©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

brief communications

nature methods  |  ADVANCE ONLINE PUBLICATION  |  �

­sensitive,­but­mapping­all­connections­requires­probing­each­of­the­approximately­1,000­brain­areas­multiple­times.­Thus,­an­esti-mated­10,000­brains2­would­need­to­be­injected,­imaged,­aligned­to­a­common­reference­and­analyzed.­Alternatively,­one­could­image­a­single­brain­at­a­resolution­sufficient­to­trace­all­axons­at­once.­The­set­of­all­axonal­trajectories­would­then­comprise­a­map­of­all­inter-areal­connections.­Currently­only­electron­microscopy­provides­sufficient­resolution­to­follow­individual­axons­reliably­in­densely­stained­samples.­Following­neural­processes­through-out­a­whole­brain­requires­special­sample­preparation­techniques­that­allow­the­entire­brain­to­be­fixed,­stained­and­embedded.

There­are­two­broad­strategies­for­whole-brain­staining:­per-fusion­and­immersion.­In­the­first­case,­the­animal’s­vasculature,­from­which­no­part­of­the­brain­is­generally­farther­than­a­few­tens­of­micrometers­ (ref.­3),­ is­used­ to­ introduce­ fixation­and­staining­ chemicals­ into­ the­ tissue.­ Transcardial­ perfusion­ has­long­been­the­method­of­choice­for­the­initial­fixation­of­whole­brains4.­Perfusion­of­staining­solutions­has­rarely­been­used,­even­though­fixation­and­staining­can­be­achieved5,6­using­perfusion­with­osmium­tetroxide­(OsO4),­commonly­used­in­the­past­as­the­sole­fixative­before­aldehyde­fixation4­was­introduced.­However,­OsO4­alone­gives­relatively­poor­membrane­contrast4,7,­and­OsO4­perfusion­results­in­incomplete­and­capricious­staining­of­white­matter5,6,­a­serious­problem­if­long-range­axons­are­to­be­traced.­The­other­way­to­introduce­chemicals­into­tissue­is­by­immer-sion,­which­relies­on­diffusion­to­transport­the­molecules­from­

staining and embedding the whole mouse brain for electron microscopyShawn Mikula, Jonas Binding & Winfried Denk

the development of methods for imaging large contiguous volumes with the electron microscope could allow the complete mapping of a whole mouse brain at the single-axon level. We developed a method based on prolonged immersion that enables staining and embedding of the entire mouse brain with uniform myelin staining and a moderate preservation of the tissue’s ultrastructure. We tested the ability to follow myelinated axons using serial block-face electron microscopy.

Mapping­the­mouse­inter-areal­connectome,­also­called­the­pro-jectome1­because­it­contains­all­projections­between­areas,­would­profoundly­advance­our­understanding­of­whole-brain­function­by­showing­precise­pathways­of­information­flow.­Currently,­the­most­comprehensive­map­is­only­10%­complete­and­is­based­on­a­meta-analysis­of­the­rodent­connectivity­literature2,­mostly­tracer­injection­ studies.­ Local­ injection–based­ analysis­ can­ be­ very­

Max-Planck­Institute­for­Medical­Research,­Heidelberg,­Germany.­Correspondence­should­be­addressed­to­S.M.­(Shawn.Mikula@mpimf-heidelberg.mpg.de)­or­­W.D.­(Winfried.Denk@mpimf-heidelberg.mpg.de).Received 14 May; accepted 28 SepteMbeR; pUbLiSHed ONLiNe 21 OctObeR 2012; dOi:10.1038/NMetH.2213

a

c d e

bBregma AP –1.26 mm

AP

–1.

26

d*

c

800 nm 800 nm 300 nm

mit

TEM

Axm

60 nmSEM SEM

b

1 mm 50 µm

5 µm5 µm

SEM

*

**

**

e*

5 µm

figure � | Whole mouse brain stained with wbPATCO and embedded with Quetol. (a) Block-face image of a whole-brain cross-section cut at the level of bregma −1.26, coated with 5 nm platinum-carbon and imaged with scanning electron microscopy (SEM) using secondary-electron detection. Inset, horizontal view of the entire mouse brain after embedding; block dimensions are 6 × 8.5 × 14 mm3. AP, anterior-posterior. (b) Single image tile (green box in a). Inset, magnified subregion indicated by the blue box. (c,d) High-magnification SEM images from labeled asterisks of the block face in a, demon-strating ultrastructural preservation of deep (c) and superficial (d) regions. Note membrane discontinuities (asterisks) and postsynaptic densities (arrowheads). (e) High-magnification transmission electron microscopy (TEM) image of a 70-nm section taken from a region in the dorsolateral striatum (labeled asterisk in a) in a different sample. Note the intra-period and major dense lines in the inset in e. mit, mitochondrion; Ax, axon; m, microtubules.

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

�  |  ADVANCE ONLINE PUBLICATION  |  nature methods

brief communications

the­ sample’s­ surface­ to­ its­ interior.­ This­approach,­ which­ has­ been­ used­ to­ stain­and­embed­whole­mouse­brains­for­light-­microscopic­ observation8,9,­ scales­ unfa-vorably­ with­ sample­ size­ because­ diffu-sion­ times­ increase­ quadratically­ with­distance10,11.­ The­ brain­ is,­ therefore,­commonly­cut­into­pieces­at­most­a­few­millimeters­in­size­after­the­initial­perfusion­fixation12,13.­For­whole-brain­connectivity­­analysis,­this­is­unsuitable­because­such­cutting­always­leads­to­surface­ damage,­ which­ in­ all­ likelihood­ precludes­ the­ reliable­­tracing­of­thin­axons­from­one­piece­to­the­adjacent­one.­Here­we­report­a­procedure­that­is­based­mostly­on­long­immersion­steps­ and­ results­ in­ whole-brain­ samples­ that­ are­ stained­ and­­embedded­well­enough­to­reliably­follow­myelinated­axons.

Aside­ from­ the­ initial­ transcardial­ perfusion,­ we­ based­ our­approach­on­ sample­ immersion.­To­compensate­ for­ the­about­tenfold­increase­in­diffusion­distance­due­to­the­larger­size­of­a­whole-brain­sample,­we­increased­the­periods­of­immersion­to­48­h­for­each­staining­step­(Supplementary Protocol).­Because­the­ effective­ diffusion­ constants­ of­ the­ staining­ chemicals­ in­brain­ tissue­ cannot­ be­ larger­ than­ that­ for­ small­ solutes­ in­­aqueous­solutions,­roughly­10−9­m2­s−1­(ref.­14),­diffusion­cannot­­equilibrate­concentrations­over­a­distance­of­much­more­ than­­10­mm­in­48­h­(the­whole­adult­mouse­brain­is­less­than­10­mm­in­width­and­height).

