strategies to improve the antigenicity,ultrastructure preservation and

ARTICLE IN PRESS

European Journal of Cell Biology 89 (2010) 285–297

Contents lists available at ScienceDirect

European Journal of Cell Biology

0171-93

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.de/ejcb

Strategies to improve the antigenicity, ultrastructure preservation andvisibility of trafficking compartments in Arabidopsis tissue

York-Dieter Stierhof a,n, Farid El Kasmi b

a Center for Plant Molecular Biology (ZMBP), Microscopy, University of Tubingen, Auf der Morgenstelle 5, 72076 Tubingen, Germanyb Center for Plant Molecular Biology (ZMBP), Developmental Genetics, University of Tubingen, Auf der Morgenstelle 3, 72076 Tubingen, Germany

a r t i c l e i n f o

Keywords:

Arabidopsis

Freeze-substitution

High-pressure freezing

Immunofluorescence labelling

Immunogold labelling

Quantum dot

trans-Golgi network

Tokuyasu cryosection

Trafficking

VAMP727

35/$ - see front matter & 2009 Elsevier Gmb

016/j.ejcb.2009.12.003

esponding author. Tel.: +49 7071 297 6662;

ail address: [email protected]

a b s t r a c t

Immunolabelling of (ultra)thin thawed cryosections according to Tokuyasu is one of the most reliable

and efficient immunolocalisation techniques for cells and tissues. However, chemical fixation at

ambient temperature, a prerequisite of this technique, can cause problems for samples, like plant tissue,

because cell walls, hydrophobic surfaces and intercellular air slow down diffusion of fixative molecules

into the sample. We show that a hybrid technique, based on a combination of cryofixation/freeze-

substitution and Tokuyasu cryosection immunolabelling, circumvents the disadvantages associated

with chemical fixation and results in an improved ultrastructure and antigenicity preservation of

Tokuyasu cryosections used for light and electron microscopic immunolabelling (as shown for Myc- or

mRFP-tagged proteins, KNOLLE and carbohydrate epitopes). In combination with the most sensitive

particulate marker systems, like 1-nm gold or quantum dot markers, we were able to obtain a

differentiated labelling pattern which allows a more detailed evaluation of plant Golgi, trans-Golgi

network and multivesicular body/prevacuolar compartment markers (COPI-specific gCOP, the ADP-

ribosylation factor GTPase ARF1, ARA7/RabF2b and the vacuolar sorting receptor VSR). We also discuss

possibilities to improve membrane contrast, e.g., of transport vesicles like COPI, COPII and clathrin-

coated vesicles, and of compartments of endosomal trafficking like the trans-Golgi network.

& 2009 Elsevier GmbH. All rights reserved.

Introduction

Plants pose considerable problems in many respects if oneaims at a life-like preservation of cellular fine structure, e.g., fortransmission electron microscopy. In general, plant tissues aredifficult to preserve by chemical fixation at ambient temperaturefor both ultrastructural analysis and immunolocalisation experi-ments. The cellular turgor pressure, which stabilises the livingcells and tissues, immediately collapses during chemical fixation,thereby causing the collapse or rupture of the often thincytoplasmic strands in vacuolated cells. Intercellular air spaces,cell walls and hydrophobic surfaces, like waxes (e.g., of pollengrains, anthers, ovules, leaves), drastically slow down diffusion offixatives. This increases the inherent limitations of chemicalfixation like the time consuming diffusion of fixatives into thesample, the selectivity of the different fixatives used for cross-linking, and the pH-related and osmotic changes caused by thefixation buffer, the properties of which can never be correct for allcellular compartments.

H. All rights reserved.

fax: +49 7071 29 5135.

n.de (Y.-D. Stierhof).

Immunolabelling of (ultra)thin thawed cryosections accordingto Tokuyasu is one of the most reliable and efficient immunolo-calisation techniques for cells and tissues (Griffiths, 1993;Griffiths et al., 1983; Griffith and Posthuma, 2002; Humbel andStierhof, 2008; Liou et al., 1996; Slot and Geuze, 2007; Tokuyasu,1973, 1978, 1997; Webster et al., 2008) for two main reasons.Firstly, antigens are fixed with low concentrations of aldehydesonly and remain in an aqueous environment prior to immunola-belling (provided they are not extracted) and secondly, theaccessibility of antigens at the thawed cryosection surface isbetter when compared to that of the resin section surface becauseantigens are not embedded in a crosslinked resin matrix.However, chemical fixation is a prerequisite for stabilising theultrastructure before cryoprotectant infiltration (= partial dehy-dration) and after thawing of the cryosections for immunolabel-ling. The disadvantages of chemical fixation at ambienttemperature are especially obvious when tissues, like antherscontaining pollen grains or developing seeds containing embryos,have to be processed for immunolocalisation experimentsbecause structures and antigens can be dislocated or get lostbefore proper arrest by chemical crosslinking (Ripper et al., 2008).

High-pressure freezing, the most important cryofixationtechnique for plants, followed by freeze-substitution (Humbel,2008; Otegui et al., 2001; Studer et al. 1989) circumvents the

www.elsevier.de/ejcb

dx.doi.org/10.1016/j.ejcb.2009.12.003

mailto:[email protected]

ARTICLE IN PRESS

Y.-D. Stierhof, F. El Kasmi / European Journal of Cell Biology 89 (2010) 285–297286

disadvantages of chemical fixation and dehydration at ambienttemperature, provided intercellular air (which is not compatiblewith high-pressure freezing because it can be compressed, incontrast to the liquid cytoplasm) can be removed and the samplesize is compatible with the inherent physical restrictions of thiscryofixation technique, as visible ice crystal damage can beavoided only up to specimen thickness of 200 mm.

It has been shown for resin section immunolabelling thatcryofixation/freeze-substitution can preserve sensitive antigensand structures better than chemical fixation and dehydration atambient temperature (Hess, 2007; Humbel and Schwarz, 1989).However, there are also antigens that cannot be immunolabelledeither in thawed cryosections or after cryofixation, freeze-substitution and resin embedding. Those antigens may besensitive to chemical fixation at ambient temperature or solventsand resin components or they cannot be detected after cryofixa-tion and resin embedding due to the limited number of epitopesaccessible at the resin section surface. Therefore, a hybridtechnique was developed for both, difficult-to-fix antigens anddifficult-to-fix specimens by combining the Tokuyasu thawedcryosection labelling method with an initial cryofixation step inorder to benefit from the advantages of both methods (Ripperet al., 2008; Stierhof et al., 2008). After cryofixation, e.g., by high-pressure freezing, the entire sample is dehydrated and simulta-neously fixed during freeze-substitution. In contrast to theconventional procedure, freeze-substitution allows the applica-tion of different fixatives and fixative combinations. In addition toformaldehyde (FA), glutaraldehyde (GA) and acrolein, uranylacetate (UA) and osmium tetroxide (OsO4) can be used (Stierhofet al., 2008; van Donselaar et al., 2007). Then, after raising thetemperature to 0 1C, the sample has to be rehydrated andpostfixed with aldehydes, before it is further processed forcryosection labelling, namely infiltrated with a cryoprotectantfollowed by conventional freezing and cryosectioning.

Strategies to improve labelling density in order to localisesparsely distributed antigens are of attracting interest in im-munohistochemistry. We will compare the most sensitiveparticulate EM markers available like 1-nm gold and quantumdot (Qdots) markers and will re-evaluate some controversiallydiscussed plant endosomal markers in the light of improvedprocessing and labelling techniques.

A specific problem of high-pressure frozen and freeze-substituted plant samples is the poor visibility of the bilayerstructure of many membranes especially in the case of well frozensamples embedded in resin or processed for cryosection im-munolabelling (for resin sections, see (Donohoe et al., 2006; Hess,2003, 2007; Murata et al., 2002); for thawed cryosections, see(Ripper et al., 2008; van Donselaar et al., 2007)). To improve thevisibility of the membrane bilayer, different freeze-substitutionprotocols were evaluated in respect to the ultrastructuralappearance of Golgi stacks, trans-Golgi network (TGN) andendosomes.

Materials and methods

Specimens used in this study

Ecotype Col-0 (wild type) and mutant Arabidopsis thaliana

seedlings were grown on Murashige and Skoog (MS) medium +1%sucrose for 3-5 days. GFP-KNOLLE- and ARA7/RabF2b-GFP-expressing lines were provided by A. Volker, H. Wolters and G.Jurgens (Developmental Genetics, ZMBP, University of Tubingen)(Reichardt et al., 2007). VHA-a1-GFP- and VHA-a1-mRFP-expres-sing lines have been described (Dettmer et al., 2006; von derFecht-Bartenbach et al., 2007). The coding sequence of VAMP727

was cloned into the KNOLLE expression cassette (Muller et al.,2003) in order to tag it with a 1� Myc tag at the N-terminus andto express it in mitotic cells. Incubation of seedlings withbrefeldin A (BFA; 50 mM, 90 min; Invitrogen, Karlsruhe, Germany)was done in cell culture dishes in 1 ml basal medium (Dettmeret al., 2006).

