Octopamine metabolically reprograms astrocytes to confer neuroprotection against α-synuclein

Significance

Octopamine is a well-established neurotransmitter in invertebrates involved in flight or fight responses. In mammals, its function has remained elusive given that its role was replaced by epinephrine. Here, we uncover a unique function of octopamine in the mammalian brain as a key metabolite diving the switch from protective to toxic astrocytes by harnessing their metabolism toward aerobic glycolysis. Pathological alterations in octopamine levels have been found in α-synucleinopathies such as Parkinson’s disease, as well as a range of psychiatric diseases including schizophrenia and bipolar disorder. Therefore, our work has therapeutic implications since we show that pharmacological modulation of octopamine pathway protects neurons against α-synuclein neurodegeneration.

Abstract

Octopamine is a well-established invertebrate neurotransmitter involved in fight or flight responses. In mammals, its function was replaced by epinephrine. Nevertheless, it is present at trace amounts and can modulate the release of monoamine neurotransmitters by a yet unidentified mechanism. Here, through a multidisciplinary approach utilizing in vitro and in vivo models of α-synucleinopathy, we uncovered an unprecedented role for octopamine in driving the conversion from toxic to neuroprotective astrocytes in the cerebral cortex by fostering aerobic glycolysis. Physiological levels of neuron-derived octopamine act on astrocytes via a trace amine-associated receptor 1–Orai1–Ca²⁺–calcineurin-mediated signaling pathway to stimulate lactate secretion. Lactate uptake in neurons via the monocarboxylase transporter 2–calcineurin-dependent pathway increases ATP and prevents neurodegeneration. Pathological increases of octopamine caused by α-synuclein halt lactate production in astrocytes and short-circuits the metabolic communication to neurons. Our work provides a unique function of octopamine as a modulator of astrocyte metabolism and subsequent neuroprotection with implications to α-synucleinopathies.

Astrocytes, the most abundant glial cells in many parts of the central nervous system (1), play active roles in maintaining the neuronal connectome and homeostasis. They regulate synaptic transmission and neuronal processing by recycling neurotransmitters from the synaptic space (2, 3); are essential partners for synaptogenesis, synapse function, and synaptic plasticity (3, 4); play a key role in the adaptive–protective response against oxidative stress (5, 6); and participate in neuronal energy metabolism by shuttling lactate to neurons during memory and learning (7–9). While astrocyte activity is necessary to maintain neuronal health and protect neurons from toxic insults, excessive astrocyte activity (also known as reactive astrocytosis), triggered by high and sustained cytosolic Ca²⁺, has been shown to have detrimental effects on neurons (10–12). Although much of our current understanding has focused on defining transcriptional signatures as well as the morphological states associated with toxic astrocytes (inflammatory, fibrous morphology, A1) vs. protective astrocytes (antiinflammatory, protoplasmic morphology, A2) (13), recent in vivo studies suggest that these signatures do not provide a full accurate description (14). Therefore, elucidating the extracellular signaling molecules as well as the molecular signaling mechanisms which underlie the conversion of astrocytes from neuroprotective to neurotoxic might hold important clues to understand the pathoprogression of neurons in neurodegenerative diseases.

Aggregation of α-synuclein (α-syn), a small lipid-binding protein, plays a central role in several neurological diseases collectively known as synucleinopathies, which include Parkinson’s disease (PD) and dementia with Lewy bodies (DLB) (15). Several studies have provided strong evidence supporting a causative role of high cytosolic Ca²⁺ as a key pathological feature in PD (16–20). A key downstream protein responsible for sensing Ca²⁺ levels and transducing them into cellular responses is the highly evolutionarily conserved Ca²⁺-dependent serine/threonine phosphatase, calcineurin (21). We and others have shown that reducing calcineurin activity in neurons confers neuroprotection against α-syn in models of PD and DLB (17–20). A hallmark of these synucleinopathies is astrocytosis (10). Further, pathological levels of Ca²⁺ have been shown to be a key feature of promoting astrocytosis (22). Whether the neuroprotective effects of reducing calcineurin activity against the toxic effects of α-syn are mediated by astrocytes are unknown.

Here, using transcriptomic and metabolomic analyses, along with functional assays in primary cortical neuron–astrocyte cocultures, cortical and hippocampal astrocyte-only cultures, ex vivo brain slices, as well as in a rodent model of α-syn pathobiology, we uncovered a central role of calcineurin activity in mediating the conversion of astrocytes from neuroprotective to neurotoxic by enabling lactate exchange to neurons via the trace amine, octopamine. We found that reducing calcineurin activity in neurons undergoing α-syn proteotoxic stress leads to physiological release of octopamine. Octopamine acts on astrocytes to metabolically rewire them toward lactate production and secretion. Lactate import into neurons via the monocarboxylase transporter2 (MCT2) is used as an energy substrate to increase the ATP deficits caused by α-syn and prevent neurodegeneration. Conversely, we found that if high calcineurin activity is not attenuated in neurons, it leads to neuronal release of high pathological levels of octopamine, which prevents lactate production in astrocytes, short-circuiting the metabolic communication to neurons, and thus driving neurodegeneration. Our work has elucidated an unprecedented role for octopamine in mammals as a key modulator of toxic vs. protective astrocytes by harnessing their metabolism toward aerobic glycolysis and hence neuroprotection with therapeutic implications to α-synucleinopathies.

Results

Partial Inhibition of Calcineurin Stimulates Lactate Production, Which Protects Against α-Syn Proteotoxicity.

