Song-Induced ZENK Gene Expression in Auditory Pathways of Songbird Brain and Its Relation to the Song Control System

The Journal of Neuroscience, November 1994, 14(11): 6652-6666

Song-induced ZENK Gene Expression in Auditory Pathways of Songbird Brain and Its Relation to the Song Control System

Claudio V. Mello’ and David F. Clayton2 ‘Laboratory of Animal Behavior, The Rockefeller University, New York, New York 10021 and *Department of Cell and Structural Biology, University of Illinois, Urbana, Illinois 61801

The ZENK gene encodes a zinc-finger-containing transcrip- Marler and Peters, 1977; Welty and Baptista, 1988; Godard, tional regulator and can be rapidly activated in songbird brain 1991). Considerableprogress has been achieved in defining the by presentation of birdsong (Mello et al., 1992). Here we map neural substrateof normal singing behavior by usinga combi- the areas of the songbird forebrain that show this genomic nation of brain lesions,electrophysiological recordings and an- response to birdsong, using in situ hybridization. After 30 atomical tract-tracing techniques.A seriesof distinct intercon- min of song presentation ZENK mRNA levels reach a peak nected brain nuclei is necessaryfor normal song production in the caudomedial telencephalon, in areas adjacent to or (Nottebohm et al., 1976, 1982); theseinclude a neostriatal nu- closely related with primary auditory structures. These areas cleus,HVC (high vocal center), an archistriatal nucleusto which include subfields of field L (Ll and L3), the caudomedial HVC projects (RA or robustus archistriatalis), and RA’s pro- neostriatum (NCM), the caudomedial hyperstriatum ventrale jections onto the motoneurons of the XII cranial nerve that (CMHV) anterior to field L, the caudal paleostriatum, and two innervate the syrinx, the vocal organ of songbirds.An alternative field L targets, HVC shelf and RA cup. In contrast, ZENK pathway leading from HVC to RA includes a seriesof paleo- induction is absent in some areas that show a response to striatal, thalamic, and neostriatal nuclei, and seemsto be es- song by other measures and where ZENK induction might sential for songlearning, but not necessarilyfor songproduction have been expected. These include the direct thalamo-re- in adults (Bottjer et al., 1984; Sohrabji et al., 1990; Scharff and cipient field L subfield L2, and the nuclei of the circuit in- Nottebohm, 1991). Various of these “song control nuclei” un- volved in the acquisition and production of learned song. dergo anatomical changesassociated with the development of These results demonstrate that ZENK induction following song production in juveniles (Konishi and Akutagawa, 1985; song presentation occurs only in a subset of areas physi- Nordeen and Nordeen, 1988) and with subsequentseasonal ologically activated by song, and draw attention to areas changesin singingbehavior in speciessuch as the canary (Not- previously unsuspected as related to processing of complex tebohm, 198 1, 1989; Nottebohm et al., 1986; Alvarez-Buylla auditory stimuli. Based on what is known about ZENK func- et al., 1990). tion in mammalian systems (Christy et al., 1989; Cole et al., Despite considerableprogress in understandingsong produc- 1989; Wisden et al., 1990), we speculate that areas revealed tion, relatively little is known about the neural circuitry re- by ZENK induction might correspond to sites where critical sponsible for song perception and discrimination, or for the neuronal modifications occur in response to birdsong pre- formation of memoriesrelated to the songsthe bird hears.An- sentation, possibly leading to the formation of song-related atomical studieshave revealed that the ascendingauditory path- memories. ways in birds share many homologies with mammalsand in- [Key words: song control circuit, birdsong, immediate-ear- clude the midbrain nucleus MLd (homolog of the mammalian ly gene, ZENK, field L, NCM] inferior colliculus) and the thalamic nucleus Ov (homolog of the medial geniculate;see Karten, 1967, 1968). Projectionsfrom Ov then reach the telencephalonthrough the primary auditory Songbirds have emerged, in the last 3 decades,as a powerful projection area, field L (Karten, 1968; Kelley and Nottebohm, system to study the mechanismsby which experience leadsto 1979). However, the anatomy and function of higher auditory the modification of behaviors. Birds must hear and remember processingstations in the avian telencephalonis poorly under- songsfrom other individuals in order to perform various tasks stood. Recent studiesof field L cytoarchitectonic organization essentialfor their survival and reproduction, such as songleam- in songbirds, for example, reveal that it is a complex structure ing, territorial defense,and courtship behavior (Kroodsma, 1976; consisting of various subdivisions whose specific connections and functions are not yet well defined (Fortune and Margoliash, 1992a). Metabolic studies and electrophysiological recordings received Nov. 16, 1993; revised Apr. 19, 1994; accepted Apr. 27, 1994. have demonstrated that units in the avian field L have a ton- We gratefully acknowledge the unfailing support and continual inspiration provided by Fernando Nottebohm over many years; without his generous and in- otopic organization and show low selectivity to complex audi- quisitive spirit, this marriage of molecular biology and neuroethology never would tory stimuli (D. Bonke et al., 1979; Mtiller and Leppelsack, have proceeded. We also thank Fernando Nottebohm for providing birds from 1985; Mtiller and Scheich, 1985); auditory responseswith a the Rockefeller University Field Research Center, with partial support of a grant from the National Institutes of Health (DCO0182). Specific grant support for the higher selectivity for complex stimuli have been describedin experiments presented here was provided to D.F.C. by the Whitehall Foundation areasadjacent to and receiving an input from field L, for instance and the NIH (NS25742). the caudomedial HV (Mtiller and Leppelsack, 1985). Correspondence should be addressed to Claudio V. Mello, Laboratory ofAnimal Behavior, The Rockefeller University, 1230 York Avenue, New York, NY, 1002 1. Initial studieson field L connectivity in songbirdsrevealed Copyright 0 1994 Society for Neuroscience 0270-6474/94/146652-l 5$05,00/O projections to the shelf and cup areasimmediately adjacentto The Journal of Neuroscience, November 1994, f4(11) 6653

