Dopamine Triggers the Maturation of Striatal Spiny Projection Neuron Excitability during a Critical Period

SUMMARY

Neural circuits are formed and refined during childhood, including via critical changes in neuronal excitability. Here, we investigated the ontogeny of striatal intrinsic excitability. We found that dopamine neuro-transmission increases from the first to the third postnatal week in mice and precedes the reduction in spiny projection neuron (SPN) intrinsic excitability during the fourth postnatal week. In mice developmentally deficient for striatal dopamine, direct pathway D1-SPNs failed to undergo maturation of excitability past P18 and maintained hyperexcitability into adulthood. We found that the absence of D1-SPN maturation was due to altered phosphatidylinositol 4,5-biphosphate dynamics and a consequent lack of normal ontogenetic increases in Kir2 currents. Dopamine replacement corrected these deficits in SPN excitability when provided from birth or during a specific period of juvenile development (P18–P28), but not during adulthood. These results identify a sensitive period of dopamine-dependent striatal maturation, with implications for the pathophysiology and treatment of neurodevelopmental disorders.

In Brief

Lieberman et al. demonstrate that ontogenetic increases in striatal dopamine release are required for the postnatal development of direct pathway spiny projection neuron excitability via control of Kir2 potassium currents.

INTRODUCTION

Selecting an appropriate action in response to environmental stimuli is critical to learning and survival. The basal ganglia encompass subcortical circuits essential to action selection. Perturbations of basal ganglia development are suggested to enhance susceptibility to drug use and neuropsychiatric disease later in life (Spear, 2000). The elucidation of mechanisms that regulate basal ganglia development is thus important for unveiling the pathophysiology of multiple disorders.

The main input nucleus of the basal ganglia—the striatum—undergoes significant postnatal development during the first through the fourth postnatal weeks (Tepper et al., 1998). The principal neuron of the striatum is the spiny projection neuron (SPN). SPNs participate in two output pathways: (1) the direct pathway containing SPNs that express D1 dopamine (DA) receptors and project to the basal ganglia output nuclei of the substantia nigra reticulata and globus pallidus interna (dSPN) and (2) the indirect pathway including SPNs that express D2 DA receptors and project to the globus pallidus externa (iSPN) (Gerfen and Surmeier, 2011).

In the adult, the SPN receives glutamatergic input from the cortex and thalamus that, together with DA inputs from the midbrain, provides a neurophysiological basis for motivational control of movement (Gerfen and Surmeier, 2011). Adult SPNs are characterized by low firing frequencies and a high threshold for action potential firing (Wilson and Groves, 1981; Wilson and Kawaguchi, 1996). Thus, only simultaneous activity from convergent synaptic inputs provides sufficient excitation to drive adult SPNs to fire and control action. Immature SPNs, however, are hyperexcitable and, therefore, do not generate mature firing patterns (Tepper et al., 1998). Although SPN hyperexcitability and consequent inappropriate action selection has been widely suggested to be present during adulthood in neuropsychiatric diseases, such as schizophrenia (Cazorla et al., 2012), Parkinson’s disease (Azdad et al., 2009; Fieblinger et al., 2014; Warre et al., 2011), Huntington’s disease (Klapstein et al., 2001), and drug abuse (Dong et al., 2006), little is known about the postnatal development of SPN firing patterns and how they acquire mature neurophysiological properties.

SPN intrinsic excitability is shaped by a complement of potassium channels. At hyperpolarized potentials, inwardly rectifying potassium channels (primarily Kir2.1 and Kir2.3; Cazorla et al., 2012; Karschin et al., 1996; Shen et al., 2007; referred to here-after as Kir) comprise the primary conductance. By driving the resting membrane potential toward the potassium reversal potential, these channels shape the summation of dendritic inputs and filter the response to excitatory glutamatergic inputs (Mermelstein et al., 1998; Nisenbaum et al., 1994; Shen et al., 2007). Sufficient coordinated synaptic input can bring the SPN resting potential toward the action potential firing threshold, where Kir currents are inactive and additional potassium currents, such as I_A, I_D, and I_M, control firing patterns (Nisenbaum et al., 1994; Shen et al., 2004, 2005; Wilson and Kawaguchi, 1996).

SPN excitability decreases over the juvenile and adolescent period from postnatal day 10 (P10) to P40 in rodents (Gertler et al., 2008; Tepper et al., 1998). P28 marks the first age with ‘‘mature’’ SPN up and down states, inward rectification, levels of excitatory input, and input resistance during in vivo recordings (Choi and Lovinger, 1997; Tepper et al., 1998). Although Kir (Gertler et al., 2008) and I_D (Surmeier et al., 1991) currents have been shown to increase during postnatal development, the ontogenesis of adult firing patterns in the striatum has not been established.

DA transmission into the striatum also changes during postnatal development. While DA axons arrive in the striatum at embryonic day 14 (E14) and are capable of calcium-dependent DA release by P0 (Ferrari et al., 2012; Voorn et al., 1988), the total levels of striatal DA increase from adolescence to adulthood (Matthews et al., 2013; Stamford, 1989). This postnatal maturation of DA neuro-transmission occurs during a period associated with the onset of multiple neuropsychiatric disorders, suggesting that changes in presynaptic DA may induce developmental changes in dopaminoceptive regions, although this has not been fully explored (Galiñanes et al., 2009; Kozorovitskiy et al., 2015).

Here, we report that DA transmission is responsible for the induction of the maturation of direct pathway SPN intrinsic excitability and must act during a critical window within the juvenile period to achieve the normal adult SPN state.

RESULTS

SPN Excitability Decreases in Parallel in dSPNs and iSPNs in the Juvenile Period

Adult SPNs maintain a hyperpolarized resting membrane potential and high latency to spike. Both in vivo and ex vivo recordings have demonstrated that SPNs are hyperexcitable after birth and acquire their mature phenotype during the first four postnatal weeks (Gertler et al., 2008; Peixoto et al., 2016; Tepper et al., 1998). Here, we examined the time course of these changes in acute brain slices from BAC-D1tomato (D1T) mice (Shuen et al., 2008). We conducted whole-cell recordings from D1T⁺ (dSPNs) and D1T⁻ (putative iSPNs) SPNs in the dorsal striatum during the second, third, and fourth postnatal weeks and adulthood (~4 months) to assess whether and how excitability is decreased throughout the juvenile and early adolescent period. We selected ages P12–P14, P18, and P28 for examination based on the following rationales: P12–P14 occur around eye opening (Evans et al., 2015) and during a period of increased excitatory synaptogenesis (Peixoto et al., 2016; Tepper et al., 1998) and DA neuron development (Oo and Burke, 1997); P18 occurs immediately before weaning and at the end of excitatory synaptogenesis in the striatum (Peixoto et al., 2016); and P28 marks the first age with ‘‘mature’’ up and down states during in vivo recordings of SPNs (Tepper et al., 1998). We also measured SPN excitability at 4 months to compare the juvenile and adult states.

We found that both D1T⁺ and D1T⁻ SPNs undergo reductions in excitability between P12 and P28. Resting membrane potential (RMP) and input resistance (R_in) decreased over this period in both dSPNs and iSPNs, while rheobase increased (Figures 1A–1D) (RMP: age p < 0.0001, subtype p = 0.94, interaction p = 0.58; R_in: age p < 0.0001, subtype p = 0.03, interaction p = 0.0618; rheobase: age × subtype interaction p = 0.045). These changes were associated with a difference in the number of action potentials fired in response to a range of injected current (current-response) (Figure 1E). By P28, parameters of SPN excitability were stable and not different from those measured at P120. Significant differences in rheo-base, input resistance, and current-response curve were also observed between dSPNs and iSPNs, similar to previous reports (Figures 1C and 1D) (Cepeda et al., 2008; Gertler et al., 2008). Thus, SPNs undergo a substantial decrease in excitability during the second and third postnatal weeks, with direct pathway neurons exhibiting a more profound decrease during the third and fourth weeks.

Figure 1. dSPNs and iSPNs Undergo Age-Dependent Reductions in Excitability.

(A) Representative traces of whole-cell current-clamp recordings from D1T⁺ (black) or D1T⁻(gray) SPNs in acute brain slices made from mice aged P18 or P28.

(B–D) Resting membrane potential (RMP) (B), rheo-base (C), and input resistance (D) from D1T⁺ or D1T⁻ SPNs from mice aged P12, P18, P28, or P120.

(E and F) Number of action potentials fired by D1T⁺ (E) and D1T^– (F) SPNs in response to current injection. Data analyzed by two-way ANOVA followed by Bonferroni post hoc tests. **p < 0.01, ****p < 0.0001. P12: D1T⁺ n = 11 cells (3), D1T⁻ n = 10 cells (3); P18: D1T⁺ n = 18 cells (3), D1T⁻ n = 18 cells (3); P28: D1T⁺ n= 16 cells (3), D1T⁻ n = 10 cells (3); P120: D1T⁺ n = 13 cells (3), D1T⁻ n = 10 cells (2).

