Random or selective neuroanatomical connectivity. Study of the distribution of fibers over two populations of identified interneurons in cerebral cortex

Brain Research Protocols 14 (2005) 67 – 76 www.elsevier.com/locate/brainresprot

Protocols

Random or selective neuroanatomical connectivity. Study of the distribution of fibers over two populations of identified interneurons in cerebral cortex Marjolein Vinkenooga,1, Michel C. van den Oevera, Harry B.M. Uylingsa,b, Floris G. Wouterlooda,* a

Graduate School Neurosciences Amsterdam, Research Institute for Neurosciences Vrije Universiteit Medical Center, Department of Anatomy, MF-G-136, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands b Netherlands Institute for Brain Research, Meibergdreef 33, 1105 AZ Amsterdam, The Netherlands Accepted 29 September 2004 Available online 5 November 2004

Abstract We present a neuroanatomical tracing method in a stereological approach to study the proportional distribution of fibers of a particular projection over two chemically different populations of neurons. The fiber projection from the presubiculum to the medial division of the entorhinal cortex of the rat serves as a model projection. Potential target interneurons express calcium binding proteins, either parvalbumin or calretinin. The three markers were simultaneously stained in one and the same histological section. The procedure is according to a threephase procedure, i.e., in vivo tracer injection phase, histology phase, laserscanning phase. Steps involved are: (1) Surgical application to the presubiculum (injection) of the neuroanatomical tracer, biotinylated dextran amine (BDA), with the purpose of labeling fibers innervating the entorhinal cortex. After surgery, transport of the tracer takes place during the one-week survival period; (2) Fluorescence detection of the labeled fibers through staining with fluorochromated avidin (avidin-Alexa Fluor 488k [green fluorescence]); (3) Simultaneous Immunofluorescence detection of two interneuron markers (using the appropriate primary antibodies and secondary antibodies conjugated to the fluorochromes Alexa Fluor 594k [red fluorescence] and Alexa Fluor 633k [infrared fluorescence]); (4) Acquisition of lowmagnification images in a confocal laserscanning microscope and the preparation on a computer of a montage image covering the entire entorhinal cortex; (5) Overlaying this montage with a sampling grid; (6) Acquisition at high magnification of Z-series of confocal images in a statistical valid way based on this grid. Each marker was visualized in its own laser excitation/emission channel: 488, 568 and 647 nm; (7) Image processing and 3D reconstruction followed by evaluation of the results. The present approach can be used to examine whether or not a particular class of chemically identified neurons receives preferential innervation by a particular fiber projection. D 2004 Elsevier B.V. All rights reserved. Theme: Cellular and molecular biology Topic: Staining, tracing and imaging techniques Keywords: Neuroanatomical tracing; Subiculum; Triple staining; Morphometry; Confocal laserscanning; Double-immunofluorescence; 3D-reconstruction

1. Type of research

* Corresponding author. Fax: +31 20 444 8054. E-mail address: [email protected] (F.G. Wouterlood). 1 Current address: Division of Clinical Immunology and Rheumatology, AMC Medical Research BV/University of Amsterdam, Meibergdreef 9,1105 AZ Amsterdam, The Netherlands. 1385-299X/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.brainresprot.2004.09.005

Axons in the CNS conveying information to cortical or subcortical areas distribute in these terminal areas among neurons of different morphologies and chemical identities. A question basic to our understanding of neuronal circuits is whether the axon terminals in a terminal area are engaged in synaptic contacts randomly with members of the entire

68

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

population of neurons or whether they selectively form contacts with specific types of neurons (Fig. 1). Fibers that end in a particular area may selectively target principal neurons, interneurons or both. Even if, for instance, fibers contact only interneurons, the question repeats itself because the innervating fibers may selectively target neurons with a very specific chemical makeup. An additional possibility is that part of the fibers form contacts with neurons of chemical type A while the remaining fibers form contacts with neurons belonging to chemical class B and not with members of class C, and so on. The classical approach to unravel selective-or-not innervation is the application of an anterograde neuroanatomical tracer in combination with a second, target-neuron specific labeling technique [4,17,25,26,30]. Alternatively, one may intracellularly record from neurons and follow up with visualization and immunofluorescence characterization or detection [18,33]. The question of selective innervation, and especially that of the proportional innervation of different classes of neurons by a given fiber projection can be studied quite well when all the participating elements: fibers, potential target neurons, are visualized in one and the same histological preparation. The solution presented here is a combination of neuroanatomical tracing, a double immunofluorescence staining, high-resolution confocal laserscanning and post-acquisition 3D reconstruction in one and the same histological section [31]. We illustrate our method in rat brain by means of the fiber projection from the presubiculum to the entorhinal cortex [2,8,20,21,23]. The presubicular fibers end in the entorhinal cortex predominantly in layer III, i.e. a layer rich in interneurons expressing calcium-binding proteins such as parvalbumin and calretinin [13,28,30]. Parvalbumin and calretinin in entorhinal cortex are mutually exclusive markers: neurons express either parvalbumin or calretinin [30]. The question is whether a single type or both types of

calcium binding protein expressing neurons receives preferential innervation and to what degree.

2. Time required (a) Neuroanatomical experiment: surgery: 1 h. (b) Survival time to allow sufficient transport of neuroanatomical tracer and robust labeling: approximately 1 week. (c) Perfusion, fixation, sectioning: 1 day. (d) Immunofluorescence staining and detection of tracer, preparation of histological slides: 3 days. (e) Confocal laserscanning (image acquisition): 1 day. (f) Three-channel post-acquisition computer processing: 1 day.