Antibodies and markers used

Primary antibodies

Rabbit anti-mRFP (1:50, anti-DsRed, Clontech, USA), rabbitanti-GFP IgG (1:500; #TP401, Torrey Pines Biolab Inc., EastOrange, USA), rabbit anti-KNOLLE serum (1:1500; (Lauber et al.,1997)), mouse anti-Myc monoclonal IgG 9E10 (1:500; Santa CruzBiotechnology); mouse IgG CCRC-M1 (1:5; (Zhang and Staehelin,1992); Carbosource Services, University of Georgia, USA), rabbitanti-gCOP serum (1:500; (Movafeghi et al., 1999)), rabbit anti-ARF1 IgG (1:500; (Pimpl et al., 2000)); rabbit anti-clathrin IgG(1:100; antibody against plant clathrin heavy chain peptide;(Kim et al., 2001; Dhonukshe et al., 2007)); rabbit anti-VSR IgG(1:50; (Tse et al., 2004)).

Markers

Goat anti-rabbit F(ab’)2 coupled to Nanogold (1:60; Nano-probes, Stony Brook, USA); goat anti-mouse F(ab’)2 coupled toNanogold (1:60; Nanoprobes), goat anti-mouse IgG coupled toultrasmall colloidal gold (1:30; Aurion; Wageningen, The Nether-lands), goat anti-mouse IgG-12 nm gold (1:30; Dianova, Hamburg,Germany), protein A-6 nm gold (1:200; (Slot and Geuze, 1985)),goat-anti-rabbit F(ab’)2-Qdot655 (1:10: Molecular Probes/Invitrogen, Karlsruhe, Germany), goat anti-rabbit F(ab’)2-Qdot525(1:10; Molecular Probes); goat anti-rabbit IgG coupled to Cy3(1:800; Dianova).

Cryofixation /freeze-substitution for Tokuyasu croysection labelling

Root tips and degassed cotyledons and anthers were high-pressure frozen (HPM 010; Bal-Tec, Balzers, Lichtenstein) inaluminium planchettes filled with 1-hexadecene (Merck Sharpand Dohme) (Studer et al., 1989). Samples were freed from1-hexadecene (below �100 1C) and transferred into 2-ml cryo-tubes placed in a Leica FS unit (at �90 1C) filled with acetonecontaining 2% water, 0.075-0.1% OsO4 (EMS, Hatfield, PA, USA),0.5% GA (from a 10% stock solution in acetone; catalogue no.16530; EMS, Fort Washington, PA, USA), and 0.1–0.2% UA (AgarScientific, Stansted, Essex, UK), as well as 0.5–2.5% methanol(deriving from UA stock solutions (20% UA in methanol); (Haweset al., 2007)). After 50–70 h at �90 1C, samples were kept at�60 1C for 8 h, and then further warmed to �35 1C. After 8–10 hat �35 1C, samples were washed five times (30–45 min each)with acetone containing 2% water and 0.5% GA. Thereaftersamples were brought to �20 1C (for 10 min), and then furtherwarmed to 0 1C. Between �20 1C and 0 1C, acetone/GA wasreplaced in two steps (50% acetone, 0.38% GA, �20 1C; 10%acetone, 0.25% GA, 0 1C, 10 min each) by water containing 0.25%GA. Root tips were kept for another 30–45 min in water with0.25% GA, Arabidopsis anthers and cotyledons for 60-90 min at0 1C. Then samples were washed once with water and twice withwater containing 50 mM glycine (to inactivate residual reactivealdehyde groups) and further processed for conventional cryo-sectioning according to (Tokuyasu, 1997).

In some cases the protocol was changed. Samples were fixedwith 0.5% GA and 0.5% UA without OsO4 in ethanol or acetone andpostfixation was omitted.

ARTICLE IN PRESS

Y.-D. Stierhof, F. El Kasmi / European Journal of Cell Biology 89 (2010) 285–297 287

Samples were infiltrated step by step with sucrose/PVP(1.8 M sucrose (Merck)/20% PVP (Sigma, PVP10-100G);(Tokuyasu, 1989)). After infiltration, specimens were mountedonto stubs, frozen in liquid nitrogen, transferred into thecryochamber (Leica EMFCS) of a Leica Ultracut UCT cryoultrami-crotome and sectioned at �80 1C for light microscopy(300–500 nm) or �110 1C to �115 1C for electron microscopy(70–100 nm) with a diamond knife (Diatome, Biel, Switzerland).Cryo-sections were picked up with a drop consisting of equalparts of sucrose/PVP and 2.1 M sucrose or a mixture of 2.1 Msucrose and 1.8% methyl cellulose (Liou et al., 1996) andtransferred onto coverslips or pioloform- and carbon-coated gridsfor subsequent immunofluorescence or immunogold labelling,respectively.

Samples processed for resin embedding were freeze-substi-tuted in acetone or ethanol containing 0.5% GA and 0.3-0.5% UA.At �35 1C, samples were washed and infiltrated in several stepswith Lowicryl HM20 before polymerisation with UV light.

Immunolabelling and embedding of cryofixed samples

Thawed 70–100 nm cryo-sections or Lowicryl HM20 sectionstransferred to coated grids were blocked with 1% milkpowder/0.5% bovine serum albumin (BSA) in phosphate-buffered saline(PBS) for 20 min. After blocking, sections were incubated withantibodies diluted in blocking buffer (see above) for 1 h, andwashed 5 times with blocking buffer (5 min each). Boundantibodies were detected with gold, Qdot, or fluorescence markers(see above). After washing (twice with blocking buffer, threetimes PBS; 5 min each) the antigen–antibody–marker complexeswere stabilised with 0.5% GA in PBS for 5 min and washed sixtimes with water (total 25 min).

Gold markers were silver enhanced with HQ Silver(Nanoprobes) for approximately 8�8.5 min (Nanogold) or withR-Gent (Aurion) for 45 min (ultrasmall gold) or 30 min (PAG-6).Thereafter, sections were washed six times (4 min each) inwater. Thawed cryosections were stained with 1% aqueousUA for 10 min, and embedded in 1.8% methyl cellulose (catalogueno. M-6385; Sigma-Aldrich) containing 0.3–0.45% UA. Resinsections were stained with 1% aqueous UA for 5 min and leadcitrate for 30 s.

For immunofluorescence labelling, semi-thick (300�400 nm)cryosections mounted onto coverslips were labelled as describedabove. Fluorochrome-conjugated markers (see above) were usedinstead of gold markers. Sections were counterstained with DAPI(4�,6-diamidino-2-phenylindole) (1 mg/ml; Sigma) and embeddedin Mowiol 4.88 (Calbiochem) containing the anti-fading reagentDABCO (25 mg/ml; Sigma).

Immunofluorescence-labelled sections were viewed with aZeiss Axioplan objective (�63 magnification; NA (numericalaperture) 1.4). Filters used: GFP, band pass 470/40, dichroicmirror 495 LP, band pass 525/50; Cy3, band pass 545/25, dichroicmirror 565 LP, band pass 605/70; mRFP, band pass 565/30,dichroic mirror 585, band pass 620/60; DAPI, band pass 377/50,dichroic mirror BS 409, band pass 447/60) (Zeiss, Gottingen,Germany; AHF Analysentechnik, Tubingen, Germany). Ultrathincryo- and resin sections were viewed in a LEO 906 transmissionelectron microscope at an accelerating voltage of 80 kV.

Immunolabelling of chemically fixed samples

Root tips were transferred into MTSB (microtubule stabilisingbuffer: 50 mM Pipes, 5 mM EGTA and 5 mM MgSO4, pH 7),containing 4% formaldehyde (Merck Sharp and Dohme) and fixedfor 30-60 min. Thereafter, they were additionally fixed with 8%

formaldehyde in 0.1 M Pipes (Merck Sharp and Dohme) (pH 7.1)for 1�2 h. After washing with PBS, samples were embedded in10% gelatine in PBS and infiltrated with sucrose/PVP (see above).

Cryofixation/freeze-substitution for ultrastructural analysis

For ultrastructural analysis, root tips and degassed antherswere high-pressure frozen as described above and freeze-substituted in acetone containing 2% OsO4 (50�70 h at �90 1C,8 h at �60 1C, 8 h at �35 1C). To improve contrast, samples werekept for 1�2 h at 0 1C before they were washed with acetone andembedded in epoxy resin (Roth, Karlsruhe, Germany). In somecases, 0.1% gallic acid (Merck), 0.5% UA or 2% water was added tothe freeze-substitution medium. Ultrathin sections were stainedwith 1�2% UA in 50% ethanol (Epstein and Holt, 1963) for10�20 min and lead citrate (3�5 min) (Venable and Coggeshall,1965).