Striatal Evoked DA Release Increases in the Juvenile Period

Why does SPN excitability change during the postnatal period? We hypothesized that a change in modulatory neurotransmission may induce these changes. Previous work has demonstrated differences in DA release dynamics between the immediate postnatal, adolescent, and adult periods (Ferrari et al., 2012; Matthews et al., 2013; Stamford, 1989), although a systematic characterization of evoked synaptic DA release dynamics during juvenile period has not been reported.

We measured electrically evoked DA release events in the dorsal striatum by fast-scan cyclic voltammetry (FSCV) in acute brain slices prepared from mice aged P8 to P110 (Figure 2A) (Sulzer et al., 2016). We chose P8, an age at which DA neuron firing rates are similar to those in adult levels (Ferrari et al., 2012), to determine whether increases in DA release preceded changes in the maturation of SPN excitability. Analysis revealed a significant effect of age on the amplitude of peak DA release (Figure 2B).

Figure 2. Electrically Evoked DA Release Increases in the Dorsal Striatum over the Juvenile Period.

(A) Representative traces of CV recordings in the dorsal striatum from P8 (gray) or P28 (black) mice. Applied stimuli are indicated by the downward pointing arrows.

(B) Peak electrically evoked DA release is plotted for five age groups over the juvenile period.

(C) Decay half-life of evoked DA transients shows no significant effect of age.

(D) A significant increase in the paired-pulse ratio (interstimulus interval: 3 s). Data were analyzed by one-way ANOVA ****p < 0.0001. P8: n = 10 (PPR: 6) (3), P10: n = 13 (PPR: 12) (3), P18: n = 13 (PPR: 6) (3), P28: n = 12 (PPR: 10) (3), P110: n = 11 (3).

(E) Representative western blots against TH, DARPP32, and actin.

(F) Levels of TH and DARPP32 normalized to actin expressed relative to levels in the adult. n = 4–6 per age group. Data were analyzed by two-way ANOVA followed by post hoc Bonferroni tests. A significant effect of age (**p < 0.01) was observed for TH, but not for DARPP32.

See also Figure S1.

The increased evoked extracellular DA concentration measured by FSCV could be due to either increased presynaptic DA release or decreased DA reuptake (Schmitz et al., 2001, 2002). To address this, we measured the t_1/2 (width at half-height) of DA peaks following electrical stimuli as a functional estimate of DA reuptake. We observed no effect of age on t_1/2 when analyzed by one-way ANOVA (Figure 2C), suggesting that, along with increased DA release, DA reuptake increases during postnatal development (Jones et al., 1995; Moll et al., 2000). Consistently, nomifensine, a DAT blocker, affected peak height and t_1/2 similarly at P14 and P28 (Figure S1).

To analyze why DA release increases postnatally, we examined short-term presynaptic plasticity by measuring the amplitude of DA release evoked by paired electrical pulses. We observed a significant effect of age on paired-pulse ratio (2^nd/1^st peak) (Figure 2D), suggesting either a larger releasable pool or faster refilling rate of release-competent synaptic vesicles. To examine a basis for an increased releasable pool of DA, we probed striatal tissue lysates for tyrosine hydroxylase (TH), the rate-limiting enzyme in DA synthesis, using DARPP-32, a protein involved in the SPN response to DA signaling, and actin as loading controls. While no effect of age was observed on DARPP-32 or actin levels, TH levels increased significantly over the same time course as that of electrically evoked DA release measured by FSCV (Figures 2E and 2F). Overall, our data indicate that DA synthesis and the number and/or quantal size of DA released during synaptic vesicle fusion increases over the P8–P30 period, with the most dramatic increases occur-ring at ages less than P20.

SPNs from Mice Developmentally Lacking Nigrostriatal DA Projections Do Not Undergo Ontogenetic Reduction in Excitability

While DA axons arrive in the striatum during the embryonic period (Voorn et al., 1988), our results indicate that DA release dynamics continue to mature throughout the first four weeks postnatally (Figure 2). These ontogenetic increases in DA release immediately precede major postnatal reductions in SPN excitability, suggesting that DA could induce these changes (Figures 1 and 2). To address this hypothesis, we utilized the Pitx3KO mouse, which harbors a loss-of-function mutation upstream of the Pitx3 gene, leading to developmental loss of DA neurons in the substantia nigra pars compacta by P0–P1 (Nunes et al., 2003). Thus, dorsal striatal SPNs in the Pitx3KO mouse develop in the absence of postnatal dopaminergic signaling (Figures 3A and 3B).

Figure 3. SPNs from Pitx3KO Mice Fail to Undergo Age-Dependent Reductions in Excitability.

(A) Representative images (scale bar, 100 mm) of TH immuno-fluorescence in the dorsal striatum of Pitx3WT and Pitx3KO mice. n = 3 WT, 4 KO.

(B) Representative traces of evoked DA measured with FSCV in the dorsal striatum of Pitx3WT (n = 6) and Pitx3KO (n = 4) mice.

(C) Traces from whole-cell current-clamp recordings of SPNs from the dorsal striatum of P28 Pitx3WT or Pitx3KO mice.

(D–F) RMP (D), rheobase (E), input resistance (F) of SPNs from Pitx3WT or Pitx3KO mice aged P14, P18, P28, and P110.

(G) Height of delayed ramp during square current injection 10–20 pA below the rheobase from P28 Pitx3WT and Pitx3KO mice.

(H) Latency to fire the first action potential in response to a current injection at the rheobase from P28 Pitx3WT and Pitx3KO mice.

(I and J) Current-response curves for SPNs from P18 (I) and P28 (J) Pitx3WT and Pitx3KO mice. Data were analyzed by two-way ANOVA followed by Bonferroni post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001. WT mice: P14 n = 7 cells (2), P18 n = 8 cells (2), P28 n = 24 cells (3), P110 n = 15 cells (3); KO mice: P14 n = 12 cells (3), P18 n = 8 cells (2), P28 n = 25 cells (3), P110 n = 14 cells (3).

(K) Representative traces of spontaneous excitatory postsynaptic events (sEPSCs) from P28 Pitx3WT and Pitx3KO SPNs.

(L and M) sEPSC frequency (L) and amplitude (M) show no difference between P28 Pitx3WT and Pitx3KO SPNs. Pitx3WT: n = 22 (3); Pitx3KO: n = 28 (3).

See also Figure S2.

We first confirmed that nigrostriatal projections are not present in Pitx3 mice. Consistent with the literature, we found that P28 Pitx3KO mice did not express TH-labeled DA axons in the dorsal striatum (Figure 3A) and that electric stimulation of the dorsal striatum did not elicit DA release (Figure 3B). We then conducted whole-cell recordings from SPNs in acute brain slices from P14, P18, and P28 Pitx3WT and Pitx3KO mice. We chose P14 as the earliest age to examine in this experiment as this is the beginning of a period of excitatory synaptogenesis (Tepper et al., 1998) and the end of a robust period of dopaminergic axon retraction (Oo and Burke, 1997). SPNs became significantly less excitable from P14 to P18, as demonstrated by a reduced RMP, an increased rheobase, and reduced input resistance in both genotypes (Figures 3C–3F) (RMP: age × genotype interaction p = 0.04; R_in: age × subtype interaction p = 0.001; rheobase: age × subtype interaction p = 0.0002). Pitx3WT SPNs then underwent further reductions in excitability from P18 to P28, as demonstrated by reduced RMP and input resistance and increased rheobase (Figures 3B–3D). In striking contrast, however, P28 Pitx3KO mice showed no further decrease in excitability after P18 and displayed significantly elevated RMP and input resistance and decreased rheobase compared to P28 WT SPNs (Figures 3B–3D). Adult SPNs also demonstrated a long delay to first spike in response to square current pulses and, at sub-threshold current levels, displayed a voltage ramp response (see Figure S3 for example). We measured the magnitude of the voltage ramp in response to current injections 10–20 pA below the rheobase in SPNs from Pitx3WT and Pitx3KO mice at age P28 and found no difference (Figure 3G, p = 0.9795). Similarly, no effect of genotype was observed on the latency to fire the first action potential in response to a square current injection at the rheobase (Figure 3H, p = 0.9941). Finally, more action potentials were triggered in response to injected current in Pitx3KO than Pitx3WT at P28, but not at P18 (Figures 3I and 3J) (P18: current p < 0.0001, genotype p = 0.5323, current × genotype p = 0.9997; P28: current × genotype p = 0.0067).

We then recorded from SPNs of 4-month-old Pitx3WT and Pitx3KO mice to address whether the hyperexcitable SPN phenotype of the Pitx3KO mouse persisted into adulthood. A significant effect of genotype was observed for rheobase, RMP, and input resistance (Figures 3C–3F). These data indicate that SPN excitability matures in DA-independent (P14–P18) and DA-dependent (>P18) stages, and the absence of striatal DA causes hyperexcitability of SPNs in adulthood.

