CNQX-sensitivity of prolonged glutamatergic corticostriatal responses

124 identified SPNs were recorded for the present study from PD30-60 BAC D₁ or D₂ eGFP mice and from rats of similar age (n = 70 dSPNs and n = 54 iSPNs). Figure 1A illustrates dSPNs neurons expressing eGFP and one neuron double labelled with biocytin (red CY3; see: Methods). Similar photomicrograph for a D₂-eGFP neuron is shown in Figure 1B (scales = 10 μm). The red trace in Figure 1C is a control suprathreshold response, after a single cortical stimulus, in a dSPN. The superimposed black trace is the response obtained, with the same stimulus, after addition of 10 μM CNQX (an AMPA/KA-receptors antagonist) to the bath saline. Superimposed green and black traces in Figure 1D illustrate the same experiment in an iSPN. Inset in Figure 1D illustrates a voltage-clamp recording after cortical stimulation: the initial PSC is followed by a barrage of PSCs of smaller amplitude; such a discharge may last hundreds of milliseconds. Subtractions of the CNQX-sensitive components can be seen in Figure 1F (dSPNs) and 1G (iSPNs). The duration of the CNQX-sensitive components (hundreds of milliseconds) preclude the possibility that they could be explained by monosynaptic events following a single stimulus. In addition, the responses exhibited latency partitions: a preceding early phase - corresponding to monosynaptic PSPs (Figure 1F, G left arrows) [3, 14, 15, 18, 21, 22, 26, 38–42], followed by a late component (Figure 1F, G arrows). A late component denotes the continuous activation of cortical neurons because: first, it is evoked by stimulating within the cortex mostly in a parasagittal corticostriatal preparation, secondly, it is sensitive to CNQX (an antagonist of glutamatergic synapses), third, it is not monosynaptic but extends along time lasting hundreds of milliseconds after the stimulus is over [30–32]; a characteristic of polysinaptically derived responses. Because these prolonged responses are not caused by repetitive stimulation [35] but by a single stimulus, it is logical to infer that they may be caused by recurrent firing within a local microcircuit [23, 33], which has been shown to be the manifestation of interconnected neurons within a neuronal ensemble [35]. Because in parasagittal slices the amount of thalamic fibers and terminals present in both the striatum and cortex would be an invariant variable it cannot be explained by current diffusion to the thalamus [12, 14, 22]. Because these responses cannot be evoked by intrastriatal stimulation with the same stimulus (see below), we infer they arise from local cortical ensembles [30].

Area under corticostriatal responses in dSPNs decreased from 12,960 ± 288 mV · ms in control to 7,907 ± 103 mV · ms during CNQX for about 40% reduction (Figure 1E; n = 6; ***P < 0.001). In iSPNs CNQX reduced the response from 8,619 ± 1,033 mV · ms to 5,158 ± 613 mV · ms, again for about a 40% reduction (Figure 1E; n = 6; *P < 0.05). The similarities between dSPNs and iSPNs responses are that in both cases late responses last enough to be considered polysynaptic and contribute to about half the complex corticostriatal responses. A difference between dSPNs and iSPNs corticostriatal responses is that average areas under iSPNs responses are significantly smaller than those under dSPNs (P < 0.001; cf. Figure 2C and H) [25].

Distinct synaptic contributions of KA- and AMPA-receptors during corticostriatal responses in SPNs

Post-synaptic KA-receptors are present in striatal neurons [2, 39, 43]. Here, we asked whether some class of KA-receptors could be synaptically activated by cortical afferents during suprathreshold corticostriatal responses (i.e.: with endogenous glutamate release). The physiological response of native KA-receptors may have different kinetics than AMPA-receptors [2, 44, 45] and we wanted to observe if this is true for corticostriatal responses. ACET (see Methods) has been reported as a highly selective antagonist for GluK1 KA receptor subunits in expression systems [46–48] and there are GluK1 subunits in the striatum [5, 49] as reported by PCR techniques. GluK2, 3 and 5 subunits have also been detected [1, 43, 50, 51]. KA-receptor heteromers made of different subunits have been reported to be blocked by ACET and other antagonists at low micromolar concentrations [52–54]. Splice and edited variants of these subunits associated with auxiliary proteins composing native receptors have unknown affinities for available antagonists [46, 47, 53]. Due to these considerations, we decided to try ACET to investigate its actions on corticostriatal physiological responses (0.1-1 μM ACET; see Methods) [47].