nuclei HVC and RA, respectively (Kelley and Nottebohm, 1979) suggesting a pathway for auditory information to reach the nuclei that control song production. Because these nuclei are an- atomically well defined and have an obvious function in song behavior, several studies have been directed at testing the hy- pothesis that these song control centers also have a role in song perception and discrimination. From the work of a number of individuals, it is now apparent that selective electrophysiological responses can be evoked by the sound of specific songs at various levels throughout the song control pathway, all the way to the syringeal musculature (Katz and Gurney, 1981; Margoliash, 1983, 1986; Williams and Nottebohm, 1985; Doupe and Kon- Figure 1. Northernblot analysisof the ZENK gene. Shown is an x-ray ishi, 1991; Margoliash and Fortune, 1992; Vicario and Yohai, film autoradiogramof a blot containingsize-fractionated RNA isolated 1993). The identification of these auditory responses confirms from the forebrainof canaries(A andB) or zebrafinches (C), hybridized that the song control pathway receives auditory input and may to a 32P-labeledsingle-stranded antisense riboprobe for the ZENK gene. be tuned to specific song patterns. However, it does not yet A contains30 wug,of total RNA, and B and C contain3 ueof polyA+ RNA, final washconditions are 0.1 x SSPE,0.1% SDS’it 65°C(hy- establish a functional role for these responses, nor does it es- bridization protocolas described in Claytonet al., 1988). tablish whether the response specificity arises within the song control pathway or in other auditory processing centers that may ultimately impinge upon the song control pathway. striatum (NCM), ZENK geneactivation seemsto be specificto More suggestive evidence that the song control nuclei have a song stimulation, in that conspecific songselicit a greater re- function in song perception and behavioral modification has sponsethan heterospecific songs,and pure auditory tones do been provided by Brenowitz (1991) who showed that lesions not causeit (Mello et al., 1992). This responseappears to be directed at HVC will disrupt the selectivity of the behavioral specifically linked to perceptual processesas opposedto motor response of females to conspecific song. However, these “HVC” activity, as it occurs independent of song production by the lesions were not limited to HVC proper but also included sur- birds that hear the song.Although the functional significanceof rounding neo- and hyperstriatum, areas that could also be in- the ZENK responseis yet unknown, it neverthelessprovides a volved in auditory processing and perception (B. A. Bonke et way to map sitesin the brain that undergo a physiological re- al., 1979; Kelley and Nottebohm, 1979; Saini and Leppelsack, sponseto a songstimulus. Since the areaswhere we detect this 1981; Fortune and Margoliash, 1992; Mello et al., 1992). Fur- responsediffer from the areasone might have expected based thermore, behavioral responses by females to conspecific song on previous studies, the present methodology may highlight can be elicited in species in which females lack a complete song heretofore unappreciatedaspects of the organization of the song control circuit, such as zebra finches (Clark and Nottebohm, responsive circuitry. The goal of this report is to provide a 1990). Thus, it seems likely that important aspects of song per- detailed anatomical description of sitesof ZENK induction fol- ception may be mediated by areas outside the motor pathway lowing songpresentation. controlling song production. A significant barri& to further studies of the neural circuitry responsible for song perception or discrimination is the lack of Materials and Methods an adequate methodology. Studies based on brain lesioning are Animals. Adult malebirds of two specieswere used, 34 zebrafinches limited by the technical challenges of placing spatially confined (Taeniopygiq guttata) and 10 canahes(Serinus can&us), bredat the lesions, and are only meaningful in the context of information RockefellerUniversitv Field ResearchCenter (Millbrook. NY) or ob- tained from CanaryBird Farms(Englishtown, hJ). Zebra’finchesused provided by anatomical and electrophysiological methods. in kinetic analysiswere in similar hormonalconditions, as estimated Studies based on conventional anatomical and electrophysio- by the amountof singingactivity, beakcolor and testis size; all canaries logical techniques are challenged by the complexity of brain analyzedwere sacrificed in the spring(May), and bloodtestosterone areas involved in higher auditory processing (presumably ho- levelsand testis size were assessed to evaluatehormonal status. Song stimulation. Songplaybacks were performed under the same mologous to parts of the mammalian sensory and association conditionsas described in Mello et al. (1992).Tape recorded songs from cortices) and by the difficulties of recording from awake, freely threedifferent conspecific individuals, not includingthe bird’sown song, behaving birds. wereplayed in the sequence1-2-3-l-2-3 and so on, after 1d of acoustic A new approach for mapping brain regions responsive to song, isolation.The durationof eachsong presentation was 15 set, andthe that circumvents some of the limitations of these other con- presentationswere separated by 45 setof silence.Thus, each minute of playbackcorresponds to one 15 set exposureto song.To definethe ventional techniques, has recently been afforded by our obser- exact parametersof the songplayback (playback duration and animal vation that exposure to tape-recorded birdsong causes a rapid survival time) for the mappingstudies, we first performeda kinetic increase in a specific mRNA (“ZENK”) in the forebrain of ca- analysisof ZENK inductionby songin NCM, asdescribed in the first naries and zebra finches (Mello et al., 1992). The ZENK mRNA sectionof Results. In situ hybridization and image analysis. At the end of the song encodes a protein that is believed to function in the transcrip- stimulationperiod, birds were euthanized by decapitation,and brains tional regulation of other genes (Milbrandt, 1987; Christy et al., wererapidly dissectedand frozen for analysisof ZENK mRNA levels 1988, 1989; Lemaire et al., 1988; Sukhatme et al., 1988) and by in situ hybridization. Ten micrometercryostat sectionswere hy- its induction may represent an early step in a mechanistic cas- bridized essentiallyas describedby Clayton et al. (1988),with ?S- cade leading to long-term cellular changes underlying learning labeledantisense riboprobes derived from a canary ZENK cDNA clone (Mello et al., 1992).Sections were exposed to x-ray filmsfor 2-4 weeks, and memory (Goelet et al., 1986). The activation of this gene or dippedin NTB2 emulsion(Kodak) andexposed for 4-8 weeks;sec- following song occurs in relatively discrete portions of the song- tions were then counterstained with cresyl violet for microscopic anal- bird brain. In at least one of these areas, the caudomedial neo- ysis. As shown in Figure 1, hybridization of RNA blots to ZENK probes 6654 Mello and Clayton l ZENK Expression in the Songbird Brain after Song

3.5 0 p 3.0 10- 1 0 0 a- 0

0 6- : a

0 4- 0

2- 3 ok-‘-‘-‘-‘-‘-‘.‘-‘-‘.’ 0 10 20 30 40 50 60 70 0 5 10 15 20 25 30 35 40 45 50 SURVIVAL TIME (min.) PLAYBACK DURATION (min.) Figure2. Time course ofZENK mRNA induction after song playback. ZENK n-fold induction in the caudomedial neostriatum (NCM) ofzebra Figure 3. Timecourse of ZENK mRNA inductionafter songplayback: finches is plotted as a function of survival time after short-duration (10 effectof playbackduration. ZENK n-fold inductionin NCM of zebra min) playback ofconspecific song. Frozen brain sections were hybridized finchesis plotted asa functionof playbackduration; all animalswere with ‘5S-labeled riboprobes for ZENK and exposed to x-ray films; optical sacrificed 40 min after beginning of playback. Frozen brain sections density measurements in NCM were taken from autoradiograms and were hybridizedwith %-labeledriboprobes for ZENK and exposedto normalized to levels in unstimulated controls. Data points represent x-ray films; optical densitymeasurements in NCM weretaken from individuals (n = 2/time point); unstimulated controls are plotted as autoradiograms and normalized to levels in unstimulated controls. Data open circles at time = 0. points represent individuals (n = Z/time point); unstimulated controls are plotted asopen circles at time = 0. under identical conditions results in detection of a single, clean band of about 3.5 kb for both species tested, confirming the specificity of Results these reaction conditions (Northern blot procedure as described in Clay- Kinetics of ZENK induction by song ton et al., 1988). For anatomical description of ZENK induction patterns, song-stim- As a first step before extensive anatomical mapping, we con- ulated zebra finches were sectioned in the parasagittal (five birds) or the ducted two experiments where we defined the kinetics of the coronal plane (three birds), and compared to two unstimulated control ZENK responseto song, in order to establish a stimulation birds in each plane; eight song-stimulated canaries were analyzed in the paradigm that would result in a high and reproducible ZENK parasagittal plane an&compared to two unstimulated controls. For ki- induction. Densitometric analysis of in Situ hybridization au- netic analysis of ZENK induction. 22 zebra finches were stimulated as described

