Experiments on embedding media for electron microscopy

Over the last 5 years new embedding materials have been developed for electron microscopy. Certain of these media employ as an important element in the final embedding mixture a water miscible component, which has been used by some workers to replace the conventional alcohol dehydration step. Progress in this field has recently been reviewed by Glauert.³  Of the characteristics of the water miscible media, some offer possible improvement over methacrylate, but other features leave much to be desired when compared in ease of handling with epon or araldite. For example: glycol methacrylate (GMA) makes an extremely hard and brittle final block and the polymer swells considerably on contact with water; with durcupan it is not easy to obtain a block which is of satisfactory composition and hardness for cutting.

Bernhard and Leduc¹  have achieved some success by adding a proportion of the usual methacrylate mixture to glycol methacrylate. We collaborated in 1961 and 1962 in experiments designed to test the value of the newer embedding media in our hands, and we have attempted to improve the cutting properties and preservation of the tissue in the final block in 2 principal ways:

(a) by using the water miscible resin component as a dehydrating agent, and embedding the tissue in another monomer mixture of satisfactory properties with which it is miscible;

(b) by making mixtures of the water miscible components with other water immiscible or partly water miscible plastics, and determining the polymerization characteristics and other properties of the mixtures.

The results reported in this paper are largely concerned with the data obtained under the heading (b).

During these investigations it was necessary to obtain some basic information regarding the miscibility with one another of the reagents often used in embedding technique for electron microscopy. From these studies it has.become clear that there, are interesting means available for the truly comparative study of the effects of the dehydrating and embedding steps upon the final appearance of the tissue in the electron microscope.

The tests were performed by placing 3 ml of each of 2 components in a test-tube, shaking gently until the contents were mixed, and standing at room temperature for 24 h. Each test-tube was inspected regularly for any signs of layering or inhomogeneity. Ó f the components tabulated here (table 1), most were well miscible, and showed no signs of layering after 24 h.

From these observations it would appear likely that one may dehydrate tissue with either GMA or X133/2097 and embed the tissue in either methacrylate or epon or vestopal. We have tested this possibility using tissue fixed in either osmium tetroxide or formalin, dehydrated in graded steps of GMA or X133/2097, and passed via a 1:1 mixture of dehydrating agent and embedding material, into the final supporting medium. Analysis of the morphological effects of these treatments is not complete, but thus far no special difficulties have been encountered in the normal technical procedures, and tissue preservation seems adequate.

There is clearly a very large number of possible combinations which could be tested. Indications of the possible value of different groups of combinations can however be obtained by studying the results of mixing together the different complete resin mixtures recommended in the literature. We chose, arbitrarily, to mix them in the proportion 1:1, and we used the following basic ‘mixes’:

1. Epon ‘C’ = EC = Luft’s complete resin mixture, 5A plus 5B with 2% v/v DMP 30. (Luft.)⁴ 

2. ‘Bartl’ = BC = 97 parts commercial GMA (Rohm & Haas) plus 3 parts of a 16-66% w/v (NH₄)₂S₂O₈ solution (to give 0-5% final concentration). (Rosenberg, Bartl, and Leško.)⁷ 

3. ‘Methacrylate’ = MC = 9 parts butyl methacrylate plus 1 part methyl methacrylate (stabilizer removed) plus 2% w/v benzoyl peroxide.

4. ‘Leduc’ = LC = 7 parts‘Bartl’plus 3 parts ‘Methacrylate’.

5. ‘Stàubli’ = SC = 2-5 parts Xi33/2097 plus 5-5 parts hardener 964 plus o-6 parts accelerator 960. (Staubli.)⁹ 

6. ‘Vestopal’ = VC = 50 parts vestopal W, 0-5 parts initiator, 0-5 parts activator. (Ryter and Kellenberger.)⁸ 

The mixtures were made in accordance with the original published methods at the concentrations stated above. The resin mixtures listed above were mixed thoroughly in pairs 1:1 in test-tubes, in accordance with table 2, and inspected for any signs of layering after several hours at room temperature. Each resultant double mixture was given the designation listed in the table. Most pairs proved to be well miscible, but MV and SV show layering after 24 h at room temperature. ES is a rather viscous combination.

