Portal de Periódicos da CAPES

Paint It Black (or Red, or Green): Optical and Genetic Tools Illuminate Inhibitory Contributions to Cortical Circuit Function

2010; Cell Press; Volume: 67; Issue: 5 Linguagem: Inglês

10.1016/j.neuron.2010.08.039

1097-4199

Andrea R. Hasenstaub, Edward M. Callaway,

Photoreceptor and optogenetics research

Visual responses in the cortex are generated by the interactions of feedforward thalamic inputs with local inhibitory and excitatory circuitry, but the exact roles of different cell types in establishing response selectivity are unclear. Two papers in this issue of Neuron (Kerlin et al. and Runyan et al.) combine two-photon imaging with guided electrical recordings to measure orientation tuning in molecularly defined groups of interneuron types. Visual responses in the cortex are generated by the interactions of feedforward thalamic inputs with local inhibitory and excitatory circuitry, but the exact roles of different cell types in establishing response selectivity are unclear. Two papers in this issue of Neuron (Kerlin et al. and Runyan et al.) combine two-photon imaging with guided electrical recordings to measure orientation tuning in molecularly defined groups of interneuron types. Since the discovery of orientation tuning in the visual cortex by Hubel and Wiesel more than 50 years ago, there has been a nearly constant inquiry into how inhibitory and excitatory cortical circuits shape the untuned inputs that arrive from the lateral geniculate nucleus of the thalamus. Recent advances in optical and genetic techniques are allowing us to begin answering this question with unprecedented throughput and precision—quite literally shedding light on the operation of cortical circuits. Cell type targeted expression of fluorescent protein, combined with two-photon (2P) imaging of intracellular calcium dynamics (Stosiek et al., 2003Stosiek C. Garaschuk O. Holthoff K. Konnerth A. Proc. Natl. Acad. Sci. USA. 2003; 100: 7319-7324Crossref PubMed Scopus (979) Google Scholar), promises to let us simultaneously examine response properties in an unbiased sample of tens or hundreds of molecularly classifiable neurons. Conversely, 2P-guided electrical recordings from fluorescent neurons (Margrie et al., 2003Margrie T.W. Meyer A.H. Caputi A. Monyer H. Hasan M.T. Schaefer A.T. Denk W. Brecht M. Neuron. 2003; 39: 911-918Abstract Full Text Full Text PDF PubMed Scopus (164) Google Scholar) may facilitate selective sampling from specific cell types of interest. Two papers in this issue of Neuron, by Kerlin and Andermann et al. (Kerlin et al., 2010Kerlin A.M. Andermann M.L. Berezovskii V.K. Reid R.C. Neuron. 2010; 67 (this issue): 858-871Abstract Full Text Full Text PDF PubMed Scopus (353) Google Scholar) and Runyan, Schummers, and Van Wart et al., (Runyan et al., 2010Runyan C.A. Schummers J. Van Wart A. Kuhlman S.J. Wilson N.R. Huang Z.J. Sur M. Neuron. 2010; 67 (this issue): 847-857Abstract Full Text Full Text PDF PubMed Scopus (171) Google Scholar) use these complementary approaches to investigate the tuning of different types of cortical inhibitory interneuron. Although the papers reach different conclusions regarding the strength and prevalence of orientation tuning among inhibitory neurons, the apparent contradiction in their results may arise from differences in the types of neurons that the two groups sampled. This implies that these disparate findings may be more compatible than they initially appear. Cortical inhibitory neurons do not make long-range connections, but are instead involved strictly in local processing or computation. It is thus of interest to know whether and how inhibition provided by local interneurons might shape the orientation tuning of excitatory projection neurons, which do send outputs to the higher cortical areas more directly involved in generation of visual behaviors. There has been little consensus in the literature regarding this question. Until recently, most studies have focused on the cat visual cortex, in which the inhibitory currents converging on excitatory neurons may or may not be tuned for the orientation observed in their spiking output (Mariño et al., 2005Mariño J. Schummers J. Lyon D.C. Schwabe L. Beck O. Wiesing P. Obermayer K. Sur M. Nat. Neurosci. 2005; 8: 194-201Crossref PubMed Scopus (218) Google Scholar, Monier et al., 2003Monier C. Chavane F. Baudot P. Graham L.J. Frégnac Y. Neuron. 2003; 37: 663-680Abstract Full Text Full Text PDF PubMed Scopus (288) Google Scholar, Priebe and Ferster, 2005Priebe N.J. Ferster D. Neuron. 