ZENK induction after song: general pattern The anatomical distribution of ZENK mRNA signal was de- termined by analyzing both x-ray film autoradiograms and emulsion-dipped serial brain sections after hybridization with ZENK riboprobe. We examined both zebra finches and canaries, and identified brain regions where ZENK hybridization signal a increased in birds exposed to song compared to unstimulated controls. The lower and upper diagrams of Figure 4 indicate the levels of coronal and parasagittal brain sections shown in Figures 5 and 6, respectively, where zebra finches exposed to song (mid- 1 dle columns) are compared to unstimulated finches (right col- P umns). The patterns of ZENK induction observed in adult male zebra finches (shown in this report) and in adult canaries in the spring (not shown) are remarkably similar. As can be seen in the series ofcoronal sections (Fig. 5, compare middle and right columns), ZENK induction after song presentation is bilateral and restricted to the medial portions of the telencephalon, at the more caudal levels. The area of high ZENK signal consists of the medial portions of the neostriatum and d hyperstriatum ventrale (Fig. 54-F) as well as the caudodorsal paleostriatum (Fig. 5G; see also Fig. 6) and portions of the archistriatum (better seen in Fig. 6). A central or core negative t P area within the neostriatum, which appears as a horizontal stripe surrounded by areas showing significant ZENK induction, can be seen in Figure 5C-E. In general, ZENK signal tends to be denser in more medial and dorsal regions, and lateral boundaries of areas showing high ZENK signal can be clearly seen in these autoradiograms (Fig. 5A-F). When sections are stained with Figure 4. Diagrammatic schemes of the songbird brain. Top, Dorsal cresyl violet, no obvious boundaries such as laminae or differ- view of the brain; lines represent levels of parasagittal sections (A-G) ences in cell number or density corresponding to the lateral shown in Figure 6. Bottom, Diagramof a parasagittalsection at the border of the high ZENK region are seen within the neostriatum levelindicated by lineA in the toppanel; lines represent levels of coronal sections(A-H) shownin Figure5. or hyperstriatum ventrale, although thicker sections might be necessary to provide more anatomical detail. There seem to be no gross interhemispheric differences; the difference seen in Fig- (Fig. 6A) and can be easily overlooked in the more medial sec- ure 5A is more likely the result of the plane of sectioning, since tions since it completely separatesfrom the rest of the brain the left hemisphere of this brain was cut at a level slightly more when brains are cut in a vibratome. This protuberance consists caudal than the right one. of a central core, field L, and the surrounding caudomedial The parasagittal series(Fig. 6) confirms this general pattern neostriatum and hyperstriatum. These areas are individually and demonstratesthat the most marked ZENK induction occurs analyzed in further detail below. in the caudal telencephalon. This effect is most pronounced in the more medial sections(Fig. 6A), where a homogeneouslyhigh Caudomedial hyperstriatum ventrale (CMHV) signal covers a relatively large area of the caudal telencephalon The caudomedial HV is seenin coronal sectionsextending me- formed by the caudomedial hyperstriatum ventrale (CMHV) dially and ventrally toward the midline, betweenthe hippocam- and neostriatum (NCM). The central or core negative region is pus and the caudomedialneostriatum; it constitutes most of the very conspicuousin this plane of section. In the more medial dorsal area of high ZENK expressionafter songpresentation in sections, the negative region is stripe-shaped(Fig. 6A-E) and Figure X-F. In medial parasagittal sections, HV constitutes extends caudodorsally from the caudodorsal tip of the paleos- the rostrodorsal portions of the drop-shapedarea that shows triatal complex (Fig. 6D,E) toward the dorsal neostriatum, fol- high (sevenfold above control levels) ZENK induction (Fig. 6A). lowing a course roughly parallel with the lamina that separates At more lateral levels, the caudal HV extends rostrally, and the hyperstriatum ventrale and the neostriatum (Fig. 6B-E); in eventually becomescontinuous with the more rostra1portions more lateral sections,this areaassumes a boomerangor inverted of HV; high ZENK levels are restricted to the caudal portions, V-shape (Fig. 6F,G); as describedmore fully below, this central closely adjacent to field L (Fig. 6C-G). core correspondsto portions of the auditory thalamo-recipient zone (field L). Caudomedial neostriatum (NC&f) ZENK induction following song occurs, thus, in a structure In medial parasagittalsections (200-500 pm from the midline), that might be describedas a medial protuberance of the caudal NCM is easily defined by its natural boundaries: the ventricular telencephalon. This protuberance is bound dorsomedially by zone dorsally, ventrally and caudally; rostrodorsally, it is sep- the hippocampus (Figs. 5A-H, 6A-C) and lies over the cere- arated from HV by the lamina hyperstriatica and medial por- bellum caudally (Figs. 5A,B, 6A), and over caudal diencephalic tions of field L (Fig. 6A; see also Mello et al., 1992). In even structures more rostrally (Figs. 5D, 6A). When sectionedin the more medial sections(less than 100 Frn from the midline) field parasagittalplane, this structure hasan ovoid or drop-like shape L and HV are absent and NCM has a rounder shapeand is A

