Influence of cortical synaptic input on striatal neuronal dendritic arborization and sensitivity to excitotoxicity in corticostriatal coculture

Corticostriatal coculture from embryonic mouse brain was used to study the effect of varying cortical-to-striatal neuron ratio on striatal neuronal properties at 18 days in vitro. Reduced cortical-striatal plating ratio resulted in enhanced loss of striatal spiny projection neurons (SPNs), reduced SPN dendritic complexity and length, altered synaptic/extrasynaptic NMDA receptor distribution, and increased susceptibility to excitotoxic death. Thus cortical-striatal plating ratio influences SPN structure and function, which should be considered in experimental design.

Abstract

Corticostriatal cocultures are utilized to recapitulate the cortex-striatum connection in vitro as a convenient model to investigate the development, function, and regulation of synapses formed between cortical and striatal neurons. However, optimization of this dissociated neuronal system to more closely reproduce in vivo circuits has not yet been explored. We studied the effect of varying the plating ratio of cortical to striatal neurons on striatal spiny projection neuron (SPN) characteristics in primary neuronal cocultures. Despite the large difference in cortical-striatal neuron ratio (1:1 vs. 1:3) at day of plating, by 18 days in vitro the difference became modest (∼25% lower cortical-striatal neuron ratio in 1:3 cocultures) and the neuronal density was lower in the 1:3 cocultures, indicating enhanced loss of striatal SPNs. Comparing SPNs in cocultures plated at a 1:1 vs. 1:3 ratio, we found that resting membrane potential, input resistance, current injection-induced action potential firing rates, and input-output curves were similar in the two conditions. However, SPNs in the cocultures plated at the lower cortical ratio exhibited reduced membrane capacitance along with significantly shorter total dendritic length, decreased dendritic complexity, and fewer excitatory synapses, consistent with their trend toward reduced miniature excitatory postsynaptic current frequency. Strikingly, the proportion of NMDA receptors found extrasynaptically in recordings from SPNs was significantly higher in the less cortical coculture. Consistently, SPNs in cocultures with reduced cortical input showed decreased basal pro-survival signaling through cAMP response element binding protein and enhanced sensitivity to NMDA-induced apoptosis. Altogether, our study indicates that abundance of cortical input regulates SPN dendritic arborization and survival/death signaling.

NEW & NOTEWORTHY

Corticostriatal coculture from embryonic mouse brain was used to study the effect of varying cortical-to-striatal neuron ratio on striatal neuronal properties at 18 days in vitro. Reduced cortical-striatal plating ratio resulted in enhanced loss of striatal spiny projection neurons (SPNs), reduced SPN dendritic complexity and length, altered synaptic/extrasynaptic NMDA receptor distribution, and increased susceptibility to excitotoxic death. Thus cortical-striatal plating ratio influences SPN structure and function, which should be considered in experimental design.

since the groundbreaking finding that immature neurons can develop and be maintained in vitro (Fischbach 1970; Harrison 1910), cultured neuronal systems as models of their in vivo counterparts have played an essential role in neuroscience investigation for over a century (Millet and Gillette 2012). Generally, neuronal cultures are generated from primary neurons dissociated from a single brain region such as cortex, hippocampus, or striatum; however, a culture from one type of tissue cannot reproduce any interregional connections that exist in the brain. This shortcoming was addressed when coculture of neurons from different brain tissues was developed, providing a valuable tool for studying the interactions of neurons from distinct brain regions (Kaufman et al. 2012; Segal et al. 2003; Snyder-Keller 2004). The coculture system is also used to study neuron/nonneuron cell interactions, such as neuron-muscle (Fischbach 1970) or neuron-glia (Li et al. 2014; Melcangi et al. 1997), or to distinguish the contribution to disease phenotype of pre- vs. postsynaptic neuronal expression of a disease-causing gene (Milnerwood et al. 2012; Parsons et al. 2014). Despite these achievements, a few questions remain, including that of how to optimize the ratio of different neurons in culture to more closely represent the physiological condition in the brain.

Studies of neurodegenerative diseases suggest that synaptic alterations precede overt clinical manifestations and neuronal loss (Marchetti and Marie 2011; Parsons and Raymond 2014; Raymond et al. 2011). In Huntington's disease (HD), an inherited disorder in which the striatum shows the most severe degeneration, one of the earliest pathogenic changes is impairment of glutamatergic cortical connections to GABAergic striatal medium-sized spiny projection neurons (SPNs; Cepeda et al. 2003; Joshi et al. 2009; Milnerwood and Raymond 2007; Milnerwood et al. 2010). Consequently, corticostriatal coculture is superior to striatal-only culture as an in vitro model for investigating pathogenic mechanisms in such a neurodegenerative disorder. Moreover, SPNs in coculture are morphologically and synaptically more developed and similar to striatal neurons in the brain (Kaufman et al. 2012; Lalchandani and Vicini 2013; Segal et al. 2003; Tian et al. 2010). However, coculture preparation methods vary between studies, and the influence of cortical-to-striatal neuron plating ratio in coculture on striatal SPN physiology, including dendritic arborization, NMDA-type glutamate receptor subcellular distribution, and survival-death signaling—an important focus of HD research—has not been explored. In this article, we address this question by varying the coculture ratio of cortical to striatal neurons and comparing striatal SPN physiology and morphology, receptor distribution, pro-survival signaling, and sensitivity to cell death.