We consider the low temperature (37°C) overnight incubation step, which appears in many impregnation procedures, a very significant one. It ensures maximum impregnation of the tissue with plastic of somewhat lowered viscosity, and also initiates polymerization gently. Omission of such a step of ‘prolonged soaking’ at 37°C or similar temperature has a direct effect, as one would expect, upon the final curing time, by whatever means. We therefore routinely included in our test procedures a stage at 37° C. Capsules were filled with the double mixtures, incubated overnight at 37° > C after capping, and then finally polymerized either in an incubator at 60° C, or under UV at 4° C. The results are given in tables 3 to 5 (appendix). When these experiments are carried out in glass test-tubes in larger quantity, rather than in gelatin capsules, results are similar, but not identical. This is most noticeable in the cases where water is added to the resin mass, and it seems likely that the gelatin capsule acts as a significant ‘sink’ for the water present during polymerization. The tabulated results are based on the results obtained in capsules, as capsules are part of the normal working procedure in electron microscopy. There are often differences between rates of polymerization under UV or at 60° C. The precise character of the blocks during trimming and sectioning can be slightly different after the two treatments. However, UV and 60° C treatments are not invariably different in timing or results (e.g. LC table 3).

De Bruijn (unpublished) had previously demonstrated that an apparatus of the type illustrated in fig. 1 could be used at 4° C and below to achieve good polymerization of methacrylate in gelatin capsules. The apparatus is based on that published by Müller.⁶  At —io° C the rise in temperature within the polymerizing plastic in the capsules was held below 17° C, and it is likely that use of a cold room at —40° C would maintain the temperature of tissue in the capsule at o° C or below. This would be valuable for histochemical studies of enzymatic activity. As the possibility of enzyme histochemistry was one of our reasons for investigating the water miscible resins, we felt that it was necessary to include UV polymerization in our studies. Either a Philips HPK 125 type 126036 ultraviolet lamp or a Philips ultraviolet MLU 300 W 220-240 V type 57265 F/28 may be used in this type of apparatus. In all the experiments reported in this paper we used the latter type of lamp. The fan may be placed either above the lamp or below the capsules. The preferred direction for the air flow is towards the lamp, away from the capsules. Lamp to capsule distance was minimally 10 cm.

In table 3 we give the data for curing times and condition of polymer mixtures in order to offer a guide to other workers, but no more than a guide. Clearly the exact timings will vary from laboratory to laboratory, in accordance with uncontrolled variables. Curing times of over 10 h with UV and of over 48 h at 60° C we generally consider as unsuitable for routine procedures. Polymer mixtures which were opaque, cloudy, or inhomogeneous were not regarded as suitable for further study, although it is not proven that cloudiness of supporting medium will affect the preservation of tissue or the electron microscopic image of such tissue. Nor can it be assumed that a cloudy polymer is unsuitable for sectioning, as shown by the mixture EV (table 6).

The most promising double mixture observed in the experiments tabulated is the mixture 1:1 epon C and Leduc (EL), which appears to have very much the consistency of epon, but also contains the water miscible GMA. Since epon 812 itself contains a high proportion of water miscible resin (30% aquon, Gibbons² ), it seemed likely that a satisfactory polymer with some water content might be made of this combination. We have tested mixtures with final water contents of between o to 20%. The miscibility of the components remains unaffected up to about 10% added water. Above this concentration, separation of the components of the mixture becomes noticeable. The mixtures are designated EL⁹⁵, EL⁹⁰, and EL⁸⁰, for final water contents of about 5%, ^Io%, and 20%, and the polymerization data are listed in table 3. One may be able, by the use of a mixture like EL, to support tissues in plastic for thin sectioning with an appreciable quantity of water (optimally 5 to 10%) still present.

A second important feature is that the mutual miscibility of the GMA, methacrylate, and epon components and the satisfactory nature of the final polymer EL, permits one to formulate experiments which will demonstrate comparatively the effect on the final image of different stages and reagents used in processing. Material can be fixed in the same manner, divided into portions, and submitted to different dehydration procedures, prior to embedding in the same final supporting medium. We have conducted experiments along these lines, which suggest that useful information can be obtained from such studies.

One would not necessarily expect a complex mixture like EL containing 3 different plastics to form a better polymer than a mixture of like plastics, for example of epoxy-type resins. It is interesting, therefore, to note the results obtained for the combination ES containing epon 812 and the water miscible epoxy resin X133/2097. The components are well miscible, but form a viscous combination, which is difficult to handle. However, better penetration into the tissue block may be assured by an antemedium of 1,2 epoxy-propane, with which it is well miscible. Reduction of the hardener in either the ‘Staubli’ or the ‘epon C’ might also reduce viscosity. The consistency, trimming and cutting properties of the final polymer are much more encouraging than for the original Staubli mixture, and more closely approach epon. Very recently, W. Staubli¹⁰  published a procedure in which X133/2097 (durcupan) is used as a dehydrating agent prior to embedding in araldite. We did not test combinations of X133/2097 with araldite, as we assumed that epon would offer us the same features as araldite.

