3D Reconstruction of the Antennal Lobe of
|Help SiteNavigator Index Where Am I? Search Page Status||Contact Us|
Pierre Philippe Laissue1, Christian Reiter2, Peter Robin Hiesinger2, Stephan Halter1, Karl-Friedrich Fischbach2 and Reinhard F. Stocker1
1 University of Fribourg, Insitute of Zoology and Program in Neuroscience, CH-1700, Fribourg,
2 University of Freiburg, Institute for Biology III, Schaenzlestr. 1, D-79104 Freiburg, Germany
Originally published in: Laissue et al., J. Comp. Neurol. 405, 543-552
Copyright © 1999 by John Wiley & Sons, Inc. All rights reserved
Proceed directly to models and movies (skip this page)
We present the first three-dimensional map of the antennal lobe of D. melanogaster, based on confocal microscopic analysis of glomeruli stained with the neuropil-specific monoclonal antibody nc82. The analysis of confocal stacks allowed us to identify glomeruli according to the criteria shape, size, position and intensity of antibody labeling. Forty glomeruli were labeled by nc82, eight of which have not been described before. Three glomeruli previously shown exclusively by backfills were not discernible in nc82 stainings. Six glomeruli consist of distinct, but contiguous structural units, termed 'compartments'. Glomerular variability observed occasionally between males and females is in the same range as between individuals of the same sex, suggesting the lack of a significant sexual dimorphism in the glomerular pattern. An important goal of this work was to create these 3D reference models of the antennal lobe, which are all accessible on-line.
On this page, we explain the basic concept of identification aids (i.e. the color code of arrays and glomerular classes) and the different items available (surface models as well as movies of volume rendered views and confocal stacks).
Each confocal stack of an individual antennal lobe was analyzed by
(In the following explanations, this order is reversed)
Abbreviations: AL: Antennal Lobe. VRML: Virtual Reality Modeling Language
To facilitate identification, the 40 glomeruli were assigned to 5 glomerular 'arrays'. An array is defined by a prominent 'core' glomerulus, i.e., D, DM3, V, VA1, VA2, and 5 to 9 glomeruli in its vicinity. By comparing their positions relative to the core glomerulus, these glomeruli can be reliably identified in frontal sections, even when they are not discrete. Glomeruli of an array are recognizable in any image from a frontally sectioned series as long as the array's core glomerulus is visible (except for the posteriormost DM3 array).
The 3 glomerular classes distinguish all glomeruli according to the criteria shape, size, position and intensity of antibody labeling. Class 1 glomeruli are easily identifiable since they are constant in all four criteria. Class 2 glomeruli are less well demarcated, as they vary in either shape, size or position. Last but not least, class 3 glomeruli are often poorly defined and vary in more than one criterion. With this classification, all class 2 and 3 glomeruli can be identified by comparison with landmark glomeruli in their vicinity.
Colors indicate the 5 glomerular arrays V (green), VA1 (blue), VA2 (yellow), D (purple) and DM3 (red), each of which is defined by a corresponding 'core' glomerulus (i.e. glomerulus V is the core glomerulus of the V array, D the core glomerulus of array D etc.). Different shading indicates the 3 classes of glomerular distinctness. Dark colors: 'landmark' glomeruli (class 1); intermediate colors: `class 2' glomeruli; light colors: ill-defined glomeruli (class 3). In summary, the VA1 and VA2 arrays comprise the anterior portion of the AL, the V and D arrays occupy its central part and the DM3 array includes the posterior end of the AL.
In surface models, all glomeruli are shown as reconstructed surfaces. These idealized surfaces were demarcated by hand in the primary dataset, the confocal stack. These models can be interactively manipulated.
The 'Virtual Reality Modeling Language' (VRML) is an ISO-standardized 3D description language that allows viewing and manipulation of the reconstructed object with freely distributed software. VRML was developed for presentation of 3D data on the Internet and is thus ideally suited for creating a 3D anatomical atlas available to the research community. Furthermore, the new VRML standard (VRML97, utf8) allows interactive programming such as the embedding of original and interpolated confocal data into the surface model (see: Surface models, VRML2 model of male specimen 1. This permits reslicing of the primary dataset of frontal sections of the AL in the horizontal and sagittal plane.
The initial viewpoint of all surface models is always the same: As seen from anterior. Lateral is to the right, dorsal on top.
Volume rendering utilizes algorithms (e.g. 'raytracing') to calculate views of the entire 3D object. In contrast to surface rendering, these movies can not be manipulated interactively. We applied raytracing to obtain standardized views of the datasets from a series of different angles as defined by the position of a virtual camera. Direct comparison of unprocessed confocal stacks is often complicated by slightly varying angles of sectioning. These difficulties are avoided by visualizing the individual confocal stack as a volume rendered 3D object.
The confocal stack is the primary dataset (meaning the original, unprocessed data) and consists of a series of digital images from front to back of an individual antennal lobe. These series were taken at 3 images/µm with 512x512 pixel resolution, resulting in an average of 130 images per AL (and a size of 40 MB).
Close comparison of AL reconstructions from males and females revealed that intrasexual structural variability is in the same range as between the two sexes, in terms of glomerular number and distribution. Hence, neither the glomerular nor the SC pattern appear to be clearly sexually dimorphic. Also, studying ALs at various adult ages from 4 hours to 10 days after eclosion did not reveal obvious differences in the glomerular pattern.
For practical use, glomeruli in an nc82 stained AL can be identified by applying the map in two different ways: By comparing the nc82 label of the unidentified confocal image series to an identified stack (as in Fig. 1 of the printed publication), or by manipulating the 3D surface models presented here.
Proceed to models and movies
|Flybrain Atlas 3D Flybrain 3D Models Antennal Lobe|
All contents are copyright © 1995-2000 Flybrain or their original publication (as noted).
Copyright and use policy
AB00200Page last modified: July 25, 2000 by Managers.