3. Materials 3.1. Subjects Female Wistar rats (Harlan, Zeist, The Netherlands), body weight 180 g (n = 2). 3.2. Surgery The neuroanatomical tracer was iontophoretically applied: constant current power supply capable of maintaining a positive pulsed 5 AA DC current on micropipettes with a tip diameter of 10 – 20 Am. Micropipettes were pulled with a home-made pipette puller from bososilicate glass capillaries, outer diameter 1.5 mm, inner diameter 0.86 mm (GCI150F-15, Harvard Apparatus, Edenbridge, Kent, UK). 3.3. Perfusion, sectioning For the purpose of precisely controlling the flow and hydrostatic pressure of the fixative during perfusion we used a compressed air driven perfusion device developed by Jonkers et al. [7]. This device was built in the machine shop of the medical school. Sections were cut on a vibrating microtome (model VT1000S, Leica Microsystems, Heidelberg, Germany). 3.4. Image acquisition

Fig. 1. Principle of nonselective or selective innervation. The target area contains cell types a and h. Fibers from area g arrive in the area and form contacts. (A) Situation in which the fibers innervate in a nonselective way neurons belonging to both types of target neuron. (B) Situation in which fibers innervate selectively, e.g., only cells of type a.

Confocal laserscanning was conducted on a Leica TCSNT instrument (Leica Microsystems, Heidelberg, Germany) equipped with an Ar–Kr gas laser providing excitation of fluorochromes at wavelengths 488, 568 and 647 nm: three detection channels. In order to avoid crosstalk we acquired images as pairs of simultaneously acquired images (scanning at 488–568 nm followed by scanning at 488–647 nm). In order to perform 3D reconstruction it is necessary to scan volumes of tissues. As a consequence we always acquired

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

Z-series (stacks) of images. Each observation thus consisted of three Z-series: one stack in the 488 nm detection channel (after verification of correct spatial duplication, one of the originally two Z-stacks at 488 nm was discarded), one stack in the 568 nm detection channel, and the third stack in the 647-nm detection channel. 3.5. Sampling scheme Before the actual Z-scanning at high magnification we took for the purpose of preparing a montage surveying the entire area of interest (medial entorhinal cortex) single, overlapping images of the medial entorhinal cortex at low magnification (20) (in each channel) and with the detector pinhole fully open. Since labeling of presubiculum fibers was seen mostly in layer III of the medial entorhinal cortex we limited our sampling to this layer. Because calretinin immunoreactive cells appeared rather scarcely distributed in this layer, we sampled all layer-III calretinin cells for study. The much more abundant parvalbumin immunoreactive cells were sampled in a random systematic way, using a sampling grid overlay with 100100 Am squared units. We sampled 33 cells expressing calretinin and 31 cells expressing parvalbumin. These cells were obtained from three sections from the brain of one rat [we are currently applying the same scheme in research projects using different markers]. 3.6. Post acquisition image processing Image stacks were deconvoluted on a Silicon Graphics Octane workstation (SGI, Mountain View, CA) by means of the Huygens II deconvolution software package (Scientific Volume Imaging, Hilversum, the Netherlands, website at http://www.svi.nl). 3.7. 3D reconstruction Immediately after deconvolution, we 3D reconstructed the involved structures on the Silicon Graphics Octane workstation with Amira 3D visualization/modeling software (Konrad Zuse Zentrum for Information Technology, Berlin, Germany; website at http://amira.zib.de; program distribution by www.amiravis.com), and merged the results into one three-color 3D reconstruction. 3.8. Chemicals (a)

Anesthesia: Ketamine (Ket, Aesco, Boxtel, The Netherlands) Xylazine (Rompun, Bayer, Brussels, Belgium) Lidocaine (10% spray; Astra Pharmaceutica BV, Zoetermeer, The Netherlands) Sodiumpentobarbital (Nembutal, Ceva, Paris, France). (b) Neuroanatomical tracer: Biotinylateddextran amine (BDA, MW 10,000; Molecular Probes, Eugene, OR). (c) Laboratory chemicals: Merck pro analysis.

69

(d) Detergent: Triton X-100 (Sigma, St. Louis, MO). (e) Normal sera, antibodies, fluorochromated chemicals/ antibodies: Normal goat serum: Dako, Copenhagen, Denmark, Mouse anti-parvalbumin: Sigma, Rabbit anti-calretinin: Chemicon, Temecula, CA, Streptavidin conjugated to fluorochrome Alexa Fluork 488: Molecular Probes, Goat anti-rabbit conjugated to Alexa Fluork 594: Molecular Probes, Goat antimouse conjugated to Alexa Fluork 633: Molecular Probes.