H The Journal of Neuroscience, November 1994, 74(11) 6657

surrounded on all sides by ventricular zone (not shown). In more LI and L3. Significant induction is observed in these parts lateral sections, the caudal neostriatum becomes gradually larg- of field L. One of the patches showing high ZENK induction er, assuming a drop-like shape together with the caudomedial (4.5-fold) lies ventrocaudally to the stripe-shaped negative core HV and eventually becoming continuous with the rostra1 parts and dorsocaudally to the paleostriatum and occupies the con- of the neostriatum (Fig. 6%D). At these more lateral levels, cavity of the boomerang or inverted V-shape (Fig. 6E-G); it ZENK signal is more diffuse and a stripe-shaped area negative thus closely matches L3 (Fortune and Margoliash, 1992a). A for ZENK appears within the neostriatum (Fig. 6A-D). In co- significant induction (2.5-fold) is also seenin the region anter- ronal sections, NCM can be defined as a thin neostriatal layer odorsal to the negative core, an area that probably corresponds with high ZENK signal that separates the core negative region to Ll (Figs. 5C-E, 6D,E; and compare with Fortune and Mar- (L2) from the midline (Fig. SC-E); more caudally, ZENK ex- goliash, 1992a). pression occurs in a larger area of the neostriatum that is caudal The relationships between field L subfieldsare observed in to field L (Fig. 54,B). closer detail in Figure 7. The diagram on the upper left comer representsa camera lucida drawing of a cresyl-stained parasagittal section at the same level as the autoradiogram in Figure Field L 6E (notice the reverse orientation of the figures). L2a is easily ZENK signal in parts of the neostriatum around the negative defined as a field that extends from the caudodorsaltip of the core presents a patchy appearance, especially in the parasagittal paleostriatum toward the dorsal neostriatum and runs parallel plane (Fig. 6E-G). Recent studies on the organization of the to the caudal end of the lamina hyperstriatica; it contains small avian brain provide evidence suggesting that some of these areas and darkly staining cells, typically organized in short rows or correspond to divisions of field L, the primary forebrain audi- columns aligned with the major axis of L2a. Other subfields tory area. Field L is a complex aggregation of subfields that have were defined with relation to L2a and the laminae hyperstriatica been variously defined by morphological and electrophysiolog- (LH) and medullaris dorsalis (LMD), both readily identified ical criteria. Fortune and Margoliash (1992a) recently presented with cresyl violet staining. Caudal to the anglebetween L2a and a detailed cytoarchitectonic study in the zebra finch in which the caudal part of the LMD (which separatesneostriatum from field L was divided into five subregions: Ll, L2a, L2b, L3 and paleostriatum) is L3, and immediately rostra1 to L2a is Ll, L (or L of Rose). Detailed differences in cytoarchitectonic or- whoserostra1 boundaries are not well-defined. The areaenclosed ganization are not easily seen in thin sections after in situ hy- by the hatched rectangle in the diagram is shown in detail in bridization; L2a, however, is readily identified. The localization Figure 7, A and B (bright- and dark-field views, respectively, of of the remaining subdivisions is inferred with reference to L2a, a section from a songstimulated bird) and Figure 7C (dark-field as well as to the laminae that separate the hyperstriatum ventrale view of a section from an unstimulated control). Some signalis from the neostriatum (LH or lamina hyperstriatica) and the seenin the unstimulated Ll and L3 (Fig. 7C) but a dramatic paleostriatum from the neostriatum (LMD or lamina medullaris increase in the number of positive cells occurs in both fields dorsalis). By comparing Figure 6 with the diagrams in Figure 2 after song presentation (Fig. 7B), but not in L2a. High power of their report (e.g., our Fig. 6F corresponds approximately to views demonstrate positive cells in Ll (Fig. 70, right, arrow their Fig. 2B,C), it becomes clear that the patchy pattern ob- points to a positive cluster) and L3 after songpresentation (Fig. served corresponds, at least in part, to various subregions of 7E, upper half, positive cells in clusters or in isolation) but not field L. The occurrence of differential patterns of gene induction in L2a (seethe left half of Fig. 70). within field L corresponding to Fortune and Margoliash’s divisions demonstrates that the various field L subdivisions are Relationship to song control nuclei composed of cells with different properties, and lends further ZENK induction following songis not observed in songcontrol support to Fortune and Margoliash’s classification. From this nuclei HVC (Figs. 5, 6, 8, 9), RA (Figs. 5, 6, 8, 9) and NIf (not comparison we draw the following observations. shown), although significant induction is observed in areasim- L2a, L26, field L of Rose. Induction is not observed in these mediately adjacent to these(as describedin more detail below), areas. The core or central area of negative signal corresponds nor does it occur in Area X (Fig. 6) DLM (not shown), and to subfields L2a (thin stripe oriented obliquely from the dor- L-MAN (Fig. 6). socaudal tip of the paleostriatum toward the dorsocaudal neo- HVC shelf: A marked induction (fourfold on x-ray film au- striatum, Fig. 6&E), L2b (the apex of the boomerang or in- toradiograms) can be seenin the caudodorsalneostriatum, most verted V-shaped negative core region in Fig. 6F,G) and the prominently in the area immediately adjacent to HVC (Fig. 8). lateral portions of field L of Rose (the caudal leg of the inverted The neostriatum underlying HVC receives a direct input from V-shape in Fig. 6G, compare with definition by Fortune and field L, possibly from subfieldsLl and L3 (Kelley and Notte- Margoliash, 1992a); it is unclear whether ZENK induction oc- bohm, 1979; Fortune and Margoliash, 1992b; Mello, Okuhata, curs in the more medial portions of field L of Rose, since this and Nottebohm, unpublishedobservations) and hasbeen called region was not readily identified in our material. the shelf region. As can be seenfrom the dark-field images,the

Figure 5. ZENK mRNA induction in the zebra finch brain following exposure to song. A-H represent a series of coronal sections, from +0.25 (A) to +2.5 mm (H) in the anteroposterior axis, as indicated in Figure 4 (bottom). Left column, Diagrams of camera lucida drawings of the sections whose autoradiograms are shown in the middlecolumn. In situ hybridization autoradiograms at levels corresponding to the diagrams on the left columndepict brains ofa song-stimulated (middlecolumn) and an unstimulated (rightcolumn) bird. A, archistriatum; Cb,cerebellum; E, ectostriatum; H, hyperstriatum; Hp, hippocampus and parahippocampal area; Hv, hyperstriatum ventrale; HVC, high vocal center; L2, subfield L2 of field L, N, neostriatum; PA, paleostriatum augmentatum; PP, paleostriatum primitivum; RA, nucleus robustus archistriatalis. Scale bar, 2 mm. 6658 Mello and Clayton l ZENK Expression in the Songbird Brain after Song

NCM

Figure 6. ZENK mRNA induction in the zebra finch brain following exposure to song. A-G represent a series of parasagittal sections, from 0.25 mm (A) to 2.2 mm (G) lateral to the midline, as shown in Figure 4 (toppanel). Left column,Diagrams of camera lucida drawings of the sections whose autoradiograms are shown in the middZecolumn. In situ hybridization autoradiograms at levels corresponding to the diagrams on the left The Journal of Neuroscience, November 1994, 14(11) 6659