Red trace in Figure 2A shows a suprathreshold corticostriatal response in a dSPN. Superimposed black trace shows a reduction of control response during application of ACET. A subsequent application of 10 μM CNQX reveals an additional reduction due to blockade of remaining AMPA-receptors or KA-receptors insensitive to ACET (pale grey trace). Figure 2B shows a similar experiment during intrastriatal suprathreshold stimulation in the same dSPN. Histogram in Figure 2C shows that these actions were significant: blockade of KA-receptors in dSPNs decreased the area under the corticostriatal response from 10,633 ± 340 mV · ms to 8,649 ± 231 mV · ms for a 19% reduction (Figure 2C; n = 6; **P < 0.01). The subsequent blockade of AMPA-receptors reduced the remaining response to 5,640 ± 56 mV · ms for a supplemental 35% reduction (Figure 2C; n = 6; ***P < 0.001). ACET plus CNQX reduced the corticostriatal response by about one half: 45 ± 4% (P < 0.001), comparable to percent reduction obtained by CNQX alone (see above). Blockade of KA-receptors in dSPNs also decreased the area under the much briefer response obtained after intrastriatal stimulation (Figure 2B) from 2,031 ± 102 mV · ms to 1,337 ± 67 mV · ms for a 34% reduction (Figure 2C; n = 6; **P < 0.01). A subsequent application of CNQX reduced the remaining response to 241 ± 9 mV · ms or 82% (Figure 2C; n = 6; ***P < 0.001). Whole reduction by ACET plus CNQX blocked most intrastriatal response: ca. 90% (P < 0.001); implying different percent compositions of these components after corticostriatal (ca. 45%) or intrastriatal stimulation (ca. 90%).

Similar experiments were performed in iSPNs (Figure 2F-J), where green trace is the control and black trace shows the reduction produced by ACET. Subsequent addition of CNQX (pale grey trace) shows an additional reduction induced by blockade of remaining AMPA- or KA-receptors insensitive to ACET. Blockade of KA-receptors in iSPNs decreased the area under the corticostriatal synaptic response from 9,822 ± 330 mV · ms to 8,258 ± 430 mV · ms for a 16% reduction (Figure 2H; n = 6; *P < 0.05). A subsequent blockade of AMPA receptors reduced the remaining response to 4551 ± 397 mV · ms for a reduction of 45% (Figure 2H; n = 6; *P < 0.02). The combined action of both blockers in the original response amounted for about one half of the response: 46 ± 3% block (P < 0.01), not significantly different than the reduction obtained with CNQX alone (see above). ACET reduced the area under the response to intrastriatal stimulation from 2,203 ± 110 mV · ms to 1,842 ± 92 mV.ms for a 16% reduction (Figure 2H; n = 6; P < 0.1). Subsequent blockade of AMPA-receptors with CNQX reduced the remaining response to 254 ± 11 mV · ms for a larger block: 86 % reduction (Figure 2H; n = 6; ***P < 0.001).

Based on the above results we conclude that there are KA-receptors that respond to synaptic activation of cortical afferents (i.e.: endogenous glutamate release) in both dSPNs and iSPNs. Because the selectivity of ACET is still under debate (see above) we do not know whether the observed participation correspond to most or only some postsynaptic KA-receptors present in SPNs. Secondly, most glutamatergic actions during intrastriatal stimulation responses in dSPNs depend on AMPA/KA-receptors at −80 mV holding potential (Figure 2A, D, C). In contrast, response to ACET by iSPNs was more variable, although very clear in some cases (Figure 2G, J). But corticostriatal responses dissipate any doubt about ACET-sensitive responses in both dSPNs and iSPNs. However, in both neuronal classes only about half the corticostriatal response could be blocked by these antagonists. Therefore, cortical stimuli appear able to recruit a more complex response than intrastriatal stimuli, suggesting that the arrangement of cortical inputs is capable to activate more receptor classes [25, 42, 55]. In contrast, responses to intrastriatal stimulation go away almost completely after CNQX plus ACET, suggesting that recruited axons lack the necessary circuit arrangement for the response to build up.

