1 What drives the plasticity of brain tissues?
The behavioral sensitivity of transient and persistent processes of cellular adaptation, and the fact that both brain activity and learning are involved in the behavioral events that appear to drive this adaptation naturally led to the following questions: What causes changes in neurons, glia and vasculature? Can we rule out causes such as hormonal or metabolic responses to behavioral manipulations? Can we then attribute experience-induced neuronal and non-neuronal plasticity to learning or to brain activity?
It is possible to imagine that cellular responses to experience that could appear to be due to learning could actually be responses to stress or related metabolic consequences. Certainly stress can affect the morphology of neurons in some brain regions (e.g., Sapolsky, 1996), although the astroglial changes in the hippocampal formation that are correlated with adrenal hypertrophy appear to be dissociated from visual cortex changes in complex environment research that are correlated with experience (Sirevaag, et al., 1991). To examine the roles of learning vs. other consequences of training (e.g., activity, stress, etc) on neuronal changes, we have utilized paradigms in which the effects of learning would be focused in particular regions of the brain for which other regions could serve as control or comparison samples.
In one early study, Chang and Greenough (1982) compared rats trained in a complex series of changing maze patterns that learned with either the same eye always occluded or with occlusion of alternate eyes on successive days. In rodents, approximately 90-95% of retinal ganglion cell axons project to the opposite side of the brain (Lund, 1965). The unilaterally-occluded rats, therefore, should have most training-related activity restricted to the hemisphere opposite the open eye, whereas the bilaterally-alternating occlusion should allocate the learning-related activity about equally to both hemispheres. Both groups had been previously subjected to transection of the corpus callosum, a “split-brain” procedure that eliminates communication between the two hemispheres of the brain. Control animals were surgically operated, divided into unilaterally- and bilaterally- occluded groups, and subsequently handled but not trained.
The results of this study indicated that dendritic branching, a correlate of synapse number, was increased in the non-occluded hemisphere of the unilaterally trained group, compared to the occluded hemisphere, and in both hemispheres of the alternately-trained group relative to the non-trained group. These results indicate that non-specific effects such as stress, which would have been distributed equally between hemispheres in the unilaterally-trained group, can not account for experience-induced plasticity. Rather, either training or training-related activity drives dendritic plasticity.
As another example of the dissociation between non-specific and specific effects of experience on plasticity, Greenough et al. (1985) examined activated vs. non-activated hemispheres of rats trained to reach for a food reward with alternating forepaws or with only one forepaw. Dendritic branching in somatosensory-somatomotor cortex in these animals was compared to untrained controls. For deep pyramidal neurons of the type that control forelimb activity, apical dendritic branching was greater in the hemisphere opposite trained forelimbs than in ipsilateral homologous areas. Dendritic arborization in the group trained with alternating forepaws was greater than untrained controls. These findings parallel those described for visual cortex by Chang and Greenough (1982). In contrast, while there were effects of training on more superficial pyramidal neurons, these effects were not restricted to the activated side in unilaterally-trained animals (Withers and Greenough, 1989). These observations of morphological changes expressed in extragranular layers of the untrained cortex (as well as the trained side) are consistent with the idea that plasticity of intercortical connections plays a key role, possibly acting as the locus of plasticity initiation within the cortex [Gilbert, 1979 #191;Gilbert, 1992 #192;Hess, 1994 #224].
The work of Chang and Greenough (1982).and Greenough et al. (1985) suggests that studies indicate that learning or some other aspect of training-related activity drives morphological change in neurons. Non-specific effects such as globally-acting hormonal or metabolic differences were not the causal force behind these morphological changes, as such non-specific effects would be expected to alter comparable regions of the brain whether or not they were selectively activated by unilateral training.
Although these studies suggest that the artifact of nonspecific stress or metabolic effects on training-related changes may be small, we should continue to consider this possibility and to run appropriate controls. However, even if this artifact is ruled out, the issue remains that any learning involves at least some brain activity, and if learning is localized in the brain (e.g., to one hemisphere), the activity is likely to be localized as well. Is brain adaptation learning-related, or can activity alone drive plastic neural change?
