A NEW STAINING METHOD (OTO) FOR ENHANCING CONTRAST OF LIPID-CONTAINING MEMBRANES AND DROPLETS IN OSMIUM TETROXIDE-¥IXED TISSUE WITH OSMIOPHILIC THIOCARBOHYDRAZIDE (TCH)

ARNOLD M. SELIGMAN, HANNAH L. WASSERKRUG, and JACOB S. HANKER. From the Departments of Surgery, Sinai Hospital of Baltimore, Inc., and The Johns Hopkins University School of Medicine, Baltimore, Maryland

Although the introduction of fixation of tissue with grids, I without a supporting membrane, to a 1 osmium tetroxide by Palade (8) ushered in the hot aqueous solution of TCH 2 for 1 hr at 50°C, ~ modern era of electron microscopy of biological followed by several washes with hot water (initial materials, a need for greater contrast and greater temperature 50°C) for 10 to 15 min to remove un- resolution in visualizing membranous structures bound TCH. The sections are then exposed to Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021 has stimulated the introduction of methods for osmium tetroxide again, resulting in further depo- staining with heavy metal salts, in order to enhance sition of osmium. The best results are obtained by the delineation of the fine architecture of cells. exposing the sections to osmium tetroxide vapor 4 This is especially necessary because of the poor for 1 hr in a closed vessel suspended in a water contrast of osmium tetroxide-fixed tissue when the bath at 60°C as described earlier (11), or very epoxy resins are used as embedding materials (9). good results may be obtained with perhaps a little These methods have relied upon specific affinities less contrast and considerably less cost by treating of heavy metal salts for various macromolecular the grids with a 2 % solution of osmium tetroxide components of the cell such as proteins, poly- I Any inert metal may be used. When copper grids saccharides, nucleoproteins, or lipoproteins (3, 9, are used, TCH binds to surface oxides resulting, on 17, 18). Such staining techniques do not neces- exposure to osmium tetroxide, in blackening of the sarily enhance the staining of the same components grid and in a tendency for the black deposits to scale of the cell originally stained by the osmium tetrox- and contaminate the specimens. This could be only ide, although interesting and important staining partly eliminated by prior washing of copper grids effects have been produced thereby. with acetic acid. Gold, stainless steel, or nickel grids In the course of developing cytochemical meth- were most useful. ods for electron microscopy, based upon the prin- 2 Thiocarbohydrazide may be purchased from Distil- ciple of designing histochemical reagents that lation Products Industries, Rochester, New York, or yield osmiophilic end products (6, 10, l 1), we have Polysciences, Inc., Rydal, Pennsylvania. 8 This procedure and the washing procedure were discovered a new way to enhance especially the conducted in a porcelain spot plate. Each depression contrast of the lipid components of the cells stained had a capacity of 1 ml and held 1 to 2 grids. During by osmium tetroxide during the initial fixation. On the incubation at 50°C, the spot plate on a tray was treatment with an excess of thiocarbohydrazide covered with a staining dish to cut down evaporation. (H2NNHCSNHNH2, TCH), one end of the mole- 4 It was found convenient to place the grids in an open cule attaches to the osmium in the tissue and, LKB grid box which was placed on a rack in the vessel when this is followed by exposure to osmium containing osmium tetroxide vapor. The plastic of tetroxide, more osmium is bound to these sites. the grid box blackened but no deterioration or scaling The new method (OTO) utilizes this bridging occurred on repeated use. After exposure to osmium phenomenon (5) and, although contrast is in- tetroxide, the grids were transferred to a fresh grid box and exposed to air for several minutes to rid creased thereby in all osmiophilic components of them of excess osmium tetroxide. They were exam- tissue, greatest enhancement of contrast is pro- ined with an RCA EMU-3H at 50 kv with a 45 /~ duced in tissue components holding the most objective aperture. Electron micrographs were taken osmium, i.e., lipid. on medium contrast plates with 2 sec' exposure and The OTO method consists of exposing ultrathin a reading of 20 on the photometer. Prints were pre- sections of Araldite-embedded, osmium tetroxide- pared by Mr. Michael Friedman on No. 4 contrast fixed tissue (8) on gold, stainless steel, or nickel paper at constant background density.