Results and discussion

Tokuyasu cryosection labelling after cryofixation, freeze-substitution

and rehydration

The ultrastructure and antigenicity of plant tissues which areexceptionally difficult to preserve for Tokuyasu cryosectionlabelling by conventional chemical fixation at ambient tempera-ture, were better preserved when the hybrid technique (seeMaterials and methods) was used. Anthers containing pollengrains and developing seeds with embryos were well preserved,cells and nuclei showed no shrinkage artefacts, and extractionwas reduced (Ripper et al., 2008). Cotyledons also showed a wellpreserved ultrastructure and strong labelling of the GFP-taggedsubunit a1 of the vacuolar H+-ATPase (VHA-a1-GFP) at the TGN(Fig. 1A-C) similar to root tip labelling (Dettmer et al., 2006;Reichardt et al., 2007). In the case of VHA-a1-GFP- or VHA-a1-mRFP-expressing pollen grains inside anthers, some, but notall the Golgi stacks were associated with a TGN which waslabelled (Fig. 1D, E). In addition, vesicle clusters, often separatedfrom Golgi stacks, were decorated with gold markers (Fig. 1E, F;for comparison, a corresponding epoxy resin-embedded vesiclecluster of pollen cells is shown in Fig. 1G).

Remarkably, difficult-to-fix antigens/epitopes were also betterpreserved. Myc (myc-VAMP727, an R-SNARE; (Ebine et al., 2008))as well as mRFP tags (VHA-a1-mRFP) can be detected even afterGA and OsO4 fixation during freeze-substitution and postfixationwith GA (Figs. 1F, 2A-E). If GA appears to be deleterious toantigens, FA instead of GA can be used for postfixation(not shown). Myc and mRFP tags are very sensitive, even to theweak fixative FA in the case of conventional Tokuyasu cryosectionlabelling, and cannot be labelled on resin sections even aftercryofixation and freeze-substitution (not shown). This is also truefor the cytokinesis-specific plant syntaxin KNOLLE ((Lauber et al.,1997; Reichardt et al., 2007; Ripper et al., 2008); see below).Myc-tagged VAMP727 localised mainly to multivesicular bodies(MVBs), autophagosome-like structures and larger vesicles, withsimilar labelling pattern and density with and without OsO4

fixation (Fig. 2A-C). Although located mainly in compartmentsdestined for fusion with the vacuolar membrane and exclusivelysynthesised in large amounts in mitotic cells (because expressedunder the KNOLLE promoter), myc-VAMP727 was present in arelatively large number of root tip cells, suggesting that theprotein is not only very stable, but might also be efficientlyrecycled back to MVBs and autophagosomes. mRFP-taggedVHA-a1 was found on the TGN in root tips (Fig. 2D, E) and

ARTICLE IN PRESS

Fig. 1. Ultrathin cryosection labelling after high-pressure freezing, freeze-substitution (0.1% OsO4, 0.3% UA, 0.5% GA) and rehydration (0.25% GA). (A�C) Immunogold

labelling of a cotyledon of A. thaliana. (A) light microscopic image of a 300-nm cross-section through a cotyledon. (B) Ultrathin cryosection cut from the identical block.

(C) VHA-a1-GFP labelling (rabbit anti-GFP, silver-enhanced Nanogold) at the TGN. (D�G) VHA-a1-GFP (D, E) or mRFP labelling (F) (rabbit anti-GFP or rabbit anti-mRFP,

respectively; silver-enhanced Nanogold) in pollen cells. (D) TGN labelling, (E, F) vesicle cluster labelling, (G) corresponding vesicle cluster after high-pressure freezing,

freeze-substitution (2.5% OsO4) and epon embedding. In addition, a COPI-like vesicle can be seen (arrowhead). e, epidermis; er, ER; g, Golgi stack; mt, microtubule; t, TGN.

Bars: 10 mm (A), 5 mm (B), 250 nm (C�G).


showed a labelling pattern comparable to VHA-a1-GFP in pollengrains (Fig. 1F). As an example for carbohydrate labelling, we usedthe xyloglucan-specific monoclonal antibody (mAb) CCRC-M1(Zhang and Staehelin, 1992). Carbohydrate antigens often cannotbe properly fixed by aldehydes and, therefore, run the risk ofbeing extracted in conventionally processed thawed cryosections.Incubation with mAb CCRC-M1 resulted in strong labelling of cellwalls and secretory vesicles at the rims of trans-Golgi cisternae, atthe TGN and close to the plasma membrane (Fig. 2F, G).

We also tested a number of Golgi, TGN and MVB/prevacuolarcompartment (PVC)-specific proteins, partly with controversiallydiscussed labelling patterns, in the root tip cells of high-pressurefrozen and freeze-substituted samples that were processed forTokuyasu cryosection labelling. g-COP (g-coatamer protein, acomponent of the COPI (coatamer protein I) coat) was found atthe cis-face of the Golgi stack (as shown by Pimpl et al. (2000)),but mainly at the rims of Golgi cisternae, and even at the TGN(Fig. 2H, I) (for comparison, see also (Staehelin and Kang, 2008)).MVBs/PVCs, e.g., located close to the TGN were never labelled(Fig. 2I). Immunogold labelling of the ADP-ribosylation factor

Fig. 2. Ultrathin cryosection labelling (A. thaliana root tip cells) using silver-enhanced

Myc-tagged VAMP727 detected on the membranes of small vacuoles, MVBs, and autoph

(B, C) freeze-substitution with 0.1% OsO4, 0.3% UA, and 0.5% GA, in all three cases rehydr

detected in the TGN (rabbit anti-mRFP IgG). (D) Freeze-substitution with OsO4, GA, UA

rehydration without postfixation. (F, G) Carbohydrate labelling (mouse IgG CCRC-M1)

membrane in secretory vesicles, and in the cell wall (G). (H, I) gCOP labelling (rabbit an

labelled. (K, L) ARF1 labelling (rabbit anti-ARF1) on TGN and Golgi stack. (K) Freeze-subs

(L) Freeze-substitution with 0.5% UA, 0.5% GA, rehydration without postfixation. (M�O)

on clathrin-coated vesicles. (M) freeze-substitution with OsO4, GA, UA, rehydration in

without postfixation, (O) gold/silver clusters represent clathrin-coated buds or ves

mitochondrion; mvb, multivesicular body; pm, plasma membrane; sv, secretory vesicl

(ARF1) GTPase (originally described as cis-Golgi marker (Pimplet al., 2000) or Golgi stack marker (Robinson et al., 2007)) resultedin strong TGN labelling and weaker labelling over Golgi cisternae(Fig. 2K, L). Clathrin heavy chain, a component of the clathrin coat,could be localised to vesicular structures in the TGN and to theplasma membrane (Fig. 2M-O) (Dhonukshe et al., 2007). Anti-bodies against the GFP-tagged Rab GTPase ARA7/RabF2b-GFP(Fig. 3A, B) or against the vacuolar sorting receptor VSR (Fig. 3C)which were shown to bind to the MVB/PVC membrane (Haaset al., 2007; Hinz et al., 2007; Reichardt et al., 2007; Tse et al.,2004) and additionally to the trans-Golgi (VSR; (Hinz et al., 2007))also often labelled the TGN in thawed cryosections (especially inthe case of VSR labelling) (Fig. 3A�C). The occasionally observedpresence of MVBs in the close vicinity of the TGN furthercomplicates the interpretation of confocal colabelling studies(Figs. 2I and 3D). In the light of the presence of many TGN andtransport vesicle-specific markers at the TGN and of the recentlyobserved high TGN mobility (Crowell et al., 2009), the plant TGNseems to be a highly complex and mobile sorting station for bothendocytosed and recycled material, as well as newly synthesised

Nanogold after high-pressure freezing, freeze-substitution and rehydration. (A�C)

agosomes (mouse anti-myc IgG). (A) Freeze-substitution with 0.5% UA and 0.5% GA,

ation and postfixation were performed with 0.25% GA. (D, E) mRFP-tagged VHA-a1

, rehydration in the presence of GA, (E) freeze-substitution with OsO4, GA and UA,

of small vesicles at the TGN, at the rims of Golgi stacks (F, G), close to the plasma

ti-gCOP) at the rims of the Golgi stack, the cis-Golgi and the TGN. The MVB is not

titution with 0.1% OsO4, 0.3% UA, 0.5% GA, rehydration in the presence of 0.25% GA.

Clathrin labelling (rabbit anti-heavy chain) at the TGN, the plasma membrane and

the presence of GA, (N, O) freeze-substitution with OsO4, GA and UA, rehydration

icles (arrow heads point to some clusters). cw, cell wall; g, Golgi stack; m,

e; t, TGN; v, small vacuole. Bars: 250 nm (A�D; E�M; N).

