Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Dense EM-based reconstruction of the interglomerular projectome in the zebrafish olfactory bulb


The dense reconstruction of neuronal circuits from volumetric electron microscopy (EM) data has the potential to uncover fundamental structure–function relationships in the brain. To address bottlenecks in the workflow of this emerging methodology, we developed a procedure for conductive sample embedding and a pipeline for neuron reconstruction. We reconstructed 98% of all neurons (>1,000) in the olfactory bulb of a zebrafish larva with high accuracy and annotated all synapses on subsets of neurons representing different types. The organization of the larval olfactory bulb showed marked differences from that of the adult but similarities to that of the insect antennal lobe. Interneurons comprised multiple types but granule cells were rare. Interglomerular projections of interneurons were complex and bidirectional. Projections were not random but biased toward glomerular groups receiving input from common types of sensory neurons. Hence, the interneuron network in the olfactory bulb exhibits a specific topological organization that is governed by glomerular identity.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Figure 1: EE embedding.
Figure 2: Reconstruction of neurons by CORE.
Figure 3: Quantification of reconstruction accuracy.
Figure 4: Subcellular distribution of synapses.
Figure 5: Neurons of the larval zebrafish OB.
Figure 6: Associations between neurons and glomeruli.
Figure 7: Synaptic innervation of glomeruli by INs.
Figure 8: Organization of interglomerular projections of INs.


  1. 1

    Briggman, K.L. & Bock, D.D. Volume electron microscopy for neuronal circuit reconstruction. Curr. Opin. Neurobiol. 22, 154–161 (2012).

    CAS  Google Scholar 

  2. 2

    Hayworth, K.J. et al. Imaging ATUM ultrathin section libraries with WaferMapper: a multi-scale approach to EM reconstruction of neural circuits. Front. Neural Circuits 8, 68 (2014).

    PubMed  PubMed Central  Google Scholar 

  3. 3

    Denk, W. & Horstmann, H. Serial block-face scanning electron microscopy to reconstruct three-dimensional tissue nanostructure. PLoS Biol. 2, e329 (2004).

    PubMed  PubMed Central  Google Scholar 

  4. 4

    Hayworth, K.J., Kasthuri, N., Schalek, R. & Lichtman, J.W. Automating the collection of ultrathin serial sections for large volume TEM reconstructions. Microsc. Microanal. 12, 86–87 (2006).

    Google Scholar 

  5. 5

    Lichtman, J.W. & Denk, W. The big and the small: challenges of imaging the brain's circuits. Science 334, 618–623 (2011).

    CAS  Google Scholar 

  6. 6

    Denk, W., Briggman, K.L. & Helmstaedter, M. Structural neurobiology: missing link to a mechanistic understanding of neural computation. Nat. Rev. Neurosci. 13, 351–358 (2012).

    CAS  Google Scholar 

  7. 7

    Titze, B. & Denk, W. Automated in-chamber specimen coating for serial block-face electron microscopy. J. Microsc. 250, 101–110 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  8. 8

    Robinson, V.N.E. The elimination of charging artefacts in the scanning electron microscope. J. Phys. E Sci. Instrum. 8, 638–640 (1975).

    Google Scholar 

  9. 9

    Mathieu, C. Beam-gas and signal-gas interactions in the variable pressure scanning electron microscope. Scanning Microsc. 13, 23–41 (1999).

    Google Scholar 

  10. 10

    Egerton, R.F., Li, P. & Malac, M. Radiation damage in the TEM and SEM. Micron 35, 399–409 (2004).

    CAS  PubMed  Google Scholar 

  11. 11

    Braitenberg, V. & Schüz, A. Cortex: Statistics and Geometry of Neuronal Connectivity (Springer, 1998).

  12. 12

    White, J.G., Southgate, E., Thomson, J.N. & Brenner, S. The structure of the nervous system of the nematode Caenorhabditis elegans. Phil. Trans. R. Soc. Lond. B 314, 1–340 (1986).

    CAS  Google Scholar 

  13. 13

    Helmstaedter, M. et al. Connectomic reconstruction of the inner plexiform layer in the mouse retina. Nature 500, 168–174 (2013).

    CAS  Google Scholar 

  14. 14

    Kim, J.S. et al. Space-time wiring specificity supports direction selectivity in the retina. Nature 509, 331–336 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15

    Helmstaedter, M., Briggman, K.L. & Denk, W. High-accuracy neurite reconstruction for high-throughput neuroanatomy. Nat. Neurosci. 14, 1081–1088 (2011).

