Background

Locus coeruleus (LC) is implicated in anxiety and depression (Itoi and Sugimoto, 2010), and is often the first place where Alzheimer’s disease (AD)-related pathology appears (Mather and Harley, 2016). Noradrenaline (NA) is known to inhibit amyloid-induced oxidative stress, and caspase activation in cortex/hippocampus (Counts and Mufson, 2010). Thus, the impairment of LC-noradrenergic system is potentially involved in the AD pathogenesis (Ross et al., 2015; Chen et al., 2022).

It was reported that cytosolic NA in the soma/dendrites of LC neurons is degraded by monoamine oxidase A (MAO-A) into 3,4-dihydroxyphenyl-glycolaldehyde (DOPEGAL) (Burke et al., 1999), which generates apoptotic free radicals (Burke et al., 1998) and activates asparagine endopeptidase (AEP) to produce Aβ and hypephosphorylated tau, consequently leading to AD (Burke et al., 1999; Kang et al., 2020). However, there appears normally no NA except dopamine in the cytosol of LC neurons because NA is produced exclusively in cytoplasmic vesicles by dopamine-β-hydroxylase (DBH) (Potter and Axelrod, 1963) from dopamine which is taken up into cytoplasmic vesicles by vMAT2.

LC neurons release NA following action potential not only from the axon terminals but also from their cell bodies as autocrine (Huang et al., 2007), and autocrine-released NA activates α2 adrenergic receptor (α2-AR) expressed on the soma/dendrites (Aoki et al., 1994), resulting in activation of α2-AR-coupled inwardly rectifying K⁺ (GIRK) channel (Arima et al., 1998) to cause autoinhibition (spike-afterhyperpolarization) (Williams et al., 1985). Thus, overexcitation of LC neurons is normally prevented by the autoinhibition.

As autocrine-released NA is taken up by NA-transporter (NAT) (Torres et al., 2003), uptaken-NA may be the only source of free NA to be metabolized by MAO-A in the cytosol of LC neurons. However, the amount of free NA to be metabolized by MAO-A in the cytosol is thought to be very small due to a much lower affinity of NA to MAO-A compared to vMAT2 (Costa, 2012). Therefore, it is not clear how free NA to be metabolized by MAO-A is increased in the soma/dendites of LC neurons in AD pathogenesis. Because the passive NA-leakage occurs from cytoplasmic vesicles in a manner that achieves a dynamic equilibrium with active NA-storage into cytoplasmic vesicles by vMAT2 (Eisenhofer et al., 2004), increased NA-uptake by NAT in response to increased NA-autocrine following [Ca²⁺]_i increases or overexcitation may increase free NA by facilitating the passive NA-leakage from cytoplasmic vesicles. Furthermore, the affinity of MAO-A for NA may be modifiable as MAO-A activity is Ca²⁺-dependent (Cao et al., 2007). Then, a question arises whether or not impairment of autoinhibition that causes overexcitation and [Ca²⁺]_i increases can consequently increase AEP activity through the production of DOPEGAL.

Neuronal activity in LC neurons is transiently increased by stress (Valentino and Foote, 1988) through activation of corticotropin releasing factor (CRF) receptor (Valentino et al., 1983; McCall et al., 2015) together with glutamate receptors (Valentino et al., 2001). However, even if firing activity is increased by synaptic activation (Barcomb et al., 2022), autoinhibition would occur to suppress overexcitation in LC neurons. Nevertheless, it is known that long-term overexcitation due to chronic stress ultimately causes degeneration of LC neurons (Kitayama et al., 2004). Then, this raises an important question whether and how stress impairs autoinhibition to cause sustained overexcitation. The whole mechanisms for the impairment of LC would be revealed by solving these questions. Here we examined whether the autoinhibition mediated by α2A-AR-coupled GIRK-I is impaired by overexcitation in LC neurons or chronic stress, to clarify the mechanisms for NA accumulation and/or MAO-A activity increase.

Results

Spike frequency adaptation in LC neurons is mediated by α2-AR-coupled GIRK channel

LC neurons invariably display spike-frequency adaptation (Figure 1A, B). We have hypothesized that the autoinhibition would occur following respective spike firing and contribute to the generation of the spike-frequency adaptation and the afterhyperpolarization seen at the offset of current pulses.

Figure 1.

When the cumulative number of spikes was plotted against the elapsed time from the onset of the current pulse every 50 ms, the cumulative spike number increased non-linearly in a saturation manner (Figure 1C). This non-linear relationship was fitted with a saturation curve defined by Monod equation (see Extended Methods and Materials) in order to quantify the spike-frequency adaptation. Because there was no significant difference in b between the two groups of spike trains, the profile of the normalized cumulative spike number vs the elapsed time remains the same regardless of firing frequency, suggesting that spike-frequency adaptation occurs in the same way.