neostriatum under HVC in unstimulated controls contains a This area receives a direct projection from the thalamic relay few ZENK positive cells (Fig. 8B) but the number of these cells nucleus ovoidalis (Kelley and Nottebohm, 1979) and seems to is dramatically increased in song-stimulated animals (Fig. 84. be reciprocally connected with some parts of the auditory fore- The exact shape of this area as well as the number of cells that brain (B. A. Bonke et al., 1979; Mello, Okuhata, and Notte- are activated in response to song varies from animal to animal bohm, unpublished observations). and in some cases the distribution of positive cells under HVC MLd. The only area outside the forebrain where a significant is more homogeneous than that shown in Figure 8.4. ZENK ZENK induction (2.5fold) is consistently observed is the nu- induction in the neostriatum underlying HVC is shown in fur- cleus mesencephalicus lateralis, pars dorsalis, or MM (Fig. IO), ther detail in Figure 9, A and C, representing bright-field images a subdivision of the avian midbrain that relays ascending au- of the same section as shown in Figure 8A. It can be clearly seen ditory information to thalamic centers and is thought to be that ZENK-positive cells occur immediately adjacent to HVC homologous to the mammalian inferior colliculus (Karten, 1967). and respect its boundaries, although a few cells are localized right at the transition between HVC and its surrounding fibrous Forebrain areas where ZENK is not induced layer (Fig. SC, arrow); positive cells in this region are relatively Telencephalic regions where very little or no apparent ZENK small, compared with the large negative cells within HVC (for induction occurs include the song control nuclei (above), an- instance, the neuronal cluster within HVC depicted by the ar- terior and lateral N (as well as some of its dorsal portions), the rowhead in Fig. 9C). anterior and lateral portions of all subdivisions of the hyper- RA cup. ZENK is 3.5 times higher in the archistriatum sur- striatum (H), the anterior paleostriatal complex (LPO and PA), rounding RA after song presentation compared to controls (Figs. the paleostriatum primitivum (PP), primary sensory recipient 8, 9). Here, again, although a few positive cells are present in zones other than field L (nucleus basalis, ectostriatum), the hip- this area in unstimulated controls (Fig. 8E), the number of these pocampus and parahippocampal areas and most of the archis- cells rises dramatically after song presentation (Fig. 80). Under triatum (Figs. 5, 6). In addition, no ZENK induction is seen bright-field (Fig. 9B,D, same section as shown in 8D), the ma- within the main nucleus of the auditory thalamus (n. ovoidalis); jority of ZENK positive cells are seen in clusters that localize hypothalamic structures were not examined in detail. rostroventrally to RA, in contrast to the large negative cells within RA; some positive cells are present within the capsule that surrounds RA and assume an elongated form in a direction Discussion tangential to RA (Fig. 9D, arrow). This general area surrounding The studies described here provide a map of brain regions in RA seems to colocalize at least partly with a projection area which a specific cellular response (increased transcription of the from the auditory field L, which has been named the RA cup ZENK gene) occurs after songbirds are exposed to the sound of (Kelley and Nottebohm, 1979). The archistriatum next to RA conspecific birdsong. We have previously shown that, in the also receives an input from the dorsal neostriatum (HVC shelf) caudomedial neostriatum, this response does not occur follow- and originates descending projections to lower auditory centers ing exposure to simpler auditory stimuli like tones, and that a (Mello, Okuhata, and Nottebohm, unpublished observations). significantly lower response occurs following heterospecific song NIX ZENK induction occurs in areas in close apposition to playbacks (Mello et al., 1992). Thus, these results provide insight NIf (field Ll). To test whether ZENK induction occurs in NIf, into the neurocircuitry that may be involved in aspects of au- cells were retrogradely labeled in this nucleus with injections of ditory processing such as the recognition of species-specific vo- fluorogold in HVC in two animals; clearly backfilled cells were calizations. restricted to the area of low ZENK induction, next to the neg- ZENK encodes a sequence-specific DNA binding protein that ative core of field L (not shown). In one case, NIf (a nucleus probably regulates the expression of a number of other genes difficult to visualize in thin cresyl-stained sections) could be (Christy et al., 1989). Rapid induction of the ZENK gene has directly identified in emulsion-dipped sections and absence of been observed in mammalian nervous tissue following both grains on NIf cells was confirmed. growth factor stimulation and membrane depolarization (Mil- brandt, 1987; Saffen et al., 1988; Bartel et al., 1989; Cole et al., Other areas of ZENK induction 1989; Wisden et al., 1990). However, the functional significance Intermediate neostriatum. A marked induction occurs in a dis- of this response is not yet known. ZENK induction is clearly crete circular spot in the neostriatum adjacent to and anterior not an obligatory or universal concomitant of neural stimula- to field L (Ll) and NIf (Fig. 6G); its appearance and shape in tion; this has been shown, for instance, in the hippocampus, the autoradiograms suggest that it could correspond to a discrete where ZENK induction seems to be specifically coupled to nucleus within the neostriatum; it is not obvious, however, in NMDA receptor activation (Cole et al., 1989). In fact, among cresyl violet-stained sections. several genes that show similarly rapid responses to cell stim- Paleostriatal complex. Another area that shows an increase ulation, induction of ZENK appears to be more specifically in ZENK mRNA (fourfold above control levels) is the caudo- associated with patterns of cell activation that lead to neuronal dorsal portion of the paleostriatal complex, more specifically change (Cole et al., 1989; Wisden et al., 1990). Thus, ZENK the paleostriatum augmentatum (PA; Figs. 5F,G; 6E-G; 7B,E). activation may represent an initial step in a cascade of events c column depict brains of a song-stimulated (middlecolumn) and an unstimulated (rightcolumn) bird. A, archistriatum; BS, brainstem; Cb,cerebellum; H, hyperstriatum; Hp, hippocampus and parahippocampal area; Hv, hyperstriatum ventrale; HVC, high vocal center; L2a, subfield L2a of field L, L2, subfield L2 of field L, LPO, lobus paraolfactorius; M, midbrain; MAN, lateral nucleus magnocellularis of the anterior neostriatum; N, neostriatum; NC, neostriatum caudale; P, paleostriatum; PA, paleostriatum augmentatum; PP, paleostriatum primitivum; RA, nucleus robustus archistriatalis; T, thalamus; A’, area X. Orientation: dorsal is up and anterior is to the left. Scale bar, 2 mm. 6660 Mello and Clayton l ZENK Expression in the Songbird Brain after Song

Figure 7. ZENK mRNA induction after song presentation in the zebra finch field L complex. At the top left is a diagrammatic scheme of a parasagittal brain section at the level of field L, the boxedarea is shown in detail in A-C. A, Bright-field view of emulsion-dipped section stained with cresyl violet after in situ hybridization with ZENK riboprok, Dashedlines are drawn over the laminae that separate L subdivisions and the The Journal of Neuroscience, November 1994, 14(11) 6681

HVC

Figure 8. ZENK mRNA induction in the HVC shelf and RA cup after song presentation. At the top rightis a diagrammatic scheme of a parasagittal brain section at the level of nuclei HVC and RA. The area enclosed by the upper rectanglecontains HVC and surrounding tissue and is shown in detail in A-C; the area enclosed by the lower rectanglecontains the rostroventral part of RA and surrounding tissue and is shown in D and E. A- E, Dark-field views of emulsion-coated zebra finch brain sections after in situ hybridization with ZENK riboprobes. Notice high hybridization signal in the areas adjacent to HVC and RA after song stimulation (A and D), compared with unstimulated controls (I3 and E). C, No significant hybridization signal is seen when a section adjacent to A is hybridized with sense-strand probe. HVC, high vocal center; RA, nucleus robustus archistriatalis. Scale bar, 200 pm.

paleostriatum. B and C, Dark-field view of the same area as shown in A in a song-stimulated and a control bird respectively; notice marked ZENK induction in fields Ll, L3, and P, but not in L2a. D, High-power view of the area enclosed by the smallrectangle in A: dashedline is drawn over the lamina that separates L2a (left) and Ll (right); notice grains over cells in Ll (arrowindicates ZENK-positive cell cluster) but not in L2a. E, High-power view of the area enclosed by the largerectangle in A: dashedline is drawn over lamina that separates L3 and PA, notice grains over cells in L3 and PA. Ll, L2a, and L3, subdivisions of field L, PA, paleostriatum augmentatum. Scale bars: A-C, 200 pm; D and E, 50 pm. 6662 Mello and Clayton l ZENK Expression in the Songbird Brain after Song

I-;.‘--... . _,

Figure 9. ZENK mRNA induction in the HVC shelf and RA cup after song presentation. Zebra finch brain sections at the level of nuclei HVC and RA are shown, coated with autoradiographic emulsion after in situ hybridization with ZENK riboprobes and counterstaining with cresyl violet. A and Bare bright-field views of the same sections as shown in Figure 8, A and D, respectively; dashed lines are drawn over the boundaries between HVC and RA and adjacent tissues. C and D, High-power views of the areas enclosed by the rectangles in A and B, respectively; labeled cells are seen in the tissues adjacent to HVC and RA, but not within HVC and RA. Arrows indicate labeled cells in the transition zones between HVC (C) and RA (D) and adjacent tissues; arrowhead points to an unlabeled neuronal cluster within HVC (C). HVC, high vocal center; RA, nucleus robustus archistriatalis. Scale bars: A and B, 200 pm; C and D, 30 pm.