With regard to this question, we have muscle as a model: with sufficient activity, muscle will hypertrophy, the particular details varying with the extent and pattern of activation. One can suppose that activation of brain tissues in association with peripheral activity might similarly trigger neuronal, glial or vascular hypertrophy of the sort seen in rats after complex environment housing. Learning, in contrast, likely causes changes in neurons and glia, but perhaps not capillaries as is harder to imagine that changes in vasculature would play very specific roles in learning.
2 Activity vs. Learning
To more directly address the issue of whether activity or learning causes structural changes in the brain Black et al. (1990) designed a paradigm in which adult rats were given the opportunity for either 1) a substantial amount of learning with relatively little physical activity (AC below), 2) a substantial amount of physical activity with relatively little learning (FX and VX below) or 3) minimal opportunity for physical activity or learning (IC below). ACrobatic rats (AC) completed a multi-element elevated obstacle course that required learning significant motor skill while providing only limited exercise. Forced eXercise (FX) rats ran on a treadmill, reaching durations of 60 minutes a day, exercising but with very little opportunity for learning. Voluntary eXercise (VX) rats had access to running wheels attached to their cages and were the only group to exhibit increased heart weight, a sign of aerobic exercise. Inactive Condition (IC) rats were merely removed from their cages for brief daily experimenter handling, providing neither activity nor learning.
JEFF, you stopped here!!!
In initial studies focusing on cerebellar cortex, Black et al. (1990) both the density of blood vessels and the number of synapses per neuron (Purkinje cells were used as a basis for this calculation). For the number of synapses per neuron, as depicted in Figure 3A, the learning group, AC, exceeded the other 3 groups, which did not differ, suggesting that when learning takes place, new synapses are formed. By contrast, as Figure 3B shows, the FX and VX groups both had a greater density of blood vessels than the AC or IC groups, which did not differ; this suggests that the formation of new blood vessels was driven by neural activity, and not by learning.
FIGURE OUT KLEIM”S REFERENCES.
Subsequent work indicated that these effects were not limited to cerebellar cortex. Kleim et al. (1996; ANYOTHER REF?) described synaptogenesis and changes in synapse morphology in association with the same AC motor learning procedure in forelimb area of primary sensory-motor cortex of rats. One of the first morphological changes to occuris an increase in the average size of PSDs, which occurs within two days after beginning training. After five days of training, an increase in the number of synapses per neuron was detected, and the average size of synapses had decreased, possibly because the new synapses were, on average, smaller than the pre-existing synapse population. This increased synapse number was maintained across the remaining days of training. As training progressed, the average synapse size increased again, suggesting that the new synapses were growing larger or that the population of synapses overall was increasing in size. A schematic interpretation of this process is depicted in Figure X. JEFF, we’ve revamped your figure a bit, and prefer to use this rather than the two graphs – 1 published, 1 unpublished. These findings suggest that at least two independent mechanisms of synaptic plasticity, the formation of new synapses and the enlargement of existing synapses, are activated in parallel during motor learning
Jeff’s comment from above goes best here somehow:
It is likely that angiogenesis is driven by increased activity within a specific brain region as the repeated performance of unskilled movements such as those produced during exercise causes increases in capillary density (Black et al., 1990; Isaacs et al., 1992 #230; Kleim et al., 2002). Further, changes in blood vessel density can occur independent of changes in synapse number (Black et al., 1990) and, as noted above, these changes are reflected in functional measures of blood flow and its responsiveness to oxygen demand (Swain et al., in press).