METHODS

Cell culture.

All procedures and animal care were approved by the University of British Columbia, according to the guidelines of the Canadian Council for Animal Care. All cultures were prepared from wild-type FVB/N embryonic day 18 (E18) mouse cortical and striatal tissue, as previously described (Parsons et al. 2014); each E18 brain yielded ∼1.5 million striatal and 3 million cortical neurons. Briefly, after the mother was anesthetized, the embryos were extracted, and their brains were quickly removed and dissected in ice-cold Hank's balanced salt solution (Invitrogen). Tissues were chopped, followed by digestion in trypsin (Invitrogen) and dissociation in trypsin inhibitor solution (Invitrogen). Striatal neurons were labeled by transfection with yellow fluorescent protein (YFP) construct on a β-actin promoter (a gift from A. M. Craig, University of British Columbia) as follows: 1.5 million striatal cells were suspended in 100 μl of electroporation buffer (Mirus Bio, Madison, WI) mixed with 2 μg of DNA and then electroporated (program 05, Amaxa Nucleofector; Lonza, Walkersville, MD) in a cuvette, according to the manufacturer's instructions. Transfected striatal neurons then were blended with 1.5 or 0.5 million cortical neurons, and the mixture was diluted in DMEM (Invitrogen) with 10% fetal bovine serum (Sigma) to give the same final concentration of cells for the two different mixtures. Five hundred microliters of the mix were plated in each well with an approximate cell density of 650 cells/mm² in a 24-well plate. Each well had a 12-mm coverslip (Marienfeld, Lauda-Königshofen, Germany) precoated with poly-d-lysine hydrobromide (P6407; Sigma) in borate buffer (12.5 mM borax, 50 mM boric acid; Sigma). Two to four hours after the cells were plated, the medium was switched to 500 μl of plating medium (Neurobasal medium, B27, glutamine, and penicillin/streptomycin; Gibco). At 3 to 4 days in vitro (DIV3-4), an extra 500 μl of plating medium were added to each well, and subsequent half-medium changes occurred every 3–7 days thereafter. All experiments were conducted at DIV17-19. YFP-positive cells with SPN morphological characteristics were chosen for assessment in both immunocytochemical and electrophysiological experiments, as previously described (Kolodziejczyk and Raymond 2016).

Electrophysiology.

Whole cell patch-clamp recording was carried out under voltage clamp. The Axopatch 200B amplifier (Axon Instruments) and pClamp 10.2 software (Molecular Devices, Palo Alto, CA) were used to acquire the data. Coverslips were transferred to the recording chamber and perfused with external bath solution prepared with (in mM) 167 NaCl, 2.4 KCl, 10 glucose, 10 HEPES, 2 CaCl₂, 1 MgCl₂, 0.05 picrotoxin (PTX), and 0.0003 tetrodotoxin (TTX); pH 7.3 with NaOH, 310–320 mosM (Parsons et al. 2014). The miniature excitatory postsynaptic currents (mEPSC), recorded at a holding potential of −70 mV, were digitized at 10 kHz and filtered at 1 kHz. Recording electrodes (3–6 MΩ) were filled with an internal solution consisting of (in mM) 145 K-gluconate, 1 MgCl₂, 10 HEPES, 1 EGTA, 2 adenosine 5′-triphosphate magnesium salt, and 0.5 guanosine 5′-triphosphate sodium salt; 280 mosM. The access resistance was typically between 10 and 20 MΩ but not higher than 25 MΩ. To assess SPN excitability, TTX and PTX were omitted from the recording solution, and action potential firing rate was measured in response to a series of 400-ms current pulses of amplitudes ranging from −250 to 300 pA, incremented by 50 pA. For the input-output curve, the same protocol was applied except that the bath solution contained TTX (0.3 μM) and PTX (50 μM). The curve was plotted based on the mean values of stable membrane potentials recorded during the final 100 ms of each current pulse.

To record whole cell and extrasynaptic NMDA currents, coverslips were placed in bath solution consisting of (in mM unless stated otherwise) 167 NaCl, 2.4 KCl, 10 glucose, 10 HEPES, 2 CaCl₂, 10 μM MgCl₂, 100 μM PTX, 0.3 μM TTX, and 10 μM glycine; pH 7.3 with NaOH, 310–320 mosM. Internal recording solution consisted of (in mM) 130 cesium methanesulfonate, 5 CsCl, 4 NaCl, 1 MgCl₂, 10 HEPES, 5 EGTA, 5 lidocaine, 0.5 GTP, 10 sodium phosphocreatine, and 5 adenosine 5′-triphosphate magnesium salt; pH 7.2, 290 mosM. Rapid application of drugs or wash solution was achieved by using a perfusion system with a theta tube (Harvard Apparatus, Saint-Laurent, QC, Canada). N-methyl-d-aspartate (NMDA; 1 mM; M3262; Sigma-Aldrich) was applied for 3 s every 30 s at a holding potential of −70 mV while the initial whole cell NMDA current was recorded and to measure extrasynaptic NMDA current after synaptic NMDA receptors were blocked with MK801 (10 μM; Sigma). Blockade of synaptic NMDA receptors was achieved by holding cells at −80 mV, superfusing for 2 min with 4-aminopyridine (4-AP; 10 μM; Tocris) to maximize synaptic activity, and superfusion with 4-AP together with MK801 for 3 min, followed by washout for 30 s. Cells in which the access resistance changed by >25% were discarded. The peak current was normalized to cell capacitance (C_m), and charge transfer density was calculated as total area of evoked NMDA current (pA × ms) normalized to cell capacitance.