In addition, subtractions of the KA- and CNQX-sensitive components (Figure 2D, E, I, J) disclose different time courses: the ACET-sensitive component is slowly rising and decaying and preferentially contributes during the late response. In contrast, the remaining CNQX-sensitive component participates in the fast early and the slow late response, suggesting that initial monosynaptic input is followed by a late polysynaptic barrage [30–32]. To conclude, a main difference between intrastriatal and corticostriatal responses is the duration and magnitude of the late response. More prolonged responses cannot be obtained stimulating within the striatum (even when GABA_A-receptors are blocked, see below), suggesting a different arrangement or origin of afferents and synapses responding to the same stimulus at each site.

Because ACET may present some affinity problems (see above), we performed complementary experiments using the selective AMPA receptor antagonist, 25 μM GYKI 52466 (GYKI). The intention was to antagonize AMPA-receptors without affecting KA-receptors [1, 48, 56, 57]. However, saturating concentrations of GYKI do affect KA-receptors [46, 47, 54]. Therefore, we used non-saturating concentrations of GYKI since our main aim was to find out whether AMPA-receptors are the ones responsible for the early and late latency components during corticostriatal responses.

Red traces in Figure 3A, B show control suprathreshold corticostriatal and intrastriatal responses, respectively, in a dSPN. Addition of GYKI (black traces in Figure 3A, B) reduced the responses. Histogram in Figure 3C shows that GYKI blockade was significant in both cases. Subtracted GYKI-sensitive components (Figure 3D and E; from experiments in Figure 3A and B, respectively) confirm that AMPA-receptors are responsible for early and late components of the response, suggesting that synaptic inputs arrive at different latencies; after the monosynaptic event that follows a single stimulus. As expected, the GYKI-sensitive fraction of the response to intrastriatal stimulation shows an almost complete absence of the late component with the same stimulus; as corresponding to a synaptic response evoked by a single stimulus.

Figure 3F and G show similar experiments in an iSPN, green traces denote controls and black traces denote recordings after 25 μM GYKI. Figure 3H shows that GYKI-induced reduction of the response was significant. As in dSPNs, subtracted GYKI-sensitive components show separate early and late components in the corticostriatal response, but only early responses after intrastriatal stimulus.

GYKI decreased the area under the corticostriatal response of dSPNs from 10,221 ± 292 mV · ms to 8,356 ± 313 mV · ms for an 18% reduction (Figure 3C; n = 7; **P < 0.01). In contrast, blockade of AMPA receptors in dSPNs during intrastriatal stimulation decreased the response from 3,314 ± 166 mV · ms to 1,500 ± 75 mV · ms for a 55% reduction (Figure 3B,C; n = 6; ***P< 0.001). GYKI decreased the area under the corticostriatal response of iSPNs from 6,815 ± 50 mV · ms to 5,374 ± 257 mV · ms for a 21% reduction (Figure 3H; n = 6; **P < 0.01). The action of GYKI on the responses to intrastriatal stimulation was from 2,306 ± 115 mV.ms to 1,486 ± 74 mV · ms (Figure 3H; n = 6; **P < 0.01) for about 36% reduction.

To conclude, the GYKI-sensitive fraction appears more important in the early latency component appearing after intrastriatal stimulation, while late latency components only appear clearly during corticostriatal responses. Taken together, the above data supports the idea that cortical stimulation activates cortical inputs in a way different than that employed during intrastriatal stimulus. One way to explain this difference has been posited in vivo[23]: cortical stimulus activates a group of interconnected excitatory neurons that converge onto the same postsynaptic SPNs. The arrival of these inputs at different latencies would prolong the responses. It is also observed that while AMPA/KA receptors almost completely explain the synaptic responses to intrastriatal stimulation, in the case of corticostriatal entries AMPA/KA receptors only explain about half the response. To explain this behavior one evidence has been given: polysynaptic activation not only involves cortical neurons [30] but also striatal GABAergic inputs (interneurons, other SPNs) eliciting a mixed excitatory plus inhibitory polysynaptic response [25]. Further experiments are needed to identify all neuronal classes participating. An additional component would be the contribution of NMDA-receptors.

