Long-term depression (LTD) is one of the most widely investigated models of the synaptic mechanisms underlying learning and memory. Previous research has shown that induction of LTD in the neocortex decreases measures of pyramidal cell dendritic morphology in both layers III and V. Here, we investigated the effects of LTD induction on 1) the time course of recovery of synaptic efficacy, 2) movement representations, 3) cortical thickness and layer V neuron density, and 4) the density of excitatory and inhibitory synapses in layer V of sensorimotor neocortex. Rats carried a stimulating electrode in the midline corpus callosum and a recording electrode in the right sensorimotor neocortex. Each rat received either low-frequency stimulation composed of 900 pulses at 1 Hz or handling daily for a total of 20–25 days. Callosal–neocortical evoked potentials were recorded in the right hemisphere before and after stimulation or handling. Our results show that LTD induction lasts for 3 weeks and results in smaller motor maps of the caudal forelimb area. We did not observe any reduction in neocortical thickness or neuron density. There was a reduction in the density of excitatory perforated synapses and an increase in the density of inhibitory synapses in layer V of the sensorimotor neocortex, thereby providing a general mechanism for the reduction in motor map size. This study sheds light on the interaction between an artificial model of learning, receptive field characteristics, and synaptic number in the sensorimotor cortex.
Changes in the efficacy of synaptic transmission have attracted a great deal of attention because of their potential role in the development of neural circuitry as well as learning and memory (Abraham and Bear 1996; Chen and Tonegawa 1997). Long-term depression (LTD), defined as a persistent weakening of synaptic efficacy (Christie and others 1994; Bi and Poo 1998), is currently one of the most widely investigated models of the synaptic mechanisms underlying learning. Low-frequency stimulation (LFS) that leads to LTD (Froc and others 2000; Werk and others 2006) is also a potential tool to reduce the rate of spontaneous seizures in people with epilepsy (Goodman 2004) and is thought to work by reducing synaptic efficacy and interfering with dynamic network behavior (Khosravani and others 2003; Goodman and others 2005).
LFS has been shown to reliably induce LTD in the rat sensorimotor cortex both in vitro (Tsumoto 1992; Hess and Donoghue 1994) and in vivo (Froc and others 2000; Froc and Racine 2005), although a complete time course of recovery in the in vivo preparation has yet to be described. Because LTD can be reliably induced in the neocortex of awake, freely behaving rats (Froc and others 2000; Froc and Racine 2004, 2005; Monfils and Teskey 2004), we have an opportunity to investigate its effects on the organization of movement representations in the neocortex. Movement representations have been found to proportionately increase with increases in synaptic efficacy (Monfils and Teskey 2004). However, there have been no reports of changes in movement representations with reductions in synaptic efficacy. Thus, we tested the hypothesis that weakening synaptic efficacy would reduce the size of the caudal forelimb area that responds to intracortical microstimulation-induced movements. The caudal forelimb area was chosen for analysis because it has well-defined borders, the capacity for internal reorganization, and the shifting of boundaries (Nudo and others 1990; Kleim and others 1998; Flynn and others 2004; Monfils and others 2005).
Given that spaced and repeated LFS over days involves delivering a substantial amount of current, determining if there is stimulation-induced damage should be evaluated. This is particularly important for those who advocate its use in people with epilepsy (Goodman 2004). Thus, we also examined whether there was a reduction in cortical thickness and neuronal loss to layer V. A previous study using the Golgi-Cox method showed that in vivo LTD induction was associated with decreases in dendritic length, branching, and spine density in layer V (Monfils and Teskey 2004). However, an ultrastructural analysis of synaptic subtypes in layer V has yet to be done and could provide convergent anatomical evidence of a reduction in synaptic efficacy. We also examined excitatory (perforated and nonperforated) and inhibitory synaptic subtypes of pyramidal cells in sensorimotor layer V to evaluate if changes in these inputs may contribute to the reduction in synaptic efficacy. This study addresses the relation between synaptic weakening, receptive field characteristics, neuronal loss, and changes in synaptic subtypes in the rat sensorimotor neocortex.