4~ B R I E F N 0 T E S (unbuffered) at 50°C for 1 hr followed by a brief is not possible at present, but results of analyses of wash. Although silver ion will bridge to tissue- reaction products of TCH and OsO4 will be pub- bound osmium through TCH, the deposits at high lished later. It is possible that osmium tetroxide magnification were not amorphous as with osmium that is reduced to insoluble lower Os oxides by tetroxide, but were granular. Lead citrate (0.5%), strong reducing groups of some components of however, could be used in place of osmium tetrox- tissue may not react with TCH. Although the pre- ide in the T-O procedure. Membranes were well cise structures of these reaction products are un- stained although the contrast wa~ not as good as known, we can probably safely assume that os- with osmium tetroxide. mium tetroxide may have a different structure Hypothetical formulation of the reaction of ex- when attached to lipid (Fig. 1) from that produced cess TCH with lipid fixed in osmium tetroxide, when osmium tetroxide has reacted with protein, based upon Criegee's first reaction product (un- nucleoprotein, or polysaccharide. Since fixation of stable) in the oxidation of a double bond by OsO4 the latter macromolecules in osmium tetroxide is (2, 16), is shown in Fig. 1. Since sections at this much less in degree than that of lipid (1, 4, 16), stage have no more density than the original os- apparent intensification by the OTO method of mium tetroxide-fixed sections, it is evident that the osmium tetroxide stain that is attached to lipid Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021 subsequent treatment with osmium tetroxide re- structures suggests that predominantly quantita- suits in attachment of more osmium to the un- tive factors are involved. This was shown in model reacted end of bound TCH to yield an increase in experiments with filter paper impregnated with contrast (Fig. 2). The OTO method results in samples of unsaturated lipid, protein, nucleic acids, striking increase in density of the lipid component and carbohydrate (5). Fig. 2 diagrammatically of membranous structures such as cristae and outer shows how a greater amount of osmium tetroxide limiting membranes of mitochondria, plasma bound to lipid than to protein could result in an membranes, nuclear envelopes, endoplasmic retic- increase of the total amount of osmium on lipid by ulum, and lipid inclusion bodies in mitochondria, TCH bridging (5) and make a poorly delineated droplets, or vesicles. Examples of these are shown thick double membrane appear as two distinct thin in Figs. 3 to l0 from kidney, heart, brain, and membranes or enhance and sharpen the appear- small intestine of the rat. In each plate both os- ance of this double membrane. There is an ap- mium tetroxide-fixed tissue (top) and OTO- parent disagreement concerning the precise treated tissue (bottom) are shown. The observa- localization of osmium in the membranes of os- tion of increased density and increased resolution mium tetroxide-fixed tissue. It has been assumed of membranes by making them more distinctly (15) that the hydrophilic end of the lipid com- visible is very significant. The enhanced staining ponent of the cell membrane is the site of the makes the membranes appear thinner as well as accumulation of osmium, as opposed to the older concept that accumulation is in the unsaturated shcw greater contrast (Fig. 4). These results could fatty chain of the lipid (1). Recent work by be due to reaction of the TCH with osmium Stoeckenius and Mahr (4), which attempts to tetroxide that is bound to lipid and is still capable reconcile this controversy, suggests that primary of reaction with TCH (Fig. 1, formula I) to give reaction of osmium tetroxide does in fact occur hyFothetical formula II, which could be reduced with unsaturated lipid and, after reduction, shifts by excess TCH to hypothetical formulas III or IV. to the polar end of the fatty chain. Whether this TCH could also react similarly with diesters and shift will be established as true or not is of no im- polymeric esters of lipid-bound osmium cited by portance in understanding the increase in con- Stoeckenius and Mahr (16). Furthermore, we can trast produced in membranes with the OTO assume that all the reactive groups of TCH (which method and studied at the magnifications pre- is used in excess) are not consumed in this first re- sented here. Subsequent study with our method action because the TCH bound to osmium at higher magnification may help resolve the tetroxide-fixed tissue (formulas II, III, or IV) is controversy. Inasmuch as the evidence for capable of reducing and binding additional os- accumulation of osmium at the polar end of the mium tetroxide, the last step in producing the lipid is indirect (15), we prefer at present not to OTO phenomenon of contrast enhancement. The specify the exact site of attachment to lipid as in formulation of the last step in the OTO procedure Fig. 2.

