Distribution of Glycogen in the Developing Salmon (Salmo Salar L.)

It has been pointed out by Needham (1931, I, 355) that results of histochemical observations upon glycogen in fish have not necessarily been confirmed by chemical estimations. Even in one of the rare cases quoted in which glycogen is supposed to have been estimated in fish eggs, it seems that a mistake has been made in translation. Fauré Fremiet & Garrault (1922) are reported, by Hayes & Hollett (1940), to have recorded 0·34% glycogen in the trout egg. What the former actually stated was ‘Nous n’avons pas pu déceler la présence de glycogène dans l’œuf de Truite, mais nous avons trouvé une faible quantité (0·34 pour 100) de sucre réducteur que nous avons dosé en glucose’; which seems to mean that they found 0·34% glucose and no glycogen.

Information on the subject of glycogen content is less satisfactory for developing teleosts than it is for Amphibia. Attention has been concentrated upon development in the latter group through the study of Organization Centres. Methods for estimating small quantities of glycogen have been applied, and there is general agreement between results of histochemical and chemical investigations, e.g. Needham (1931, 2, p. 1046), in spite of differences over the question of a glycogen decrease in the dorsal lip of the blastopore of Amphibia during invagination.

There is reason, therefore, for the application of recent improvements in histochemistry and quantitative estimations to the study of teleosts.

The developing eggs and larvae of the salmon (S. salar) have been chosen here for investigation by both methods. In 1941 sagittal sections were prepared from material in different stages of development and examined for glycogen. At this time attention was drawn to a paper by Hayes & Hollett (1940) which deals with the carbohydrate content of eggs and larvae of S. salar up to the end of the yolk-sac period. To obtain total carbohydrates, including glycogen, they used water extraction in a Soxhlet apparatus for 2−3 hr. Results obtained do not agree in certain particulars with the distribution of glycogen as seen in sections.

Their method is not well adapted for determining small quantities of glycogen. Carruthers & Ling (1934), for instance, do not regard water extraction of glycogen and total sugars as satisfactory for work on liver tissue with less than 4 % glycogen.

Untoward circumstances prevented the carrying out of estimations by a modified Pflüger method until 1944−5, but these being completed, the results are included, for convenience, in the first part of the paper, although in point of time they were obtained three years later than the series of prepared sections. On the whole these results are in agreement with the distribution of glycogen as shown in the histological sections.

The larvae were kept in a hatching apparatus similar to that described by Moult (1939); main water, which is of good quality, was run through the trays.

Stages I-XII, for histological work, were obtained during the winter of 1941-2 from ova fertilized at the River Dee (Wales) Fish Hatchery on 21 November, and a second series, samples 1−12, for glycogen estimated in 1944−5, from ova fertilized there on 23 November 1944.

Varying conditions of water temperature during each season made it advisable to find a basis for comparing the rate of development in the two series.

A usual method adopted is the thermal unit, originally proposed by Wallich (1900), who states, however, that chief dependence should not be on the unit but on the appearance of the various stages.

In the present investigation the thermal unit was not found to be of any service. For instance, in 1941−2, the larvae hatched out at 615−632 thermal units, in 1944−5 at 897 thermal units.

Battle (1944) has noted similar discrepancies in the salmon. She states on p. 116 that ‘the variation in the thermal units required to reach the same status in these (her own) two series, throws some doubt on their validity as criteria for determining comparable stages in embryos’.

Indeed, I have found the excellent drawings accompanying her paper much more useful in identifying various stages.

The advantages of drawings of easily identified stages has been recognized by Pelluet (1944), and a series he has produced for the salmon has been helpful.

In order to synchronize the two series dealt with in the present paper the following plan has been adopted :

Samples 1−12 (representing the estimations) have been plotted along the base line of Text-fig. 1 and spaced according to the number of days each sample was taken before or after hatching. Stages I-XII (representing the histological sections) have been put in the same figure, but the spacing of them is not related directly to the time scale. Each stage is placed in a position relative to the samples, which seems warranted by comparison of the development reached in the two series. To take the simplest example, stage VI followed immediately on the hatching out of the majority of the larvae in 1941−2. It corresponds, therefore, with sample 7, which was also taken immediately after the hatching in 1944−5.