Materials and Methods
A total of 48 male Long-Evans rats weighing between 350 and 425 g at the time of electrode implantation surgery were used. The rats were obtained from the University of Calgary Breeding Colonies and were housed individually in clear plastic cages in a colony room that was maintained on a 12-h on/12-h off light cycle with lights on at 7:00 AM. All experimentation was conducted during the light phase. Rats were maintained on Lab Diet no. 5001 (PMI Nutrition International, Brentwood, MO) and water ad libitum and were handled and cared for according to the Canadian Council on Animal Care guidelines.
Twisted-wire bipolar stimulating and recording electrodes were constructed and implanted according to the methodology previously reported (Teskey and others 2002). Rats were anesthetized with a combination of 50.0 mg/kg ketamine and 8.8 mg/kg xylazine at 1.0 ml/kg injected intramuscularly. Two percent lidocaine was administered subcutaneously at the incision site. Two bipolar electrodes were chronically implanted according to the stereotaxic coordinates of Swanson (1992). The stimulating electrode was implanted 1.0 mm anterior to bregma, on the midline, in the callosal matter. The recording electrode was implanted 1.0 mm anterior to bregma and 4.0 mm lateral to midline in the right frontal neocortex. Electrophysiological monitoring was performed during the implantation in order to adjust the dorsal–ventral placements to maximize evoked response amplitude. Histological analysis revealed that the stimulating electrode was located in the corpus callosum and the recording electrode was located in the right caudal forelimb area of sensorimotor cortex for each rat. The upper pole of the recording electrode was located in layer II/III, and the lower pole in deep layer V or upper layer VI.
The gold-plated amphenol pins connected to the electrodes were inserted into a 9-pin McIntyre connector plug and attached to the skull with 4 stainless steel screws and dental cement. One of the stainless steel screws served as the ground reference. Experimental procedures commenced 7 days after surgery.
Experimental and Treatment Groups
All rats were implanted with chronic indwelling stimulating and recording electrodes and were divided into 2 groups that received either LFS (N = 23) or handling (N = 25). In Experiment 1, rats that received 20 sessions of LFS (n = 6) or handling (n = 6) were followed for 5 weeks to determine the time course of recovery of evoked potential responses following induction of LTD. In Experiment 2, rats received 20–25 sessions of LFS (n = 11) or handling (n = 13) and then had their movement representations determined using intracortical microstimulation. A range of stimulation sessions was used so that all rats would have their motor maps derived 1 day after their last stimulation session. In Experiment 3, rats received 20 sessions of LFS (n = 6) or handling (n = 6) and were then sacrificed, their brains processed for cortical thickness, neuron counts from layer V, and quantification of synapse type (perforated, nonperforated, inhibitory) and density from electron micrographs.
Evoked Potentials and LFS
Baseline input–output measures were conducted for 2 consecutive days prior to stimulation or handling. The input–output measures were obtained by delivering pulses of increasing intensity to the callosal white matter and recording the resultant evoked potentials from the frontal neocortical electrode. The stimulation pulses consisted of biphasic rectangular waves, 200 μs in duration and separated by 200 μs. Ten stimulation pulses were delivered at each of the 10 ascending logarithmic intensities (32, 46, 68, 100, 147, 215, 316, 464, 681, and 1000 μA) at a frequency of 0.1 Hz. This type of recording was necessary to ensure that the evoked responses were stable and to identify whether the stimulation threshold to induce a response was constant pre- and post-LFS. Stimulation voltages were computer generated and converted to amperage via a constant current and isolation unit (A-M Systems, Carlsborg, WA). The recorded signals were filtered at half amplitude, below 0.5 Hz and above 100 Hz, and then amplified 1000 or 2000 times (Grass Neurodata Acquisition System Model 12). The analog signals were digitized at a sampling rate of 5 points/ms, and the averaged evoked potential, at each intensity, was stored to a computer hard disk (Datawave, Longmont, CO). All input–output measures were conducted while the rats were awake but behaviorally quiescent. This was necessary because ongoing behavior can dramatically affect the size and shape of the evoked potentials (Racine and others 1975; Vanderwolf and others 1987; Teskey and Valentine 1998).