leadingto long-lastingcellular changesthat could underlie learn- auditory relay nucleusovoidalis, the primary forebrain auditory ing and memory (Goelet et al., 1986). projection area (field L2), and the various nuclei of the motor The use here of ZENK induction as a mapping tool for an- pathway for songproduction (discussedfurther below). ZENK alyziingbrain areasthat display functional responsesto a specific induction thus seemsto reveal a subsetof siteswhere physio- sensorystimulus is analogous,in principle, to the use of elec- logical activation occurs;the compelling speculationis that these trophysiological recordings or indicators of metabolic activity may specifically representsites undergoing plastic changein the (e.g., 2-deoxyglucoseuptake, or blood flow measurements)for circuit, although this remains to be proven. similar purposes.However, activity maps generated by these other techniques do not necessarilycoincide with ZENK in- Summary of ZENK induction pattern duction maps. Indeed, we observed no ZENK responsein sev- A schematicdiagram demonstrating brain regionsrevealed by era1areas that have been previously describedto show physi- ZENK induction following song(shaded areas) is shownin Fig- ologic responsesto song stimulation, including the thalamic ure 11. Two general effects are observed. The Journal of Neuroscience, November 1994, 14(11) 6663

Figure 10. ZENK mRNA inductionin MLd after songpresentation. At the lowerright is a drawingrepresenting a transverse section of a zebra finchbrain at the level of MM; the boxed area is shownin detail in A-C. A andB, Emulsion-coatedbrain sections from songstimulated (A) and unstimulated(B) birds,after in situ hybridization with riboprobesfor ZENK andcounterstaining with cresylviolet, viewedunder dark-field. C, Bright-fieldview of A. A, archistriatum:BS, brainstem;Cb, cerebellum;Hp, hippocampus;MM, nucleusmesencephalicus lateralis, pars dorsalis; N, neostriatum.Scale bar, 230pm.

Induction in the caudomedialtelencephalon, in areasadjacent playbacks result in a habituation of the responseto the specific to or closely related with primary auditory structures,including song used, but not to other songs(Mello and Clayton, unpub- some subfieldsof field L (Ll and L3, shown in the diagram as lished observations). a ring around field L core, L2), the caudomedial neostriatum Among ZENK positive areas,attention is immediately drawn (NCM) and the caudomedial hyperstriatum ventrale (CMHV) to the most caudomedialportions of HV andN, for the following anterior to field L, the caudal paleostriatum (PC),and two areas reasons.(1) The most marked ZENK induction seemsto occur that overlap with field L targets, HVC shelf and RA cup (note there; the fact that basal levels are very low in these areasin that theselast two areasare also closely apposedto songcontrol the unstimulated controls may contribute significantly to that nuclei). effect. (2) The close proximity to the primary auditory area A lack of induction in some areasknown to be activated by suggeststhe possibility of auditory input and a function in au- songand where induction might have been expected, thesein- ditory processing.The caudomedial HV has been previously clude (1) the direct thalamo-recipient field L subfield, L2, and shown to be reciprocally connected with field L (B. A. Bonke (2) the nuclei of the songsystem involved in the acquisition and et al., 1979) in the guinea fowl; in addition, auditory responses production of learned song. with high selectivity for complex stimuli have beenrecorded in Note that this anatomical pattern wasobserved in adult birds starlings in this area and in the neostriatum close to field L sacrificed 30 min after beginning a playback of tape-recorded (Miiller and Leppelsack, 1985). (3) Cytoarchitectonic bound- conspecificsong; changes in the pattern might conceivably occur ariesof theseareas are easily identified in cresyl-stainedsections; if various aspectsof the experimental paradigm were changed, this allows precisedelineation for densitometric measurements suchas duration of playback, familiarity of the bird to the song in the evaluation of the time courseand specificity of the ZENK stimulus and the hormonal or developmental state of the bird. response. No obvious variations in this general pattern have been seenin either preliminary seasonal(canaries in October vs May) or Involvement of NC34 in auditory processing developmental (juvenile vs adult zebra finches) comparisons NCM, in particular, has not been previously describedas di- (Mello and Clayton, unpublished observations), although a rectly involved in song production or perception. To address quantitative analysishas not beenperformed. However, in other the possibility that NCM may be involved in auditory process- recent experiments with adult birds we have found that longer ing, we have begun an analysis of NCM connectivity usingan- 6664 Mello and Clayton - ZENK Expression in the Songbird Brain after Song