The spike-frequency adaptation observed was almost completely abolished by a blocker of α2-AR, atipamezole (Figure 1D, E), which subsequently largely decreased pulse-afterhyperpolarization (pulse-AHP) (Figure 1D, E, blue and red stars). The cumulative spike number increased non-linearly in a saturation manner in the control condition while it increased almost in a linear manner after atipamezole application (Figure 1F). In the relationship between a and b examined in LC neurons, atipamezole invariably and significantly shifted both a and b in a larger direction, and there was a significant difference in the relationship between a and b (Figure 1G).

Bath application of NA induced a GIRK-I in response to the ramp pulse (Figure 1H, red trace) whereas addition of atipamezole to NA almost completely abolished the NA-induced GIRK-I (Figure 1H, blue trace), showing that GIRK-I reversed at –97 mV. The amplitudes of the inward component of NA-induced GIRK-I were invariably larger than those of the outward component, consistent with the inwardly rectifying property of GIRK-I. Atipamezole significantly decreased the amplitudes of the inward and outward components (Figure 1I). Taken together, NA autocrine is likely to be caused following respective spikes in the LC neuron and NA is accumulated around the LC neuron to activate α2-AR-coupled GIRK channels, to cause spike-frequency adaptation by enhancing AHP.

Ca²⁺-dependent rundown of NA-induced GIRK-I in LC neurons

In PC12 cells, heterologously expressed α2A-ARs undergo internalization following agonist application (Taraviras et al., 2002). Such internalization of α2A-ARs would result in an abolishment of GIRK-I or desensitization given the functional coupling between α2A-AR and GIRK as in LC noradrenergic neurons. However, in LC neurons, NA-induced GIRK-I showed almost no rundown/desensitization after repeating the ramp pulse 20 times in the persistent presence of NA (Figure supplement S1A-D). Then, we next examined whether GIRK-I shows rundown under the condition of “increased excitability” or increased [Ca²⁺]_i, because acute/chronic stress increases the excitability of LC neurons (Borodovitsyna et al., 2018; Campos-Lira et al., 2018) and various channels/receptors are known to be internalized in a Ca²⁺-dependent manner (Beattie et al., 2000; Aziz et al., 2012; Zaccor et al., 2020).

To examine the Ca²⁺-dependency of the rundown of GIRK-I, the respective ramp pulses to evoke GIRK-I were preceded by various pulse trains which would induce Ca²⁺ currents differentially (compare Figure supplement 1F and Figure 2B). After examining various pulse protocols (Figures supplement 1 and 2), the best protocol to induce a prominent rundown of GIRK-I was found as shown in Figure 2B. First, bath application of NA induced GIRK-I (Figure 2A red trace; 2) that showed the inwardly rectifying profile as revealed by subtraction of the control response from that obtained after NA application (Figure 2D, red trace; 2 ̶ 1). Subsequently, the combined command pulse (Figure 2B, upper trace) was applied every minute. The positive pulse trains evoked large inward Ca²⁺ currents in the presence of extracellular 30 mM TEA (Figure 2B, lower traces) whereas similar trains of the same positive pulse in the absence of TEA hardly induced Ca²⁺ currents (Figure supplement 1F, lower trace). With repetition of the combined command pulse 20 times, the NA-induced GIRK-I (Figure 2C, red trace; 2) was decreased to almost the same current as the control response (Figure 2C, blue trace; 3). Indeed, no GIRK-I remained after 20 times repetition of the combined command pulse as revealed by the subtraction between trace 1 and trace 3 (Figure 2D, blue trace; 3 ̶ 1). Both the amplitudes of the inward and outward components of the NA-induced GIRK-I promptly decreased to zero along with repetition of the combined command pulse (Figure 2E, blue and red circles; Figure 2F). Thus, it is likely that NA-induced GIRK-I displays a prominent rundown in a Ca²⁺-dependent manner. Indeed, we further confirmed that these rundowns were completely dependent on the number and application timing of the command pulse in the presence or absence of TEA (see Figure 2 and Figures supplement 1 and 2), which would differentially increase Ca²⁺ transients in amplitude and time course (Figure supplement 3).

Figure 2.

To investigate the effects of the rundown of GIRK-I on the spike-frequency adaptation, spike trains were examined before and after the voltage-clamp experiment (Figure 2A-F). The rundown of GIRK-I invariably resulted in an abolishment of spike-frequency adaptation (Figure supplement 4), similar to the effects of atipamezole on spike-frequency adaptation (Figure 1). Thus, Ca²⁺ increases can cause the rundown of NA-induced GIRK-I and almost abolished the spike-frequency adaptation, leading to an increase in the excitability of LC neurons. These observations suggest that [Ca²⁺]_i increases cause rundown of NA-induced GIRK-I to suppress autoinhibition, leading to excitability increases that cause further increases in [Ca²⁺]_i to further cause a rundown of the GIRK-I.

The agonist-dependency of the rundown of NA-induced GIRK-I was also confirmed with α2A-AR blocker and GIRK blocker; α2-AR blockade but not GIRK blockade prevented the rundown of NA-induced GIRK-I caused by excitability increases (see Figures supplement 5 and 6).

