Shaping Early Networks to Rule Mature Circuits: Little MiRs Go a Long Way

Neuron

Previews Armcx1 likely recruits stationary mitochondria into a motile pool. Armcx1 overexpression also triggered a substantial increase in axons growing out of the explants, an effect dependent on its mitochondrial localization. Furthermore, Armcx1 overexpression increased the amount of motile mitochondria in axons and neurite length of embryonic cortical neurons. Consistently, knockdown of Armcx1 decreased the motile pool of mitochondria in axons and reduced neurite growth. This dual role of Armcx1 promoting mitochondrial transport and neurite outgrowth also applied to adult RGCs after an optic nerve crush injury. Armcx1 overexpression induced a higher number of regenerating axons after injury and promoted neuronal survival, suggesting that Armcx1 may not only promote axon regeneration but also protect injured neurons from cell death. The positive effect of Armcx1 on axon regeneration was not due to the limited regenerative ability of wild-type RGCs since Armcx1 overexpression also increased the regeneration capacity of axons in comparison to PTEN mutant RGCs. Remarkably, the high regeneration capacity of PTEN-deleted RGCs overex-

pressing Armcx1 approached levels seen for dKO RGCs. Moreover, Armcx1 overexpression in PTEN mutant RGCs primarily increased regeneration of axons from non-aRGCs, even though PTEN inhibition selectively promotes regeneration from aRGCs. Does upregulation of Armcx1 in RGCs of dKO mice underlie their high regeneration capacity? To test this, Cartoni et al. knocked down Armcx1 in dKO mice and found that the ability of high axon regeneration is inhibited and cell survival is reduced. Accordingly, injury-induced elevation of Armcx1 levels is critical for increased neuronal survival and axon regeneration of the dKO model. Taken together, Han et al. and Cartoni et al. demonstrate a critical role of mitochondrial transport for axon regeneration and provide valuable insights into intrinsic molecular mechanisms facilitating the upregulation of mitochondrial transport after injury and the regeneration capacity of neurons (Figure 1). As such, these insights may provide an emerging foundation for designing efficient neural repair strategies. Future work will be needed to explore how DLK-1 and Armcx1 mobilize mitochondrial transport into axons following neuronal injury.

REFERENCES Bradke, F., Fawcett, J.W., and Spira, M.E. (2012). Nat. Rev. Neurosci. 13, 183–193. Cartoni, R., Norsworthy, M.W., Bei, F., Wang, C., Li, S., Zhang, Y., Gabel, C.V., Schwarz, T.L., and He, Z. (2016). Neuron 92, this issue, 1294–1307. Han, S.M., Baig, H.S., and Hammarland, M. (2016). Neuron 92, this issue, 1308–1323. He, Z., and Jin, Y. (2016). Neuron 90, 437–451. Holt, C.E., and Schuman, E.M. (2013). Neuron 80, 648–657. Mar, F.M., Simo˜es, A.R., Leite, S., Morgado, M.M., Santos, T.E., Rodrigo, I.S., Teixeira, C.A., Misgeld, T., and Sousa, M.M. (2014). J. Neurosci. 34, 5965– 5970. Rawson, R.L., Yam, L., Weimer, R.M., Bend, E.G., Hartwieg, E., Horvitz, H.R., Clark, S.G., and Jorgensen, E.M. (2014). Curr. Biol. 24, 760–765. Rishal, I., and Fainzilber, M. (2014). Nat. Rev. Neurosci. 15, 32–42. Schwarz, T.L. (2013). Cold Spring Harb. Perspect. Biol. 5, http://dx.doi.org/10.1101/cshperspect. a011304. Sun, F., Park, K.K., Belin, S., Wang, D., Lu, T., Chen, G., Zhang, K., Yeung, C., Feng, G., Yankner, B.A., and He, Z. (2011). Nature 480, 372–375. Zhou, B., Yu, P., Lin, M.Y., Sun, T., Chen, Y., and Sheng, Z.H. (2016). J. Cell Biol. 214, 103–119.