4. Detailed procedure 4.1. Animals, surgery All animal treatment was in accordance of the rules set for experimentation on animals by the European Community and the Dutch Government. The animals were deeply anesthetized with 1 ml/kg body weight of 4 parts ketamine (1% of a solution of Ket) mixed with 3 parts xylazine (2% solution of Rompun) and mounted in a stereotaxic frame. The skull was exposed, the periost anesthetized with lidocaine, and the a hole was drilled in the skull over the desired X–Y location. A micropipette filled with tracer (BDA, 5% in 10 mM phosphate buffer, TBS, pH 7.25) was lowered to the desired Z coordinate and the tracer was injected bilaterally into the presubiculum (coordinates derived from the stereotaxic brain atlas by Ref. [14] Bregma-7.0, lateral 1.4, dorsoventral 4.0 mm [from the brain surface], angle in the coronal plane 258). Successful injections were obtained by the application to the micropipette (tip diameter 10–20 Am) of a positive pulsed 5 AA DC current (7 s on /7 s off) during 10–15 min. The presubiculum is an elongated and curved cell area. We aimed at the ventralmost portion of this structure. Using the injection parameters above, we obtained injection spots that nicely fitted within the boundaries of the ventral presubiculum. After injection the pipette was left in situ for 10 min. After retraction of the micropipette the skin over the wound was sutured and the animal was allowed to recover. The survival period post surgery was 1 week. 4.2. Perfusion–fixation, sectioning, storage After the survival period, the rats received an overdose of sodium pentobarbital intraperitoneally (60 mg/kg body weight). They were then transcardially perfused, first with 100 ml of physiological saline solution of 38 8C, pH 6.9, immediately followed by 1000 ml of 4% freshly depolymerized paraformaldehyde, 0.1% glutaraldehyde and 0.1% of a saturated picric acid solution in 125 mM phosphate buffer, pH 7.4 (room temperature). Perfusion was carried out with a compressed air-driven device that permits a controlled hydrostatic pressure of 30 kPa [7]. The thoracic aorta was clamped to ensure that all fixative was directed at

70

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

the upper part of the body. Perfusions conducted in this way typically take place within a time slot of 45 min. Immediately after perfusion, the brain was removed from the skull and cut with the vibrating microtome into 50-Am thick sections in the horizontal plane. Sections were collected in 125 mM phosphate buffer, pH 7.4 (vials kept on melting ice). If sections were to be stored for prolonged periods, they were infiltrated with 20% glycerin and 2% DMSO in 125 mM phosphate buffer, pH 7.4, and placed in a freezer ( 20 8C). 4.3. Triple fluorescence staining procedure Incubation was in a free floating regime, with vials containing approximately 10 sections in 1000 Al incubation medium. Continuous gentle agitation was provided to prevent the contents of the vials from settling on the bottom. In between all incubation steps the sections were thoroughly rinsed with incubation buffer: 50 mM Tris/ HCL buffer with 0.875% sodium chloride, 0.5% Triton X100, pH 8.0 (TBS-TX); 310 min. Steps were as follows. (a)

Pre-incubate for 1 h at room temperature with 5% normal goat serum and then incubate for at least 48 h at 4 8C with a cocktail of primary antibodies: mouse anti-parvalbumin (1:2000) and rabbit anti-calretinin (1:500). (b) Incubate for at least 24 h at 4 8C with a cocktail consisting of streptavidin conjugated to the fluorochrome Alexa Fluork 488 (1:200), goat anti-rabbit IgG conjugated to Alexa Fluork 594 (1:100) and goat anti-mouse IgG conjugated to Alexa Fluork 633 (1:200). (c) Rinse with Tris buffer (6.06 g/l aqua dest, pH 7.4), mount in Tris buffer with gelatin (0.2 g/100ml Tris buffer, pH 7.4) and dry thoroughly in the dark at room temperature overnight, or for two hours in an at 40 8C. Completely dried sections were dipped for a few seconds in 100% toluene and coverslipped with DPX (Fluka Chemie, Buchs, Switzerland). After completion of the procedure, slides were stored in a freezer at 20 8C. 4.4. Immunohistochemistry and instrument controls Control sections were prepared to verify that the binding of primary and secondary antibodies was specific and that no crosstalk occurred between the fluorochromes. This kind of procedure has been discussed at length for doubleimmunofluorescence stained sections by us in this journal [29]. In brief, we incubated sections with only one of the two primary antibodies against calcium binding proteins, divided these sections over three vials and continued incubation with a mixture of one corresponding and one non-corresponding or with both fluorochromated secondary

antibodies. Also we incubated with buffer vehicle only and hereafter with both fluorochromated secondary antibodies. As it turned out, the fluorescence emitted by Alexa Fluork 594 showed crosstalk in our confocal instrument with that of Alexa Fluork 633, whereas fluorescence emitted by Alexa Fluork 488 did not crosstalk with that of the other two fluorochromes. Because of this and because our instrument lacked the option of spectral separation of mixed signals we always stained the transported BDA with avidin-Alexa Fluork 488, the parvalbumin with Alexa Fluork 633, and the calretinin with Alexa Fluork 594. In addition we studied in the confocal instrument the sections with pairs of two laser emission settings (488–568 and 488–647 nm, see below). Since triple staining is extremely dependent on highly specific immunostaining without even the smallest trace or suspicion of cross-reactivity, a control in which the primary antisera are replaced with pre-immune serum of the relevant animal species may provide additional confidence in the results (this control not done by us in the present experiment, but suggested by one of the reviewers and certainly worth doing). The control sections used to check instrument performance consisted of sections containing BDA-labeled fibers stained with a cocktail of three fluorochromated streptavidins: respectively, Alexa Fluork 488, Alexa Fluork 594, and Alexa Fluork 633. The purpose of this triple staining was to calibrate the instrument and check possible chromatic aberration. 4.5. Observations, analysis (a)