We­initially­used­single-step­incubations­of­brain­hemispheres­in­OsO4­(0.5–4%),­using­unbuffered­and­buffered­(phosphate,­Tris,­cacodylate­or­collidine)­solutions.­None­of­the­buffers­improved­penetration­compared­to­the­unbuffered­case.­When­using­unbuff-ered­2%­OsO4­at­20­°C,­the­penetration­depths­at­different­incu-bation­times­(data­not­shown)­were­consistent­with­an­effective­diffusion­constant­of­roughly­2­×­10−10­m2­s−1,­slower­than­free­diffusion­in­water­but­much­faster­than­the­value­of­roughly­10−11­m2­s−1­reported­for­liver10.­Provided­that­the­quantity­of­OsO4­was­sufficiently­high­(>250­mg)­and­the­incubation­times­sufficiently­long­(>24­h),­staining­was­uniform­throughout­the­whole­hemi-sphere,­but­the­contrast­was­generally­poor­for­membranes­and­myelin­(Supplementary Fig. 1a).­A­further­increase­in­incubation­

time­(>96­h)­caused­a­decrease­in­membrane­contrast.­The­inclu-sion­of­organic­solvents­did­not­facilitate­penetration­and­led­to­a­precipitation­and­progressive­darkening­of­the­staining­solutions,­presumably­by­reducing­OsO4­(ref.­15).

To­improve­contrast,­we­examined­enhanced­staining­proto-cols,­first­using­small­tissue­pieces­(<1­mm)­and­short­incubation­times­(90­min).­Guided­by­the­results,­we­then­used­hemispheres­or­ whole­ brains­ and­ long­ incubation­ times­ (24–48­ h).­ The­­protocols­and­comparative­results­are­given­in­Supplementary Tables 1 and 2.­ By­ combining­ a­ single­ OsO4­ step­ with­ sub-sequent­enhancement­using­uranyl­acetate­or­lead­aspartate,­we­slightly­increased­membrane­contrast­(Supplementary Fig. 1b).­­The­ OsO4-thiocarbohydrazide-OsO4­ (OTO)­ method7­ yielded­samples­with­high­membrane­contrast­at­short­incubation­times,­but­with­incubation­times­of­24–48­h­it­suffered­from­cytosol­darkening­ (Supplementary Fig. 1c),­ which­ reduced­ mem-brane­contrast,­masked­synapses­and­resulted­in­bubble­forma-tion,­possibly­due­to­N2­gas­liberated­from­hydrazino­groups­in­­tissue-bound­thiocarbohydrazide­(TCH)16.­The­rOTO­method,­which­uses­ferrocyanide-OsO4­(‘reduced­osmium’;­refs.­17,18)­instead­of­OsO4­alone­in­the­first­incubation­step,­substantially­increased­membrane­contrast­over­the­OTO­method,­but­stain­penetration­ was­ poor­ even­ at­ 48-h­ incubation­ time,­ and­ the­­tissue­ultrastructure­was­damaged­(data­not­shown).­By­embedding­and­imaging­samples­at­various­points­of­the­staining­protocol,­­we­ determined­ that­ the­ poor­ penetration­ resulted­ from­ the­reduced-osmium­step.

Because­ the­ OTO­ method,­ despite­ suffering­ from­ nonuni-form­staining­and­embedding­problems,­resulted­in­good­mem-brane­contrast­for­short­incubation­times,­we­examined­similar­­staining­ methods­ that­ also­ use­ TCH19.­ The­ periodic-acid–­thiocarbohydrazide-OsO4­(PATCO)19­method­is­a­precursor­of­

Lateral 0.5 mm

Corpus callosum

a

c

b

d

e

71 µ

m

72 µm82 µm

Center cube

y

x

x

y

z

z

y

5 µm 5 µm

ECCC

V1DHC

VPL

IC

Amyg

DpMe

EC

CC

V1

DHC

VPLIC

Amyg

DpMe

figure � | Traceability analysis of eight regions of interest (ROIs) in the mouse brain. (a) Parasagittal view showing projections of ROIs. (b) Horizontal view of ROIs through a transparent brain. (c) Serial block-face electron microscopy stack from the corpus callosum, cut down the middle, with 50 traced axons emerging, randomly colored. (d) Single xy image from the stack. (e) yz reslice from the center of a cube (same as in c) showing locations of the 50 seed tracings (cyan points) for the axons shown in c. Amyg, amygdala; CC, corpus callosum; DHC, dorsal hippocampal commissure; DpMe, deep mesencephalic nucleus; EC, external capsule; IC, internal capsule; V1, primary visual cortex (superficial layers); VPL, ventroposterolateral nucleus of the dorsal thalamus. Color coding in a and b denotes major brain subdivisions (cerebral cortex, olfactory bulb, cerebellum, brainstem, superior colliculus and so on).

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature methods  |  ADVANCE ONLINE PUBLICATION  |  �

brief communications

the­OTO­method­but­uses­periodic­acid­instead­ of­ OsO4­ in­ the­ first­ incubation­step­and­was­originally­designed­to­stain­glycogen­and­other­polysaccharides.­We­found­that­it­also­enhanced­myelin­con-trast­ at­ least­ twofold­ (Supplementary Fig. 1d),­ possibly­ because­ of­ the­ high­levels­of­glycolipids­and­cerebrosides­ in­myelin20.­By­using­buffered­periodic-acid­and­TCH­solutions,­we­obtained­uniform­and­ complete­ staining­ throughout­ the­brain.­We­will­use­‘wbPATCO’­to­refer­to­the­PATCO­method­that­we­optimized­for­whole-brain­staining.

For­embedding,­we­first­tested­an­Epon­812­substitute,­glycid­ether­100­(Serva),­and­we­found­that­it­became­viscous­within­several­hours,­thereby­preventing­complete­infiltra-tion.­We­next­explored­the­use­of­low-viscosity­epoxies­and­found­that,­with­an­immersion­time­of­12­h­for­each­dehydration­and­epoxy-­monomer­ infiltration­ step,­ both­ Spurr’s­ resin21­ and­ Quetol22­yielded­whole-brain­blocks­of­uniform­mechanical­quality.­Image­contrast­did­not­depend­on­the­embedding­medium.­Embedded­samples­ were­ about­ 3–4%­ larger­ (in­ length)­ than­ they­ were­­following­fixation.