For general comparisons the following data may be given:

Once the larvae are hatched the stages of development can easily be identified from Pl. 2, figs. 4−6 and Pl. 4, fig. 13.

The amounts of glycogen in the embryos and in the yolk-sacs were estimated separately. The eggs or larvae were first weighed alive and then thrown into hot alcohol. This stopped enzyme action and also coagulated the yolk contents, so that the yolk-sac could be cut from the blastoderm or embryo without serious loss of material. In this separation a fair amount of yolk was included with the embryo, to make sure that no true blastoderm or embryo tissue was included under the heading ‘yolk’, particularly the liver which lies couched on the yolk.

The precaution actually caused a division into : (a) embryonic tissue plus a certain amount of yolk-sac, and (b) rest of the yolk-sac and yolk. A part of the protoplasmic vitelline membrane network was also included under yolk-sac and, after the anterior and lateral margins of the blastoderm had grown over the yolk, there would be included, also, ectoderm of the embryo and later mesoderm and endoderm. In fact the yolk in the salmon is finally enclosed by embryonic tissue which forms the sac.

It is clear from the figures of Hayes & Hollett (1940) that the amounts of glycogen in salmon larvae are comparatively small, and it is necessary to adopt methods suitable to such conditions. About fifty larvae were taken for each of the earlier samples, but with an increase of glycogen later, numbers were correspondingly reduced or only part of the sample was used for final estimation.

Extraction and hydrolysis were carried out by the method of Good, Kramer & Somogyi (1933) with the following minor modifications:

• Two volumes of alcohol were added to the mother liquid of caustic potash.

• After preliminary dissolving in water and reprecipitation with alcohol, the material thus obtained was dissolved once more in water and neutralized to litmus with 1 % sulphuric acid. Yolk precipitated by the neutralization was filtered with careful washings.

Determination of glucose, obtained by hydrolysis, was made by the copperiodometric reagent of Somogyi (1937), devised for the estimation of small quantities.

The solution is practically reagent 50 of Shaffer & Somogyi with added sodium sulphate. The 200 g. of the latter per litre recommended by Somogyi seems to be excessive, since about 167 g. of material crystallized out from the solution under laboratory conditions of temperature. The decanted solution, when tested against known strengths of glucose, however, gave reduction equivalents very similar to those in the original paper. As an instance the reading for 0·5 mg. of glucose, after boiling for 20 min., was 4·68 c.c. of 0·005 N thiosulphate compared with 4·60 c.c. of 0·005 N thiosulphate.

In the samples the possibility of reducing substances other than glucose being present during boiling with the copper reagent cannot be dismissed. Young & Inman (1938) have reported small amounts of cystin and glucosamine in the dried egg capsule of the salmon. Although pyrocatechol, derived from glucosamine by boiling with caustic potash, reduces cupric copper, it is soluble in both water and alcohol and will therefore stay in solution and be discarded when the centrifuged tubes are decanted. Cystin may be less easily disposed of by the method of glycogen extraction adopted, but any effect it might have on the low alkalinity reagent of Somogyi would be at a minimum.

The specific action of amylase upon polysaccharides has therefore been used as an additional means of testing for glycogen in the tissues under examination. The sensitivity of enzyme action involves such factors as temperature and the purity of enzyme and substrate, together with incomplete hydrolysis (e.g. Sahyan & Alberg (1931) found that takadiastase at 38° C. for 4 hr., hydrolysed only 52% of the original pure glycogen). Quantitative estimations, therefore, would need to rest upon a basis of further investigation.

Nevertheless, it is useful in the present work to demonstrate by the use of amylase, the presence or absence of glycogen in the various samples, as a further check upon the validity of the glycogen distribution as shown by estimations and in sections.