    CAS  Google Scholar 

  16. 16

    Varshney, L.R., Chen, B.L., Paniagua, E., Hall, D.H. & Chklovskii, D.B. Structural properties of the Caenorhabditis elegans neuronal network. PLoS Comput. Biol. 7, e1001066 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  17. 17

    Takemura, S.Y. et al. A visual motion detection circuit suggested by Drosophila connectomics. Nature 500, 175–181 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  18. 18

    Helmstaedter, M. Cellular-resolution connectomics: challenges of dense neural circuit reconstruction. Nat. Methods 10, 501–507 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  19. 19

    Fantana, A.L., Soucy, E.R. & Meister, M. Rat olfactory bulb mitral cells receive sparse glomerular inputs. Neuron 59, 802–814 (2008).

    CAS  PubMed  PubMed Central  Google Scholar 

  20. 20

    Axel, R. The molecular logic of smell. Sci. Am. 273, 154–159 (1995).

    CAS  PubMed  Google Scholar 

  21. 21

    Wilson, R.I. & Mainen, Z.F. Early events in olfactory processing. Annu. Rev. Neurosci. 29, 163–201 (2006).

    CAS  PubMed  PubMed Central  Google Scholar 

  22. 22

    Friedrich, R.W. Neuronal computations in the olfactory system of zebrafish. Annu. Rev. Neurosci. 36, 383–402 (2013).

    CAS  PubMed  Google Scholar 

  23. 23

    Tapia, J.C. et al. High-contrast en bloc staining of neuronal tissue for field emission scanning electron microscopy. Nat. Protoc. 7, 193–206 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  24. 24

    Deerinck, T.J. et al. Enhancing serial block-face scanning electron microscopy to enable high resolution 3-D nanohistology of cells and tissues. Microsc. Microanal. 16, 1138–1139 (2010).

    CAS  Google Scholar 

  25. 25

    Briggman, K.L., Helmstaedter, M. & Denk, W. Wiring specificity in the direction-selectivity circuit of the retina. Nature 471, 183–188 (2011).

    CAS  Google Scholar 

  26. 26

    Pinching, A.J. & Powell, T.P. The neuropil of the glomeruli of the olfactory bulb. J. Cell Sci. 9, 347–377 (1971).

    CAS  PubMed  Google Scholar 

  27. 27

    Fuller, C.L., Yettaw, H.K. & Byrd, C.A. Mitral cells in the olfactory bulb of adult zebrafish (Danio rerio): morphology and distribution. J. Comp. Neurol. 499, 218–230 (2006).

    PubMed  Google Scholar 

  28. 28

    Miyasaka, N. et al. Olfactory projectome in the zebrafish forebrain revealed by genetic single-neuron labelling. Nat. Commun. 5, 3639 (2014).

    PubMed  Google Scholar 

  29. 29

    Li, J. et al. Early development of functional spatial maps in the zebrafish olfactory bulb. J. Neurosci. 25, 5784–5795 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  30. 30

    Miyasaka, N. et al. From the olfactory bulb to higher brain centers: genetic visualization of secondary olfactory pathways in zebrafish. J. Neurosci. 29, 4756–4767 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  31. 31

    Rosselli-Austin, L. & Altman, J. The postnatal development of the main olfactory bulb of the rat. J. Dev. Physiol. 1, 295–313 (1979).

    CAS  PubMed  Google Scholar 

  32. 32

    Yaksi, E., Judkewitz, B. & Friedrich, R.W. Topological reorganization of odor representations in the olfactory bulb. PLoS Biol. 5, e178 (2007).

    PubMed  PubMed Central  Google Scholar 

  33. 33

    Wiechert, M.T., Judkewitz, B., Riecke, H. & Friedrich, R.W. Mechanisms of pattern decorrelation by recurrent neuronal circuits. Nat. Neurosci. 13, 1003–1010 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  34. 34

    Mack-Bucher, J.A., Li, J. & Friedrich, R.W. Early functional development of interneurons in the zebrafish olfactory bulb. Eur. J. Neurosci. 25, 460–470 (2007).