Apoptotic assay.

SPN sensitivity to excitotoxicity was assessed as previously described (Buren et al. 2014). Briefly, in a 37°C, 5% CO₂ humidified incubator, cultures were treated with NMDA at concentrations of 30 and 50 μM in at least 200-µl conditioned medium for 15 min, washed twice with new plating medium, switched back to the conditioned medium, and incubated for 1 h. Cultures were then fixed with 4% paraformaldehyde + 4% sucrose (30 min), permeabilized with 1% Triton X-100 (Sigma) in PBS (PBST; 5 min), and incubated with 10% normal goat serum (NGS; Vector Laboratories) at room temperature (RT; 40 min) in 0.03% PBST. To detect YFP, the neurons were incubated with a chicken polyclonal anti-green fluorescent protein (GFP) antibody (1:1,000–2,000; ab13970; Abcam) in 2% NGS PBST at RT for 3 h. Neurons were then washed 3 times with PBST, incubated with a goat anti-chicken Alexa Fluor 488 antibody (1:1,000; A-11039; Invitrogen) in 2% NGS PBST at RT for 1.5 h, washed again 3 times with PBST, stained with 5 μM Hoechst 33342 (Invitrogen) at RT for 10 min in PBST, and washed again 3 times with PBST, and the coverslips were mounted on slides with Fluoromount-G (0100-01; SouthernBiotech). Cells with round, small, densely compacted nuclei and few neuritic processes were counted as apoptotic neurons, whereas cells with loosely compacted, large nuclei and extensive dendritic projections were regarded as healthy cells (Buren et al. 2014). At least 200 YFP-positive cells were counted in each condition of each culture batch under a ×63 oil-immersion lens (1.4 numerical aperture) with a Zeiss Axiophot epifluorescence microscope.

Immunostaining.

Basic characterization of cell populations was carried out using Hoechst 33342 to assess nuclear morphology, an antibody against microtubule-associated protein 2 [MAP2; 1:200; MA5-12823; Thermo Scientific; Alexa Fluor 568 (1:1,000; A-11031; Invitrogen) was used as secondary antibody] to label all neurons, and an antibody against dopamine- and cAMP-regulated phosphoprotein 32 [DARPP-32; 1:500; 2306S; Cell Signaling Technologies; Alexa Fluor 488 (1:1,000; A-11008; Invitrogen) was used as secondary antibody] to identify striatal SPNs. At DIV17-19, cells were fixed, permeabilized, blocked, and immunostained with primary antibodies in 2% NGS PBST at RT for 3 h. Neurons were then washed, incubated with secondary antibodies in 2% NGS PBST at RT for 1.5 h, washed, stained with Hoechst 33342, washed, and mounted. Basic characterization images were acquired in ×20 fields using a Zeiss Axiophot epifluorescence microscope. Live cells were differentiated from dead cells on the basis of their nuclear morphology (as described above; Buren et al. 2014) and counted with the Cell Counter plugin in ImageJ software (1.47v; NIH, Bethesda, MD) with or without merging channels for overlapping signals. Results are displayed in Table 1 as average numbers ± standard error in a single ×20 field. A total of 77–78 ×20 fields from 5 batches of cocultures were used for live cell and DARPP-32 staining analysis, whereas 15–18 ×20 fields from 2 culture batches were assessed for MAP2 staining statistics.

Table 1.

Characterization of cell densities in corticostriatal cocultures plated at 1:1 and 1:3 ratios

	Cortical-Striatal Neuron Plating Ratio
	1:1	1:3
Live cells	30.0 ± 2.3	22.9 ± 1.5*
MAP2+	21.3 ± 2.2	16.4 ± 1.2†
DARPP-32+	3.4 ± 0.2	3.3 ± 0.3
MAP2+ and DARPP-32−	17.2 ± 1.7	12.7 ± 1.0‡

Values are average (±SE) numbers of cells in a single ×20 microscope field; n values in legend indicate the numbers of ×20 fields counted, with the numbers of culture batches shown in parenthesis. Despite the same plating density, both the number of live cells and the number of MAP2-positive (MAP2⁺) cells per ×20 field are significantly lower in 1:3 compared with 1:1 cocultures [live cell density: n = 77(5) for 1:1 and n = 78(5) for 1:3 plating ratio; paired t-test,

^*P = 0.0011; MAP2⁺ cell density: n = 15(2) for 1:1 and n = 18(2) for 1:3 plating ratio; paired t-test,

^†P = 0.0216]. Surprisingly, the density of DARPP-32⁺ cells is similar between conditions even though the MAP2⁺ but DARPP32⁻ cell density is low in 1:3 cocultures [DARPP-32⁺ cell density: n = 77(5) for 1:1 and n = 78(5) for 1:3 plating ratio; paired t-test, P = 0.5761; MAP2⁺ and DARPP32⁻ cell density: n = 15(2) for 1:1 and n = 18(2) for 1:3 plating ratio; paired t-test, ‡P = 0.0172].

