Efferent Axonal Projections of the Habenular Complex in the Fire-Bellied Toad Bombina orientalis

The habenular complex and its associated axonal pathways are often thought of as phylogenetically conserved features of the brain among vertebrates despite the fact that detailed studies of this brain region are limited to a few species. Here, the gross morphology and axonal projection pattern of the habenular complex of an anuran amphibian, the fire-bellied toad Bombina orientalis, was studied to allow comparison with the situation in other vertebrates. Axonal pathways were traced using biocytin applications in dissected brain preparations. The results show that the rostral part of the left dorsal nucleus is enlarged in this species, while the rostral ventral nucleus and caudal parts do not show left-right size differences. Biocytin applications revealed widespread axo- nal projections of the habenular complex to the posterior tuberculum/dorsal hypothalamic region, ventral tegmen- tum, interpeduncular nucleus (IPN), and raphe median. Ad- ditionally, axons targeting the lateral hypothalamus origi- nated from the ventral habenular nuclei. The results also suggest an asymmetrical pattern of projection to the IPN in the rostral part of the habenular complex, where the left ha- benula preferentially targeted the dorsal IPN while the right habenula preferentially targeted the ventral IPN. The caudal habenular nuclei showed no asymmetry of projections as both sides targeted the ventral IPN. Comparison of the ha- benular complex axonal connectivity across vertebrates ar- gues against strong phylogenetic conservation of the axonal projection patterns of different habenular nuclei.

development [Puelles and Rubenstein, 2003]. This brain region has recently been implicated in the processing of information related to aversive events [Matsumoto and Hikosaka, 2009; Stama- takis and Stuber, 2012; Amo et al., 2014; Lawson et al., 2014; Hennigan et al., 2015]. The habenula often displays left-right asymmetry in size, neurochemistry, and/or con- nectivity [Concha and Wilson, 2001; Villalón et al., 2012]. Medial and lateral nuclei have been recognized in the ha- benular complex of mammals, and these nuclei can be further subdivided into subnuclei based on connectivity or neurochemistry [Herkenham and Nauta, 1977, 1979; Andres et al., 1999; Geisler et al., 2003; Aizawa et al., 2011; Wagner et al., 2014]. Nuclei homologous to the mamma- lian medial and lateral nuclei have been proposed in the habenular complex of other vertebrate groups [Kemali and Làzàr, 1985; Amo et al., 2010; Aizawa et al., 2011; Stephenson-Jones et al., 2012].Across vertebrates, the habenular complex and its pro- jections are thought to be very conservative features of the brain [Butler and Hodos, 1996]. However, this claim is based on few examples of detailed studies of habenular complex anatomy in different vertebrates. For example, studies providing an account of the axonal projections of the different habenular nuclei are limited to laboratory rodents [Herkenham and Nauta, 1979; Kim, 2009; Quina et al., 2015], 1 species of lizard [Distel and Ebbesson, 1981], the zebrafish [Aizawa et al., 2005; Amo et al., 2010], and 1 species of lamprey [Stephenson-Jones et al., 2012]. Efferent projections of the avian habenular complex have not been studied and studies in amphibians are incom- plete (see below). Despite this paucity of studies, differ- ences in habenular complex axonal projection patterns have already been noted across vertebrates.