production systems. To clarify the role of NCM in auditory processing,other experiments are necessary,including a study of the effects of several classesof auditory stimuli (bird’s own song, conspecific song played backward, white noise, etc.) on ZENK induction and electrophysiological recordings to deter- mine the responseproperties of units within NCM. Significance of induction in other brain regions Most other brain areas showing high ZENK induction in responseto songhave also been described as intimately related to auditory pathways. Fields Ll and L3 may receive a sparse direct projection from the auditory thalamus,but they also constitute the main target of field L2. This suggeststhat theseareas could represent higher order stations for auditory processing than field L2; the fact that more complex responsesand longer latenciesare seenas one moves a recording electrode away from a central area that correspondsto field L2 (Miiller and Leppel- sack, 1985; Heil and Scheich, 1991) is consistentwith this idea. The HVC shelf and the RA cup both receive a projection either directly from field L or from some immediately adjacent area (Kelley and Nottebohm, 1979); the shelf and cup may represent sitesof converging auditory inputs (Mello, Okuhata, and Not- tebohm, unpublished observations), and their close proximity to songcontrol nuclei suggeststhat they could then contribute AUDITORY to the song-selectiveauditory responsesseen within the song t control circuit. The caudal portion of the paleostriatal complex (to medullary nuclei involved in receives a direct projection from the thalamic relay nucleus song and respiratory control) ovoidalis (Kelley and Nottebohm, 1979), whereas MM is a Figure II. Schematic summary diagram representing areas where the central component of the ascending auditory pathway and is ZENK geneis induced(shaded areas) in responseto playbacksof con- generally consideredthe avian homolog of the mammalian in- specificsong. The ring aroundthe field L core(L2) representsLl and L3, the shadedurea under HVC representsthe shelfregion and the area ferior colliculus (Karten, 1967). aroundRA representsthe cup region.Thin solidarrows represent the connectionsof the main avian ascendingauditory pathway(Karten, Lack of ZENK responsein field L2 and the songcontrol 1967, 1968; Kelley and Nottebohm, 1979) and the telencephalic pro- nuclei jections of field L in canaries (Kelley and Nottebohm, 1979). Thicksolid arrowsrepresent connections of the descendingsong control pathway ZENK induction in responseto songseems to occur in specific (Nottebohm et al., 1976). Dashedarrows represent preliminary results subregionsof the brain that are closely related to auditory struc- observed after injectionsof anterogradeor retrogradetracers in NCM. tures and could thus be involved in various aspectsof auditory No attemptwas made to representall knownconnections or the densities of these connections; it is also mostly unknown whether they are ex- processing.However, not all regions of the brain involved in citatory or inhibitory. CMHV, caudomedialhyperstriatum ventrale; auditory processingare necessarily revealed by ZENK induc- HVC, high vocal center; L, core of field L (L2); mesenc,mesencephalon; tion: L2a and the songcontrol nuclei constitute clear examples MLd, nucleus mesencephalicus lateralis, pars dorsalis; NCM, caudo- of areas where no ZENK induction is detected despitethe fact medial neostriatum; Ov, nucleus ovoidalis; PC,caudal paleostriatum; that stimulation from song playbacks is known to reach these RA, nucleus robustus archistriatalis; telenc,telencephalon; thud,thalamus. regions. Field L is the primary forebrain thalamo-recipient zone for auditory stimuli (Karten, 1968; B. A. Bonke et al., 1979; Kelley terograde and retrograde tracers (Mello, Okuhata, and Notte- and Nottebohm, 1979). The heaviest telencephalic termination bohm, unpublished observations). Preliminary results, as of the auditory thalamus correspondsto L2a, which then pro- summarized in Figure 11, indicate that NCM is an integral part jects to surroundingfields, Ll and L3 (Karten, 1968; B. A. Bonke of the telencephalic auditory circuitry. Injections of retrograde et al., 1979; Saini and Leppelsack, 1981). Cells in L2a are char- tracers in NCM result in the retrograde labeling of cells in the acterized by very low responseselectivity and respondreadily, thalamus (in close apposition to nucleus ovoidalis), in field L, both electrophysiologically and metabolically, to various classes in the caudomedial HV (HVCM) and in the caudodorsal pa- of auditory stimuli, from simple to complex (including species- leostriatum (PC). Injections of anterograde tracers result in la- specific vocalizations), as is expected of an area that may func- belled fibers within HVCM, PC, and more lateral portions of tion primarily as a relay to higher-order auditory regions (D. the neostriatum, including the area under HVC (shelf), a field Bonke et al., 1979; Scheich et al., 1979; Braun et al., 1985; L target that could contribute an input to HVC (Katz and Gur- Miiller and Leppelsack, 1985; Miiller and Scheich, 1985). Nev- ney, 1981; Mello, Okuhata, and Nottebohm, unpublished ob- ertheless,no ZENK induction was observed in L2a (or in L2b, servations). In addition, units in NCM respondto auditory stim- a subdivision of L2 that specifically receivesinput from nucleus ulation (S. Okuhata, S. Volman, and D. Vicario, personal semilunaris paraovoidalis, a subdivision of the auditory thal- communications).The position NCM apparently occupieswith- amus closely related to ovoidalis; see Wild, 1987). Similarly, in the forebrain auditory pathways suggeststhat NCM could ZENK induction was notably absent in the nuclei of the song represent part of the interface between song perception and control circuit, which respond metabolically to auditory stim- The Journal of Neuroscience, November 1994, f4(11) 6665 ulation (Braun et al., 1985) and where electrophysiological re- ties-specific calls and synthetic stimuli. I. Tonotopy and functional sponses with a high selectivity for species-specific song have zones of field L. J Comp Physiol 132:243-255. Bottjer SW, Miesner EA, Arnold AP (1984) Forebrain lesions disrupt been reported (Margoliash, 1983, 1986; Williams and Notte- development but not maintenance of song in passerine birds. Science bohm, 1985; Doupe and Konishi, 1991; Margoliash and For- 224:901-903. tune, 1992; Vicario and Yohay, 1993). Braun K, Scheich H, Schachner M, Heizmann CW (1985) Distribution In another study, we have found a specific lack of ZENK of parvalbumin, cytochrome oxidase activity and 14C-2-deoxyglucose response in primary telencephalic receptive fields and in an- uptake in the brain of the zebra finch. I. Auditory and vocal motor systems. Cell Tissue Res 240:101-l 15. drogen-concentrating nuclei within the song control circuit fol- Brenowitz E (1991) Altered perception of species-specific song by lowing metrazole-induced seizures, despite widespread ZENK female birds after lesions of a forebrain nucleus. Science 251:303- induction throughout the rest of the telencephalon (Mello and 305. Clayton, unpublished observations). Taken together, our find- Christy B, Lau L, Nathans D (1988) A gene activated in mouse 3T3 cells by serum factors encodes a protein with “zinc finger” sequences. ings suggest that ZENK induction has been uncoupled from Proc Nat1 Acad Sci USA 85:7857-7861. cellular depolarization in these specific areas of the telenceph- Christy B, Lau L, Nathans D (1989) DNA binding site of the growth alon. The mechanism of this uncoupling, and its functional factor-inducible protein Zif268. Proc Nat1 Acad Sci USA 86:8737- significance, are both unknown at this time. It also remains to 8741. be seen whether other classes of genes responsive to depolar- Clark SJ, Nottebohm F (1990) Perception of birdsong by female zebra finches and canaries. Sot Neurosci kbstr 16:2.1106 - ization can be activated in areas that lack a ZENK response. Clayton DF, Huecas ME, Sinclair-Thompson EY, Nastiuk KL, Not- However, we have observed that at least one other immediate tebohm F (1988) Probes for rare mRNAs reveal distributed cell early gene, c-jun, is induced in a pattern roughly similar to subsets in canary brain. Neuron 1:249-26 1. ZENK, either by song stimulation or metrazole (Mello, Nastiuk, Cole AJ, Saffen DW, Baraban JM, Worley PF (1989) Rapid increase of an immediate early gene messenger RNA in hippocampal neurons and Clayton, unpublished observations). If immediate early gene by synaptic NMDA receptor activation. Nature 340:474-476. activation represents a necessary step in long-term activity-de- Doupe AJ, Konishi M (199 1) Song-selective auditory circuits in the pendent modifications of neural circuitry, as others have sug- vocal control svstem of the zebra finch. Proc Nat1 Acad Sci USA 88: gested, a selective shutting off of their induction in certain areas 11339-11343.- like primary sensory receptive fields and motor control path- Falls JB (1982) In: Acoustic communication in birds (Kroodsma DE, Miller DH. eds). DD 237-278. New York: Academic. ways (such as the song control circuit) might provide a mech- Fortune E, Margoliash D (1992a) Cytoarchitectonic organization and anism for stabilizing essential sensory and motor representa- morphology of cells of the field L complex in male zebra finches tions in the brain. (Taenopygia g&tutu). J Comp Neurol 325:388404 Fortune E, Margoliash D (1992b) Multiple auditory pathways into HVc. Sot Neurosci Abstr 18:2.1193. Conclusion Godard R (1991) Long-term memory of individual neighbors in a Electrophysiological responses recorded in nuclei of the song migratory songbird. Nature 350:228-229. control pathway in response to song playbacks are highly selec- Goelet P, Castelucci V, Schacher S, Kandel E (1986) The long and the short of long-term memory-a molecular framework. Nature 332: tive, with preferences for conspecific songs and for the bird’s 419422. own song (Margoliash, 1986; Doupe and Konishi, 199 1; Vicario Heil P, Scheich H (199 1) Functional organization ofthe avian auditory and Yohay, 1993). This has led investigators to implicate this cortex analogue. II. Topographic distribution of latency. Brain Res pathway in aspects of song perception and processing. ZENK 539:121-125. induction patterns, however, draw attention to larger brain Karten HJ (1967) The organization of the ascending auditory pathway in the pigeon (Columba livia). I. Diencephalic projections of the in- regions that are intimately related with the primary auditory ferior colliculus (nucleus mesencephalis lateralis, pars dorsalis). Brain area and may process song information before it reaches the Res 6~409427. song system. Together with data derived from anatomical stud- Karten HJ (1968) The ascending auditory pathway in the pigeon (Co- ies (Mello, Okuhata, and Nottebohm, unpublished observa- lumba livia). II. Telencephalic projections of the nucleus ovoidalis thalami. Brain Res 11:134-153. tions), these results suggest that complex telencephalic circuits Katz LC, Gurney ME (198 1) Auditory responses in the zebra finch’s may be involved in the neural processing of auditory stimuli motor system for song. Brain Res 221: 192-l 97. such as song. Thus, it is possible that the selectivity to song Kellev DB. Nottebohm FJ (1979) Proiections of a telenceohalic au- playbacks observed by neurophysiological recording within nu- ditory nucleus-field Llin the canary: J Comp Neurol 183!455469. Konishi M, Akutagawa E (1985) Neuronal growth, atrophy and death clei of the song system results partly from auditory processing in a sexually dimorphic song nucleus in the zebra finch brain. Nature occurring in these other areas. It should be very interesting to 315:145-147. examine in detail the electrophysiological response properties Kroodsma DE (1976) Reproductive development in a female song- of these areas to complex auditory stimuli, and whether and bird: differential stimulation by quality of male song. Science 192: how these responses can be modified. 574-575. Lemaire P, Revelant 0, Bravo R, Chamay P (1988) Two mouse genes encoding potential transcription factors with identical DNA-binding References domains are activated by growth factors in cultured cells. Proc Nat1 Alvarez-Buylla A, Kim JR, Nottebohm F (1990) Birth of projection Acad Sci USA 85:46914695. neurons in adult avian brain may be related to perceptual or motor Margoliash D (1983) Acoustic parameters underlying the responses learning. Science 249:1444-1446. of song-specific neurons in the white-crowned sparrow. J Neurosci Bartel DP, Sheng M, Lau LF, Greenberg ME (1989) Growth factors 3:1039-1057. and membrane depolarization activate distinct programs of early re- Margoliash D (1986) Preference for autogenous song by auditory neu- sponse gene expression: dissociation of fos and jun induction. Genes rons in a song system nucleus of the white-crowned sparrow. J Neu- Dev 3:304-3 13. rosci 6:1643-1661. Bonke BA, Bonke D, Scheich H (1979) Connectivity of the auditory Margoliash D, Fortune E (1992) Temporal and harmonic combina- forebrain nuclei in the guinea fowl (Numida meleagris). Cell Tissue tion-sensitive neurons in the zebra finch’s HVc. J Neurosci 12:4309- Res 200:101-121. 4326. Bonke D, Scheich H, Langer G (1979) Responsiveness of units in the Marler P, Peters S (1977) Selective vocal learning in a sparrow. Science auditory neosttiatum of the guinea fowl (Numida meleagris) to spe- 198:519-527. 8666 Mello and Clayton l ZENK Expression in the Songbird Brain after Song