BRIEF NOTES 425

I 1 HC - 0 0 Hc + 050 4 ll II Os FiC HC__ O \O 1 I I

I HC -0\ 0 IS,// HC -O \\ NNHCSNHNH 2 I II

TCH Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021

1 HC- O NNHCSNHNH 2 HC-0 OS 41 NNHCSNHNH2 HI- o \ NNHCSNHNH 2 HC- O 1 III IV

FIGURE 2 Diagrammatic representation of the OTO method used on a double membrane (Fig . 2 a) com- FIGURE 1 Hypothetical formulation of the reaction posed of protein (P) and lipid (L) . In Fig . 2 b, the of excess TCH with lipid fixed in osmium tetroxide, difference in the degree of osmication of the two com- based upon Criegee's first reaction product in the ponents (P and L) in osmium tetroxide is represented oxidation of a double bond by OsO4 (2, 16) . The hy- by a factor of 1 :2, although the actual ratio would pothetical reaction products II, III, and IV are formu- be nearer to 1 :10 . In Fig. 2 c, the attachment of TCH to components of the membrane fixed in osmium lated to reveal their capability of reacting and binding tetroxide is not quantitated . In Fig. 2d, the enhance- further OsO4, the last step in the OTO method . ment, most marked with lipid, of bridging osmium through TCH to the tissue-bond osmium, is demon- strated .

FIGURES 3 to 10 Rat organs were fixed in Palade's osmium tetroxide (8), embedded in Araldite, cut in ultrathin sections, and mounted on copper grids (top figure of each plate) for comparison with sections mounted on gold grids and treated with TCH followed by wash and exposure to osmium tetroxide (bottom figure of each plate) .

FIGURES 3 and 4 Kidney . Control, Fig. 3 ; OTO method, Fig. 4 . Note enhancement of contrast of mitochondrial cristae, mitochondrial dense granules, plasma membrane, and nuclear envelope. Also, note some lightening of the mitochondrial matrix .

426 B R I E F N 0 T E S Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021

BRIEF NOTES ~27 Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021

FmVRnS 5 and 6 Heart in cross-section. Control, Fig. 5; OTO method, Fig. 6. Note enhancement of contrast of mitochondrial membranes, and lightening of the initochondrial matrix. Note increased density of the myofilaments and of the sarcotubules.

428 Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021

FIGVRES 7 and 8 Cerebral cortex. Control, Fig. 7; OTO method, Fig. 8. Note enhan trast of all membranous structures and of lipid droplets and myelin.