First, Bianco and Wilson [2009] observed that the lateral habenular nuclei appear to show a less evolutionarily conserved pat- tern of axonal projections compared to the medial ha- benular nuclei. Second, Kuan et al. [2007] proposed that the dorsoventral asymmetry of the habenula to midbrain projection is a unique feature of teleost fishes. Thus, evo- lutionary changes in the habenular complex could have been underestimated and studies of representative spe- cies within understudied groups of vertebrates could prove helpful in understanding the evolution of habenu- lar complex connectivity.The present study aimed to characterize in detail theaxonal projections of the habenular complex in an am- phibian, the anuran Bombina orientalis, in order to com- pare it with other vertebrate groups. B. orientalis is par- ticularly interesting for this purpose because it belongs to a basal group in anuran phylogeny [Pyron and Wiens, 2011] and could provide useful information to elucidate the tetrapod brain morphotype [e.g., Northcutt, 1995] when considering that the more basal caecilians and uro- deles underwent substantial secondary brain simplifica- tion during their evolutionary history [Roth et al., 1997; Schmidt and Wake, 1997]. Experimental studies of ha- benular complex connections have previously been con- ducted in the frog Rana esculenta [Kemali et al., 1980; Kemali and Guglielmotti, 1982; Kemali and Làzàr, 1985; Guglielmotti and Fiorino, 1998], but a precise picture of the projection patterns of different nuclei has yet to be achieved in any amphibian species. In frogs, the habenu- lar complex displays dorsal and ventral nuclei on each side of the brain and the left dorsal nucleus shows con- spicuous lateral and medial divisions. The studies of R. esculenta concluded that both dorsal and ventral nuclei send axonal projections to the interpeduncular nucleus (IPN) and some axons continue their course caudally be- yond this brain region. However, the region of termina- tion of these axons has not been identified. Kuan et al. [2007] also showed that the dorsal habenula of larval am- phibians (anuran R. clamitans and urodele Ambystoma maculatum) send axons to the IPN. There remains a need to ascertain the targets of axons projecting beyond the IPN in amphibians and to verify the presence or absence of projections to the hypothalamus and rostral ventral midbrain present in other vertebrates (e.g., rat [Herken- ham and Nauta, 1979]; lizard [Distel and Ebbesson, 1981]; lamprey [Stephenson-Jones et al., 2012]).

This was at- tempted by using tract tracing of habenular pathways with biocytin, a sensitive tracer substance that can be used for precise applications in in vitro brain preparations of amphibians.Forty-six adult fire-bellied toads of both sexes were used in the present study. The animals were bought from a local supplier (Na- tional Reptile Supply, Mississauga, ON, Canada) and maintained at a temperature of 21 °C under a photoperiod of 12 h of light fol- lowed by 12 h of darkness (lights on at 07:00 h). The toads were housed in groups in glass tanks (37 × 22 × 25 cm) with gravel sub- strate, broken clay pots, and flat stones for cover. They had con- tinuous access to water and were fed crickets (Acheta domesticus) lightly dusted with calcium and vitamin powder ad libitum once weekly. The experimental procedures were approved by the Uni- versity of Guelph animal care committee under the guidelines of the Canadian Council on Animal Care.ProceduresAll experiments were carried out in vitro in isolated brain prep- arations. After deep anesthesia by immersion in a solution of 0.1% tricaine methanesulfonate (Argent Chemical Laboratories, Red- mond, WA, USA), the animals were quickly decapitated, the low- er jaw was removed, and the skull was opened from the roof of the mouth to enable brain dissection. The dissection was performed in Ringer solution consisting of Na+ 129 mM, K+ 4 mM, Ca2+ 2.4 mM, Mg2+ 1.4 mM, Cl– 115 mM, HCO – 25 mM, glucose 10 mM, bubbled with 95% O2/5% CO2 until a pH of 7.3 was achieved. Tract tracing of neural pathways was achieved in 2 ways: (1) by manual applica- tion of biocytin crystals (B4261, Sigma-Aldrich, St. Louis, MO, USA) directly to the lightly lesioned surface of the brain outside of the Ringer bath, and (2) by iontophoretic injection of a 2% solution of biocytin dissolved in 0.3 M KCl. Lesioning of the brain surface was achieved using a pulled glass pipette with a broken tip. Ionto- phoresis was achieved by loading a pulled glass pipette (tip broken at 10–15 μm) with the solution and passing a pulsed current of 4 μA (on/off every 5 s) for 10–20 min.

Additionally, labeling of single neurons by intracellular injections of the 2% biocytin solu- tion was conducted in an attempt to clarify some axonal projection patterns. For the latter, brains were pinned down in a chamber perfused with oxygenated Ringer solution (6 mL/min) and pulled micropipettes with sharp tips were advanced in brain tissue while a 200-ms hyperpolarizing current of 0.2 nA was applied every sec- ond. The current was injected and potential was monitored using an electrometer (Duo 773, World Precision Instruments, Sarasota, FL, USA). Electrode impedance ranged between 80 and 120 MΩ. The penetration of neurons was identified by rebound action po- tential activity and followed by the injection of biocytin by passing a pulsed current of 1 nA for 4 min.After biocytin applications, the brains were stored in oxygen- ated Ringer solution for 5–6 h at room temperature and then at 4 °C overnight. On the next day, the brains were fixed in 2% para- formaldehyde and 2% glutaraldehyde, and then 50-μm-thick transverse sections were cut on a VT1200 vibrating microtome (Leica Biosystems, Wetzlar, Germany). Biocytin was visualized by means of an avidin-biotin horseradish peroxidase complex (Vec- tor Laboratories, Burlingame, CA, USA) by using diaminobenzi- dine (Sigma-Aldrich) as the chromogen with heavy metal intensi- fication achieved by adding 0.03% nickel sulfate and cobalt chlo- ride to the solution [Adams, 1981].