To address whether these changes in intrinsic excitability were homeostatic in response to reduced glutamatergic synaptic inputs, we measured spontaneous excitatory postsynaptic current (sEPSC) frequency and amplitude at P28 and observed no difference between genotypes (Figures 3K–3M). These findings suggest that in the absence of developmental DA signaling, excitatory inputs onto SPNs function normally, but SPN intrinsic excitability is increased.

Kir Currents Do Not Undergo Ontogenetic Increases in SPNs from Pitx3KO Mice

To identify which ion channels may be responsible for the hyper-excitability of SPNs from Pitx3KO mice, we measured currents elicited following voltage steps from a holding potential of −60 mV (Figure 4A). We observed a difference between SPNs from Pitx3WT and Pitx3KO mice following hyperpolarizing voltage steps in a range where currents are predominantly mediated by Kir2 channels, while there was no difference between genotype following depolarizing voltage steps (Figure 4A). These results are consistent with our current-clamp recordings indicating a difference in RMP and input resistance, which are mediated by Kir2 channel in SPNs, but not in spike latency or subthreshold ramping, which are mediated by other K⁺ currents that dominate at depolarized voltages (Figure 3). We then measured the current elicited by hyperpolarizing voltage steps in Pitx3WT and Pitx3KO SPNs at P14, P18, P28, and adulthood (~P110) to see whether ontogenetic increases in Kir2 currents fail to occur in Pitx3KO SPNs. The current elicited by a hyperpolarizing voltage step from −60 mV to −150 mV was not different between genotype at ages P14 and P18 but was significantly greater in Pitx3WT SPNs at P28 and P110 (Figure 4B) (age × genotype p = 0.0005), mirroring the age-dependent changes in current-clamp parameters that we observed in Pitx3KO mice (Figure 3). Additionally, we found no significant difference in membrane capacitance between WT and Pitx3KO SPNs (Figure 4C), but a significant difference between genotypes was observed for the Kir2 current density at P28 (Figure 4D).

Figure 4. Kir Current Increases with Age in Pitx3WT SPNs, but Not SPNs from the Pitx3KO Mouse.

(A) IV curve of currents elicited from Pitx3WT (black) and Pitx3KO (red) SPNs following 10 mV voltage steps from −60 mV.

(B) Current elicited with a voltage step from −60 mV to −150 mV plotted against age. Data were analyzed with two-way ANOVA followed by Bonferroni post hoc tests.

(C and D) Capacitance was unchanged (C) but inward current density was significantly lower (D) in P28 Pitx3KO SPNs. Data in (C) and (D) were analyzed by two-tailed t tests. WT mice: P14 n = 7 cells (2), P18 n = 8 cells (2), P28 n = 24 cells (3), P110 n = 13 cells (3); Pitx3KO mice: P14 n = 12 cells (3), P18 n = 8 cells (3), P28 n = 25 cells (3), P110 n = 12 cells (3).

(E) Representative voltage-clamp recordings before and following addition of 1 mM CsCl₂ and subtracted traces from SPNs recorded in slices from P28 Pitx3WT and Pitx3KO mice.

(F–H) IV curves from Pitx3WT (F) and Pitx3KO (G) mice before and following addition of CsCl₂ and CsCl₂ subtracted currents (H). Pitx3WT: n = 8 cells (3); Pitx3KO: n = 14 cells (4).

To directly address whether these differences in the current voltage (IV) curve were mediated by differences in Kir2 currents, we measured current elicited in the presence of tetrodotoxin (TTX; to block sodium channels) with and without cesium chloride (CsCl₂), which blocks Kir2 channels (Cazorla et al., 2012; Mermelstein et al., 1998). The Kir2 currents were extracted by subtracting the currents in the presence of CsCl₂ from those in the absence of CsCl₂. CsCl₂ significantly decreased currents elicited by hyperpolarizing voltage steps from −60 mV in both Pitx3WT and Pitx3KO SPNs at P28 (Figures 4E–4G) (WT: current 3 CsCl₂ p < 0.0001; KO: current × cesium p < 0.0001). CsCl₂-sensitive currents, however, were reduced in Pitx3KO SPNs compared to Pitx3WT, indicating a role for Kir2 channels in mediating the hyperexcitability of Pitx3KO SPNs (Figure 4H) (current 3 genotype p < 0.0001). The reversal potential and voltage dependence of Kir2 currents were not different in SPNs from Pitx3WT and Pitx3KO mice (Figure 4H). In summary, Kir2 channel activity failed to increase in SPNs from Pitx3KO mice after P18, suggesting a possible mechanism for SPN hyperexcitability due to an absence of DA during development.

Increased Extracellular DA before P18 Does Not Accelerate Maturation of SPN Excitability

To test whether DA-independent maturation prior to P18 might be limited due to low levels of DA, we examined mice deficient for the DA uptake transporter (DAT KO mice), which demonstrate elevated extracellular DA (Gainetdinov et al., 1999; Giros et al., 1996). We recorded from SPNs at P18 in DAT WT or DAT KO mice and measured SPN excitability. We found no difference in RMP, rheobase, input resistance, or maximal Kir current between genotypes (Figure S2). Further, no difference was observed in the current-response curve, consistent with previous reports that SPN excitability is normal in adult DAT knockdown mice (Wu et al., 2007). These data indicate that the insensitivity to DA before P18 is not due to a low level of extracellular DA but to a developmental change in postsynaptic response to DA.

The Maturation of Intrinsic Excitability of dSPNs, but Not iSPNs, Is Disrupted in Pitx3KO Mice

DA induces distinct biochemical signaling cascades in dSPNs and iSPNs. In dSPNs, DA activates the D1 receptor to stimulate cAMP production, while in iSPNs, DA activates the D2 receptor to inhibit cAMP production (Beaulieu and Gainetdinov, 2011). To determine whether DA leads to the selective maturation of SPN excitability in one or both subtypes, we crossed Pitx3KO mice with mice carrying the BAC-D1-tomato allele (Shuen et al., 2008). We confirmed D1-Tomato expression in Pitx3KO mice (Figure 5A).

Figure 5. dSPNs, but Not iSPNS, Are Hyperexcitable in P28 Pitx3KO Mice.

(A) Immunohistochemistry for TH and D1-tomato demonstrates the absence of TH immunoreactivity in the dorsal striatum of P28 Pitx3KO mice but similar numbers of D1-tomato positive cells. Scale bar, 50 μm.

(B–D) RMP (B), input resistance (C), rheobase (D) in D1T⁺ and D1T⁻ SPNs from P28 Pitx3 WT (black) and Pitx3 KO (red) mice. Data were analyzed by two-way ANOVA followed by Bonferroni post hoc tests.

(E) Sample traces from D1T⁺ and D1T⁻ SPNs from P28 Pitx3WT and Pitx3KO mice.

(F and G) Current-response curves from D1T⁺ (F) and D1T⁻ (G) SPNs from Pitx3WT and Pitx3KO mice. D1T⁺ WT: n = 52 cells (10), D1T⁻ WT: n = 9 cells (3); D1T⁺ KO: n = 50 cells (7), D1T^-KO: n = 17 cells (4).

(H) Representative dendritic skeletons of D1T⁺ SPNs from Pitx3WT and Pitx3KO mice.

(I) No effect of genotype was observed in cumulative dendritic length, p = 0.5686.

(J) Sholl analysis revealed no effect of genotype on dendritic arborization. Pitx3WT: n = 14 cells (4); Pitx3KO: n = 17 cells (3).

(K–N) sEPSC frequency (K), amplitude (L), and decay tau (M) and rise tau (N) measured in whole-cell recordings of D1-SPNs (V_hold = −70 mV) showed no difference between genotype. Pitx3WT: n = 13 (4); Pitx3KO: n = 12 (4).

We prepared acute brain slices from P28 Pitx3WT;D1T or Pitx3KO;D1T mice and measured SPN excitability. D1T⁺ SPNs from Pitx3KO;D1T mice displayed a significantly depolarized RMP (genotype × subtype interaction p = 0.22; genotype p = 0.035; subtype p = 0.9966), elevated input resistance (genotype × subtype interaction p = 0.0025), and decreased rheobase (genotype × subtype interaction p = 0.0002) compared to D1T⁺ SPNs from Pitx3WT;D1T mice (Figures 5B–5D). D1T⁻ SPNs, in contrast, showed no difference between the genotypes (Figures 5B–5D). We observed a significant difference in the current-response curve between D1T⁺ SPNs (genotype × current interaction p < 0.0001) from Pitx3WT and Pitx3KO mice, but not between D1T⁻ SPNs (genotype × current interaction p = 0.9776; genotype p = 0.5993; subtype p < 0.0001) (Figures 5F and 5G). These results demonstrate that DA is preferentially required for dSPN maturation in the juvenile period, while iSPNs mature normally in the absence of DA.