Shaping Early Networks to Rule Mature Circuits: Little MiRs Go a Long Way Andre Marques-Smith,1,2,3 Emilia Favuzzi,1,2,3 and Beatriz Rico1,2,* 1Centre

for Developmental Neurobiology, Institute of Psychiatry, Psychology and Neuroscience Centre for Neurodevelopmental Disorders King’s College London, London, SE1 1UL, UK 3Co-first author *Correspondence: [email protected] http://dx.doi.org/10.1016/j.neuron.2016.12.014 2MRC

Normative cortical processing depends on precise interactions between excitatory and inhibitory neurons. In this issue of Neuron, Lippi et al. (2016) identify miR-101 as a master regulator coordinating molecular programs during development that ultimately impact the activity of mature networks. Neural computation relies on the precise organization of synaptic connections among different neuronal subtypes. Inter-

actions between excitatory pyramidal neurons and inhibitory GABAergic interneurons are particularly important, as

1154 Neuron 92, December 21, 2016 ª 2016 Published by Elsevier Inc.

neuronal circuits can only operate effectively within certain bounds of excitation and inhibition (Isaacson and Scanziani,

Neuron

Previews 2011). This is critical not only for the information processing that supports animal behavior but also because overstepping these boundaries can lead to neurodevelopmental and neurological disorders, including autism, schizophrenia, and epilepsy (Paz and Huguenard 2015; Marı´n 2016). During brain development a plethora of turbulent events will frame mature neural circuits: endogenous spontaneous rhythms give way to sensory-driven activity, GABA switches polarity, canonical circuits are formed, potentiated, and refined, and eventually synapses elevate their threshold for plasticity, narrowing integration windows to become fast, precise reporters of spiking activity. Each of these processes is regulated by dynamic programs of gene expression, which are tuned by neural activity in a bidirectional manner. What could quickly become a neural cacophony actually plays out as a beautifully orchestrated symphony; transcriptional programs regulate expression of ion channels, neurotransmitter receptors, and transporters, restraining patterns of network activity and controlling the transition between them. The intimate association of several such developmental processes—e.g., dendritic arbor elaboration and synapse formation—and the need to concertedly switch transcription on or off for different genes requires centralized regulation of gene cohorts to effect on-going neural genetic programs. MicroRNAs (miRs) are small non-coding RNAs that function as post-transcriptional regulators of gene expression holding the ability to simultaneously regulate multiple genes in the context of complex regulatory networks (McNeill and Van Vactor, 2012). miRs provide mechanisms of regulation that are fast, flexible, and reversible and as such, well-suited for the complexities of neural circuit wiring. They appear thus as ideal candidates to tightly regulate and tune developmental gene programs during the assembly of neuronal circuits. In a series of elegant experiments, Lippi et al. (2016) discover that microRNA 101 (miR-101) synergistically regulates expression of several genes for the common goal of constraining excitation in hippocampal circuits. Lippi et al. (2016) carried out a thorough screening to identify sequenced miRs in the developing hippocampus at

postnatal day 12 (P12), a critical developmental window, curating a list of candidates well suited for neural developmental processes. Then, based on: (1) abundance, (2) upregulation during development, (3) enrichment in Argonaute complexes (Ago, effector of miR function), and (4) published targets of miRs involved in neural differentiation, they identified miR-101. Transient (P2–P9) and localized inhibition of miR-101 resulted in a lasting adult phenotype characterized overall by hyper-excitability. Adult hippocampal pyramidal neurons displayed increased firing rates in vivo, as well as elevated frequency and amplitude of spontaneous excitatory post-synaptic currents (sEPSCs) in vitro. Calcium imaging experiments revealed higher proportions of active neurons at any one time, as well as an overall increase in frequency of calcium (putative spiking) events. By using behavioral tests that depend on hippocampal function, Lippi et al. (2016) showed that blocking miR-101 in early postnatal life—but not in adult—led to lasting deficits in context-dependent associative memory, spatial working memory, and spatial episodic-like memory. These findings are particularly relevant for neurodevelopmental disorders, as they link the transient early inhibition of miR-101 to impaired cognitive function in the adult. To identify the mechanism by which miR-101 regulates the establishment of a balanced network, they searched for miR-101 targets. Using a combination of in vitro and in vivo approaches, Lippi et al. (2016) revealed several candidates, including the sodium-potassium-chloride co-transporter 1 (NKCC1). Across multiple brain regions, downregulation of this chloride importer and upregulation of the chloride exporter KCC2 underlies the developmental shift in chloride reversal potential and consequent maturation of GABAergic signaling from depolarizing to hyperpolarizing (Ben-Ari, 2002). Indeed, blocking miR-101 in vivo resulted in increased NKCC1 expression by release from miR101 repression and a relatively depolarized EGABA at P8. In contrast, KCC2 expression was unchanged, suggesting that a distinct developmental genetic program regulates KCC2 levels. By disrupting miR101-NKCC1 interaction without affecting other miR-101 targets, Lippi et al. (2016)