We leveraged well-established models of α-synucleinopathy to understand the contribution of calcineurin activity to astrocytosis in the cerebral cortex (17, 18, 23). Primary rat cortical cultures, consisting of both glutamatergic neurons and astrocytes (hereafter referred to as neuron–astrocyte cocultures), were transduced with a lentivirus-expressing human α-syn A53T, a mutation causing an autosomal-dominant form of PD (24) (hereafter referred to as α-syn) or empty vector as control. In these neuron–astrocyte cocultures, α-syn expression was restricted to neurons where it is endogenously expressed (25, 26), via the regulation of the synapsin promoter. After transduction, neuron/astrocyte cocultures were treated with a previously established subsaturating neuroprotective dose of the calcineurin-specific inhibitor, FK506 (17, 18), or with vehicle as control. Differences in neuron viability were assessed by counting the number of microtubule-associated protein 2 (MAP2)-positive cells (SI Appendix, Fig. S1 A and B). Consistent with our previous findings, α-syn caused approximately 50% neuron loss at 5 wk posttransduction (17, 18). Also consistent with our previous findings, decreasing calcineurin activity with FK506 protected neurons against α-syn toxicity without having an effect on control neuron–astrocyte cocultures (SI Appendix, Fig. S1 A and B). The neuronal rescue in these neuron–astrocyte cocultures was also accompanied by a reduction in phosphorylated α-syn S129, a posttranslationally modified form of α-syn associated with pathology (27) (SI Appendix, Fig. S1 C–E). Importantly, the neuroprotective effects of FK506 were not due to a decrease in α-syn or calcineurin expression (SI Appendix, Fig. S1 C–F).

We next asked whether expression of α-syn can cause reactive astrogliosis, a hallmark of neuronal stress/damage, previously unexplored in this model. Expression of α-syn in the neuron–astrocyte cocultures led to an increase in glial fibrillary acidic protein (GFAP) expression, a marker for reactive astrocytes in the neighboring astrocytes (SI Appendix, Fig. S1 A and B). However, treatment with FK506 decreased GFAP expression without affecting the number of astrocytes (SI Appendix, Fig. S1 A, B, and G). Importantly, we detected no differences in α-syn concentrations in the supernatants of these neuron–astrocyte cocultures (SI Appendix, Fig. S1H). These data indicate that astrocyte reactivity is a response to the proteotoxic stress caused by α-syn expression in neurons and not due to a local effect of α-syn expression in astrocytes.

To investigate whether calcineurin activation also played a role in astrocyte reactivity in vivo, we used a mouse model of α-syn pathobiology. In this synucleinopathy model, human α-syn A53T is driven by the Ca²⁺/calmodulin-dependent kinase II (CaMKII)–tTA promoter and is therefore highly expressed in the cerebral cortex, a brain region highly affected in DLB (23). Control and α-syn-expressing animals were treated with vehicle or two single doses of FK506 4 d apart for a final brain content of 40 ng/g (2 ng/mL), a dose range below the standard saturating inhibitory calcineurin doses (28) (SI Appendix, Fig. S1I). In the cerebral cortex, treatment with FK506 was accompanied by a reduction in phosphorylated α-syn S129, the posttranslationally modified form of α-syn associated with pathology (27), without alteration in the expression of α-syn or calcineurin (Fig. 1 A and B and SI Appendix, Fig. S1 J–L). As expected, α-syn caused an increase in GFAP expression compared to control animals (Fig. 1 C and D). However, treatment with FK506 decreased GFAP expression in α-syn transgenic animals without affecting the total number of astrocytes (Fig. 1 C–E).

Fig. 1.

Partial inhibition of calcineurin induces neuroprotection by promoting lactate synthesis, which prevents α-syn toxicity in neurons and astrocytes. (A) Representative western blot for α-syn, phosphoserine-129 α-syn, and calcineurin from cerebral cortex lysates of CaMKII–Cre (control) and CaMKII–Cre–α-syn (α-syn) animals injected twice with FK506 or vehicle (Dimethyl Sulfoxide or DMSO) 4 d apart and killed at day six. Actin serves as loading control. (B) Densitometry quantitation from the WB in A of p-S129 α-syn/α-syn/actin. Data are normalized to control cultures treated with vehicle (DMSO). N = 3; **P < 0.01; ****P < 0.0001; one-way ANOVA, post hoc Tukey test. (C) Representative immunohistochemistry for GFAP of matched sections from cerebral cortex of animals in A. (Scale bar is 20 µm.) (D) Quantification of GFAP fluorescence intensity (F.I) and (E) GFAP-positive nuclei from animals in A N ≥ 5; n.s =nonsignificant; *P < 0.05; one-way ANOVA, post hoc Tukey test. (F) Lactate metabolite from rat primary cortical neuron–astrocyte cocultures transduced with either control lentivirus or α-synA53T driven by the synapsin promoter and treated with either vehicle (DMSO) or subsaturating (0.2 µM) doses of FK506 and processed for 5 to 6 wk post transduction. Representative data from 3 independent experiments run in duplicates; *P < 0.05; one-way ANOVA, post hoc Tukey test. (G) Lactate metabolite from cerebral cortex of control and α-syn transgenic animals in A treated with FK506, N ≥ 5; *P < 0.05; one-way ANOVA, post hoc Tukey test. (H) ATP content from rat primary cortical neuron–astrocyte cocultures described in (F) treated with exogenous lactate, FK506, or both 5 to 6 wk post transduction. Raw values were normalized to control—DMSO. N = 6 to 8; *P < 0.05; unpaired t test with Bonferroni correction. (I) Representative immunofluorescence image for MAP2+ and GFAP+ from rat primary cortical neuron–astrocyte cocultures in (H); (scale bar, 50 μM.) (J) GFAP-positive fluorescence intensity (F.I)/nuclei from cortical neuron–astrocyte cocultures in (H) normalized to control—DMSO-treated conditions. N = 3; 300 to 600 cells/biological replicate; ***P < 0.001; one-way ANOVA, post hoc Tukey test. (K) Number of MAP2-positive neurons normalized to control—DMSO-treated conditions described in H. N = 6 to 8; ***P < 0.001, one-way ANOVA, post hoc Tukey test.