The sections were lightly counterstained with cresyl violet, dehydrated in ethanol, cleared in xylene, and coverslipped before examination under the micro- scope.The assessment of biocytin application sites and charting of retrograde labeling and axonal projections was done using a DM1000 light microscope equipped with a drawing tube (Leica). The intensity of axonal projections and retrograde labeling was as- sessed qualitatively by a single observer (F.L.). Axonal projections were described as notable or weak, with weak labeling correspond- ing to the presence of only 1 or 2 axons with limited varicosities in a given brain region and notable labeling corresponding to the presence of multiple axons with abundant varicosities along their length in a brain region. Retrograde labeling was described as strong, moderate, or weak depending on the number of cell bodies labeled in a given brain region, as exemplified in online supple- mentary Figure S1 (see for all online suppl. material). Photomicrographs were scanned using an Eclipse 90i upright microscope (Nikon, Tokyo, Japan) equipped with a Retiga 2000R digital camera (QImaging, Surrey, BC, Canada), modified (sized and cropped) and optimized for pre- sentation (brightness and contrast) using Adobe Photoshop CS3 (Adobe Systems, San Jose, CA, USA). Analysis of potential asym- metry of the habenular complex was done by measuring the sur- face area of habenular divisions on serial sections using the con- tour drawing function in Neurolucida version 11.02.1 (MBF Bio- science, Williston, VT, USA) on the Eclipse 90i microscope. Such comparison of the rostral habenula involved the 4 most ros- tral sections, while the caudal habenula involved the 3 most caudal sections. Statistical analyses were done in Prism version 5.04 (GraphPad Software Inc., San Diego, CA, USA). The neuroana- tomical framework used for presentation of the data is based on published accounts in R. perezi [Puelles et al., 1996], R. catesbeiana [Neary and Northcutt, 1983], and the fire-bellied toad [Laberge and Roth, 2007; Laberge et al., 2008].

Figure 1 illustrates the extent of the habenular complex in the fire-bellied toad brain. The specimen chosen for this purpose received a large application of biocytin that covered part of the ventral tegmentum (vTEG) and the dorsoventral extent of the IPN and rostral raphe median (mRaphe). The abundant retrograde labeling that result- ed from this tracer application allowed a clear outline of the extent of the habenular complex. The labeling shows equivalent staining in the left and right caudal habenula, but more prominent staining in the left rostral habenula. The higher magnification inset included in Figure 1F shows details of the dorsal and ventral nuclei on each side of the brain as well as the lateral and medial divisions vis- ible in the middle left dorsal habenula nucleus. Such divi- sions in the left dorsal habenula nucleus were also noted in the frog R. esculenta [Kemali and Làzàr, 1985; Gugliel- motti and Fiorino, 1998]. The dorsal nuclei (especially in their middle part) are structured as cellular rims sur- rounding central neuropils, while ventral nuclei are made of homogeneously distributed cell bodies. Note that there is high interindividual variability in the divisions of the dorsal habenular nuclei in their middle part.Five specimens displaying abundant retrograde label- ing in the habenular complex were chosen for morpho- metric analysis of the habenular nuclei on each side of the brain. Note that the volumes of the lateral and medial di- visions in the middle left dorsal habenula nucleus were not calculated because their boundaries were difficult to estimate.

A 2-tailed paired t test showed that the volume of the left habenula (mean ± SD: 0.017 ± 0.0009 mm3) is larger than the right habenula (0.013 ± 0.0008 mm3; t4 = 15.8, p < 0.0001). Further analysis showed no difference between volumes of the left and right caudal habenula (left caudal: 0.0032 ± 0.0005 mm3, right caudal: 0.0027 ± 0.0006 mm3; t4 = 1.7, p = 0.18) or rostral ventral habenu- la (left rostral ventral: 0.0042 ± 0.0003 mm3, right rostral Surface areas were converted to volumes by multiplying values bysection thickness. Comparison of the left and right habenular vol- ventral: 0.0044 ± 0.0003 mm3; t4 = 1.2, p = 0.30). How- umes involved all sections comprising the habenular complex. ever, the volume of the left rostral dorsal habenula (0.005 habenula nucleus in panel B is from a postfixation artifact. AT, anterior thalamic nucleus; CT, central thalamic nucleus; dHb, dor- sal habenula nucleus; lat. dHb, laterodorsal habenula subnucleus; lHYP, lateral hypothalamic nucleus; LT, lateral thalamic nucleus; med. dHb, mediodorsal habenula subnucleus; MP, medial palli- um; POA, preoptic area; SCN, suprachiasmatic nucleus; TE, tha- lamic eminence; vHb, ventral habenula nucleus; VL, ventrolateral thalamic nucleus; VT, ventral thalamic nucleus; 2sp, second spinal nerve. ± 0.0005 mm3) was larger than that of the right rostral dorsal habenula (0.002 ± 0.0004 mm3; t4 = 13.9, p = 0.0002). Therefore, left-right asymmetry of the habenular complex in the fire-bellied toad is the result of an enlarged rostral part of the dorsal nucleus on the left side. Efferent Axonal Projections of the Habenular Complex Figure 2 shows a typical example of the axonal projec- tions revealed by biocytin applications directly to the ha- benular nuclei. In this case, the tracer application in- volved both dorsal and ventral nuclei along most of the rostrocaudal extent of the habenula on the left side of the brain (Fig. 2A, B, B′). Output axons of the habenular nu- clei form the fasciculus retroflexus, which runs from the dorsal diencephalon to the posterior tuberculum and dorsal hypothalamus region on the side ipsilateral to the application site (Fig. 2C). Some axonal varicosities are ob- served near the rostral vTEG and caudal dorsal hypothal- amus, but no clear axon branches are seen there (Fig. 2D). Abundant varicose axons were seen in the ventral IPN at a more rostral level (Fig. 2E) and both the ventral and dorsal IPN more caudally in this case (Fig. 2F). Finally, varicose axons extended caudally into the ventral part of mRaphe of the medulla oblongata (Fig. 2G). Axons cross over the brain midline multiple times to invade both sides of the IPN and they maintain this central, bilateral loca- tion in mRaphe.Table 1 summarizes the results of 16 anterograde la- beling experiments. The extent of these biocytin applica- tion sites is detailed in online supplementary Figure S2. The majority of biocytin applications to the habenular complex revealed consistent axonal projections to the posterior tuberculum/dorsal hypothalamus region (Fig. 3b), rostral vTEG (Fig. 3c), IPN (Fig. 3d–e), and mRaphe (Fig. 3f). Projections to mRaphe were limited to the anterior part of this region. Axonal projections to the contralateral habenula neuropil were often seen, but no obvious pattern suggesting a topographic organization emerged when comparing different applications. Projec- tions to the lateral hypothalamus (lHYP; Fig. 3a), and possibly the preoptic area (POA), appear to originate from the ventral habenular nuclei, when taking into ac- count the potential projections due to inclusion in some application sites of parts of the dorsal thalamus which are known to send axons to the lHYP/POA [Laberge et al., 2008]. Axons found in the commissural pretectum and the median thalamic neuropil may also be due to inclu- sion of parts of the thalamus in some application sites, but this problem would require further investigation due to a limited sample size of applications displaying such con- nections.An asymmetric pattern of axonal projections to the IPN was noticed: while applications to the caudal haben- ula targeted the ventral part of the IPN, applications to the rostral part of the habenula showed a left-right side dif- ference. The rostral habenula on the left side preferen- tially targeted the dorsal IPN (Fig. 3d), while the rostral habenula on the right side preferentially targeted the ven- tral IPN (Fig. 3e). This pattern is reminiscent of the dor- soventral asymmetry of projections of left and right dor- sal habenulae in zebrafish [Aizawa et al., 2005; Amo et al., 2010] and contrasts with the symmetrical habenular pro- jections to IPN seen in larval amphibians [Kuan et al., 2007]. This dorsoventral asymmetry of projections to IPN in the fire-bellied toad was confirmed by intracellular labeling of neurons. In 1 case, 1 neuron with its cell body in the rostral part of the left dorsal habenula sent 2 axons (or axon collaterals) into the ipsilateral fasciculus retro- flexus and targeted the whole rostrocaudal extent of the IPN in its dorsal part (Fig. 4a). In another case, 2 neurons labeled by the same intracellular injection had their cell bodies in the rostral part of the right dorsal habenula and sent axons that ended in the ventral part of the IPN (Fig. 4b). Few successful intracellular injections of haben- ular complex neurons were obtained due to difficulties with intracellular recording in this brain region using a dorsal approach in intact brain preparations.Biocytin is also taken up at synaptic sites and moved back toward the cell body to produce retrograde labeling tion in panel e is the same as in d, but in a different animal. Insets High-power micrographs showing examples of axonal varicosities (arrowheads). Scale bars, 0.5 mm (a–f), 0.05 mm (insets). The specimens from which the micrographs were taken are Hb 11 (a), Hb 7 (b, c), Hb 12 (d), Hb 6 (e), and Hb 1 (f). See Table 1 for a full description of the axonal projection sites. of neurons. Retrograde labeling resulting from applica- tions of biocytin to the habenular complex is summarized in online supplementary Table S1. This analysis showed that the main afferent brain region to the habenular com- plex in the fire-bellied toad is the bed nucleus of the pal- lial commissure/thalamic eminence continuum (12 out of 16 applications), a projection that was previously de- scribed by Laberge and Roth [2007]. In this previous axon targeting the ventral IPN (arrowheads in the bottom panel). The locations of the micrographs are indicated by boxes on sche- matic transverse brain sections to the right. The cresyl violet coun- terstaining procedure did not function properly in a due to wax contamination of the solvent used to prepare the slides. Scale bars,0.1 mm. of lHYP – or the ventral thalamus – just dorsal of lHYP – did not produce any retrograde labeling in the habenular complex. Additionally, retrograde labeling experiments suggest a strong projection of the rostral ventral habenular nuclei to mRaphe. Finally, it should be noted that retro- grade labeling in the rostral dorsal habenula was strongest in the tracer application that involved the dorsal IPN, which is hard to reach when applying crystalline biocytin on the surface of the brain.Table 2 shows retrograde labeling of neurons in the ha- benular complex following biocytin applications to other brain regions (n = 21 experiments). The extent of these biocytin applications sites is detailed in online supplemen- tary Figure S3 and the qualitative scale of retrograde label- ing intensity is illustrated by an example in online supple- mentary Figure S1. These results confirm the strong con- nection between the caudal habenula and the IPN/mRaphe. In accordance with the projection pattern described above, applications lateral to the IPN or in the caudal part of mRaphe did not label any neurons in the habenular com- plex. The results also confirm a moderate projection of the ventral nuclei to lHYP. Applications restricted to the POA and suprachiasmatic hypothalamic nucleus – just rostral. Discussion Figure 5 summarizes the axonal projections of the ha- benular complex in the adult fire-bellied toad. The varia- tion in axonal projections between habenular nuclei in- volves both left-right differences in innervation of the IPN and differences between the dorsal and ventral nuclei in innervation of lHYP. The results also suggest extensive overlap of axonal projections between nuclei. The mor- phological asymmetry of the anuran habenular complex detected previously [Kemali and Làzàr, 1985; Gugliel- motti and Fiorino, 1998] and in our analysis is restricted to the rostral part of the dorsal nucleus.The amount of retrograde labeling following biocytin applications depended on the depth of lesions made to the brain surface in the case of crystalline applications, and whether more than 1 region was included in the applica- tion site in all cases. Conclusions reached here were in- ferred from consistent patterns between different applica- tions and the use of both anterograde and retrograde trac- ing for confirmation of connections between brain regions. One projection site from the habenular complex that proved difficult to confirm in the fire-bellied toad is lHYP. All anterograde tracing experiments where vari- cose axons were seen in lHYP or POA, except one, in- volved retrograde labeling of neurons located in the ven- tral thalamus. It is possible that retrogradely labeled neu- rons in the ventral thalamus could send axons to lHYP, rather than neurons of the habenular complex. However, we reject this possibility because tracer injections in the ventral thalamus itself, in its medial or lateral parts, did not result in retrograde labeling in the habenular com- plex. Overall, evidence from biocytin applications involv- ing the POA, suprachiasmatic nucleus, and lHYP suggest that only the caudal parts of lHYP receive axonal input from the habenular complex, and that this input is lim- ited to the ipsilateral ventral nuclei. The present study clarified and expanded the known habenular complex axonal projection sites in anuran am- phibians. If the results in the fire-bellied toad are broadly applicable to the situation in other anurans, then in addi- tion to the IPN [Kemali and Làzàr, 1985; Kuan et al., 2007], the anuran habenular complex also targets