Channels: Slides were introduced in the confocal scanning microscope. Fluorescence emission was measured in three channels: channels #1, #2 and #3, configured around laser excitation at 488, 568 and 647 nm, respectively. A channel is a specific configuration of the confocal instrument with the purpose to detect a particular fluorochrome: laser, emission wavelength, dichroic mirror, filter set and detector. Note that Alexa Fluork 488 was detected in channel #1, Alexa Fluork 594 in channel #2, and Alexa Fluork 633 in channel #3. Please note that the Leica instrument is fitted with variable shortpass, bandpass and longpass filtering. Filtering was chosen such that crosstalk was eliminated [24]. (b) Calibration: Before the scanning of the experimental sections we took sample images of the triple-stained dcalibrationT sections at high magnification in channels #1, #2 and #3. The purpose of this action was to check by means of 3D reconstruction whether lateral or axial color shift might occur in the optical pathway of the instrument. Major color shift is revealed when images in the three channels are inspected at acquisition time in doverlayT mode, whereas subtle color shift becomes visible only after 3D reconstruction. In case of subtle

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

color shift the 3D reconstructed fiber in one of the channels is slightly spatially shifted compared with the other channels. (c) Experimental sections: Next, the experimental sections were analyzed. In this part of the procedure we checked first at low magnification whether the injection of the tracer was centered in the presubiculum and whether labeled fibers were present in the target area (the medial division of the entorhinal cortex). Next we took single, overlapping images of the medial entorhinal cortex at low magnification (20) and in each channel. (d) Montages: These overlapping images were used to prepare a montage surveying the entire portion of the entorhinal cortex where BDA labeled fibers, parvalbumin- and calretinin-immunoreactive cells were present. Thus, we made three separate montages: one montage showing the extent of the fiber labeling (illumination at 488 nm), one montage of the same area displaying the distribution of calretinin immunoreactive neurons (illumination at 568 nm) and one montage of the same area showing the distribution of the parvalbumin immunoreactive cells (illumination at 647 nm). These montages were used to determine the sampling areas (see next section). 4.6. Sampling strategy for high-magnification image acquisition Since labeled fibers were mostly seen in layer III of the medial entorhinal cortex, our sampling had to be restricted to this layer. Because in entorhinal layer III calretinin immunoreactive cells appear rather scarcely distributed, we sampled (guided by the montages) all layer-III calretinin cells for study. Parvalbumin immunoreactive cells were abundant in layer III compared with calretinin immunoreactive cells. Thus, for parvalbumin neurons we adopted the following sampling strategy (Fig. 2). A 100100 Am unit grid overlay was placed over the area of interest of the parvalbumin montage in order to systematic randomly select parvalbumin-immunoreactive neurons for close inspection [5]. Starting at a randomly selected 100100 Am unit (this unit randomly taken from the first 10 units overlying the fiber termination area), each tenth unit overlying the area with fiber termination was inspected for the presence of a parvalbumin-immunoreactive neuron. Each grid unit has two dinclusionT lines and two dexclusionT lines [5]. Only parvalbumin neurons whose cell bodies were present completely within a grid unit and those touching and overlying one of the dincludeT lines were investigated at high magnification (Fig. 2). It should be mentioned here that the drying of the sections on slide and the subsequent embedding in mounting medium results in a shrinkage of the section in the Z-direction of approximately 70% (in the

71

Fig. 2. Sampling strategy depicted in a montage made in the 647 nm channel (parvalbumin) of the entorhinal cortex. Horizontal section. A 100100 Am unit grid overlay was superimposed on the low magnification montage. A starting unit was selected randomly between the first and tenth unit. From this starting unit, every 10th unit overlying layer III of MEA was selected (i.e., the area containing presubiculum fibers). MEA= medial entorhinal area, LEA= lateral entorhinal area, PrS = presubiculum, Pas = parasubiculum, Sub = subiculum. The big white arrow indicates the fiber projection: presubiculum to MEA.

confocal microscope it is easy to measure the thickness of the section; on average this thickness measures 15 Am compared with the original 50 Am cutting increment in the vibrating microtomy). Thus, tissue deformation in the Zdirection is almost certainly present [3]. Because we noted in the Z-direction no parvalbumin cells overlapping and complying with the inclusion criterion, we decided not to apply a correction for tissue distortion in the Z-direction. In total we systematically sampled a total of 33 calretinin cells and 31 parvalbumin cells collected from three sections of one rat (case no. 2001011). The results in terms of percentages of contacts recorded by us must therefore be considered as preliminary. 4.7. High magnification Z-scanning: twin stacks After the sampling areas had been established and cells for sampling selected, laserscanning could begin. Scanning was done with a high-resolution objective lens (63 water, NA 1.20, electronic zoom 8), using illumination with 488 and 568 nm laser light (BDA and calretinin) and with 488 and 647 nm laser light (BDA and parvalbumin). To avoid the channel crosstalk reported above we took in the final image acquisition Z-series of images in the 488 nm/ 568 nm channel pair and, after returning the microstage to the start Z-position, immediately afterwards a Z-series of images in the 488 nm/647 nm channel pair. The pairs of image series thus recorded are further called twin stacks. Note that this procedure included obtaining duplicate image series in the 488-nm channel. Because of the pair–pair image acquisition we could use the two 488-nm image series to check correct overlay, i.e. whether the 488,