elegantly demonstrate that miR-101 regulation of NKCC1 mRNA alone was responsible for the delayed maturation of the GABA reversal potential. Giant depolarizing potentials (GDPs) synchronize activity and promote synaptic plasticity between pyramidal neurons (Alle`ne et al., 2008). Furthermore, early GABAergic activity is required for dendritic elaboration (Cancedda et al., 2007). Therefore, a sustained depolarizing action of GABA in miR-101 blocking experiments could affect both synaptic stabilization and dendritic development leading to an exuberant excitatory network. However, specific de-repression of NKCC1 without affecting other miR101 in vivo only explained the increased rate of synchronous calcium events and a modest elevation in miniature EPSC frequency in vitro, causing no discernible effect on overall rate of calcium, proportion of active ensembles, or double synchronized events, hallmark features of the miR-101 phenotype. Given the partial phenotype of prolonged NKCC1 expression, Lippi et al. (2016) hypothesized that the effect of miR-101 inhibition was achieved through multi-level targeting of several genes within a biological network. They explored this possibility by combining the top targets for miR-101 into groups, according to their known developmental ontology effects (‘‘Pre-synaptic,’’ ‘‘Glial,’’ and ‘‘Excitability’’). In addition to NKCC1, the ‘‘Presynaptic’’ group included two genes involved in the formation and stabilization of presynaptic inputs, Ank2 and Kif1a. Lippi et al. (2016) elegantly dissected the contribution of these genes, finding that NKCC1 targeting by miR-101 limits dendritic length while complementary repression of Kif1a and Ank2 is required to restrict excitatory synaptic density. As a result, continued expression of NKCC1 and the genes in the ‘‘Pre-synaptic’’ group mimicked the increased levels of activity, mainly because of the occurrence of more asynchronous calcium events. Next, de-repression of NKCC1 and two genes, the cholesterol transporter Abca1 and the hydrolase Ndrg2 (‘‘Glial’’ group), enriched in glial cells with a role in neurite growth, was responsible for the increase in the size of cell ensembles recruited in each synchronous event. Releasing the expression of genes involved in regulating neuronal excitability (‘‘Excitability’’ Neuron 92, December 21, 2016 1155