    PubMed  Google Scholar 

  35. 35

    Lindsay, S.M. & Vogt, R.G. Behavioral responses of newly hatched zebrafish (Danio rerio) to amino acid chemostimulants. Chem. Senses 29, 93–100 (2004).

    PubMed  Google Scholar 

  36. 36

    Zhu, P., Frank, T. & Friedrich, R.W. Equalization of odor representations by a network of electrically coupled inhibitory interneurons. Nat. Neurosci. 16, 1678–1686 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  37. 37

    Kiyokage, E. et al. Molecular identity of periglomerular and short axon cells. J. Neurosci. 30, 1185–1196 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  38. 38

    Braubach, O.R., Fine, A. & Croll, R.P. Distribution and functional organization of glomeruli in the olfactory bulbs of zebrafish (Danio rerio). J. Comp. Neurol. 520, 2317–2339, Spc1 (2012).

    Google Scholar 

  39. 39

    Braubach, O.R. et al. Experience-dependent versus experience-independent postembryonic development of distinct groups of zebrafish olfactory glomeruli. J. Neurosci. 33, 6905–6916 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  40. 40

    Dynes, J.L. & Ngai, J. Pathfinding of olfactory neuron axons to stereotyped glomerular targets revealed by dynamic imaging in living zebrafish embryos. Neuron 20, 1081–1091 (1998).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. 41

    Baier, H. & Korsching, S. Olfactory glomeruli in the zebrafish form an invariant pattern and are identifiable across animals. J. Neurosci. 14, 219–230 (1994).

    CAS  PubMed  PubMed Central  Google Scholar 

  42. 42

    Eyre, M.D., Antal, M. & Nusser, Z. Distinct deep short-axon cell subtypes of the main olfactory bulb provide novel intrabulbar and extrabulbar GABAergic connections. J. Neurosci. 28, 8217–8229 (2008).

    CAS  PubMed  PubMed Central  Google Scholar 

  43. 43

    Stocker, R.F., Lienhard, M.C., Borst, A. & Fischbach, K.F. Neuronal architecture of the antennal lobe in Drosophila melanogaster. Cell Tissue Res. 262, 9–34 (1990).

    CAS  PubMed  PubMed Central  Google Scholar 

  44. 44

    Chou, Y.H. et al. Diversity and wiring variability of olfactory local interneurons in the Drosophila antennal lobe. Nat. Neurosci. 13, 439–449 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  45. 45

    Olsen, S.R., Bhandawat, V. & Wilson, R.I. Divisive normalization in olfactory population codes. Neuron 66, 287–299 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  46. 46

    Banerjee, A. et al. An interglomerular circuit gates glomerular output and implements gain control in the mouse olfactory bulb. Neuron 87, 193–207 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  47. 47

    Hansen, A. et al. Correlation between olfactory receptor cell type and function in the channel catfish. J. Neurosci. 23, 9328–9339 (2003).

    CAS  PubMed  PubMed Central  Google Scholar 

  48. 48

    Friedrich, R.W. & Korsching, S.I. Combinatorial and chemotopic odorant coding in the zebrafish olfactory bulb visualized by optical imaging. Neuron 18, 737–752 (1997).

    CAS  Google Scholar 

  49. 49

    Walton, J. Lead aspartate, an en bloc contrast stain particularly useful for ultrastructural enzymology. J. Histochem. Cytochem. 27, 1337–1342 (1979).

    CAS  PubMed  Google Scholar 

  50. 50

    Westerfield, M. The Zebrafish Book. A Guide for the Laboratory use of Zebrafish (Danio rerio) (University of Oregon Press, 2000).

  51. 51

    Tallini, Y.N. et al. Imaging cellular signals in the heart in vivo: cardiac expression of the high-signal Ca2+ indicator GCaMP2. Proc. Natl. Acad. Sci. USA 103, 4753–4758 (2006).

    CAS  PubMed  Google Scholar 

  52. 52

    Akerboom, J. et al. Optimization of a GCaMP calcium indicator for neural activity imaging. J. Neurosci. 32, 13819–13840 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  53. 53

    Ahrens, M.B., Orger, M.B., Robson, D.N., Li, J.M. & Keller, P.J. Whole-brain functional imaging at cellular resolution using light-sheet microscopy. Nat. Methods 10, 413–420 (2013).