Sholl analysis was conducted in ImageJ using the Sholl analysis plugin (http://fiji.sc/Sholl_Analysis). The dendritic trees of striatal SPNs were visualized by immunostaining against DARPP-32 and were traced with the segmented line tool in ImageJ. Concentric circles with radii from 10 to 200 μm, stepped by 5 μm, were centered on the soma, and the number of intersections with the traced lines was measured. The total dendritic length is a summation of lengths of all traced lines.

Analysis of excitatory synapses was conducted with the same immunostaining protocol, with the following exceptions: PBS was used for wash instead of PBST; methanol (−20°C, 5 min) and PBST (RT, 5 min) were applied for permeabilization; primary antibodies were the chicken polyclonal anti-GFP antibody (to label striatal SPNs, as above), a mouse anti-postsynaptic density-95 (PSD-95) antibody (1 h at RT, 1:1,000; MA1-045; Thermo Scientific), and a guinea pig anti-vesicular glutamate transporter 1 (vGlut1) antibody (4°C overnight, 1:4,000; AB 5905; Chemicon) in 2% NGS PBST; and secondary antibodies were Alexa Fluor 488 goat anti-chicken, Alexa Fluor 568 goat anti-mouse, and aminimethylcoumrin-conjugated donkey anti-guinea pig (1:100; 706-155-148; Jackson ImmunoResearch) at RT for 1.5 h. Coverslips were washed with PBS three to four times before being mounted. Images were acquired with a Zeiss Axiophot epifluorescence microscope under a ×63 oil-immersion lens (1.4 numerical aperture) by taking 10–15 sections with 0.23-μm steps in the z-plane with the same exposure time for each channel on each sample. The three best-focused sections were extracted to average, using the extended depth of focus function, and exported as gray-value TIFF files for individual channels. Blinded to conditions and using the green channel, the experimenter chose three 15- to 30-μm segments of primary and/or secondary dendrite at a distance of 40 to 100 μm from the soma with the polygon selection tool in ImageJ. The PSD-95 and vGlut1 images were thresholded manually within the chosen dendritic segments to remove background staining and to isolate puncta. Binary images of puncta were generated and analyzed for puncta density and colocalization using the Analyze Particles tool and the colocalization plugin of ImageJ (Parsons et al. 2014). Average values of the three chosen areas of interest from individual cells were used in the final analyses.

For measuring phosphorylated cAMP response element binding protein (pCREB), a reflection of pro-survival signaling, we used the chicken polyclonal anti-GFP antibody (to identify striatal SPNs, as above) and a mouse monoclonal anti-pCREB antibody [1:500; 05-667; Millipore; Alexa Fluor 568 (A-11031; Invitrogen) was used as secondary antibody]. Cells were stained as described above. Images for pCREB signal analysis were acquired with a Zeiss Axiophot epifluorescence microscope under a ×63 oil-immersion lens. Nuclear-to-cytoplasmic pCREB ratios were obtained by comparing the average pCREB signal intensity in the nucleus to that of three randomly chosen areas covering most of the cytoplasmic region (as illustrated in Fig. 6Ai) using ImageJ.

Fig. 6.

Reduction in cortical neuron proportion reduces pro-survival signaling of striatal SPNs and enhances their vulnerability to excitotoxic injury. Ai: typical photomicrographs of striatal SPNs (soma) labeled with YFP, whose signal is enhanced with green secondary antibody; cell nuclei are visualized with Hoechst (blue), and pCREB staining is shown in red. Aii: the nuclear-to-cytoplasmic pCREB ratio is significantly lower in SPNs in cocultures plated at 1:3 vs. 1:1 [n = 67(4) for both conditions; paired t-test, *P < 0.05]. Bi: representative photomicrographs of live and apoptotic striatal SPNs in corticostriatal cocultures plated at 1:1 and 1:3 ratios. Live neurons have a large, loosely compacted nucleus and show abundant dendritic processes, whereas a small, round, compacted nucleus associated with few if any dendritic processes is a signature of an apoptotic cell. Bii: 1:3 corticostriatal coculture exhibits a higher proportion of apoptotic cells after NMDA challenge (culture batch: n = 5, 5, and 4 for both 1:1 and 1:3 plating ratios at 0, 30, and 50 μM NMDA; 2-way ANOVA: interaction, *P = 0.0325; ratio, ***P = 0.0006; and NMDA concentration, ***P < 0.0001. Bonferroni's posttests: *P < 0.05 at 30 μM; **P < 0.01 at 50 μM NMDA).

Data analysis.

Pooled data are presented as means ± SE. Figures were prepared with Prism 5 (GraphPad Software), and graphs were created with Adobe Illustrator CS5 (Adobe Systems). In all figures, n values represent the numbers of cells analyzed in given measurements, followed by the culture batch numbers in parentheses unless notified.

RESULTS

Influence of cortical-striatal plating ratio on survival and basic membrane properties of striatal SPNs after 18 days in culture.