NMDA-receptor contribution in corticostriatal and intrastriatal responses of SPNs

Red trace in Figure 4A illustrates a control suprathreshold corticostriatal response obtained after a single stimulus in a dSPN and superimposed black trace shows the response obtained for the same stimulus after addition of an antagonist of NMDA-receptors, 50 μM APV, to the superfusion. APV decreased the magnitude of the response [14, 15, 18, 33, 35]. The same experiment is shown for suprathreshold responses to intrastriatal stimulation in the same neuron (Figure 4B). Histogram in Figure 4C shows that in both cases, APV-blockade was significant. Figures 4D and E show NMDA-sensitive responses (corresponding to experiments in Figure 4A and B, respectively). The APV-sensitive component was larger in corticostriatal responses. In contrast to the KA-sensitive component, the APV-sensitive component was fast rising. This can be explained by monosynaptic activation of NMDA-receptors [3, 14, 15, 18, 38, 41]. In contrast to the GYKI-sensitive component (or the CNQX-sensitive component after ACET-blockade), the APV-sensitive response displays a persistent plateau depolarization instead of separate early and late components. Therefore, this response greatly explains the shape of the suprathreshold corticostriatal response in dSPNs and may be due to a number of factors: the slower kinetics of NMDA-responses, their capacity to produce plateau-potentials, and their capacity to activate intrinsic inward currents [15, 18, 33, 55, 58, 59]. In contrast, the response to intrastriatal stimulation greatly resembles the gradual NMDA-mediated monosynaptic potential described in many neurons [58].

Figures 4F and G show a similar experiment in iSPNs. APV decreased corticostriatal and intrastrial dependent responses in various neurons, although decrease of responses to intrastriatal stimulation was more variable (Figure 4H). Interestingly, although subtracted APV-sensitive component in the corticostriatal response of iSPNs is of long duration (Figure 4I), the response is of lower amplitude (P < 0.001) than that of dSPNs and more similar to that found in other neurons [58]. Two phenomena have been reported to explain these differences: first, shorter and fewer dendrites in iSPNs make them more excitable [42], so that iSPNs are more prone to fire autoregenerative spikes during suprathreshold synaptic responses [25]. In turn, autoregenerative events trigger a stronger repolarization that reduces the amplitude of corticostriatal responses in iSPNs [58, 60]. In support to this more integrative explanation, stimulation of dendritic spines locally with uncaged glutamate produces similar dendritic plateau potentials in both dSPNs and iSPNs [55].

In a sample of dSPNs, the area under suprathreshold cortical response goes from 11,256 ± 436 mV · ms to 6,395 ± 300 mV · ms after APV for a 43% reduction (Figure 4C, n = 9; ***P < 0.001). The decrease of intrastriatal responses was from 3,099 ± 155 mV · ms to 2,334 ± 117 mV · ms for a 24% reduction (Figure 4C; n = 6; *P < 0.05). The decrease in the area under the corticostriatal response in iSPNs after APV goes from 6,921 ± 206 mV · ms to 5,456 ± 266 mV · ms for a reduction of about 21% (Figure 4H; n = 6; **P < 0.01), while during responses to intrastriatal stimulation the reduction was from 2,722 ± 136 mV · ms to 2,280 ± 114 mV · ms for an average 16% decrease (Figure 4G,H; n = 6; P < 0.1), that is, variability and small amplitude of the response precluded statistical significance with this sample size, underlying a main difference with corticostriatal responses. Nonetheless, one example with a clear APV-sensitive component is illustrated in Figure 4J.

It was concluded that each cell class configures its response to glutamate differently, not by possessing different assortments of glutamate receptor classes in their synaptic contacts, but by the different use they make of the prolonged time window conferred by long-lasting synaptic activation: dSPNs apparently maintain larger APV-sensitive components (43%) than iSPNs (21%). This difference between dSPNs and iSPNs in part explains a smaller area under the response in iSPNs. In addition, prolonged time windows for synaptic integration can only be generated when stimuli are delivered in the cerebral cortex; where polysynaptic activation is virtually inescapable [30]. It cannot be generated when stimulus is delivered within the striatum: dSPNs (24%) and iSPNs (16%).