β-Arrestin is involved in the rundown of NA-induced GIRK-I

We further examined whether the internalization of α2-AR is involved in the rundown of NA-induced GIRK-I or not. As β-arrestin/AP2 complex is known to be involved in the agonist-dependent internalization of GPCR (DeGraff et al., 1999; Cottingham et al., 2011), we examined the effects of barbadin, which prevents AP2 from binding with β-arrestin by forming barbadin/AP2 complex, on the rundown of NA-induced GIRK-I. With the patch pipettes filled with barbadin-including internal solution, it was examined whether NA-induced GIRK-I displays rundown in response to the same combined command pulse (Figure 2H, upper trace) as that in the control experiment (Figure 2B, upper trace). The NA-induced GIRK-I (Figure 2G-I, red traces) was hardly decreased (Figure 2H, I, blue traces) after repeating the combined pulse 20 times, in contrast to the control rundown (Figure 2C, D). Barbadin largely suppressed the rundown of the both components of GIRK-I (Figure 2J, K). These results suggest that the rundown of NA-induced GIRK-I is caused by the internalization of α2-AR-coupled GIRK channels.

Restraint stress alters α2A-AR expression in LC neurons

We next investigated whether the expression of α2A-AR is altered in LC neurons by applying restraint stress (RS) to mice. After mice were subjected to daily 12-h RS for 3 days followed by non-stress condition for 2 days, the immunoreactivity for α2A-ARs (see Figure supplement 7) seen along the membrane of TH-positive LC neurons was apparently lower compared to the control (Figure 3A, B). Consistent with this observation, Western blot analysis revealed significantly lower protein levels of α2A-ARs in LC neurons obtained after 3-day RS compared to the control (Figure 3C). Especially when α2A-ARs expression was examined in the membrane fraction identified using Na⁺/K⁺-ATPase, the immunoreactivity for α2A-ARs expression was apparently lower in LC neurons of RS mice than in those of the control mice (Figure 3D, E; *, compare the rightmost lower panels). Consistent with this observation, Western blot analysis revealed a significantly lower protein level of α2A-ARs in the membrane fraction of LC neurons of RS mice compared to the control whereas there was no significant difference in the cytosol between the control and RS mice (Figure 3F, G). Thus, the internalization of α2A-ARs is not necessarily accompanied by increases in cytosolic levels of α2A-ARs because the internalized receptors are either recycled or degraded (Koenig and Ikeda, 1996; Takei et al., 1996).

Figure 3.

RS reduces NA-induced GIRK-I

As 3-day RS caused the internalization or downregulation of α2A-ARs, we next examined if the daily 12-h RS for up to 5 days can differentially cause reductions of NA-induced GIRK-I in LC neurons. NA-induced GIRK-I decreased with an increase in the RS period (Figure 4A-E). The amplitudes of inward and outward components of GIRK-I decreased with the increase in the period of RS (Figure 4F, G), in a way that is described by a saturation function (red interrupted lines). The amplitudes of inward and outward components of GIRK-I in 1-day mice were significantly smaller than those of the control mice, and those in 3-day RS mice were significantly smaller than those in 1-day RS mice while no significant difference in those between 3-day and 5-day RS mice (Figure 4F, G). Thus, the downregulation of GIRK-I was increased as RS period is increased, but in a saturation manner.

Figure 4.

Quantitative PCR analysis revealed that LC neurons expressed much higher mRNA level of α2A-AR than that of α2C-AR (Figure 4H) and also expressed higher mRNA level of GIRK1 compared to GIRK2 while GIRK3 was hardly expressed (Figure 4I). Consistent with the significant decreases in NA-induced GIRK-I in the mice subjected to 3-day RS, mRNA levels of α2A-AR, GIRK1 and GIRK2 in mice subjected to 3-day RS were significantly smaller than those of the control mice (Figure 4J-L). Thus, RS downregulated the expression/production of α2A-AR-coupled GIRK1 prominently in LC neurons.

Histological and biochemical changes in LC neurons after 5-day RS

Next, immunohistochemical experiments demonstrated that 5-day RS appeared to slightly increase the expressions of TH and MAO-A in LC neurons (Figure 5A). Furthermore, AEP expression also appeared to increase in cytosol after 5-day RS compared to the control (Figure 5B). Because TH catalyzes the rate limiting step in the synthesis of catecholamines, if the RS enhances TH activity, free NA concentration in LC neurons would be directly and greatly increased. Therefore, we first carefully examined whether TH is really upregulated in RS mice or not by performing Western blot analysis. No significant increase in the protein level of TH was observed in 5-day RS mice compared to the control (Figure 5C, upper panel; Figure 5D). This result clearly indicates that TH activity was not significantly upregulated in 5-day RS mice, in spite of the TH immunohistochemical experiment. We further examined whether protein levels of MAO-A and AEP are increased or not by 5-day RS. Western blot analysis revealed significant increases in the protein levels of MAO-A (Figure 5C, lower panel; Figure 5E) and of pro-AEP and active AEP (Figure 5F-H) in LC neurons after 5-day RS compared to the control. Western blot analysis also revealed a significant increase in the expression of tau and tau N368 fragment after 5-day RS (Figure 5I-K). Thus, an increased activity of active AEP caused a production of tau N368 fragments as observed at 37 and 35 kDa (Figure 5I), which is consistent with a previous report (Zhang et al., 2014). Relative protein levels of the tau N368 fragments (Figure 5K) were the mean values obtained from the two band intensities, which may reflect the two N-terminal fragments (upper and lower bands) generated by AEP cleavage at Asn-368 of the 4 and 3 repeat-tau isoforms, respectively (Goedert et al., 1989). Finally, Y-maze test revealed that 5-day RS decreased spatial memory (Figure supplement 8).