Neuron

Previews

group), along with NKCC1, tors, operate at the individual increased the number of cell level? The simplest hydouble events. Thus, each pothesis is that genes regugroup of genes accounted for lating E/I balance are responunique aspects of the multisive to neural activity. miRs tier regulatory control of miRhave indeed been previously 101 and together they dictate linked to activity (McNeill and the precise code for a Van Vactor, 2012). Could balanced development of neumiR-101, for instance, sense ral circuits (Figure 1). chronic increases in excitAlthough previous studies atory activity and increase have proposed that miRs repression of its downstream function in shaping the targets? neuronal landscape (see Additional work will be McNeill and Van Vactor, needed to examine whether 2012), the work of Lippi et al. miR-101 plays a similar role (2016) constitutes the first in other brain regions such demonstration that simultaas the neocortex. This will neous regulation of multiple help to determine whether target genes by a single miR the regulatory developmental during a critical developprogram described by Lippi mental window orchestrates et al. (2016) represents a genconvergent molecular proeral mechanism to constrain grams that ultimately sculpt a excitation in the brain. This stable mature neuronal is particularly relevant since network (Figure 1). Also, it is neural circuits show exquisite Figure 1. miR-101 Regulates the Development of Neural Circuits important to emphasize that fine structure, with spatially Shaping Mature Networks in Adult this study has been carried proximal cells often particiout using in vivo models pating in completely different where the cellular context is intact, gests that subtler forms of compensation microcircuits and subnetworks (Lee demonstrating a more physiological func- occurred and went undetected, prevent- et al., 2014). These channels of information of the miR. In sum, Lippi et al. (2016) ing the emergence of pathology. An tion may not have the same ratio of excitareveal here a set of interesting results attractive possibility is that miR-101 itself tion and inhibition and may differentially with implications not only for miR biology regulates inhibitory synapse formation impact neural function. Could miRs help and function, but also for the regulation of while it constrains excitation, and sculpt an additional level of circuit-speciexcitatory-inhibitory balance and neuro- blockade of its action prevented emer- ficity, beyond cell-type rules of innervadevelopmental processes. gence of an inhibitory compensatory tion? Some data in Lippi et al. (2016)’s It remains unknown why the long-last- response. It will be interesting to deter- work hints at pathway-specific regulation, ing effects caused by early transient mine the role of miR-101 in different types e.g., discrepant effect of NKCC1 demiR-101 blockage were not compen- of GABAergic cells. Inhibition synchro- repression on the secondary branches in sated homeostatically. It is well docu- nizes and sharpens excitatory responses CA1 and in CA3 or over-representation mented that neurons and networks are in many brain areas, and its impairment of mossy fiber input. It would be of highly reactive to, and capable of could increase ‘‘noise’’ in learning and great interest to extend these observacompensating for, changes in their excit- cognition, partially accounting for some tions with pathway and cell-type-specific atory-inhibitory environment (Xue et al., of the observed cognitive effects of miR- methods. Because of its ability to regulate multi2014). It is surprising therefore that the 101 inhibition described by Lippi et al. ple key aspects of brain development, it hippocampal network did not respond (2016). to unfettered excitation through release What determines the changes in the is not surprising that miR-101 has a role from miR-101 regulation by increasing in- expression of miR-101 in the first place? in many neuropsychiatric disorders (Lippi hibition. Indeed, increases in excitation Is it the result of specific activity patterns et al., 2016, Figure 1). Interestingly, occurred in the absence of proportional or is it intrinsically determined? Pyramidal the prevalent view in the field is that changes in inhibitory currents, suggesting neurons receive inhibition in proportion to although development is a continuous the presence of exuberant excitatory their afferent synaptic excitation levels, process, there are specific sensitive wincircuits rather than dis-inhibition. This is meaning E/I balances across cells are dows—‘‘critical periods’’—in which modparticularly intriguing since miR-101 is stable even though afferent excitation ifications in network organization have also expressed in interneurons. Interest- levels differ widely (Xue et al., 2014). long-lasting impact over the lifespan ingly, the lack of epileptiform activity in How does the genetic regulation of E/I (Marı´n, 2016). These sensitive periods such an excitable network in itself sug- balance, through miR-101 and other ac- are pivotal milestones for the assembly 1156 Neuron 92, December 21, 2016

Neuron

Previews of excitatory and inhibitory circuits. Therefore, understanding how the relative bounds of excitation and inhibition are developmentally established, maintained, and shifted is an exciting topic of research, increasingly attracting interest in Neuroscience. Indeed, unveiling the main regulators of these processes might be key for early interventions to restore normal brain function (Marı´n, 2016). Future work uncovering the function of miRs in neural circuit development promises to shed light on potential therapeutic targets for neurodevelopmental disorders.

ACKNOWLEDGMENTS Work in B.R. laboratory is supported by the European Research Council (ERC-2012-StG 310021). B.R. is Wellcome Trust Investigator. REFERENCES Alle`ne, C., Cattani, A., Ackman, J.B., Bonifazi, P., Aniksztejn, L., Ben-Ari, Y., and Cossart, R. (2008). J. Neurosci. 28, 12851–12863. Ben-Ari, Y. (2002). Nat. Rev. Neurosci. 3, 728–739.