ARTICLE IN PRESS


ARTICLE IN PRESS

Fig. 3. MVB marker and KNOLLE labelling (A. thaliana root tip cells) of ultrathin cryo- and resin section labelling using silver-enhanced Nanogold. (A, B) ARA7/RabF2b-GFP

labelling (rabbit anti-GFP; ultrathin cryosection after high-pressure freezing, freeze-substitution (0.1% OsO4, 0.3% UA, 0.5% GA) and rehydration (0.25% GA) on MVB

membrane (A) and TGN (B). (C) VSR labelling (rabbit anti-VSR; ultrathin cryosection after high-pressure freezing, freeze-substitution (0.1% OsO4, 0.3% UA, and 0.5% GA) and

rehydration with 0.25% GA) on MVB membrane and TGN. (D) MVB associated with TGN (high-pressure freezing, freeze-substitution and epoxy resin embedding). (E)

KNOLLE labelling with rabbit anti-KNOLLE serum (chemically fixed (4% FA, 45 min) ultrathin cryosection) on growing cell plate in wild-type cells. (F) KNOLLE labelling with

rabbit anti-KNOLLE serum (ultrathin Lowicryl-HM20 section, after high-pressure freezing, freeze-substitution (0.5% UA, 0.5% GA) and low-temperature resin embedding) in

wild-type cells. Note: no labelling. (G) KNOLLE labelling with rabbit anti-KNOLLE serum (ultrathin cryosection after high-pressure freezing, freeze-substitution (0.1% OsO4,

0.3% UA, 0.5% GA) and rehydration (0.25% GA)) on growing cell plate in wild-type cells. (H) KNOLLE labelling with rabbit anti-KNOLLE serum (ultrathin Lowicryl-HM20

section, after high-pressure freezing, freeze-substitution (0.5% UA, 0.5% GA) and low-temperature resin embedding) on growing cell plate in cells expressing a GFP-tagged

KNOLLE protein. cp, cell plate; cw, cell wall; er, ER; g, Golgi stack; m, mitochondrion; mvb, multivesicular body; t, TGN. Bars: 250 nm (A�D), 500 nm, (E�H).


material which is destined for the vacuole or for the plasmamembrane (or cell plate) (Dettmer et al., 2006; Lam et al., 2007b;Otegui and Spitzer, 2008; Reichardt et al., 2007; Richter et al.,2009; Robinson et al., 2007, 2008; Staehelin and Kang, 2008). Itwould be interesting to study possible subdomains in more detail,using markers with partly overlapping labelling patterns, likeVHA-a1, RabA2/A3, RabA4b, SCAMP1 and SCAMP2 (Chow et al.,2008; Lam et al., 2007a; Staehelin and Kang, 2008; Toyooka et al.,2009).

Immunogold labelling experiments on the cytokinesis-specificplant syntaxin KNOLLE are difficult to interpret. Thawed cryosec-tion labelling of chemically fixed root tips with KNOLLE-specific

Fig. 4. Thin (300-400 nm) cryosection labelling (A. thaliana root tip cells) using Cy3-con

rabbit anti-KNOLLE serum (after high-pressure freezing, freeze-substitution (0.5% UA, 0

arrows to completed cell plates. Fluorescent spots representing TGN and MVBs can be d

differential interference contrast image. (C, D) Clathrin (C) and ARF1 labelling (D) (chem

ARF1). The Cy3 fluorescence is imaged together with the GFP fluorescence of VHA-a1

compartments showing Cy3 and GFP colocalisation. The DNA is labelled with DAPI (b

showing red mRFP-fluorescence of VHA-a1-mRFP. The DNA is labelled with DAPI (blue)

compartment of a root tip treated with BFA. (F) ARF1 immunolabelling (Cy3 fluorescen

(0.1% OsO4, 0.5% GA, 0.2% UA, 2% H2O in acetone) and rehydration) expressing an YFP-t

The BFA compartments (red) are surrounded by Golgi stacks (green). Golgi stacks are of

Note the well preserved nuclei in (A, E, F). e, epidermis; n, nucleus. Bars: 10 mm.

rabbit antibodies (Lauber et al., 1997) was only successful aftershort FA fixation (4% FA, 30–45 min; Fig. 3E). In contrast,cryosections from samples after high-pressure freezing, freeze-substitution, and rehydration allowed KNOLLE labelling even afterOsO4, UA, and GA fixation during freeze-substitution and post-fixation with GA at 0 1C (Fig. 3G). We were not able to label thisantigen with KNOLLE-specific rabbit antibodies in root tips afterhigh-pressure freezing, freeze-substitution and Lowicryl (HM20or K11M) embedding (Fig. 3F). This was not a copy numberproblem because the growing cell plate contains huge amounts ofthis syntaxin (see Figs. 3G and 4A). Surprisingly, GFP-taggedKNOLLE fusion proteins (GFP-KNOLLE; (Reichardt et al., 2007))

jugated markers and GFP-, YFP- or mRFP-fluorescence. (A, B) KNOLLE labelling with

.5% GA) and rehydration) of mitotic cells. Arrowhead points to a growing cell plate,

etected. Nuclear and mitochondrial DNA are stained with DAPI. (B) Corresponding

ically fixed (4% FA followed by 8% FA) samples; rabbit anti-clathrin or rabbit anti-

-GFP. The CY3 images are slightly shifted to the left to facilitate identification of

lue). (E) Cryosection from a high-pressure frozen and freeze-substituted root tip

. Inset (two-fold enlarged) shows the mRFP fluorescence of VHA-a1-mRFP in a BFA

ce, red) in a BFA-treated seedling (after high-pressure freezing, freeze-substitution

agged version of N-ST (YFP fluorescence, green) which localises to the trans-Golgi.

ten curved in BFA-treated cells (arrowheads). The DNA is labelled with DAPI (blue).

ARTICLE IN PRESS


ARTICLE IN PRESS


could be labelled with KNOLLE-specific antibodies even in Low-icryl sections after high-pressure freezing and freeze-substitution(Fig. 3H). One reason could be that the fused GFP somehowstabilises the epitopes detected by the KNOLLE antiserum, whichwas raised against recombinant protein lacking the hydrophobicCOOH terminus (Lauber et al., 1997). Interestingly, the degrada-tion of the GFP-tagged syntaxin seemed to be decelerated incomparison to wild-type KNOLLE (A. Volker, I. Reichardt,G. Jurgens, Y.-D. Stierhof, unpublished observations). This demon-strates an additional problem of the use of tagged proteins. Inaddition to overexpression, mislocalisation and functional im-pairment, tags may also change the dynamic behaviour of thetagged protein, e.g., by interference with a fast degradationprocess.

It turned out that the visibility of the membrane bilayer wasgreatly enhanced when the freeze-substitution cocktail wassupplemented with low amounts (2–4%) of water (Reichardtet al., 2007; Ripper et al., 2008; van Donselaar et al., 2007),although the sample was completely rehydrated upon completionof the freeze-substitution process. As too much water in thesubstitution medium could cause a delay in the substitutionprocess, which could result in ice crystal damage (for furtherdetails see below), water was added later during freeze-substitu-tion, 48 h after starting freeze-substitution at –90 1C, but 24 hbefore raising the temperature (at that time, freeze-substitutionshould be completed). In this case, the bilayer visibility wasimproved, but did not reach the quality of the samples incubatedwith water during the whole freeze-substitution process. Theaddition of 0.1% gallic acid or tannic acid (which have been shownto improve membrane contrast of samples embedded in epoxyresin(Giddings, 2003) at –90 1C for two days, before exchanging gallicor tannic acid for OsO4, did not improve the bilayer visibility incryosections (not shown). It should be noted that OsO4 concen-tration may be critical because some of the antigens were moresparsely labelled when OsO4 was added to the freeze-substitutionmedium or when the OsO4 concentration was raised to 0.5–1%(Ripper et al., 2008).

Importantly, this hybrid technique also allows the reliablecryosection immunolabelling of adult nematodes and Drosophila

embryos (Ripper et al., 2008). These organisms are protected by avery resistant cuticle which makes chemical fixation at ambienttemperature nearly useless without microwave treatment oropening the cuticle of these organisms. Also, in this case, theaddition of water to the freeze-substitution medium as well as thepostfixation step, is necessary to obtain an improved ultrastruc-ture preservation (Ripper et al., 2008; Stierhof et al., 2008).