LFS protocols previously found to reliably induce chronic LTD in the neocortex of awake, behaving rats were used in the present study (Froc and others 2000; Monfils and Teskey 2004; Werk and others 2006). LTD was induced with single biphasic square-wave pulses, with pulse widths that were 200 μs in duration at a pulse intensity of 1000 μA base to peak and a frequency of 1 Hz. This LFS was applied once daily for 15 min (900 pulses). Follow-up evoked potentials were obtained 24 h after the last LFS or handling session. Electrographic activity and the rats' behavior were monitored during application of the LFS and for 1 min thereafter. There was no evidence of epileptiform activity during or shortly after the LFS. During the LFS, the rats occasionally displayed forelimb movements driven by the individual pulses.
Evoked potentials obtained pre- and poststimulation to the callosum (or handling) were measured for change in the monosynaptic component. The monosynaptic component peaks at latencies of 9–11 ms (Chapman and others 1998; Froc and others 2000), therefore the peak amplitude (mV) was determined between those latency ranges for each rat. The polysynaptic component was not analyzed in this study because in response to LFS, it has been previously found to decrease in a proportional manner with the monosynaptic component (Monfils and Teskey 2004) yielding little incremental information. All potentials were examined but only those evoked at 681 μA were statistically analyzed.
Experiment 1—Time Course
Two days following the 20th LFS session, rats had their input–output evoked responses measured. Thereafter, evoked potentials were measured at weekly intervals on days 9, 16, 23, 30, and 37. Twenty-four hours following the last input–output measurement, rats were euthanized with intracardial injection of 0.35 ml of 240 mg/ml sodium pentobarbital.
Experiment 2—Movement Representations (Motor Maps)
One day following the last stimulation or handling session, standard intracortical microstimulation techniques (Stoney and others 1968; Nudo and others 1990; Kleim and others 1998) were used to generate detailed maps of the sensorimotor neocortical forelimb regions.
Twenty-four hours prior to surgery, rats were food deprived but had free access to water. Rats initially received intraperitoneal injections of ketamine (100 mg/kg) and xylazine (5 mg/kg). Two percent lidocaine was administered subcutaneously at the incision site. Supplemental injections of ketamine alone (25 mg/kg) or a cocktail of both ketamine (17 mg/kg) and xylazine (2 mg/kg) were delivered intraperitoneally as required throughout the surgical procedure to maintain a constant level of anesthesia, as indicated by breathing rate, vibrissae whisking, and a foot reflex in response to a gentle pinch.
A 7 × 5–mm craniotomy, contralateral to the electrode assembly, exposed the sensorimotor neocortex of the left hemisphere. The window extended 4 mm anterior and 3 mm posterior from bregma and from midline to 5 mm lateral of midline. Using an 18-gauge needle, a small puncture was made in the cisterna magna to reduce pressure due to edema. Dura was carefully removed, and 37.4 °C silicone fluid (Factor II, Inc., Lakeside, AZ) was used to cover the cortical surface. A ×32 digital image of the exposed portion of the brain was captured using a Stemi 2000-C stereomicroscope (Carl Zeiss, Thornwood, NY), digital camera (Cohu, San Diego, CA), and Image J software (U.S. National Institutes of Health, Bethesda, MD) and displayed on a computer. A grid composed of 500 × 500 μm squares was overlaid on the digital image. Penetration points were chosen at the intersections of the grid lines and at a central point in the middle of each square (interpenetration distance of 353 μm), except when located over a blood vessel. Microelectrodes made from borosilicate glass capillaries (World Precision Instruments, Sarasota, FL) on a micropipette puller (Kopf, Tujunga, CA) were filled with 3.5 M NaCl and beveled at a 30° angle to yield a 3-μm tip diameter with impedance values between 1.0 and 1.5 MΩ. Penetrations of the neocortex by the glass electrode were guided by a microdrive (Narishige, Tokyo, Japan) to a depth of 1550 μm from the cortical surface, corresponding to the cell body region of neocortical layer V. Electrical stimulation was delivered via an isolated stimulator and