2005; 45: 133-145Abstract Full Text Full Text PDF PubMed Scopus (186) Google Scholar), and inhibitory neurons themselves may or may not be tuned for orientation (Cardin et al., 2007Cardin J.A. Palmer L.A. Contreras D. J. Neurosci. 2007; 27: 10333-10344Crossref PubMed Scopus (127) Google Scholar, Hirsch et al., 2003Hirsch J.A. Martinez L.M. Pillai C. Alonso J.M. Wang Q.B. Sommer F.T. Nat. Neurosci. 2003; 6: 1300-1308Crossref PubMed Scopus (139) Google Scholar, Nowak et al., 2008Nowak L.G. Sanchez-Vives M.V. McCormick D.A. Cereb. Cortex. 2008; 18: 1058-1078Crossref PubMed Scopus (72) Google Scholar), perhaps depending on the cortical layer in which they reside (Martinez et al., 2005Martinez L.M. Wang Q.B. Reid R.C. Pillai C. Alonso J.M. Sommer F.T. Hirsch J.A. Nat. Neurosci. 2005; 8: 372-379Crossref PubMed Scopus (143) Google Scholar) or their position relative to other orientation-tuned neurons (Schummers et al., 2002Schummers J. Mariño J. Sur M. Neuron. 2002; 36: 969-978Abstract Full Text Full Text PDF PubMed Scopus (109) Google Scholar). More recent studies taking advantage of genetic targeting in mice have found little evidence for orientation tuning of most inhibitory neurons (Liu et al., 2009Liu B.H. Li P.Y. Li Y.T. Sun Y.J.J. Yanagawa Y. Obata K. Zhang L.I. Tao H.W. J. Neurosci. 2009; 29: 10520-10532Crossref PubMed Scopus (109) Google Scholar, Sohya et al., 2007Sohya K. Kameyama K. Yanagawa Y. Obata K. Tsumoto T. J. Neurosci. 2007; 27: 2145-2149Crossref PubMed Scopus (172) Google Scholar), although a small fraction may be well tuned (Niell and Stryker, 2008Niell C.M. Stryker M.P. J. Neurosci. 2008; 28: 7520-7536Crossref PubMed Scopus (588) Google Scholar). How can these discrepancies, both within and between species, be reconciled? One possibility is that cats and mice might be fundamentally different. More broadly, these studies are complicated by the bewildering diversity of cortical inhibitory circuit elements. Different types of inhibitory interneurons appear to be specialized for targeting the apical tuft, shaft, basal dendrites, soma, or axon initial segment of pyramidal neurons, as well as for inhibiting one another in specific circuits (reviewed in Markram et al., 2004Markram H. Toledo-Rodriguez M. Wang Y. Gupta A. Silberberg G. Wu C. Nat. Rev. Neurosci. 2004; 5: 793-807Crossref PubMed Scopus (1943) Google Scholar). These anatomical differences are related to differences in expression of peptides, calcium binding proteins, and ion channel genes, although the mapping is not one-to-one. For instance, the majority of interneurons expressing the calcium binding protein parvalbumin (PV) are the soma-targeting, fast-firing “basket” neurons identifiable in intracellular and extracellular recordings. However, parvalbumin is also expressed by axon-targeting, fast-firing “chandelier” cells, as well as by a rarer population of interneuron- and shaft-targeting, burst-firing “multipolar-bursting” cells (Blatow et al., 2003Blatow M. Rozov A. Katona I. Hormuzdi S.G. Meyer A.H. Whittington M.A. Caputi A. Monyer H. Neuron. 2003; 38: 805-817Abstract Full Text Full Text PDF PubMed Scopus (236) Google Scholar). The peptides somatostatin (SOM) and vasoactive intestinal peptide (VIP), similarly, each label multiple (although nonoverlapping) inhibitory neuron types. Do these cell types play different roles in shaping visual responses? Perhaps some are important for generating orientation tuning, while others are instead important for other aspects of cortical processing, such as regulating response gain or global network state. Alternatively, inhibition might produce orientation tuning, but through mechanisms that do not require the inhibition itself to be explicitly tuned. Definitive answers to these questions have been hampered by the fact that many of the key molecular and anatomical features that distinguish among these interneuron types cannot be detected solely from electrical recordings of their activity. Further, many of these types are rare and unlikely to be encountered frequently in untargeted recordings. The two papers in this issue of Neuron tackle these difficulties by using transgenic mice and optical methods to sample visual responses from better-defined subsets of inhibitory interneurons. The first study comes from the laboratory of Clay Reid, where pioneering work using two-photon calcium imaging in the visual cortex has led to several seminal papers on cortical functional micro-organization. This paper introduces an impressive new technological achievement: the ability to identify, in postmortem tissue, the same neurons that were functionally assayed in vivo. This allows imaged neurons to be further classified by antibody staining to determine their molecular identity (their Figure 2). The authors performed 2P calcium imaging to determine the response properties of nearly all the cells in a cortical volume. These studies were performed in the Gad67-GFP mouse line (Tamamaki et al., 2003Tamamaki N. Yanagawa Y. Tomioka R. Miyazaki J.I. Obata K. Kaneko T. J. Comp. Neurol. 