72

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

568 and 647 nm image series had been taken in exactly the same tissue volume. Each Z-series of a twin stack consisted of approximately 50 images, taken at a Z-increment of 80 nm, 512512 pixels at 8-bit sampling, dsimultaneous modeT, Z-scanning was thus confined to the volume in which a cell and its possible appositions were visible, usually in the order of magnitude of 4.0–5.0 Am. Occasionally, more than one parvalbumin expressing cell was present in a grid unit. In such a case, we scanned only one cell, i.e. the one best fitting the inclusion–exclusion criteria. The reason for this decision was that Z scanning with the 647 nm laser results in considerable bleaching of the section. Scanning a second cell in a tissue volume previously Z-scanned would therefore amount to scanning part of a partly bleached cell. Errors in the image acquisition, for instance caused by chromatic aberration, were corrected for according to the procedure outlined in detail elsewhere [31]. Calibration sections were scanned during each image acquisition session. At the conclusion of each acquisition session, all twin stacks were copied to cd-rom awaiting further processing. 4.8. Image processing: deconvolution Image files were copied to the hard drive of a Silicon Graphics Octane workstation and processed with HuygensII deconvolution software. Deconvolution is sometimes referred to as ddeblurringT or dimage restorationT [1,6,16, 19,22]. Huygens II deconvolution uses a Maximum Likelihood Estimation (MLE) algorithm that takes into account the confocal instrument’s point spread function (PSFi), which is a mathematical description of the statistical distribution of photons emitted by a point source onto the detector via the CLSM’s pinhole. The MLE algorithm uses the PSFi to calculate for each pixel of each acquired image in a Z-series the statistical likelihood of the exact origin of the photons emitted by its fluorescent source. The result is a statistically reliable, improved version of the originally acquired image. The deconvolution program can however only process one Z-series of images at a time since each Zseries has its own PSFi (a PSFi is wavelength-dependent). The three Z-series of images that comprise one observation therefore had to be processed separately from each other. Via the dimportT function of the 3D reconstruction software it was possible to merge the three separately deconvoluted Z-series of images. 3D-Reconstruction was performed using Amirak visualization/modeling software. This versatile surface rendering program enables the operator to display a 3D-surface rendering of the reconstructed 488, 568 or 647 nm Z-series of images either separately or in overlay (import, merge) mode and, more importantly, it allows real-time rotation along all three spatial axis as well as a zoom option such that reconstructed cells, dendrites, fibers and boutons can be inspected from all directions. Furthermore Amira allows the

quick comparison of the 3D reconstructions with the original stacks of bitmapped images. 4.9. Criteria for identifying boutons and contacts Because axons in general and presubicular projection fibers in particular show along their course in the terminal area continuous variations of their diameters, we defined a bouton as follows: a varicosity emerging from a BDA labeled fiber with or without a pedicle, or a swelling of the shaft of the fiber displaying a diameter of at least three times the diameter of the fiber in between two subsequent swellings. An apposition between a bouton of a BDA labeled projection fiber and a target neuron was considered a contact if the following criteria were fulfilled: (a)

A bouton had to be present in (swelling) or on (axon terminal) a BDA labeled fiber. (b) This swelling had to be in juxtaposition with a cell body, dendrite or other part of a target neuron. (c) In the 3D reconstruction the juxtaposition had to be complete. If by inspection at different angles via rotation an optically empty space or slit in between the bouton and the target cell could be seen, then this particular juxtaposition was disqualified as a contact. Only if there was no optically empty space of slit at any angle of inspection, then the juxtaposition was accepted as a contact.

5. Results 5.1. BDA injection sites, projection patterns and termination areas Each BDA injection resulted in the labeling of a large number of neurons in the presubiculum. The iontophoretic delivery of the BDA provides a means to control the size of the injection spot. In our experience, the size of the injection area is controlled within some limits by adapting the size of the micropipette tip and the iontophoretic injection parameters [15]. The cell bodies of the BDA labeled neurons were confined within the boundaries of the presubiculum. Relatively thick labeled fibers with few varicosities (dpassing fibersT) originated from these neurons and traversed the adjacent deep layers of the medial division of the entorhinal cortex (MEA) under oblique angles with respect to the pial surface. Terminal fields of the fiber projections were organized in MEA into two discrete terminal fields, i.e., one plexus in layer I, and a particularly dense plexus in layer III. Fiber and terminal labeling in MEA did not extend into the adjacent lateral division of the entorhinal cortex (LEA). In all cases there was labeling also in layers I and III of the contralateral MEA, albeit less dense than on the ipsilateral side. The

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

individual boutons of the BDA labeled fibers were very prominent at the high magnification applied for searching and recording in the confocal microscope. The labeled terminal boutons had globular and oval shapes. Boutons formed en passant were present as spindle-like swellings on the labeled fibers. Diameters of boutons varied between 0.5 and 1.0 Am. 5.2. Parvalbumin and calretinin cell and dendrite labeling Layer III of the medial entorhinal cortex is richly populated with parvalbumin immunoreactive neurons. In brief, the majority of these neurons were medium sized to large multipolar neurons that emitted long aspiny dendrites in a radiate fashion in all directions. In addition we noted a dense plexus of parvalbumin fibers and terminals in layer III. Our present findings are in agreement with those published earlier [28]. Calretinin immunopositive neurons were much less abundant than parvalbumin immunopositive neurons; the calretinin neurons were often of the bipolar type, with aspiny dendrites generally oriented perpendicularly to the layer containing the parent cell body. The dendrites of the calretinin cells penetrated into layers above and below layer III. The findings on calretinin expression are in agreement with those published previously by us [30].