consisted of 13 monophasic cathodal pulses, each 200 μs in duration, delivered at a frequency of 333 Hz, and repeated every second. Rats were maintained in a prone position, with the limb contralateral to the stimulation side being supported by placing one finger below the elbow joint and elevating the forelimb. This allowed for visual detection of all possible forelimb (digit, wrist, elbow, or shoulder) movements. At each site, the minimal threshold required to elicit a movement was recorded, and a color-coded dot representing the movement was placed on the digital image. Penetration sites that failed to elicit a movement at any current intensity, up to a maximum of 60 μA, were defined as nonresponsive. To determine movement threshold, current intensity began at 0 μA, was rapidly increased until a movement was detected and then decreased until the movement was no longer present. No more than 10 trains of pulses were delivered to a single site to determine movement. The border of the forelimb motor map was defined first and was characterized by any nonforelimb movement that included head and neck, jaw, vibrissa, or nonresponsive sites. The more central map points were then determined in an effort to reduce the likelihood of the intracortical microstimulation affecting the border points of the map (Nudo and others 1990). The level of anesthesia was also assessed by revisiting positive response sites to check for changes in movement thresholds as mapping progressed. Mapping sessions lasted between 102 and 216 min after which the rats were euthanized as described above.
Movement Representation Analysis
Canvas (version 9.0.1) imaging software (ACD Systems Inc., Miami, FL) was used to calculate the aerial extent of the caudal forelimb area. The caudal forelimb area is separated from the rostral forelimb area by a band of nonresponsive or nonforelimb sites (Kleim and others 1998) and was chosen for analysis because previous work demonstrated its capacity to undergo reorganization following behavioral manipulations (Kleim and others 1998; Remple and others 2001) or stimulation (Nudo and others 1990; Teskey and others 2002; Monfils and others 2004, 2005). The proportion of distal and proximal movements that occupied the caudal forelimb area was then calculated. The mean stimulation threshold for each movement category (proximal vs. distal) was also calculated.
Experiment 3—Cortical Thickness, Neuron Density, Synapse Type, and Quantification
Twenty-one days following the last stimulation or handling session, at a time when the stimulation group showed LTD, rats were deeply anaesthetized with sodium pentobarbital (120 mg/kg) and transcardially perfused with 2% paraformaldehyde/2.5% glutaraldehyde in 0.1-M phosphate-buffered saline.
The left hemisphere was sectioned (300 μm) in the coronal plane on a vibratome and viewed under a light microscope at low power (×2). Three coronal samples of tissue extending from pia to white matter were taken from motor cortex homotopic to the recording electrode, in the left hemisphere. Blocks were taken between 0.5 and 1.0 mm anterior to bregma and included tissue 2.5–4.0 mm lateral to midline. These tissue blocks were washed in 0.1-M cacodylate, postfixed in 2% osmium tetroxide/1.5% potassium ferrocyanide in 0.1-M cacodylate buffer for 2 h, and en bloc stained with 2% uranyl acetate for 45 min. Samples were then dehydrated through a series of alcohols before being transferred into propylene oxide and gradually embedded in Eponate resin. All tissue samples were coded with respect to treatment condition to ensure treatment-blind analysis.
Cortical Thickness Analysis
In the areas to be quantified for neuron density, cortical thickness measurements were taken from those sections at ×5 magnification using a light microscope with a ruler built into the eyepiece. Section placement was manipulated until the ruler spanned the midline of the tissue, perpendicular to the white matter and the pia. Measurements were taken from the dorsal edge of the white matter of corpus callosum to the pial membrane. Thickness was recorded according to the smallest increments on the ruler (1 unit = 10 μm), and the unit measurement was multiplied by 200 to translate thickness into micrometers. For each rat, a mean cortical thickness measurement was calculated from each of 3 serial sections 1 μm apart.