For this purpose, when the stage was reached in quantitative estimations for hydrolysis, the content of the centrifuge tube was first dissolved in 10·0 c.c. of distilled water; 5·0 c.c. of this solution was then pipetted into a second centrifuge tube, so that each tube contained one-half of the particular sample.

The original tube was dried off again and used for acid hydrolysis and subsequent glucose estimation. The second tube was also dried, and 5·0 c.c. of filtered saliva (1 : 10) + toluene was added before keeping in an oven for 24 hr. at 38° C. An advantage of using saliva is that whereas it hydrolyses glycogen, any galactogen which happens to be present will remain unattacked (May, 1931)..

At the end of this time, 2 vol. of ethyl alcohol was added to the tube, so that any glycogen still intact, dextrins formed and any other material soluble in water and insoluble in 66 % alcohol, which had come through the various precipitations with 2 vol. of alcohol, following the original boiling with caustic potash, would be thrown out of solution again.

Any substance remaining in solution in the 66 % alcohol, therefore, would have been produced by the specific action of saliva, and it remained to identify the substance, If a sample of pure glycogen is boiled in caustic potash and put through this process with saliva and subsequent addition of 2 vol. of alcohol, what is held in the final solution is not maltose, as might be expected, but glucose. This was established by the formation of typical glucosazone crystals when the solution was concentrated and boiled with phenylhydrazine hydrochloride and sodium acetate. Estimations of glucose showed the yield of glycogen to be low and variable, amounting from 10 to 33% of the original glycogen taken.

The glucosazone reaction was also obtained from salmon larvae by taking 150 larvae left over at the end of the experiment and submitting them to the same procedure. This was also true for a similar number of larvae that had been fixed in Bouin-Allen and preserved in 70 % alcohol for a year. In view of the facts already given, the glucose could only have been obtained from glycogen originally in the tissues of the salmon larvae.

The most satisfying procedure would have been to apply the glucosazone test to every sample of embryo and yolk-sac taken throughout the larval series. It was obvious, however, from the low returns given by the action of saliva upon pure glycogen, that in these samples small amounts of glucose would be involved, which would be insufficient to support the glucosazone reaction.

The Somogyi copper iodine reagent, therefore, was used instead as a qualitative test for the presence or absence of a reducing substance, after the use of saliva as outlined above.

Using filtered saliva as a blank, all results were positive, and from the evidence presented it may be inferred that the reducing substance was glucose. The lowest reading was the equivalent of 0·01 mg. glucose for half the amount of yolk-sac and yolk taken for sample 2.

The results of the glycogen estimations are given in Table 1. Although, as already explained, the division into blastoderm or embryo and yolk-sac is somewhat arbitrary, the average wet weight and glycogen percentages have been included in their appropriate columns.

The actual weight of glycogen is much the same in blastoderm and in yolk-sac for samples 1 and 2. It is greater in the yolk-sac in samples 3, 4, 5 and 6 and in the embryo in samples 8, 9, 10, 11 and 12, as might be expected, since the yolk is gradually disappearing and the embryo growing at its expense.

Nevertheless, there is an actual rise in the weight of yolk-sac glycogen in the later samples, which reflects a quickening in deposition.

The glycogen percentages on wet weight give a different picture. They are usually a little higher in the blastoderm or embryo than in the yolk-sac. Since the yolk itself adds much to the total weight of the yolk-sac and carries no glycogen in its main bulk, the percentage figures suggest that glycogen may, actually, be more concentrated in, or near, the embryonic tissues of the yolk-sac than in the embryo.

Results obtained by Hayes & Hollett (1940) for S. salar differ from the above in the following particulars :

They state that ‘no glycogen was detected in the yolk-sac except on two occasions and then in such minute quantities—of the order of 0·01 mg.—that it might well be attributed to error’.

The first appearance of glycogen recorded by them was about 3 weeks after fertilization. Taking into account a slightly longer time between fertilization and hatching in the developing salmon of the winter 1944−5, this would mean an appearance about 5 days after sample 3 (1944) was taken, i.e. about the time of stage III.