It might be argued that running on a treadmill or in a wheel is sufficiently different from skilled movements involved in the AC task that this difference, rather than the learning difference, XXXXXXXX. However, even in paradigms in which the movements were very similar, the acquisition of skill has been selectively associated with the addition of synapses (Kleim et al., 1996; Kleim et al., 1998c; Kleim et al., 2002 this last reference may be the only one to support that statement, or another may be needed JEFF, can you address this??). Similarly, the acquisition of skilled forelimb movements resulted in reorganization of forelimb movement representations within motor cortex (Nudo et al., 1996; Kleim et al., 1998a) while extensive repetition of unskilled movements (Plautz et al., 2000; Kleim et al., In press) and forelimb strength training (Remple, et al., 2001) were without effect. In contrast, strength training increased synapse number in the ventral spinal cord but motor skill training was without effect (Kleim, et al., 2001).
Differential patterns of plasticity can also be observed across types of learning that seemingly use similar neuronal pathways. For example, motor skill training that altered cerebellar cortex (in work noted above) did not detectably alter synapse number within the deep cerebellar nuclei (Kleim, et al., 1998) whereas eye blink conditioning that did not detectably affect cerebellar cortex (by one measure: Anderson et al. XXXX) did affect the deep nucleus (Bruneau, et al., 2001). It is perhaps not surprising that different types of learning would be differently represented in the same structures, given the lateralized effects of training described above (e.g., Chang & Greenough 1982), but there are remarkably few demonstrations of this.
Similar specificity of plasticity is seen in subpopulations of neurons within the same region: specificity of the plasticity can even be reduced to subpopulations of neurons within the same brain region. For example, complex housing caused dendritic hypertrophy within cerebellar Purkinje cells but not granule cells (Floeter and Greenough, 1979). Reach training caused dendritic hypertrophy within layer II/III of the motor cortex that was restricted to a specific class of pyramidal cells (Withers and Greenough, 1989). Finally, plasticity can even be restricted to specific afferents onto individual neurons. Complex motor skill training increased parallel fiber synapses onto Purkinje cells but not climbing fibers (Kleim, et al., 1998). Eyeblink conditioning increased the number of excitatory synapses within the anterior interpositus without altering inhibitory synapse numbers (Bruneau, et al., 2001). This result is particularly interesting because the same system, with two excitatory afferents to the cortex and a single inhibitory output to the deep nuclei, is showing different patterns/locations of plasticity under different training conditions; perhaps when learning involves the integration of an array of body positioning movements and counter-forces the detailed representation of the body evident in the afferent parallel fiber system handles the task, whereas when the focal disinhibition of a response to a sensory stimulus is involved, this takes place at the output level of the cerebellar cortex. Similarly strength training increased excitatory but not inhibitory axosomatic synapses within the ventral spinal cord (Kleim, et al., 2001). This degree of specificity is consonant with the concept of local “synaptic tags” that designate locations where synapses are to be modified, or perhaps generated (e.g., Frey and Morris, 1997).
A further dissociation of synaptic plasticities deserves mention. While some studies have suggested that synaptogenesis may be associated with long-term potentiation (e.g., [Lee et al., 1980 #233;Chang & Greenough, 1984 #231]; Engert and Bonhoeffer, 1999; Maletic-Savatic, et al., 1999; Andersen and Soleng, 1998), others have shown explicit dissociations between electrically-induced LTP and behaviorally-induced synaptogenesis in receptor mutants [Rampon et al. 2000 #87]. The dissociation seems the more powerful result: synapse addition may mediate or be involved in some aspects of LTP, but synapse addition need not involve an LTP-like process for its induction.
Primary neurons and the cortex: Is the current approach overly-restrictive?
