Dendritic Spine Research—Putting Meat on the Bones
3 February 2006. Dendritic spines are rapidly becoming a darling of neuroscientists. They are the main site for excitatory synaptic transmission in the brain, and their waxing and waning underlie sophisticated modulations of neuronal circuits in development (Bock et al., 2005), not to mention mechanisms of learning and memory, such as long-term potentiation (see Matsuzaki et al., 2004) and long-term depression (see Zhou et al., 2004). All of the above make them particularly interesting to researchers studying schizophrenia.
Despite the budding interest, not much is known about where, how, or why these synaptic stems grow, but three independent basic science studies published January 19 shed some light on all these questions. And a study by David Lewis and colleagues at the University of Pittsburgh, published January 10, reports reduced expression of Cdc42 and Duo, important regulators of spine morphology, in schizophrenia.
Spirals of spines
First, the "where" of dendritic spines. John O’Brien and Nigel Unwin from the Medical Research Council Laboratory of Molecular Biology in Cambridge, England, reported in last week’s PNAS online that the distribution of dendritic spines is far from random, at least in the Purkinje neurons of the cerebellum. Rather, spines grow in elaborate and regular linear arrays, and they actually trace short-pitch helical paths around the dendrites.
O’Brien and Unwin coated Purkinje neurons from the mormyrid fish with a lipophilic fluorescent dye, and then examined tissue slices with a confocal microscope. Dendritic spines in the fish are less dense than in the mammalian brain, and the cerebellum is more regular. The authors found that in regions of low density, spines were evenly spaced along the dendrites, typically at about 0.54 μM apart. At high density this pattern was not easy to identify, but the authors detected a different pattern that also indicates regular spine spacing. In the fish, the Purkinje dendrites are long, straight, and parallel, forming a planar “palisade,” and the authors sometimes found lines of side-by-side spines in periodic groupings.
Using image analysis, O’Brien and Unwin generated a “diffraction pattern,” which revealed that spines along individual dendritic shafts trace a good approximation of a helix. The helix had a pitch of about 1.25 μM, which would fit about three spines per turn. “The helical ordering of spines gives rise to a set of similar surface lattices, the dimensions of which lead to approximately equal sampling of the surrounding space by the spineheads,” write the authors. In other words, it would seem the spines are arranged to maximize the probability that the dendritic arbor would interact with any afferent axon.
Extending these observations to mammals, the authors examined both the “weaver” mouse, a mutant that has fewer axons innervating the Purkinje cells, which have relatively straight dendritic shafts, and wild-type mice. The mammalian spines were more variable in shape, but even so, the authors detected a periodic spacing of about 0.58 μM. They also detected a helical pattern that was, again, slightly longer in pitch than in the fish. Because the periodicity and the pitch were very similar in all three cases, the authors conclude that this ordering of spines is an inherent property of dendrites that is not influenced by external factors, such as presynaptic activity. They also suggest that the periodicity indicates the involvement of a filamentous protein, such as actin. “Giant actin-binding proteins, such as nebulin, are more than long enough to span the interval between successive spines and could play a role in creating regularly spaced templates for the growth of filopodia, from which mature spines are thought to develop,” they write.
Myosin—not just for muscles anymore
Which brings this story straight to the “how” of dendritic spines. Morgan Sheng, at the Picower Institute for Learning and Memory at Massachusetts Institute of Technology, and colleagues there and elsewhere, report in the January 19 Neuron that myosin IIB plays a critical role in the formation of spines.
Myosin II was once dismissed as having no role in spine formation because experiments showed that the myosin inhibitor 2,3 butanedione-monoxime (BDM) had no effect on morphology or motility of dendritic spines. But BDM was subsequently shown to be a poor inhibitor of myosin IIB, one of the two isoforms found in non-muscle cells. That revelation, plus recent proteomic data suggesting the postsynaptic density is loaded with the protein (see Jordan et al., 2004; Peng et al., 2004) prompted Sheng and colleagues to reinvestigate a role for myosin IIB in spines.
First author Jubin Ryu and colleagues first confirmed the proteomic data using immunohistochemistry to show that in cultured hippocampal neurons, the vast majority of neuronal myosin IIB congregates with postsynaptic density 95, a marker of synaptic sites. Then, to test if this myosin has functional significance, Ryu and colleagues dosed the neurons with blebbistatin, a molecule that specifically inhibits the ATPase activity of myosin II. Ryu and colleagues found that within minutes of adding the chemical, the mushroom shaped heads of the dendritic spines started to disappear, and by about half an hour, all that was left of most spines were long, thin filopodia. Knocking down myosin IIB by RNA interference had a similar effect.
Spine dynamics have long been linked to actin filaments, but these data indicate that growth and regression of actin filaments are not the only means to control spine morphology. Sheng and colleagues suggest that myosin, which helps maintain tension in actin filament networks in other scenarios, such as cytokinesis, may have the same effect in spines. By providing tangential force at the spine head membrane, myosin might counteract the outward push generated by growth of actin filaments, creating a mushroom head, much like the bud in yeast. In fact, without myosin, budding in yeast is disrupted. Data in Sheng’s paper supported this idea. Time-lapse confocal microscopy showed that the normal extension and retraction of the filopodia is reduced to just extension in the presence of blebbistatin, suggesting that myosin does rein in the filopodia.
Perhaps one of the most interesting aspects of myosin IIB is that its effects on morphology seem tightly linked to the spines’ function. When Ryu and colleagues measured excitatory postsynaptic potentials in blebbistatin-treated cells, they found that amplitudes were down by more than half. This may well be related to loss of AMPA receptors, because addition of the myosin II inhibitor led to depletion of these receptors.
