anatomic staining_2008

Embed Size (px)

Citation preview

  • 8/7/2019 anatomic staining_2008

    1/12

    A R T I C L E

    A Rapid Method Combining Golgi and Nissl Staining to StudyNeuronal Morphology and Cytoarchitecture

    Nadia Pilati, Matthew Barker, Sofoklis Panteleimonitis, Revers Donga, and Martine Hamann

    Department of Cell Physiology and Pharmacology (NP,MB,SP,MH) and Departments of Infection, Immunity, andInflammation and Medical and Social Care Education (SP,RD), Leicester University, Leicester, United Kingdom

    S U M M A R Y The Golgi silver impregnation technique gives detailed information on neu-

    ronal morphology of the few neurons it labels, whereas the majority remain unstained. In

    contrast, the Nissl staining technique allows for consistent labeling of the whole neuronal

    population but gives very limited information on neuronal morphology. Most studies char-

    acterizing neuronal cell types in the context of their distribution within the tissue slice

    tend to use the Golgi silver impregnation technique for neuronal morphology followed by

    deimpregnation as a prerequisite for showing that neurons histological location by sub-

    sequent Nissl staining. Here, we describe a rapid method combining Golgi silver impregnation

    with cresyl violet staining that provides a useful and simple approach to combining cellular

    morphology with cytoarchitecture without the need for deimpregnating the tissue. Our

    method allowed us to identify neurons of the facial nucleus and the supratrigeminal nucleus,

    as well as assessing cellular distribution within layers of the dorsal cochlear nucleus. With this

    method, we also have been able to directly compare morphological characteristics of neu-

    ronal somata at the dorsal cochlear nucleus when labeled with cresyl violet with those ob-

    tained with the Golgi method, and we found that cresyl violetlabeled cell bodies appear

    smaller at high cellular densities. Our observation suggests that cresyl violet staining is in-

    adequate to quantify differences in soma sizes. (J Histochem Cytochem 56:539550, 2008)

    K E Y W O R D S

    Lucifer yellow

    neuron

    dendrite

    soma

    THE GOLGI SILVER impregnation method is a powerfulmethod still routinely used for studying neuronal mor-phology (Hani et al. 2007; Mendizabal-Zubiaga et al.2007). Its usefulness for quantitative analysis of labeledneurons is limited by its capricious nature. In contrast,the Nissl staining approach allows for the visualizationof all somata in appropriately prepared tissue sectionswhile being poor in labeling neuronal processes. Thus,a combined approach using the Golgi silver impregna-tion technique and the Nissl staining method would

    allow for the establishment of the detailed morphologi-cal profiles of neurons within a nucleus or a laminarstructure. Although it is possible to counterstain cellspreviously treated with the Golgi silver impregnation

    technique with the Nissl stain (Werner and Brauer1984; Werner et al. 1986,1989), a drawback with thisapproach is that the Golgi-stained neurons must bedeimpregnated before counterstaining with Nissl stains.Under these conditions, most Golgi-labeled neuronslose their morphological characteristics before theytake up the Nissl stain. Recently, Friedland et al. (2006)described staining for cresyl violet around neurons la-beled with the Golgi method that was rather faint andsuggested that pH changes during the Golgi staining

    might reduce affinity for cresyl violet. We report herethe first successful combination of Golgi and cresylviolet staining methods that allowed us to simulta-neously characterize the morphology of individualneurons together with the cytoarchitecture of the ratdorsal cochlear nucleus (DCN). Our method also al-lowed us to identify neuronal somata within thebrainstem or cerebellum and to directly compare themorphological characteristics of the somata when la-beled with cresyl violet and with the Golgi silver im-pregnation method.

    Correspondence to: Martine Hamann, Department of Cell Physi-ology and Pharmacology, Medical Sciences Building, PO Box 138,University Road, Leicester LE1 9HN, UK. E-mail: [email protected]

    Received for publication November 15, 2007; accepted Febru-ary 7, 2008 [DOI: 10.1369/jhc.2008.950246].

    Deposited to PMC for immediate release

    C The Histochemical Society, Inc. 0022-1554/08/$3.30 539

    Volume 56(6): 539550, 2008

    Journal of Histochemistry & Cytochemistry

    http://www.jhc.org

  • 8/7/2019 anatomic staining_2008

    2/12

  • 8/7/2019 anatomic staining_2008

    3/12

    taken at a 0.5-mm interval throughout the slice andput into a single-focused image with the OlympusFluoview software.

    Analysis

    Nissl-stained Cells. Analysis of cell density and cellsoma surface areas was performed on two to three slicesper rat, and three rats were used per staining condition.The whole DCN was cut from its caudal to its rostralend, and DCN slices were analyzed at three depthsthroughout the nucleus (i.e., 100, 250, and 400 mmdeep). Analysis of cell numbers and cell soma surfaceareas was performed using rectangular areas (89 355 mm) placed as a fixed matrix (as in King et al. 2002)of three rows and three columns, and the three rowsrepresented the three layers: the external (molecular),intermediate (fusiform), and internal (deep) layers. Inthe vertical plane, each rectangle was separated by a

    60-m

    m gap between the molecular and the fusiformlayer and by a 100-mm gap between the fusiform andthe deep layer. In the horizontal plane, rectangles wereseparated from each other by a 90-mm gap. Analysis ofcell density and cell soma surface area was performedon all cells that were in the optical focal plane lyingwithin or were crossing the inclusion boundaries, usinga method similar to that of King et al. (2002). Cell den-sity was measured per area and not volume. Cell somasurface was analyzed for all cell profiles contained

    within the six rectangles (two rectangles per layer). Cellsoma surface analysis was done by outlining thesoma border but excluding the emerging dendritesand axon using the method and the terms described byMcDonagh et al. (2002) and Zwaagstra and Kernell(1981). Only cell soma surface areas $39 mm2 corre-

    sponding to a diameter $7 mm were included in theanalysis. This lower limit was set to include granulecells known to have diameters ofz79 mm (Mugnainiet al. 1980b; Alibardi 2003). A diameter lower limitof 7 mm also excluded glial cells from the counting(Skoglund et al. 1996). Area localization was based onthe rat stereotaxic atlas of Paxinos and Watson (1998).