To determine the effect of cortical neuron abundance on the properties of striatal SPNs, we cocultured cortical and striatal neurons at a cortical-striatal plating ratio of 1:1 vs. 1:3 and then compared these two conditions at DIV18 for all experiments. Cell population density at DIV18 was characterized using Hoechst-stained nuclear morphology to determine density of live cells, MAP2 staining to identify all neurons, and an antibody against DARPP-32 as a marker of striatal SPNs (Kolodziejczyk and Raymond 2016; Wu et al. 2016). Results are shown in Table 1. We found that the average number of live cells per ×20 field, including both cortical and striatal neurons as well as MAP2-negative cells, was significantly lower in cocultures plated at the 1:3 vs. 1:1 ratio, as was the density of MAP2-positive cells (∼23%). Moreover, despite a 50% higher plating density of striatal neurons in the 1:3 coculture condition, the density of DARPP-32-positive cells was similar at DIV18 for 1:3 and 1:1 cocultures. Taken together, our results suggest that cocultures plated at the lower cortical ratio (1:3) undergo greater loss of striatal SPNs. As well, the relatively low ratio of DARPP-32-positive neurons compared with MAP2-positive cells (all neurons) in both coculture conditions suggests that striatal SPNs are generally more fragile in these cocultures and/or slow to fully mature (i.e., some may not express detectable levels of DARPP-32 by DIV18). Although we cannot make definite conclusions from these results about the ratio of cortical glutamatergic pyramidal neurons to striatal SPNs at DIV18, our data indicate that the density of live MAP2-positive/DARPP-32-negative (mainly cortical) neurons at DIV18 is only ∼25% lower in the cocultures plated at a 1:3 vs. 1:1 ratio, whereas there is no significant difference in striatal SPN density.

To determine how a change in cortical input affects basic membrane properties of SPNs, we measured the resting membrane potential, action potential firing rates, current-voltage (I-V) response, membrane capacitance, and input resistance of SPNs in corticostriatal cocultures plated at the 1:1 and 1:3 ratios. SPNs in the cocultures plated at the lower cortical ratio showed significantly reduced membrane capacitance (Fig. 1A), suggesting that a reduction in cortical-striatal neuron ratio decreases the soma size and/or dendritic growth of striatal SPNs. However, the input resistance (Fig. 1B) and resting membrane potential (Fig. 1C) were similar for SPNs in the two cocultures, indicating that the reduction of cortical neurons does not significantly alter these basic striatal neuronal membrane properties. In addition, SPN excitability curves (Fig. 1, Di and Dii) were similar, and the I-V response curve (Fig. 1, Ei and Eii) only trended toward depressed in the cocultures plated at the lower cortical ratio. Together, these results show that the cortical-striatal neuronal plating ratio has only a modest impact on the SPN basic membrane properties.

Fig. 1.

Cortical-to-striatal neuron ratio affects some SPN membrane properties in coculture. The numbers of cells analyzed are presented as n values with the culture batch numbers in parentheses. A: SPN membrane capacitance is significantly smaller in corticostriatal cocultures plated at 1:3 than at 1:1 ratio [n = 19(3) in 1:1 and n = 20(4) in 1:3 ratio; unpaired t-test, *P < 0.05]. B and C: membrane input resistance [B; n = 19(3) in 1:1 and n = 20(4) in 1:3 ratio; unpaired t-test, P > 0.05] and resting membrane potential [C; n = 19(3) for both ratios; unpaired t-test, P > 0.05] are similar for the 2 coculture conditions. D: current-induced firing rate is not significantly different. Di: representative traces of membrane potential changes under current injections. Dii: pooled data for current injection-induced firing rates [n = 19(3) for 1:1 and n = 18(3) for 1:3 ratio; repeated-measures 2-way ANOVA: interaction, P = 0.8296; cortical-striatal ratio, P = 0.6781; and current, ****P < 0.0001]. E: I-V curve is not significantly different between the 2 coculture conditions in the presence of TTX and PTX. Ei: representative traces for membrane potential changes in response to a range of current injections (I_inject). Eii: pooled data for I-V curve [n = 24(4) for 1:, and n = 22(4) for 1:3 ratio; repeated-measures 2-way ANOVA: interaction, P = 0.9989; cortical-striatal ratio, P = 0.1402; and current, ***P < 0.0001].

Impaired SPN dendritic arborization in cocultures plated at the lower cortical ratio.

The reduction in SPN membrane capacitance in the cocultures plated at the lower cortical ratio suggests that cortical neuronal abundance influences striatal SPN dendritic length, soma size, and/or spine density. To test this, we assessed dendritic arborization by visualizing striatal SPNs with antibody staining against DARPP-32. Our results showed a significant reduction in the total dendritic length of SPNs in cocultures plated at the lower cortical ratio (Fig. 2, A and B), suggesting underdevelopment of the dendritic arbor. Consistent with that result, Sholl analysis indicated a reduced complexity of dendritic arborization in cocultures plated at the 1:3 ratio (Fig. 2C). Altogether, these data suggest that a relative change in cortical neuron abundance affects SPN neurite growth and/or maintenance.

Fig. 2.

Change in cortical relative to striatal cells regulates SPN dendritic arborization. A: representative tracings of dendritic arbor of SPNs from both conditions. B: SPNs grown with lower cortical cell ratio (1:3) have shorter total dendritic length compared with the other condition [n = 44(4) and 45(4) for 1:1 and 1:3 conditions, respectively; paired t-test, ***P < 0.0001]. Moreover, dendritic complexity indicated by Sholl analysis is reduced significantly in coculture plated at lower cortical ratio [C; n = 44(4) and 45(4) for 1:1 and 1:3 conditions, respectively; repeated-measures two-way ANOVA: interaction, ***P < 0.0001; ratio, *P = 0.0388; and radius, ***P < 0.0001; Bonferroni's posttests: *P < 0.05 at 55 and 70 μm].