2003; 467: 60-79Crossref PubMed Scopus (949) Google Scholar), in which GFP is expressed in nearly all GABAergic cells, allowing the authors to use this GFP expression to comprehensively identify inhibitory neurons. Subsequent antibody staining for the marker proteins PV, SOM, and VIP allowed for the partial classification of these inhibitory neurons into established nonoverlapping cell type groups. The authors found that on average, compared to excitatory neurons, inhibitory neurons were poorly tuned for orientation and responded to a broader range of spatial frequencies. These observations are consistent with previous findings of relatively poor tuning of inhibitory neurons in the mouse primary visual cortex (Liu et al., 2009Liu B.H. Li P.Y. Li Y.T. Sun Y.J.J. Yanagawa Y. Obata K. Zhang L.I. Tao H.W. J. Neurosci. 2009; 29: 10520-10532Crossref PubMed Scopus (109) Google Scholar, Sohya et al., 2007Sohya K. Kameyama K. Yanagawa Y. Obata K. Tsumoto T. J. Neurosci. 2007; 27: 2145-2149Crossref PubMed Scopus (172) Google Scholar). The findings here extend those observations to identify not only PV cells (a class that likely includes the soma-targeting fast-spiking interneurons previously characterized through electrophysiological recordings [Niell and Stryker, 2008Niell C.M. Stryker M.P. J. Neurosci. 2008; 28: 7520-7536Crossref PubMed Scopus (588) Google Scholar] and which is further examined in Runyan et al. below) but also SOM and VIP cells, whose visual responses have not previously been characterized. Interestingly, although all three inhibitory cell groups were poorly tuned relative to excitatory neurons, there were some small but significant differences between these groups: VIP cells were somewhat better tuned for orientation, and SOM for direction, when compared to PV cells (their Figure S4). These differences were, on average, modest and may have resulted from the relatively sharp tuning observed in a small subset of the recorded cells (their Figure S4). This raises the possibility that the functional diversity within these groups may be linked to the diversity of the cell types within each group (Markram et al., 2004Markram H. Toledo-Rodriguez M. Wang Y. Gupta A. Silberberg G. Wu C. Nat. Rev. Neurosci. 2004; 5: 793-807Crossref PubMed Scopus (1943) Google Scholar), an issue which recurs in the context of the findings from the Runyan et al. paper. Noting the striking difference in orientation tuning of inhibitory neurons in mouse when compared to previously published results from cats, Kerlin and Andermann et al. postulated that this might be related to the lack of orientation columns in mice (Sohya et al., 2007Sohya K. Kameyama K. Yanagawa Y. Obata K. Tsumoto T. J. Neurosci. 2007; 27: 2145-2149Crossref PubMed Scopus (172) Google Scholar, Wang et al., 2006Wang K.H. Majewska A. Schummers J. Farley B. Hu C. Sur M. Tonegawa S. Cell. 2006; 126: 389-402Abstract Full Text Full Text PDF PubMed Scopus (176) Google Scholar). If inhibitory interneurons indiscriminately collect inputs from excitatory neurons in their immediate vicinity, then cat interneurons that are located within a distinct orientation column, where their neighbors are all similarly tuned, would collect input from a relatively uniform population and mirror that tuning. Since mice, unlike cats, lack orientation columns, a mouse interneuron collecting inputs indiscriminately from its neighbors will usually collect from a diverse population and therefore have poor orientation tuning (e.g., gray inhibitory neuron in Figure 1). A similar indiscriminate input hypothesis has been previously proposed to explain differences in the tuning of inhibitory input to excitatory neurons located in discrete orientation domains, versus those at pinwheel centers where neighboring neurons have a broad range of orientation preferences (Schummers et al., 2002Schummers J. Mariño J. Sur M. Neuron. 