Lee, S.H., Marchionni, I., Bezaire, M., Varga, C., Danielson, N., Lovett-Barron, M., Losonczy, A., and Soltesz, I. (2014). Neuron 82, 1129–1144. Lippi, G., Fernandes, C.C., Ewell, L.A., John, D., Romoli, B., Curia, G., Taylor, S.R., Frady, E.P., Jensen, A.B., Liu, J.C., et al. (2016). Neuron 92, this issue, 1337–1351. Marı´n, O. (2016). Nat. Med. 22, 1229–1238. McNeill, E., and Van Vactor, D. (2012). Neuron 75, 363–379.

Cancedda, L., Fiumelli, H., Chen, K., and Poo, M.M. (2007). J. Neurosci. 27, 5224–5235.

Paz, J.T., and Huguenard, J.R. (2015). Nat. Neurosci. 18, 351–359.

Isaacson, J.S., and Scanziani, M. (2011). Neuron 72, 231–243.

Xue, M., Atallah, B.V., and Scanziani, M. (2014). Nature 511, 596–600.

Healing Pains of the Past Using Neuronal Transplantation XiaoTing Zheng1 and Sunil P. Gandhi1,2,* 1Department

of Neurobiology and Behavior, University of California, Irvine, CA 92697, USA for Neurobiology of Learning and Memory, University of California, Irvine, CA 92697, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.neuron.2016.12.012 2Center

Yang et al. (2016) show that transplantation of GABAergic inhibitory neurons into the amygdala boosts the persistence of fear extinction in mice. Transplantation was found to degrade perineuronal nets on endogenous inhibitory neurons and enhance synaptic plasticity in host amygdala. In the movie ‘‘Inside Out,’’ Riley starts out as a bubbly pre-teen whose psyche is powered by joyful core memories. Traumatic experiences, however, threaten to recolor Riley’s core memories and destabilize her personality. Luckily, Riley overcomes her hard experiences and regains a healthy emotional balance to her core memories. What if we could neutralize traumatic memories in adulthood by bringing back the mutability of juvenile emotional memories? In this issue of Neuron, Yang et al. (2016) use inhibitory neuron transplantation to erase fear memories in adult mice as if they were juveniles once again. The transplantation of GABAergic inhibitory neurons has emerged as a powerful new approach to treat a diverse array of brain disorders in large part because these precursors disperse and

integrate widely in the brain (Chohan and Moore, 2016). Transplantation of inhibitory neurons into the hippocampus has been shown to suppress seizures and alleviate anxiety (Hunt et al., 2013). Inhibitory neuron transplantation into adult visual cortex restores spatial vision to amblyopic mice that were visually deprived in early childhood (Davis et al., 2015). This technique has also been shown to ameliorate rodent models of schizophrenia, Alzheimer’s, Parkinson’s, and neuropathic pain (Chohan and Moore, 2016). Now we can add to the list enhanced fear extinction. Yang et al. (2016) harvested inhibitory progenitor cells from the medial ganglionic eminence (MGE) of embryonic mice and transplanted them into the young adult amygdala (weeks 5–7). These progenitors dispersed in the basolateral and central

amygdala and survived but had no effect on the density of host inhibitory neurons. The transplanted cells developed markers for mature inhibitory neuron cell types, established normal excitability, and integrated synaptically into the host circuit. Yang and colleagues next used a classic paradigm for assessing fear memories. Mice were conditioned to fear a neutral stimulus (CS) using a foot shock (US). After the fear conditioning, mice went through two days of exposure to the CS in a new context without the foot shock causing the temporary extinction of the fear. After a week, fear for the CS returns when the normal young adult mice are re-exposed in both the extinction and conditioning contexts (Figure 1A). To determine whether MGE transplantation could facilitate fear erasure in adult mice, Yang et al. (2016) fear conditioned the animals at

Neuron 92, December 21, 2016 ª 2016 Elsevier Inc. 1157