From this it can be’ concluded that they found no glycogen in their samples which correspond to stages-I and II of the present series, and their first record would be about stage III.

Further comparison between the two series can be made by examining the graphs of total glycogen per egg or larva included in Text-fig. 1. Hayes & Hollett based their results upon an egg of 130 mg. at fertilization, but this does not inconvenience the comparison.

The important difference is that glycogen was present in all samples from 44 days before hatching onwards, whereas the first record of Hayes & Hollett was 22 days before hatching.

While the present amounts are small and show little change in the embryo or the yolk-sac until 10 days after hatching, their series rises from zero, 28 days before hatching, through the time of hatching and onwards with two check periods, 6−10 days after hatching and about 32 days after hatching.

Both investigations agree in a rapid increase in amounts of glycogen between 20 and 39 days after hatching, the latter being the time of the last record in the present series, and this is the period of maximum fat absorption in developing salmon eggs (Hayes & Ross, 1936).

The finding of glycogen in the yolk-sac and of its early appearance in development agrees with the distribution of glycogen as seen in the histological sections to be described. It depended upon the use of methods which were devised to estimate small amounts of glycogen. It may be noted, however, that the steady increase recorded by Hayes & Hollett (1940) from 28 days before to 6 days after hatching, not found in the 1944−5 series, covers the period from stage III to stage VI in the histological series, during which glycogen first appears in muscular tissue (stage III) and then in the liver (stage V).

Sagittal sections were obtained which included the so-called animal and vegetable poles of the fertilized egg, and in later stages they cut through the longitudinal axis of the developing larva. Sections taken in the median vertical plane were always used for obtaining the figures, although lateral ones were also used for examination.

Material was fixed in Bouin-Allen or picro-dioxan solutions, upgraded through N butyl alcohol in both cases, and imbedded in paraffin wax m.p. 58° C., to which 1 % of ceresin had been added. Both fixatives, used in this way, gave similar pictures of the distribution of glycogen in stained sections.

Each wax block was shaved down until a thin slice had been removed from the imbedded tissue and then left overnight in a mixture of 70 % alcohol and glycerine (9 : 1), or in Mollifex (B.D.H.) for yolk softening. After this treatment it was possible to cut a number of serial sections including the yolk, before the splintering of the latter gave warning that further immersion was necessary. The sections were cut comparatively thick (about 10µ).

With the above methods the chief technical difficulty was not concerned with the cutting of complete sections, but with sticking them to the glass slides before staining and mounting in Canada balsam. Although several methods were tried besides the usual routine of using egg albumen, the only successful one was to immerse the slides and sections in % celloidin. If required this was finally removed by acetone before mounting.

The staining methods used for demonstrating glycogen were : (a) Lugol’s iodine-potassium iodide solution (strong formula), the sections being counterstained with alcoholic light green, (b) Best’s carmine, (c) Bauer’s reaction, using chromic acid followed by Schiff’s reagent. This is a particularly useful method because it provides its own counterstain, general tissue being blue in contrast with the red-violet colouring of glycogen. The deep brown stain of the iodine, the brilliant red colour of Best’s carmine and the characteristic colour reaction of Bauer act as a check upon each other when the distribution of glycogen is being considered.

No less important is the use of control sections, prepared by immersion in warm water and afterwards leaving for at least half an hour in saliva at 40° C. A clear solution of salivary amylase may be obtained by rinsing out the mouth with about 100 c.c. water at a time, and filtering the product before use.

The check sections should be prepared and examined with great care after they have been stained and mounted. The almost complete absence of deep-staining reactions following this procedure suggests a thorough removal of glycogen from sections. It is known that hydrolysis of glycogen to maltose (or later glucose) by salivary amylase is not complete. It can only be assumed that glycogen, and dextrins formed, are freed in some way from the action of the fixative so that they are no longer insoluble.