Figure 5.

Discussion

We have found that LC neurons display spike-frequency adaptation which is caused by the autoinhibition mediated by α2A-AR-coupled GIRK-I following respective spikes (Figure 1) and that α2A-AR-coupled GIRK-I showed Ca²⁺-dependent rundown due to the internalization of α2A-AR (Figure 2, Figure supplement 5A-D and Figure supplement 6E-H), leading to the abolishment of spike-frequency adaptation and the excitability increase (Figure supplement 4). Immunohistochemical examination and Western blot analysis revealed that α2A-AR expression in cell membrane was decreased following RS (Figures 3 and 4J). Consistently, NA-induced GIRK-I was decreased in LC neurons with an increase in the period of RS (Figure 4A-G). Following chronic RS, protein levels of MAO-A, pro-/active-AEP, and tau and tau N368 fragment were significantly increased (Figure 5A-C, E-K).

Based on these results, we propose a novel mechanism for how chronic RS impairs autoinhibition and increases cytosolic free NA to be metabolized by enhanced Ca²⁺-dependent MAO-A activity, to consequently increase AEP activity and produce tau N368 protein (Figure 6). In non-stress mice, autocrine released NA is slowly taken up by NAT after binding to α2-ARs to cause autoinhibition (Figure 6A), while in chronic RS mice, autocrine-released NA may be directly and rapidly taken up by NAT due to the absence (internalization) of α2-ARs (Figure 6B). Such increased rate of re-uptake of NA by NAT would increase free NA concentration by facilitating the leakage of NA from cytoplasmic vesicles in the cell body, which would be metabolized by upregulated MAO-A activity to DOPEGAL, as revealed by the increased activity of AEP in chronic RS mice (Figure 6B).

Figure 6.

Spike frequency adaptation

LC neurons display spike-frequency adaptation. The mechanism for the spike-frequency adaptation may be different among neurons while M currents are usually responsible for the adaptation in hippocampal pyramidal neurons (Aiken et al., 1995). In LC neurons, the spike-frequency adaptation was mediated by α2A-AR-coupled GIRK-I as the spike-frequency adaptation was abolished by atipamezole (Figure 1D-G) and by causing the rundown of GIRK-I (Figure supplement 4). These results suggest that autocrine NA release following respective spikes activated α2A-ARs expressed in the soma/dendrites of LC neurons, and subsequently activated α2A-AR-coupled GIRK-I to cause autoinhibition as spike-frequency adaptation. Thus, spike-frequency adaptation clearly reflects NA autocrine.

Rundown of α2A-AR-coupled GIRK-I

α2A-AR-coupled GIRK-I displayed Ca²⁺-dependent rundown in the presence of agonist. The degree and time course of rundown of α2A-AR-coupled GIRK-I were dependent on the number and timing of positive pulses applied under voltage-clamp condition to cause Ca²⁺ currents (Figure 2A-F and Figures supplement 1E-J, 2, 3). Given the Ca²⁺-dependency of the rundown of autoinhibition, the excitability of LC neurons would increase as LC neurons were excited more such as under RS condition, resulting in a vicious cycle between excitability increase accompanied by [Ca²⁺]_i increase and rundown of autoinhibition/Ca²⁺-dependent internalization of α2A-ARs.

Internalization of α2A-ARs

Activation of GPCR by agonist binding usually causes the formation of GPCR/β-arrestin complexes. Subsequently, the GPCR/β-arrestin complex is assembled through the interaction between β-arrestin and AP2 in clathrin-coated pits (CCPs), which is formed by the association of AP2 with phosphatidylinositol 4,5-bisphosphate (PIP2) at the plasma membrane, leading to endocytosis of GPCRs (Pierce and Lefkowitz, 2001). PIP2 is locally produced from the membrane phosphatidylinositol (PI) by PI4/PIP5 kinases in a Ca²⁺-dependent manner (Sarmento et al., 2014). Barbadin prevented the stable formation of sufficient GPCR/β-arrestin/AP2 complexes in CCPs as barbadin prevents AP2 from binding with β-arrestin by forming barbadin/AP2 complex, thereby hampering GPCR internalization (Beautrait et al., 2017). In the present study, intracellular barbadin effectively suppressed the rundown of GIRK-I (Figure 2G-K), indicating that the rundown of GIRK-I was caused by the internalization of α2A-ARs in LC neurons.