429 Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021

FIGum~s 9 and 10 Small intestine. Control, Fig. 9; OTO method, Fig. 10. Note enhancement of con- trast of all membranous structures, of internal structure of the mierovilli, and especially of lipid droplets in vesicles. When we substituted for TCH other sulfur- original attachment to tissue-bound osmium. This containing osmiophilic reagents which were type of attachment also may occur with TCH, but capable of bridging metals, we noted that some of the additional hydrazino group is capable of re- these agents were without effect, such as thio- action with osmium tetroxide. semicarbazide (H2NNHCSNH2), or increased Application of a second T-O procedure upon a density generally but not membrane contrast, first OTO method resulted in further increase in such as p-chlorothiophenol, or decreased the density, without further improvement in contrast density of the mitochondrial matrix resulting in or in delineation of structure. However, when some increase in membrane contrast, such as viewed in a 100 kv beam, the sections so treated hydrazine (H2NNH2), 1,4-dithiothreitol (HSCH2 looked similar to the sections treated with the CHOHCHOHCH2SH) and 1,2-dithioglycerol original OTO method at 50 kv. (HSCH~CHSHCH2OH), or decreased the density Since hydrazone formation is facilitated by acid of the matrix as well as increased the density of or alkaline pH, the role of acid was tested. The membranes producing maximum membrane reaction of TCH with osmium tetroxide-fixed contrast, such as carbohydrazide, (H2NNHCO- sections was unaffected by 1% acetic acid but NHNH2, CH) and hydrogen sulfide (H2S). None could be slightly reversed by washing with 50% Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021 of the reagents was as striking as TCH. It was our acetic acid solution. However, after completion impression that some of the reagents were capable of the OTO method, 50% acetic acid wash was of decreasing the density of the mitochondrial without effect. Although more water soluble, matrix. These were TCH, CH, hydrazine, and TCH mono- or dihydrochloride, or diacetate gave hydrogen sulfide. These reagents seemed to us very poor results when compared with those with capable of removing osmium, from osmium TCH, in spite of the use of concentrations higher tetroxide-fixed tissue, selectively from com- than 1%. ponents other than lipid, such as protein, thus Affinity of TCH for other metal ions bound to lightening the matrix of mitochondria around tissue was also studied with the electron micro- membranes, which contributed to their sharpness. scope. Glutaraldehyde-fixed tissue, embedded in This way of increasing contrast was much less Araldite, cut in ultrathin sections, placed on gold effective than the increased density of the bridging grids and stained with 5% solutions of lead, reaction. That this occurs with TCH may be seen uranyl, mercury, copper, zinc, or chromate salts, by comparing the matrix of the mitochondria in or 0.1% K2PdC14, was washed and treated with Figs. 3 and 4. The fact that good contrast can be 1% TCH followed by osmium tetroxide vapor obtained with TCH, CH, and H2S, suggests that, (T-O procedure). Acid was avoided in the wash- in addition to osmiophilic sulfur, a bifunctional ing process, to prevent reversal. Exposure to os- hydrazide or coordination to two metal atoms mium tetroxide resulted in enhancement of the (bridging) is also required and may be the most particular metallic stain as far as contrast was con- important feature, as revealed by the quite good cerned, but increased sharpness of membranes was contrast at low density seen with CH. However, not produced by any of these metals as compared neither TCH, CH, or H2S increases contrast when with the OTO method. Significant contrast was used alone without the final step of exposure to conferred on metal stains (such as copper and zinc osmium tetroxide vapor. This means that selective ions) which are not by themselves particularly removal of osmium by these agents must play a good stains for electron microscopy. The most minor role, if any, in producing the striking striking increase in contrast occurred when the features of the OTO method. T-O procedure was applied to sections stained Not only was thiosemicarbazide (TSC) unable with basic lead acetate or uranyl acetate. How- to yield the OTO phenomenon when substituted ever, greater delineation of membranes or visual- for TCH, but it proved to be a fairly good block- ization of new structures was not produced by this ing agent in preventing TCH from producing the process. The results were no better when osmium OTO reaction when osmium tetroxide-fixed tetroxide-fixed tissue was stained with these metal tissue was exposed to TSC prior to performing the salts before applying the T-O procedure. It is also T-O procedure. This indicates that although TSC fair to point out that the OTO method itselt is a bidentate ligand, both the hydrazino group applied to ultrathin sections of glutaraldehyde- and the thiocarbonyl group are involved in the fixed tissue did not give results that in any way