We then sought to further examine the role of excitatory synaptic inputs and dendritic arborization on intrinsic excitability in Pitx3KO SPNs. The size and complexity of the dendritic arbor have previously been linked to the intrinsic excitability of SPNs (Cazorla et al., 2012; Gertler et al., 2008). To address whether the hyperexcitability in Pitx3KO D1T⁺ SPNs arises from altered dendritic morphology, we reconstructed neurobiotin-filled cells by confocal microscopy (Figures 5H–5J). We measured dendrites at P28, the first age at which Pitx3KO D1T⁺ SPNs were found to be hyperexcitable (Figures 3 and 5). We found no effect of genotype on the cumulative dendritic length or on the complexity of the dendritic tree as measured by Sholl analysis (Figures 5H–5J; genotype 3 distance interaction p > 0.9999; distance p < 0.0001; genotype p = 0.7748). These data indicate that hyperexcitability of D1T⁺ SPNs is independent of altered dendritic complexity in Pitx3KO mice.

The D1T reporter further provided a means for a direct assessment of excitatory inputs in D1T⁺ SPNs. We found no effect of genotype on sEPSC frequency, amplitude, or rise and decay time (Figures 5K–5N), further confirming an absence of excitatory synaptic deficits at P28.

We then assessed which SPNs exhibit decreased Kir currents. We found a significant effect of genotype on Kir currents only in D1T⁺ SPNs (Figures 6A–6C), whereas no difference was found in D1T⁻ SPNs between Pitx3WT and Pitx3KO mice (compared on one panel for clarity, D1: genotype 3 current p < 0.0001; D2: genotype × current p > 0.9999, genotype p = 0.9689, current p < 0.0001). We conclude that Kir currents are specifically reduced in D1T⁺ SPNs from Pitx3KO mice, but not in D1T⁻ SPNs.

Figure 6. Altered PIP₂ Dynamics Mediate Lower Kir2 Current in D1T⁺ SPNs from Pitx3KO Mice.

(A and B) Sample voltage-clamp recordings from D1T⁺ (A) and D1T⁻ (B) SPNs from P28 Pitx3WT and Pitx3KO mice following a protocol to elicit inward currents.

(C) IV curves for inward current from D1T⁺ and D1T⁻ SPNs from both genotypes. Data were analyzed by two-way repeated-measure ANOVA followed by Bonferroni post hoc tests. Preplanned post hoc tests comparing genotypes for each subtype showed significant genotype effects between D1T⁺ SPNs, but not D1T⁻ SPNs. *p < 0.05, ***p < 0.001, ****p < 0.0001. D1T⁺ WT: n = 11 cells (3), D1T⁻ WT: n = 9 cells (3); D1T⁺ KO: n = 22 cells (4), D1T⁻ KO: n = 17 cells (4).

(D) Sample voltage-clamp traces from Pitx3WT and Pitx3KO D1T⁺ SPNs immediately at membrane rupture (t = 0 s) and after dialysis of 1 mM neomycin (neo; t = 180 s).

(E) Percent change in inward currents after dialysis with neo or with neo and 50 μM diC8-PIP₂ (PIP₂) compared to no drug control. Experiments including or compared to PIP₂ were analyzed after 12 min. Data were analyzed with one-way ANOVA followed by Bonferroni post hoc test. **p < 0.01. WT neo (3 min): n = 7 cells, KO neo: n = 9 cells; WT neo (12 min): 7 cells, WT neo + PIP₂: n = 7 cells.

(F) Sample traces from Pitx3WT and Pitx3KO D1T⁺ SPNs immediately at membrane rupture (t = 0 min) and after dialysis of 50 mM PIP₂ (t = 12 min).

(G) PIP₂ significantly increases inward currents in Pitx3KO, but not Pitx3WT, mice. Data were analyzed with a two-tailed unpaired t test. **p < 0.01. WT: n = 9 cells; KO n = 11 cells.

See also Figures S3–S5.

Changes in additional potassium channels may contribute to the hyperexcitability of D1T⁺ SPNs from Pitx3KO mice. The K_v1.2 and K_v4.2 channels mediate I_D and I_A potassium currents, respectively, and control the timing and number of action potentials (Nisenbaum et al., 1994; Shen et al., 2004, 2005; Wilson and Kawaguchi, 1996). We measured two parameters, which are controlled by these channels: the latency to fire the first action potential and voltage ramping in response to subthreshold current injections. We found no significant effect of genotype on these two parameters in D1T⁺ SPNs from P28 mice (Figures S3A–S3D; see also Figures 3G and 3H). We further investigated the I_D current with voltage-clamp recordings, as this current increases during the postnatal development of SPNs (Surmeier et al., 1991). We found no difference in I_D current density between D1T⁺ SPNs from P28 Pitx3WT and Pitx3KO mice (Figures S3E and S3F). In the same subset of cells, however, Pitx3KO SPNs had a significantly reduced Kir2 current density (Figure S3G). The ratio of maximal I_D to maximal Kir2 current for each cell was significantly greater in Pitx3KO mice than in Pitx3WT mice (Figure S3H). The protein level of K_v1.2, the main channel underlying I_D in SPNs (Shen et al., 2004), did not differ in striatal lysates between Pitx3WT and KO mice at P28 (Figures S3I and S3J). These data suggest that Kir2 currents, but not the other K⁺ currents measured, fail to undergo postnatal maturation in dSPNs from Pitx3KO mice.

Increasing PIP₂ Levels, but Not cAMP Production, Rescues Decreased Kir Currents in dSPNs from Pitx3KO Mice

Kir currents are dynamically regulated in SPNs via changes in protein expression (Cazorla et al., 2012), phosphorylation status (Zhao et al., 2016), and lipid interactions (Shen et al., 2007). One possible explanation for the reduced Kir current in Pitx3KO SPNs is a reduction in total Kir2 protein level or a change in the relative levels of the two Kir2 subunits expressed in SPNs, Kir2.1 and Kir2.3 (Cazorla et al., 2012; Karschin et al., 1996; Shen et al., 2007). We found no difference in protein levels of Kir2.1 and Kir2.3 in striatal lysates from Pitx3WT and Pitx3KO mice by immunoblot (Figure S4). Alternatively, relatively more Kir2.3, which has a smaller single-channel conductance than Kir2.1, could explain diminished whole-cell Kir2 currents (Hibino et al., 2010). The relative contribution of each subunit to whole-cell Kir2 currents can be assayed by determining the barium sensitivity of Kir2 currents (Hibino et al., 2010; Liu et al., 2001; Schram et al., 2003). We did not observe a difference in the IC₅₀ of barium for whole-cell Kir2 currents between D1T⁺ SPNs from Pitx3WT and Pitx3KO mice (Figure S4). We note, however, that the absolute difference in Kir2 currents between genotypes is also present in this cohort, and the remaining current, comprised primarily of K_leak, after perfusion of 1 mM BaCl₂, did not differ. These data suggest that the reduced Kir2 currents in D1T⁺ SPNs from Pitx3KO mice do not arise from reduced Kir2 protein or differences in Kir2 subunit composition.

Alternatively, modulation of Kir2 current by PKA/cAMP signaling could be disrupted in Pitx3KO mice. However, we found no difference in PKA activity, and forskolin did not rescue Kir currents in Pitx3KO mice (Figure S5).

We next addressed whether alterations in Kir2 interactions with its partner, phosphatidylinositol 4,5-biphosphate (PIP₂), could explain reduced whole-cell Kir2 currents in Pitx3KO D1T⁺ SPNs. PIP₂ is required for proper Kir2 channel function, as it promotes channel opening (Hibino et al., 2010). PIP₂ interacts with Kir2 electrostatically, and this interaction can be disrupted by addition of neomycin to the intracellular pipette solution (Xie et al., 2008). We found that neomycin significantly reduced Kir2 currents in D1T⁺ SPNs from P28 Pitx3WT mice, but not P28 Pitx3KO mice (Figures 6D and 6E). Further addition of a water-soluble, short-chain analog of PIP₂, diC8-PIP₂, blocked the inhibitory effect of neomycin on WT neurons, confirming the specificity of neomycin’s action on Kir2 channels via PIP₂ (Figure 6E, right). To further analyze this mechanism, we included diC8-PIP₂ alone in the intracellular pipette solution and measured Kir2 currents. Remarkably, we found that diC8-PIP₂ significantly increased Kir2 currents in D1T⁺ SPNs from Pitx3KO mice but had no effect in D1T⁺ SPNs from Pitx3WT mice (Figures 6F and 6G). These results are consistent with reduced PIP₂-Kir2 interactions, leading to diminished whole-cell Kir2 currents in D1T⁺ SPNs from Pitx3KO mice.