Staining­quality­and­uniformity­were­assessed­for­three­brains­prepared­ using­ the­ wbPATCO­ method.­ We­ acquired­ a­ set­ of­block-face­ images­ at­ 80-nm­ pixel­ size­ that­ covered­ an­ entire­­coronal­cross-section­through­one­whole-brain­sample­(WB1405,­Fig. 1).­ Upon­ close­ inspection­ of­ the­ individual­ image­ tiles­­(Fig. 1b),­we­observed­that­staining­and­tissue­preservation­were­uniform­across­the­entire­cross-section.­Ultrastructural­preserva-tion­was­moderately­good­and­consistent­in­deep­and­superficial­regions­of­the­sample­(Fig. 1).

To­ analyze­ traceability­ of­ myelinated­ axons­ in­ wbPATCO­­samples,­ we­ acquired­ serial­ block-face­ electron­ microscopy­(SBEM)­ stacks­ from­ several­ locations­ distributed­ widely­throughout­ the­ brain­ (Fig. 2a,b).­ No­ cutting­ problems­ were­­apparent­(Fig. 2c–e­and­Supplementary Video 1),­thus­indicating­

that­epoxy­infiltration­and­polymerization­were­complete­through-out­the­sample.­To­determine­axon­traceability­as­a­function­of­imaging­ resolution,­ we­ traced­ the­ SBEM­ stacks­ at­ the­ native­­40-nm­voxel­size­and­down-sampled­to­80-nm­and­120-­nm­reso-lution­(Figs. 2c and 3).

Axons­ were­ generally­ straight­ and­ followed­ a­ dominant­direction­ in­ white­ matter,­ but­ they­ followed­ more­ tortuous­paths­ with­ no­ apparent­ direction­ preference­ in­ gray­ matter­(Fig. 3a–c­and­Supplementary Video 2).­On­average,­ there­were­ 0.92­ nodes­ on­ each­ of­ the­ traced­ white-matter­ axons­(10.14­ nodes­ per­ mm,­n­ =­ 200­ axons),­ with­ no­ correlation­apparent­ in­node­ locations­ for­neighboring­axons.­For­gray­matter,­the­number­of­nodes­was­slightly­larger­(1.21­nodes­per­axon­traced,­11.58­nodes­per­mm,­n­=­200­axons),­presumably­because­of­the­higher­rate­of­branching.­Although­the­distribu-tions­of­axon­calibers­did­not­vary­substantially­among­regions­of­interest,­the­node-of-Ranvier­lengths­showed­more­variation,­ranging­from­1­to­4­μm­(Fig. 3d).­For­example,­node-of-Ranvier­length­ in­the­amygdala­was­1.39­±­0.84­µm­(±­ s.d.,­n­=­30),­whereas­ in­ the­dorsal­hippocampal­ commissure­ it­was­3.54­±­0.91­µm­(n­=­27).­We­found­a­mean­inner-axon­caliber­of­­873­nm­for­white­matter­(n­=­200­axons),­which­is­substantially­larger­than­the­510­nm­reported­for­subcortical­white­matter23­­

VPL of dorsal thalamus

Internal capsule

ba

c 6

5

Nod

e-of

-Ran

vier

leng

th (

µm)

4

3

2

1

00 500 1,000

Axon caliber (nm)1,500

Gray matter

Whitematter

d CCDHCECICDPMEVPLAMGV1

2,000 2,500

Gray matter

0.5f

Total

Resolution

Internode

120 nm80 nm

120 nm80 nm

40 nm

40 nmE

rror

freq

uenc

y

Axon caliber (nm)0 500

1,500 2,0001,000

0.4

0.3

0.2

0.1

0

Axon caliber (nm)

White matter

0.08

e

0.06

0.04

Err

or fr

eque

ncy

0.02

0

500 1,50002,000

Total

Inte

rnod

e

1,000

figure � | Analysis of axon morphological diversity and tracing error rate. (a–c) Axons traced in the ventroposterolateral nucleus (VPL) of the dorsal thalamus (a) with corresponding volumetric representation (b), and axons traced in the internal capsule (c). Fifty axons are displayed per region, with five axons highlighted in different colors to emphasize individual axon morphologies in relation to axon bundles and groupings (a,c). Nodes of Ranvier are indicated by small gray spheres. (d) Quantification of node-of-Ranvier length versus axon caliber in all eight regions of interest for axons containing a node of Ranvier with matching paranodes contained in the serial block-face electron microscopy cube. Ellipses indicate the s.d. of the data (n = 339 axons) for white- and gray-matter regions. (e,f) White-matter (e) and gray-matter (f) tracing error rates for myelinated axons of different axon-diameter bands and data display resolutions (n = 200 axons for each plot).

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

�  |  ADVANCE ONLINE PUBLICATION  |  nature methods

brief communications

and­460­nm­for­the­corpus­callosum24.­This­discrepancy­far­exceeds­ the­ overall­ linear­ expansion­ of­ 3–4%­ seen­ in­ our­­ sample,­ but­ it­ is­ possible­ that­ axons­ shrank­ more­ in­ their­preparation­or­expanded­more­in­our­preparation,­or­that­the­sampled­areas­were­different.

Error­rates­for­tracing­differed­for­white­and­gray­matter­but­did­not­depend­much­on­axon­diameter­for­axons­with­diameters­greater­than­500­nm­(Fig. 3e,f).­At­the­40-nm­native­resolution,­tracing­errors­were­confined­to­nodes­of­Ranvier,­with­about­one­error­for­every­80­nodes­(one­error­per­32­and­17­nodes­at­80-­and­120-nm­resolution,­respectively,­n­=­200­axons)­in­white­matter­and­one­error­per­30­nodes­(10­and­6­nodes­at­80­and­120­nm,­­n­ =­ 200­ axons)­ in­ gray­ matter­ (Supplementary Table 3).­ At­­80-nm­resolution,­there­were­still­no­errors­in­internode­segments­for­white­matter,­but­one­error­occurred­for­every­100­mm­of­internode­length­in­gray­matter.

We­ have­ demonstrated­ that­ the­ whole­ mouse­ brain­ can­ be­stained­and­embedded­in­a­way­that­is­suitable­for­ultramicro-tomy­and­electron­microscopic­imaging.­The­contrast­is­uniform­and­sufficient­ for­serial­block-face­ imaging.­Myelinated­axons­are­traceable­with­low­error­rates.­A­further­reduction­in­error­rate­is­expected­if­a­complete­mutual­assignment­of­paranodes­is­performed­on­the­basis­of­their­geometrical­parameters,­such­as­direction­and­diameter.

At­ similar­axon­calibers,­white-­and­gray-matter­ error­ rates­differ,­possibly­owing­to­differences­in­tortuosity:­in­gray­matter,­when­encountering­a­node­of­Ranvier,­the­matching­paranode­cannot­be­found­by­simply­continuing­along­the­same­direction.