Chronic RS increases the activities of MAO-A and pro-/active-AEP in the cell body

Chronic cold stress reduced α2-AR mediated inhibition of LC neurons (Jedema et al., 2008), and chronic social stress downregulated α2-ARs in the LC (Flugge, 1996). Our present results were consistent with those previous findings: Following RS, α2A-ARs expression was decreased due to the internalization of α2A-ARs (Figure 3), together with the decreases in NA-induced GIRK-I (Figure 4A-G) which mediates the autoinhibition. As long as RS persists, the impairment of autoinhibition leads to the persistent overexcitation in LC neurons. This would further increase NA autocrine, because respective spikes were accompanied by NA autocrine as reflected in spike-frequency adaptation or autoinhibition.

As extracellular NA is mostly taken up into cytosol by NAT (Torres et al., 2003), it is likely that under chronic RS the increased autocrine-released NA is directly and more rapidly taken up by NAT due to the absence of α2A-ARs (Figure 6B). Such an increase in the rate of NA uptake by NAT is similar to the effects of uptake inhibitor of transmitter ligand on the activity of receptors. After re-uptake of NA by NAT, a large portion of NA in the cytosol is restored in cytoplasmic vesicles by vMAT2 while only a small portion of NA can escape from vMAT2 to be metabolized by MAO-A (Eisenhofer et al., 1992), because vMAT2 has a much higher affinity for NA than MAO-A (Bloom, 2006; Costa, 2012). However, there is an additional component of NA to be metabolized by MAO-A, as NA can leak from cytoplasmic vesicles into cytosol (Eisenhofer et al., 2004). When re-uptake of NA by NAT is facilitated due to the internalization of α2-ARs in response to NA autocrine, active NA storage into cytoplasmic vesicles by vMAT2 would increase. Then, the rate of NA leakage from cytoplasmic vesicles would also increase because passive NA leakage from cytoplasmic vesicles and active NA storage into cytoplasmic vesicles by vMAT2 are in a dynamic equilibrium in cytoplasmic vesicles (Eisenhofer et al., 2004). Subsequently, such increases in the rate of NA leakage would result in increases in a MAO-A metabolite, DOPEGAL (Goldstein, 2021) (Figure 6B). Indeed, it has previously been demonstrated that a partial inhibition of NAT in human subjects reduced the original level of a derivative of DOPEGAL (3,4-dihydroxyphenylglycol; DHPG) in CSF by more than 50% (Kielbasa et al., 2015). Thus, free NA to be metabolized by MAO-A would be increased when autocrine-released NA is taken up by NAT. Therefore, under the chronic RS condition, more NA leaked from cytoplasmic vesicles is likely to be metabolized by MAO-A, which activity is enhanced Ca²⁺-dependently and by chronic RS (Cao et al., 2007) (Figures 5C, E and 6B), to DOPEGAL compared to the control condition, as revealed by the increased protein levels of pro-AEP and active-AEP (Figure 5B, F-H). Thus, re-uptaken NA by NAT may largely be degraded to DOPEGAL by the increased activity of MAO-A. Subsequently, DOPEGAL converts pro-AEP into active-AEP to produce tau N368 protein (Figure 5I-K).

It has never been addressed how DOPEGAL production was caused in the cell body of LC neurons while the DOPEGAL production in the cell bodies of LC neurons is well established in AD patients (Burke et al., 1999; Kang et al., 2020) and in transgenic AD model mice (Kang et al., 2020). We demonstrated for the first time that 5-day RS can impair the autoinhibition by causing the internalization of α2A-AR and consequently induces active-AEP to produce tau N368 fragment. Indeed, spatial memory impairment was detected in 5-day RS mice (Figure supplement 8). It is likely that, in addition to NAT inhibitor, the suppression of the internalization of α2A-AR has a strong translational potential for developing drugs that prevent the degeneration of LC.

Limitations of this study

One may argue against the usage of 3-week-old juvenile mice to study the neurodegenerative disorder which is normally found in aged ones. However, a stringent autoinhibition together with potent NAT activity were previously demonstrated in adult rats using intracellular recordings (Williams et al., 1985). Thus, regardless of the age of mice, LC neurons showed the autoinhibition that is mediated by the activity of α2A-AR-coupled GIRK channels. We also confirmed that spike-frequency adaptation reflecting autoinhibition can be induced in adult mice as prominently as in juvenile mice regardless of sexes (Figure supplement 9), although we did not investigate the sex differences in RS effects (Leistner and Menke, 2020). The downregulation of autoinhibition by the internalization of α2A-AR is agonist- and Ca²⁺-dependent regardless of the age of mice. However, it is not known if there is any difference in the susceptibility of α2A-AR internalization to RS between juvenile and aged mice. Thus, a further study may be necessary to examine if autoinhibition in LC neurons is more easily downregulated in aged mice compared to juvenile ones.