73

5.3. Contacts between BDA labeled boutons and parvalbumin- or calretinin-expressing neurons Numerous contacts were observed between boutons of presubicular fibers and parvalbumin neurons. In total we reconstructed 31 parvalbumin cells and 33 calretinin cells in layer III of MEA and we closely inspected these cells in after deconvolution and 3D reconstruction at high magnification under all viewing angles (3608). Sixty-four percent of the sampled parvalbumin cells were seen to receive contacts by boutons of BDA labeled fibers, while of the calretinin cells, 90% received such contacts. Of the parvalbumin subpopulation cells that received contacts, neurons received on average 2.3 contacts per cell. Calretinin immunopositive neurons received slightly more contacts (on average 3.0 contacts per cell in the calretinin subpopulation that received contacts). Parvalbumin cells received contacts mostly on their cell bodies (80% of all contacts on cell bodies) and, to a lesser degree, on their dendrites whereas calretinin cells received contacts mostly on their dendrites (68% of all contacts with dendrites). In several cases we noted labeled fibers forming contacts with both parvalbumin and calretinin-immunofluorescent cells. Representative 3D reconstructions of contacts between BDA labeled fibers and parvalbumin and calretinin neurons are shown in Figs. 3C and D.

Fig. 3. Triple staining. (A) Extended focus image after image acquisition. Green fibers: BDA; blue structures: parvalbumin, red structures: calretinin. Bar 25 Am. (B) 3D reconstruction of the involved structures after deconvolution. Same colors as in (A). Bar = 25 Am. (C) 3D reconstructed contacts between presubiculum boutons (B) and a parvalbumin-expressing cell body (S). Fibers indicated with arrows. Bar = 0.5 Am. (D) 3D reconstructed contacts between a presubiculum bouton (B) and a calretinin-expressing dendrite (D). Arrows point at BDA-labeled fiber. Bar = 0.5 Am.

74

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

6. Discussion With the present method we studied the distribution of a particular contingent of labeled fibers and their contacts with neurons belonging to two different chemical classes. Our observations were focused on triple-stained sections. These sections contained fibers labeled with the neuroanatomical tracer, and they contained as well populations of potential target neurons each characterized by the expression of its exclusive marker, i.e., parvalbumin or calretinin. By virtue of the presence in the same section of structures labeled with each of the three markers we were able to establish whether there was preferential innervation of neurons belonging to one of the populations of possible target neurons, and in addition to determine the proportional distribution of the incoming fibers over the individual parvalbumin- and calretinin-immunoreactive target neurons. Finally, we were able to study which compartments of the target neurons were involved in the contacts made by the fibers. We found a slight preference of presubiculum fibers to form contacts with calretinin neurons compared to parvalbumin neurons. It should be mentioned here that neurons in layer III of MEA expressing parvalbumin far outnumber neurons expressing calretinin. In absolute numbers, more parvalbumin expressing neurons may therefore receive contacts from presubiculum fibers than calretinin expressing neurons. Parvalbumin neurons received most contacts on their cell bodies next to contacts on their dendrites, and calretinin cells received most contacts on dendrites next to contacts on their cell bodies. Thus, contacts of boutons of presubiculum fibers showed in relative terms asymmetry in their distribution among parvalbumin- or calretinin-immunoreactive neurons. The distribution of contacts of boutons over cell compartments was different in parvalbumin neurons compared with calretinin neurons. The major advantage of the present approach versus that of using adjacent sections is that in the present approach the fiber contingent under study is a fixed component and therefore invariable. When two adjacent sections were to be subjected to analysis, one would in essence deal with two separate, albeit strongly related, fiber projections. This variable factor is eliminated in our approach. In addition to this the distribution of target neurons expressing their specific (second and third) markers is also a fixed component in our approach. If adjacent sections were to be used for analysis, the distributions of these markers would vary as well. In addition to their small absolute numbers, calretinin neurons for example express in sections of the medial entorhinal cortex quite some variation in their distribution compared with that of the much more abundant parvalbumin neurons. The present study is not the first study in which it is attempted to determine whether or not target neurons are exclusively innervated. Previously we have focused on attempts to label two fiber projections possibly converging

on the same area or onto the same identified neuron populations [9–11]. The methods used in these studies were all based on conventional light microscopy. The present method however exploits the high resolution of laserscanning microscopy [31–33]. The complementary postacquisition 3D reconstruction to map and study contacts adds the invaluable tool of inspection of potential contacts from every conceivable angle of view. In contrast, the conventional light microscope is limited to only one viewing direction, i.e. the orthoscopic view. Another approach of studying the distribution of a fiber projection with respect to target neurons is by means of separate experiments in which the injection of a tracer is combined with intracellular injection or immunostaining for a particular neuron marker [12,27] and refined with confocal laserscanning [18]. In this type of experiment the accurate replication of the injection site, amount of tracer deposited and amount of tracer taken up and transported always show wide variability. As a consequence the results in this type of experiment are not easily quantifiable. The confocal laserscanning microscope that we used, with its monochromatic excitation light and narrow and controllable emission bandwidth is an eminent instrument to allow the observation of three fluorescence signals independently in one and the same section. Rigorous postacquisition processing of Z-series of confocal images completed with the 3D reconstruction of fibers and targets allows close inspection and reliable identification whether a bouton of a fiber is in apposition with a target neuron process or not [31,33]. Because confocal laserscanning and the subsequent post-acquisition computer processing is such a powerful tool, one has to take a series of precautions ranging from measures to counteract cross-reactivity of antisera [29] to procedures to avoid or deal with crosstalk of fluorochromes during detection in the confocal instrument [24]. Although we studied the distribution of fibers among target neurons belonging to two different chemical categories, the present analysis does not solve the question whether an individual fiber that contacts a neuron of population A also contacts a neuron of population B. At this point, there are several possibilities. One of these is that individual presubiculum neurons innervate either entorhinal parvalbumin-or calretinin expressing neurons. Alternatively, axon collaterals from the same presubiculum neuron could target both parvalbumin and calretinin neurons. A third possibility is a mixture of the above possibilities. Although we did not specifically pursue this type of 3D reconstruction, the present histological preparation lends itself very well to the 3D-reconstruction of single BDA labeled fibers and following these along their course down to their contacts with the processes or cell bodies of neurons expressing one of the two simultaneously studied markers. We found several BDA labeled fibers very close to parvalbumin and calretinin neurons, but we did not further sample or 3D reconstruct these appositions with the purpose