Neuron Density Analysis
One block from each rat was randomly chosen, and 120 serial 1-μm sections were taken using a diamond knife on an ultramicrotome. Every second section was mounted on slides (8 serial sections/slide), stained with Toluidine blue, and cortical layer V identified (area of largest pyramidal cells, top of bottom 1/3 of cortex). Openlab, a computer-assisted microscope and stereology software package (Improvision Inc., Lexington, MA), was used to obtain image files. The physical disector method (Sterio 1984) was used to obtain unbiased estimates of neuron density. Pairs of sections within each slide series were photographed and compared in succession. The first section in the series was considered the “reference” section and the second the “lookup” section, and then the second section became the reference section and the third the lookup section, and so on. Within an unbiased counting frame, the number of neuronal nuclei present in the reference section, but not the lookup section (Q−), was counted. The disector volume of tissue (Vdis) through which cells were counted was calculated by Vdis = (Aframe)(H)(2), where Aframe is the area of the counting frame, H is the section thickness (1 μm) multiplied by the number of sections, and 2 is a constant to account for the 1-μm separation between sections counted. Approximately 60 neurons were counted per rat. Neuronal density (NvNeuron) was determined by NvNeuron = Q−/Vdis.
All serial sections were counted by 2 independent scorers. The images were coded to assure that quantification was performed in a blind fashion. In the few instances, where the 2 scores disagreed the images were reexamined. If a classification error was detected it was corrected. In the rare cases, where judgments diverged an average score was entered.
Electron Microscopy and Morphological Analysis
Following 1-μm sectioning, a small pyramid was trimmed into layer V of the motor cortex using a 1-μm Toluidine blue section from that block as a guide. From the pyramid, 20 silver/gray, serial sections (color indicates ∼70 nm thick) were taken using a diamond knife and an ultramicrotome. Five to eight serial sections were collected on Formvar coated, slotted, copper grids and stained with lead citrate/uranyl acetate. Five grids were obtained from each rat. A 4-picture photomicrograph montage (×8000 negative magnification) was taken from the same position within each section.
Fifteen montages were attempted for each rat. Negatives of micrographs were enlarged, developed, printed to maximum size with the image contained on a 20.3 × 25.4–cm picture, and the final magnification (×17 200) determined. Enlargement was consistent throughout the experiment. The physical disector method (Sterio 1984) was then used to estimate synapse density from these micrographs. As above, pairs of electron micrographs (∼70 nm apart) within the series were compared, and the number of synapses present in the reference section but not the lookup section counted, within an unbiased counting frame of known area (determined to be 112.64 nm2 based on magnification and actual area of counting frame). Excitatory synapses were defined by the presence of a prominent postsynaptic density as identified by an asymmetrical accumulation of dense material on the cytoplasmic face of the postsynaptic element and a thin layer on the presynaptic membrane with at least 3 round vesicles. Excitatory synapses with at least one discontinuity in the postsynaptic density in one or more serial sections, and no vesicles in close apposition with the presynaptic membrane at the site of the break, were classified as perforated synapses. Synapses with a continuous postsynaptic density in all serial sections were classified as nonperforated. Inhibitory synapses were identified by the presence of at least 3 cigar-shaped vesicles and no obvious postsynaptic density but rather a thin, symmetrical layer of dense material on the pre- and postsynaptic membranes (Peters and others 1991) (see Fig. 1). Approximately 100 synapses were counted per rat. Two independent scorers who were blind to experimental conditions quantified all micrographs for number of synapses. In the few instances, where the 2 scores disagreed the micrographs were reexamined. If a classification error was detected it was corrected. In the rare cases, where judgments diverged an average score was entered.
A mixed-factorial analysis of variance (ANOVA) with Tukey's protected T follow-up tests was conducted on the evoked potential time course data. Two-tailed independent t-tests were conducted on the evoked potential threshold, anesthetic dosage, intracortical microstimulation threshold, cortical thickness, and neuron density measures. One-tailed independent t-tests were conducted on the total size of the caudal forelimb area movement representations, size of the proximal and distal movement representations, and number of excitatory and inhibitory synapses, consistent with our a priori hypotheses that the maps would decrease in size and that there would be a reduction in the number of excitatory synapses and an increase in the number of inhibitory synapses in the LTD induction group relative to the control group.