    Golgi- and Lucifer Yellowstained Cells. The numberof Golgi-labeled cells was estimated throughout thetotal DCN, and both left and right sides of three brainswere used for every staining condition reported in Fig-ure 1, giving a total of 6174 slices analyzed per condi-

    tion. The surface of the non-neuronal (a-specific) Golgistaining (defined as all the black to orange precipitates)was estimated in 8 DCN slices per rat from both leftand right sides, giving a total number of 24 slices ana-lyzed per condition (three rats per condition). Surface(or profile) areas were measured using the Image-J1.36 freehand selection tool (National Institutes ofHealth; Bethesda, MD), allowing selection of thecontour of the DCN, the nonspecific staining, and thecell soma surface area together, with the Image-J 1.36

    Figure 1 Golgi staining of dorsal co-chlear nucleus (DCN) neurons is de-pendent on the chromating solution.Summary histograms representing thefollowing: white, mean 6 SD numberof labeled cells per DCN slice (n56);

    black, mean 6 SD percentage of areanot specifically stained relative to thetotal DCN area (n56). CH, chloral hy-drate; GT, glutaraldehyde; PF, parafor-maldehyde. Inset at left is an exampleof a neuron-specific labeling high-lighted by arrow (potassium dichro-mate solution was dissolved in PBS,pH 7.6, and 4% paraformaldehyde plus2% chloral hydrate). Inset at right is anexample of nonspecific staining high-lighted by white asterisks (potassiumdichromate dissolved in H2O and 5%paraformaldehyde). Bar 5 50 mM.

    A Method Combining Golgi and Nissl Staining 541

  • 8/7/2019 anatomic staining_2008

    4/12

    area calculator plug-ins, allowing estimating the sur-face areas.

    Statistics

    Data are expressed as mean 6 SD. Unless specifiedotherwise, data were analyzed by one- or two-factor

    ANOVA (general linear model) tests followed byTukeys test using the statistical software ezANOVA(http://www.sph.sc.edu/comd/rorden/ezanova/home.html) or Mini Tab 14. Post hoc power analysis wasperformed using Gpower 2.0 software (http://www.psycho.uni-duesseldorf.de/aap/projects/gpower/).

    Results

    We started by setting up optimal staining conditions forlabeling individual neurons of the rat DCN using therapid Golgi silver impregnation method before combin-ing it with cresyl violet staining so that we could char-

    acterize the morphology of individual neurons withinthe context of layers of the DCN.

    Cell Labeling Using the Rapid Golgi Method

    We used the rapid variant of the Golgi silver impregna-tion method to label neurons of the DCN over 4 days.This labeling method can be divided into two steps.First, cells were impregnated with both trivalent andhexavalent chromium ions over 2 days, and this isknown as the chromation step. This process requiresthe presence of aldehydes and is known to be influencedby time, temperature, and pH (Kopsch 1896; Colonnier1964). The chromation step was followed by a precipi-

    tation step also lasting 2 days, in which the chromiumions react with heavy metal ions, silver in this case, togive a chromogenic reaction product that serves as thevisual marker of the labeled cells. We optimized ourprotocols by experimenting with various aldehyde do-nors during the chromation step, namely paraformal-dehyde, glutaraldehyde, or chloral hydrate, and alsousing phosphate buffer to give us a pH range of 5.87.6. In another set, the chromation step was carried outin the absence of buffer with or without paraformal-dehyde (5%). The quality of the Golgi staining wasassessed by determining whether the staining was con-fined to neurons (pointed with the arrow in Figure 1,

    left inset) or to nonspecific staining (shown as asterisksin Figure 1, right inset). The stained neuron in Figure 1showed a clearly visible cell body with a pyramidalshape and dendrites, whereas nonspecific labeling inFigure 1 had an undefined patchy presentation insteadof a neuronal shape. Figure 1 also shows a histogramplot illustrating the effects of pH and aldehyde on neu-ronal and non-neuronal staining by plotting the aver-age number of labeled neurons per DCN slice (whitebars) and the surface area of the non-neuronal stainingrelative to the total DCN area (black bars). We quan-

    tified the number of labeled cells and the extent of thenon-neuronal staining and compared the differentconditions using one-factor ANOVA and Tukey tests.We found that the highest number of labeled cells wereobtained when we used either PBS, pH 7.6, with 5%paraformaldehyde or PBS, pH 7.6, with 4% parafor-maldehyde and 2% chloral hydrate (Figure 1, whitebars; Tables 1 and 2, left columns). Using a PBS solu-tion, pH 7.6, with 5% paraformaldehyde also led toreduced non-neuronal labeling (Figure 1, black bars;Tables 1 and 2, right columns). Using a PBS solution,pH 7.6, with glutaraldehyde instead of paraformalde-hyde reduced the number of labeled cells (Figure 1),suggesting that glutaraldehyde is not ideal for neuronalstaining with the Golgi silver impregnation method.Similarly, more acidic pH conditions or aqueous solu-tions produced only few labeled neurons (Figure 1;