It should be stressed that immunofluorescence microscopy canalso benefit from improved preparation techniques. The qualityand reliability of immunofluorescence labelling depends not onlyon the physical resolution of the microscope and objective used,but also on the ultrastructure preservation, which is sometimesoverlooked in light microscopy. Using thin (300–400 nm)cryosections from cryofixed samples offers several advantages:(i) The ultrastructure preservation is optimised (cryofixed sam-ple), (ii) fluorescence-tagged markers, the most sensitive markersavailable, can be used (if immunolabelling experiments usingfluorescence markers fail to produce any signal, it is not worthperforming the more elaborate immunogold labelling for TEM, asgold markers are usually less sensitive (Hermann et al., 1991;Stierhof et al., 1995)), (iii) thin sections make high lateral and axialresolution possible (at least in z direction even better than in aconfocal microscope), without blurred signals from regions not infocus, and (iv) large areas can be inspected to obtain an overviewof the overall labelling pattern and to judge background ((Humbeland Stierhof, 2008; Mori et al., 2006; Takizawa and Robinson,

2003); see also (Schwarz and Humbel, 2007) for the usefulness ofultrathin resin section labelling for correlative light and electronmicroscopic studies). Fig. 4A and B show immunofluorescenceand differential interference contrast (DIC) images of a 400-nmcryosection of a root tip from a wild-type seedling, whichillustrate the potential of this labelling technique. The antibodyagainst the cytokinesis-specific KNOLLE syntaxin labelled cellplates, TGN and MVBs of cells in different stages of mitosis.Interestingly, thawed cryosections (after chemical fixation or aftercryofixation, freeze-substitution and rehydration) allow thesimultaneous detection of GFP, YFP, or mRFP fluorescencetogether with labelled antigens. Figs. 4C and 4D show 300-nmcryosections from chemically fixed samples, which were labelledfor clathrin and ARF1, respectively, with a Cy3-coupled secondaryantibody. In addition, the GFP fluorescence of the TGN markerVHA-a1-GFP, and DAPI fluorescence were imaged. The Cy3 imageswere slightly shifted to the left side in order to facilitateidentification of the co-labelling patterns. Fig. 4E and F showcryosections from high-pressure frozen, freeze-substituted andrehydrated root tips. Cells with YFP fluorescence from N-a-2,6-sialyltranferase fused to YFP (N-ST-YFP; (Grebe et al., 2003)) wereimaged together with fluorescence-labelled anti-ARF1 antibodies(Fig. 4F). The roots that were treated with the fungal toxin BFA(Fig. 4F), which causes the reversible aggregation of TGN andendosomal vesicles, revealed the typical BFA compartments,which are composed of large vesicle aggregates. In BFA-treatedcells the Golgi stack often adopted a curved structure which couldbe visualised using the N-ST-YFP fluorescence, even in lightmicroscopic images (Fig. 4F). In addition, Golgi stacks oftensurrounded the BFA compartments which were labelled withanti-ARF1 antibodies. The fluorescence images also showed thatARF1 (Fig. 4F) and VHA-a1-mRFP (Fig. 4E, inset) in the BFAcompartments were not evenly distributed. In Fig. 4E the mRFPfluorescence of VHA-a1-mRFP is imaged together with DAPI-stained DNA. An interesting aspect is the fact that in thin sections,GFP, YFP and mRFP fluorescence from the inner parts of the rootare detected without the inherent problems of confocalmicroscopy, like the scattering of the incident beam as well asof the fluorescence signal in a thick sample, which resultsin reduced signal intensity and resolution. In this way wenoticed that, in root tips, many TGN-associated antigens weremost strongly expressed in epidermal (and root cap) cells, andthat their expression decreased towards the central parts of theroot tip.

Improved labelling density obtained with 1-nm gold markers,

quantum dot markers and weak fixation

We often prefer 1-nm gold markers over larger gold. Itsenhanced sensitivity, in combination with the sensitive Tokuyasuthawed cryosection labelling technique, allows the detection ofsparse antigens or compartments harbouring only minuteamounts of antigens (Baschong and Stierhof, 1998; Stierhof,2008). KNOLLE labelling of internal vesicles of MVBs (Fig. 5A, B)and carbohydrate labelling in small secretory vesicles and in thecell wall (Fig. 5C, D) nicely demonstrate the improvement inlabelling density. This effect is especially strong when not only thesection surface is labelled (Fig. 5C, D), but also internal structuresof loosely packed vesicles (Fig. 5A, B).

We also compared 1-nm gold marker (Nanogold or 1-nmcolloidal gold; (Baschong and Stierhof, 1998; Hainfeld and Furuya,1992; Leunissen and Van de Plas, 1993; Stierhof, 2008)) labellingdensity with quantum dot (Qdot) marker labelling density(a marker useful for correlative light and electron microscopy;(Deerinck, 2008; Giepmans et al., 2005)). Qdot markers, smallsemiconductor nanocrystals, show a size-dependent fluorescence

ARTICLE IN PRESS

Fig. 5. Use of 1-nm gold and Qdot markers for ultrathin cryosection labelling after high-pressure freezing, freeze-substitution and rehydration (A�D) or conventional

chemical fixation (E�H). (A, B) Labelling of KNOLLE-GFP-containing vesicles inside MVBs (rabbit anti-GFP) using silver-enhanced 6-nm gold-protein A (A) or Nanogold as

markers (B). (C, D) Labelling of carbohydrates at the TGN and at the rims of Golgi stacks (mouse IgG CCRC-M1) using 12-nm gold coupled to IgG (C) or silver-enhanced

Nanogold (D). (E, F) Labelling of GFP-tagged VHA-a1 (rabbit anti-GFP) on the TGN using silver-enhanced Nanogold (E) or non-enhanced Qdot 655 (F) as markers.

Arrowheads point to some of the marker clusters. (G, H) Labelling of YFP-tagged N-ST (rabbit anti-GFP) on the trans-Golgi using silver-enhanced ultrasmall gold (G) or

silver-enhanced Qdot 525 (H). g, Golgi stack; mvb, multivesicular body; pm, plasma membrane; t, TGN. Bars: 250 nm (A�D; E, F; G, H).


colour, are largely resistant to bleaching and are electron dense.Large Qdot markers, like the red fluorescent Qdot 655-F(ab’)2

(core diameter �10–15 nm, conjugate diameter �15–25 nm;(Stierhof, 2008)), could be directly identified in cryosections, e.g.,after VHA-a1-GFP labelling (Fig. 5F, compared to Nanogoldlabelling, Fig. 5E). However, the contrast of Qdot 655 wasrelatively weak and its visibility was further reduced afterembedding the labelled cryosections in heavy metal (UA)-containing methyl cellulose. In contrast to the large Qdot 656,the smallest commercial Qdot conjugates presently available(the green fluorecent Qdots 525- and 565-F(ab’)2 (core diameter�3-5 nm, conjugate diameter �13 nm)) could only be detectedafter silver enhancement. Silver enhancement cannot only beapplied to colloidal gold and Nanogold but also to Qdots (Stierhof,2008). Fig. 5H shows a root cryosection labelled for N-ST-YFP withsilver-enhanced Qdot 525-F(ab’)2. For comparison, a similarsection labelled with silver-enhanced 1-nm colloidal gold isshown (Fig. 5G). It turned out that the labelling density wascomparable to that of 1-nm gold markers (compare Figs. 5E to 5Fand Figs. 5G to 5H). The small Qdots could only be silver-enhanced by very few silver enhancers like HQ Silver or theDanscher solution (Danscher, 1981; Stierhof, 2008). If necessary,the silver layer can be stabilised by treatment with gold chloride(Pohl and Stierhof, 1998; Stierhof, 2008).

We also got a remarkably enhanced labelling density whenpostfixation was omitted (and sometimes also, when OsO4

concentration was reduced). However, a higher labelling densitywas often achieved at the expense of ultrastructural preservationand appearance. Examples are mRFP-labelling (VHA-a1-mRFP,compare Figs. 2E to 2D), ARF1 labelling (compare Figs. 2L to 2K)and clathrin labelling (compare Fig. 2N, O to Fig. 2M). Thelabelling patterns indicated that these proteins were of limitedrelevance for the characterisation of potential subdomains of theTGN, because obviously the whole TGN was extremely stronglylabelled. The situation is even more complicated if the plant TGNcan quickly adapt to different cargo molecules as suggested byStaehelin and Kang (2008).

It should be mentioned that there is still no evidence forsubstantial penetration of antibodies and markers into wellpreserved cryosections (for review see (Stierhof and Schwarz,1989)), which is also true for well preserved cryosectionsobtained from cryofixed and freeze-substituted samples. This hasunfavourable implications for immuno-electron tomography,which is usually performed on thick sections, because theoutermost regions of the thawed cryosections, namely thesection surface, are predominantly labelled (e.g., (Mari et al.,2008)). The limited antibody and marker penetration wasprobably the reason that (destructive) BH4Na treatment was used

ARTICLE IN PRESS


to facilitate infiltration of immunoreagents to some extent(Zeuschner et al., 2006).