lHYP, the posterior tuberculum/dorsal hypothalamus region, rostral vTEG, and rostral mRaphe. The projection to mRaphe was anticipated from the finding of habenula ax- ons projecting beyond the IPN in R. esculenta by Kemali and Làzàr [1985]. The results in the fire-bellied toad also suggest that the topography of habenular complex axonal projections to the IPN is variable within amphibians. The symmetric IPN innervation by right and left dorsal ha- benulae shown by Kuan et al. [2007] differs from the IPN dorsoventral innervation topography of the rostral ha- benular nuclei seen here. Since Kuan et al. [2007] studied larval amphibians and newborn mice, it is unclear if this difference is due to ontogenetic differences in IPN inner- vation patterns or species differences. Nevertheless, this finding suggests that a distinct dorsoventral patterning of left and right habenular projections to the IPN is not unique to teleosts. However, our confirmation of distinct rostral habenular projections to the ventral and dorsal IPN obtained by anterograde labeling in the fire-bellied toad is limited to retrograde labeling from a single appli- cation to the deep IPN and 2 intracellular injections. Fur- ther experiments such as tracer applications restricted to the dorsal and ventral IPN would be needed for convinc- ing confirmation of the dorsoventral asymmetry of this projection in adult amphibians.Considering that the habenular complex and its pro- jections are thought of as conservative features of the brain across vertebrates, the plesiomorphic (or basal) state of habenular projection patterns should be easy to determine. However, when comparing known axonal tar- gets of the habenular complex across vertebrates in a phy- logenetic context (Table 3), some difficulties arise. Name- ly, there are fewer targets, and distinct targets between habenular nuclei, in lampreys and teleosts compared to amphibians (see Table 3 for references). Additionally, both habenular nuclei show differences in axonal targets across groups, arguing against the proposal of Bianco and Wilson [2009] suggesting that medial habenular nuclei show a more conservative evolutionary pattern in verte- brates. A reconsideration of the habenular complex as a vertebrate brain region with more labile axonal connec- tions invites different evolutionary scenarios. In one sce- nario, the patterns of habenular complex efferents in fish could be derived features evolved from more widespread ancestral habenular projections, the latter of which would have been kept in amphibians. Habenular complex pro- jections are especially simpler in teleosts, where they are limited to a single brain region for each habenula nucleus [Amo et al., 2010]. The divergent topography of habenu- lar projections to IPN between lampreys (rostrocaudal topography) and teleosts (dorsoventral topography) also suggests that habenular projection patterns in fish saw some evolutionary changes. In an alternative scenario, ancestral amphibians would have gained many habenula axonal targets compared to the simpler situation in fish, and from there evolution would have again refined the projections of the medial and lateral habenular nuclei to a set of specific brain regions in amniotes (here, lizards and rodents for which we have data). A final evolutionary scenario posits that the lineage leading to modern am- phibians, not the common ancestor of tetrapods, would have gained many habenula axonal targets. In this regard, it is interesting to note the similarities of habenular com- plex projections between lampreys and amniotes. Ste- phenson-Jones et al. [2012] showed that the habenular nuclei in the river lamprey (Lampetra fluviatilis) have similarly broad brain targets as in amniotes, with the ex- ception of projections to the raphe, which are absent. In this final scenario, the habenular complex in basal verte- brates would have targeted many brain regions with a dis- tinct pattern between nuclei somewhat similar to the situ- ation in lampreys and amniotes, and from there teleosts saw a reduction in habenula targets, while modern am- phibians saw an increase in habenula targets for both the dorsal and ventral nuclei. Conclusion The present study expanded the known habenular projection targets in anuran amphibians to include the hypothalamus, tegmentum, and median raphe. The breadth of axonal targets shown for both habenular nu- clei in the fire-bellied toad is unique among vertebrates that have been studied so far and presents a problem for establishing a plausible scenario for the evolution of ha- benular complex efferents. The finding suggests that con- servation of habenular complex connectivity among ver- tebrates could have been overestimated, and opens ques- tions about the evolution of this brain structure. More research on the connectivity and neurochemistry of ha- benular nuclei in animals strategically placed in verte- brate phylogeny will be needed to elucidate this problem. In the meantime, the use of caution is suggested when proposing homologies between brain nuclei in different groups even when dealing with a seemingly conserved structure such as the habenular Biocytin complex.