To elucidate the downstream effectors of calcineurin that convey protective responses, we took a transcriptional approach using RNA sequencing (RNA-Seq) (29). Neuron–astrocyte cocultures transduced with neuronal expressed α-syn were treated with neuroprotective subsaturating doses of FK506 (0.2 µM). As controls, we utilized previously established nonprotective saturating doses of FK506 (2 µM) (17, 18) or vehicle alone, and RNA was extracted before the onset of α-syn toxicity. In these neuron–astrocyte cocultures, α-syn expression caused a relatively modest effect on transcription, with only 233 genes being significantly differentially expressed when compared to control (SI Appendix, Fig. S2 A and B and Dataset S1). The down-regulated genes were associated with pathways which are known to be impaired by α-syn, including ion transport, synaptic signaling, and vesicular transport (30–32) (SI Appendix, Fig. S2B). The up-regulated genes were associated with nuclear factor-kappa-light-chain enhancer of activated B cell (NF-ΚB) signaling and apoptotic processes (SI Appendix, Fig. S2B), consistent with inflammatory and proapoptotic responses also previously associated with α-syn toxicity (33). To understand the biological nature of the transcriptional changes across drug conditions, we performed a cluster analysis on the union of differentially expressed genes found in each condition compared to vehicle-treated control neuron–astrocyte cocultures (SI Appendix, Fig. S2A). In neuron–astrocyte cocultures, saturating and subsaturating doses of FK506 caused a similar and moderate effect relative to the vehicle control. In sharp contrast and in response to α-syn, only the neuroprotective subsaturating doses of FK506 resulted in a striking effect on transcription, with 5,164 genes being significantly differentially expressed when compared to both control and α-syn. FK506’s effects on α-syn were specific to the neuroprotective dose since the saturating nonneuroprotective doses caused the opposite effect. Importantly, the robust transcriptional signature caused by FK506 was not due to an increase in cell number or changes in the ratios of neurons and astrocytes compared to control treated with FK506 (SI Appendix, Fig. S1G).

The most prominent down-regulated genes in α-syn neuron–astrocyte cocultures treated with subsaturating neuroprotective doses of FK506 included those belonging to amine and ATP synthesis-coupled transport. In contrast, the most prominent up-regulated genes corresponded to axonal fasciculation and insulin-like growth factor (IGF-1) receptor signaling pathway. Upregulation of prosurvival and IGF-1 signaling pathway genes served as a proof of principle that the RNA-Seq results support a protective program in response to partial inhibition of calcineurin in α-syn (SI Appendix, Fig. S2 A and C–J). IGF-1 signaling pathway has also been shown to be neuroprotective at least partially due to inhibition of neuroinflammation (34, 35). To validate these RNA-Seq-based genes, we preformed real-time PCR (qPCR) for IGF-1 in the neuron–astrocyte cocultures and in the cerebral cortex of α-syn transgenic animals. α-Syn caused a modest but significant transcriptional increase in IGF-1 levels compared to that of control, with a concordant increase in IGF-1 protein levels in both neuron–astrocyte cocultures and transgenic animals (SI Appendix, Fig. S2 K–M). However, treatment with FK506 in α-syn-expressing neuron–astrocyte cocultures and transgenic animals further increased mRNA and protein IGF-1 levels (SI Appendix, Fig. S2 K–M).

Upregulation of genes involved in prosurvival pathways due to partial inhibition of calcineurin agreed with a neuroprotective program against α-syn. The surprising part was that the transcriptional neuroprotective program also included downregulation of mitochondrial complex I subunit genes critical for ATP synthesis-coupled transport and oxidative phosphorylation, which neurons rely on (SI Appendix, Fig. S2A and Dataset S1). Downregulation of mitochondrial complex I can compromise aerobic respiration and lead to a shift toward glycolysis. We thus performed metabolomics to investigate whether a shift from oxidative phosphorylation to glycolysis was taking place. Reduction of calcineurin activity with FK506 in α-syn neuron–astrocyte cocultures caused a significant reduction in all tricarboxylic acid cycle (TCA) metabolites with a modest effect in control neuron–astrocyte cocultures (SI Appendix, Fig. S3 A and B and Dataset S2). Reduction of calcineurin activity in the cerebral cortex of FK506-treated α-syn transgenic mice caused a significant reduction of only phosphoenol pyruvate (SI Appendix, Fig. S3C and Dataset S2). However, one metabolite that was consistently elevated in FK506-treated α-syn expressing both neuron–astrocyte co-cultures and the cerebral cortex was lactate (Fig. 1 F and G and Dataset S2). An increase in lactate together with a decrease in TCA metabolites suggests a shift in respiration toward glycolysis. To confirm the shift in metabolism caused by FK506 in α-syn-expressing neuron–astrocyte cocultures, we measured oxygen consumption rates in these cultures (SI Appendix, Fig. S4 A–G). Respiration measurements were not significantly changed across conditions and treatments. This result is perhaps not surprising considering that cells can be heterozygous for many complex I or electron transport chain genes without having a drastic effect on basal oxygen consumption (36). However, there was a trend toward an increase in the extracellular acidification rate, a surrogate index of glycolytic flux in the α-syn neuron–astrocyte cocultures treated with FK506 (SI Appendix, Fig. S4G). Consistent with this trend by both RNA-Seq and qPCR, α-syn neuron–astrocyte cocultures treated with FK506 expressed higher levels of hypoxia-inducible transcription factor, responsible for the upregulation of glycolytic genes (37) (SI Appendix, Fig. S4H and Dataset S1). Moreover, in both α-syn-expressing neuron-astrocyte cocultures and the cerebral cortex of transgenic mice treated with FK506, expressed higher levels of lactate dehydrogenase 5 (LDH5), one of the five LDH isoenzymes critically involved in promoting glycolysis and highly expressed in astrocytes (38) (SI Appendix, Fig. S4 I and J and Dataset S1).