Visualisation of membranes of trafficking compartments in

high-pressure frozen, freeze-substituted and epoxy resin-embedded

plant samples

The problem of membrane (bilayer) visualisation after high-pressure freezing and freeze-substitution is not unique to

Fig. 6. Ultrastructural appearance of endosomal structures after high-pressure freezing

showing three layers (A, B; arrows) and vesicle budding (B) at the rim of the ESCRT c

vesicles close to the plasma membrane (arrowhead, C: pollen grain) or close to the TGN

COPII-like vesicle (F; arrow) and different stages of COPII budding (F� I; arrowheads). F

cells after resin embedding (K) (section thickness �50 nm) and in weakly fixed (extra

COPI vesicles and COPI-like vesicles, the arrow to a non-coated vesicle. (M, N) Ultr

substitution protocols. (M) Freeze-substitution was performed in 2.5% OsO4, 0.5% UA an

0.5% UA, 0.5% GA and 2% H2O. The arrowheads in (M) and (N) point to a COPI-like vesic

substitution medium was removed after incubating the samples for 1 h at 0 1C. c, cis-Go

t, TGN; v, small vacuole. Bars: 250 nm (A�D; F� I; K�N), 50 nm (E).

immunolabelled thawed cryosections. It is well known that thevisualisation of many membranes in plant cells and other walledcells, after high-pressure freezing, freeze-substitution and epoxyresin embedding, is quite difficult and often not very reproducible((Donohoe et al., 2006; Giddings, 2003; Hess, 2003, 2007; Murataet al., 2002); for other organisms, (Buser and Walther, 2008;Hawes et al., 2007)). Therefore, it was tempting to comparesamples that were identically cryofixed and freeze-substituted ina similar way in regard to the appearance of ultrastructuraldetails, in particular membranes. The bilayer structure of plasma

, freeze-substitution and epoxy resin embedding. (A, B) MVBs with ESCRT complex

omplex (arrowhead). Freeze-substitution with 2.5% OsO4. (C�E) Clathrin-coated

(root tip, D, E; arrowheads). Freeze-substitution in acetone with 2.5% OsO4. (F� I)

reeze-substitution with 2.5% OsO4. (K, L) Large numbers of small vesicles in pollen

cted) cryosections (L) (section thickness �100 nm). Arrowheads point to budding

astructural appearance of Golgi stacks and TGN after applying different freeze-

d 0.5% GA. (N) Freeze-substitution was performed in acetone containing 2.5% OsO4,

le. The insets show the same vesicles, two-fold enlarged. In both cases the freeze-

lgi; cw, cell wall; er, ER; g, Golgi stack; m, mitochondrion; pm, plasma membrane;

ARTICLE IN PRESS


and vacuolar membranes and MVB membranes, together with theESCRT complex (Fig. 6A, B) could be easily identified whensamples were freeze-substituted in acetone with OsO4 withoutthe addition of water. TGN and ER membranes and BFAcompartment membranes were more difficult to visualise,which is also true for small coated and uncoated transportvesicles. One reason could be that, in Arabidopsis cells, manyvesicles like COPI (Figs. 1G and 6N inset), COPII (Fig. 6F–I) andclathrin-coated vesicles (Fig. 6C–E) as well as uncoated vesicleshave a diameter of not more than 40–60 nm (Fig. 6E, F, K, N) (seealso (Dhonukshe et al, 2007; Donohoe et al., 2007; Robinson et al.,2007; Staehelin and Kang, 2008)), which is smaller than theaverage thickness of a resin section. Therefore, thinner sectionsshould give a sharper, better resolved image. Indeed, carefulinspection of cryofixed and freeze-substituted samples showedthat these vesicles are present, sometimes in large numbers, inplant cells (Fig. 6K, L).

Especially well frozen samples exhibit a reduced membranevisibility. Sometimes TGN and ER can just be identified by missingribosomes (not shown; see also (Hess, 2007; Donohoe et al.,2006)). Accordingly, samples with reduced amounts of water, likeplasmolysed cells or mature pollen grains, appeared to bewell-frozen, but showed a suboptimal bilayer visibility(not shown, see also (Hess, 2007)). This is in clear contrast tomaturing pollen grains with an excellent ultrastructure preserva-tion and bilayer visibility even without the addition of waterto the freeze-substitution medium (acetone with 2.5% OsO4)(Figs. 1G and 6C, K).

Some reports have shown that the addition of water to thefreeze-substitution medium results in the improved visibility ofcellular membranes (for epoxy resin sections, (Buser and Walther,2008; Hess, 2007); for cryosections after high-pressure freezing,freeze-substitution, and rehydration, (Reichardt et al., 2007;Ripper et al., 2008; Stierhof et al., 2008; van Donselaar et al.,2007)). However, it was also argued that the additional water mayprevent freeze-substitution at �90 1C, because, at �90 1C, acetonecan dissolve only 1% water (Humbel et al., 1983; Zalokar, 1966).Consequently, freeze-substitution may be delayed and the re-moval of water would take place only at higher temperatures. Thiseffect may result in ice crystal damage, because, at temperaturesabove �90 1C the cellular water could recrystallise, therebyimpairing the cellular ultrastructure (Zechmann et al., 2007;Steinbrecht, 1981). This would also explain why the overallcontrast and especially the membrane contrast are enhanced incells showing moderate ice crystal damage (not shown). Thegeneral variability in membrane bilayer visibility may also be dueto the amount of water in the sample and the amount of liquid inthe freeze-substitution medium. Substitution in 1 ml medium maypromote the ‘‘water effect’’, if too much water is introduced by thesample itself. To explore this problem in more detail, we added 2%water to the freeze-substitution medium (acetone with 2.5% OsO4)after two days at �90 1C without water. Then the sample wasincubated for another 24 h before raising the temperature. In thiscase, the membrane contrast was slightly enhanced (similarly tothat of cryosections from samples after high-pressure freezing,freeze-substitution and rehydration), however, significant effectscould not be observed. The strongest effect of water added to thefreeze-substitution cocktail was detected in samples freeze-substituted in 2.5% OsO4 and 0.5% UA and washed at 0 1C. Withoutwater, membrane visibility was worse compared to that ofsamples freeze-substituted in OsO4 alone, whereas with 2% water,the bilayer visibility was clearly improved (Fig. 6M, N; see alsoCOPI-vesicle in inset). Unfortunately, often also slight ice crystaldamage could be observed. This water effect on membranevisibility was not observed in all tissues. When water was addedto mature pollen grains in the substitution medium, ice crystal

damage was not detected, but membrane visibility was notsignificantly improved. Interestingly, a similar membrane contrastimproving effect of water in the freeze-substitution medium wasnot detected in high-pressure frozen, freeze-substituted andLowicryl-embedded Arabidopsis root tips. The use of 0.l% gallicacid during freeze-substitution before OsO4 treatment resulted indarker structures, but the contrast was very homogeneous und notbetter compared to that obtained with OsO4 alone (not shown, fora more detailed discussion see (Giddings, 2003; Hess, 2007)).Therefore, we preferred OsO4 fixation alone during freeze-substitution because, in general, it resulted in an acceptablevisibility of the ultrastructural details of different plant tissues,e.g., root tips, embryos, female gametophytes and anthers withmaturing pollen grains. However, it seems to be worthwhile tovary freeze-substitution protocols, if the ultrastructure appearanceis not satisfying. An additional possibility to enhance membranevisibility is the use of alcoholic UA staining solutions for resinsections, e.g., UA in 50% ethanol or in methanol (Epstein and Holt,1963; Stempak and Ward, 1964).

Conclusion

Cryofixation and freeze-substitution of difficult-to-fix planttissue combined with Tokuyasu thawed cryosection immunola-belling does not only improve the quality of ultrastructurepreservation but also that of antigenicity preservation. Incombination with the most sensitive marker systems available,together with a large variety of possible fixatives, it is possible todetect both fixation-sensitive antigens as well as low amounts ofantigens that may otherwise escape detection.

A surprisingly large number of TGN-associated antigens isinvolved in the sorting of newly synthesised or recycled materialto the plasma membrane and to the MVB and tonoplast and alsoin retrograde transport. This indicates a high plasticity of thiscompartment but hampers the identification of potential sub-domains. In addition, large numbers of relatively small transportvesicles which are difficult to visualise and characterise hamperthe detailed deciphering of trafficking pathways in plants. Theproblem of visualisation of the membrane bilayer structure afterhigh-pressure freezing and freeze-substitution is still not satis-factorily solved especially in the case of well frozen samples,either for immunolabelled cryosections or resin sections. Theaddition of water to the freeze-substitution medium does notalways improve the ultrastructure appearance of resin-embeddedtissues and may result in mild ice-crystal damage.

Acknowledgements

We are grateful to Karin Schumacher (Heidelberg Institute forPlant Sciences, Developmental Biology, University of Heidelberg,Heidelberg, Germany), and Axel Volker, Hanno Wolters and GerdJurgens (ZMBP, Developmental Genetics) for providing VHA-a1-GFP-, VHA-a1-mRFP-, and GFP-KNOLLE-expressing Arabidopsis

lines, and David G. Robinson (Heidelberg Institute for PlantSciences, Cell Biology, University of Heidelberg, Heidelberg,Germany) and Inhwan Hwang (Division of Molecular and LifeSciences, Pohang University of Science and Technology, Pohang,Korea) for providing antibodies. Development and distributionof antibody CCRC-M1 was supported in part by NSF grantsDBI-0421683 and RCN-0090281. We thank Heinz Schwarz(Max-Planck-Institute for Developmental Biology, Tubingen,Germany) for providing the high-pressure freezer, Dagmar Ripper,Barbara Maier and Stefan Pfeffer for excellent technical work, andHeinz Schwarz and Karin Schumacher for critically reading the

ARTICLE IN PRESS


manuscript. This work was supported by the Deutsche For-schungsgemeinschaft through SFB 446/Z2.