Following the reasoning applied to the mixture EL, we have tested whether it is possible to polymerize ES mixtures with a moderate water content of between o to 20%. The mixtures for which results are listed in table 3 are designated ES⁹⁵, ES⁹⁰, ES⁸⁰, for final water contents of about 5%, 10%, and 20%. We also tested the effect of added water upon the original mixture of Staubli⁹  by diluting the component X133/2097 (durcupan A) with appropriate quantities of water. The presence of water in the final original mixture of Staubli does not prevent polymerization, but generally gives a polymer which is too soft for sectioning. From the data in table 3 it is clear that above a final concentration of about 10%, water also makes the mixture ES too soft. But below 10%, especially if only 5% or less is present, ES is a useful polymer which trims and sections easily. There are some indications that water is actively excluded from resin masses,, particularly, during heat polymerization, and subsequently lost by evaporation. One cannot therefore assume that the entire amount of added water is present in the final cured mass of resin. The softness of the ES mixtures with a higher water content may perhaps be overcome by using a harder epon C mixture, or by using a ratio E: S of 2:1 instead of 1:1, but we have not yet tested this.

All our observations emphasize, a point which appears throughout the experiments: complex mixtures of plastics, hardeners, plasticizers, and catalysts do not behave in a predictable manner, and their properties can only be evaluated by empirical test. For example, although, as shown in tables 1 and 3, methacrylate and vestopal are immiscible, the mixture LV, which contains glycol methacrylate as well as methacrylate and vestopal, does not show the layering and inhomogeneity which might be expected, either on mixing or on polymerization (table 3), and sections well (table 6).

It seems likely that the polymerization of one catalysed monomer may directly affect the polymerization of another associated with it. Thus, as shown in table 4, where we assess the percentage of polymerization after a constant time (15 h) at 37° C, there may be mutual acceleration in the combinations of methacrylate with GMA (LC), and of both with epon (EL). Our assessments are based on the proportion of the resin mass in the capsule which became firm, usually in the lower part of the capsule, the rest remaining liquid. Furthermore, from this table it can be seen that the other epoxy resin, X133/2097, in combination with either methacrylate (MS) or glycol methacrylate (BS), and catalysts, suffers inhibition, but that this is much reduced when it is mixed with both together (LS). It is possible that there may be direct interaction between polymerizing chains. It is also possible that catalysts may interact directly with each other. It might be expected that catalysts for one plastic could cause polymerization in another. In order to go some way to disentangle these interactions, we obtained the data listed in table 5. From this table and other experiments we can draw the following conclusions:

(a) Neither benzoyl peroxide nor ammonia persulphate is an effective catalyst for epon 812 alone.

(b) In increased concentration, neither benzoyl peroxide nor ammonia persulphate promotes the polymerization of a 1:1 mixture of GMA and epon 812 since the times given in table 5 are much the same as those given in table 3 for the control mixes.

(c) There is no direct interaction between GMA and epon 812 during polymerization.

(d) DMP 30 is not an effective catalyst for X133/2097 or for mixtures of X133/2097 and epon 812. The addition of X133/2097 alone to epon C therefore tends to slow down its normal rate of polymerization.

(e) Accelerator 960, the normal catalyst for X133/2097, is also a catalyst for epon 812, particularly when in combination with its curing agents DDSA (dodecenyl succinic anhydride) or MNA (methyl nadie anhydride). Hence it also catalyses mixtures of epon and Xi33/2097.

(f) DMP 30 is not an effective catalyst for vestopal W nor is accelerator 960. It can be shown that vestopal W alone can be hardened completely by 15 h at 37° C followed by 4 h under UV, and the times for the normal control mix VC listed in table 3, are shorter than those in table 5 with added DMP 30.

(g) Benzoyl peroxide is the normal initiator for the usual vestopal mixture, but it is interesting to note that a doubled concentration (2%) of benzoyl peroxide alone produces a hard polymer in less time than the normal proportions of initiator and activator.

(h) Cobalt naphthenate, the activator of vestopal W, is not an effective catalyst for the Stàubli mixture.

(i) As a mixture of X133/2097 alone and ‘Vestopal’ hardens quickly without layering, but the SV mixture layers considerably during curing, it may be inferred that the hardener DDSA is immiscible with vestopal. This is confirmed by direct experiment.