The­ultrastructure­in­our­samples­is­preserved­well­enough­to­permit­the­recognition­of­unmyelinated­processes­and­synapses;­however,­low­membrane­contrast­and­excessive­cytosol­staining­prevent­the­reliable­tracing­of­unmyelinated­processes­over­long­distances,­making­the­wbPATCO­method­unsuitable­for­mapping­complete­neural­circuits.

At­this­point­it­is­unclear­what­fraction­of­the­projectome­can­be­recovered­by­following­only­myelinated­axons­because­not­all­long-range­axons­are­myelinated.­The­myelination­of­modula-tory­projections­is­sparse­(or­absent).­Estimates­from­subcortical­white­matter­in­the­adult­mouse­brain­suggest­that­about­a­third­of­projection­axons­are­myelinated23.­An­open­question­is­whether,­aside­ from­ dopaminergic­ projections,­ there­ are­ areas­ that­ are­connected­by­only­unmyelinated­fibers­or­whether­all­inter-areal­connections­contain­a­mixture­of­myelinated­and­unmyelinated­fibers.­If­the­latter­is­the­case,­a­map­of­all­myelinated­axons­in­a­single­mouse­brain­would­yield­a­relatively­complete­catalog­of­all­inter-areal­projections.

methodsMethods­and­any­associated­references­are­available­in­the­online­version­of­the­paper.

Note: Supplementary information is available in the online version of the paper.

acknoWledgmentsWe thank I. Sonntag and M. Helmstaedter for help with traceability analysis, S. Hillmer and U. Mersdorf for help with transmission electron microscopy, K. Briggman, S.K. Mikula and A. Scherbarth for help with staining procedures, B. Titze for help with conductive coating, J. Tritthardt for developing electronic circuits and M. Mueller for help with scanning electron microscopy–related software. We also thank the following student tracers: M. Diemer, C. Domnick, J. Hanne, P. Hofmann, A. Ivanova, H. Jakobi, A. Klein, J. Löffler, J. Nassel, J. Trendel and P. Weber. This work was supported by the Max Planck Society and the Deutsche Forschungsgemeinschaft (DFG).

author contributionsS.M. and W.D. designed the study and devised the analysis; S.M. carried out the experiments; J.B. and W.D. devised the aberration correction algorithm; S.M. analyzed the data; and S.M. and W.D. wrote the paper.

comPeting financial interestsThe authors declare competing financial interests: details are available in the online version of the paper.

Published online at http://www.nature.com/doifinder/�0.�0�8/nmeth.����. reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Kasthuri, N. & Lichtman, J.W. Nat. Methods �, 307–308 (2007).2. Bohland, J.W. et al. PLoS Comput. Biol. 5, e1000334 (2009).3. Tsai, P.S. et al. J. Neurosci. �9, 14553–14570 (2009).4. Sabatini, D.D., Bensch, K. & Barrnett, R.J. J. Cell Biol. �7, 19–58

(1963).5. Dalton, A.J., Kahler, H., Striebich, M.J. & Lloyd, B. J. Natl. Cancer Inst.

��, 439–461 (1950).6. Palay, S.L., McGee-Russell, S.M., Gordon, S. Jr. & Grillo, M.A. J. Cell Biol.

��, 385–410 (1962).7. Seligman, A.M., Wasserkrug, H.L. & Hanker, J.S. J. Cell Biol. �0, 424–432

(1966).8. Li, A. et al. Science ��0, 1404–1408 (2010).9. Chung, J.R. et al. Front. Neuroinform. 5, 29 (2011).10. Dempster, W.T. Am. J. Anat. �07, 59–72 (1960).11. Hagström, L. & Bahr, G.F. Histochem. Cell Biol. �, 1–4 (1960).12. Bozzola, J.J. & Russell, L.D. Electron Microscopy: Principles and Techniques

for Biologists. (Jones & Bartlett Learning, 1999).13. Hayat, M.A. Principles and Techniques of Electron Microscopy: Biological

Applications, 4th edn. (Cambridge University Press, 2000).14. Van Holde, K.E. J. Phys. Chem. 6�, 1574–1577 (1959).15. Bahr, G.F. Exp. Cell Res. 7, 457–479 (1954).16. Guha, P.C. & De, S.C. J. Chem. Soc. Trans. ��5, 1215–1218 (1924).17. De Bruijn, W.C. in Proc. 4th Eur. Reg. Conf. Electron Microsc. (ed.

Bocciarelli, D.S.) 11–65 (Rome: Tipografia Polyglotta Vaticana, 1968).18. Karnovsky, M.J. in Proc. 11th Meet. Am. Soc. Cell Biol. 146 (1971).19. Seligman, A.M., Hanker, J.S., Wasserkrug, H., Dmochowski, H. & Katzoff, L.

J. Histochem. Cytochem. ��, 629–639 (1965).20. Quarles, R.H., Macklin, W.B. & Morell, P. in Basic Neurochemistry:

Molecular, Cellular, and Medical Aspects, 7th edn. (eds. Siegel, G., Albers, R.W., Brady, S. & Price, D.) Ch. 4, 51–72 (Academic, 2006).

21. Spurr, A.R. J. Ultrastruct. Res. �6, 31–43 (1969).22. Kushida, H. J. Electron Microsc. (Tokyo) ��, 197 (1974).23. Partadiredja, G., Miller, R. & Oorschot, D.E. J. Neurocytol. ��, 1165–1179

(2003).24. Sturrock, R.R. Neuropathol. Appl. Neurobiol. 6, 415–420 (1980).

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature methodsdoi:10.1038/nmeth.2213

online methodsAnimals used.­A­total­of­45­adult­(10-­to­36-week-old)­wild-type­(C57BL/6)­male­mice,­maintained­at­a­12-h­dark-light­cycle­under­standard­chow,­were­used.­Out­of­these,­three­(10-week-old)­mice­were­used­for­whole-brain­block-face­imaging­and­traceability­analysis.­The­remaining­animals­were­used­for­general­staining-protocol­testing­and­optimization.

Mice­were­anesthetized­with­isoflurane­(Baxter)­inhalation­and­killed­by­transcardially­perfused­fixative.­One­brain­(WB1405)­was­used­for­acquiring­whole-brain­coronal­block-face­mosaics­at­80-nm­pixel­size­for­the­purpose­of­assessing­staining­and­embed-ding­quality­and­uniformity.­The­remaining­two­brains­(WB1406­and­WB1407)­were­used­for­extracting­small­(<1­mm3)­regions­of­ interest­ (ROIs)­ for­ SBEM­ imaging25.­ All­ procedures­ were­approved­by­the­local­animal­care­and­use­committee­and­were­in­accordance­with­the­laws­of­animal­experimentation­issued­by­the­German­federal­government.