References

Baschong, W., Stierhof, Y.-D., 1998. Preparation, use and enlargement of ultrasmallgold particles in immunoelectron microscopy. Microsc. Res. Tech. 42, 66–79.

Buser, C., Walther, P., 2008. Freeze-substitution: the addition of water to polarsolvents enhances the retention of structure and acts at temperatures around�60 1C. J. Microsc. 230, 268–277.

Chow, C.M., Neto, H., Foucart, C., Moore, I., 2008. Rab-A2 and Rab-A3 GTPasesdefine a trans-Golgi endosomal membrane domain in Arabidopsis thatcontributes substantially to the cell plate. Plant Cell 20, 101–123.

Crowell, E.F., Bischoff, V., Desprez, T., Rolland, A., Stierhof, Y.-D., Schumacher, K.,Gonneau, M., Hofte, H., Vernhettes, S., 2009. Pausing of Golgi bodies onmicrotubules regulates secretion of cellulose synthase complexes in Arabi-dopsis. Plant Cell 21, 1141–1154.

Danscher, G., 1981. Histochemical demonstration of heavy metals: a revisedversion of the sulphide silver method suitable for both light and electronmicroscopy. Histochemistry 71, 1–16.

Deerinck, T.J., 2008. The application of fluorescent quantum dots to confocal,multiphoton, and electron microscopic imaging. Toxicol. Pathol. 36, 112–116.

Dettmer, J., Hong-Hermesdorf, A., Stierhof, Y.-D., Schumacher, K., 2006. VacuolarH+-ATPase activity is required for endocytic and secretory trafficking inArabidopsis. Plant Cell 18, 715–730.

Dhonukshe, P., Aniento, F., Hwang, I., Robinson, D.G., Mravec, J., Stierhof, Y.-D.,Friml, J., 2007. Clathrin-mediated constitutive endocytosis of PIN auxin effluxcarriers in Arabidopsis. Curr. Biol. 17, 1–8.

Donohoe, B.S., Mogelsvang, S., Staehelin, L.A., 2006. Electron tomography of ER,Golgi and related membrane systems. Methods 39, 154–162.

Donohoe, B.S., Kang, B.-H., Staehelin, L.A., 2007. Identification and characterizationof COPIa- and COPIb-type vesicle classes associated with plant and algal Golgi.Proc. Natl. Acad. Sci. USA 104, 163–168.

Ebine, K., Okatani, Y., Uemura, T., Goh, T., Shoda, K., Niihama, M., Morita, M.T.,Spitzer, C., Otegui, M.S., Nakano, A., Ueda, T., 2008. A SNARE complex unique toseed plants is required for protein storage vacuole biogenesis and seeddevelopment of Arabidopsis thaliana. Plant Cell 20, 3006–3021.

Epstein, M.A., Holt, S.J., 1963. The localization by electron microscopy of HeLa cellsurface enzymes splitting adenosine triphosphate. J. Cell Biol. 19, 325–326.

Giddings, T.H., 2003. Freeze-substitution protocols for improved visualization ofmembranes in high-pressure frozen samples. J. Microsc. 212, 53–61.

Giepmans, B.N.G., Deerinck, T.J., Smarr, B.L., Jones, Y.Z., Ellisman, M.H., 2005.Correlated light and electron microscopic imaging of multiple endogenousproteins using quantum dots. Nat. Methods 2, 743–749.

Grebe, M., Xu, J., Mobius, W., Ueda, T., Nakano, A., Geuze, H.J., Rook, M.B., Scheres,B., 2003. Arabidopsis sterol endocytosis involves actin-mediated trafficking viaARA6-positive early endosomes. Curr. Biol. 13, 1378–1387.

Griffith, J.M., Posthuma, G., 2002. A reliable and convenient method to storeultrathin thawed cryosections prior to immunolabeling. J. Histochem.Cytochem. 50, 57–62.

Griffiths, G., 1993. Fine Structure Immunocytochemistry. Springer-Verlag, Berlin.Griffiths, G., Simons, K., Warren, G., Tokuyasu, K.T., 1983. Immunoelectron

microscopy using thin frozen sections: application to studies of theintracellular transport of Semliki forest virus spike glycoprotein. MethodsEnzymol. 96, 466–486.

Haas, T.J., Sliwinski, M.K., Martınez, D.E., Preuss, M., Ebine, K., Ueda, T., Nielsen, E.,Odorizzi, G., Otegui, M.S., 2007. The Arabidopsis AAA ATPase SKD1 is involvedin multivesicular endosome function and interacts with its positive regulatorLYST-INTERACTING PROTEIN5. Plant Cell 19, 1295–1312.

Hainfeld, J.F., Furuya, F.R., 1992. A 1.4 nm cluster covalently attached to antibodiesimproves immunolabeling. J. Histochem. Cytochem. 40, 177–184.

Hawes, P., Netherton, D.L., Muller, M., Wileman, T., Monaghan, P., 2007. Rapidfreeze-substitution preserves membranes in high-pressure frozen tissueculture cells. J. Microsc. 226, 182–189.

Hermann, R., Schwarz, H., Muller, M., 1991. High precision immunoscanningelectron microscopy using Fab fragments coupled to ultra-small colloidal gold.J. Struct. Biol. 107, 38–47.

Hess, M.W., 2003. Of plants and other pets: practical aspects of freeze-substitutionand resin embedding. J. Microsc. 212, 44–52.

Hess., M.W., 2007. Cryopreparation methodology for plant cell biology. MethodsCell Biol. 79, 57–100.

Hinz, G., Colanesi, S., Hillmer, S., Rogers, J.C., Robinson, D.G., 2007. Localization ofvacuolar transport receptors and cargo proteins in the Golgi apparatus ofdeveloping Arabidopsis embryos. Traffic 8, 1452–1464.

Humbel, B.M., 2008. Freeze-substitution. In: Cavalier, D., Spehner, D., Humbel, B.M.(Eds.), Handbook of Cryo-Preparation Methods for Electron Microscopy. CRCPress, Boca Raton, pp. 319–341.

Humbel, B.M., Schwarz, H., 1989. Freeze-substitution for immunochemistry. In:Verkleij, A.J., Leunissen, J.L.M. (Eds.), Immuno-Gold Labelling in Cell Biology.CRC Press, Boca Raton, pp. 115–134.

Humbel, B.M., Stierhof, Y.-D., 2008. Cryo-sectioning according to Tokuyasu. In:Cavalier, A., Spehner, D., Humbel, B.M. (Eds.), Handbook of Cryo-PreparationMethods for Electron Microscopy. CRC Press, Boca Raton, pp. 467–498.

Humbel, B.M., Marti, D., Muller, M., 1983. Improved structural preservation bycombining freeze substitution and low temperature embedding. Beitr.Elektronenmikroskop. Direktabb. Oberfl. 16, 585–594.

Kim, Y.W., Park, D.S., Park, S.C., Kim, S.H., Cheong, G.W., Hwang, I., 2001.Arabidopsis dynamin-like 2 that binds specifically to phosphatidylinositol 4-phosphate assembles into a high-molecular weight complex in vivo andin vitro. Plant Physiol. 127, 1243–1255.

Lam, S.K., Siu, C.L., Hillmer, S., Jang, S., An, G., Robinson, D.G., Jiang, L., 2007a. PiceSCAMP1 defines clathrin-coated, trans-Golgi-located tubular-vesicular struc-tures as an early endosome in tobacco BY-2 cells. Plant Cell 19, 296–319.

Lam, S.K., Tse, Y.C., Robinson, D.G., Jiang, L., 2007b. Tracking down the elusive earlyendosome. Traffic 12, 497–505.

Lauber, M.H., Waizenegger, I., Steinmann, T., Schwarz, H., Mayer, U., Hwang, I.,Lukowitz, W., Jurgens, G., 1997. The Arabidopsis KNOLLE protein is acytokinesis-specific syntaxin. J. Cell Biol. 139, 1485–1493.

Leunissen, J.L.M., Van de Plas, P., 1993. Ultrasmall gold probes and cryoultrami-crotomy. In: Hyatt, A.D., Eaton, B.T. (Eds.), Immuno-Gold Electron Microscopyin Virus Diagnosis and Research. CRC Press, Boca Raton, pp. 327–348.

Liou, W., Geuze, H.J., Slot, J.W., 1996. Improving structural integrity of cryosectionsfor immunogold labeling. Histochem. Cell Biol. 106, 41–58.