Mello CV, Vicario DS, Clayton DF (1992) Song presentation induces neostriatum of the starling (Sturnus vulgaris). J Comp Neurol 198: gene expression in the songbird forebrain. Proc Nat1 Acad Sci USA 209-229. 8968 18-6822. Scharff C, Nottebohm F (199 1) A comparative study of the behavioral Milbrandt J (1987) A nerve growth factor-induced gene encodes a deficits following lesions of various parts of the zebra finch song possible transcriptional regulatory factor. Science 238:797-799. system: implications for vocal learning. J Neurosci 11:2896-2913. Miiller CM, Leppelsack H (1985) Feature extraction and tonotopic Scheich H, Bonke BA, Bonke D, Langner D (1979) Functional or- organization in the avian auditory forebrain. Exp Brain Res 59:587- ganization of some auditory nuclei in the guinea fowl demonstrated 599. by the 2-deoxyglucose technique. Cell Tissue Res 204: 17-27. Mtiller SC, Scheich HP (1985) Functional organization of the avian Sohrabji F, Nordeen EJ, Nordeen KW (1990) Selective impairment auditory field L-a comparative 2-deoxyglucose study. J Comp Phys- of song learning following lesions of a forebrain nucleus in the juvenile iol 156:1-12. zebra finch. Behav Neural Biol 53:5 l-63. Nordeen KW, Nordeen EJ (1988) Projection neurons within a vocal Sukhatme VP, Cao X, Chang LC, Tsai-Morris CH, Stamenkovitch D, motor pathway are born during song learning in zebra finches. Nature Ferreira PCP, Cohen DR, Edwards SA, Shows TB, Cut-ran T, LeBeau 344:149-151. MM, Adamson ED ( 1988) A zinc finaer-encodina aene coreeulated Nottebohm F (198 1) A brain for all seasons: cyclical anatomical changes with c-fos during growth and differeniiation, and after cellular de- in song control nuclei of the canary brain. Science 2 14: 1368-l 370. polarization. Cell 53:37-t3. Nottebohm F (1989) From birdsong to neurogenesis. Sci Am 260:74- Vicario DS, Yohay KH (1993) Song-selective auditory input to a 79. forebrain vocal control nucleus in the zebra finch. J Neurobiol 24: Nottebohm F, Stokes T, Leonard CM (1976) Central control of song 488-505. in the canary, Se&us cunarius. J Comp Neurol 165:457486. Welty JC, Baptista L (1988) The life of birds. New York: Saunders. Nottebohm F, Kelley DB, Paton JA (1982) Connections of vocal Wild J M (1987) Nuclei of the lateral lemniscus project directly to the control nuclei in the canary telencephalon. J Comp Neurol207:344- thalamic auditory nuclei in the pigeon. Brain Res 408:303-307. 357. Williams H, Nottebohm F (1985) Auditory responses in avian vocal Nottebohm F, Nottebohm ME, Crane L (1986) Developmental and motor neurons: a motor theorv for song.-- uerceotion_ in birds. Science seasonal changes in canary song and their relation to changes in the 229~279-282. anatomy of song-control nuclei. Behav Neural Biol 46:445-47 1. Wisden W, Errington ML, Williams S, Dunnett SB, Waters C, Hitchcock

Saffen DW, Cole ?IJ, Worley PF, Christy BA, Ryder K, Baraban JM D, Evan G, Bliss TVP, Hunt SP (1990). I Differential exuression of (1988) Convulsant-induced increase in transcription factor messen- immediate early genes in the hippocampus and spinal cord. Neuron aer RNAs in rat brain. Proc Nat1 Acad Sci USA 85:7795-7799. 4:603-6 14. Sa&i KD, Leppelsack H ( 198 1) Cell types of the auditory caudomedial