Antipsychotic drugs (APDs) are widely used to treat psychosis in schizophrenia 1,2 and symptoms in other
neuropsychiatric conditions 3. Blocking dopamine
signaling onto 60–80% of striatal D2 receptors (D2r) is thought to underlie their therapeutic efficacy
4,5.
However, studies showing symptom improvement at a broader range of D2r blockade (i.e. 16–95%) 6–10 suggest that this mechanism is poorly predictive of
antipsychotic response. Assessment of APD function at a circuit level has proven more predictive of behavioral
efficacy in humans 11–13 and rodents
14. In human patients, the effects of APDs on striatal
circuitry are assessed from fMRI BOLD signal, which is inhibited by APDs 11,12. This clinical
response is at odds with expectations, given the physiology of dopamine receptor expressing striatal cells that
intracellularly couple to either Gi/o (D2r) or Gs/o (D1 receptors, D1r), thereby leading to
cell inhibition and stimulation, respectively 15,16. Since APDs are antagonists or inverse agonists of D2r
17, but largely spare D1r, the hyperdopaminergic
signaling thought to underlie psychosis shifts to D1r in the presence of APDs, since D2rs are occupied, and the net
striatal response is expected to be excitation 15.
Thus, despite the broad clinical application of APDs, the neurobiology underlying their psychomotor effects is
unclear and difficult to explain based on mere D2r blockade.
To clarify these mechanisms, we used single cell in vivo calcium (Ca2+) imaging to analyze
activity of D1r- and D2r-expressing neurons in freely moving mice at baseline and in response to an acute
intraperitoneal injection of the typical APD haloperidol (HAL 0.5 mg/kg, Fig. 1A), a widely prescribed APD
with high affinity for D2r 18. Striatal cells respond
to HAL within minutes 19, contributing to changes in
brain structure 20 and symptoms 21 within hours. We determined the earliest psychomotor
effects of HAL in relationship with D1r and D2r neuronal responses in the nucleus accumbens core (NAcc), a prominent
striatal structure involved in spontaneous locomotion in animals 22,23 and in psychosis
and antipsychotic responses in humans 11–13, 24.
Because medium spiny neurons (MSNs) are the most prevalent neuronal type in the striatum (~ 95%) in humans and
animals 25 for simplicity we will refer to labeled
cells as MSNs throughout the text. We used D1- and D2-cre transgenic mice 26 to obtain selective expression of the Ca2+ sensing
fluorophore GCaMP6f in D1- or D2-MSNs (Fig. 1B), recorded with gradient refractive index lenses connected to a
head mounted miniature microscope (Fig. 1C and methods). We measured Ca2+ events from a total of 568
MSNs (246 D1-MSNs and 322 D2-MSNs from 8 mice each) over 45 min (15 min baseline and 30 min after HAL
or saline injection (Fig. 1D-E) while animals moved freely in an open field to which they were previously
habituated. At baseline D1-MSNs were more active than D2-MSNs (Fig. 2A). In animals that received saline, D1-
but not D2-MSN activity correlated positively with locomotion (Fig. S1). An acute injection of HAL did not reduce
spontaneous locomotion relative to saline-injected animals (Fig. 2B) and did not induce catalepsy, but
moderately impacted reward reactivity (Fig S2-3). Within the first 5 min HAL depressed Ca2+ events
in D2-MSNs compared to their baseline activity (Fig. 2C-D) and compared to saline-injected animals (Fig. S4A)
and the reduction was maintained throughout the recording session. Quite the opposite, HAL acutely elevated
Ca2+ events in D1-MSNs during the first 5 min after injection compared to their baseline activity,
although D1-MSN activity was gradually reduced over the course of the recording session (Fig. 2C-D). After HAL
injection, D2-MSN activity was positively correlated with locomotion (Fig. 2E). No correlation was found
between D1-MSN activity and locomotion after HAL (Fig. 2F). These data suggest that D1-MSN activity may impact
motor behavior at baseline, but in the presence of an APD, NAcc D2-MSNs drive spontaneous locomotion.