Historically, investigations into neuronal plasticity have focused, almost exclusively, on modifications in the structure and function of primary neurons and circuits. While the importance of these systems can not be overlooked, it belies the fact that neurons not directly involved in such pathways far out number those that do. Seress et al. (2001) have reported that 95% of terminals forming asymmetric synapses with parvalbumin-positive (assumed to be GABAergic) dendrites in the dentate and strata pyramidale and lucidum of CA3 are from granule cells. Such observations beg the question as to why large projection neurons (e.g., pyramidal cells) are the primary focus of so many studies while historically “modulatory” cells are largely ignored. It has been shown that modulatory systems can have dramatic effects on the function of primary neurons. For example, receptive field size in primary sensory cortex has been shown to be sensitive to pharmacological disinhibition [Jacobs, 1991 #187;Tremere, 2001 #185]. The source of such inhibition likely stems from extragranular layers of cortex as rapid (physiologically-defined) changes in both visual (Trachtenberg, et al., 2000; Trachtenberg and Stryker, 2001) and somatosensory cortices (Diamond, et al., 1994) have been reported to occur prior to the expression of modifications in layer IV. Finally, observations by Gilbert and colleagues further suggest that horizontal connections may be the impetus?? to reorganization of visuo- and topo-graphic maps (Gilbert and Wiesel, 1979; Gilbert, 1992).
Much like the argument that additional scrutiny of “modulatory” neurons appears warranted, we also believe that it is both naïve and foolish to emphasize the role of cerebral cortex so heavily. That is, the cerebral cortex is often-times considered “where” changes occur, ignoring the fact that in a diverging pathway (e.g., somatosensory projections to S1), minor modifications in subcortical connectivity would have much larger effects at the level of the cortex. The idea that modifications of subcortical areas of the brain express plasticity is not a novel one and has been shown to occur in areas from the spinal cord (Devor and Wall, 1978) to striatum (Comery, et al., 1995; 1996). In general, plasticity appears to be a fundamental property of the nervous system and as such, the contribution of areas beyond those most readily accessible need to be fully considered.
Finally we return to a point made at the outset and the subsequent presentation of supporting data: in both cortical and subcortical areas there is ample and growing evidence for substantial plasticity of non-neural cells such as oligodendrocytes, astrocytes and vasculature. In general we do not know what intermediary events relate behavioral or physiological activity to changes in glial or vascular structure and function. The morphological and functional changes may be transient or lasting and may be associated with learning-driven processes such as synaptogenesis or with vascular responses to activity. Astrocytic changes that appear to selectively accompany synaptogenesis (e.g., Anderson 94) may require activity for their maintenance (Kleim in revision). We are still working to understand the behavioral forces that regulate these non-neuronal plasticities as well as the cellular and molecular events that mediate them. Perhaps most important, we must continue to work on their functional roles.
Summary and Conclusion
A view that is emerging is that the brain has multiple forms of plasticity that must be governed, at least in part, by independent mechanisms. This view is illustrated by 1) the apparent separate governance of some non-neural changes by activity, in contrast to synaptic changes driven by learning, 2) the apparent independence of different kinds of synaptic changes that occur in response to the learning aspects of training, 3) the occurrence of separate patterns of synaptic plasticity in the same system in response to different task demands, 4) apparent dissociations between behaviorally-induced synaptogenesis and LTP and 5) reports of an increasing number of nonsynaptic forms of plasticity in neurons. The historical focus of research and theory in areas ranging from learning and memory to experiential modulation of brain development has been heavily upon synaptic plasticity since shortly after the discovery of the synapse. Based upon available data, it could be argued that 1) synaptic, and even neuronal, plasticity is but a small fraction of the overall range of changes that occur in response to experience and that 2) we are just beginning to understand the importance of these other forms of brain plasticity. Appreciation of this aspect of the brain adaptive process may allow us to better understand the capacity of the brain to tailor a particular set of changes to the demands of the particular experiences that generated them.
******* THIS IS THE FORMAL END OF THE CHAPTER ********
Bill’s still working on the rest.
I checked ref’s up to here.
Role of non-neuronal changes in learning / activity based plasticity. – MAINTENANCE!!!!
One is Brenda's 1994? Paper showing the correlation between synapse number and astrocyte Vv. The other is Jeff's astrocyte persistence paper. Discussed already?????Now covered in pre-conclusion paragr