Sample preparation.­The­optimized­whole-brain­staining­pro-tocol,­wbPATCO,­was­performed­as­follows­(the­methods­used­for­exploratory­preparations7,19,26–29­are­given­in­Supplementary Table 1).­After­anesthesia,­the­mouse­was­placed­supine­and­a­horizontal­cut­was­made­into­the­upper­abdomen.­First­the­dia-phragm­was­cut,­then­the­ribs­were­cut­laterally­and­the­chest­flap­was­folded­back­using­a­hemostat.­The­heart­was­exposed­and­held­in­place­using­serrated­curved­forceps­(Moria­MC31,­Fine­Science­Tools)­while­a­22-gauge­cannula­was­carefully­inserted­into­the­left­ventricle­and­secured­using­an­aneurysm­clamp­(Fine­Science­Tools).­The­right­auricle­was­pierced,­and­the­mouse­was­perfused­transcardially­at­a­rate­of­0.5­ml/s­with­40­ml­of­a­0.418­M­(10%)­solution­of­acrolein30­(Sigma)­in­0.1­M­sodium­cacodylate­buffer­(Serva),­pH-adjusted­to­7.4­with­HCl­and­chilled­to­2­°C.­We­chose­acrolein­as­the­fixative­because­with­glutaraldehyde,­but­not­with­acrolein,­we­occasionally­saw­artifacts­in­the­form­of­ densely­ stained­ spots­ inside­ the­ myelin­ sheath.­ The­ mouse­was­subsequently­decapitated,­the­muscle­and­skin­tissue­around­the­cranium­and­the­cervical­vertebrae­were­removed­with­small­scissors­ (Fine­Science­Tools),­and,­ starting­ from­the­vertebrae­and­working­rostrally,­the­vertebrae­and­small­bone­fragments­were­carefully­removed­using­#2­forceps­until­ the­entire­brain­was­exposed­dorsally­and­laterally.­During­this­procedure,­a­trans-fer­pipet­was­periodically­used­to­drop­fixative­solution­onto­the­brain­to­prevent­dehydration.­The­cranial­nerves­were­cut­with­small­scissors­before­we­removed­the­brain­and­placed­it­into­an­acrolein­solution­for­post-fixation­(48­h­at­2­°C)­followed­by­five­8-h­rinses­with­50­ml­of­0.1­M­cacodylate­buffer.­Brains­were­then­either­used­for­whole-brain­staining,­hemispherectomized­with­a­scalpel­blade­and­used­for­hemisphere­staining­or­diced­into­small­cubes­(<1­mm3)­for­small-sample­staining.

For­all­immersion­steps­the­samples­were­kept­in­closed­cen-trifuge­tubes­(Sarstedt),­which­were­discarded­after­each­staining­step.­Small­samples­were­incubated­in­1­ml­of­staining­solution­for­90­min­in­2-ml­centrifuge­tubes­(Sarstedt)­on­a­rotator.­Rinse­cycles­consisted­of­three­10-min­immersions­in­2­ml­of­0.1­M­cacodylate,­pH­7.4,­at­20­°C.­Embedding­of­small­samples­was­performed­as­ described­ below­ except­ that­ volumes­ were­ 2­ ml­ and­ dura-tions­were­30­min­each.­Hemispheres­were­incubated­in­12.5­ml­­of­ staining­solution­ for­24­h­ in­15-ml­centrifuge­ tubes­during­gentle­agitation.­Rinse­cycles­between­successive­staining­steps­

consisted­of­four­4-h­immersions­in­25­ml­of­0.1­M­cacodylate,­pH­7.4,­at­20­°C.­Embedding­of­hemispheres­was­performed­as­described­below­except­that­volumes­were­25­ml­and­durations­were­6­h­each.­Whole­brains­were­incubated­in­25­ml­of­staining­solution­for­48­h­in­50-ml­centrifuge­tubes­during­gentle­agitation.­Rinse­cycles­between­successive­staining­steps­consisted­of­four­8-h­immersions­in­50­ml­of­0.1­M­cacodylate,­pH­7.4,­at­20­°C.

For­the­wbPATCO­stain,­the­whole­brain­was­first­placed­into­a­ solution­ containing­ 90­ mM­ periodic­ acid­ (Sigma)­ in­ 0.1­ M­­cacodylate­buffer­(Serva)­at­pH­7.4,­at­2­°C­for­48­h.­It­was­then­rinsed­four­times,­immersed­for­48­h­in­a­solution­of­100­mM­­TCH­ (Sigma)­ in­ 0.1­ M­ cacodylate­ buffer­ at­ pH­ 7.4,­ held­ at­­50­ °C,­ rinsed­ four­ times­and­ finally­ immersed­ for­48­h­ into­a­solution­of­80­mM­(2%)­OsO4­(Electron­Microscopy­Sciences)­in­water­at­20­°C.

For­ embedding,­ brains­ were­ first­ dehydrated­ by­ immersing­them­sequentially­for­12­h­each­in­a­series­of­acetone-water­mix-tures­ (50%,­ 75%,­ 100%­ and­ again­ 100%­ acetone)­ followed­ by­epoxy-monomer­ infiltration­by­ immersion­(12­h,­25­ml­each)­into­mixtures­of­acetone­and­epoxy­resin­(50%,­100%­and­100%­epoxy)­with­gentle­agitation.­All­dehydration­and­infiltration­steps­were­performed­at­20­°C.­Epoxy­resins­were­either­Spurr’s21,­using­ERL-4221­as­a­replacement­for­ERL-4206,­or­Quetol22,­and­con-sisted­of­the­following­formulations:­for­Spurr’s,­10­g­ERL-4221­(Serva),­5­g­DER­736­(Serva),­26­g­NSA­(Electron­Microscopy­Sciences)­and­0.2­g­DMAE­(Serva);­for­Quetol,­10­g­Quetol­651­(Electron­Microscopy­Sciences),­20­g­NSA,­0.15­g­DMAE.­The­brain­was­then­placed­into­a­custom-made­rectangular­silicone­mold­of­dimensions­6­mm­×­8.5­mm­×­18­mm­or­7­mm­×­10­mm­×­­18­mm­and­cured­at­70­°C­for­at­least­24­h.­We­used­Spurr’s­for­embedding­WB1406­and­Quetol­for­WB1405­and­WB1407.