Mari, M., Bujny, M.V., Zeuschner, D., Geerts, W.J., Griffith, J., Petersen, C.M., Cullen,P.J., Klumperman, J., Geuze, H.J., 2008. SNX1 defines an early endosomalrecycling exit for sortilin and mannose 6-phosphate receptors. Traffic 9,380–393.

Mori, M., Ishikawa, G., Takeshita, T., Goto, T., Robinson, J.M., Takizawa, T., 2006.Ultrahigh-resolution immunofluorescence microscopy using ultrathin cryo-sections: subcellular distribution of caveolin-1alpha and CD31 in humanplacental endothelial cells. J. Electron Microsc. (Tokyo) 55, 107–112.

Movafeghi, A., Happel, N., Tai, G.-H., Robinson, D.G., 1999. Arabidopsis Sec21p andSec23p homologs: probable coat proteins of plant COP-coated vesicles. PlantPhysiol. 119, 1437–1446.

Muller, I., Wagner, W., Volker, A., Schellmann, S., Nacry, P., Kuttner, F., Schwarz-Sommer, Z., Mayer, U., Jurgens, G., 2003. Syntaxin specificity of cytokinesis inArabidopsis. Nat. Cell Biol. 5, 531–534.

Murata, T., Karahara, I., Kozuka, T., Thomas Jr., H.G., Staehelin, L.A., Mineyuki, Y.,2002. Improved method for visualizing coated pits, microfilaments, andmicrotubules in cryofixed and freeze-substituted plant cells. J. Electron.Microsc. 51, 133–136.

Otegui, M.S., Spitzer, C., 2008. Endosomal functions in plants. Traffic 9, 1589–1598.Otegui, M.S., Mastronarde, D.N., Kang, B.-H., Bednarek, S.Y., Staehelin, L.A., 2001.

Three-dimensional analysis of syncytial-type cell plates during endospermcellularization visualized by high-resolution electron tomography. Plant Cell13, 2033–2051.

Pimpl, P., Movafeghi, A., Coughlan, S., Denecke, J., Hillmer, S., Robinson, D.G., 2000.In situ localization and in vitro induction of plant COPI-coated vesicles. PlantCell 12, 2219–2236.

Pohl, K., Stierhof, Y.-D., 1998. Action of gold chloride (‘‘gold toning’’) on silver-enhanced 1 nm gold markers. Microsc. Res. Tech. 42, 59–65.

Reichardt, I., Stierhof, Y.-D., Mayer, U., Richter, S., Schwarz, H., Schumacher, K.,Jurgens, G., 2007. Plant cytokinesis requires de novo secretory trafficking butnot endocytosis. Curr. Biol. 17, 2047–2053.

Richter, S., Voss, U., Jurgens, G., 2009. Post-Golgi traffic in plants. Traffic 10,819–828.

Ripper, D., Schwarz, H., Stierhof, Y.-D., 2008. Cryo-section immunolabelling ofdifficult to preserve specimens: advantages of cryofixation, freeze-substitutionand rehydration. Biol. Cell 100, 109–123.

Robinson, D.G., Herranz, M.C., Bubeck, J., Pepperkok, R., Ritzenthaler, C., 2007.Membrane dynamics in the early secretory pathway. Crit. Rev. Plant Sci. 26,199–225.

Robinson, D.G., Jiang, L., Schumacher, K., 2008. The endosomal system of plants:charting new and familiar territories. Plant Physiol. 147, 1482–1492.

Schwarz, H., Humbel, B.M., 2007. Correlative light and electron microscopy usingimmunolabeled resin sections. Methods Mol. Biol. 369, 229–256.

Slot, J.W., Geuze, H.J., 1985. A new method of preparing gold probes for multiple-labeling cytochemistry. Eur. J. Cell Biol. 38, 87–93.

Slot, J.W, Geuze, H.J., 2007. Cryosectioning and immunolabeling. Nat. Protocols 2,2480–2491.

Staehelin, L.A., Kang, B.H., 2008. Nanoscale architecture of endoplasmic reticulumexport sites and of Golgi membranes as determined by electron tomography.Plant Physiol. 147, 1454–1468.

Steinbrecht, R.A., 1981. Experiments on freezing damage with freeze substitutionusing moth antennae as test objects. J. Microsc. 125, 187–192.

Stempak, J.G., Ward, R.T., 1964. An improved staining method for electronmicroscopy. J. Cell Biol. 22, 697–701.

Stierhof, Y.-D., 2008. Immunolabelling of ultrathin sections with enlarged1 nm gold or Qdots. In: Cavalier, A., Spehner, D., Humbel, B.M. (Eds.),Handbook of Cryo-Preparation Methods for Electron Microscopy. CRC Press,Boca Raton, pp. 587–616.

Stierhof, Y.-D., Schwarz, H., 1989. Labeling properties of sucrose-infiltratedcryosections. Scanning Microsc. Suppl. 3, 35–46.

Stierhof, Y.-D., Hermann, R., Humbel, B.M., Schwarz, H., 1995. Use of TEM, SEM,and STEM in imaging 1-nm colloidal gold particles. In: Hayat, M.A. (Ed.),Immunogold-silver Staining. Principles, Methods, and Applications. CRC Press,Boca Raton, pp. 97–118.

Stierhof, Y.-D., van Donselaar, E., Schwarz, H., Humbel, B.M., 2008. Cryo-fixation,freeze-substitution, rehydration and Tokuyasu cryo-sectioning. In: Cavalier, A.,

ARTICLE IN PRESS


Spehner, J., Humbel, B.M. (Eds.), Handbook of Cryo-Preparation Methods forElectron Microscopy. CRC Press, Boca Raton, pp. 343–366.

Studer, D., Michel, M., Muller, M., 1989. High pressure freezing comes of age.Scanning Microsc. Suppl. 3, 253–268.

Takizawa, T., Robinson, J.M., 2003. Ultrathin cryosections: An important tool forimmunofluorescence and correlative microscopy. J. Histochem. Cytochem. 51,707–714.

Tokuyasu, K.T., 1973. A technique for ultracryotomy of cell suspensions andtissues. J. Cell Biol. 57, 551–565.

Tokuyasu, K.T., 1978. A study of positive staining of ultrathin frozen sections.J. Ultrastruct. Res. 63, 287–307.

Tokuyasu, K.T., 1989. Use of poly(vinylpyrrolidone) and poly(vinyl alcohol) forcryoultramicrotomy. Histochem. J. 21, 163–171.

Tokuyasu, K.T., 1997. Immunocytochemistry on ultrathin cryosections. In: Spector,D.L., Goldman, R.D., Leonwand, L.A. (Eds.), Cells: A Laboratory Manual. ColdSpring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 131-1–131-27.

Toyooka, K., Goto, Y., Asatsuma, S., Koizumi, M., Mitsui, T., Matsuoka, K., 2009. Amobile secretory vesicle cluster involved in mass transport from the Golgi tothe plant cell exterior. Plant Cell 21, 1212–1229.

Tse, Y.C., Mo, B., Hillmer, S., Zhao, M., Lo, S.W., Robinson, D.G., Jiang, L., 2004.Identification of multivesicular bodies as prevacuoloar compartments inNicotiana tabacum BY-2 cells. Plant Cell 16, 672–693.

van Donselaar, E., Posthuma, G., Zeuschner, D., Humbel, B.M., Slot, J.W., 2007.Immunogold labelling of cryo-sections from high-pressure frozen cells. Traffic8, 471–485.

Venable, J.H., Coggeshall, 1965. A simplified lead citrate stain for use in electronmicroscopy. J. Cell Biol. 25, 407–408.

von der Fecht-Bartenbach, J., Bogner, M., Krebs, M., Stierhof, Y.-D., Schumacher, K.,Ludewig, U., 2007. Function of the anion transporter AtCLC-d in the trans-Golginetwork. Plant J. 50, 466–474.

Webster, P., Schwarz, H., Griffiths, G., 2008. Preparation of cells and tissues forimmuno EM. Methods Cell Biol. 88, 45–58.

Zalokar, M., 1966. A simple freeze-substitution method for electron microscopy.J. Ultrastruct. Res. 15, 469–479.

Zechmann, B., Muller, M., Zellnig, G., 2007. Membrane associated qualitativedifferences in cell ultrastructure of chemically and high pressure cryofixedplant cells. J. Struct. Biol. 158, 370–377.

Zeuschner, D., Geerts, W.J.C., van Donselaar, E., Humbel, B.M., Slot, J.W., Koster, A.J.,Klumperman, J., 2006. Immuno-electron tomography of ER exit sites revealsthe existence of free COP II-coated transport carriers. Nat. Cell Biol. 8, 377–383.

Zhang, G.F., Staehelin, L.A., 1992. Functional compartmentation of the Golgiapparatus of plant cells: Immunocytochemical analysis of high-pressurefrozen and freeze-substituted sycamore maple suspension culture cells. PlantPhysiol. 99, 1070–1083.

strategies to improve the antigenicity,ultrastructure preservation and

Documents