Aerobic glycolysis is a well-established metabolic mechanism by which cancer cells sustain rapid production of energy and metabolic intermediates that allows their survival and proliferation (39). In the brain, astrocytes have been shown to generate lactate through aerobic glycolysis to provide neurons with a fast source of energy to cope with the high energetic demands posed during memory and learning (7, 40). To test whether lactate production is the mechanism by which partial inhibition of calcineurin confers neuroprotection against α-syn toxicity, control and α-syn-expressing neuron–astrocyte cocultures were treated with a range of lactate concentrations from nanomolar to millimolar and assayed for ATP content as a surrogate for viability. As previously established (17, 18), α-syn caused a reduction in ATP relative to control; however, treatment with 1 and 5 mM exogenous lactate rescued the ATP deficits (Fig. 1H and SI Appendix, Fig. S4K). Importantly, the exogenous lactate concentrations that rescued ATP are physiologic concentrations of lactate in mammals, which typically range between 1 and 5 mM in basal state (41, 42). The protective effects of 1mM lactate were also accompanied by an increase in total number of neurons as well as by a decrease in astrocyte reactivity (Fig. 1 I–K and SI Appendix, Fig. S4 L–N). The protective effects of lactate treatment were calcineurin dependent since cotreatment with FK506 and lactate showed an epistatic relationship in α-syn-expressing neuron–astrocyte cocultures, without any evidence of additive effect regarding the number of neurons and astrocyte reactivity (Fig. 1 I–K and SI Appendix, Fig. S4 L–N). Together, our data indicate that partial inhibition of calcineurin under α-syn proteotoxic stress transcriptionally rewires metabolism to promote aerobic glycolysis and lactate-dependent neuronal survival.

MCT2 Expression Regulated by Calcineurin Mediates Lactate Neuronal Survival and Astrogliosis in α-Syn-Expressing Cultures.

Monocarboxylase transporters (MCTs) are responsible for transporting lactate across membranes. RNA-Seq and qPCR data from the neuron–astrocyte cocultures showed that α-syn caused a decrease in the expression of the neuronal selective MCT2 (43, 44) when compared to control (SI Appendix, Fig. S5A and Dataset S1). Further, treatment with FK506 in both α-syn-expressing neuron–astrocyte cocultures and α-syn transgenic mice increased MCT2 levels by qPCR and immunostaining (Fig. 2 A and B, SI Appendix, Fig. S5 B and C, and Dataset S1). To determine whether regulation of MCT2 by calcineurin activity is the mechanism underlying lactate import in neurons to protect against α-syn toxicity, we took a knockdown approach. Control and α-syn-expressing neuron–astrocyte cocultures were cotransduced with an inducible shRNA targeting rat MCT2 or scrambled sequence as control (50% reduction levels of rat MCT2, SI Appendix, Fig. S6A) and treated with exogenous lactate, FK506, or combination of both. Reduction of MCT2 in the neuron–astrocyte cocultures had a marginal effect on neuronal and astrocyte health assayed by three distinct measurements: ATP content, total number of MAP2+ neurons, and GFAP expression (SI Appendix, Fig. S6 B–E). However, reduction of MCT2 in α-syn-expressing neuron–astrocyte cocultures abolished the protective effects of FK506, lactate, or the combination of both (Fig. 2 C–G). Notably, the lack of neuronal rescue by lactate and FK506 due to reduction of MCT2 in α-syn-expressing neuron–astrocyte cocultures led to an increase in astrocyte reactivity (Fig. 2 D and G). This result indicates that the protective effect of lactate relies on neurons and that the reactive state of astrocytes depends on the neuronal health. To investigate whether MCT2 expression was altered in human disease, we examined MCT2 expression in fixed postmortem tissue from humans diagnosed with two common synucleinopathies with prominent pathology in the cerebral cortex and cognitive deficits: DLB and concurrent PD and DLB. Superior frontal cortex from four to five disease cases and age-matched nondiseased controls were immunostained for MCT2 (Fig. 2 H and I and SI Appendix, Fig. S7). While most controls exhibited high MCT2 staining in the superior frontal cortex, MCT2 immunoreactivity was decreased in DLB and PD/DLB cases, consistent with a downregulation of MCT2 expression. Together, our results support an association between MCT2 expression and α-syn pathology in human disease and indicate that calcineurin regulates MCT2 expression in neurons to enable lactate import and survival against α-syn proteotoxicity.

Fig. 2.