    Table 1, left column).Cell Labeling Using the Combined Cresyl VioletRapid

    Golgi Method

    Brains were previously labeled with the Golgi methodusing the optimal chromating conditions in a PBS solu-tion, pH 7.6, with 5% paraformaldehyde, and slicescontaining the DCN were subsequently counterstainedwith cresyl violet. In addition to acting as a control stainof the neuronal tissue, cresyl violet allowed us to study

    Table 1 One-factor ANOVATukey tests comparing the labeling

    obtained in various chromating solutions

    PBS (pH 7.6) and 5% PF vs

    condition below

    Number of

    labeled cells

    Percentage of

    aspecific staining

    H2O, 5% PF p,0.0001 p50.18

    H2O p,0.0001 p50.09

    PBS (pH 5.8), 5% PF p,

    0.0001 p5

    0.29PBS (pH 7.4), 5% PF p,0.0001 p,0.05

    PBS (pH 7.6), 5% PF, 2% GT p,0.0001 p,0.01

    PBS (pH 7.6), 4% PF, 2% CH p50.74 p,0.05

    p values refer to comparing the chromating condition in PBS, pH 7.6, and 5%paraformaldehyde vs the other conditions. p values are reported for the num-berof labeledcells perDCN slice andthe nonspecificallystained area relative tothe total DCN area. PF, paraformaldehyde; GT, glutaraldehyde; CH, chloralhydrate; DCN, dorsal cochlear nucleus.

    Table 2 One factor ANOVATukey tests comparing the labeling

    obtained in various chromating solutions

    PBS (pH 7.6), 4% PF and

    2% CH vs condition below

    Number of

    labeled cells

    Percentage of

    aspecific staining

    H2O, 5% PF p,0.001 p50.60

    H2O p,0.0001 p50.78

    PBS (pH 5.8), 5% PF p,0.0001 p50.32

    PBS (pH 7.4), 5% PF p,0.001 p50.55

    PBS (pH 7.6), 5% PF, 2% GT p,0.001 p50.08

    PBS (pH 7.6), 5% PF p50.74 p,0.05

    p values refer to comparing the chromating condition in PBS, pH 7.6, 4%paraformaldehyde, and 2% chloral hydrate vs the otherconditions.p values arereported for the number of labeled cells per DCN slice and the nonspecificallystained area relative to the total DCN area. See Table 1 for abbreviations.

    542 Pilati, Barker, Panteleimonitis, Donga, Hamann

  • 8/7/2019 anatomic staining_2008

    5/12

    the silver-impregnated neurons in the context of theDCN laminated structure. While staining the tissue withcresyl violet, studies report first hydrating the sliceswith distilled water, then staining them with cresylviolet, and finally dehydrating them by using progres-sively graded solutions of ethanol, namely 5070% and

    100% (Gittins and Harrison 2004; Friedland et al.2006). In our experience, we found that this commonlyused approach produced cracking within Golgi-labeledtissue, leading us to modify our protocol such thatsections were transferred directly to the cresyl violetsolution without prior hydration and from the cresylviolet solution directly to 100% ethanol for 2 min.

    Figure 2 shows the morphologies of silver-impregnatedneurons before (Figure 2A) and after (Figure 2B)counterstaining the tissue with cresyl violet. The cresylviolet stain shows laminated areas of higher density oflabeling suggestive of cell layering within the DCN,whereas Golgi-stained cells showed typical neuronal

    morphologies. We also used the combined Golgicresylviolet staining in slices containing the cerebellum andother brainstem nuclei easily recognizable by the cresylviolet staining. Figure 3A shows cerebellar Purkinjecells aligned in the Purkinje cell layer, with their typicalspiny dendritic tree oriented toward the molecular layer(Palay and Chan-Palay 1974; Friedland et al. 2006).

    Figure 2 Combined cresyl violet andGolgi labeling of DCN neurons andcytoarchitecture. (A) Photomicro-graph showing two DCN fusiformneurons and a giant cell stained withthe Golgi method. In B, the same fusi-form cells (Fu) and giant cell (Gi) areshown after counterstaining the slicewith cresyl violet. The chromatingsolution contained potassium dichro-mate dissolved in PBS, pH 7.6, and 5%paraformaldehyde. (CF) Photomicro-graphs of DCN cell types obtained af-ter labeling slices with the combinedGolgicresyl violet method (same con-ditions as for B). (C) Arrow points toanother fusiform cell (Fu) in the fusi-form cell layer with its large elongatecell body and its basal dendrites lyingin the deep layer, whereas the apicaldendrites are oriented toward the

    molecular layer. (D) Arrow indicatesa cartwheel cell (Cw) with its smalloval cell body between the fusiformand the molecular layer, an axon ex-tending into the fusiform layer, anda large spiny dendritic tree in themolecular layer. (E) Arrow points to agranule cell (gr) in the deep layer withtwomain dendritesending in claw-likeprotuberances. (F) Arrow indicates agiant cell (Gi) with its large soma ex-ceeding 30 mm diameter and manythick dendrites, both located in thedeep layer.