    CAS  Google Scholar 

  54. 54

    Ahrens, M.B. et al. Brain-wide neuronal dynamics during motor adaptation in zebrafish. Nature 485, 471–477 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  55. 55

    Zhu, P., Fajardo, O., Shum, J., Zhang Schärer, Y.-P. & Friedrich, R.W. High-resolution optical control of spatiotemporal neuronal activity patterns in zebrafish using a digital micromirror device. Nat. Protoc. 7, 1410–1425 (2012).

    CAS  PubMed  Google Scholar 

  56. 56

    Cuntz, H., Forstner, F., Borst, A. & Hausser, M. One rule to grow them all: a general theory of neuronal branching and its practical application. PLoS Comp. Biol. (2010).10.1371/journal.pcbi.1000877

Download references


We thank C. Bleck, B. Anderson and B. Titze for inputs and comments on the manuscript, and we thank W. Ong, L. Ong and R. Ong for help with tracer training and management. This work was supported by the Novartis Research Foundation, the Human Frontiers Science Program (HFSP) and the Swiss National Science Foundation (SNF).

Author information




A.A.W. participated in all tasks. He developed EE embedding, CORE, PyKNOSSOS and analysis procedures. He acquired images from the larval OB, founded ariadne-service, supervised tracers, reconstructed all neurons, annotated synapses, analyzed the data, and participated in writing the manuscript. C.G. participated in sample preparation, methods development, and image acquisition. T.M. participated in sample preparation and image acquisition of the stack from the adult OB. L.S. participated in sample preparation. R.W.F. participated in data analysis and wrote the manuscript.

Corresponding author

Correspondence to Rainer W Friedrich.

Ethics declarations

Competing interests

Part of the results disclosed herein have been included in European patent application EP14736451 and US patent application US14897514. A.A.W. is the founder and owner of ariadne-service.

Integrated supplementary information

Supplementary Figure 1 SBEM imaging and EE embedding.

(a) Images of the same EE-embedded sample acquired in high vacuum using a lvSEM (left; EL = 1.5 keV, De = 17 enm–2) or a vpSEM (right; EL = 2 kV, De = 19 enm–2). Nominal resolution was 4 × 4 nm2 pixels. (b) Images of the same sample (zebrafish telencephalon) acquired at different rates (0.2 MHz – 2 MHz; EL = 1.5 keV; 4 × 4 nm2 pixels). Beam current was adjusted to maintain De within a narrow range (88 enm–1 for 50 pA; 71 enm–1 otherwise). (c) Images of the same sample (zebrafish telencephalon) acquired at different pressures in a vpSEM and in a lvSEM. HV, high vacuum. Other imaging parameters were identical (EL = 2 keV; acquisition rate 200 kHz; beam current ca. 60 pA; pixel size 9 × 9 nm2). Each image is a subregion of the third image in a small stack (50 images; 2,048 × 2,048 pixels; 30 nm section thickness) taken in the same field of view. Images were autoscaled and low-pass filtered to attenuate shot noise. (d) Signal-to-noise ratio (SNR) as a function of spatial frequency for images from four different samples acquired at different pressures in a vpSEM and in a lvSEM. Three samples (adult telencephalon, adult OB, larval OB) were embedded in EE while one sample (larval OB) was embedded in standard EPON (non-EE). For each sample and condition, small stacks of images (usually 50 images; 2,048 × 2,048 pixels; 30 nm section thickness) were acquired in the same five regions of interest (ROIs) using fixed parameters (EL = 2 keV; acquisition rate 200 kHz; beam current ca. 60 pA; pixel size 9 × 9 nm2). Each stack was inspected for regular sectioning and charging artifacts. Regular sectioning without charging artifacts could not be achieved in the non-EE-embedded sample at pressures <20 Pa. These stacks were therefore excluded from quantitative SNR analysis. To determine the SNR, two-dimensional spatial power spectra were computed, averaged over images and ROIs, collapsed into one dimension by averaging in circular bins, and expressed in dB (10 dB correspond to a factor of 10). Power spectra of noise images (electron beam turned off) were computed using the same procedure and subtracted. The resulting spectra therefore represent the logarithm of the signal power-to-noise power ratio (SNR) as a function of spatial frequency. To compare SNR curves for samples from the same tissue (larval OB), curves for the EE-embedded sample (Larval OB, EE) are copied onto the corresponding curves for the non-EE-embedded sample (Larval OB, non-EE; right panel) as dashed lines.