Whole-brain block-face imaging.­One­sample­(WB1405)­was­used­ for­ acquiring­ whole-brain­ coronal­ block-face­ mosaics­ at­­80-nm­pixel­size­to­assess­staining­and­ultrastructural­preserva-tion.­Excess­epoxy­and­ the­anterior­portion­of­ the­brain­were­removed­using­an­EM­sample­trimmer­(EM­Trim,­Leica).­The­cross-sectional­surface­was­then­smoothed­using­an­8-mm-wide­diamond­knife­(JumboHisto,­Diatome)­at­a­3.5°–clearance-angle­setting­on­the­microtome,­advances­of­70–100­nm­between­cuts­and­1.0­mm/s­cutting­speed.­The­smoothed­block­face­was­then­coated­ with­ a­ 5-nm-thick­ layer­ of­ platinum-carbon­ using­ an­­electron-beam­ evaporator­ (MED­ 20,­ Bal-Tec)­ to­ allow­ charge­­dissipation­during­imaging.

The­smoothed­cross-sectional­surface­was­not­planar­for­two­reasons.­First,­the­sample­moves­in­an­arc­on­the­ultramicrotome,­which­leads­to­a­cylindrical­shape;­and­second,­the­sample­con-tinues­to­shrink­under­vacuum­slightly­but­unevenly­after­cutting­and­coating.­This­means­ the­ focus­has­ to­be­corrected­during­mosaic­ acquisition.­ To­ provide­ approximately­ correct­ starting­values­for­the­autofocus­algorithm­used­during­the­actual­acqui-sition,­we­determined­the­correct­working­distance­(WD)­for­all­points­lying­on­an­evenly­spaced­(1-mm)­grid­by­manual­focusing,­and­we­then­interpolated­between­points­with­cubic­splines.­Next,­we­automatically­acquired­all­tiles­needed­for­the­mosaic­using­a­custom-written­Matlab­(MathWorks)­script­that­controlled­the­scanning­electron­microscopy­(SEM,­Merlin,­Zeiss)­and­the­exter-nal­scan­generator­(Atlas,­Fibics/Zeiss).­Prior­to­the­acquisition­of­each­tile,­the­script,­after­determining­which­part­of­the­tile­had­

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature methods doi:10.1038/nmeth.2213

the­best­contrast,­called­a­custom­aberration-correction­proce-dure31­for­both­WD­and­astigmatism.

Images­were­acquired­using­5.0-keV­landing­energy,­1-nA­beam­current,­2-µs­pixel­dwell­time­and­6.0-mm­WD.­Secondary­elec-trons­were­detected­using­the­built-in­Everhart-Thornley­(ET)­detector,­with­the­grid­voltage­set­to­400­V.­To­prevent­secondary­electrons­from­being­drawn­into­the­column,­we­inserted­a­cus-tom­aperture­(diameter­of­2­mm)­into­the­end­cap.­The­tile­size­was­4,000­×­4,000­pixels,­which­results­in­a­330-µm­field-of-view.­Substantially­larger­tiles­could­not­be­used­without­incurring­a­loss­of­resolution­at­ the­ image­corners.­We­used­a­5%­overlap­between­tiles,­which­was­enough­to­calculate­tile­offsets­with­sub-pixel­precision­by­cross-correlation.­The­offsets­were­then­used­in­a­global­alignment­and­stitching­procedure­based­on­iteratively­minimizing­the­weighted­(by­the­reciprocal­residuals)­sum­of­the­squared­residual­shifts32,33.­The­data­set­comprised­975­tiles­that­were­combined­into­a­final­mosaic­153,000­×­98,000­pixels­in­size.­All­image­processing­was­performed­using­custom­Matlab­code.­The­final­mosaic­is­available­for­online­viewing­(http://homes.mpimf-heidelberg.mpg.de/~smikula/blockface/)­using­a­custom­javascript/dHTML­interface34.

TEM imaging.­Myelin-sheath­and­synaptic­ultrastructure­were­assessed­at­high­resolution­by­transmission­electron­microscopy­(TEM).­ For­ this,­ a­ small­ cube­ from­ the­ dorsolateral­ striatum­was­dissected­out­from­an­embedded­whole-brain­sample­with­a­scalpel­blade­and­mounted­on­a­metal­stub.­Then,­70-nm-thick­sections­ were­ cut­ using­ an­ ultramicrotome­ (Ultracut,­ Leica),­mounted­on­Formvar-coated­copper­TEM­grids­(Plano­GmbH),­post-stained­in­2%­(50­mM)­aqueous­uranyl­acetate26­and­lead­citrate35­ for­10­and­5­min­each,­respectively,­and­imaged­on­a­transmission­electron­microscope­(JEM­1400,­JEOL)­at­80-keV­electron­energy,­with­images­recorded­on­a­CCD­camera.

Serial block-face imaging.­For­SBEM­imaging,­a­total­of­eight­ROIs­ were­ selected­ from­ WB1406­ and­ WB1407.­ The­ random­selection­ of­ a­ white-­ or­ gray-matter­ ROI­ was­ performed­ by­generating­pseudo-random­3-tuples­until­ the­bregma-referred­coordinates­corresponding­to­the­3-tuple­pointed­to,­respectively,­a­white-­or­gray-matter­location­inside­of­the­brain,­as­determined­by­consulting­the­Paxinos­mouse­atlas36.­This­was­repeated­until­coordinates­ for­all­ROIs­were­obtained.­We­then­ trimmed­the­sample­coronally,­starting­from­the­anterior­end­of­the­brain­to­the­approximate­location­of­the­most­anterior­ROI,­smoothed­and­imaged­the­block­face.­Once­the­ROI­had­been­located­in­the­SEM­using­visual­cues,­it­was­repeatedly­imaged­with­high­beam­current­to­mark­the­ROI­on­the­block­face.­This­mark­was­then­used­to­visually­identify­the­ROI­in­a­dissecting­scope,­and­a­small­volume­of­sample­at­the­ROI­was­cut­out­with­a­scalpel­blade,­mounted­on­a­stub­and­imaged­with­SBEM­on­a­QuantaFEG­200­(FEI).­The­whole-brain­ sample­was­ then­retrimmed­ to­ the­next­ROI­and­the­process­was­repeated.­For­each­ROI,­a­data­set­of­approxi-mately­(80­µm)3­was­acquired­at­40-nm­isotropic­voxel­size.­SBEM­imaging­parameters­were­2.8-keV­primary­beam­electron­energy,­­155-pA­beam­current,­8-µs­pixel­dwell­ time,­6.0-mm­WD­and­20-Pa­water­vapor­pressure.­Backscattered­electrons­were­detected­

using­ a­ custom­ silicon-diode­ detector­ (AXUV,­ International­Radiation­Detectors)­combined­with­a­custom­current­amplifier.