Cortical-striatal ratio affects the density of excitatory synapses onto SPNs.

To determine if the cortical-striatal ratio in cocultures influences excitatory cortical synapses onto SPNs, we measured spine density and excitatory synapse density along dendritic segments >40 μm away from the soma with immunocytochemistry, staining the pre- and postsynaptic markers vGlut1 and PSD-95, respectively. Our results showed a trend toward reduction in spine density per unit length (∼15%; Fig. 3, A and Bi) and a significant decrease in the density of colocalized puncta of PSD-95 and vGlut1 along the dendrites (∼24%; Fig. 3, A and Bii) in cocultures plated at the 1:3 compared with 1:1 ratio. Interestingly, the density along dendrites of both individual PSD-95 and vGlut1 puncta only showed a trend toward reduction in the cocultures plated at the lower cortical ratio (Fig. 3, Ci and Cii), suggesting that cortical-striatal ratio does not significantly affect development of major pre- and postsynaptic elements of striatal excitatory synapses, but only synapse formation in culture.

Fig. 3.

Cortical-to-striatal neuron ratio in coculture impacts dendritic excitatory synapse density. A: representative photomicrographs of immunostaining for YFP, PSD-95, and vGlut1. White lines, generated in ImageJ, outline the shapes of spines and dendrites in higher magnification graphs. Green arrows indicate colocalized PSD-95 and vGlut1 puncta (far right). Bi: spine density along the SPN dendrites >40 μm away from soma shows a nonsignificant (n.s.) decrease in 1:3 compared with 1:1 cocultures [n = 45(3) for both 1:1 and 1:3 plating ratios; paired t-test, P = 0.1127]. Bii: density of colocalized PSD-95 and vGlut1 puncta along dendrites is lower in 1:3 than in 1:1 plating ratio [n = 45(3) for both 1:1 and 1:3 conditions; paired t-test, *P = 0.0213]. C: individual PSD-95 (Ci) and vGlut1 (Cii) puncta densities along dendrites trend lower for 1:3 condition compared with SPNs from 1:1 corticostriatal cocultures, but neither is statistically significant [PSD-95 puncta density: n = 45(3) for both 1:1 and 1:3 conditions; paired t-test, P = 0.0679; vGlut1 puncta density: n = 45(3) for both conditions; paired t-test, P = 0.089].

Miniature EPSC frequency recorded from SPNs in cocultures plated at the lower cortical ratio showed a strong trend toward a lower mean (∼26% decreased; P = 0.1389 by nonparametric t-test) compared with cocultures plated at the higher cortical ratio (Fig. 4, A and Bi). The cumulative probability analysis of the event frequency showed a shift toward lower frequency distribution in cocultures plated at the lower cortical ratio (Fig. 4Bii; significant interaction effect by repeated-measures 2-way ANOVA). Although the mean mEPSC amplitude was almost identical (Fig. 4Ci), a cumulative probability plot suggested a difference in distribution of amplitudes with slightly more smaller amplitude events for the 1:3 cocultures (Fig. 4Cii; significant interaction effect by repeated-measures 2-way ANOVA). Together, these data suggest that a change in the cortical-striatal ratio has little effect on SPN synaptic AMPA receptor current but does impact excitatory synapse number in cocultures.

Fig. 4.

The proportion of cortical neurons to striatal SPNs in coculture affects SPN miniature excitatory postsynaptic current (mEPSC) frequency. A: typical current traces of mEPSCs at DIV18 show a decreased number of events in SPNs from coculture plated at a 1:3 ratio. Bi: mean mEPSC frequency trends lower in SPNs in 1:3 than in 1:1 cocultures [n = 39(6) for 1:1 and n = 45(7) for 1:3 ratio; nonparametric t-test, P = 0.1389]. Bii: cumulative probability of interevent intervals shows a rightward shift, toward longer interevent intervals for SPNs in 1:3 coculture [n = 39(6) for 1:1 and n = 45(7) for 1:3 ratio; repeated-measures 2-way ANOVA: interaction, ****P < 0.0001; cortical-striatal ratio, P > 0.05]. Ci: mean amplitude of striatal mEPSCs is similar in the 2 different cocultures [n = 39(6) for 1:1 and n = 45(7) for 1:3 ratio; unpaired t-test, P = 0.9347], but its cumulative probability plot (Cii) shows a small difference at higher mEPSC amplitudes between the 2 coculture conditions [n = 39(6) for 1:1 and n = 45(7) for 1:3 ratio; repeated-measures 2-way ANOVA: interaction, ****P < 0.0001; ratio, P > 0.05].

Level of cortical abundance influences the NMDA receptor distribution on SPN cell membrane.