Supplementary Figure 2 High-speed SBEM imaging of a large volume.

Single image from a large stack acquired at high speed (adult zebrafish OB; EE-embedding; EL = 1.5 keV; De = 15.9 – 17.3 enm–2; 2 MHz; 9 × 9 × 25 nm3 voxels; 173 × 288 × 98 μm3; 3,918 sections; high vacuum; lvSEM). The image consists of 40 tiles, each with 4,096 × 4,096 pixels. Inset shows zoom-in.

Supplementary Figure 3 PyKNOSSOS.

Screenshot of PyKNOSSOS in tracing mode. The four viewports on the left show reslices through the original image stack (here, the upper left shows the x-y plane, the lower right is orthogonal to the traced process shown in red, and the remaining two viewports are orthogonal to the lower right viewport. Other viewport assignments are possible). The fifths viewport shows reconstructions. Zoom levels, translations, rotations, and superimpositions with image data can be adjusted in real time. Panels on the right provide access to different functions for tracing, annotation and visualization (see manual in Supplementary Methods). A tab for synapse annotation is shown on the lower right. Images in the viewports are from the olfactory bulb of a zebrafish larva (4.5 dpf) and were acquired with a vpSEM in high vacuum (EL = 2 keV, De = 17.5 enm–2, 9.25 × 9.25 × 25 nm voxels).Ergonomic software is important to minimize the amount of human labor in circuit reconstruction. We generated a software package for manual skeleton tracing, synapse annotation and visualization with two goals in mind. First, it should be written in a widely used programming language to facilitate modifications and extensions by users. Second, it should allow for the real-time visualization of large and complex reconstructions (hundreds of neurons) together with raw image data. Existing software packages used for neuron reconstruction include KNOSSOS (; Helmstaedter, M., et al. [2011] Nat. Neurosci. 14, 1081-1088), CATMAID (; Saalfeld, S., et al. [2009] Bioinformatics 25, 1984-1986), ilastik (; Sommer, C., et al. [2011] 8th Internat. Symp. Biomed. Imaging [ISBI] Proceedings, 230-233), and others. KNOSSOS and CATMAID are specifically designed for large-scale reconstruction projects. Inspired by these tools, we created new software that is written in PYTHON and uses the VTK library for 3D visualization ( The software was developed by close interactions between programmer and user and is named PyKNOSSOS because the user interface is closely related to KNOSSOS. A manual with screenshots can be found in Supplementary Software Information. The code for PyKNOSSOS is available at Further information is available at features of PyKNOSSOS include the following:1. Efficient rendering of reconstructed neurons using VTK. Using the visualization toolkit (VTK;, PyKNOSSOS can interactively visualize hundreds of reconstructed neurons in 3D on a standard desktop computer together with the underlying image data. This functionality is important for efficient data browsing, tracing and error correction. For higher-level analyses, visualization in PyKNOSSOS can be integrated into custom workflows. Moreover, PyKNOSSOS can be used to generate 3D renderings and animated displays of raw data and reconstructions.2. Multi-resolution view. Like KNOSSOS, PyKNOSSOS dynamically loads cubes of data into RAM as users navigate through a volume. This allows users to browse through large datasets (>1TB) with minimal RAM requirements. Storing multiple sets of cubes at different resolutions allows for seamless zooming through large volumes and the extraction of virtual reslices with arbitrary orientation and zoom-level.3. Virtual reslicing of the raw data orthogonal to local processes. Tracing of neurites and branch point detection should be particularly efficient when users view a section through the EM data volume that is orthogonal to the process being traced. Such virtual reslices should also facilitate the identification of synapses because presynaptic vesicles, the synaptic density, and the postsynapse are contained in the same view. PyKNOSSOS automatically calculates a rotation-minimized “locally orthogonal” section (Wang, W., et al. [2008] ACM Trans. Graph. 27, 2) during tracing and presents it in a separate viewport, in addition to the cardinal cross-sections and the imaging plane. We found that the “locally orthogonal” view facilitates the detection of branch points and increases tracing speed.4. Synapse annotation tools. PyKNOSSOS includes tools to define the location and direction of a synapse by three successive clicks on the presynaptic process, the synaptic density, and the postsynaptic process. In addition, synapses can be assigned to user-defined classes by a confidence level. Furthermore, pre-calculated “flight paths” can be loaded to automatically visit all branches of a reconstructed neuron. Using this mode, only two clicks on the synaptic density and postsynapse are required to define a synaptic connection. These features, together with the “locally orthogonal” view, facilitate manual synapse annotation. Optionally, each synapse can be assigned to a predefined class and annotated with a confidence level.Note that some of these functionalities have recently been incorporated also into other software packages.