Traceability analysis.­For­axon­traceability­analysis,­SBEM­data­sets­(2,048­×­1,768­×­1,800­voxels),­originally­acquired­at­40­nm,­were­also­prepared­as­data­sets­down-sampled­to­80­and­120­nm­by­selecting,­respectively,­every­second­or­third­voxel­in­x,­y­and­z.­­Using­ a­ custom­ Matlab­ script,­ data­ sets­ were­ split­ into­ cubes­­(128­ ×­ 128­ ×­ 128­ voxels)­ as­ needed­ for­ the­ Knossos­ software­(www.knossostool.org/).­Using­Knossos,­50­axons­were­randomly­chosen­among­those­passing­through­the­central­region­of­the­­40-nm-resolution­data­set.­Each­of­those­axons­was­independently­traced­by­five­undergraduate­tracers­at­each­resolution.

To­quantify­myelinated-axon­traceability,­tracing­errors­were­detected­by­manually­comparing­each­ individual­axon­tracing­to­the­40-nm­consensus­tracing­of­the­same­axon.­The­consen-sus­tracing­was­the­largest­subset­of­tracings­(at­least­two)­that,­­in­both­tracing­directions,­agreed­with­each­other­until­they­either­reached­the­surface­of­the­SBEM­data­set­or­a­node­of­Ranvier­located­within­10­µm­of­the­nearest­data­set­surface.­The­latter­condition­accounted­mainly­for­cases­in­which­a­node­of­Ranvier­extended­beyond­the­data­set.­Because­the­unmyelinated­stretch­of­axon­at­a­node­had­low­contrast,­tracing­relied­heavily­on­find-ing­the­matching­paranode.­If­no­such­node­could­be­found­or­if­no­two­tracers­agreed­on­which­paranode­to­select,­the­correct­paranode­was­likely­outside­of­the­data­set.­Other­possibilities­are­that­the­tracing­had­reached­a­terminal­branch­or­the­initial­segment,­both­of­which­are­unmyelinated.

Although­it­ is­theoretically­possible­to­end­up­with­two­sets­of­ two­ tracings­ for­ each­ axon­ that­ are­ consistent­ within­ but­­inconsistent­between­sets,­this­was­never­observed.­Errors­were­classified­by­whether­or­not­they­occurred­at­nodes­of­Ranvier.­Consensus­ tracings­ of­ collaterals­ were­ not­ included­ in­ the­­analysis­because­they­were­ infrequent­(3.5%­of­total­ tracings),­occurring­mainly­in­V1.­We­computed­error­rates­per­millimeter­by­dividing­the­total­number­of­errors­by­the­combined­total­path­lengths­for­the­respective­axons,­using­the­median­path­lengths­of­their­respective­40­nm­tracings.­The­axon­diameter­was­computed­using­the­mean­of­ten­measurements­randomly­selected­along­the­internodes­of­the­axon.

25. Denk, W. & Horstmann, H. PLoS Biol. �, e329 (2004).26. Terzakis, J.A. J. Ultrastruct. Res. ��, 168–184 (1968).27. Walton, J. J. Histochem. Cytochem. �7, 1337–1342 (1979).28. Willingham, M.C. & Rutherford, A.V. J. Histochem. Cytochem. ��, 455–460

(1984).29. Deerinck, T.J. et al. Microsc. Microanal. �6 (suppl. 2), 1138–1139 (2010).30. Luft, J.H. Anat. Rec. ���, 305 (1959).31. Binding, J., Mikula, S. & Denk, W. Microsc. Microanal. (in the press).32. Holland, P.W. & Welsch, R.E. Commun. Stat. Theory Methods 6, 813–827

(1977).33. Baumann, M. et al. in Proc. 6th PIMS Ind. Problem Solving Workshop

(ed. Macki, J.) Ch. 1, 1–25 (PIMS, 2002).34. Mikula, S., Trotts, I., Stone, J.M. & Jones, E.G. Neuroimage �5, 9–15

(2007).35. Reynolds, E.S. J. Cell Biol. �7, 208–212 (1963).36. Franklin, K.B.J. & Paxinos, G. The Mouse Brain in Stereotaxic Coordinates

3rd edn. (Academic, 2008).

Staining and Embedding the Whole Mouse Brain for

Electron Microscopy

Shawn Mikula, Jonas Binding, Winfried Denk

Supplementary Item Title or Caption

Supplementary Figure 1 Comparison of white matter in hemispheres stained using different long-

incubation (24 hr) protocols

Supplementary Table 1 Staining protocols

Supplementary Table 2 Outcomes for various staining protocols

Supplementary Table 3 Nodal and internodal error rates for white and gray matter at different

resolutions

Supplementary Protocol Step-by-step whole-brain sample preparation protocol

Nature Methods: doi:10.1038/nmeth.2213

Supplementary Figure 1

Supplementary Figure 1. Comparison of white matter in hemispheres stained using different long-incubation (24 hr) protocols.

(a) Osmium tetroxide (Os) incubation. (b) Osmium tetroxide incubation followed by uranyl-acetate enhancement (Os->UA). (c)

Osmium tetroxide, followed by thiocarbohydrazide, and then by a second osmium tetroxide incubation (OTO). (d) Periodic acid,

followed by thiocarbohydrazide, and then by osmium tetroxide (wbPATCO). See text for details. Insets: magnified sub-regions.

Nature Methods: doi:10.1038/nmeth.2213

Supplementary Table 1

incubation steps references

staining protocol primary secondary tertiary

solution temp

(°C)

solution temp

(°C)

solution temp

(°C)

Os 80 mM OsO4 20 - - - -

Os -> PbAsp 80 mM OsO4 20 lead aspartate 20 - - 34

Os -> UA 80 mM OsO4 20 50 mM

uranyl acetate

20 - - 31

wbPATCO 90 mM periodic acid,

0.1 M cacodylate, pH

7.4

2 100 mM

TCH, 0.1M

cacodylate,

pH 7.4

50 80 mM

OsO4

20 19

OTO 80 mM OsO4 20 100 mM TCH 50 80 mM

OsO4

20 7

rOTO 80 mM OsO4,

72 mM ferrocyanide,

0.1 M cacodylate, pH

7.4

20 100 mM TCH 50 80 mM

OsO4

20 35,36

Supplementary Table 1. Staining protocols. All staining steps use ddH2O as solvent.

Nature Methods: doi:10.1038/nmeth.2213

Supplementary Table 2

small samples, 90 min incubations hemispheres or whole brains, 24-48 hr incubations

staining

protocol

membrane

contrast

myelin

contrast

staining

uniformity

tissue

preservation

membrane

contrast

myelin

contrast

staining

uniformity

tissue

preservation

Os + + +++ +++ + + +++ ++

Os ->

PbAsp

++ ++ +++ +++ ++ ++ ++ ++

Os ->

UA

++ ++ +++ +++ ++ ++ ++ ++

PATCO ++ +++ +++ +++ ++ +++ +++ ++

OTO +++ +++ +++ +++ - - + -

rOTO +++ +++ +++ +++ + + - -

Supplementary Table 2. Outcomes for various staining protocols, ranging from very strong (+++) to very weak (-).