It is only possible to make general statements about the distribution of glycogen in sections. Magnification under the microscope tends to give a disproportionate idea of the actual quantity that is being examined. Further, the stained material may not be pure glycogen. The latter may be adsorbed on the surface of other material, or even form a loose complex with it. Examination of check sections, in which there is no distinctive staining reaction, is useful in revealing such conditions. A good example is found in larvae at stage V. Here there are inlets in yolk lying close to the yolk-sac cells. Discrete material in these inlets is heavily stained after iodine, Best’s carmine and Bauer. In hydrolysed sections there is some unstained substance still present in the position marked in other sections by staining reactions.

Such conditions are found in subsequent stages and are very marked in stage IX, where, in addition, granular material is left behind in those places on the edge of the yolk cells where, apparently, there had been solid clots of glycogen.

The three main centres of glycogen concentration as seen in stained sections are the liver, the muscular system and the cells lining the yolk-sac.

Glycogen has a wider distribution in the tissues ; it appears, for instance, in the branchiae, but observation has been concentrated upon the three centres named, and these are examined in turn through the various stages of development, except stages I and II, which will be dealt with in their entirety since, in them, the tissues are not yet differentiated.

Stage I (Pl. 2, fig. I ; Pl. 3, fig. 7). At this stage the fertilized egg is enclosed by a tough outer membrane or chorion. Inside this, and spreading over the yolk, is a protoplasmic vitelline membrane. Between the two membranes is a perivitelline space, which also overlies a multicellular blastoderm resting on periblast at the animal pole. The chorion hardens when in contact with water and, according to Bogucki (1930), it is then impermeable, or pratically so, to colloidal solutions. He further states that the yolk releases colloidal substance into the perivitelline space. The fluid there is slightly opalescent and is said to contain protein. Certainly it has a strong reaction to stains used for glycogen, so that in section the perivitelline space shows up as a deep brown or red continuous band, according to the staining method used. A similar deposition of glycogen is apparent in the fertilized frog’s egg; Needham (1931, 2, 1044) states that it is partly eliminated into the perivitelline space and almost wholly used up by cells of the embryo before gastrulation. As for the origin of this glycogen in stage I, it may come from the yolk, as suggested above, but some glycogen is already at the periphery of the yolk in unfertilized salmon ova.

Glycogen is also concentrated in the tightly packed blastoderm cells. It is granular in appearance and concentrated towards the cell walls, so that in certain areas the staining reaction marks out the pattern of the cells.

Cells nearest the chorion contain most glycogen, so that, on the whole, there is a gradation in amount from the cells adjacent to the periblast to the heavily stained contents of the perivitelline space. There are also small granules of glycogen in the protoplasm of the underlying periblast.

Stage II (Pl. 2, fig. 2). There is the beginning of blastoderm differentiation in the position where the dorsal lip and the primitive endoderm will appear, but the blastoderm is still a number of cells deep in sagittal section, and the deposition of the glycogen is much the same as in stage I except that the perivitelline space is less deeply stained. Conditions round the yolk have not changed except that stained granules appear in certain spaces that seem to be formed by local shrinkage of yolk from the vitelline membrane.

Deposition of Glycogen in the Muscular System

It is to be considered that the distribution hereunder described in the later stages is that in larvae which have taken violent action before being caught for fixation.

Stage III (Pl. 4, fig. 14). There has been rapid development between stages II and III. The embryonic mesoderm is already separated into segmentally arranged compartments or myotomes, and certain of the derived myoblasts have started to divide and form elongated cells which span the length of the myotome. The myoblasts assume a syncytial character, and the small amount of glycogen that is present appears as isolated granules or streaks across a syncytium. These streaks do not span the myotomes in many cases, but they are elongated and therefore give the impression that they are connected with the formation of muscle cells, if not actually contained in them.

Stage IV. Although sections prepared from tissue fixed in silver nitrate at this stage show striped muscle fibres in certain restricted areas, the site of the muscular system is occupied mainly by elongated cells containing large nuclei and stretching from one myoseptum to the next.