Other similar inferences may be drawn from the tables, and tested. The value of the tables lies in the multiplicity of treatments which are shown to be available to prepare tissue for electron microscopy. Such a variety of techniques may be of considerable help in studying refractory tissues and should also be valuable in forming assessments of the results obtained by current procedures.

Mere polymerization of a mixture does not constitute an adequate indication of its suitability for histological electron microscopy. It is necessary to embed well-known tissues in the media, cut sections, and assess the quality of the supporting medium by the study of micrographs. This, however, is an extremely time-consuming task, if applied to many different mixtures. It is possible to shorten the task by considering some ancillary features which are more quickly assessed. Certain of these may be considered under the heading ‘trimming and sectioning properties’. The criteria are highly subjective, but are descriptive of properties rapidly appreciated by anyone used to handling blocks for ultra-microtomy. Table 6 lists our appreciation of the experimental mixtures in these terms.

All assessments of trimming properties were made using fresh razor blades on cold blocks, i.e. blocks which had been at room temperature for many hours. All assessments of sectioning quality were made by the same observer working with a Huxley ultramicrotome and glass knives, using 20% acetone/water in the boat. It is worth pointing out that the elaborate saws and grinding tools which have been advocated by some workers are not necessary to trim even the hardest blocks. If a refractory block is returned to the 60° C oven until warm, or washed in warm running tap water, and trimmed shortly after, all trimming and shaping operations can be done with a single fresh razor blade. Sectioning should be delayed after such a warm trimming procedure until the block has thoroughly cooled.

We have not completed a full survey of osmium, formalin, and permanganate fixed tissues in the experimental mixed embedding media. However, as mentioned above, and as can be seen from the tables, the most promising mixture of all is that designated EL, epon/Leduc, containing Luft’s⁴  complete epon mixture and Bernhard and Leduc’s¹  mixture of conventional methacrylates and glycol methacrylate. By examining Krebs 2 ascites tumour cells embedded in EL as test objects, after fixation by centrifugation through each of the 3 fixatives, we are satisfied that this medium does not produce polymerization damage. Membrane preservation is as satisfactory as it is after using epon itself. The medium ‘clears’ in the beam to about the same extent as epon, and may be used satisfactorily on naked grids, for it is of high stability in the beam. It is less viscous than epon ‘C’ and therefore easier to handle during infiltration and embedding. Experiments on different methods of pre-treatment before final embedding in EL show that with conventional osmium fixation and dehydration prior to embedding in EL, there is virtually no difference from the images seen in ordinary epon, figs. 2 and 3. Preservation as good as any reported in the literature, is obtained after formalin or permanganate. Given a final supporting medium of high stability, which EL would appear to be, our experiments seem to indicate that the most significant steps after fixation are those involved in dehydration. There is a suggestion that the embedding medium may affect, to some extent, the final ‘staining’ or ‘contrasting’ of the embedded tissue with uranyl acetate, as compared with tissue in epon, but this would seem to apply mainly to large lipid containing droplets. The majority of cellular elements are stained normally. Tissue embedded in EL, like epon, can be stained satisfactorily with either Nile blue or Sudan black solutions for light microscopy (McGee- Russell and Smale).⁵ 

We have demonstrated that epon can be mixed with certain other plastics during dehydration sequences or in the final embedding medium. Tissues may therefore be brought through water miscible plastics into a final medium of epon, or water miscible plastic mixed with epon, so improving the sectioning qualities of otherwise difficult media. In comparing dehydration sequences experimentally, it is sometimes more satisfactory if the final embedding medium is exactly the same. The mixture designated EL in this paper is suitable for experiments of this kind. We have shown that some combinations of plastic will tolerate the presence of water in the final embedding medium. This may mean that a wide range of histochemical techniques can be applied fruitfully to sections of tissue embedded, after suitable processing, in epon, or in combinations of epon with other plastics.

We wish to thank Mr. C. J. G. Smale and Miss H. Belham for their valuable technical assistance. We are grateful to Dr. F. K. Sanders for providing many samples of Krebs 2 ascites cells, and for the hospitality he extended to W. C. de Bruijn within his Unit, which enabled us to continue a stimulating collaboration; also to the Medical Research Council for the facilities of their Laboratories at Carshalton. W. C. de Bruijn’s visits to this country in 1961 and 1962 were made possible through the generosity of the Dutch Organisation for Pure Scientific Research, Z.W.O.,’s-Gravenhage, Holland, and the City Council of Rotterdam.