Nature Methods: doi:10.1038/nmeth.2213

Supplementary Table 3

nodes of Ranvier inter-node segments total

errors/mm

resolution

(nm)

errors/node (%) errors/mm errors/segment (%) errors/mm

wh

ite

mat

ter

40 1.25 0.13 0 0 0.13

80 3.15 0.32 0 0 0.32

120 5.87 0.603 0.75 0.077 0.68

gra

y m

atte

r

40 3.36 0.39 0 0 0.39

80 9.86 1.14 0.083 0.0096 1.15

120 15.96 1.86 2.74 0.33 2.19

Supplementary Table 3. Nodal and internodal error rates for white and gray matter at different resolutions.

Nature Methods: doi:10.1038/nmeth.2213

Supplementary Protocol

Materials

Reagents

Sample Adult whole mouse brain.

Staining and fixation Isoflurane (Baxter)

Cacodylate buffer (Serva)

Acrolein (Sigma)

Periodic acid (Sigma)

Osmium tetroxide (Serva)

TCH (Sigma, cat. no. T-2137)

Dehydration and infiltration Water (double-distilled)

Acetone (Sigma)

Embedding o ERL-4221 (Serva)

o DER 736 (Serva)

o NSA (Electron Microscopy Sciences)

o DMAE (Serva)

Or

o Quetol 651 (Electron Microscopy Sciences)

o NSA (Electron Microscopy Sciences)

o DMAE (Serva)

Equipment

Perfusion #2 forceps (Fine Science Tools)

Coarse and fine scissors (Fine Science Tools)

Nitrile gloves

Plastic tray (generic)

Nature Methods: doi:10.1038/nmeth.2213

Hemostat (Fine Science Tools)

Serrated curved forceps (Moria MC31, Fine Science Tools)

22-gauge cannula with syringe attachment (use a needle and remove the sharp tip)

Aneurysm clamp (Fine Science Tools)

40 ml syringe

Transfer pipet (generic)

Razor blade (VWR)

50 ml centrifuge tubes

Lab rocker (VWR)

EM processing and embedding Centrifuge tubes (Sarstedt)

Embedding oven

Custom made silicone molds with dimensions 7 mm x 10 mm x 18 mm

Block trimming Leica EM trimmer

Reagent setup

Buffer Add 4.28 g sodium cacodylate per 50 ml water, and bring volume up to 100 ml with water, to get a final

concentration of 0.2 M sodium cacodylate buffer. Adjust the pH to 7.4 with hydrochloric acid and store in

the fridge. Add equal parts water to make up 0.1 M buffer as needed.

Fixative Add 20% acrolein to 50 ml of 0.2 M sodium cacodylate buffer and bring volume up to 100 ml with water.

Osmium tetroxide

Periodic acid Add 90 mM periodic acid to 0.1 M cacodylate buffer, as prepared above.

Thiocarbohydrazide Add 100 mM thiocarbohydrazide to 0.1M cacodylate buffer, as prepared above.

Procedure

Fixation by perfusion -- Time: < 5 min After anesthesia, place the mouse supine in a plastic tray and make a horizontal cut into the upper

abdomen. Cut into the diaphragm, then cut the ribs laterally. Fold the chest flap back using a hemostat.

The heart should then be exposed with visible auricles. Hold the heart in place gently using serrated,

curved forceps. Insert a 22-gauge cannula attached to the syringe into the left ventricle and secure it in

Nature Methods: doi:10.1038/nmeth.2213

place using an aneurysm clamp. Pierce or cut the right auricle using a needle or a pair of fine scissors.

Begin perfusing the mouse using the syringe at a rate of 0.5 ml / sec with 40 ml of a 0.418M (10%)

solution of acrolein in 0.1M sodiuim cacodylate buffer. When the fixative has all flowed through the

mouse, carefully dissect out the brain using scissors and forceps. Proceed to the post-fixation step.

Dissection – Time: 10 min Decapitate the mouse and remove the muscle and skin tissue around the cranium and cervical vertebrae

with small scissors and forceps. Remove skull bone with #2 forceps beginning at the back of the skull and

moving rostrally until the entire surface of the brain is exposed. Periodically drop fixative solution onto

the surface of the brain using a transfer pipet to prevent tissue dehydration and to facilitate quick fixation.

Invert the brain and, using small scissors, cut the cranial nerves to separate the brain from the remaining

skull tissue.

Post-fixation and rinses -- Time: 88 hrs Place the brain into 50 ml of acrolein solution (the same fixative used in the transcardial perfusion) for

post-fixation for 48 hrs at 2°C. Rinse the brain five times for eight hours using 50 ml of 0.1M cacodylate

buffer. If the brain is to be used for small samples, dice the brain into small cubes using a razor blade

during the rinses. For the procedure using small samples, shorten the staining times according to the

manuscript. Following is the procedure for whole brain staining.

Staining– Time: 208 hrs (between 8 and 9 days) 1. Place the whole mouse brain in a 50 ml tube containing 25 ml of 90 mM periodic acid in 0.1M

cacodylate buffer for 48 hrs at room temperature. Gently agitate by placing the tube on a

laboratory rocker on a slow speed.

2. Rinse the brain four times for eight hours each in 50ml of caocdylate buffer at room temperature,

as prepared above.

3. Transfer the brain to a fresh container and, for 48 hrs, immerse it in a solution of 100 mM

thiocarbohydrazide in 0.1 M cacodylate buffer at 50°C.

4. Rinse the brain again four times in buffer for eight hours each and change containers.

5. Immerse the brain in a solution of 2% osmium tetroxide in water at room temperature for 48 hrs.

6. Proceed to the embedding step.

Embedding – Time: 120 hrs (5 days)

Dehydration Step the tissue through a series of water / acetone mixtures (50%, 75%, 100%, 100% acetone) of 50 ml

each, with each step lasting 12 hrs at room temperature.

Infiltration Infiltrate the epoxy resin with 12 hr steps of 25 ml each (50%, 100%, 100% epoxy) with gentle agitation

on a lab rocker at room temperature.

For Spurr’s resin:

10g ERL-4221

5g DER 736

26g NSA

Nature Methods: doi:10.1038/nmeth.2213

0.2g DMAE

For Quetol resin:

10g Quetol 651

20g NSA

0.15g DMAE

Place the brain into a custom made silicone mold and cure in a laboratory oven at 70°C for at least 24 hrs.

When the block is hard, trim it according to desired dimensions with the EM Leica trimmer.

Nature Methods: doi:10.1038/nmeth.2213