The last conclusion seems more dramatic when no receptor antagonists are used, repetitive discharge is allowed and traces are compared by superimposition (Figure 5): Comparison of suprathreshold corticostriatal responses shows clear differences between dSPNs and iSPNs (Figures 5A-D with insets) [25]. In contrast, superimposition of responses to intrastriatal stimulation cannot distinguish between dSPNs and iSPNs. Differences in the areas under the responses were: 10,612 ± 242 mV · ms (n = 19) for corticostriatal and 2,561 ± 128 mV · ms (Figure 5E, G; n = 12, ***P < 0.001) for intrastriatal stimulus in dSPN. In iSPNs, corticostriatal response area was: 6,527 ± 283 mV · ms (n = 16; P < 0.001); [25] and intrastriatal responses were 2,345 ± 117 mV · ms (Figure 5F, G; n = 10; ***P < 0.001). That is, there were significant differences in the corticostriatal responses between dSPNs and iSPNs (***P < 0.001) [25] but there were no significant differences in responses to intrastriatal stimulation between dSPNs and iSPNs. Therefore, antidromic activation of dispersed cortical axons within the neostriatum cannot activate interconnected converging cortical neurons, perhaps, because most interconnected neurons are in the vicinity of the ones stimulated first during a cortical stimulus [7, 13, 23]. Nevertheless, peak amplitude of responses had no significant differences between dSPNs and iSPNs, although, as reported previously [25], duration at half amplitude does show significant differences in corticostriatal responses between dSPNs and iSPNs. Here, we report that these differences are lost for intrastriatal responses (Figure 5G).

Still, it can be argued that one reason for these differences in the responses depend on stimulation site: it may be that intrastriatal stimulation preferentially activates GABAergic inputs, which in turn, prevents the generation of prolonged synaptic responses. Experiments in Figure 6 show that this is not the case. Here, responses to corticostriatal stimulus are shown (Figure 6 left column) [25] to compare with responses to intrastriatal stimulus of the same strength, in the same cells (Figure 6 right column). Colored traces are controls and superimposed black traces are the responses to the same stimulus obtained after adding 10 μM bicuculline, a GABA_A-receptor antagonist, to the bath saline. Blockade of GABA_A-receptors not only did not prolong synaptic responses to intrastriatal stimulation in dSPNs, but instead, they reduced their duration even more. In dSPNs, a depolarizing GABAergic component contributes to the responses to both cortical and intrastriatal stimulation (Figure 6A, B) [25]. Subtracted bicuculline-sensitive responses show that activation of GABAergic inputs produce two different responses: they decrease the response of dSPNs at the beginning (Figure 6C, D) and enhance the response in a later phase; as it has been modeled [24]. Note that cortical stimuli are more efficient than intrastriatal stimuli to activate bicuculline-sensitive inputs onto dSPNs. In a sample of dSPNs, the area under the late latency component of the suprathreshold corticostriatal response was reduced from 8,268 ± 1014 mV · ms to 3,754 ± 1105 mV · ms after bicuculline for a 45% reduction (Figure 6A, C, n = 8; P < 0.025). The decrease of intrastriatal responses was from 1,874 ± 176 mV · ms to 960 ± 124 mV · ms for a 48% reduction (Figure 6B,D; n = 6; P < 0.001). On the contrary, areas under the corticostriatal responses in iSPNs after bicuculline were enhanced from 5,757 ± 623 to 9,560 ± 665 mV.ms for a 66% increase (Figure 6E, G; n = 11; P < 0.001), while during responses to intrastriatal stimulation the enhancement was from 1,988 ± 195 mV · ms to 3,035 ± 358 mV · ms for an average 52% increase (Figure 6F, H; n = 6; P < 0.002).

Two observations are evident: first, blockade of GABAergic inputs with bicuculline decreases the late depolarizing part of both corticostriatal and intrastriatal responses in dSPNs, while it enhanced both corticostriatal and intrastriatal responses in iSPNs [25]. Secondly, in both dSPNs and iSPNs activation of GABAergic polysynaptic inputs is larger for cortical than striatal stimulation [61] in absolute terms; although similar in percentage. In other words, dissimilarities in responses as a function of stimulation site could not be explained by differences in GABAergic activation.