A micro-chaperone for Limk1
Together, these two papers reveal new insights into where and how spines form, but they don’t address the question of why. That’s left up to Michael Greenberg and colleagues at Children’s Hospital and Harvard Medical School in Boston, and also at the Medical University of Vienna, Austria. First author Gerhard Schratt and colleagues reported in the January 19 Nature that spine volume is regulated by a microRNA.
Schratt and colleagues investigated the role of the brain microRNA-134 (miR-134), finding that it is localized near synaptic sites in dendrites on cultured hippocampal neurons. When overexpressed, it decreased the volume of dendritic spines. This result prompted the authors to search the genome for genes this small RNA might target. One of the sequences that partially matches the sequence of miR-134 is in the 3’ untranslated region of the gene for Lim domain-containing protein kinase (Limk1). The microRNA not only binds to this site, but it also inhibits translation of the kinase, Schratt and colleagues report.
This finding may be of particular interest to schizophrenia researchers. Ablation of Limk1 leads to cognitive defects in mice (see Meng et al., 2002), and the kinase has been linked to two molecules that are heavily implicated in synaptic plasticity, brain-derived neurotrophic factor (BDNF) and actin. The former induces expression of Limk1, while actin polymerization can be regulated by the Limk1 substrate, cofilin. Actin, of course, is involved in the extension of spine filopodia. In fact, Schratt and colleagues found that reporter genes only respond to BDNF if they carry the miR-134 binding site, indicating that the neurotrophin may somehow relieve repression of Limk1 translation.
The authors speculate that miR-134 may keep Limk1 mRNA in a dormant state while it is being transported to synaptic sites. “Upon synaptic stimulation, the release of BDNF may trigger activation of the TrkB/mTOR signaling pathways, which inactivates the miR-134-associated silencing complex by an as-yet-unknown mechanism, leading to enhanced Limk1 protein synthesis and spine growth,” suggest the authors.
Cytoskeletal proteins perturbed in schizophrenia
Dendritic spines have also caught the attention of some schizophrenia researchers (see, e.g., Law et al. 2004), as has synaptic dysfunction in general (for review, see Harrison and Weinberger, 2005; Owen et al., 2005). Previously, David Lewis and colleagues demonstrated layer-specific reductions in spine density in the dorsolateral prefrontal cortex in people with schizophrenia (Kolluri et al., 2005; Glantz and Lewis, 2000). In the present paper, first author Justin Hill, Takanori Hashimoto, and Lewis put a finer focus on these abnormalities, exploring whether they correlate with changes in expression of any of five proteins known to be involved in the regulation of spine morphology.
Hill and colleagues report reductions of mRNA for two proteins, Cdc42 and Duo, in schizophrenia versus control postmortem tissue. The changes were detected in gray matter, but not white matter, and correlated with spine density in individual study subjects. The RhoGTPase Cdc42 is known to be critical for the formation of filopodia, and Duo, a GDP-GTP exchange factor, is thought to be involved in maintaining mature spines and the apposition between pre- and postsynaptic sites. Curiously, the researchers did not detect corresponding changes in proteins closely related to these two, such as RhoA, Rac1, or drebrin. In an effort to control for possible effects of antipsychotic medications, the researchers administered drugs to macaques but found no changes in mRNAs of the proteins of interest from either first- or second-generation antipsychotics (haloperidol and olanzapine). "Taken together, a decrease in the expression of both Cdc42 and Duo mRNAs would be expected to result in a reduction in the formation of new spines and an impairment in the maintenance of existing, mature spines," they write. It might be surprising that the other cytoskeletal mRNAs assayed were unchanged, given their close relationships to Cdc42 and Duo in regulating spines. But, as the authors point out, changes in levels of Duo, for example, could affect activation of its downstream target Rac1, which helps maintain spine stability, without necessarily changing levels of the target protein.
Because their earlier work had found lower spine density in dendrites of the pyramidal cells of layer 3, but not those of layer 6, Hill and colleagues guessed that they might find corresponding differences in the distribution of Cdc42 and Duo mRNAs. This did not turn out to be the case—a finding that the authors interpret to support the idea that changes in the expression of the two cytoskeletal proteins are causal, and not a consequence of changes of spine density. If Cdc42 and Duo expression had been found to be reduced only in layer 3, they reason, this would have left open the possibility that spine density reductions precede changes in expression of the proteins. But what, then, causes the selective reduction in these proteins, and of spines, in layer 3? Differential activation could be the explanation: The pyramidal cells of layers 3 and 6 have quite different sources of input, in part due to radically different dendritic arbor patterns. As excitatory input is required for spine formation, the authors speculate that a reduction of excitatory input to only layer 3 might underlie reduced spine density in its pyramidal cells.—Tom Fagan and Hakon Heimer.
Hill JJ, Hashimoto T, Lewis DA. Molecular mechanisms contributing to dendritic spine alterations in the prefrontal cortex of subjects with schizophrenia. Mol Psychiatry . 2006 Jan 10 ; Abstract
O’Brien J, Unwin N. Organization of spines on the dendrites of Purkinje cells. PNAS early edition. 19 January, 2006. Abstract
Ryu J, Liu L, Wong TP, Wu DC, Burette A, Weinberg R, Wang YT, Sheng M. A critical role for myosin IIB in dendritic spine morphology and synaptic function. Neuron. 2006 Jan 19;49(2):175-82. Abstract
Schratt GM, Tuebing F, Nigh EA, Kane CG, Sabatini ME, Kiebler M, Greenberg ME. A brain-specific microRNA regulates dendritic spine development. Nature. January 19, 2006;439:283-289. Abstract