Effects of LFS on the Evoked Responses
Representative responses recorded at baseline and 2 days following 20 sessions of LFS, evoked in the right motor cortex by an ascending series of stimulation intensities to the corpus callosum, are shown in Figure 2. Examination of all field potentials showed that the mean (±standard error of mean [SEM]) minimum amount of current (μA) to elicit the evoked potentials in the control (51.0 ± 5.8) and LFS (54.7 ± 6.3) groups were not significantly (t10 = 0.43, P = 0.68) different from one another and did not change over the course of the experiment. Furthermore, the largest proportional reduction in evoked potential peak height was elicited by 681 μA of current. For this reason, the potentials evoked by 681 μA are the only ones reported here.
Consistent with previous reports, LFS resulted in a LTD that was characterized by a decrease in the size of the monosynaptic component of the evoked potential (Froc and others 2000; Froc and Racine 2004; Monfils and Teskey 2004; Werk and others 2006). A mixed-factorial ANOVA revealed a significant (F1,10 = 24.74, P < 0.0006) effect of group, a nonsignificant (F4,4 = 1.62, P = 0.19) effect of day, and a significant (F4,40 = 4.03, P < 0.008) group by day interaction. Tukey's protected T follow-up tests indicated significant differences between control and LTD groups at 2 (P < 0.01), 9 (P < 0.01), 16 (P < 0.01), and 23 (P < 0.05) days following the last stimulation session. There was no significant difference between the 2 groups, 30 days after stimulation (see Fig. 3).
Effects of LTD Induction on Motor Representations
The mean (±SEM) amounts of ketamine, as a function of body weight and duration of surgery (expressed in ml/kg/min), administered to the control (0.0101 ± 0.0008) and LTD (0.0109 ± 0.0017) rats were not significantly (t18 = 0.38, P = 0.71) different. Furthermore, the mean (±SEM) minimum (threshold) current intensities required to elicit forelimb motor responses were not significantly (t21 = 1.57, P = 0.13) different across all responsive points in the control (26.09 ± 1.46) and LTD (29.53 ± 1.64) groups.
The total mean neocortical area of caudal forelimb movement representations in depressed rats was 30% smaller than control. This contraction of the forelimb movement area occurred in all directions relative to the control area (Fig. 4). Induction of LTD resulted in a significant (t22 = 3.073, P < 0.01) reduction in the size of the total caudal forelimb area compared with control (Fig. 4). When comparing the size of proximal (shoulder and elbow) and distal (wrist and digit) representations, we observed that LTD induction resulted in a significant (t22 = 3.057, P < 0.01) decrease in the size of distal representations relative to control. The reduction in the distal representations accounted for 93% of the total reduction in size of the motor maps in the LTD group (Fig. 5).
Effects of LTD Induction on Cortical Thickness, Neuron Density, and Synapse Type
The mean (±SEM) cortical thickness of the control and LTD rats was not significantly (t10 = 1.01, P = 0.33) different from each other. Likewise, neuron density in layer V of control and LTD rats was not significantly (t8 = 0.28, P = 0.79) different from each other (Fig. 6). Because the neuron density and cortical thickness measures did not differ, no gross change in neuropil was presumed to have occurred. Therefore, synapse densities were expressed as density per unit volume.
LTD induction did result in a significant (t8 = 1.95, P < 0.05) reduction in the density of excitatory perforated synapses relative to control. Although there was an increase in the density of excitatory nonperforated synapses in the LTD induction group relative to control, it was not a statistically significant (t8 = 1.4, P = 0.11) effect. There was a significant (t8 = 1.87, P < 0.05) increase in the density of inhibitory synapses in the LTD induction group compared with control (Fig. 7).
