NEWS AND VIEWS

and we look forward to the innovative 5. Greenwald, A.G., McGhee, D.E. & Schwartz, J.L.K. 11. Ito, T.A. & Urland, G.R. J. Pers. Soc. Psychol. 85, approaches that social neuroscience will con- J. Pers. Soc. Psychol. 74, 1464–1480 (1998). 616–626 (2003). 6. Fazio, R.H., & Olson, M.A. Annu. Rev. Psychol. 54, tinue to bring to these problems. 12. Brendl, C.M., Markman, A.B. & Messner, C. J. Pers. 297–327 (2003). Soc. Psychol. 81, 760–773 (2001). 7. Karpinski, A. & Hilton, J.L. J. Pers. Soc. Psychol. 81, 1. Richeson, J.A. et al. Nat. Neurosci. 6, 1323–1328 13. De Houwer, J. J. Exp. Soc. Psychol. 37, 443–451 774–788 (2001). (2003). (2001). 8. Nosek, B.A., Banaji, M.R. & Greenwald, A.G. Group 2. Botvinick, M.M., Braver, T.S., Barch, D.M., Carter, C.S. 14. Meyer, D.E. & Kieras, D.E. in Attention and & Cohen, J.D. Psychol. Rev. 108, 624–652 (2001). Dynam. 6, 101–115 (2002). Performance XVII. Cognitive Regulation of 3. Cacioppo, J.T. et al. J. Pers. Soc. Psychol. 85, 9. Rothermund, A.W., & Wentura, D. Z. Exp. Psychol. 650–661 (2003). 48, 94–106 (2001). Performance: Interaction of Theory and Application 4. Gehring, W.J. & Knight, R.T. Nat. Neurosci. 3, 10. Pratto, F. & John, O.P. J. Pers. Soc. Psychol. 61, (eds. Gopher, D. & Koriat, A.) 17–88 (MIT Press, 516–520 (2000). 380–391 (1991). Cambridge, Massachusetts, 1999).

Sensory-motor control: a long-awaited behavioral correlate of presynaptic inhibition

P Ken Rose & Stephen H Scott

Presynaptic inhibition of cutaneous afferents influences sensory-motor responses in the . In-vivo recordings in monkeys now show that this process suppresses the transmission of cutaneous signals generated during volitional movement.