2002; 36: 969-978Abstract Full Text Full Text PDF PubMed Scopus (109) Google Scholar). To explore the hypothesis that this mechanism might be related to the poor tuning of inhibitory neurons in mouse, Kerlin and Andermann et al. compared the orientation tuning of GFP-labeled inhibitory neurons to that of the GFP-negative excitatory neurons imaged nearby. They found close correlations between the orientation tuning and spatial frequency tuning of individual inhibitory neurons and the average tuning of neighboring excitatory neurons (their Figures 6, S5, and S6). Such correlations were not observed between individual excitatory neurons and their neighbors, where the population average was often strikingly different from the individual. These observations are consistent with the possibility that, regardless of species, inhibitory neurons collect inputs from their neighbors without preference for orientation or spatial frequency tuning (Figure 1, gray oval), while excitatory neurons are more selective. The second paper comes from Runyan, Schummers, and Van Wart et al. in the laboratory of Mriganka Sur. This lab has previously published numerous important studies investigating the contributions of local cortical network activity in general, and inhibition in particular, to visual tuning in the cat. In contrast to Kerlin and Andermann et al.'s aim of unbiased sampling of all interneuron types, Runyan, Schummers, and Van Wart et al. aimed to selectively sample specific interneuron types—in this case, parvalbumin-positive neurons in the superficial layers—in order to examine in greater detail the response properties of this restricted set of neurons. The authors used a Cre-dependent virus in a Pv-Cre mouse line to drive fluorescent protein expression in a subset of these interneurons (their Figure 1) and used 2P imaging to guide juxtacellular recordings either to fluorescently labeled neurons or to randomly selected neurons in the surround. They found that 5 of the 11 PV neurons recorded had an orientation selectivity index (OSI) greater than 0.5, while the remaining cells were more poorly tuned (their Figure 2). This is a strikingly different result from previous electrophysiological studies, in which only a small minority of fast-spiking, presumed PV-positive cells had an OSI greater than 0.5 (Niell and Stryker, 2008Niell C.M. Stryker M.P. J. Neurosci. 2008; 28: 7520-7536Crossref PubMed Scopus (588) Google Scholar) and differs from the findings of Kerlin and Andermann et al., who found a relatively small fraction of inhibitory neurons with OSIs greater than 0.5 (their Figure S4). This difference suggests that Runyan et al. tapped into a particular subpopulation of PV neurons that is well tuned. Consistent with this hypothesis, they find a link between action potential waveform and OSI: the PV neurons with the poorest orientation tuning had the narrow, deep action potentials typical of fast-spiking interneurons, while the neurons with the best orientation tuning had broader or shallower action potentials (their Figure S1). This evidence suggests that the diversity of orientation tuning strengths they observe may be related to diversity of cell types rather than to position in the local network (Figure 1, red hexagon versus gray oval). Both groups addressed the important question of the extent to which calcium imaging accurately reflects the tuning of action potential responses. Runyan, Schummers, and Van Wart et al. report a correlation between orientation selectivity index and spiking response rate, with more selective cells having lower response rates (their Figure S4). This raises the possibility that calcium imaging might systematically undersample the lowest-firing, most selective neurons. Consistent with this possibility, their calcium-imaging measurements of excitatory neurons also indicate relatively poor orientation tuning (their Figure 3). However, this contrasts with the calcium-imaging data of Kerlin and Andermann et al.: in these data, the distribution of orientation selectivity measured with calcium imaging includes many highly tuned neurons (their Figures 5 and S4) and is closely matched to the distribution of orientation selectivity previously measured with electrical recordings (Niell and Stryker, 2008Niell C.M. Stryker M.P. J. Neurosci. 2008; 28: 7520-7536Crossref PubMed Scopus (588) Google Scholar). Furthermore, Kerlin and Andermann et al. directly relate individual neurons' action potentials to calcium changes by performing juxtacellular recordings on a subset of the imaged neurons (their Figure 3). They show a nearly linear relationship between spiking and calcium changes over a large range of firing rates, although the calcium signal may saturate at the highest firing rates. Their data also shows that, for each of the three excitatory and five inhibitory neurons tested, electrophysiology and calcium imaging yielded nearly identical measures of orientation tuning (their Figure 3). This indicates that under the conditions used by Kerlin and Andermann et al., calcium signals reliably measured the orientation tuning of cells' spiking responses, while Runyan, Schummers, and Van Wart et al. directly demonstrate that their calcium measures underestimate orientation tuning. While the source of this discrepancy is unknown, it does call for caution when interpreting calcium-imaging data. Both groups also explored the possibility that the Gad67-GFP mouse line (Tamamaki et al., 2003Tamamaki N. Yanagawa Y. Tomioka R. Miyazaki J.I. Obata K. Kaneko T. J. Comp. Neurol. 