The main observation of this study is that under identical intracortical microstimulation conditions, the average size of motor maps was reduced in rats that underwent LTD induction. Thus, the caudal forelimb area of the neocortex that was activated by spaced and repeated LFS of the callosum has undergone some functional change. This functional change reduced the ability to elicit forelimb movements from regions of neocortex that produced movements in control rats in response to similar stimulation. Because the thresholds for obtaining intracortical microstimulation-induced movements were equivalent in the LTD and control groups, the changes in synaptic efficacy and map size are not the result of a nonspecific desensitization of the neocortex (Kleim and others 1998). Furthermore, the positive relationship between synaptic efficacy and the size of movement representations is in general agreement with our previous results that showed that increased synaptic efficacy leads to increased motor map area (Monfils and others 2004).
At the level of the sensorimotor neocortex, the corpus callosum is composed of axons from pyramidal cells in layer V (and layer III). Thus, patterned electrical stimulation applied to the corpus callosum results in both orthodromic and antidromic activation of those pyramidal cells. Layer V pyramidal cells also send axons to neighboring layer V cell dendrites, and functional synaptic coupling of a small fraction of these neurons is correlated with the existence of synaptic boutons at existing touch sites (Kalisman and others 2005). Pre- and postsynaptic interactions within layer V form specific functional microcircuits that are likely responsible for the motor maps because intracortical microstimulation is probably exciting layer V horizontal fibers rather than via directly exciting the few corticospinal neurons (Jankowska and others 1975; Lemon and others 1987). Thus, we conclude that changes in synaptic efficacy within layer V mediate, to some degree, the overall size of the motor map.
Previous studies have shown that changes to the internal organization of motor maps can be induced within the caudal forelimb area following skilled motor training (Kleim and others 1996, 1998, 2002). Skilled motor training was found to induce an increase in distal representations at the expense of proximal representations, but the overall size of the caudal forelimb area was not altered. The results reported here suggest that LTD induction yielded an important difference from skilled training: the loss of distal representations and a consequent reduction in overall map size. In fact, the retraction of the boundaries resulting in a smaller map in this study can be accounted for by the loss of the wrist and digit representations. In the case of skilled training, the increase in the distal representations is thought to occur because training on the pellet-reaching task requires preferential skilled use of the wrist and digits (Kleim and others 1998, 2004). It is also possible that distal representations are both more plastic and more sensitive to LFS-induced alterations in neural connectivity. We have previously reported that seizures repeatedly induced with electrical stimulation resulted in an increase in distal representations (Teskey and others 2002), but this was not the case when high-frequency theta-burst stimulation was used to induce long-term potentiation (Monfils and others 2004). Future research should be undertaken to determine if the observed differences in the plasticity of proximal and distal representations are affected by electrode placement and/or differential responses to a range of stimulation frequencies.
It is important to point out that the area of neocortex, defined as the caudal forelimb area, is highly dependent upon the methods used during mapping. Stimulation parameters such as current intensity, pulse train, stimulation frequency, as well as the anesthetic level can all affect the motor cortex maps (Donoghue and Wise 1982; Nudo and others 1990; Kleim and others 1998). Although the baseline maps reported in any experiment are dependent upon the techniques being used, the size of caudal forelimb area movement representations for control rats in the present study is consistent with those of previous reports that used similar stimulation parameters (Gioanni and Lamarche 1985; Castro-Alamancos and Borrel 1993; Kleim and others 1998; Monfils and others 2004). Furthermore, in the present study, the map reduction was not dependent on anesthetic levels or the sensitivity, as measured by the movement threshold at positive response sites, of the map area to intracortical microstimulation.
A key question is whether the LFS used to induce LTD caused damage and whether it is that damage that is responsible for the reduced evoked potential amplitude. There is evidence that suggests it is unlikely that the LTD that we observed in this study was due to stimulation-induced damage. In the present study, we did not observe changes in the threshold for an evoked response indicating that it is unlikely that damage occurred at the level of the stimulated axons. We also observed a recovery of the evoked potential size to control levels by 30 days post-LFS that extends the anecdotal report that LTD persists for at least 2 weeks (Froc and others 2000). Furthermore, repeated stimulation with the same-sized pulses at higher frequencies results in synaptic potentiation (Racine and others 1995; Trepel and Racine 1998; Monfils and others 2004), and the neocortex can be potentiated after the induction of LTD (Froc and Racine 2004). In the present study, we also examined both neocortical thickness and layer V neuron density, and in both cases, we failed to find differences in those measures between control and LTD induction conditions. Thus taken together, the electrophysiological results and the lack of anatomical changes suggest that it is highly unlikely that stimulation-induced damage occurred in our preparation.