A survey of popular neuroscience textbooks extension movements of the wrist. This is no results suggest that cutaneous input to the http://www.nature.com/natureneuroscience suggests that presynaptic inhibition is either small feat and is arguably one of the most spinal was suppressed during voli- rare or a figment of the imagination of phys- challenging experimental preparations in tional movement. This leads to two impor- iologists and anatomists of the 1960s. To the neuroscience today. The cells that the tant questions: what is the source of this contrary, considerable progress has been authors studied receive monosynaptic con- suppression and how does it occur? made in unraveling the cellular and molecu- nections from large-diameter cutaneous Seki et al.5 answer the first question by lar details of this neural process1,2,but little afferents. These cells occupy a strategic loca- showing that descending commands to the is known about its role in behavior3,4.A tion (Fig. 1); they are the first cells in the spinal cord are at least partially responsible for report in this issue by Seki et al.5 provides an spinal cord to relay information from suppressing cutaneous input to the spinal important functional face to presynaptic mechanoreceptors to the brain, and they are cord. The GABAergic shown in inhibition by showing how it modulates also the first cells in a circuit that terminates Figure 1 are known to receive connections cutaneous afferent input to spinal on spinal motor neurons. By regulating the from descending systems and primary affer- during behavior. flow of cutaneous signals at this location, ents supplying the skin2.Thus, activity in In its most conventional form, presynaptic descending commands can simultaneously either of these connections could inhibit the © Group 2003 Nature Publishing inhibition involves axo-axonic made influence motor behavior and the perception responses of first-order interneurons during by GABAergic interneurons2.Although the of somatosensory stimuli. The key question wrist movement. The transmission of cuta- precise actions of GABA remain open to is under what circumstances do descending neous signals was not reduced during passive debate, the end result is clear: GABAergic commands use presynaptic inhibition via movements, suggesting that descending com- activity reduces neurotransmitter GABAergic interneurons to regulate this mands were responsible for presynaptic inhi- release from the postsynaptic . As com- flow of information? bition during active movement. However, the pared to inhibitory synapses directly on The authors5 used several approaches to key observation is that the effect of nerve stim- the postsynaptic neuron, these axo-axonic show that presynaptic inhibition influences ulation was reduced by 20% even before the synapses selectively reduce input from a partic- the transmission of cutaneous input to onset of movement. This pre-movement ular presynaptic neuron without influencing spinal interneurons. First, the effect of stim- modulation could not be generated by periph- other inputs to the same postsynaptic neuron. ulating a cutaneous nerve on the activity of eral afferents and provides definitive proof Seki et al.5 studied the behavioral features spinal neurons was task dependent (Fig. 1). that descending commands are at least par- of presynaptic inhibition by applying proce- When the monkey actively flexed or tially responsible for modulating cutaneous dures pioneered in the lumbar spinal cord of extended its wrist, cell discharge increased input to the spinal cord during movement. anesthetized cats to the cervical spinal cord up to 8-fold, but the influence of simultane- The trickiest part of the study was to of monkeys trained to execute flexion and ous nerve stimulation decreased by 50% demonstrate that modulation of cutaneous from the rest condition. In contrast, compa- input during movement is due to presynaptic rable passive wrist movements did not affect rather than postsynaptic inhibition. The Ken Rose is in the Department of Physiology and the influence of nerve stimulation. Because finding that cutaneous responses are sup- Stephen Scott is in the Department of Anatomy and cell discharge was similar during active and pressed at the very time that the interneurons Cell Biology, Centre for Neuroscience Studies, passive movements, changes in cutaneous are highly active during extension and flexion Queen’s University, Kingston, Ontario K7L 3N6, input cannot be simply due to increased suggests that postsynaptic inhibition is not Canada. refractoriness caused by high-frequency dis- responsible for modulation of cutaneous e-mail: [email protected] charge during active movements. Rather, the nerve input. If postsynaptic inhibition is not

NATURE NEUROSCIENCE VOLUME 6 | NUMBER 12 | DECEMBER 2003 1243 NEWS AND VIEWS

Figure 1 Key neurons involved in the relay of cutaneous input, via first-order interneurons in the spinal cord, to segmental motor systems and higher sensory systems. The numbers (1–4) identify the anatomical loci of critical experiments performed by Seki et al.5 that led to the conclusion that transmission from cutaneous afferents to first-order neurons is dynamically regulated during voluntary motor tasks by presynaptic GABAergic interneurons.