MCT2 expression regulated by calcineurin mediates lactate neuronal survival and astrogliosis in α-syn-expressing cultures. (A) Representative immunohistochemistry for MCT2 from cerebral cortex of matched sections of CaMKII–Cre (control) and CaMKII–Cre–α-syn (α-syn) animals injected twice with FK506 or vehicle (DMSO) 4 d apart and killed at day six. [Scale bar is 50 µm (40×).] (B) Quantification of average MCT2 fluorescence intensity (F.I) per cell from A. N ≥ 5; **P < 0.01; one-way ANOVA, post hoc Tukey test. Data were normalized to control vehicle (DMSO). (C and D) Representative immunofluorescence for MAP2+ (C) and for GFAP+ (D) from rat primary cortical neuron–astrocyte cocultures cotransduced with either control lentivirus or α-synA53T driven by the synapsin promoter and a doxycycline-inducible shRNA for MCT2 or shRNA scrambled sequence as control. All cultures were treated with either vehicle (DMSO), subsaturating doses of FK506 (0.2 µM), lactate (1 mM), or combination of both lactate and FK506 once a week for 5 wk post transduction. (Scale bar, 50 μM.) These cultures were quantified for number of MAP2+ cells for neurons (E), ATP content (F) and GFAP fluorescence intensity (F.I)/nuclei (G). Data were normalized to control-ShCtrl-DMSO condition. N = 4 to 6; *P < 0.05; **P < 0.01; ****P < 0.0001; unpaired t test with Bonferroni correction. (H) Representative immunohistochemistry for MCT2 in the superior frontal cortex area in human DLB and PD/DLB cases and age-matched controls. An average of 5 sections from 5 control and 6 disease cases were analyzed. [Scale bar is 50 µm (40×).] (I) MCT2 intensity was measured per cell from H. N = 5 to 6; **P < 0.01; one-way ANOVA, post hoc Dunnett’s multiple comparison test.

Octopamine Mediates Ca²⁺ Influx in Astrocytes via the G-Protein-Coupled Receptor Trace Amine-Associated Receptor 1 (TAAR1) and Orai1 Ca²⁺ Channel.

Decreasing calcineurin activity in α-syn-expressing neuron–astrocyte cocultures promotes lactate production and survival (Figs. 1 and 2). Neurons are mostly oxidative, whereas glial cells, namely astrocytes and oligodendrocytes, are predominantly glycolytic and hence lactate producers (45, 46). This suggests that the astrocytes in these neuron–astrocyte cocultures are responsible for the production of lactate. But, how do astrocytes sense the need from neurons to produce lactate in response to α-syn proteotoxic stress in a calcineurin-dependent manner? To elucidate the factor that mediates the metabolic exchange between astrocytes and neurons, we performed metabolomics on the supernatants from the neuron–astrocyte cocultures (Fig. 3A and Dataset S3). We detected a total of 186 metabolites across conditions, with 9 being significantly different between control and α-syn-expressing neuron–astrocyte cocultures. Of these 9 metabolites, enriched for catecholamine synthesis, 3 reverted to control when treated with subsaturating protective doses of FK506: octopamine, tryptamine, and DOPET (3,4-dihydroxyphenylethanol) (Fig. 3A). However, only octopamine significantly changed in a manner corresponding to calcineurin toxic vs. protective activity: high amounts of octopamine under toxic levels of calcineurin activity [both, no drug treatment, corresponding to high calcineurin activity as previously described (17, 18), and 2 µM FK506, corresponding to complete inhibition of calcineurin activity as previously described (17, 18)] and lower amounts of octopamine under protective levels of calcineurin activity [0.2 µM FK506, corresponding to partial inhibition of calcineurin activity as previously described (17, 18)].

Fig. 3.

Octopamine is a calcineurin-dependent metabolite released under α-syn proteotoxicity that mediates Ca²⁺ influx in astrocytes. (A) Metabolite screen from supernatants of rat primary cortical neuron–astrocyte cocultures transduced with either control lentivirus or α-synA53T driven by the synapsin promoter and treated with either vehicle (DMSO) or subsaturating (0.2 µM) doses of FK506 5 to 6 wk posttransduction. Metabolites that were significantly different (P < 0.05) between control and α-syn cultures were enriched for catecholamine synthesis by metabolite set enrichment analysis from MetaboAnalyst (McGill University). From these, only 3 hits were significantly different between α-syn and α-syn + FK506: octopamine, tryptamine, and DOPET. Representative data from 3 independent experiments run in duplicates; *P < 0.05; one-way ANOVA, post hoc Dunnett’s multiple comparison test. (B) Ca²⁺ imaging of primary rat cortical astrocyte-only cultures treated with a range of octopamine concentrations (0 to 10 µM). N = 3; 7 to 10 cells/biological replicate. (C) Representative single-cell Ca²⁺ imaging traces of primary rat cortical astrocyte-only cultures treated with 100 nM octopamine (N = 6). (D) Ca²⁺ imaging of primary rat cortical astrocyte-only cultures cotreated with octopamine (100 nM) and the TAAR1 receptor inhibitor, EPPTB (500 nM), washed, and then rechallenged with octopamine (100 nM). N = 3; 7 to 10 cells/biological replicate. (E) Ca²⁺ imaging of wild type (WT) or Orai1-KO (Orai1_fl/fl, GFAP-Cre) primary mouse cortical astrocyte-only cultures treated with octopamine (100 nM). N = 3; 7 to 10 cells/biological replicate. (F) Ca²⁺ influx rates of WT and Orai1-KO (Orai1_fl/fl, GFAP-Cre) transients from E. N = 3; 7 to 10 cells/biological replicate. (G) Ca²⁺ imaging of WT or Orai1-KO (Orai1_fl/fl, GFAP-Cre) primary mouse hippocampal astrocyte-only cultures treated with octopamine (100 nM) in the absence of extracellular Ca²⁺ and in 2mM Ca²⁺. N = 3; 7 to 10 cells/biological replicate.