NMDA receptor distribution and activity at synaptic and extrasynaptic sites have been shown to determine neuronal survival vs. cell death signaling (Hardingham and Bading 2010). Since, in our study, the decrease of cortical neuronal abundance relative to striatal neurons in cocultures downregulated the density of glutamatergic synapses on striatal SPNs (Fig. 3Bii), we were interested to know whether this reduction would shift the balance of synaptic and extrasynaptic NMDA receptors in striatal neurons. To address this question, we measured the total whole cell NMDA receptor current and compared this to the extrasynaptic NMDA receptor current isolated by specifically blocking synaptic receptors with MK801 within each cell (Fig. 5Ai). Our results showed that the charge transfer density (Fig. 5, Aii and Bi) and peak current density (Fig. 5Ci) of striatal SPNs were similar for total whole cell NMDA receptor current in both 1:1 and 1:3 coculture conditions; however, the proportion of current carried by extrasynaptic NMDARs was significantly larger in the 1:3 cocultures (Fig. 5, Bii and Cii), suggesting that increased cortical input upregulates synaptic and downregulates extrasynaptic NMDA receptors. Altogether, these data demonstrate that a change in the amount of cortical neurons relative to SPNs does not impact the overall density of NMDA receptors on the cell membrane, but it does influence their surface membrane distribution.

Fig. 5.

Reduced abundance of cortical neurons enhances proportion of extrasynaptic NMDA receptors in cocultured SPNs. Ai and Aii: representative current traces before and after MK801 application to block synaptic NMDA receptors (NMDARs), and in response to NMDA application, respectively. Black traces represent total (synaptic and extrasynaptic) currents, whereas light gray traces stand for currents through extrasynaptic NMDARs (Aii). The charge transfer density of the total NMDAR current (Bi) is similar [n = 10(4) for 1:1 and n = 9(3) for 1:3 ratio; P = 0.3206], but the proportion of current carried by extrasynaptic NMDARs (Bii) is significantly higher in cocultures with a 1:3 ratio [n = 10(4) for 1:1 and n = 9(3) for 1:3 ratio; ***P < 0.0001]. Moreover, the whole cell NMDAR peak current density (Ci) is similar in both coculture conditions [n = 10(4) for 1:1 and n = 9(3) for 1:3 ratio; P = 0.1221], whereas the proportion of peak current mediated by extrasynaptic NMDARs (Cii) is higher in 1:3 than in 1:1 cocultures [n = 10(4) for 1:1 and n = 9(3) for 1:3 ratio; ***P < 0.0001]. Unpaired t-tests were conducted in all statistical analyses.

Increased cortical-striatal ratio upregulates SPN basal level pro-survival signaling and resistance to apoptosis.

Because our study suggested that cortical neuronal abundance impacts NMDA receptor membrane distribution on SPNs, a factor that determines the neuronal survival-death signaling balance, we wondered whether altering the proportion of cortical to striatal neurons would change SPN vulnerability to a harmful stimulus. We measured the basal level of pro-survival signaling as reflected by activation of the master pro-survival transcriptional regulator, nuclear pCREB, by assessing the ratio of nuclear to cytoplasmic pCREB (pCREB ratio; Buren et al. 2014; Kaufman et al. 2012; Fig. 6Ai). The pCREB ratio measured in SPNs was significantly lower in cocultures plated at the lower cortical ratio (Fig. 6Aii), suggesting that the reduction in cortical-striatal neuronal ratio decreases the basal level of pro-survival signaling through pCREB. We also exposed cocultures plated at 1:1 and 1:3 ratios to different concentrations of NMDA in conditioned medium at DIV17-19 for 15 min and waited for 1 h to let the cells respond (Fig. 6Bi). Our result showed that the apoptotic cell percentage was significantly higher in cocultures plated at the lower cortical ratio (Fig. 6Bii), indicating that the reduction of cortical neurons renders striatal SPNs more vulnerable to an apoptotic stimulus. Together, these data support our conclusion that the proportion of cortical neurons determines the cell survival capacity of striatal SPNs in vitro.

DISCUSSION

Cocultures composed of neurons that originate from distinct brain regions that are synaptically connected in vivo have become a popular platform to study these connections in a well-controlled in vitro system. Although most previous work has combined cortical and striatal neurons in equal numbers to generate corticostriatal connections in culture, we hypothesized that a lower proportion of cortical neurons could more closely recapitulate conditions in the striatum in vivo, which receives robust glutamatergic cortical (and thalamic) afferents but is composed of >90% GABAergic medium-sized spiny projection neurons (SPNs). In this study, we compared corticostriatal cocultures with two different concentrations of cortical neurons to understand how the ratio of cortical to striatal neurons impacts striatal SPN physiology, dendritic morphology, membrane receptor distribution, and vulnerability to excitotoxic stress. Our results indicate that manipulation of cortical-striatal neuron ratio does not substantially alter many of the striatal neuronal basic membrane properties. However, striatal SPNs in coculture with a lower cortical neuron proportion (plating ratio of 1:3) showed reduced membrane capacitance and dendritic arborization, decreased numbers of excitatory synapses, and enhanced vulnerability to an excitotoxic challenge associated with reduced basal pCREB and an increased proportion of extrasynaptic NMDA receptors.

It is interesting that despite the reduction in cortical neurons and excitatory synapses on striatal SPNs in cocultures plated at the 1:3 ratio, the SPN input resistance, resting membrane potential, and excitability were not substantially changed. On the other hand, SPN resting potential showed a strong trend toward being more hyperpolarized in the 1:3 cocultures. One possible reason for this difference may be in the contribution of leak and/or inwardly rectifying potassium channels. Moreover, in our coculture system, the resting membrane potential was considerably depolarized (approximately −51 to −55 mV) compared with the dominant “downstate” resting membrane potential in vivo (approximately −86 mV; Mahon et al. 2006), and striatal SPNs showed an apparent lack of inwardly rectifying potassium current (Fig. 1Di), which would be consistent with the absence of dopaminergic input in our cocultures, as suggested by a previous study (Cazorla et al. 2012). Further experiments would help clarify these potential differences. Still, the resting potential of SPNs in coculture with the lower ratio of cortical neurons was relatively closer to that of the in vivo condition, as previously reported (Lalchandani et al. 2013).