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76

to verify the presence of a contact. Systematic study of this type of innervation may be undertaken using the labeling of individual presubiculum neurons together with the present triple fluorescence procedure. In vitro living brain slices are attractive in this respect since the topography of the presubiculo-entorhinal projection is known in detail [2]. Brain slices can be cut such that the entire connectivity is completely available in a single slice. A labeled presubiculum axon can be traced all the way down from its parent cell body to its branches and terminations in layer III of the medial entorhinal cortex. Because in entorhinal layer III of such a preparation all members of possible connectivity will be available for confocal laserscanning, 3D reconstruction here may provide further insight into to the question of preferential innervation of parvalbumin and calretinin neurons by presubiculum neurons.

7. Essential literature references J.B. Pawley (ed), Handbook of Biological Confocal Microscopy, 2nd ed. Plenum Press, New York, 1995.

Acknowledgments We are much indebted to Peter Goede for his perfect blend of enthusiasm, good humor and skillful immunohistochemical assistance. Access to the confocal scanning instrument was granted by kind permission of the Department of Pathology, Vrije University Medical Center. Images A and B of Fig. 3 were acquired by Amber Boekel. The continuous involvement and advice by Nico Blijleven with respect to hardware and software matters related to deconvolution and 3D reconstructions is much appreciated. Menno Witter is thanked for his supportive discussions and criticism on various drafts of this manuscript.

References [1] M. Bertero, P. Boccacci, G.J. Brakenhoff, F. Malfanti, H.T.M. van der Voort, Three-dimensional image restoration and super-resolution in fluorescence confocal microscopy, J. Microsc. 157 (1990) 3 – 20. [2] M. Caballero-Bleda, M.P. Witter, Regional and laminar organization of projections from the presubiculum and parasubiculum to the entorhinal cortex: an anterograde tracing study in the rat, J. Comp. Neurol. 328 (1993) 115 – 129. [3] D. Gardella, W.J. Hatton, H.B. Rind, G.D. Rosen, C.S. von Bartheld, Differential shrinkage and compression in the Z-axis: implications for optical disector counting in vibratome-, plastic- and cryosections, J. Neurosci. Methods 124 (2003) 45 – 59. [4] H.J. Groenewegen, F.G. Wouterlood, Light and electron microscopic tracing of neuronal connections with Phaseolus vulgaris-leucoagglutinin (PHA-L), and combinations with other neuroanatomical techniques, in: F.G. Wouterlood, A.N. van den Pol, A. Bjfrklund, T. Hfkfelt (Eds.), Analysis of Neuronal Microcircuits and Synaptic Interactions, Handbook of Chemical Neuroanatomy, vol. 8, Elsevier Biomedical Press, Amsterdam, 1990, pp. 47 – 124. Chap. 2.