Supplementary Figure 4 EM stack of the larval OB.

(a) Image showing cross-section through the entire larval OB (EL = 2 keV, De = 17.5 enm–2, 9.25 × 9.25 × 25 nm voxels; high vacuum; vpSEM), assembled from multiple tiles (green). Adjacent tiles were combined at the center of the overlapping areas. ON, olfactory nerve; S, silver particles. (b) Reslices through a volume of 512 × 512 × 512 voxels in the area outlined by the white square in (a). Green arrows depict the line where two tiles were combined. (c) Examples of synapses, including a pair of reciprocal synapses (bottom).

Supplementary Figure 5 Tracing performance as a function of experience.

(a) Tracing speed of the same 8 tracers (mean ± s.d.) at two levels of experience (‘low’ and ‘high’; see below). (b) Accuracy of individual tracings, measured by precision, recall and AL. Each data point quantifies the accuracy of the reconstruction of one neuron by one tracer (n = 37 neurons). Gray lines connect data points representing reconstructions of the same neurons by the same tracers at different levels of experience. Experience of individual tracers is color-coded. Box-plots show median, 25th percentile, and 75th percentile. Statistical comparisons were performed using a Wilcoxon rank-sum test. P-values, U-values and degrees of freedom (df) were as follows: Precision: P = 0.10, U = 618, df = 74; Recall: P = 0.00004, U = 306, df = 74; AL: P = 0.03, U = 487, df = 74.To quantify the effect of long-term experience on performance, eight tracers traced a total of 37 neurons within one month using KNOSSOS. The same tracers then traced the same neurons again using KNOSSOS approximately one year later. Between the first and the second tracings, the mean experience of the tracers increased from 412 ± 410 h (mean ± s.d.) to 1,042 ± 609 h. Precision and recall (Fig. 3a) of the individual tracings were measured relative to consolidated reconstructions generated by CORE. The average tracing speed increased slightly but not significantly (p = 0.55; Wilcoxon rank-sum test) (a). Precision did not change significantly but recall and AL improved (b). Hence, accuracy increased with long-term experience because recall errors (missed processes) decreased. Note that any interpretation of accuracy measures should consider that tracing speed and accuracy may vary depending on neuronal morphology, sample preparation methods, image quality and tracing instructions (Helmstaedter, M., et al. [2011] Nat. Neurosci. 14, 1081-1088; Mikula, S., and Denk, W. [2015] Nat. Methods 12, 541-546).

Supplementary Figure 6 Estimation of the number of non-reconstructed OB neurons.

(a) Somata near the presumptive caudal boundary of the OB. Colors mark somata of fully reconstructed OB neurons (blue), somata of putative OB neurons (light blue; dendrite directed towards OB), somata of reconstructed cells that do not project into the OB or are not neurons (bright red; non-OB), and somata of putative non-OB cells (pale red; no dendrite towards OB). (b) Number of cells in each class as a function of their minimal distance to an OB neuropil region. Colors as in (a). Light blue line shows an exponential fit to the number of OB neurons as a function of distance. Vertical dashed line shows the distance beyond which no further OB neurons are expected.In order to estimate the total number of OB neurons we focused on the presumptive boundary of the OB where somata of OB neurons are directly adjacent to, and partially intermingled with, a large number of telencephalic somata. 95 additional candidate cells near the presumptive boundary were reconstructed with redundancy 3. Among these, 21 cells had neurites in the OB and were fully reconstructed using CORE. The remaining cells did not project into the OB or were non-neuronal cells. We observed that primary neurites of OB neurons were always directed towards the OB whereas primary neurites of other neurons pointed in other directions. We used this criterion to estimate the number of OB neurons not included in the set of 1,096 reconstructed neurons. We marked the position of approximately 1,000 additional somata near the presumptive boundary of the OB, determined their distance to the closest neuropil region in the OB, and sorted them into 2.5 μm bins. For all somata up to 20 μm that were not reconstructed before we traced the initial portion of their primary neurite, if any. 24 of these cells were classified as putative OB neurons because they had a primary neurite oriented towards the OB. The number of OB neurons (reconstructed and putative) decreased sharply with increasing distance up to 20 μm (b). An exponential fit predicted that approximately one additional putative OB neuron should be found in subsequent distance bins (≥ 20 μm). We therefore estimate the total number of putative OB neurons that were not included in our reconstructions to be 25. The total number of OB neurons was therefore estimated to be 1,047 (1,001 + 21 + 25). Out of these, 1,022 (98%) were reconstructed by CORE.