It has also been reported that re-uptake of NA by NAT in LC neurons may decrease with age (Sanders et al., 2005; Mitchell et al., 2017). However, NAT expression in LC in 3-week-old juvenile mice was only slightly larger (by 4%) than in adult mice although the difference was significant (Mitchell et al., 2017). Potent NAT activity stringently affecting autoinhibition was reported in adult rats (Williams et al., 1985). Furthermore, partial inhibition of NAT in human subjects largely reduced the original level of a derivative of DOPEGAL in CSF as described above. Thus, based on the present study, it is likely that increased re-uptake of NA by NAT following excessive NA autocrine in the absence of α2A-AR, induced in response to chronic RS, resulted in the increase in active-AEP and subsequent production of tau N368. However, a further experiment to directly assess NAT activity would be necessary to firmly justify our hypothesis.

Conclusion

Our study unveiled the cellular mechanism for the degeneration of LC neurons under chronic RS as follows: Chronic RS impaired the autoinhibition in LC neurons by inducing Ca²⁺- and agonist-dependent internalization of soma/dendritic α2A-AR coupled with GIRK channels. This subsequently caused overexcitation/[Ca²⁺]_i increases and increased Ca²⁺-dependent MAO-A activity and active AEP activity, leading to a production of tau N368 fragment together with an impairment of spatial memory. These results clearly indicate that chronic RS increases cytosolic free NA to be metabolized by MAO-A, presumably by facilitating the re-uptake of autocrine-released NA due to the absence/internalization of α2A-AR. Thus, in addition to NAT inhibitor, the suppression of α2A-AR internalization may have a strong translational potential for developing drugs that prevent the degeneration of LC.

Methods

Restraint stress

Three-weeks-old male C57BL/6 mice (JAPAN HAMAMATSU SLC, Hamamatsu, Japan) were used and housed at 24 ± 2°C. Food and water were available ad libitum. The mice in the stressed groups were restrained in a modified 50 ml clear polystyrene conical centrifuge with multiple air holes for ventilation. The mice were subjected to 12-h restraint stress (RS; 20:00 p.m. to 8:00 a.m. of the next day) for 1, 2, 3 and 5 days. In the previous study, the duration of RS widely ranged from 0.5-h (Gray et al., 2010) to 24-h (Chu et al., 2016) while the RS with the duration less than 12-h has been applied daily for a period (6-h for 2 weeks (Sakaguchi and Nakamura, 1990); 8-h for up to 16 weeks (Sugama et al., 2016); 12-h for 2 days (Zhang et al., 2008)). As the prolonged stress is known to be a potential risk factor for AD (Bisht et al., 2018), the protocol of daily 12-h RS is relevant for the study of functional and cytological disorders of LC neurons. Once the restraint ended, the mice were returned to their home cages with access to food and water ad libitum. The control group remained in their home cages without the restraint procedure.

Slice preparation and electrophysiological recordings

Slices were prepared from 3-weeks-old male C57BL/6 mice to investigate the effects of RS for varying periods on GIRK-I, and from C57BL/6 mice of either sex at 16–23 postnatal days (PD) to investigate the effects of [Ca²⁺]_i increases on GIRK-I, similar to our previous study (Toyoda et al., 2022). Using an Axopatch 200B (MDS Analytical Technologies, Sunnyvale, CA), whole-cell recordings were made from LC neurons that were viewed under Nomarski optics (BX51WI; Olympus, Tokyo, Japan). The internal solution of the patch pipettes had the following ionic composition (in mM): 123 K-gluconate, 18 KCl, 10 NaCl, 2 MgCl₂, 2 ATP-Na₂, 0.3 GTP-Na₃, 10 HEPES, and 0.2 EGTA; pH 7.3. The membrane potential values given in the text were corrected for the junction potential (10 mV) between the internal solution for the whole-cell recording (negative) and the standard extracellular solution. The pipette resistances were 4–6 MΩ. The series resistance was < 10 MΩ. All recordings were made at room temperature (RT). Records of currents and voltages were low-pass filtered at 5 kHz (3-pole Bessel filter), digitized at a sampling rate of 10 kHz (Digidata 1322A, MDS Analytical Technologies) and stored on a computer hard disk. Spike-frequency adaptation was quantitatively estimated by fitting with a saturation curve defined by Monod equation (y = (a * x) / (b + x)) with a saturation level (a) and a half saturation constant (b). The amplitudes of inward and outward components of NA-induced GIRK-I were measured at –130 mV and –60 mV.

Procedure for immunohistochemical examination

C57BL6/J male mice (4-5 weeks) were perfused transcardially with 5 ml of PBS, followed by 10 ml of a freshly prepared 4% paraformaldehyde in PBS. The whole brain including LC (–9 to –10 mm from bregma) was dissected out and immersed in 4% paraformaldehyde solution for 4 h at 4°C fallowed by cryoprotection with 30% (w/w) sucrose in PBS.