In invertebrates, octopamine is a well-characterized neurotransmitter that plays an important role in fight or flight responses (47). Even though octopamine’s function was replaced by epinephrine during mammalian evolution, it is still present at trace amounts (nanomolar concentrations) (48). While octopamine does not act as a mammalian neurotransmitter given that its neuronal release is not dependent on an action potential (49), it can still modulate neuronal function by a yet unidentified mechanism. Moreover, the importance of octopamine is highlighted by human disease where its levels are severely deregulated in PD and a range of psychiatric diseases including schizophrenia and bipolar disorder (50, 51). To address whether octopamine is the secreted neuronal metabolite responsible for modulating lactate metabolism in a Ca²⁺-dependent manner in astrocytes, we first tested its ability to trigger Ca²⁺ mobilization. Primary cortical astrocyte-only cultures were loaded with the ratiometric Ca²⁺ indicator Fura-2 and challenged with a range of octopamine concentrations (Fig. 3B). While concentrations as low as 1 nM were sufficient to elicit Ca²⁺ influx, the pattern of Ca²⁺ mobilization changed according to the dose. At 100 nM octopamine, Ca²⁺ oscillations were detected (Fig. 3 B and C). However, concentrations below 100 nM triggered a monotonic low-magnitude Ca²⁺ response, whereas concentrations above 100 nM triggered a monotonic high-magnitude Ca²⁺ response (Fig. 3B).

In mammals, octopamine has been shown to bind the G-protein-coupled receptor TAAR1 (52). While more prevalent in neurons, TAAR1 has also been shown to be expressed in astrocytes (53). To investigate whether octopamine’s ability to mobilize Ca²⁺ was dependent on TAAR1 binding, we treated the astrocyte-only cortical cultures with the selective and reversible TAAR1 antagonist N-(3-Ethoxyphenyl)-4-(1-pyrrolidinyl)-3-(trifluoromethyl)benzamide (EPPTB) (54). EPPTB abolished the Ca²⁺-dependent response of octopamine and this inhibition was reversible since it could be competed with increasing concentrations of octopamine (Fig. 3D and SI Appendix, Fig. S8 A and B).

Ca²⁺ release-activated Ca²⁺ (CRAC) channels, composed of Orai1 and stromal interaction protein 1, have been recently shown to be a major route of Ca²⁺ entry in astrocytes (55). To investigate whether the Ca²⁺-dependent effects of octopamine were mediated by the CRAC channel, we utilized astrocytes from a mouse where Orai1 was deleted exclusively in astrocytes (Orai1_{fl/fl, GFAP-Cre}). This deletion of Orai1 obliterates the channel function (56). While wild type (WT) primary cortical astrocyte-only cultures displayed oscillatory Ca²⁺ responses in the presence of octopamine, Orai1_{fl/fl, GFAP-Cre} astrocyte-only cultures did not respond to octopamine (Fig. 3 E and F). Importantly, octopamine’s Orai1 dependence was also observed in hippocampal astrocyte-only cultures (Fig. 3G), suggesting that the Ca²⁺-mediated effects of octopamine are widespread in other regions of the brain important for cognition. Together, these results indicate that octopamine mediates Ca²⁺ influx in astrocytes via TAAR1–Orai1 and that octopamine concentration dictates the amplitude and frequency of Orai1-dependent Ca²⁺ waves.

Octopamine Mediates Lactate Production in Astrocytes in a TAAR1–Orai1–Ca²⁺–Calcineurin-Dependent Manner.

To determine whether lactate synthesis is octopamine’s ultimate effector response, cortical astrocyte-only cultures were treated with a range of octopamine concentrations in a time-dependent manner. Lysates were collected longitudinally for lactate measurements using a lactate dehydrogenase luciferase-based assay. Concentrations as low as 0.5 nM to 100 nM of octopamine elicited robust lactate secretion; however, concentrations of 500 nM halted this effect (Fig. 4A). To investigate whether octopamine’s effect on lactate secretion was TAAR1–Orai1–calcineurin dependent, cortical astrocyte-only cultures were cotreated with octopamine in the presence of the calcineurin inhibitor FK506, TAAR1 inhibitor, EPPTB, or in the absence of Orai1 Ca²⁺ channel. While treatment with 100 nM of octopamine elicited robust lactate secretion, lactate synthesis was compromised in astrocytes cotreated with EPPTB, FK506, or in Orai1_{fl/fl, GFAP-Cre} (Fig. 4B).

Fig. 4.

Octopamine mediates lactate production in astrocytes in a TAAR1–Orai1–Ca²⁺–calcineurin-dependent manner. (A) Lactate measurements from WT and Orai1 KO (Orai1_fl/fl, GFAP-Cre) primary cortical astrocyte-only cultures treated with a range of octopamine concentrations (0.5 to 500 nM) over time. N = 9; *P < 0.05, one-way ANOVA, post hoc Tukey test. (B) Lactate measurements from WT and Orai1 KO (Orai1_fl/fl, GFAP Cre) primary cortical astrocyte-only cultures pretreated for 30 min with either FK506 (1 µM) or EPPTB (200 nM) and subsequently challenged with octopamine (100 nM) for 4 h. N = 3. *P < 0.05; one-way ANOVA, post hoc Dunnett’s multiple comparison test. (C) Heat map from the union of differentially expressed genes by RNA-Seq from primary cortical astrocyte-only cultures treated with vehicle, octopamine (100 nM), or octopamine (100 nM) + FK506 (1 µM). All data are relative to the vehicle treated (DMSO). Log10(P-value): Group 1= −14.5, Group 2= −10.1, Group 3= −8.5, Group 3= −28.2. (D) Measurements of generated lactate from ex vivo brain slices extracted from WT and Orai1_{fl/fl, GFAP-Cre} mice. After 3 d in culture, slices were treated with either 1× PBS or 100 nM octopamine. Samples of the media were taken at 4 h, 8 h, and 24 h post treatment. Normalized measurements were presented as fold change from baseline value measurements taken from vehicle-treated slices. N = 20 slices, N = 4 mice. (E) Measurements of rate of lactate production as calculated by the flux of lactate concentration between timepoints from D. (F) Measurements of the rate of lactate production with spatial separation from D. Slices were binned by stereotaxic coordinates lateral to bregma to show relative rates of lactate production in both spatial and temporal dimensions. (G) Representative immunohistochemistry from hippocampus CA3 for GFAP and DAPI of ex vivo brain slices from (D) 24 h posttreatment with either vehicle (Phosphate Buffered Saline or PBS) or 100 nM octopamine. (Scale bar, 50 µm.) (H) Quantification of GFAP fluorescence relative to the DAPI fluorescence as a measure of astrocyte reactivity from G. N = 3. **P < 0.01, ****P < 0.0001; one-way ANOVA, post hoc Dunnett’s multiple comparison test. (I) Representative immunofluorescence from prefrontal cortex for GFAP and DAPI of ex vivo brain slices from (D) 24 h posttreatment with either vehicle (PBS) or 100 nM octopamine. (J) Quantification of GFAP fluorescence relative to the DAPI fluorescence from I as a measure of astrocyte reactivity. N = 3. ****P < 0.0001; one-way ANOVA, post hoc Dunnett’s multiple comparison test. (Scale bar, 50 µm.)