To further remark this point, experiments in Figure 7 show that main known classes of striatal interneurons are powerfully activated by a single stimulus from the cortex. The suprathreshold corticostriatal response of fast spiking (FS) interneurons (Figure 7A) [19, 62–64] is shown in Figure 7B. It is a slowly decaying depolarization lasting hundreds of milliseconds with a high frequency spike train on top (Figure 7B, G): 305 ± 62 Hz (n = 7). In comparison, dSPNs discharge reaches 141 ± 7 Hz (Figure 7G; n = 10; P < 0.05), while the brief trains of iSPNs attain 248 ± 12 Hz under the same conditions (n = 7; NS). That is, during brief periods iSPNs may reach frequencies as high as those exhibited by FS interneurons. Mean latency for FS corticostriatal synaptic potentials (PSPs) was 1.56 ± 0.18 ms (n = 7; at 0.5X threshold; not shown), while the latency for similar responses in SPNs was 2.6 ± 0.22 ms (n = 46; P < 0.001), suggesting that both cortical and GABAergic inputs reach SPNs quasi-simultaneously [61], explaining why complex suprathreshold corticostriatal responses have mixed inhibitory and excitatory polysynaptic inputs thus comprising a feed-forward activating circuitry.

Interneurons that exhibit persistent low threshold spikes (PLTS) also exhibit prolonged corticostriatal depolarizations lasting hundreds of milliseconds after a single cortical stimulus (Figure 7C) [63, 65, 66]. However, they exhibit little output in terms of action potentials at these holding potential. Instead, they may exhibit autoregenerative events (“low threshold spikes”, arrows in Figure 7C, D), confirming that their way of activation may not necessarily arise from the polarized membrane potentials used for the present comparison (ca. -80 mV) [61]. Latency to subthreshold responses was 2.5 ± 0.3 ms (n = 6); not significantly different to that of SPNs. When action potentials are fired, initial frequency may reach 117 ± 17 Hz (Figure 7G).

Finally, tonically active neurons (TANs), known to be putative large aspiny cholinergic interneurons (Figure 7E) [65, 67–70] may respond with repetitive firing after a single cortical stimulus, with maximal frequencies of 19 ± 6 Hz (n = 9), higher than those reached with intracellular current injections: 6–15 Hz (Figure 7E, F, G), but lower than those attained by any other striatal cell class. Nevertheless, corticostriatal latency to subthreshold PSPs in these neurons was 1.2 ± 0.11 ms, briefer than that of SPNs (P < 0.001), although not significantly different than that of FS interneurons. Depolarizing responses in TANs may, on occasion, last several seconds and be lengthier than those from any other striatal neuron (inset in Figure 7F, Figure 7H).

Taken together, the present results show: first, that GABAergic inputs from interneurons and other SPNs onto postsynaptic SPNs are more efficiently activated from the cortex than from the striatum itself, confirming that different GABAergic entries cannot explain differences in duration between corticostriatal responses and intrastriatal stimulus. Secondly, by themselves, interneurons responses to corticostriatal single stimulus are a direct proof that polysynaptic activation occurs during corticostriatal responses, not only involving cortical cells [23, 30] but even striatal neurons [23, 61], explaining the GABAergic component [25]. Finally, they show that not only corticostriatal responses between SPNs differ, but a comparison of areas under the responses as well as durations at half amplitude, as those seen in histograms of Figures 7G-I, is enough to confirm that each neostriatal neuron class exhibits a particular and distinct corticostriatal response, perhaps reflecting diverse combinations of cortical connections. This point has been shown for FS interneurons [13, 19, 62–64]. Further analysis of these differences is out of the scope of the present report.

Corticostriatal responses do not depend on the cortical area being stimulated

Finally, it can be asked whether the corticostriatal responses that distinguish dSPNs from iSPNs are a result of stimulating a particular cortical location, or the use of a particular slice orientation. To answer these questions we stimulated different cortical areas with different slice orientations (Figure 8A-C): frontoparietal cortex (orientation = sagittal, n = 55 neurons, Figure 8A scheme at left), temporal cortex (orientation = horizontal, n = 15 neurons, Figure 8B) and frontal cortex (orientation = sagittal –shown–, and horizontal, n = 17 neurons, Figure 8C). It was observed that characteristic responses of dSPNs and iSPNs were maintained no matter the cortical area being stimulated [6–11] or whether the thalamus was present.

We conclude that differences in the responses were independent of the cortical area stimulated and on the assortment of receptors being activated, and more dependent on the different intrinsic properties that these neurons exhibit [60]. In fact, more differences have been documented after intrasomatic stimulation [42, 60, 71].

Prolonged corticostriatal responses

Previous evidences, in vivo and in vitro, have shown that a single cortical stimulus may initiate sequential polysynaptic activation of pyramidal cells [23, 30].