75

[5] H.J. Gundersen, E.B. Jensen, The efficiency of systematic sampling in stereology and its prediction, J. Microsc. 147 (1987) 229 – 263. [6] P.R. Hiesinger, M. Scholz, I.A. Meinertzhagen, K.-F. Fischbach, K. Obermayer, Visualization of synaptic markers in the optic neuropils of Drosophila using a new constrained deconvolution method, J. Comp. Neurol. 429 (2001) 277 – 288. [7] B. Jonkers, J. Sterk, F.G. Wouterlood, Transcardial perfusion fixation of the CNS by means of a compressed-air-driven device, J. Neurosci. Methods 12 (1984) 141 – 149. [8] C. Kfhler, Morphological details of the projection from the presubiculum to the entorhinal area as shown with the novel PHAL immunohistochemical tracing method in the rat, Neurosci. Lett. 45 (1984) 285 – 290. [9] J.L. Lanciego, P.H. Goede, M.P. Witter, F.G. Wouterlood, Use of peroxidase substrate vector VIP for multiple staining in light microscopy, J. Neurosci. Methods 74 (1997) 1 – 7. [10] J.L. Lanciego, F.G. Wouterlood, E. Erro, J.M. Gimenez-Amaya, Multiple axonal tracing: simultaneous detection of three tracers in the same section, Histochem. Cell Biol. 110 (1998) 509 – 515. [11] J.L. Lanciego, F.G. Wouterlood, E. Erro, J. Arribas, N. Gonzalo, X. Urra, S. Cervantes, J.M. Gimenez-Amaya, Complex brain circuits studied via simultaneous and permanent detection of three transported neuroanatomical tracers in the same histological section, J. Neurosci. Methods 103 (2000) 127 – 135. [12] G.E. Meredith, F.G. Wouterlood, Identification of synaptic interactions of intracellularly injected neurons in fixed brain slices by means of dual-label electron microscopy, Microsc. Res. Tech. 24 (1993) 31 – 42. [13] M. Miettinen, A. Pitk7nen, R. Miettinen, Distribution of calretininimmunoreactivity in the rat entorhinal cortex: coexistence with GABA, J. Comp. Neurol. 378 (1997) 363 – 378. [14] G. Paxinos, C. Watson, The Rat Brain in Stereotaxic Coordinates, 2nd ed., Academic Press, New York, 1986. [15] A. Reiner, C.L. Veenman, M. Honig, Anterograde tracing using biotinylated dextram amine, Neurosci. Prot. (1993) (93-050-14-01). [16] D.L. Snyder, T.J. Schulz, J.A. O’Sullivan, Deblurring subject to nonnegativity constraints, IEEE Trans. Signal Process. 40 (1992) 1143 – 1150. [17] J.F. Staiger, F.G. Wouterlood, Efferent projections from the lateral septal nucleus to the anterior hypothalamus in the rat: a study combining Phaseolus vulgaris-leucoagglutinin tracing with vasopressin immunocytochemistry, Cell Tissue Res. 261 (1990) 17 – 23. [18] J.F. Staiger, D. Schubert, W. Zuschratter, R. Kftter, H.J. Luhmann, K. Zilles, Innervation of interneurons immunoreactive for VIP by intrinsically bursting pyramidal cells and fast-spiking interneurons in infragranular layers of juvenile rat neocortex, Eur. J. Neurosci. 16 (2002) 11 – 20. [19] H.T.M. van der Voort, K.C. Strasters, Restoration of confocal images for quantitative image analysis, J. Microsc. 158 (1995) 43 – 45. [20] T. van Groen, J.M. Wyss, The connections of presubiculum and parasubiculum in the rat, Brain Res. 518 (1990) 227 – 243. [21] T. van Haeften, F.G. Wouterlood, B. Jorritsma-Byham, M.P. Witter, GABAergic presubicular projections to the medial entorhinal cortex of the rat, J. Neurosci. 17 (1997) 862 – 874. [22] W. Wallace, L.H. Schaefer, J.R. Swedlow, A workingperson’s guide to deconvolution in light microscopy, BioTechniques 31 (2001) 1076 – 1097. [23] M.P. Witter, F.G. Wouterlood, P.A. Naber, T. van Haeften, Anatomical organization of the parahippocampal-hippocampal network, in: H.E. Scharfman, M.P. Witter, R. Scwarz (Eds.), The Parahippocampal Region, Ann. N.Y. Acad. Sci., vol. 911, 2000, pp. 1 – 24. [24] F.G. Wouterlood, 3D reconstruction of neurons from multichannel confocal laserscanning image series, in: S. Wray (Ed.), Current Protocols in Neuroscience, Wiley and Sons, New York, 2004, in press. [25] F.G. Wouterlood, J.G.J.M. Bol, H.W.M. Steinbusch, Double-label immunocytochemistry: combination of anterograde neuroanatomical tracing with Phaseolus vulgaris-leucoagglutinin and enzyme immu-

76

[26]

[27]

[28]

[29]

M. Vinkenoog et al. / Brain Research Protocols 14 (2005) 67–76 nocytochemistry of target neurons, J. Histochem. Cytochem. 35 (1987) 817 – 823. F.G. Wouterlood, B. Jorritsma-Byham, P.H. Goede, Combination of anterograde tracing with Phaseolus vulgaris-leucoagglutinin, retrograde fluorescent tracing and fixed-slice intracellular injection of Lucifer Yellow, J. Neurosci. Methods 33 (1990) 207 – 217. F.G. Wouterlood, P.H. Goede, M.P.M. Arts, H.J. Groenewegen, Simultaneous characterization of efferent and afferent connectivity, neuroactive substances and morphology of neurons, J. Histochem. Cytochem. 40 (1992) 457 – 465. F.G. Wouterlood, W. H7rtig, G. Brqckner, M.P. Witter, Parvalbuminimmunoreactive neurons in the entorhinal cortex of the rat: localization, morphology, connectivity and ultrastructural, J. Neurocytol. 24 (1995) 135 – 153. F.G. Wouterlood, J.C.M. van Denderen, N. Blijleven, J. van Minnen, W. H7rtig, Two-laser dual-immunofluorescence confocal laser scanning microscopy using Cy2- and Cy5-conjugated secondary antibodies: unequivocal detection of co-localization of neuronal markers, Brain Res. Protoc. 2 (1998) 149 – 159.

[30] F.G. Wouterlood, J. van Denderen, T. van Haeften, M.P. Witter, Calretinin in the entorhinal cortex of the rat: distribution, morphology, ultrastructure of neurons, and co-localization with gamino butyric acid and parvalbumin, J. Comp. Neurol. 425 (2000) 177 – 192. [31] F.G. Wouterlood, T. van Haeften, N. Blijleven, P. Pe´rez-Templado, H. Pe´rez-Templado, Double-label confocal laserscanning microscopy, image restoration and real-time 3D-reconstruction to study axons in the CNS and their contacts with target neurons, Appl. Immunohistochem. Mol. Morphol. 10 (2002) 85 – 102. [32] F.G. Wouterlood, T. Bfckers, M.P. Witter, Synaptic contacts between identified neurons visualized in the confocal laserscanning microscope. neuroanatomical tracing combined with immunofluorescence detection of postsynaptic density proteins and target neuron-markers, J. Neurosci. Methods 128 (2003) 129 – 142. [33] F.G. Wouterlood, T. van Haeften, M. Eijkhoudt, L. Baks-Te-Bulte, P.H. Goede, M.P. Witter, Input from the presubiculum to dendrites of layer-V neurons of the medial entorhinal cortex of the rat, Brain Res. 1013 (2004) 1 – 12.