LTD induction has previously been shown to lead to a decrease in dendritic length, complexity, and spine density in both layers III and V of sensorimotor neocortex (Monfils and Teskey 2004). The results of the present study extend those observations and demonstrate that LTD induction is associated with a loss of excitatory perforated synapses in layer V. Perforated synapses are thought to be more efficacious due to the narrower synaptic cleft between pre- and postsynaptic membranes at the site of the perforation (Sirevaag and Greenough 1985) and because of the additional “edges” at the perforation sites (Greenough and others 1978). Perforated synapses also have approximately 6.5 times greater alpha-amino-3-hydroxy-5-methylisoxazole-4-proprionic acid (AMPA) receptor density, 80% higher N-methyl-D-aspartate (NMDA) receptor density, and are significantly larger than nonperforated synapses (Ganeshina and others 2004). Cumulatively, fewer numbers of perforated synapses onto a postsynaptic neuron would decrease the probability of firing that neuron and result in lower synaptic efficacy. Induction of neocortical LTD in the slice is found to be dependent on mGluRs (Kato and Yoshimura 1993), whereas in the freely moving rat it was NMDA receptor–independent (Froc and Racine 2005). LTD has been found to involve endocytosis of AMPA receptors (Carroll, Beattie, and others 1999; Carroll, Lissin, and others 1999; Luscher and others 1999), and the NMDA receptor 1 subunit of the NMDA receptor has been reported to decrease (Heynen and others 2000) following the induction of LTD. We extend those observations and demonstrate that there are a reduced number of perforated synapses. The loss of perforated synapses with their high level of AMPA and NMDA receptors could account for the reduced synaptic efficacy and the size of movement representations. Changes in γ-aminobutyric acidergic (GABAergic) transmission may also underlie both LTD and alterations in movement representations. Jacobs and Donoghue (1991) demonstrated that by decreasing intracortical inhibition through application of GABA antagonists, cortical representations could be reshaped. The increase in inhibitory synapses observed in this study may also account for the reduced size of the motor maps. Therefore, a potential neurochemical mechanism for the caudal forelimb area retraction is that adjacent cortical regions expand when preexisting lateral excitatory connections are masked by an increase in intracortical inhibition (Jacobs and Donoghue 1991). Thus, both enhancement of GABA through increased number of inhibitory synapses and reduction of glutamatergic-mediated currents through a loss of perforated synapses may underlie the synaptic depression, as well as the reduction in the size of the movement representations.
Given that changes in synaptic efficacy induced through application of patterned stimulation are thought to model the synaptic processes underlying learning and memory, the present results further support the notion that changes in synaptic efficacy provide a common mechanism of learning, memory, and motor map reorganization operating within the motor cortex. Apart from the theoretical relationship between learning, memory, synaptic efficacy, and movement representations, LFS could potentially provide a practical tool for disrupting the initiation of seizures (Khosravani and others 2003) or reversing seizure-induced neural reorganization (Goodman and others 2005). This report has shown that LTD results in the opposite effects of repeated seizures, which lead to an expansion of motor map area (Teskey and others 2002) and an increase in the density of perforated synapses (Goertzen and Teskey 2003; Teskey and others 2005). Indeed, there is evidence that LFS can “quench” the effect of seizures in experimental models (Gaito 1980; Weiss and others 1998; Adamec 1999; Goodman and others 2005).
This work was supported by Natural Sciences and Engineering Research Council and Canadian Stroke Network grants to GCT. The authors would like to thank Richard Dyck, Doug Bray, Penny VandenBerg, Zain Jivraj, Arden Lee, Lana Ozen, Erin Carter, Faith Ng, and Palki Arora for technical assistance. Conflict of Interest: None declared.