The RNA-Seq data from the neuron–astrocyte cocultures indicated that the neuroprotective effects of partial inhibition of calcineurin lead to downregulation of ATP synthesis-coupled transport, also known as oxidative phosphorylation (SI Appendix, Fig. S2A). To determine whether octopamine-mediated signaling is responsible for the transcriptional repression of oxidative phosphorylation-related genes in astrocytes, we performed RNA-Seq (29). Primary cortical astrocyte-only cultures were subjected to three different conditions: 1) lactate-inducing octopamine concentrations, 2) octopamine plus calcineurin inhibitor to establish which genes regulated by octopamine were dependent on calcineurin, and 3) control (vehicle treated). To understand the biological nature of the transcriptional changes across drug conditions, we performed a cluster analysis on the union of differentially expressed genes found in each condition compared to vehicle-treated astrocyte-only cultures. Treatment of octopamine caused a prominent downregulation of genes belonging to oxidative phosphorylation followed by those involved in ion regulation (Fig. 4C and Dataset S4). In contrast, the most up-regulated genes were those corresponding to response to growth factors, and, to a lesser extent, neurotransmitter regulation. Importantly, the regulation of genes involved in oxidative phosphorylation (group 2) and response to growth factors (group 1) was dependent on calcineurin, since complete inhibition of the phosphatase in the presence of octopamine reverted to control (vehicle treated). These data are very reminiscent of the RNA-Seq from α-syn-expressing neuron–astrocyte cocultures treated with subsaturating doses of FK506 (SI Appendix, Fig. S2A). These data indicate that the transcriptional signature generated by partial inhibition of calcineurin in the α-syn-expressing neuron–astrocyte cocultures is most likely due to physiological levels of octopamine.

To investigate whether octopamine’s effects on metabolic reprogramming and astrocyte reactivity occurred in the brain, we used ex vivo brain slices. Brain slices contain the structural and functional characteristics of synaptic connectivity of neuronal circuits while being more amenable to environmental manipulations. Brain slices containing frontal cortex and hippocampus were taken from 3-wk-old WT and Orai1_{fl/fl, GFAP-Cre} animals and subjected to octopamine treatment. We focused on the octopamine dose that triggered Ca²⁺ oscillations and a lactate effector response, 100 nM octopamine. After treatment with either vehicle or 100 nM of octopamine, samples of supernatants from live ex vivo brain slices were taken 1 h, 4 h, 8 h, and 24 h post treatment. 4 to 8 h postoctopamine treatment induced lactate production in WT slices (Fig. 4 D–F). However, lactate production was compromised in Orai1_{fl/fl, GFAP-Cre} slices (Fig. 4 D–F). The Orai1 lactate dependence was specific for brain slices containing hippocampus and frontal cortex compared to other regions in the brain (Fig. 4F). When lactate production was displayed spatiotemporally, it was evident that slices containing the prefrontal cortex (bregma +2.5 to +1.5) and the hippocampus (bregma −0.5 to −2.5) contained the highest levels of lactate generation (Fig. 4 F–H). These regions responded the most 4 h postoctopamine treatment compared to the Orai1_{fl/fl, GFAP-Cre} slices. Moreover, the increase in lactate production by octopamine was also accompanied by an increase in astrocyte reactivity in both the hippocampus and the frontal cortex (Fig. 4 G–J). Altogether, the data from primary cortical astrocyte-only cultures to ex vivo brain slices indicate that octopamine operates in a TAAR1–Orai1–calcineurin-dependent manner to regulate lactate metabolism in astrocytes.

Discussion

We had previously demonstrated the therapeutic potential of decreasing calcineurin activity in a large preclinical study in a rat PD model (17). However, the mechanism by which the downstream effectors of calcineurin activity led to neuroprotection in the cerebral cortex had not yet been elucidated. Here, using several systems from primary cortical neuron–astrocyte cocultures, cortical and hippocampal astrocyte-only cultures, patient samples, ex vivo mouse brain slices to in vivo murine models, we uncovered a feedback mechanism between astrocytes and neurons driven by calcineurin activity, which leads to neuroprotection. We showed that reducing calcineurin activity in neurons undergoing α-syn proteotoxic stress triggers the release of physiological levels of octopamine and upregulation of the MCT2 transporter. In astrocytes, physiological levels of octopamine trigger Ca²⁺- oscillatory waves via Orai1, which leads to subsequent activation of calcineurin and lactate synthesis. Secreted lactate by astrocytes is imported into neurons via MCT2. In neurons, lactate is used as an energy substrate to rescue the ATP deficits caused by α-syn proteotoxic stress and confer neuronal survival (SI Appendix, Fig. S9A). On the other hand, we found that if high calcineurin activity, caused by α-syn proteotoxic stress, cannot be attenuated, neurons release high pathological levels of octopamine. This leads to sustained activation of Orai1–Ca²⁺–calcineurin activity in astrocytes, preventing transcriptional upregulation of LDH5 and lactate synthesis. Lack of lactate import into ATP-deficient neurons due to α-syn proteotoxic stress leads to cell death and astrocytosis (SI Appendix, Fig. S9B). Therefore, our data demonstrate a unique role for octopamine in the mammalian brain as a key mediator of the metabolic rewiring of astrocytes and subsequent neuroprotection regulated by calcineurin.