at the site of recorded neurons inhibited motoneuron activity to limb muscles used to generate wrist movement (Fig. 1). This finding suggests that cutaneous afferent activity during movement would tend to inhibit ongoing motor commands. In the- ory, descending commands could compen- sate for this inhibitory cutaneous input, but the strategy appears to be that descending commands presynaptically inhibit cuta- neous signals, and thus reduce their influ- ence on the motor system. This may be related to a general strategy of reducing self- http://www.nature.com/natureneuroscience generated sensory information, in this case to guide action. Body movement is generated by muscle activity under the direct control of alpha- involved, the inhibition must be presynaptic. the findings of Seki et al.5 indicate that other motor neurons, the final common output Easy to say, hard to prove! inputs to first-order interneurons should be pathway from the spinal cord. It is easy to The solution to this obstacle lies in a well- immune to the presynaptic inhibition acting assume that the focus of descending com- described biophysical corollary of presynap- on cutaneous . Testing this possibility mands is only to control this output either tic inhibition. If we assume that the GABA will require even more demanding experi- directly or through spinal interneurons. released by the axo-axonic binds to ments. However, such efforts may reap signif- However, the work by Seki et al.5 highlights GABAA receptors, the axon will depolarize icant rewards. Separate populations of the much broader role of descending com- due to the opening of chloride channels. GABAergic interneurons form axo-axonic mands in controlling spinal processing11. (The equilibrium potential for chloride ions synapses on secondary affe- Corticospinal projections from somatosen- © Group 2003 Nature Publishing in these axons is typically more positive than rents and cutaneous afferents6.Thus, the sory cortex would provide a likely source for the resting membrane potential, so the stim- results reported by Seki et al.5 set the stage for presynaptic inhibition of cutaneous input. ulus is depolarizing.) This depolarization, a much more comprehensive understanding However, cutaneous responses were reduced commonly known as primary afferent depo- of the selective control of transmission of well before EMG activity onset, before move- larization or PAD, increases the excitability of sensory inputs to spinal interneurons. ment-related activity is normally observed in the axon. Under these circumstances, a small Because some dorsal spinal neurons proj- somatosensory cortex12,suggesting that the electrical stimulus in the vicinity of the depo- ect to the dorsal column nuclei7,presynaptic sparse motor cortical projection to dorsal larized axon will be more likely to elicit an inhibition of cutaneous input of spinal neu- spinal regions may also modulate sensory antidromic action potential. Despite formi- rons likely contributes to the well-docu- input at the spinal level13. dable technical problems, Seki et al.5 provide mented observation that perceptual By examining the activity patterns of persuasive evidence that the probability of thresholds of cutaneous input increase dur- motor neurons of eye muscles and working evoking antidromic potentials increases dur- ing self-induced movements8,9.Further, systematically backward through the oculo- ing the active phase of extension or flexion. cutaneous afferents that project directly to motor brainstem circuit that controls eye Thus, the data fulfill a critical criterion for the dorsal column nuclei are also influenced movements, Robinson and colleagues unrav- presynaptic inhibition as the mechanism by presynaptic inhibition from cortex10. eled the neural basis of oculomotor con- responsible for modulation of cutaneous sig- Therefore, presynaptic inhibition provides an trol14.Similar progress has not been attained nals on first-order interneurons. important mechanism to suppress cutaneous in the limb motor system, partially due to the What is the advantage of presynaptic inhi- input before it has any opportunity to influ- difficulty of recording the activity of neurons bition? Postsynaptic input would reduce the ence neural processing throughout the cen- in the spinal cord in awake, behaving ani- influence of cutaneous inputs, but would also tral nervous system. mals. Although limb motor control is inher- modulate all other inputs to the neuron. Feedback on the effects of motor per- ently more complex than oculomotor Presynaptic inhibition allows selective sup- formance can only be provided by afferents, control, the techniques pioneered by Seki and pression of a specific input to a neuron with- so why inhibit this feedback signal? Seki colleagues5 to record neural activity in the out influencing other synaptic inputs. Thus, et al.5 observed that intraspinal stimulation spinal cord of awake, behaving non-human

1244 VOLUME 6 | NUMBER 12 | DECEMBER 2003 NATURE NEUROSCIENCE NEWS AND VIEWS

primates will continue to lead to important 2. Rudomin, P. & Schmidt, R.F. Exp. Brain Res. 129, 9. Williams S.R., Shenasa, J. & Chapman, C.E. advances in understanding spinal function 1–37 (1999). J. Neurophysiol. 79, 947–963 (1998). 3. Hultborn, H., Meunier, S., Pierrot-Deseilligny, E. & 10. Andersen, P., Eccles, J.C., Schmidt, R.F. & Yokota, T. during volitional motor control. Perhaps it is Shindo, M. J. Physiol. (Lond.) 389, 757–772 (1987). J. Neurophysiol. 27, 92–106 (1964). time to acquaint fledgling neuroscientists 4. Capaday, C. & Stein, R.B. J. Neurosci. 6, 11. Scott, S.H. Curr. Opin. Neurobiol. (in press). with the intricacies of presynaptic inhibition 1308–1313 (1986). 12. Kalaska, J.F. & Crammond, D.J. Science 255, 5. Seki, K., Perlmutter, S.I. & Fetz, E.E. Nat. Neurosci. 1517–1523 (1992). and other features of spinal processing. We 6, 1309–1316 (2003). 13. Bortoff, G.A. & Strick P.L. J. Neurosci. 13, hope that the editors of introductory texts in 6. Jankowska, E., Slawinska, U. & Hammar, I. 5105–5118 (1993). neuroscience are listening. J. Physiol. (Lond.) 542, 287–299 (2002). 14. Robinson, D.A. in Basic Mechanisms of Ocular 7. Rustioni, A. Science 196, 656–658 (1977). Motility and Their Clinical Implications (eds. 1. Miller, R.J. Annu. Rev. Pharmacol. Toxicol. 38, 8. Shergill, S.S., Bays, P.M., Frith, C.D. & Wolpert, D.M. Lennerstrand, G. & Bach-y-Rita, P.) 337–374 201–227 (1998). Science 301, 187 (2003). (Pergamon Press, Oxford, 1975).