    A Method Combining Golgi and Nissl Staining 543

  • 8/7/2019 anatomic staining_2008

    6/12

  • 8/7/2019 anatomic staining_2008

    7/12

    diallayer and the celldensity of7.261.0 cells/1000mm2

    (n527) in the inner layer (p,0.0001, two-factorANOVA test). Cells in the intermediate layer were alsocharacterized by a larger soma surface area (112 618 mm2, n518) compared with cells within the externallayer (51 6 6 mm2, n518) and with cells within theinner layer (586 11 mm2, n518, p5 0.002, two-factorANOVA test). Cell soma surface areas were similar forthe same layer when the slices were chosen at three dif-ferent depths within the DCN (n56 rectangles and 3 rats,two-factor ANOVA test, p5 0.167). Our DCN cell layer

    organization is similar to the DCN laminar orga-nization reported in previous studies (see Discussion),and we refer to the external layer as the molecular layer,the intermediate layer as the fusiform cell layer, andthe inner layer as the deep layer. The combined Golgicresyl violet staining method therefore made it possibleto localize the different cells within the three layerswith a high degree of certainty. Figure 2 gives an over-view of some morphological cell types observed in thisstudy. Within the fusiform cell layer, fusiform cells arecharacterized by an elongated cell body of 34 6 4 mm

    diameter (n55) across their major axis and a diameterof 196 3 mm (n55) across their minor axis (Figures 2Band 2C) (Maruyama and Ohmori 2006). Fusiform cellsare also characterized by basal dendrites directed to-ward the deep layer and apical dendrites directed to-ward the molecular layer. Usually, the apical dendritedivides into several branches (Figure 2C). Cartwheelcells are found at the boundary between the molecularlayer and the fusiform cell layer and have a diameter of14 and 22 mm (n52), as well as a thick primary den-drite that gives rise to arborizations in the form of a

    tree that extends into the molecular layer (Figure 2D;Wouterlood and Mugnaini 1984). Giant cells with acell body exceeding 30-mm diameter (diameter of 38 67 mm, n56) (Zhang and Oertel 1993) are characterizedby multiple thick dendrites projecting along the deeplayer and toward the fusiform cell layer (Figures 2Band 2F). Small-diameter cells likely to be granule cells(Mugnaini et al. 1980b) with an oval cell body and twoto three primary dendrites were mainly found in thedeep layer (diameter of 116 2mm, n53) and also in themolecular layer (diameter of 96 2mm, n53). Figure 2E

    Figure 4 Analysis of the DCN architec-ture labeled with cresyl violet beforeand after Golgi staining. (A,B) Photo-micrographs of the DCN stained withcresyl violet (A) and with cresyl violetand Golgi labeling (B). In A and B,dashed lines define the boundaries be-tween the external or molecular layer(ML), the intermediate or fusiform celllayer (FL), and the inner or deep layer(DL). For each layer, areas within therectangles are analyzed (three rectan-gles shown in A). (C) Summary plotrepresenting the mean 6 SD valuesof the cell soma surface area (n51618 rectangles) and of the cell density(n52427 rectangles) for the differentlayers. Slices were labeled with cresylviolet (black circles) or with the com-bined Golgi and cresyl violet labelingmethod (white circles). The cell den-sity and soma surface areas were ana-lyzed from cresyl violetstained cells inboth conditions.

    A Method Combining Golgi and Nissl Staining 545

  • 8/7/2019 anatomic staining_2008

    8/12

    shows a granule cell with its classically described den-dritic claws (Mugnaini et al. 1980b).

    DCN Cytoarchitecture After Golgi Staining

    We checked whether the Golgi staining procedures didnot interfere fundamentally with the native cytoarchi-tecture of the DCN and quantified the cell density andthe soma surface area of cresyl violetstained sectionspreviously stained with the Golgi method. Figure 4Crepresents an analysis of the DCN architecture oncresyl violetstained cells without any previous Golgistaining (black circles) and after having previouslylabeled the tissue with the Golgi method (white circles).After Golgi and cresyl violet staining, DCN layers couldstill be distinguished by differences in their cell densityand cell soma surface area. The external or molecularlayer was still characterized by the lowest cell density of1.7 6 0.6 cells/1000 mm2 (n524 rectangles) comparedwith the cell density of 3.8 6 1.4 cells/1000 mm2 in theintermediate or fusiform layer (n524) and to the celldensity of 6.76 0.7 cells/1000 mm2 (n524) in the inneror deep layer (p,0.00001, two-factor ANOVA test).After Golgi and cresyl violet staining, cells in thefusiform layer were also characterized by a larger somasurface area (956 23 mm2, n516 rectangles) comparedwith cells within the molecular layer (55 6 9 mm2,n516) and with cells within the inner or deep layer(59 6 6 mm2, n516, p,0.00001, two-factor ANOVAtest). Within each layer, Nissl-stained cell bodies had asimilar soma surface area with or without previouslylabeling the tissue with the Golgi method (two-factorANOVA test, p 5 0.58, n55 and 6 rectangles, respec-tively, containing between 100 and 400 cells per layer).Power analysis with values of 80% and 100% for100 and 400 cells, respectively, indicates that our sam-ple size is sufficient to detect a 10% variation of cellsoma surface areas between the two staining condi-tions, further suggesting similar Nissl staining condi-tions in the presence or absence of Golgi staining. Celldensities were also similar for the same layer when theslices were chosen at three different depths within theDCN (n59 rectangles and 3 rats, two-factor ANOVAtest, p 5 0.602), supporting the idea that the layer dis-tribution is uniform throughout the DCN. Cell den-

    sity was also similar within each particular layer, withor without previously labeling tissue with the Golgimethod (two-factor ANOVA test, p5 0.73, n59 and 6,respectively), showing that Nissl-stained layer charac-teristics stayed unchanged despite Golgi labeling.