2003; 467: 60-79Crossref PubMed Scopus (949) Google Scholar), used both in the Kerlin and Andermann et al. study and in two previously published studies (Liu et al., 2009Liu B.H. Li P.Y. Li Y.T. Sun Y.J.J. Yanagawa Y. Obata K. Zhang L.I. Tao H.W. J. Neurosci. 2009; 29: 10520-10532Crossref PubMed Scopus (109) Google Scholar, Sohya et al., 2007Sohya K. Kameyama K. Yanagawa Y. Obata K. Tsumoto T. J. Neurosci. 2007; 27: 2145-2149Crossref PubMed Scopus (172) Google Scholar), might display developmental abnormalities resulting in poor orientation tuning of inhibitory neurons. Using juxtacellular recordings, Runyan, Schummers, and Van Wart et al. found that 0 out of 12 GFP-expressing inhibitory neurons in the Gad67-GFP mouse line had OSIs above 0.5 (their Figure S2). However, given that the Gad67-GFP mouse line labels diverse types of GABAergic interneurons (Tamamaki et al., 2003Tamamaki N. Yanagawa Y. Tomioka R. Miyazaki J.I. Obata K. Kaneko T. J. Comp. Neurol. 2003; 467: 60-79Crossref PubMed Scopus (949) Google Scholar), it is (as Runyan et al. note) perhaps not surprising that a relatively small sample of the entire population of inhibitory neurons would fail to contain members of a well-tuned but rare subset of PV neurons. Consistent with this possibility, Kerlin and Andermann et al. find no differences in the orientation tuning of PV+ GFP-expressing inhibitory neurons in the Gad67-GFP mouse line and PV+ neurons from wild-type mice (their Figure 5). This raises the question of what methodological differences might have allowed Runyan, Schummers, and Van Wart et al. to target PV-expressing inhibitory neurons that are clearly well tuned for orientation. The viral injections used to label neurons in the Runyan et al. paper label only a subset of PV neurons, whose exact identity likely depends on viral titer, injection depth, intrinsic tropism of the virus itself, and cell-cell differences in PV levels. In addition, recording depth may have been an important factor: Kerlin and Andermann et al. collected the bulk of their data from neurons located relatively deep in layers II and III (200–325 μm), while the targeted patch recordings of the Runyan et al. paper may have sampled more superficial neurons. These factors, particularly sample composition, may account for the systematic differences in the properties of PV neurons sampled in the two experiments. In particular, as Runyan et al. note, not all parvalbumin-positive cells are fast-spiking basket interneurons: parvalbumin is also present in an extremely superficial, relatively rare set of broad-spiking, multipolar bursting (MB) interneurons (Blatow et al., 2003Blatow M. Rozov A. Katona I. Hormuzdi S.G. Meyer A.H. Whittington M.A. Caputi A. Monyer H. Neuron. 2003; 38: 805-817Abstract Full Text Full Text PDF PubMed Scopus (236) Google Scholar). These PV/MB interneurons are seldom found in the deeper parts of layers II/III where Kerlin and Andermann et al. collected the bulk of their data, making it plausible that it is these broad-spiking PV/MB neurons that are well tuned. These discrepancies highlight that this is not a mature field: subtle differences in methodology may yield apparently contradictory results, both of which are correct. While these studies disagree on the relative proportions, identities, and mechanisms of cortical inhibitory neuron response tuning, both agree on the key point that while many inhibitory neurons are likely not well tuned, some likely are—and this work points us on the path to knowing their identities. But we should consider the counsel of former Defense Secretary Donald Rumsfeld: “As we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don't know we don't know.” (Defense.gov transcript, February 12, 2002). Even though the heterogeneity of interneurons is a known unknown, we must be cautious that the light of their fluorescent labeling does not blind us to their fundamental diversity. A comprehensive understanding of the roles of cortical inhibition will require tools that can subdivide interneurons into more selective and homogeneous populations and that extend beyond the superficial layers. Who knows how many unknown unknowns lurk still deeper in the cortex?