B R I E F ~ O T E S 431 compared in excellence with those obtained with lyric enzymes by methods recently introduced tissue initially fixed in osmium tetroxide. However, and dependent upon the production of osmiophilic osmium tetroxide fixation after brief formalde- products (6, 10). This has been shown to be helpful hyde or glutaraldehyde fixation gave good results for esterase (12) and will be demonstrated in with the OTO method. subsequent publications dealing with the detailed A comparison of the results using the OTO presentation of other methods. method with those obtained with double heavy metal salt staining methods will be published later, This investigation was supported by a research grant and will show the superiority of the former in de- (CA-02478) from the National Cancer Institute, lineation of membranes. There is much less con- United States Public Health Service, Bethesda, tamination of the sect'ons with the OTO method Maryland. and the results are more reproducible. Further- Received for publication 8 February 1966. more, increased contrast is confined to structures that were osmiophilic to begin with, and not zo new components of the tissue unstained by osmium REFERENCES fixation. Downloaded from http://rupress.org/jcb/article-pdf/30/2/424/1068110/424.pdf by guest on 26 September 2021 1. ADAMS, C. W. M., J. Histochem. and Cytochem., The T-O procedure to enhance staining by 1960, 8, 252. metals other than osmium tetroxide has found use- 2. CRIECEE, R., Ann. Chem., 1936, 562, 75. ful application in enhancing contrast, in the 3. DALTON, A. J., and ZEIOEL, R. F., J. Biophysic. electron microscope, of uranium-labeled anti- and Biochem. Cytol., 1960, 7, 409. bodies (13, 14), and may find use with uranium- 4. HAKE, T., Lab. Inv., 1965, 14, 1208. azo dyes in studies directed to determination of 5. HANKER,J. S., DEn, C., WASSERKRUG H. L., and base sequence in nucleic acids (7). Not only is SELIOMAN, A. M., Science, 1966, 152, 1631. greater contrast produced by the final step of 6. HANKER, J. S., SEAMAN, A. R., WEISS, L. P., UENO, H., BERGMAN, R. A., and SELIGMAN, exposure to osmium tetroxide, but greater stability A. M., Science, 1964, 146, 1039. in the electron beam would be expected than with 7. MOUDRIANAK1S, E. N., and BEER, M., Proc. uranium alone; and this was found to be the case Nat. Acad. Sc., 1965, 53, 564. (l 3). The T-O procedure may also be used on the 8. PALADE, G., J. Exp. Med., 1952, 95, 285. copper chelate which is the end product of the 9. PARSONS, D. F., J. Biophysic. and Biochem. Cytol., histochemical reaction for aminopeptidase, to 1961, 11, 492. give a permanent pigment useful in electron 10. SELIaMAN, A. M., Proceedings of the 2nd In- microscopy as well as in light microscopy (5). Fur- ternational Congress for Histochemistry and ther enhancement in contrast of underlying struc- Cytochemistry, t3erlin, Springer-Verlag, ture may then be afforded by performing a second 1964, 9. 11. SELIOMAN, A. M., HANKER, J. S., WASSERKRUG, T-O procedure (note Fig. 7 of reference 5). H., DMOCHOWSKI, H., and KATZOFF, L., or. We have recently found the OTO method Histochem. and Cytochem., 1965, 13, 629. especially useful in enhancing the visualization of 12. SELIOMAN, A. M., UENO, H., WASSERKRUG, H., fine structure of tissues stained for various en- and HANKER, J. S., Ann. Histochim., 1966, 11, zymes, where initial exposure to osmium tetroxide 115. must be avoided until after the enzyme reaction is 13. STERNBERGER, L. A., HANKER, J. S., DONATI, completed. For example, the cristae of mito- E. J., PETRALI, J. P., and SEL1OMAN, A. M., chondria may be more sharply delineated in fresh J. Histochem. and Cytochem., 1966, in press. heart slices stained for cytochrome oxidase by the 14. STERNBERGER, L. A., DONATI, E. N., WILSON, production of an osmiophilic pigment (6, 10), C. E., J. Histochem. and Cytochem., 1963, 11, 48. 15. STOECKENIUS,W., Circulation, 1962, 26, 1066. when the T-O procedure is subsequently super- 16. STOECKENIUS, W., and MAHR, S. C., Lab. Inv., imposed. The T-O procedure enabled us to 1965, 14, 1196. demonstrate mitochondria and other membranous 17. WATSON, M. L., J. Biophysic. and Biochem. Cytol., structures more clearly than we could obtain by 1958, 4, 727. other metal staining methods, in formalin-fixed 18. ZOBEL, C. R., and BEER, M., Int. Rev. Cvtol., or glutaraldehyde-fixed tissue stained for hydro- 1965, 18, 363.

432 B R i E F N o T E S