    Estimation of the Soma Surface Areas After Golgi

    or Cresyl Violet Staining

    Although we found no difference in the Nissl-stainedcytoarchitecture when we analyzed the cell bodieslabeled with cresyl violet, we noticed that some cells

    labeled with the Golgi method had a bigger cell somasurface area compared with the average cell somasurface area obtained with cresyl violet staining. Forexample, giant cells are characterized by a soma surfacearea of 426 6 181 mm2 (n516) while stained with theGolgi method. This value exceeds by 2.5 times the

    highest values of the cell soma area measured in cresylviolet staining condition (172 6 63 mm2, Student t-test,p 5 1 31025, n516). We represented the ability of thetwo staining methods to differently label the cell somasurface area by their cumulative distributions of thesoma surface areas (Figure 5). Cumulative distributionswere represented for each cell layer (Figures 5A5C)and seem to be different for the fusiform cell layer(Figure 5B) and the deep layer (Figure 5C), whereas nomajor difference was observed for the molecular layer(Figure 5A). This suggests that Golgi-labeled cells arebigger than cresyl violetstained cells within the fusi-form and the deep layers. This difference in the cell

    soma surface area between Golgi- and Nissl-stainedcells can be represented as the area difference betweenthe two curves represented in Figures 5A5C. In themolecular layer (Figure 5A), 50% of cells have a somasurface area 953 mm2 when labeled with cresyl violet(n567), and 50% of cells have a soma surface area960 mm2 when labeled with the Golgi method (n576).Maximal values (representing the total number ofcells) are similar for both staining conditions (reachingz100 mm2). In the fusiform cell layer (Figure 5B),50% of cells have a soma surface area 964 mm2 whenlabeled with cresyl violet (n5284), and 50% of cellshave a soma surface area 9107 mm2 (n5291) when

    labeled with the Golgi method. Cell soma surface areasdo not exceed 490 mm2 when stained with cresyl violet,whereas cell soma surface areas of Golgi-stained cellsreach 674 mm2 when stained with the Golgi method. Inthe deep layer (Figure 5D), 50% of cells have a somasurface area 958 mm2 when labeled with cresyl violet(n5488), and 50% of cells have a soma surface area981 mm2 when labeled with the Golgi method(n5377 cells). Cell soma surface area does not exceed398 mm2 when staining with cresyl violet, whereas cellsoma surface area of Golgi stained cells reach values of883 mm2. We further tested whether there was anycorrelation between this area difference and the cell

    density reported for the different layers. Figure 5Dshows a correlation (r2 5 0.83) between the differencein the cell surface soma area obtained between the twostaining methods and the cell density, indicating thatbigger cells within a dense layer are more likely to beunderestimated when quantified with the cresyl violetstaining method. We estimated the soma surface area ofcresyl violet stained cells using a standard cresyl violetstaining method that used progressive dehydration andfound similar soma surface area values using the stan-dard cresyl violet staining procedure or the abbreviated

    546 Pilati, Barker, Panteleimonitis, Donga, Hamann

  • 8/7/2019 anatomic staining_2008

    9/12

    cresyl violet staining of 55 6 17 (61 cells) and 53 6

    19m

    m

    2

    (87 cells), respectively, in the molecular layer(p5 0.39, unpaired Student t-test), 1236 54 (88 cells)and 106 6 98 mm2 (286 cells), respectively, in the fusi-form layer (p5 0.12, unpaired Student t-test), and 56617 (124 cells) and 586 24mm2 (390 cells), respectively,in the deep layer (p5 0.48, unpaired Student t-test). Wefinally filled cartwheel, fusiform, and giant cells withLucifer yellow to compare their cell soma surface areawith those obtained with the Golgi method and foundsimilar values of soma surface areas when cells werefilled with Lucifer yellow or when labeled with the Golgimethod (Figure 6).

    DiscussionThe aim of this work was to set up a rapid methodcombining Golgi and Nissl staining so that we couldstudy detailed neuronal morphology together withcytoarchitecture without using any form of intracellularlabeling such as horseradish peroxidase (Oertel et al.1990). We were able to characterize neuronal morphol-ogy directly from single tissue sections without havingto reconstruct the neurons of interest (Blackstad et al.1984). Using cresyl violet allowed us to characterize thecell layers within the DCN similar to those reported in

    previous studies from different species or different rat

    strains (Osen 1969; Brawer et al. 1974; Mugnaini et al.1980a,b; Browner and Baruch 1982; Hackney et al.1990). We used the Nissl stain alongside the Golgisilver staining method to place labeled neurons withinthose layers, and we were able to counterstain our tis-sue sections without the need to deimpregnate the silverstain as has been the case in previously published lit-erature (Pasternak and Woolsey 1975; Werner andBrauer 1984; Werner et al. 1986,1989). We also usedthis rapid method combining Golgi and Nissl stainingin other brain areas. We had to overcome two majorobstacles that attend the Golgi silver impregnationtechnique, first on its own and second, in combination

    with cresyl violet.

    Cell Labeling Using the Rapid Golgi Method

    The Golgi silver impregnation method is unreliable be-cause potassium dichromate and silver nitrate can reactunselectively, forming bulk crystals on the surface ofthe specimen in question (Pasternak andWoolsey 1975);therefore, the Golgi silver impregnation method hasbeen modified by adding either formaldehyde to thepotassium dichromate solution (Kopsch 1896) or glu-taraldehyde (Colonnier 1964) to the potassium dichro-

    Figure 5 Comparative analysis of thecell soma surface area between cresylvioletstained cells and Golgi-stainedcells. A cumulative frequency plotrepresents the cumulative percentageof cells in function ofthe soma surfacearea of cells labeled with cresyl violet(white circle) and with the Golgimethod (black triangle) for the molec-ular layer (A), the fusiform layer (B),and the deep layer (C). (D) The areabetween the two curves representedas white circles and black triangles wascalculated for the three layers andplotted against the cell density valuesof those layers. The correlation factorobtained after linear regression (r2) is0.83. ML, molecular layer; FL, fusiformlayer; DL, deep layer.

    A Method Combining Golgi and Nissl Staining 547

  • 8/7/2019 anatomic staining_2008

    10/12

    mate solution. Those methods have been shown toimprove the quality and the reaction time of the Golgistaining. Other modifications such as using microwaves

    (Marani et al. 1987; Zhang et al. 2003), altering thecomposition of the chromation solution and its pH (Vander Loos 1956; Morest and Morest 1966; Adams 1979;Grandin et al.1988;Angulo et al.1994,1996),or using avacuum (Friedland et al. 2006) improved the labeling ofneurons. We compared the quality of the Golgi stainingby altering the aldehyde types (paraformaldehyde,glutaraldehyde, or chloral hydrate) and the pH of thephosphate buffer used in the chromating solution. Weshowed that potassium dichromate added into aphosphate-buffered medium at a weak basic pH (pH7.6) with 5% paraformaldehyde favored the labeling ofneurons over nonspecific labeling within the tissue. The

    nonspecific staining is likely caused by the uncontrolledchemical reaction between potassium dichromate,which acts as the primary impregnation compound,and the chromogen,silver nitrate. Thischemical reactionisgovernedbytheratioofCr31andCr2O7

    22 ions,whichis dependent on the concentration of protons and thepresence of aldehydes (Angulo et al. 1996). Chromium(III) cross-links the carboxyl terminal of intracellularproteins and binds to Cr2O7

    22 (Angulo et al. 1996).Nevertheless, Cr31 does not react with the silver nitratein contrast to Cr2O7

    22, which produces the black silver

    chromate, thereby defining the morphological profile ofneurons. The low-quality staining obtained at moreacidicpHcouldberelatedtoanexcessofCr31 relative to

    Cr2O722

    . Good labeling conditions are therefore depen-dent on an adequate ratio of Cr31 andCr2O7

    22 (Anguloet al. 1996).

    Cytoarchitecture Labeling and Distribution of the

    Cell Types

    Our analysis of the cell density and soma surface areasallowed us to characterize three layers within the ListerHooded rat DCN similar to previous studies performedin other species or rat strains (Osen 1969; Brawer et al.1974; Mugnaini et al. 1980a,b; Browner and Baruch1982; Webster and Trune 1982; Oertel and Wu 1989;

    Hackney et al. 1990; Alibardi 2006). The peripheral(molecular) layer is characterized by a low density ofcells in Nissl staining, and this is probably because ofthe fact that it contains mainly parallel fibers and pro-cesses of fusiform cells and cartwheel cells. In contrastto the molecular layer, the fusiform and the deep layershad higher densities of cells. Another difference wasthat cells within the fusiform cell layer had larger somasurface areas in Nissl stain compared with cells of themolecular and deep layers. Our results further showedthat the laminar organization of the DCN seems to be

    Figure 6 Comparative analysis of the cell soma surface area between Lucifer yellowfilled cells and Golgi-stained cells. (AC)Photomicrographs of various DCN cells filled with the fluorescent dye Lucifer yellow. ( A) Example of a fusiform cell. (B) Example of acartwheel cell. (C) Example of a giant cell.Below each micrograph, histograms show a summary of the soma surface areas calculated for the celltype represented above, when cells were labeled with the Golgi method (black bars) and when cells were labeled with Lucifer yellow (greenbars). Data are represented as mean 6 SD from 36 neurons. p values of 0.67, 0.15, and 0.32 (Student t-test) for cartwheel cells, fusiform cells,and giant cells, respectively.

    548 Pilati, Barker, Panteleimonitis, Donga, Hamann

  • 8/7/2019 anatomic staining_2008

    11/12

    unaffected by the Golgi staining histological proce-dures. Consequently, our combined NisslGolgi stain-ing method can be used to characterize different celltypes and assign them to specific DCN layers. Fromthe observations made in this study, the predominantDCN cell types have been characterized and mapped

    within the three cell layers. Cartwheel cells are local-ized between the molecular and the fusiform cell layer,whereas fusiform cells are localized in the fusiform celllayer (Mugnaini 1985; Hackney et al. 1990; Maruyamaand Ohmori 2006). Giant cells were localized in thedeep layer (Hackney et al. 1990; Zhang and Oertel1993), and granule cells are present mainly in the deeplayer (Mugnaini et al. 1980b). The cellular compositionand localization within the Lister Hooded rat DCN issimilar to previous studies and therefore validates ourrapid method combining the Golgi silver impregnationtechnique with Nissl staining, without the need fordeimpregnation as an intervening step. Using the Golgi

    silver staining method, we were also able to label mor-phologically identified cells within the supratrigeminalnucleus that could only be located by its position rela-tive to the trigeminal motor nucleus. The presence ofpyramidal and small ovoid neurons within this struc-ture suggests the coexistence of neuronal subgroupsthat could be involved in modulating the process ofmastication (Donga and Lund, 1991).

    Estimation of the Cell Soma Surface Area

    Previous reports are based on quantifying differences insizes and shapes of neuron somata within the cochlear

    nucleus using cresyl violet staining (Seldon and Clark1991; Saada et al. 1996). Our study showed that Golgi-labeled cells had larger cell soma compared with theirequivalents labeled with Nissl stain. This could be ex-plained by Golgi precipitates bursting out of the cellbody, making the cell body appear larger, but this isunlikely to be the case because cells filled with Luciferyellow had similar soma surface areas as those mea-sured in Golgi-stained cells. It is likely that cresyl violetstaining tends to underestimate the cell soma surfacearea because of the Nissl stain being directed primarilyat the Nissl substance of the cytoplasm and not atthe boundaries of the cell (cell membrane). This

    might therefore explain why the general cell outline isunderestimated by Nissl stain, hence the discrepanciesin cell sizes between the Nissl and Golgi staining.

    Given our use of an abbreviated Nissl staining pro-cedure, it is theoretically possible that cell bodies maynot have taken the stain satisfactorily, thereby explain-ing this discrepancy. To control for this possibility, wealso measured neuronal cell bodies using the standardcresyl violet staining method that used progressivetissue dehydration (using graded ethanol solutions).When comparing neuronal staining obtained with the

    full and abbreviated dehydration methods, we foundthat cell body sizes were very similar, suggesting that wewere unlikely to have underestimated neuronal cellbody sizes as a result of the abbreviated dehydrationprocedure. Interestingly, Geisler et al. (2002) reportedan increased staining intensity of the cell bodies when

    cresyl violet at a pH of 4.55 was used in combinationwith Luxol fast blue, and it is possible that the pH of thecresyl violet solution may have affected staining inten-sity, and possibly, neuronal cell body size. Our studyshowed that bigger cells within a dense layer are morelikely to be underestimated when quantified with thecresyl violet staining method. This can be explained bythe cellular overlap at high density making the estima-tion of the cell size unreliable. In contrast, Golgi-stainedcells are easily distinguished from their background(unstained neighboring cells).

    Acknowledgments

    This work was supported by GlaxoSmithkline, Medi-search, and the Wellcome Trust.

    The authors thank Ian Forsythe and Margaret BarnesDavies for comments on the manuscript, Tim Ecclestone fromNikon Instruments for technical advice, and Mike Mulheranfor advice on power analysis.

    Literature Cited

    Adams JC (1979) A fast, reliable silver-chromate Golgi method forperfusion-fixed tissue. Stain Technol 54:225226

    Alibardi L (2003) Ultrastructural distribution of glycinergic andGABAergic neurons and axon terminals in the rat dorsal cochlearnucleus, with emphasis on granule cell areas. J Anat 203:3156

    Alibardi L (2006) Review: cytological characteristics of commissuraland tuberculo-ventral neurons in the rat dorsal cochlear nucleus.Hear Res 216217:7380

    Angulo A, Fernandez E, Merchan JA, Molina M (1996) A reliablemethod for Golgi staining of retina and brain slices. J NeurosciMethods 66:5559

    Angulo A, Merchan JA, Molina M (1994) Golgi-Colonnier method:correlation of the degree of chromium reduction and pH changewith quality of staining. J Histochem Cytochem 42:393403

    Blackstad TW, Osen KK, Mugnaini E (1984) Pyramidal neuronesof the dorsal cochlear nucleus: a Golgi and computer reconstruc-tion study in cat. Neuroscience 13:827854

    Brawer JR, Morest DK, Kane EC (1974) The neuronal architectureof the cochlear nucleus of the cat. J Comp Neurol 155:251300

    Browner RH, Baruch A (1982) The cytoarchitecture of the dorsalcochlear nucleus in the 3-month- and 26-month-old C57BL/6mouse: a Golgi impregnation study. J Comp Neurol 211:115138

    Colonnier M (1964) The tangential organization of the visual cortex.

    J Anat 98:327344Donga R, Lund JP (1991) Discharge patterns of trigeminal commis-

    sural last-order interneurons during fictive mastication in therabbit. J Neurophysiol 66:15641578

    Friedland DR, Los JG, Ryugo DK (2006) A modified Golgi stain-ing protocol for use in the human brain stem and cerebellum.J Neurosci Methods 150:9095

    Geisler S, Heilmann H, Veh RW (2002) An optimized method forsimultaneous demonstration of neurons and myelinated fibertracts for delineation of individual trunco-and palliothalamic nu-clei in the mammalian brain. Histochem Cell Biol 117:6979

    Gittins R, Harrison PJ (2004) Neuronal density, size and shape in thehuman anterior cingulate cortex: a comparison of Nissl and NeuNstaining. Brain Res Bull 63:155160

    A Method Combining Golgi and Nissl Staining 549

  • 8/7/2019 anatomic staining_2008

    12/12

    Grandin T, Demotte OD, Greenough WT, Curtis SE (1988) Perfusionmethod for preparing pig brain cortex for GolgiCox impregna-tion. Stain Technol 63:177181

    Grinevich V, Brecht M, Osten P (2005) Monosynaptic pathwayfrom rat vibrissa motor cortex to facial motor neurons revealedby lentivirus-based axonal tracing. J Neurosci 25:82508258

    Hackney CM, Osen KK, Kolston J (1990) Anatomy of the cochlearnuclear complex of guinea pig. Anat Embryol (Berl) 182:123149

    Hani SA, Al-Haidari MH, Saboba MM (2007) Neuronal types in thehuman anterior ventral thalamic nucleus: a Golgi study. Cell MolNeurobiol 27:745755

    King MA, Scotty N, Klein RL, Meyer EM (2002) Particle detection,number estimation, and feature measurement in gene transferstudies: optical fractionator stereology integrated with digitalimage processing and analysis. Methods 28:293299

    Kopsch F (1896) Erfahrungen uber die Verwfendung des Formalde-hyds bei der Chromsilber-Impragnation. Anat Anz 11:727729

    MaraniE, Guldemond JM,Adriolo PJ,Boon ME,Kok LP (1987) Themicrowave Rio-Hortega technique: a 24 hour method. HistochemJ 19:658664

    Maruyama A, Ohmori H (2006) Fluorescent labeling of rat auditorybrainstem circuits for synaptic and electrophysiological studies.J Neurosci Methods 152:163172

    McDonagh JC, Hornby TG, Reinking RM, Stuart DG (2002) Asso-ciations between the morphology and physiology of ventral-hornneurons in the adult turtle. J Comp Neurol 454:177191

    Mendizabal-Zubiaga JL, Reblet C, Bueno-Lopez JL (2007) The un-derside of the cerebral cortex: layer V/VI spiny inverted neurons.J Anat 211:223236

    Morest DK, Morest RR (1966) Perfusion-fixation of the brain withchro-mium-osmium solutions for the rapid Golgi method. Am JAnat 1118:811831

    Mugnaini E (1985) GABA neurons in the superficial layers of the ratdorsal cochlear nucleus: light and electron microscopic immuno-cytochemistry. J Comp Neurol 235:6181

    Mugnaini E, Osen KK, Dahl AL, Friedrich VL Jr, Korte G (1980a)Fine structure of granule cells and related interneurons (termedGolgi cells) in the cochlear nuclear complex of cat, rat and mouse.J Neurocytol 9:537570

    Mugnaini E, Warr WB, Osen KK (1980b) Distribution and lightmicroscopic features of granule cells in the cochlear nuclei of cat,

    rat, and mouse. J Comp Neurol 191:581606Oertel D, Wu SH (1989) Morphology and physiology of cells in

    slice preparations of the dorsal cochlear nucleus of mice. J CompNeurol 283:228247

    Oertel D, Wu SH, Garb MW, Dizack C (1990) Morphology andphysiology of cells in slice preparations of the posteroventralcochlear nucleus of mice. J Comp Neurol 295:136154

    Osen KK (1969) Cytoarchitecture of the cochlear nuclei in the cat.J Comp Neurol 136:453484

    Palay SL, Chan-Palay V (1974) Cerebellar Cortex. Cytology andOrganization. New York, Springer

    Pasternak JF, Woolsey TA (1975) On the selectivity of the GolgiCox method. J Comp Neurol 160:307312

    Paxinos G, Watson C (1998) The Rat Brain in Stereotaxic Coor-dinates. 2nd ed. San Diego, Academic Press

    Saada AA, Niparko JK, Ryugo DK (1996) Morphological changesin the cochlear nucleus of congenitally deaf white cats. BrainRes 736:315328

    Seldon HL, Clark GM (1991) Human cochlear nucleus: comparisonof Nissl-stained neurons from deaf and hearing patients. Brain Res551:185194

    Skoglund TS, Pascher R, Berthold CH (1996) Aspects of quantitativeanalysis of neurons in the cerebral cortex. J Neurosci Methods70:201210

    Uno A, Idoux E, Beraneck M, Vidal PP, Moore LE, Wilson VJ,Vibert N (2003) Static and dynamic membrane properties of lat-eral vestibular nucleus neurons in guinea pig brain stem slices.J Neurophysiol 90:16891703

    Van der Loos H (1956) Une combinaison de deux vieilles methodeshistologiques pour le systeme nerveux central. MonatsschrPsychiatr Neurol 132:330334

    Webster DB, Trune DR (1982) Cochlear nuclear complex of mice.Am J Anat 163:103130

    Werner L, Brauer K (1984) Neuron types in the rat dorsal lateralgeniculate nucleus identified in Nissl and deimpregnated Golgipreparations. J Hirnforsch 25:121127

    Werner L, Hedlich A, Koglin A (1986) Classification of neurons inthe visual cortex of the guinea pig (Cavia porcellus). A combinedGolgi-Nissl study using deimpregnation technics. J Hirnforsch27:213236

    Werner L, Winkelmann E, Koglin A, Neser J, Rodewohl H (1989) AGolgi deimpregnation study of neurons in the rhesus monkeyvisual cortex (areas 17 and 18). Anat Embryol (Berl) 180:583597

    Wouterlood FG, Mugnaini E (1984) Cartwheel neurons of the dorsalcochlear nucleus: a Golgi-electron microscopic study in rat. J CompNeurol 227:136157

    Zhang H, Weng SJ, Hutsler JJ (2003) Does microwaving enhancethe Golgi methods? A quantitative analysis of disparate staining

    patterns in the cerebral cortex. J Neurosci Methods 124:145155Zhang S, Oertel D (1993) Giant cells of the dorsal cochlear nu-

    cleus of mice: intracellular recording in slices. J Neurophysiol 69:13981408

    Zwaagstra B, Kernell D (1981) Sizes of soma and stem dendritesin intracellularly labelled a-motoneurones of the cat. Brain Res204:295309

    550 Pilati, Barker, Panteleimonitis, Donga, Hamann