We found that the reduction in the density of excitatory synapses along striatal SPN dendrites, as reflected by the colocalization of PSD-95 and vGlut1, was only ∼24% for the cocultures plated at the lower cortical ratio, whereas the number of cortical neurons per striatal neuron plated was decreased by ∼67%. However, as shown in Table 1, further examination revealed that the density of MAP2-positive cells that were DARPP-32 negative (putative cortical neurons) in our cocultures was only ∼25% lower at the 1:3 vs. 1:1 plating ratio, whereas the density of DARPP-32-positive cells (striatal SPNs) was nearly identical. These results suggest that over the course of development in vitro, cortical and striatal cells in our coculture experience a compensatory alteration in rate of differentiation and/or apoptosis, which acts to normalize the relative abundance of cells with different origins. Thus the modest reduction in functional (∼26% for mean mEPSC frequency; not significant) and morphological measures of excitatory synapses of striatal SPNs in the cocultures plated at the lower cortical ratio is roughly commensurate with the modest difference in cortical-striatal SPN cell density. That said, we still cannot exclude the possibility of alteration in cortical glutamatergic presynaptic terminals, such as a compensatory enhancement in synaptic vesicle release probability in 1:3 cocultures, since the combination of reduced SPN dendritic length and synaptic density by immunostaining would predict a more robust, and significant, reduction in mEPSC frequency than we observed.

Some features of striatal SPNs in our cocultures differ from those found in ex vivo brain slices or other coculture conditions. Recordings made from SPNs in acute striatal slice from 2-mo-old mice show mEPSC frequency of ∼2.5–3 Hz (Indersmitten et al. 2015; Kolodziejczyk et al. 2014), which includes excitatory projections from thalamus as well, whereas SPNs in our cocultures show mean mEPSC frequencies of ∼8.3 and ∼11.3 Hz for 1:3 and 1:1 plating ratios, respectively. Still, the lower cortical ratio coculture once again more closely matches the ex vivo condition than the 1:1 coculture.

Moreover, the density of spines on striatal SPNs that we observed in both sets of cocultures is ∼50% of that reported in a previous study (Tian et al. 2010). However, Tian et al. (2010) started with postnatal day 1 and 2 (P1-2) striatal tissue, which was then cocultured with E17-18 cortical neurons; in this study, we started with E18 cortical and striatal tissue. After 18 days in vitro, the striatal SPNs in our cocultures may not have achieved the same level of maturity, which could contribute to the reduced spine density. The reduced absolute cell density at the time of plating in our study compared with that of the previous study could also be a factor.

In this study, we also found that although the overall density of surface NMDA receptors, as reflected by the whole cell NMDA-evoked current normalized to cell capacitance, was similar for SPNs from the two coculture conditions, the proportion of extrasynaptic NMDA receptors was increased, suggesting a corresponding reduction in synaptic NMDA receptors, in the 1:3 cocultures. According to a variety of studies (Hardingham and Bading 2010), such a change in NMDA receptor distribution would be expected to alter cell survival/death signaling and shift SPNs toward enhanced susceptibility to harmful stimuli. Consistent with this idea, we found that basal pro-survival signaling through pCREB was reduced by ∼16%, whereas sensitivity to NMDA-induced apoptotic cell death was increased by ∼22–32% in SPNs from 1:3 cocultures. Thus, as predicted by the prevailing model (Hardingham and Bading 2010), in our coculture system the change in NMDA receptor distribution is closely correlated with dephosphorylation of nuclear CREB and weakening of neuronal health.

Together, our results demonstrate that culture conditions, in this case the plating ratio of two different neuronal types, influence key characteristics of mature cultured neurons. In our studies, cortical neuronal abundance influenced striatal SPN membrane area, dendritic arborization, excitatory synapse numbers, NMDA receptor distribution, pro-survival signaling, and susceptibility to stressful stimuli. As well, the lower cortical ratio resulted in striatal SPN resting membrane potential and mEPSC frequency that more closely matched recordings from striatal SPNs in acute brain slice or in vivo. Moreover, our data highlight the fact that changes in neuronal type proportion can affect experimental outcome measures, especially NMDA receptor distribution and vulnerability to excitotoxicity.

GRANTS

This work was supported by Canadian Institutes of Health Research Grant MOP-102517. C. Buren holds a University of British Columbia Four-Year Fellowship. M. P. Parsons held fellowship awards from the Michael Smith Foundation for Health Research and Canadian Institutes of Health Research.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.B., M.P.P., and L.A.R. conception and design of research; C.B. and G.T. performed experiments; C.B. and G.T. analyzed data; C.B., M.P.P., M.D.S., and L.A.R. interpreted results of experiments; C.B. and G.T. prepared figures; C.B. drafted manuscript; C.B., G.T., M.P.P., M.D.S., and L.A.R. edited and revised manuscript; C.B., G.T., M.P.P., M.D.S., and L.A.R. approved final version of manuscript.