Brainstem including LC regions was cut into 8-20 μm-thickness coronal sections on a freezing microtome (Leica Microsystems, Wetzlar, Germany). To examine the distribution of α2A-AR in LC neurons, mouse anti-TH (1:500; sc-25269, Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-sodium potassium ATPase (1:1000; ab76020, Abcam, Cambridge, MA, USA) and goat anti-α2A-AR (1:200; ab45871, Abcam) or rabbit anti-α2A-AR (1:500; RA14110, Neuromics, Northfield, MN, USA) were used as primary antibodies, while cy5 donkey anti-mouse IgG (1:500; cy-2500, Vector Laboratories, Burlingame, CA, USA), FITC donkey anti-rabbit IgG (1:500; 711-095-152, Jackson ImmunoResearch, West Grove, PA, USA) and Cy3 donkey anti-goat IgG (1:500; 706-165-147, Jackson ImmunoResearch) were used as secondary antibodies. We used the affinity purified IgG antibody for α2A-AR immunohistochemistry, in which antibody specificity was ensured as examined previously (Stone et al., 1998; Tan et al., 2009). To examine the expression in LC neurons, mouse anti-TH (1:500) and rabbit anti-MAO-A (1:2000; ab126751, Abcam) were used as primary antibodies, while FITC donkey anti-mouse IgG (1:500; 715-545-150; Jackson ImmunoResearch) for TH and Cy3 donkey anti-rabbit IgG (1:2000; 711-165-152; Jackson ImmunoResearch) were used as secondary antibodies. To examine the distribution of legumain in LC neurons, mouse anti-legumain (1:200; sc-133234, Santa Cruz) and rabbit anti-DBH (1:500; 22806, ImmunoStar, Hudson, WI, USA) were used as primary antibodies, while FITC donkey anti-mouse IgG (1:500; 715-545-150, Jackson ImmunoResearch) and Cy3 donkey anti-rabbit IgG (1:500; 711-165-152, Jackson ImmunoResearch) were used as secondary antibodies. The slices were coverslipped with mounting media (H1000, Vector Laboratories), and sections were visualized with a confocal microscope (LSM980, Carl Zeiss, Oberkochen, Germany) at 10x and 63x magnification.

Western blot study

Brain slices from the control or RS-treated mice were sectioned with vibratome (VT1000P, (Leica Microsystems) to obtain 500 μm-thickness coronal sections containing LC region (bregma –5.34 mm to –5.80 mm). LC dissection was done via micro punch using 1.5 mm biopsy punch (BP-15F, Kai medical, Gifu, Japan) bilaterally under light microscope. Dissected LC were quickly frozen in liquid nitrogen and stored at −80°C until processed. Because LC region was too small to collect sufficient tissue from one mouse for the analysis, the punched LC tissues were collected from two to three mice to obtain a protein lysate sample for each western blotting analysis. LC tissues were homogenized by bead homogenizer in T-PER™ Tissue Protein Extraction Reagent (78510, Thermo Fisher) containing protease and phosphatase inhibitor cocktails (MilliporeSigma, St. Louis, MO, USA). The membrane and cytosol proteins from LC lysate were isolated using Mem-PER Plus Kit (78840, Thermo Fisher) in accordance with the manufacture’s protocol. The protein concentration of the lysate was determined using a Lowry protein assay (Bio-Rad, Hercules, CA, USA). Protein samples (20 µg) were heated to 95°C for 5 min, separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and transferred to a polyvinylidene fluoride membrane using a Transblot SD apparatus (Bio-Rad). After the blots had been washed with tris-buffered saline containing Tween-20 (TBST, 10 mM Tris-HCl, 150 mM NaCl, 0.05% Tween-20), the membranes were blocked with 2% BSA and 4% skim milk/TBST at RT for 1 h and incubated at 4°C overnight with primary antibodies, followed by an anti-rabbit IgG horseradish peroxidase (HRP)-conjugated secondary antibody (1:5,000, 7074; Cell Signaling Technology, Danvers, MA, USA) at RT for 1 h. The western blot membrane was stripped the primary and secondary antibodies and reprobed with the anti-β-actin antibody followed by the anti-rabbit IgG HRP-conjugated secondary antibody. The following primary antibodies were used for the blots: rabbit anti-α2A-AR (1:500; AAR020, alomone), rabbit anti-GAPDH (1:1000; LF-PA0018, Abfrontier, Seoul, Korea), mouse anti-beta actin (1:2000, A5441, MilliporeSigma) and rabbit anti-sodium potassium ATPase (1:1000; ab76020, Abchem) (Figure 3C, F and G); rabbit anti-TH (1:1600; GTX113016, GeneTex, Irvine, CA, USA), rabbit anti-MAO-A (1:6400; ab126751, Abcam), mouse anti-legumain (AEP) (1:400; sc-133234, Santa Cruz), mouse anti-TAU-5 (1:1500, Santa Cruz Biotechnology, SC-58860) and mouse anti-beta actin (1:12800, 66009-1-Ig, Proteintec, Rosemont, IL, USA) (Figure 5C, F and I); rabbit anti-tau AEP-cleaved (N368) (1:5000, ABN1703, MilliporeSigma) and mouse anti-beta actin (1:1000, A5441, MilliporeSigma) (Figure 5I, K). Western blot analysis was performed by measuring band intensity with ImageJ 1.50i software (National Institutes of Health, Bethesda, MD, USA).

Quantitative real-time PCR

cDNA was synthesized with Super Script II Reverse Transcriptase (18064, Invitrogen, Carlsbad, CA, USA) with 1 mg of LC neuron RNA. Using a 7300 Gene Amp PCR system (Applied Biosystems, Carlsbad), quantitative real-time PCR was run in Power SYBR Green PCR master mix (4367656; Life Technologies, Carlsbad) under the following thermal cycling conditions; an initial step of 94°C/5 min, followed by 40 cycles of a series of steps of 94°C/45 s, 61°C/45 s, 72°C/45 s and ended with a series of steps of 95°C/60 s, 60°C/1 min, 95°C/15 s, 60°C/15 s. The following primers were used: α2A (forward GGAATCATGGCTGTGGAGAT, reverse CAGAAATCCCTTCCCTGTCA); α2C (forward TCATCGTTTTCACCGTGGTA, reverse GCTCATTGGCCAGAGAAAAG); GIRK1 (forward ACCCTGGTGGATCTCAAGTG, reverse GGCCACACAGGGAGTGTAGT); GRIK2 (forward CAACCTCAACGGGTTTGTCT, reverse ATCCTACCATGAAGGCGTTG); GIRK3 (forward CGTCTCACCTCTCGTCATCA, reverse CATCCACCAGGTACGAGCTT); GAPDH (forward CCAGAACATCATCCCTGCAT, reverse GCATCGAAGGTGGAAGAGTG).

Behavioral test

The control mice as non-stress group and mice exposed to RS for 1 day, 3 days and 5 days as 1-day, 3-day and 5-day RS groups were tested on the Y maze apparatus, consisted of three identical arms in a “Y” shape made of white acryl (35 cm in length, 5 cm in width, and 20 cm in height) to assess working spatial memory. All tests were performed between 9 AM and noon. Investigators were blinded to the group when performing experiments. Each mouse was placed at the extremity of the “start arm” and left for a training session of 15 minutes, with one of the 2 other arms closed with a white acrylic board (the closed arm was randomly chosen for each mouse). Then, the mouse was brought back to its cage for 5 minutes, the maze was cleaned, the closed arm was opened, and the mouse placed at the extremity of the start arm for a 5-minute test session, under video tracking. The time spent and distance traveled in each arm (start, familiar, and novel arms) were measured and analyzed with Smart 3.0 video tracking software (Panlab, Harvard Apparatus, Holliston, MA, USA). Given their natural curiosity, mice are expected to spend more time exploring the new arm.

Drug application

Atipamezole as a blocker of α2-AR and tertiapin-Q as a GIRK blocker were bath-applied at 5-10 μM and 200 nM, respectively. Barbadin as an arrestin/AP2 blocker was added to the internal solution at a final concentration of 100 μM (Beautrait et al., 2017). All chemicals except tertiapin-Q were purchased from MilliporeSigma unless specified otherwise. Tertiapin-Q was purchased from Tocris Bioscience (Bristol, UK).

Statistical Analysis

Statistical analysis was performed using STATISTICA10J (StatSoft Japan, Tokyo, Japan). Numerical data were expressed as the mean ± SD. The statistical significance was assessed using Student’s t-test, one-way ANOVA followed by post hoc Fisher’s PLSD and Wilk’s lambda. The significance level was set at 5% or less (p < 0.05). Statistical results are given as a p value, unless p < 0.001. All the statistical data contained in this study are presented in Table S1.

Data availability

All data and/or analyses generated during the current study are available from the corresponding author upon request.

Acknowledgements

This research was funded by Japan Society for the Promotion of Science (26290006 and 21K06441 to Y.K.: 20K06926 and 23K06346 to H.T.). Y.K. was supported by Brain Pool Program (NRF) funded by the Korean government (Ministry of Science and ICT). H.T. was supported by Naito Foundation. This research was partly supported by the Basic Science Research Program through the National Research Foundation of Korea grants (NRF-2018R1A5A2024418 and NRF-2021R1A2C3003334 to S.B.O) funded by the Korean government (Ministry of Science and ICT).

Additional information

Author Contributions

YK, HT, and SBO conceived and designed the study. HT, DK, BGK, and TS performed experiments. HT, DK, BGK, TS, TK, and YK analyzed data. YK, HT and SBO wrote the manuscript.

Ethics

This study was conducted following the Helsinki Declaration, and carried out according to the ARRIVE guidelines. All experimental procedures were approved and performed in accordance with the relevant guidelines of the Ethical Guidelines Committee at Osaka University, Seoul National University and Kyushu University.

Additional files

Supplemental figures