L-DOPA Supplementation during the Juvenile Period, but Not during Adulthood, Induces Normal SPN Maturation in Pitx3KO Mice

To confirm whether DA plays an important role in establishing mature SPN excitability during early postnatal development, we gave breeding pairs of Pitx3KO and Pitx3WT mice, or adult Pitx3KO mice, access to food pellets supplemented with the DA precursor L-DOPA, which is converted to DA and released from serotonergic terminals in the DA-denervated striatum (Lindgren et al., 2010; Mosharov et al., 2015), together with benserazide (bsrz) to block the peripheral conversion of L-DOPA to DA. We split these mice into three treatment groups depending on the age at treatment initiation to address whether DA was required to act during a specific age window during postnatal development. Mice in group 1 received either bsrz (referred to as B in Figure 7) or bsrz/L-DOPA (referred to as L in Figure 7) from birth to P28. Mice in group 2 received bsrz or bsrz/L-DOPA from weaning at P18 to P28, and mice in group 3 received bsrz or bsrz/L-DOPA starting at P90 to P120 (Figure 7A). Importantly, with this paradigm, we found that serum L-DOPA levels are relatively constant throughout the day (Table S2), providing an advantage over pulsatile treatment with DA receptor agonists.

Figure 7. Dopamine Replacement from Birth, but Not in the Adult, Rescues Mature SPN Excitability in Pitx3KO Mice.

(A) Schematic depicting chow treatment from birth to P28 (group 1; red), from P18 to P28 (group 2; gray), and from P90 to P110 (group 3; pink).

(B) Sample current-clamp traces from SPNs treated in groups 1–3 from the specified genotype. B or L after the group number indicates the type of chow provided (B, benserazide; L, benserazide + L-DOPA).

(C–E) RMP (C), input resistance (D), and rheobase (E) from specified groups. The effect of L-DOPA in the chow within each age was evaluated by one-way ANOVA of Pitx3KO 1B, 1L, and 2L following by Bonferroni post hoc test (RMP: p > 0.4423; R_in: F = 8.33, p < 0.0001; rheobase: F = 11.11 p < 0.0001) or two-tailed unpaired t test for 3B and 3L (RMP: p = 0.3; R_in: p = 0.7295; rheobase: p = 0.4071). Gray bar denotes the WT 1B mean and error (SEM; see also Figure S6).

(F and G) Current-response curves for groups 1B, 1L, and 2L (F) or 3B and 3L (G). Repeated-measures (RM) two-way ANOVA with Bonferroni post hoc tests between WT 1B and KO 1B, 1L, and 2L shows significant differences between WT 1B and KO 1B, but not 1L and 2L, at specified currents (F). RM two-way ANOVA shows no significant interaction between current and treatment group (G).

(H) Sample voltage-clamp traces from all groups.

(I and J) IV curve for inward currents for groups at P28 (I) and in adulthood (J). Significant differences were observed between inward currents evoked in SPNs from KO 1B compared to other groups as specified (I). No significant interaction was observed between current and treatment group (J). RM two-way ANOVA followed by Bonferroni post hoc tests were used. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. KO 1B: n = 21 cells (2), 1L: n = 28 cells (4), 2L: 1 n = 9 from 4 mice, 3B: n = 14 cells (3), 3L: n = 18 cells (3).

See also Figures S6 and S7.

To expose mice in the postnatal and juvenile period (group 1), we placed breeding pairs of Pitx3KO or Pitx3WT mice together and switched the chow to either bsrz alone or bsrz/L-DOPA. Breeding pairs and pups were maintained on the specified diet until weaning at P18. Pups were then weaned onto the diet on which they were raised and sacrificed at P28–P30 for electrophysiology or postmortem analysis of striatal monoamine content.

We first compared striatal monoamine content across genotype and treatment. In Pitx3WT mice, we found no significant effect of L-DOPA treatment on striatal tissue levels of DA or its metabolites, homovanillic acid (HVA) or dihydroxyphenylacetic acid (DOPAC) (Figure S6). In Pitx3KO mice, as expected in assays of total tissue monoamine content (Wood and Altar, 1988), we found no effect of L-DOPA treatment on DA levels but observed a significant increase in the levels of HVA and DOPAC (Figure S6). Increases in these metabolites are due to increased extracellular DA (Wood and Altar, 1988) and demonstrate that L-DOPA elevated striatal DA turnover in Pitx3KO mice.

We then determined whether L-DOPA treatment affected SPN excitability in Pitx3WT mice compared to bsrz alone. We found no significant differences between bsrz- and bsrz/L-DOPA-treated Pitx3WT mice at P28 in resting membrane potential, input resistance, rheobase, and peak inward current (Figure S6).

To examine the role of DA in the developmental maturation of SPN excitability, we examined SPN excitability after treatment from birth to P28 (KO group 1B versus KO group 1L; Figures 7C–7E). In contrast to group-1L-treated Pitx3WT SPNs, group-1L-treated Pitx3KO SPNs exhibited a trend of decreased resting membrane potential and a significantly increased rheobase and decreased input resistance compared to Pitx3KO SPNs in group 1B (Figures 7C–7E). Furthermore, current-response curves for Pitx3KO SPNs from group 1L were significantly shifted toward the right compared to group 1B (Figure 7F). We note that Pitx3WT SPNs in group 1B (reproduced from Figure S6) are shown throughout Figure 7 for comparison.

The intrinsic excitability of SPNs from Pitx3KO mice was indistinguishable from Pitx3WT mice until P18 but diverged by P28 (Figure 3) and increasing DA levels before P18 failed to accelerate the maturation of SPN excitability (Figure S2). We therefore tested whether treatment from P18 to P28 with L-DOPA was sufficient to rescue SPN excitability in Pitx3KO mice (group 2L; Figures 7C–7F). First, we confirmed that a 10-day treatment with bsrz/L-DOPA increased striatal DA turnover by HPLC compared to P28 Pitx3KO mice maintained on bsrz from birth (Figure S6). Remarkably, L-DOPA treatment from P18 to P28 had the same effect as treatment from birth (Figures 7C–7F). Thus, the presence of DA during a specific 10-day window is sufficient to induce normal maturation of SPN intrinsic excitability, but DA is not necessary prior to this developmental period.

We then addressed whether DA could lead to SPN maturation if introduced in adulthood. We placed naive adult Pitx3KO mice on bsrz or bsrz/L-DOPA chow from P90 to P110 (group 3). L-DOPA treatment had no effect on resting membrane potential, rheobase, or input resistance (Figures 7C–7E). Furthermore, L-DOPA treatment had no effect on the current-response curve in SPNs from group 3L compared to 3B (Figure 7G). Together, these results demonstrate that DA must act during a critical window in the juvenile period to provide normal maturation of SPN excitability.

Using voltage clamp, we tested whether diminished Kir currents in Pitx3KO SPNs were rescued by L-DOPA. Pitx3KO SPNs from groups 1L and 2L had significantly increased Kir currents compared to Pitx3KO SPNs from group 1B (Figures 7H and 7I). L-DOPA treatment in group 3 did not increase Kir currents (Figures 7H and 7J). Thus, L-DOPA during the juvenile period, but not during adulthood, rescued underlying deficits in Kir2 currents.

L-DOPA Treatment from P18 to P28 Rescues Pitx3KO D1T⁺ SPN Excitability and Kir2 Current via Control of PIP₂ Dynamics without Affecting D1T^– SPN Excitability or Kir Currents

We have shown that L-DOPA treatment during a critical period in SPN development rescues deficits in intrinsic excitability and Kir currents in Pitx3KO mice (Figure 7). To confirm whether L-DOPA treatment selectively rescued D1T⁺ SPN hyperexcitability in Pitx3KO mice or also affected D1T⁻ SPN excitability, we treated Pitx3KO;D1-tomato mice with bsrz (group 2B) or L-DOPA and bsrz (group 2L) from P18 to P28 and recorded SPN excitability at P28 (Figure S7B). L-DOPA treatment specifically rescued D1T⁺ SPN excitability (RMP, rheobase, and input resistance) and Kir current without affecting D1T⁻ SPN excitability or Kir current (Figures S7A–S7H).

Finally, we asked whether L-DOPA treatment from P18 to P28 would rescue Pitx3KO D1T⁺ SPN Kir currents via PIP₂ dynamics. We recorded Kir currents from Pitx3KO D1T⁺ SPNs in groups 2B and 2L with neomycin in the internal solution to disrupt PIP₂-Kir2 interactions (Figures S7I and S7J). Neomycin had no effect on Kir currents in group 2B SPNs, similar to what we observed in untreated Pitx3KO D1T⁺ SPNs (Figure 6). However, neomycin significantly inhibited Kir2 currents in cells from group 2L (Figure S7J), demonstrating that L-DOPA treatment during the appropriate developmental period rescued PIP₂-Kir2 dynamics in D1T⁺ SPNs of Pitx3KO mice.

DISCUSSION

DA neurotransmission has long been suspected to regulate the maturation of basal ganglia circuitry (Galiñanes et al., 2009; Kozorovitskiy et al., 2015). We found that evoked DA release in the striatum increases dramatically over the juvenile period. In the Pitx3KO mouse, which lacks nigrostriatal DA neurotransmission from birth, SPN excitability matured normally until P18, but then displayed persistent hyperexcitability into adulthood. DA replacement to juveniles, but not to adults, corrected SPN excitability in Pitx3KO mice, identifying a mechanism through which DA regulates basal ganglia development.