Here, we show in vitro, that a single stimulus in the cortex evokes a complex and prolonged suprathreshold response, separate from the monosynaptic response, lasting hundreds of milliseconds, and sensitive to ionotropic glutamate receptors antagonists. Besides projection neurons (SPNs), several classes of striatal neurons including FS-, PLTS-, and TAN-interneurons could powerfully be activated by cortical stimuli, supporting and extending direct evidence of polysynaptic activation that includes not only cortical but also striatal neuron pools. It is important to demonstrate in vitro, that a long-lasting arrival of inputs may underlie prolonged corticostriatal responses: microcircuit dynamics demonstrated in the corticostriatal slice [35, 36] is sustained by long-lasting up-states and recurrent burst firing [33] that allows neuronal synchronization. We infer that an up-state may be the result of the activity of a cortical ensemble. This information allows future comparisons and interpretations of changes in dynamics after decortication, or different forms of deafferentation, in control and Parkinsonian subjects.

Prolonged activation of excitatory inputs has been explained by the activation of neighboring cells by the first stimulated cortical neurons [12, 23]. Sequentially activated cortical neurons [26] would then converge onto the same postsynaptic SPNs at different times, thus shaping up a response that may last hundreds of milliseconds setting a window for synaptic integration and correlated or synchronous firing [35]. Since striatal neurons are also activated by the stimulus, both cortical and striatal inputs together shape up the striatal output conveyed by SPNs.

An obvious question is why intrastriatal stimuli could not induce these prolonged responses. Our working hypothesis, confirmed so far, is that intracortical connections cannot be stimulated in these conditions, instead, stimulation within the neostriatum activates sparsely extended cortical axons coming from distant cortical areas that would unlikely be interconnected. Antidromic activation is not dense enough to stimulate neurons that are in the vicinity of the first stimulated cells [23]. The alternative explanation, that intrastriatal stimuli better activate inhibition, was discarded in the present work: it was shown that polysynaptic activation of inhibitory entries is larger when the stimulus is within the cortex and not within the striatum.

To conclude: accepting that cortical stimulation generates recurrent bursting and correlated firing due to the sequential activation of cortical neuron sets [30, 33, 72–75], and that these neurons sets may converge in target postsynaptic sets of SPNs, then, prolonged SPNs responses reflect circuitry processing among cortical and striatal neuron ensembles [34]. Striatal ensembles can also be activated in sequence [35], perhaps, encoding motor programs and procedures. Still an improbable alternative is that some thalamic inputs may be stimulated in a non equal manner by current diffusion. However, in vivo experiments point right toward the opposite phenomenon [76]. Finally, responses obtained in vitro are not equal to those evoked in vivo: a long-lasting disinhibition due to intracortical connections activating nets of inhibitory cortical interneurons [12, 22, 23, 27] appears to be small or lacking in the present preparation.

Roles of glutamate receptors during corticostriatal responses

Prolonged corticostriatal responses displayed GYKI-, ACET-, CNQX- and APV-sensitive components in both dSPNs and iSPNs. In particular, the contribution of GYKI- and CNQX-sensitive components in both neuron classes exhibited early and late components. We conclude that both projection neuron classes have qualitatively similar assortments of glutamate receptors. Therefore, differences between the magnitudes of orthodromically activated responses (e.g., areas under the response) are not due to differences in expressed glutamate receptors as experiments with uncaged glutamate confirm [55], but by different ways of integration due to anatomical and intrinsic differences (e.g., different sets of G-protein coupled receptors and modulation - unpublished). Taken all together, these results show that up-states appear not to be a simple spatial or temporal summation of otherwise uncoordinated synaptic inputs. They are the manifestation of ensembles processing and circuitry modulation.

The ACET-sensitive component suggests either that glutamatergic synapses in SPNs have GluK1 subunits [5, 49], or that heteromers are sensitive to ACET [46, 53]. The postsynaptic presence of an ACET-sensitive component with a physiological stimulus (release of endogenous glutamate) confirms previous suggestions about their potential postsynaptic importance [2, 51].

A particular interest of the KA-receptors mediated responses is that their slow duration may ignite intracellular signaling cascades that cross-talk with other modulatory signaling cascades (e.g., dopaminergic; cholinergic). Further research is needed to observe this possibility as well as its interactions with voltage-gated currents.