Supplementary Figure 7 Association of mitral cells and interneurons with glomeruli.

(a) Left: innervation of glomeruli by mitral cells. For each mitral cell, glomeruli were ranked by decreasing intraglomerular neurite length (axon excluded). Mitral cells were sorted by increasing innervation of the glomerulus with the lowest rank (“parent glomerulus”). Most mitral cells innervate a single glomerulus, and very few mitral cells innervate more than two glomeruli. Right: same plot for interneurons. (b) Mean fraction of mitral cell neurite per glomerulus as a function of glomerulus rank. Gray curve: cumulative distribution of neurite fraction per glomerulus. Images show three examples of mitral cells with dendrites in multiple glomeruli (gray shapes). Note that innervated glomeruli are adjacent to each other. No mitral cells with multiple dendritic tufts in distant glomeruli were observed.

Supplementary Figure 8 Distance of MCs and INs to glomeruli.

Each cell was assigned to a primary glomerulus that contained most of the cell’s intraglomerular processes. Distances (mean ± s.d.) of somata were measured to the primary glomerulus and to all other glomeruli. Top: P-values of statistical comparisons (Wilcoxon rank-sum test). U-values and degrees of freedom (df) were as follows: MCs: U = 276,035, df = 12,631; All INs: U = 294,866, df = 4,114; Class 1 INs: U = 1,288, df = 323; Class 2 INs: U = 333, df = 119; Class 3 INs: U = 1,216, df = 323.MC somata were significantly closer to their primary glomerulus than to other glomeruli. A similar, albeit weaker, association between somata and primary glomeruli was observed for IN classes 1 and 3 but not for IN class 2.

Supplementary Figure 9 Organization of interglomerular MC projections.

(a) Correlation matrix showing similarities between glomerular innervation patterns of MCs. MCs were sorted as in Fig. 6c: each MC was assigned to a primary glomerulus by its maximum innervation, and MCs assigned to the same glomerulus were ranked by their relative innervation of the glomerulus. Gray lines separate MCs assigned to different primary glomeruli. Because most MCs are associated with a single glomerulus, correlations between projections patterns of MCs assigned to the same glomerulus were very high while correlations across glomeruli were low. (b) Correlation matrix showing similarities between MC innervation patterns of glomeruli, sorted by k-means clustering. Gray lines separate k-means clusters. Color-code of glomerulus labels represents OSN-defined groups (Braubach, O.R., et al. [2012] J. Comp. Neurol. 520, 2317-2339; Braubach, O.R., et al. [2013] J. Neurosci. 33, 6905-6916) as in Fig. 6a. Correlations between innervation patterns of different glomeruli are low because each MC is typically associated with a single glomerulus.

Supplementary Figure 10 Shuffling of projections abolishes the correlation structure of glomerular innervation patterns.

Red: histogram of pairwise correlations between glomerular IN innervation patterns. Blue: histogram of correlation coefficients after shuffling of relative innervations across glomeruli for each IN (mean over 100 independent shufflings). Shuffling abolished positive correlations and the distribution of correlation coefficients became narrower. Hence, the observed glomerulus-specific pattern of IN innervation in the larval OB (Fig. 8b) is inconsistent with a random innervation pattern.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–10 and Supplementary Software Information (PDF 3419 kb)

Supplementary Methods Checklist

(PDF 383 kb)

SBEM stack of an EE-embedded sample (telencephalon of adult zebrafish).

Pixel size 12 × 12 nm2; section thickness 25 nm; acquisition rate 1 MHz; EL = 1.5 keV; De = 8.7 e nm–2; high vacuum). (AVI 8528 kb)

SBEM stack from the adult zebrafish OB (Fig. 1f).