Developmental Changes in Striatal DA Release

Nigrostriatal DA fibers arrive by E14 in the mouse, and while there is a morphological absence of axonal boutons on dopaminergic axons until P14 (Voorn et al., 1988), DA axons are capable of Ca²⁺-dependent DA release in the striatum as early as P7 (Ferrari et al., 2012). Support for increased DA release comes from reports that the total striatal tissue content of DA (Coyle and Campochiaro, 1976), TH activity, and TH protein (Coyle and Campochiaro, 1976; Matthews et al., 2013) each increase during the juvenile period. DA reuptake is also thought to increase during this developmental period, as there is increased striatal DAT expression and tritiated DA uptake (Cao et al., 2007; Coulter et al., 1996; Tarazi et al., 1998).

To examine changes in evoked DA neurotransmission in striatal tissue, we performed cyclic voltammetry recordings in the dorsal striatum. We found that the total evoked DA signal increased dramatically through the juvenile period (Figure 2), with an increased amplitude and unchanged half-life of evoked DA transients. As detailed previously (Schmitz et al., 2001; Venton et al., 2003), peak height in these measurements mostly reflects the amount of evoked release, with some further increase if DAT is blocked, whereas the t_1/2 is mostly controlled by DAT activity (see also Figure S1). An increase in peak height with a similar t_1/2 as seen here is consistent with both increased release and increased DAT activity: in contrast, if DA release increased but DAT activity did not change, the falling phase of the DA transient (t_1/2) would be prolonged.

Diminished Kir2-PIP₂ Interactions Underlie dSPN Hyperexcitability in Pitx3KO Mice

DA release increases during the first four postnatal weeks, leading to reductions in dSPN intrinsic excitability. In Pitx3KO mice, which lack DA in the dorsal striatum from birth, the ontogenetic increase in Kir2 currents during postnatal development of SPNs fails to occur beyond P18 in dSPNs (Figure 4B). We propose that this deficit is responsible for the intrinsic hyperexcitability observed beginning at P28 in dSPNs from Pitx3KO mice.

Kir2 currents in SPNs play a critical role in setting the RMP and the response to inputs that drive the cell away from this potential (Mermelstein et al., 1998; Nisenbaum et al., 1994; Wilson and Kawaguchi, 1996). Decreased Kir2 currents lead to a more depolarized RMP, increased input resistance, and decreased rheobase, all of which are observed in D1T⁺ SPNs in Pitx3KO mice (Figure 5).

Interestingly, the latency to fire the first action potential and subthreshold voltage ramping, key characteristics of SPN firing patterns, were unchanged in SPNs from Pitx3KO mice. Consistently, the D-type potassium current, which also increases postnatally in SPNs (Surmeier et al., 1991) and contributes to delayed spiking (Shen et al., 2004), was not different in Pitx3KO and Pitx3WT SPNs at P28. We thus conclude that the intrinsic hyper-excitability and increased action potential firing in response to current injections arises predominantly from decreased Kir currents and the consequent elevated RMP and input resistance. Together, these results indicate that striatal DA tone during post-natal development is critical for the maturation of a specific subset of potassium channels (Kir2) and not others (Kv1.2).

Why are Kir2 currents reduced in Pitx3KO SPNs? No difference was observed in the expression of Kir2.1 or Kir2.3, the predominant Kir2 subunits in SPNs (Cazorla et al., 2012; Karschin et al., 1996; Mermelstein et al., 1998; Shen et al., 2007), suggesting that post-translational regulation of Kir2 channels is critical to this phenotype (Figure S4). We note, however, that small changes in Kir2 protein levels may not be apparent in total striatal lysates, as diminished Kir2 currents are only observed in a subset of Pitx3KO SPNs.

PIP₂ is a membrane phospholipid that modulates numerous ion channels and membrane receptors (Suh and Hille, 2008). PIP₂ is a required cofactor for Kir2.1 and Kir2.3 activity, as blocking PIP₂/Kir2 interactions pharmacologically or genetically greatly decreases Kir2 currents (Hibino et al., 2010). To address whether diminished PIP₂/Kir2 interactions were responsible for reduced Kir2 currents in D1T⁺ SPNs from Pitx3KO mice, we acutely screened PIP₂ by addition of neomycin or included a short-chain, water-soluble PIP₂ (diC8-PIP₂) in the patch pipette (Figure 6). We found that neomycin reduced Kir2 currents in WT D1T⁺ SPNs by ~30% but had no effect in Pitx3KO SPNs. Furthermore, diC8-PIP₂ increased Kir2 currents in Pitx3KO SPNs but had no effect in Pitx3WT SPNs. These findings suggest that Kir2/PIP₂ interactions are nearly saturated in D1T⁺ SPNs from Pitx3WT mice but decreased at baseline in Pitx3KO SPNs, underlying the diminished whole-cell currents observed in Pitx3KO SPNs after P18 (Figures 4 and 6).

We found that DA controls Kir currents in dSPNs during the striatal critical period identified here via regulation of PIP₂ dynamics (Figure S7). PIP₂ synthesis undergoes developmental increases in brain regions, including the hippocampus, via upregulation of its main biosynthetic enzyme, PIP5K (Unoki et al., 2012). Ontogenetic increases in PIP5K expression may be dependent on dopaminergic tone in the striatum. Alternatively, DA signaling may regulate PIP₂ synthesis via control of protein phosphatase-1 (PP1) or calcium entry through NMDA receptors (Nakano-Kobayashi et al., 2007; Unoki et al., 2012). Finally, PIP5K activity is closely regulated by Rho small GTPases (Weernink et al., 2004). Intriguingly, the DA D1 receptor signals through RhoA GTPases to control dendritic morphogenesis in the pre-frontal cortex, providing a possible connection between DA and PIP₂ (Li et al., 2015).

PIP₂ degradation via phospholipase C (PLC) is stimulated by G_q-coupled receptors. In dSPNs, the predominant G_q-coupled receptor is the muscarinic M1 receptor. Intriguingly, Ding et al. (2011) demonstrate that striatal cholinergic interneurons are hyperactive in adult Pitx3KO mice, suggesting the possibility of a hypercholinergic tone that may lead to increased PLC activity and PIP₂ degradation. However, pharmacologic M1R activation failed to inhibit Kir2 channels in dSPNs, suggesting that this may not play a prominent role in the regulation of Kir2 channels in dSPNs in Pitx3KO mice (Shen et al., 2007).

Remarkably, we found that I_Kir, but not I_D, is disrupted in D1T⁺ SPNs from Pitx3KO mice (Figure S3). K_v1.2 channels, which are the predominant source of I_D in SPNs (Shen et al., 2004), are also modulated by PIP₂ (Kruse and Hille, 2013; Rodriguez-Menchaca et al., 2012). The effect of PIP₂ on K_v1.2 differs from its effect on Kir2, however, by modulating voltage sensing and kinetics rather than current amplitude (Kruse and Hille, 2013). Thus, subtle changes in K_v1.2 channel activation may be present in Pitx3KO SPNs that were not detected here. Alternatively, the effect of disrupted PIP₂/Kir2 interactions on Kir2 currents may be indirect and mediated by Kir2 localization to cholesterol-rich membrane microdomains (Romanenko et al., 2004). The lipid raft localization of Kir2 is dependent on PIP₂ (Rosenhouse-Dantsker et al., 2014), and K_v1.2 channels are not strongly modulated by cholesterol, although it remains possible that cholesterol controls SPN excitability and that this role is regulated developmentally.

DA Induces Maturation of SPN Excitability in Development and Maintains SPN Firing Patterns in Adulthood

Here, we have identified a role for DA in regulating the maturation of D1-SPN intrinsic excitability. In the adult striatum, DA depletion leads to SPN hyperexcitability, which is thought to reflect a homeostatic compensation in response to a loss of excitatory inputs onto SPNs (Azdad et al., 2009; Fieblinger et al., 2014; Warre et al., 2011). Our results suggest that the SPN hyperexcitability following DA depletion during development may not represent a homeostatic mechanism to ensure proper striatal activity but rather the absence of DA-mediated developmental increase in Kir channel function that controls SPN excitability. In support of this hypothesis, we did not observe changes in dendritic arborization or in functional measurements of synaptic inputs at P28 in D1T⁺ SPNs from Pitx3KO mice (Figure 6). How might this be congruent with a causal link between SPN hyperexcitability and dendritic morphology reported previously (Cazorla et al., 2012)? In that study, Cazorla et al. demonstrate that reduced Kir currents in DA D2-receptor-overexpressing (D2-OE) mice leads to dendritic shrinkage in SPNs. These models differ, however, as SPN hyperexcitability in D2-OE mice occurs in cells not expressing the D2 transgene, although to a lesser degree, and was mediated by a downregulation of Kir2.1 and 2.3 protein expression (Cazorla et al., 2012). In contrast, the changes in intrinsic excitability in Pitx3KO SPNs were not present in all subtypes (Figure 5) and were independent of changes in Kir2 expression (Figure S4). The changes in SPN excitability observed in the Pitx3KO mouse indicate a fundamental role for DA in the ontogeny of corticostriatal neurotransmission.