MSN firing rate is not constant, but alternates between periods of relative silence and episodes of moderate or high
firing 27,28. For this reason, the depression of D2-MSN activity after HAL could
result from complete abolition of Ca2+ events (i.e. no firing activity) or instead could result from an
activity switch from high-to-low firing. We analyzed the cumulative frequency distributions of Ca2+
events at baseline and after HAL treatment and found that HAL did not completely suppress Ca2+ events,
but instead decreased the number of D2- and D1- MSNs firing at high frequency (Fig. 2G-H). By subdividing MSNs
into quartiles according to frequency of Ca2+ events at baseline, we found that HAL decreased the
proportion of D2-MSNs exhibiting high and moderate Ca2+ spike frequency gradually over the 30 min
recording session, whereas the proportion of D2-MSNs with low or no Ca2+ events were
ultimately ~ 80% of all D2-MSNs (Fig. 2I). No changes in firing frequency were observed in
D1-MSNs after HAL during the first 20-min of the recording session, but a significant shift was observed during the
last 10 min of recording, where the proportion of cells firing at high frequency was reduced compared to
baseline (Fig. 2J). Since NAcc MSN firing is the result of a summation of excitatory and inhibitory inputs
29,30, we compared the increased/decreased ratio of MSN activity for each
animal based on the number of neurons firing above or below the median at baseline to estimate the net effect of HAL
on MSN activity. Confirming the results in Fig. 2I-J, HAL depressed D2-MSNs (Fig. 2K), but did not impact
D1-MSNs (Fig. 2L).
D2r stimulation inhibits adenylate cyclase and cyclic AMP production through Gαi/o and endogenous dopamine
has ~ 1000x higher potency for D2r than D1r at baseline 31. D2r blockade with APDs prevents this tonic inhibition, facilitating
D2-MSN excitation 15,32,33. Furthermore,
since HAL is an inverse D2r agonist 17, depression of
D2-MSN activity in our study was unexpected. Importantly, at the dose used, HAL does not saturate all D2r 34. Thus the possibility remains that spared D2r
permitted endogenous dopamine to generate inhibitory post-synaptic currents (IPSCs) in NAcc D2-MSNs. To determine if
dopamine could elicit IPSCs in NAcc D2-MSNs in HAL treated mice, we overexpressed the G protein-coupled inward
rectifying potassium (GIRK2) channel in the NAcc of Drd2-eGFP mice using AAV2/9 -GIRK2-TdTomato (Fig. S5). Because
endogenous D2r, but not D1r on MSNs can couple to GIRK2 channels, GIRK2 functions as a sensor providing a rapid,
direct readout of IPSC-mediated synaptic D2r activation (D2r-IPSC) 35,36. As expected,
synaptic dopamine stimulation evoked D2r-IPSCs in NAcc D2-MSNs in control animals and HAL reduced it five-fold,
(Fig. 3A). The lag to IPSC onset and decay increased after HAL treatment indicating a delayed NAcc D2r-IPSCs
(Fig. 3B) and reduced rate of dopamine clearance, respectively (Fig. 3C). Importantly, we estimated that
~ 21–45% of D2r were spared by HAL depending on incoming dopaminergic transmission, with a more robust
D2r-IPSC reduction following increased dopamine transmission. Indeed, HAL suppressed only 44–66% of D2r-IPSC
at low intensity stimulation and 76–81% at higher stimulation (Fig. 3D).
Together, these data indicate that HAL effectively reduced post-synaptic dopamine signaling onto NAcc D2-MSNs, but
also that endogenous dopamine could stimulate a pool of spared NAcc D2r even in the presence of HAL. The reduced
D2r-IPSC decay indicates reduced dopamine clearance, which could prolong dopamine transmission onto both MSN
subtypes (Fig. 2C-D). While we recapitulated previous reports on partial receptor occupancy 14,34
and dopamine uptake blockade 14 after systemic HAL
treatment, we show that post-synaptic D2r blockade cannot explain the depression of NAcc D2-MSN activity, suggesting
that other mechanisms are likely to be involved in the HAL-induced suppression of D2-MSNs in vivo.
The decrease in Ca2+ conductance may derive not only from decreased D2-IPSCs 37, but also from alterations in pre- and post-synaptic excitatory
transmission. NAcc glutamatergic terminals express D2r 38 and HAL functionally competes with endogenous dopamine for binding
39. Because our conditions fall short of D2r
saturation, it is possible that stimulation of spared D2r by dopamine could have reduced NAcc glutamate release
30. To test this hypothesis, we conducted ex
vivo whole-cell electrophysiological recordings from NAcc D2-MSNs in control and HAL treated mice. We first
determined whether glutamate release probability was altered by HAL using the paired-pulse ratio (PPR) of evoked
excitatory postsynaptic currents (EPSCs). PPR was significantly reduced on D2-MSNs of HAL treated mice
(Fig. 4A), excluding that reduced excitatory transmission onto D2-MSNs contributed to D2-MSN suppression after
HAL. Next, we evaluated putative post-synaptic alterations by measuring the ratio of α
-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPArs) and N-methyl-D-aspartate receptors (NMDArs).
