm LED photostimulation adjusted for light intensity ~2.5mW/mm² at the optical fiber tip. All photostimulation experiments were done in unanesthetized, freely behaving mice.

Photometry recordings were analyzed using custom python script. Baseline drift from photobleaching artifact was corrected with an exponential decay curve. For GRAB_NE and GCaMP signals, dF/F was calculated as the linear least-squares fit of 405nm signals subtracted from the 470nm signals. The 405nm signals were not used in GRAB_DA as it does not appear to serve as an appropriate isobestic channel. Z-scores were calculated using the mean and standard deviation of a 5-10 min baseline before photostimulation.

Drugs

All drugs were administered intraperitoneally (i.p.). These include reserpine (Sigma 83580) dissolved in 0.2% glacial acetic acid to 0.5 mg/ml; nepicastat (Sigma-Aldrich SML0940) suspended as 5mg/ml in saline mixed with 1.5% DMSO, 1.5% Kolliphor; dexmedetomidine (100mcg/ml Piramal Critical Care, PSLAB-020872-00) diluted to 5mcg/ml in saline, atipazemole made to 0.1mg/ml in saline, atomoxetine hycrochloride (thermoscientific, Cat 467680010) make to 0.3mg/ml in saline, propranolol hydrochloride (Tocris Bioscience Cat. 0624) made to 1mg/ml in sterile water, prazosin hydrochloride (Tocris Bioscience Cat. 0623) made to 1mg/ml in distilled water. Drugs were administered at least 15-20min prior to experiments unless otherwise noted.

Stress behavioral assays

Tail suspension

Mice were suspended by an experimenter holding the tail for 10min ~10in above the floor.

Forced swim

Mice were transferred from a holding chamber into a 5L cylindrical container containing 4L of water (~24°C) and allowed to swim up to 15min. Mice were removed from water if there was evidence of drowning. Mice were subsequently dried on a heating pad for recovery.

Physical restraint

Mice were placed in a conical tube with a strip cut open on top to allow photometry patch cable attachment and the front tapered end cut open to allow nose access. The other end has a lid with a center hole for the tail.

Foot shock

Mice were placed in a Med Associates Fear Conditioning Chamber (NIR-022MD, 29.53cm L x 23.5cm W x 20.96cm H) with a conductive grid floor, a soundproof barrier, and an infrared lighting. They were habituated to the box for 3min before experiencing 1s 0.5mA shocks every min for 10 minutes.

Quantification and statistical analysis

All summary data are expressed as mean±SEM. Statistical significance was denoted as *p<0.05, **p<0.01, ***p<0.001, ****p<0.001 as determined by Student’s t-test, one-way or two-way repeated measure analysis of variance (ANOVA), followed by Dunnett’s post-hoc test. Statistical tests were performed in Excel and in R, and a summary of the statistical tests performed and p-values are shown in Supplemental Table 1

Supplementary Material

Supplemental table - statistics

Supplemental Table 1

Statistical summary is shown for experiments in the main and supplemental figures.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
AAV5-EF1α-DIO-EYFP	Addgene	#20756
AAV5-Syn-FLEX-rc[ChrimsonR-tdTomato]	Addgene	#62723
AAV-DJ-EF1α-DIO-GCaMP6s	Stanford Gene Vector and Viral Core	N/A
AAV-DJ-EF1α-DIO-GCaMP6f	Stanford Gene Vector and Viral Core	N/A
AAV-hSyn-GRAB_NE2m	Yulong Li (Peking University)	N/A
AAV-hSyn-GRAB_NE2h	Yulong Li (Peking University)	N/A
AAV9-hSyn-GRAB_DA2m	Addgene	#140553
Chemicals, peptides, and recombinant proteins
VECTASHIELD Hardset Antifade Mounting Medium with DAPI	Vector Laboratories	CAT#H-1800
Reserpine	Sigma	83580
Nepicastat	Sigma-Aldrich	SML0940
Dexmedetomidine	Piramal Critical Care	PSLAB-020872-00
Atipamezole	Sigma-Aldrich	A9611
Atomoxetine hydrochloride	Thermoscientifc	467680010
Propranolol hydrochloride	Tocris Bioscience	0624
Prazosin hydrochloride	Tocris Bioscience	0623
Experimental models: Organisms/strains
C57BL/6J	Jackson Laboratory	000664
Dbh-cre	Jackson Labarotory	033951
Software and algorithms
FIJI/ImageJ	NIH	https://fiji.sc/
Python 3.0	Python Software Foundation	https://www.python.org/
Illustrator CS8	Adobe	https://www.adobe.com/products/illustrator.html
Excel	Microsoft	N/A
R	GNU project	https://www.r-project.org/
Other
RZ5P fiber photometry system	Tucker-Davis Technologies	N/A
5mm or 6mm 200μm diameter optical fibers	Doric	N/A