DA as a Critical Period Signal in the Striatum

Treatment of juvenile, but not adult, Pitx3KO mice with L-DOPA rescued SPN excitability. To test whether L-DOPA enhances DA transmission differently at these ages, we measured DA and its metabolites in striatal lysates by HPLC. We found that L-DOPA increased the levels of the dopamine metabolites DOPAC and HVA, which primarily represent released DA (Lindgren et al., 2010; Mosharov et al., 2015; Wood and Altar, 1988), to a similar extent in juvenile and adult Pitx3KO mice (Figure S6). We conclude that L-DOPA treatment increases DA turnover in the striatum of both juvenile and adult Pitx3KO mice.

Difference in postsynaptic DA sensitivity may explain the specific response of juvenile Pitx3KO mice to L-DOPA that is absent in adult animals. Both DA receptor levels and DA-receptor-coupled signaling change dramatically during the juvenile period (Andersen, 2002; Teicher et al., 1995). It is possible that DA receptor dynamics and downstream signaling cascades, such as PIP₂ levels, require a minimum threshold of DA to undergo ontogenetic changes during the juvenile period to change SPN excitability. We note, however, that this threshold may be low, as the increases in DOPAC and HVA required to rescue SPN hyperexcitability in Pitx3KO mice remain lower than those observed in Pitx3WT mice. Nevertheless, without this exposure to DA, SPNs from Pitx3KO mice fail to respond to increased DA from L-DOPA supplementation during adulthood.

Why is DA unable to reduce SPN intrinsic excitability in adult Pitx3KO mice? A recent report has demonstrated dendritic atrophy in adult Pitx3KO mice (Suarez et al., 2018). It is possible that prolonged reductions in striatal DA initially lead to SPN hyperexcitability at P28 (Figure 3) and subsequently to dendritic shrinkage in adulthood (Suarez et al., 2018). Reduced glutamate signaling from synapse loss in the adult Pitx3KO mouse may prevent increased DA signaling with L-DOPA treatment from rescuing Kir2 function. In support of this hypothesis, the activity of several GTPases in SPNs are under the dual control of glutamate and DA (Girault et al., 2007), and in other cell types, these GTPases play critical roles in control of PIP₂ levels that activate Kir2 channels and reduce intrinsic excitability (Weernink et al., 2004). DA replacement increased Kir2 currents and decreased SPN excitability in Pitx3KO mice prior to P28, when glutamate signaling is not different from Pitx3WT SPNs, as indicated by the absence of difference in synaptic inputs or dendritic arborization (Figure 5). Thus, prolonged DA depletion may lead to homeostatic changes in glutamatergic signaling, prohibiting DA supplementation from rescuing SPN hyperexcitability in adulthood.

Implications for Developmental Diseases of the Striatum

We have established a timeline of ontogenesis of SPN intrinsic excitability and striatal DA neurotransmission and identified a role for striatal DA in the regulation of a postnatal decrease of D1-SPN excitability during a specific critical period. The study of normal brain development may elucidate pathogenic mechanisms (Spear, 2000), and the present findings may indicate new therapeutic directions if decreased DA signaling is identified in neurodevelopmental disorders.

STAR★METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, David Sulzer (ds43@columbia.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILSS

Animals

All mouse lines, including Pitx3^ak mice (referred to as Pitx3KO) (Nunes et al., 2003), DATKO (Giros et al., 1996) and Drd1a-tdTomato line 6 (Shuen et al., 2008), were obtained from Jackson Laboratories (Bar Harbor, Maine). All mice were maintained on a C57/BL6J background. Mice were housed in same sex groups of 2–4 on a 12-hour light/dark cycle with water and food available ad libitum. Breeding pairs were checked daily for pregnancy and new litters. Mice were used for experiments on the specified postnatal day (±1) in all experiments (Table S1). All experimental procedures were approved by the Columbia University Institutional Animal Care and Use Committee and followed NIH guidelines. No differences were observed between male and female mice, so all data were combined.

METHOD DETAILS

Subjects were randomly allocated to chow treatment group. All experiments were replicated in mice from separate litters. All electrophysiology data were obtained from 3–8 mice per group. N in the figure legend or text indicated the number of cells and the number in parentheses represents the number of mice. For immunohistochemistry and western blotting, N is the number of mice and for FSCV n is the number of slices. For FSCV, slices from at least 3 mice were analyzed for each age. Sample-sizes were estimated based on past studies from our groups. No formal power analyses were conducted. Exclusion criteria for whole-cell patch clamp recordings are as follows: series resistance and the pipette capacitance transients were monitored online and recordings were discarded when series resistance or pipette capacitance transient was unstable, series resistance greater than 20 MΩ or a change in pipette capacitance transient of >20%. For CV, slices were excluded if electrically evoked DA release varied by more than 30% during a 10 min baseline period. In some experiments, datasets were formally tested for outliers by the Grubbs’ test.

Electrophysiology

Acute striatal slices were prepared from animals at the specified ages as previously described (Borgkvist et al., 2015). Briefly, mice underwent cervical dislocation. The brain was removed and placed in ice-cold sucrose cutting solution (in mM): 10 NaCl, 2.5 KCl, 25 NaHCO₃, 0.5 CaCl₂, 7 MgCl₂, 1.25 NaH₂PO₄, 180 sucrose, 10 glucose bubbled with 95% O₂/5% CO₂ to pH 7.4. Coronal slices (250 μm) that included the striatum were collected and allowed to rest at 34°C for 30 min in artificial cerebrospinal fluid (ACSF; in mM): 125 NaCl, 2.5 KCl, 25 NaHCO₃, 1.5 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄ and 10 glucose bubbled with 95% O₂/5% CO₂ to pH 7.4. Slices were then bubbled in ACSF until recordings.

For recordings, slices were placed in a bath at 26°C with constant perfusion with oxygenated ACSF (1.5–2 mL/min). SPNs were identified in the slice using IR/DIC optics by their small cell bodies (approximately 10–15 mm diameter) and characteristic electrophysiological properties including resting membrane potential less than −65 mV, delayed spike in response to a square depolarizing pulse and lack of spontaneous action potential firing. Cells were also examined for fluorescence from the D1-Tomato allele where specified. Whole cell recordings were then established with glass pipettes (3–6 MΩ) with an internal solution of (in mM): 115 potassium gluconate, 20 KCl, 20 HEPES, 1 MgCl₂, 2 MgATP, 0.2 NaGTP adjusted to pH 7.25 with KOH, osmolarity 285 mOsm. Liquid junction potential was not corrected. Data were digitized at 10 kHz and filtered at 5 kHz. All recordings were made within 5 hr of slice preparation.

Membrane properties were extracted from current clamp recordings in which a series of hyper- and depolarizing current steps (20 pA) were injected into the cell. Input resistance was determined by taking the voltage difference between steps in which+60 and −60 pA were injected. Rheobase was determined following a ramp depolarization at 200 pA/s. Capacitance was calculated from tau extracted from a fit of a −5 mV current injection and the input resistance determined as described above. Current-response curves were determined from the number of action potentials fired over the 500 ms square current pulse (Cazorla et al., 2012).

Inward and Kir2 currents were measured in voltage clamp with a V_hold of −60 mV and a series of 10 mV voltage steps. Voltage steps lasted one second and the current measurement used was taken from the steady-state level at the end of the one second step as Kir2 currents are non-inactivating (Cazorla et al., 2012; Mermelstein et al., 1998). In Figure 4, Kir2 currents were measured by using the voltage clamp protocol described above in the presence of tetrodotoxin (TTX; 1 μM) and in the presence of both TTX and CsCl₂ (1mM). The CsCl₂ sensitive current is the Kir2 current (Cazorla et al., 2012). As no difference was observed between conditions in the CsCl₂ resistant current (see Figures 4D and 4E) but only in the current elicited with TTX alone, in Figures 5, 6, and 7, the inward current evoked with the voltage clamp protocol described above was used as a proxy for Kir2 currents without CsCl₂ subtraction (Gertler et al., 2008).

D-type currents were evoked from a V_hold of −90 mV followed by a series of steps to+30 mV (15 mV increments) and return to a V_hold of −60 mV in the presence of TTX (1 μM) (Surmeier et al., 1991). Steady-state current at the end of the 1 s voltage steps was measured.

The effect of neomycin was measured by the addition of 1 mM neomycin (Gabev et al., 1989) to the internal pipette solution. Giga-Ohm seals were established and the cell membrane was ruptured. Immediately, a voltage clamp protocol was begun with cells being held at a voltage of −60 mV and stepped to −130 mV for 500 ms every 10–30 s for 3 min (the time required for the maximal effect of neomycin on Kir2 currents; Rosenhouse-Dantsker et al., 2014). Series resistance was monitored online with a −10 mV voltage step. Percent change of Kir2 current was calculated by subtracting the relative change of Kir2 current from time 0 (membrane rupture) to t = 3 min (maximal effect of neomycin) and normalizing it to the relative change in cells patched without neomycin in the internal solution.