Pharmacological isolation of AMPAr and NMDAr currents revealed an enhanced AMPAr/NMDAr in D2-MSNs in HAL treated
mice (Fig. 4B). While increased AMPAr/NMDAr may broadly indicate enhanced synaptic strength (i.e. long-term
potentiation, LTP) in D2-MSNs, an imbalanced AMPAr/NMDAr could also result from decreased NMDAr rather than
increased AMPAr currents, which is likely to reduce synaptic strength.
To determine whether a systemic HAL injection modified NMDAr currents, we examined the current-voltage (I/V)
relationship of pharmacologically isolated NMDAr currents in D2-MSNs. NMDAr I/V curves in D2-MSNs were significantly
decreased in HAL treated mice at -20 mV compare to control (Fig. 4C), suggesting that NMDArs undergo gross
changes in voltage dependence. Moreover, the NMDAr current at + 40 mV revealed faster decay kinetics in
mice treated with HAL (Fig. 4D), independently of differences in membrane properties, as membrane capacitance
and input resistance did not differ between treatment groups (Table S2). To confirm that HAL facilitated
presynaptic, but depressed post-synaptic excitatory transmission we assessed spontaneous EPSC (sEPSC) frequency and
amplitude and found that D2-MSNs in HAL treated mice showed a significant increase in sEPSC frequency, but not
amplitude (Fig. 4E).
HAL is a potent inhibitor of the sigma receptor 40,
which is known to regulate pre- and post-synaptic glutamate transmission 41,42. To test if HAL
altered NMDA receptor function by blocking the sigma receptor we bath applied the sigma receptor agonist siramisine
and found that while siramisine significantly decreased NMDA current in D2-MSNs of control animals, this effect was
antagonized in HAL-treated animal (Fig. 4F). Together, these findings show that HAL alters NMDA excitatory
transmission likely through blockade of sigma receptor function, likely leading to alterations in synaptic
plasticity. Because synaptic plasticity relies on Ca2+ influx through NMDAr 43 we expected that the impact of HAL on NMDAr function would alter
synaptic plasticity within the striatal network. To determine whether decreased NMDAr function after HAL altered the
signature of synaptic plasticity, we measured the amplitude of field EPSPs after the application of high-frequency
stimulation (HFS) of glutamatergic afferent fibers in the NAcc. In normal conditions the application of HFS enables
Ca2+ entry through post-synaptic NMDAr and triggers NMDAr-dependent LTP 43. Accordingly, the amplitudes of field EPSPs were significantly increased
from control mice (Fig. 4E) after HFS. To examine whether this LTP was mediated by NMDAR activation, field
EPSPs were recorded in the presence of the selective NMDAR antagonist D-AP5, which inhibited synaptic strength
expression (Fig. 4E) and confirmed the NMDAR-dependency of LTP. Since HAL shortened NMDAr decay kinetics,
thereby reducing the amount of Ca2+ entry through NMDAr, we predicted that LTP magnitude in the NAcc
would be reduced by HAL. Indeed, as shown in (Fig. 4F-G), application of HFS decreased field EPSP amplitude
below baseline after HAL, demonstrating that the same protocol that potentiated synaptic transmission in control
animals, instead depressed synaptic transmission in mice receiving HAL.
We reveal for the first time rapid neuronal responses of an antipsychotic in relationship to psychomotor output,
which is largely independent of D2 receptor blockade. Importantly, locomotion was correlated with NAcc D2-MSNs
Ca2+ activity after HAL. The suppression of Ca2+ events in NAcc D2-MSNs were underlined by
shortened NMDAr offset, by antagonism of the sigma receptor and impaired LTP. Paradoxically, these physiological
effects emerged despite HAL efficiently blocking dopamine transmission onto ~ 73% of D2r, minimizing the role
of D2r blockade in an antipsychotic response. Because HAL reduced D2-IPSCs without suppressing them completely, our
findings suggest that spared D2r might also have mediated the reduced Ca2+ signaling and LTP expression.
These rapid changes in D2-MSN functional plasticity rather than D2r blockade are likely to contribute to rapid
symptom improvements 21,44.
Finally, the depression of MSN responses illuminates the neurobiology underlying reduced striatal BOLD signals
observed in human studies coincident with antipsychotic efficacy 11–13. Here we extend these fMRI studies by showing divergent D1- and D2- MSN
responses to an APD, a cell-type signature that cannot be captured by fMRI. Importantly, because in our study a main
mechanism of action of HAL was to induce synaptic meta-plasticity by blocking LTP induction, our results shed light
on why antipsychotic responses can endure even after treatment discontinuation in humans 45 and animals (Fig. S3).