Astrocytes have been shown to play a critical role in PD and many other neurodegenerative diseases mainly due to their neuroinflammatory toxic properties (57). Here, we elucidate a previously unanticipated mechanism in which astrocyte reactivity can be modulated to be protective in disease: supporting aerobic glycolysis via octopamine–Ca²⁺–calcineurin signaling-mediated pathway. This neuroprotective state was defined transcriptionally by downregulation of mitochondrial complex I genes and metabolically by an increase in lactate synthesis. Therefore, our data suggest that deciphering the metabolic state of reactive astrocytes will be key to determine whether they will become neuroprotective or neurotoxic.

Octopamine is a bona fide neurotransmitter in invertebrates involved in fight or flight responses (47). While epinephrine replaced its role in mammals, octopamine is still present. However, its function had remained elusive until now. Our data suggest a physiological role for octopamine as an “S.O.S” distress signal from energetically challenged neurons to astrocytes, to provide a fast source of energy supply in the form of lactate. While the lactate shuttle we elucidated was triggered by α-syn proteotoxic stress, lactate shuttles have been shown to occur in the hippocampus as a mechanism to cope for the high energy demands during memory consolidation and learning (40). Interestingly, the octopamine–calcineurin-dependent lactate shuttle not only allowed survival but a rewiring in neurotransmitters; from a reduction in inhibitory neurotransmission through reduction of GABA toward an increase in excitatory neurotransmission through choline derivatives (SI Appendix, Fig. S10 A and B). Loss of cholinergic tone has been associated with dementia-related neurodegenerative diseases such as PD and Alzheimer’s disease (58, 59). In fact, the drug rivastigmine is given to patients with dementia to improve cognition by increasing cholinergic tone (60). Therefore, our studies suggest a unique physiological role for octopamine–calcineurin-dependent lactate shuttles during cognition and possibly in the treatment of dementia-related synucleinopathies such as PD dementia and DLB. Moreover, our studies can have therapeutic implications. Plasma levels of octopamine have been shown to be altered in PD (50). Our data demonstrated that octopamine levels are key to trigger lactate synthesis in astrocytes, and partial inhibition of calcineurin with the FDA-approved drug FK506 enables this metabolic response. Therefore, octopamine can also serve as a biomarker to follow FK506 dosing in the clinic. Finally, it is well documented that exercise is beneficial to patients with PD (61). Physical activity can have many different effects in the body, one of which is generation of lactate (62, 63). Based on our findings, lactate treatment could be explored as a unique therapeutic route to treat PD, DLB, and potentially other neurodegenerative diseases where astrocytes play an important role.

Materials and Methods

Primary Cells, Mice, and Human Samples.

Rat primary cortical neuron–astrocyte cocultures were infected with a lentivirus-carrying α-syn driven by the synapsin promoter. WT rodent primary cortical and hippocampal astrocyte-only cultures were used. In vivo models were generated based on overexpression of α-syn in CaMKII-dependent brain areas and Orai1_{fl/fl, GFAP-Cre} mice. Brain slices were obtained from WT and Orai1_{fl/fl, GFAP-Cre} mice. Human samples from DLB patients were obtained through the Massachusetts Alzheimer Disease Research Center. Detailed description can be found in SI Appendix, Materials and Methods.

RNA-Seq.

RNA samples from transduced rat primary cortical neuron–astrocyte cocultures and WT rat primary cortical astrocyte-only cultures were used to prepare next-generation sequencing libraries with the Illumina TruSeq Stranded mRNA Library Preparation Kit. Detailed description can be found in SI Appendix, Materials and Methods.

Metabolomics.

Soluble metabolites were extracted from transduced rat primary cortical neuron–astrocyte cocultures and cerebral cortex from α-syn transgenic animals in methanol/water. Metabolites were reconstituted in 50% acetonitrile and analyzed by ultra-high-performance liquid chromatography and high-resolution mass spectrometry and tandem mass spectrometry. Detailed description can be found in SI Appendix, Materials and Methods.

Imaging Techniques.

Ca²⁺ imaging was performed using Fura-2 in rat primary cortical and mouse hippocampal astrocyte-only cultures. MAP2 and GFAP staining in neurons and astrocytes was performed by immunofluorescence. MCT2 and GFAP staining in mice and human samples was performed by immunohistochemistry. Detailed description can be found in SI Appendix, Materials and Methods.

Ex Vivo Slice Cultures.

Postnatal-aged 21-23 WT and Orai1_{fl/fl, GFAP-Cre} mice were killed using kynurenic acid to prevent excitotoxicity. Slices were acquired using a Leica VT1000 S vibratome at a thickness of 250 µm. The slices were matched and allowed to recover and equilibrate for 3 d before the experiment. Detailed description can be found in SI Appendix, Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Data, Materials, and Software Availability

RNAseq data have been deposited in GEO (GSE213875) (29).