The effect of diC8-PIP₂ (50 μM; Echelon Biosciences) was measured as described for neomycin except that Kir2 current was measured every 30 s for 12 min (maximal effect of diC8-PIP2; Rosenhouse-Dantsker et al., 2014). Percent change of Kir2 currents combining diC8-PIP₂ and neomycin were measured after 12 min.

The effect of forskolin on Kir2 currents was measured by first adding synaptic blockers (picrotoxin, CNQX and AP5) to the bath. Kir2 currents (V_hold −60 mV step to −130 mV) were measured every 30 s. Baseline was measured for 5–10 min followed by addition of forskolin (20 μM) to the perfusate. Kir2 currents were then measured for 12 min.

Cyclic Voltammetry

Electrochemical recordings of endogenous dopamine release were collected as detailed previously (Pereira et al., 2016). Striatal slices were prepared as for electrophysiological experiments except for an increased CaCl₂ concentration (2 mM). During recordings, slices were kept under constant superfusion of oxygenated ACSF (2 mL/min, 34°C). Carbon fiber electrodes (5 mm diameter, cut to ~150 μm length) were placed in the dorsolateral striatum 50–70 μm into the slice. A triangular voltage wave (−450 to+800 mV at 294 mV/ms versus Ag/AgCl) was applied to the electrode every 100 ms and the resulting current was detected with an Axopatch 200B amplifier (Axon Instruments) using a 5 kHz low-pass Bessel Filter setting and 25 kHz sampling rate. Signals were digitized using an ITC-18 board (Instrutech) and recorded with IGOR Pro 6.37 software (WaveMetrics), using custom in-house acquisition procedures (Pereira et al., 2016). Slices were stimulated with a bipolar stainless steel electrode placed ~150 μm from the recording electrode using an Iso-Flex stimulus isolator triggered by a Master-9 pulse generator (AMPI). Single pulses (100 μs × 200 μA) were applied every 2 min and once stable release was achieved, four consecutive peaks were analyzed. To test DAT blockade, the perfusion solution was switched to ACSF containing 10 μM nomifensine after stable baseline was achieved and slices were stimulated every 5 min (Jones et al., 1995). Data were analyzed using an in-house written procedure in IGOR Pro. Electrodes were calibrated by characterizing background-subtracted voltammograms in standard solutions of DA in ACSF.

DA Replacement

Custom chow was ordered from TestDiet (St. Louis, Missouri). PicoLab rodent diet 205053 was supplemented with either 0.2% w/w levodopa (Sigma-Aldrich, St. Louis, Missouri) or 0.0125% w/w Benserazide HCl (Sigma-Aldrich, St. Louis, Missouri). No effect of genotype or treatment was found on litter size (data not shown).

High-Pressure Liquid Chromatography

Brains were removed and striata were dissected rapidly on an ice-cold surface. Tissue DA, dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA) levels were determined by HPLC with electrochemical detection as previously described (Feigin et al., 2001; Mosharov et al., 2006).

Immunohistochemistry

Mice were anesthetized and perfused transcardially with ice-cold 0.9% saline (~10 mL) and then with 4% paraformaldehyde (PFA) in 0.1M phosphate buffer pH 7.4 (~40 mL). Brains were removed and post-fixed overnight in 4% PFA at 4°C. Brains were then sectioned on a vibratome (50 μm) and stored at −20°C in cryoprotectant until analysis. For immunohistochemistry, sections were permeabilized and blocked in 0.1% triton-X, 2% normal donkey serum (Jackson) in 1X TBS and then stained overnight in blocking solution at 4°C containing the following primary antibodies: rabbit anti-RFP (1:500; Rockland) and/or mouse anti-tyrosine hydroxylase (1:1000; Millipore). Secondary antibodies (1:500) were obtained from Invitrogen conjugated to Alexafluor dyes.

Western Blotting

Mice underwent rapid cervical dislocation and brains were removed onto an ice-cold surface. Striata were dissected, flash frozen in liquid nitrogen and stored at −80°. Striata were then placed in RIPA buffer (25 mM Tris HCl pH7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) and homogenized with a hand sonicator. Protein concentration was determined using the BCA protein determination assay (Thermo). Equivalent amounts of total protein (10–25μg) were then loaded onto 10% SDS-PAGE gels. Proteins were then transferred onto PVDF membranes (Immobilon FL, LI-COR). Membranes were blocked with LI-COR blocking buffer (TBS) and incubated with primary antibody overnight. Membranes were then incubated with secondary antibody (LI-COR) and imaged in the Odyssey blot image (LI-COR.

For blots of K_v1.2, Kir2.1 and Kir2.3, membranes were incubated with secondaries conjugated to horseradish peroxidase (Jackson ImmunoResearch). Membranes were then reacted with immobilon western chemiluminescent HRP substrate (Millipore) and imaged using a film developer.

QUANTIFICATION AND STATISTICAL ANALYSIS

All data analysis was conducted blinded to genotype and treatment. Electrophysiology data were analyzed offline using Clampfit software (Molecular Devices, Sunnyvale, California). Statistical analysis was conducted in GraphPad Prism 7 (La Jolla, CA). All bar graphs show the mean ± SEM. Data comparing two variables was analyzed with a two-way ANOVA. Post hoc Bonferroni tests were conducted when significant differences were found with the two-way ANOVA. Data comparing one variable among >2 groups was analyzed with one-way ANOVA and Bonferroni post-tests and among 2 groups a two-tailed t test. Data were not formally tested for parametric distribution.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-Tyrosine Hydroxylase monoclonal	Millipore	Cat # MAB5280
Rabbit anti-Red fluorescent protein polyclonal	Rockland	Cat # 600-401-379
Rabbit anti-DARPP32 monoclonal	Cell Signaling Technology	Cat # 2306S
Mouse anti-beta actin monoclonal	Novus Biologicals	Cat # NB600-501
Mouse anti-Kir2.1 monoclonal	Antibodies Incorporated	Item # 73-210; RRID: AB_11000720
Mouse anti-Kir2.3 monoclonal	Antibodies Incorporated	Item # 75-069; RRID: AB_2130742
Mouse anti-Kv1.2 monoclonal	Antibodies Incorporated	Item # 75-008; RRID: AB_2296313
Rabbit anti- Phospho-(Ser/Thr) PKA Substrate Antibody polyclonal	Cell Signaling Technology	Cat # 9621S
Donkey anti-Mouse IgG (H+L) Secondary Antibody, Alexa 488	Invitrogen	Cat # A-21202
Donkey anti-Rabbit IgG (H+L) Secondary Antibody, Alexa 594	Invitrogen	Cat # A-21207
Donkey anti-Rabbit IgG (H+L) Secondary IRDye 680LT	LI-COR	P/N 925-68023
Goat anti-Mouse IgG (H+L) Secondary IRDye 800CW	LI-COR	P/N 925-32210
Streptavidin, Alexa 488 conjugate	Invitrogen	S11223
Donkey anti-Mouse IgG (H+L) conjugated to HRP	Jackson ImmunoResearch	Code: 715-035-151
Chemicals, Peptides, and Recombinant Proteins
Tetrodotoxin citrate	Tocris	Cat # 1069
Forskolin	Tocris	Cat # 1099/10
diC8-PIP2	Echelon Biosciences	P-4508
Picrotoxin	Tocris	Cat # 1128/1G
D-AP5	Tocris	Cat # 0106
G418 disulfate salt (Neomycin)	Tocris	Cat # 4131/100
Neurobiotin tracer	Vector Laboratories	Cat # SP-1120
CNQX disodium salt	Tocris	Cat # 1045/1
Critical Commercial Assays
BCA Protein Assay Kit	Thermo Fisher Scientific	Prod # 23227
Immobilon Western Chemiluminescent HRP Substrate	Millipore	Cat # WBKLS0500
Experimental Models: Organisms/Strains
Mouse: Pitx3ak/2J	The Jackson Laboratory	Stock No: 000942
Mouse: B6.Cg-Tg(Drd1a-tdTomato)6Calak/J	The Jackson Laboratory	RRID: IMSR_JAX:016204
Mouse: WT: C57BL/6J	The Jackson Laboratory	RRID: IMSR_JAX:000664
Mouse: DAT KO	Courtesy of Marc Caron	Giros et al., 1996
Software and Algorithms
GraphPad Prism	GraphPad	RRID: SCR_002798
Igor	Wavemetrics	RRID: SCR_000325
pClamp	Molecular Devices	RRID: SCR_011323
Image Studio Lite	LI-COR	RRID: SCR_014211
ImageJ	NIH	RRID: SCR_003070

Highlights.

• Striatal dopamine release and projection neuron excitability mature postnatally

• Dopamine is required for the maturation of direct pathway SPN intrinsic excitability

• Maturation of direct pathway SPN excitability arises from PIP₂-Kir2 interactions

• Dopamine must act during a critical period in development for this maturation