Doublecortin finds its place

Magdalena Götz

Confusing results from gene deletion experiments have left the importance of doublecortin (DCX) during brain development unclear. A report in this issue establishes a definitive function for DCX and highlights limitations of gene knockout approaches.

Migration is a daunting task. Billions of neu- of a phenotype, or possibly to a phenotype with rodent Dcx mRNA. Within one day, http://www.nature.com/natureneuroscience rons in the developing brain must leave their unrelated to the gene of interest. Deleting a DCX protein expression was reduced to place of birth and travel to precise locations gene throughout the entire life span of an 20% of the levels found in cells transfected to form the circuits of a mature functioning organism can also give different results than with control RNAs. Despite the rapid nervous system. Frequent impairment of altering the gene during a certain critical time decrease, the initial migration of affected this complex process is perhaps not surpris- period. Therefore, to examine the function of neurons appeared normal. However, several ing, and migration disorders are among the DCX, which interacts with the cytoskeleton days later, most cells were stuck within the most common causes of developmental neu- and is expressed specifically in vertebrates by intermediate zone, the future rological defects1. One such defect leads to young neurons of the central and peripheral (Fig. 2a). There they remained, eventually the so-called ‘double cortex’ (or subcortical nervous system7,8,Bai et al.4 took advantage of forming the characteristic band of ectopic band heterotopia), in which neurons accu- recent advances in RNA interference (RNAi), a neurons seen in adult animals. The similar- mulate inappropriately within the white technique that allows acute targeting and ity of this phenotype to the human double matter beneath the normal cortical layers disruption of a specific RNA, while leaving cortex condition is striking, in particular (Fig. 1). A mutation in the doublecortin gene genes untouched9. given the large number of phenotypes asso- © Group 2003 Nature Publishing (DCX in humans)2 seems to be the cause, The authors delivered a short hairpin ciated with neuronal migration disorders1. but oddly, mice with a targeted deletion of RNA to rats in utero to specifically interfere Admittedly, the RNA interference technique this gene (Dcx) have an apparently normal neocortex3.It might have been a simple species difference—molecules indispensable Control DCX-RNAi for the long distances traveled by migrating I neurons in the human cortex may not be

Erin Boyle II needed for the shorter distances in a rodent brain. However, Bai et al.4 report in this issue III that DCX protein is indeed important for IV normal cortical development in rodents and provide a cautionary note for more general V gene-knockout approaches. Targeted deletions used in gene-knockout VI studies are often difficult to interpret because the insertion of foreign DNA into a genome White matter can have unexpected effects on neighboring genes5,6, leading to compensation and absence SVZ

Magdalena Götz is at the Max-Planck Institute Figure 1 The ‘double cortex’ mutation. Bai et al.4 used a short hairpin RNA to interfere with of Neurobiology, Am Klopferspitz 18A, doublecortin expression in embryonic rats and found that by adulthood, affected cells accumulated in Planegg-Martinsried, D-82152, Germany. the white matter, or were scattered among the cortical layers compared to their normal positions in the email: [email protected] cortex of control animals.

NATURE NEUROSCIENCE VOLUME 6 | NUMBER 12 | DECEMBER 2003 1245