Each image consisted of 40 tiles, each with 4,096 × 4,096 pixels. EE embedding; pixel size 9 × 9 nm2; section thickness 25 nm; acquisition rate 2 MHz; EL = 1.5 keV; De = 15.9 – 17.3 e nm–2; high vacuum; lvSEM. Image resolution has been severely reduced and only a subset of sections is shown to minimize movie size. (MP4 7215 kb)

Zoom into an image from large SBEM stack (Fig. 1f and Supplementary Movie 2).

Image consisted of 40 tiles, each with 4,096 × 4,096 pixels. Enlarged region contains overlap between adjacent tiles. EE embedding; pixel size 9 × 9 nm2; section thickness 25 nm; acquisition rate 2 MHz; EL = 1.5 keV; De = 15.9 – 17.3 e nm–2; high vacuum; lvSEM. Image quality has been slightly reduced to reduce movie size. (AVI 29578 kb)

Zoom into an SBEM stack from the larval zebrafish OB (Supplementary Fig. 4).

EE embedding; pixel size 9.25 × 9.25 nm2; section thickness 25 nm; acquisition rate 200 kHz; EL = 2 keV; De = 17.5 e nm−2; high vacuum; vpSEM. (AVI 12824 kb)

Subvolume of SBEM stack from the larval zebrafish OB (Supplementary Fig. 4).

EE embedding; pixel size 9.25 × 9.25 nm2; section thickness 25 nm; acquisition rate 200 kHz; EL = 2 keV; De = 17.5 e nm−2; high vacuum; vpSEM. The four quadrants show the four viewports of PyKNOSSOS (red: xy; blue: yz; green: xz; yellow: arbitrary). Arbitrary viewport (yellow) was set to be dorsal up and ventral down, lateral to the right and medial to the left. (AVI 22733 kb)

Distribution of somata in the OB and adjacent areas.

Somata are color-coded as in Supplementary Figure 6 (blue, reconstructed OB neurons; pale blue spheres: putative OB neurons; red, reconstructed non-OB cells; pale red, putative non-OB cells). Gray, glomeruli. Scale bar, 10 μm. (AVI 6692 kb)

1,022 neurons reconstructed in the OB of a zebrafish larva.

Color encodes soma position along dorsoventral axis. Scale bar, 10 μm. (AVI 10210 kb)

Large olfactory bulb cells (LOC).

Shaded volumes are outlines of glomeruli and color-coded according to the relative innervation by the two LOCs (Fig. 5g). Scale bars, 10 μm. (AVI 5524 kb)

Soma of a LOC.

Scale bars (black edges), 2 μm. (AVI 4215 kb)

Glomeruli of the larval OB.

Shaded volumes show glomeruli, color-coded according to OSN-defined glomerular groups (Fig. 6a). Scale bars (red edges), 20 μm. (AVI 9180 kb)

Associations between mitral cells and glomeruli.

Glomeruli are represented by large shaded volumes. Superimposed are individual mitral cells with neurites innervating three glomeruli. Mitral cell somata are represented by small shaded volumes. Long processes are axons projecting out of the olfactory bulb. Dendrites of individual mitral cells are closely associated with single glomeruli. Scale bars, 10 μm. (AVI 7042 kb)

All mitral cells and their associations with glomeruli.

Mitral cells are color-coded according to the OSN-defined group of their parent glomerulus (color code as in Fig. 6a). The parent glomerulus is the glomerulus containing the majority of intraglomerular neurites. Note that somata of mitral cells are superficial and spatially clustered by OSN-defined groups. Scale bar, 10 μm. (AVI 7737 kb)

Associations between interneurons and glomeruli.

Glomeruli are represented by large shaded volumes. Superimposed are individual interneurons with neurites innervating three glomeruli. Somata are represented by small shaded volumes. Individual interneurons associated with the same glomerulus show different morphologies. Scale bars, 10 μm. (AVI 7264 kb)

All INs and their associations with glomeruli.

INs are color-coded according to the OSN-defined group of their parent glomerulus (color code as in Fig. 6a). The parent glomerulus is the glomerulus containing the majority of intraglomerular neurites. Note that most somata of INs are located in deep layers and that spatial clustering is less pronounced than for mitral cells. Scale bar, 10 μm. (AVI 7027 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wanner, A., Genoud, C., Masudi, T. et al. Dense EM-based reconstruction of the interglomerular projectome in the zebrafish olfactory bulb. Nat Neurosci 19, 816–825 (2016).

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing