Introduction

The lentivirus human immunodeficiency virus type 1 (HIV-1) enters the central nervous system early in the course of the infection and productively infects brain-resident macrophages, but not neurons. HIV-1 infection in cerebral cortex and basal ganglia may indirectly lead to neuronal apoptosis (Adle-Biassette et al. 1995; Gelbard et al. 1995; Petito and Roberts 1995). Previous work from our laboratory has established that the phospholipid mediator, platelet-activating factor (PAF), an evanescent molecular species normally synthesized in postsynaptic neuronal membranes to function as a retrograde messenger at glutamatergic nerve terminals, is also released in greatly increased concentrations from HIV-1-infected mononuclear phagocytes (Wieraszko et al. 1993; Gelbard et al. 1994). Under these circumstances, PAF receptor activation in turn leads to NMDA receptor-dependent and independent neuronal apoptosis (Gelbard et al. 1994; Perry et al. 1998).

Interestingly, neither viral load nor postmortem quantitation of neuronal apoptosis correlates with the severity of premortem neurologic disease (Fox et al. 2000; Gray et al. 2000). These observations, and the ability of potent antiretroviral therapy to temporarily ameliorate the symptoms of HIV-1-associated dementia (HAD), have led us and others to speculate that a significant component of HAD may be due to changes in synaptic transmission mediated by proinflammatory neurotoxic factors produced by brain-resident mononuclear phagocytes. Decreased synaptic facilitation and reduced expression of neuronal cytoskeletal and synaptic proteins have been demonstrated in a severe combined immunodeficiency murine model of HIV-1 encephalitis (Zink et al. 2002; Dou et al. 2005). To gain further insight into how pathologic production of PAF might disrupt synaptic transmission in the basal ganglia, we employed a sagittal corticostriatal brain slice model (Spencer and Murphy 2000). We now demonstrate that acute application of a single, sublethal dose of PAF results in increased synaptic facilitation in the caudate-putamen following high-frequency stimulation (HFS) by a PAF receptor-mediated mechanism. However, application of PAF, at a concentration known to induce pathologic changes in synaptic architecture with loss of synaptic transmission (Bellizzi et al. 2005), during HFS abolished the ability of corticostriatal slices to achieve increased synaptic facilitation. These findings suggest that PAF may be a normal modulator of glutamatergic synapses in corticostriatal pathways but, during HIV-1 infection, may disrupt activity-dependent glutamatergic neurotransmission.

Materials and methods

Animals

Ten- to 14-week-old (25–35 g) male C57Bl/6J mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA). For all experiments, brain tissue was harvested according to the guidelines of the Animal Welfare Act (1987) and NIH policies. Briefly, mice were killed by cervical dislocation and decapitation. A total of 41 slices cut from the brains of 12 mice were included in the analysis of the present study.

Slice preparation

The brains were rapidly removed and cooled in sucrose-based artificial cerebrospinal fluid (ACSF) containing, in millimolars, 110 sucrose, 60 NaCl, 3 KCl, 1.25 NaH2PO4, 28 NaHCO3, 10 d-glucose, 0.5 CaCl2, 7 MgSO4, and 0.6 ascorbate. The brains were blocked (to obtain a sagittal slice, Fig. 1a) and fixed to a specimen stage with cyanoacrylate. Sagittal slices (400 μm in thickness), containing the medial-striatum and cortex, were cut with a vibroslicer (World Precision Instruments, Sarasota, FL, USA) equipped with a Peltier cooling system (see Fig. 1a). All slices were collected to a holding chamber maintained at room temperature. They were allowed to recover for at least an hour before a slice was transferred to an interface recording chamber (Harvard Apparatus, Holliston, MA, USA). The slice was equilibrated for 30 min prior to recording and maintained at 35°C and perfused with oxygenated ACSF (95% O2, 5% CO2) in millimolars, 125 NaCl, 5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 25 d-glucose, 2 CaCl2, 1 MgSO4, and 0.01 glycine, at a flow rate of 1.0–1.5 ml/min. The remaining slices were kept in the holding chamber until they were needed later.

Fig. 1
figure 1

A Photomicrograph of one of the slices used in experiments described in Figs. 2 and 3. Here, the stimulating and recording electrode sites are labeled in the 1.25× brightfield upper image, with a 20× inset photomicrograph that shows DiI-filled axon fibers extending from the recording site toward the stimulating site. A small DiI crystal jammed in a broken micropipette with a 20-μm tip was deposited at the recording site after the recording electrode was removed at the end of the recording session. B Example of FPs recorded in striatum in response to the electrical stimulation of white matter. The red cursors a and b are used to set the analysis window in Clampfit where the negative peak amplitude of FPs were measured using this software.

Full size image

Field potential recording

Extracellular recording electrodes, pulled on a horizontal micropipette puller (Sutter Instruments, Novato, CA, USA) and filled with 2 M NaCl and 2% pontamine blue (electrode impedance 2–4 MΩ), were placed within the striatum, located 1–2 mm from the white matter (see Fig. 1a). Field potentials (FPs), caused by action potentials (APs) and postsynaptic potentials (PSPs) response generated in the postsynaptic neurons to stimulation, were evoked by a constant current pulse stimulus (100–500 μA, 100 μs) applied to the white matter via a custom-made bipolar platinum-wire stimulating electrode placed rostro-dorsally with respect to the recording electrode. Field potentials were recorded using an Axoclamp-2B amplifier (Axon Instruments, Union City, CA, USA) and amplified further by an EX1 differential amplifier (Dagan, Minneapolis, MN, USA). Data acquisition and analyses were done by pClamp 8 software (Axon Instruments) on a Pentium III PC computer (Dell, Indianapolis, IN, USA).

Corticostriatal plasticity

Extracellular field recordings were used to assess synaptic plasticity in the corticostriatal pathway. A quick input and output function was determined by increasing the intensity of test stimuli in exponential steps from 50 to 3,200 μA (or until the peak amplitude of the negative FP reached the maximum), and the test stimuli were delivered every 30 s. The peak amplitude of the FPs was measured online by the manual placement of cursors. The intensity of stimuli that evoked FPs to one half of the maximal peak amplitude (100–500 μA, 100 μs duration) was used to establish the baseline responses before the induction of synaptic facilitation. Stable responses were recorded for 10–20 min prior to tetanic conditioning. Tetanic conditioning consisted of one train of HFS applied to the white matter via the stimulating electrode at test intensity (each train consisted of 100 stimuli at 100 Hz). Field potentials were monitored for 45–60 min after conditioning (Fig. 2b). The percentage change in the peak amplitude of the FP was determined by normalizing the mean response measured 45–50 min posttetanus (unless otherwise stated) to the mean response measured over the 5-min period immediately prior to conditioning. The values of peak amplitude of the FPs were measured by using Clampfit software and the cursors were set to include the FP response after electrical stimulation (Fig. 1b). Synaptic facilitation was defined as a stable increase in FP amplitude (>15%) 40 min posttetanus. Synaptic depression was defined as a stable decrease in FP amplitude (>15%) 40 min posttetanus. Short-term changes such as posttetanic potentiation or posttetanic depression were responses that returned to baseline values by 40 min posttetanus. Pooled data are presented as the mean normalized value of the FP ± SEM. For all data, the Student t test was used to compare the difference between mean values.

Fig. 2
figure 2

a Examples of FP traces recorded from individual slices treated either with control vehicle, cPAF 130 nM, or cPAF 1,300 nM. In each of the treatment conditions, green, blue, and red traces were recorded at 2 min before, 2 min after, and 56 min after the HFS, respectively. b Data were collected at 1-min intervals and averaged from 17 slices for control vehicle,17 slices for 130 nM cPAF, and 7 slices for 130 nM cPAF + 10 μM BN52021.

Full size image

Reagents

Carbamyl PAF (cPAF), a nonhydrolyzable analog of PAF that is approximately one half to one tenth as potent as PAF (Gelbard et al. 1994; O’Flaherty et al. 1987), and the PAF receptor antagonist gingkolide B (BN52021) were obtained from Biomol (Plymouth Meeting, PA, USA). Drugs were first dissolved in 95% ethanol as stock solution and were diluted to the effective concentration in ACSF, and bath-applied via the perfusion lines. It took about 10 min at a flow rate of 1 ml/min for reagents to reach the recording chamber.

Results

Platelet-activating factor receptor-mediated synaptic facilitation

Measuring FPs generated by corticostriatal circuitry in an experimental model with the appropriate afferent and efferent connections is technically challenging, but a number of studies have characterized the parameters of synaptic facilitation and synaptic depression and their corresponding neurochemical substrates at corticostriatal synapses using acutely prepared coronal and sagittal slices (Spencer and Murphy 2000; Calabresi et al. 2000; Partridge et al. 2000; Pisani et al. 2000; Tang and Lovinger 2000; Kerr and Wickens 2001; Smith et al. 2001). Because of issues concerning pharmacologic manipulation of dopamine receptor subtypes to modulate synaptic efficacy in a number of these studies, we chose a sagittal corticostriatal slice model described by Spencer and Murphy (2000) to investigate the effects of PAF on PSPs and APs that comprise the FPs that we can record. In this model, pharmacologic treatment is not required to achieve synaptic facilitation following HFS.

An additional level of complexity in this model system is added because the evoked potential cannot be interpreted in terms of an afferent volley of stimulation that results in the excitatory PSP and AP. In contrast to the hippocampus, the striatum has a diffuse distribution of cells (i.e., medium spiny neurons) that receive mono- and polysynaptic inputs that are not organized into discrete afferent tracts that we can stimulate. Figure 2a demonstrates examples of extracellular recordings of FPs obtained from individual striatal slices exposed to control vehicle (contained 2 mM ethanol, which is also present in both cPAF solutions), cPAF 130 nM, or cPAF 1,300 nM before (green trace), 2 min after (blue trace) and 56 min after (red trace) HFS. Immediately following the HFS, the peak amplitude of FPs could become greater, smaller, or remain the same compared to before induction of HFS, but eventually would become greater 20 min after HFS, continue to increase, and ultimately stabilize around 30 min later in most of the slices regardless of treatment conditions. Importantly, in control vehicle-treated slices without tetanic conditioning, no synaptic potentiation developed at any point during the course of the recording period (≥60 min). To assess the effects of cPAF [the carbamyl congener is used in all experiments, otherwise PAF is rapidly hydrolyzed by tissue acetylhydrolases (O’Flaherty et al. 1987), on synaptic transmission in sagittal corticostriatal slices, we used a sublethal concentration of cPAF (130 nM) applied to sagittal corticostriatal slices, based on dose–response curves of PAF-mediated apoptosis (Gelbard et al. 1994; Perry et al. 1998; Tong et al. 2001) in in vitro neuronal cell culture models from multiple brain regions. Thus, extracellular field recordings demonstrate that cPAF increases the peak amplitude of the FP after HFS for a sustained period of time relative to control vehicle (Fig. 2b). Data in Figs. 2b and 3a, b, depicting quantitative changes in the mean FPs recorded from control vehicle and various treatment conditions, were calculated from our raw data by dividing the data sets at a given time point and converting to percent change. The increased mean FP in cPAF-treated slices relative to control vehicle-treated slices rises from 5% at 20 min after the initiation of HFS to a maximum of 39% (p < 0.005) at 51 min (Fig. 2b). Interestingly, the peak amplitude of the FP increased after HFS in both control and cPAF-treated corticostriatal slices, but the specificity of the striatal response to PAF was established by coapplication of the gingkolide B PAF receptor antagonist BN52021 (Fig. 2b). Here, mean FP values after HFS were nearly identical to control vehicle-treated values.

Fig. 3
figure 3

a Data were collected as described for Fig. 2b and represent the mean FP amplitude ± SEM for 1,300 nM cPAF from eight slices. For purposes of comparison, FPs obtained from slices exposed to 1,300 nM cPAF are plotted against FPs obtained from slices exposed to 130 nM cPAF (obtained from data depicted in Fig. 2b). For clarity, control values are not plotted in this panel but are the same as data depicted in Fig. 2b. b Data were collected as described in Fig. 2b and averaged from seven slices exposed to 20 min of control vehicle and subsequently exposed to 20 min of 130 nM cPAF or nine slices exposed to 20 min of 130 nM cPAF and subsequently exposed to an additional 20 min of 130 nM cPAF. c The number of slices for each treatment condition is displayed as bar graph representations that show HFS-induced facilitation, depression, and no change of FP.

Full size image

Under these conditions, tetanic stimulation resulted in synaptic facilitation in 14/17 slices, synaptic depression in 0/17 slices, and no change in 3/17 slices (Fig. 3c). Application of 130 nM cPAF resulted in synaptic facilitation in 17/17 slices and synaptic depression or no change in 0/17 slices (Fig. 3c). Application of 130 nM cPAF + the PAF receptor antagonist BN52021 (10 μM) resulted in synaptic facilitation in 6/7 slices, synaptic depression in 0/7 slices, and no change in 1/7 slices (Fig. 3c). While we do not have a mechanistic or methodological explanation for the increase in baseline values of the normalized FP during our recordings, the PAF receptor-specific changes relative to our control data strongly support the nonartifactual nature of the observed increases in striatal synaptic facilitation following treatment with cPAF. Furthermore, we have previously demonstrated that PAF receptor message is present in striatal cell bodies and PAF receptor protein is also expressed in striatal neurons (Tong and Gelbard, unpublished data).

Having established that treatment with 130 nM cPAF augmented synaptic facilitation in response to HFS, we investigated the effects of a concentration of cPAF (1,300 nM) known to induce pathologic changes in synaptic architecture with loss of activity-dependent neurotransmission (Bellizzi et al. 2005) on the induction and maintenance of synaptic facilitation in corticostriatal slices. Here, application of 1,300 nM cPAF initially produced a sharp decrement in mean FP to 58% of control values (Fig. 3a; note for purposes of comparison, only values for low- and high-dose cPAF are plotted, the control values are plotted in Fig. 2b) 1 min after HFS. A slow rebound of the mean FP was observed 10 min after HFS such that it remained facilitated to values comparable to those measured in slices treated with control vehicle until the end of recording (60 min after HFS).

Repeated PAF receptor agonism synergizes with tetanic stimulation

We subsequently investigated whether repeated application of 130 nM cPAF to slices would have an effect on the ability of corticostriatal synapses to achieve synaptic facilitation after a second tetanic conditioning. As a control, one group of corticostriatal slices was initially treated with vehicle and HFS for a 30-min period and a treatment-free period for the next 45 min, followed by 30 min of cPAF and HFS (Fig. 3b). In contrast, the second group of slices received 30 min of cPAF and HFS, a treatment-free period for the next 45 min, and were treated with a second 30-min period of cPAF and HFS. Compared to slices that received only one application of cPAF, slices that received two consecutive applications of 130 nM cPAF had a 56% increase in FP 70 min after the first volley of HFS and a 70% increase in FP 50 min after the second volley of HFS. Interestingly, a similar sharp decrement in FP (that was observed for treatment of slices with 1,300 nM cPAF + HFS) was also seen after the second application of cPAF + HFS. Here, values for the FP were 155% of slices that received control vehicle followed by cPAF and dropped to 113% of control immediately after the second tetanic stimulation. However, in these slices, the percent of FP rebounded significantly such that maximal values were >300% of the original baseline (Fig. 3b) after the second application of cPAF and HFS in contrast to slices that only received a single application of cPAF, where maximal FP values were <200% (Fig. 3b).

Discussion

The central premise behind these experiments is the idea that we could develop a model of intact corticostriatal circuitry with well-defined, reproducible electrophysiological parameters to better understand the effects of HIV-1-induced neurotoxins on vulnerable neurons prior to their demise by apoptosis. While many other HIV-1 virotoxins (Tat and gp120) and cellular metabolites such as tumor necrosis factor alpha and nitric oxide can directly or indirectly alter the excitability of neuronal membranes (Lipton et al. 1991; Dawson et al. 1993; Gelbard et al. 1993; Magnuson et al. 1993; Beattie et al. 2002), we chose the nonhydrolyzable analog of PAF (cPAF) as our experimental HIV-1-induced neurotoxin because we have previously shown that it plays a pivotal role in both NMDA receptor-dependent and independent neuronal apoptosis (Gelbard et al. 1994; Perry et al. 1998; Bennett et al. 1998) and because it has a role in synaptic facilitation and presynaptic modulation of glutamate release (Kornecki et al. 1996; Bazan 1998). In these studies, HFS induces postsynaptic activation of glutamate receptors and increased biosynthesis of PAF, with subsequent retrograde signaling to presynaptic glutamatergic terminals.

The major afferent projection from the cerebral cortex to striatal medium spiny neurons (corticostriatal pathway) is capable of activity-dependent changes in synaptic plasticity (Calabresi et al. 1992). This pathway regulates excitatory input to the basal ganglia and is thought to be involved in motor learning and certain forms of cognition (Graybiel 1995). Although the role(s) of synaptic plasticity in regulating striatal function remain unclear, both depression and facilitation of corticostriatal synapses can be readily demonstrated in organotypic cultures and striatal slices using physiologic conditions (Spencer and Murphy 2000; Kawaguchi et al. 1989; Plenz and Kitai 1998), as well as in vivo (Charpier and Deniau 1997).

As an initial attempt to understand what role PAF might play in modulation of corticostriatal synapses, we chose low and high doses of cPAF that we believed would result in synaptic potentiation and failure of activity-dependent synaptic transmission, respectively, based on our in vitro studies of PAF-mediated effects on neuronal apoptosis, neuronal migration, and synaptic architecture and neurotransmission (Perry et al. 1998; Tong et al. 2001; Bellizzi et al. 2005), and published data on PAF-mediated intracarotid effects on leukocyte-trafficking in an in vivo model of the blood–brain barrier (Uhl et al. 1999). Thus, in our model system of cerebellar granule neuronal cultures (Tong et al. 2001), a 130-nM dose of cPAF had no effect on neuronal apoptosis, and a 1,300-nM dose acutely induced loss of normal synaptic architecture and neurotransmission (Bellizzi et al. 2005) and later induced neuronal apoptosis. Additionally, doses ≥1 μM of PAF induced arterial hypotension and increased leukocyte adhesion to cerebral microvessels (Uhl et al. 1999), conditions that might predispose toward adverse outcomes in brain parenchyma supported by this microcirculation. We reasoned that these two concentrations of cPAF might approximate a physiologic and pathophysiologic level of PAF in our model system. Application of 130 nM cPAF and HFS clearly augmented synaptic facilitation in corticostriatal synapses (Fig. 2b) by a PAF-receptor-dependent mechanism because the increase in FP was abolished by coapplication of the PAF receptor antagonist, BN52021 (Fig. 2b). These findings are in agreement with studies of PAF receptor-mediated synaptic facilitation in rat medial vestibular nuclei (Pettorossi and Grassi 2001). In this model system, application of BN520201 also reduced the amplitude of HFS potentiation; and application of the nonhydrolyzable methylcarbamyl analog of PAF to rodent brainstem slices was associated with glutamate release, as evidenced by a reduction in the paired-pulse facilitation ratio. In the sagittal corticostriatal slice model system, the peak amplitude of FP after HFS could be the result of several possibilities that are not exclusive to each other: (1) an increase in glutamate release at corticostriatal synapses, (2) a surge of dopamine release during HFS that can potentiate postsynaptic response of striatal median spiny neurons (MSNs) to cortical inputs, or (3) lowering the threshold for AP of MSNs that could occur because HFS induced more MSNs to an up-state (an electrophysiologic term that refers to lower resting membrane potential of MSNs) or resulted in an increased release of dopamine from nerve terminals. Measuring glutamate release in our model would help resolve this issue but is beyond the scope of present study because current methodology lacks the sensitivity to determine glutamate concentrations in corticostriatal slices.

Interestingly, facilitation of corticostriatal synapses by HFS alone (i.e., control vehicle-treated striatal slices) occurred gradually (Fig. 2b). The reasons for this are presently unclear but may be dependent on stimulating electrode placement in the white matter, invariant stimulation strength during HFS conditioning, and variability in response between striatal slices taken from more rostral vs caudal origins along the striatal axis (Spencer and Murphy 2000; Charpier and Deniau 1997). An equally compelling possibility is that activity-dependent changes in striatal plasticity mediated by HFS in our experimental system result in APs that, in turn, alter the intrinsic electrical properties of medium spiny neurons, decreasing the threshold for subsequent excitatory PSPs (Mahon et al. 2003). This provides a mechanistic basis for the slow “ramping up” of the baseline normalized FP amplitude that we observed during control and cPAF-treated corticostriatal slices subjected to tetanic stimulation (Fig. 2b).

A model of chronic inflammation and corticostriatal plasticity

Having demonstrated that a low dose of cPAF can augment the FP generated by corticostriatal synapses subjected to HFS, we asked the question of whether application of a lethal concentration of cPAF could damage the functional capacity of the corticostriatal pathway to facilitate in response to a tetanic stimulus. Two salient findings emerge from this experiment: (1) a sharp decrement in the FP following application of 1,300 nM cPAF + HFS and (2) failure of 1,300 nM cPAF + HFS to increase the amplitude of the FP beyond that of control vehicle-treated slices (Fig. 3a). The observed decrement in FP may be explained by the fact that GABA is released in response to both baseline test pulses and HFS; thus, the initial depression of FPs after HFS may be due to GABA inhibition. Furthermore, two repeated 130-nM cPAF doses can synergize with a tetanic stimulus to facilitate corticostriatal synapses at a significantly faster rate than corticostriatal synapses exposed to control vehicle + HFS. Furthermore, the amplitude of FPs generated by repeated doses of 130 nM cPAF + HFS is significantly greater than corticostriatal synapses exposed to either control vehicle + HFS or a single dose of 130 nM cPAF + HFS (Fig. 3b). Our data suggest that the difference in FP between striatal slices exposed to control vehicle + HFS or two repeated doses of 130 nM cPAF + HFS may be an indirect measure of the plasticity/capacity of corticostriatal synapses to achieve facilitation by release of glutamate in this system. Unfortunately, the data presented in Fig. 3b cannot be extended beyond the present kinetic analyses because corticostriatal slices do not remain viable after 4 h in the recording chamber. However, the marked change in the rate and amplitude of the FP achieved under these conditions suggests that further exposure to low doses of PAF may ultimately result in a loss of synaptic plasticity with subsequent deficits in the either the motor or cognitive functions subserved by corticostriatal circuitry. The data presented in this report may be an important step in our understanding of the continuum between normal striatal plasticity and the pathology associated with HIV-1 infection of the basal ganglia.