In glycogen-prepared sections there are stained streaks, made up of closely packed granules, which seem to be in elongated muscle cells. Where these have formed such granules also appear in a more broken up state in the intervening sarcoplasm, where this is showing, but not in the myoblasts.

The embryo can now move with some vigour. It would appear from stages III and IV that glycogen is present in muscle tissue before the striped muscle fibres have been laid down.

Stage V (Pl. 4, fig. 15). Now the muscular system is complete in the portion of the mesoderm dorsal to the notochord. There are also striped muscle fibres ventral to the chord and lying close to it ; more ventral still there remain elongated muscle cells mingled with striped fibre. The embryos are capable of strong movements within the egg capsule at this stage.

In section, glycogen is seen to be packed in the connective tissue of the myosepta (which is also a storehouse for fat) and in the myotomes, particularly near these septa. It is concentrated in the sarcoplasm between the muscle fibres, but also present as finer granules, in the fibres, particularly at the posterior end of the myotomes.

Stage VI (just hatched). The system of striped muscle fibres is finally established. Under the highest powers glycogen is found as very fine grains inside the fibres, although it also lies between them. It is concentrated towards the parts lying nearest to the myosepta, as in stage V. The septa, themselves, are also packed, so that the divisions between the myotomes are shown up clearly as a deeply stained pattern under the low power of the microscope.

Stages VII and VIII. There is glycogen in the septa between the myotomes, particularly near the dorsal surface of the embryo. It is also present in the sarcoplasm between muscle fibres, but seems to be less densely packed than in stage V, although it is concentrated in and between the anterior and posterior ends of the muscle fibres near the myosepta. The granules inside the fibres are always smaller than those outside.

Stages IX-XII (Pl. 4, fig. 16) are little changed, except that glycogen is more generally distributed in the muscle fibres, which have grown so that there is little intervening sarcoplasm. Where this occurs, however, there are still deposits of glycogen.

To summarize, glycogen is early present in embryonic tissue, when fibres are being laid down in the myoblasts, and it is present in the muscular system throughout the series. In earlier stages glycogen is more concentrated in the sarcoplasm than in the muscle fibres but this condition is reversed later, except in the areas which lie near the ends of the fibres and therefore adjacent to the myosepta. It is here and in the connective tissue septa that glycogen is most heavily concentrated, and this power of retention by the septa is shared, to some extent, by the subcutis which is continuous with them. The deposition may be so heavy that it is possibly not confined to the connective tissue cells.

According to Battle (1942) the liver anlage arises early in the development of 5. salar, when the embryo is about 4-0 mm. long. The blastoderm at this stage has overgrown slightly more than three-quarters of the yolk sphere.

In the series under discussion the anlage was first found in a larva 6 ·0 mm. in length (stage III) and near it, in one or two sections there were small granular bodies which stained deep brown with iodine.

The liver is certainly present in stage IV, but this was revealed in material that had not been fixed to retain glycogen.

Stage V. It was not until stage V, therefore, that the staining reactions for glycogen showed this substance in quantity in the liver cells, although it may have been there earlier, and it was found in all the subsequent stages examined.

Stage VI (Pl. 4, fig. 17). The liver is now conspicuous and lies deeply couched on the underlying yolk. The central portions of many of the liver cells are practically filled with glycogen, which is probably at its highest concentration at this stage.

Stage VII. The distribution of the glycogen is much the same as in stage VI except that it is somewhat less concentrated in the cells. It is not possible to say whether glycogen is in the cytoplasm only or in cell nuclei as well.

Stages VIII-XII (Pl. 4, fig. 18). There is little change in these later stages except that there is not the same amount of packing in any particular cell, so that glycogen appears to have a more even distribution throughout the organ.

Differences in the general appearance of the liver glycogen in stages VI and XII may be studied in Pl. 4, figs. 17 and 18. During the period covered by stages V-XII an increase in glycogen, as recorded by Hayes & Hollett (1940), would not be due to additional amounts in individual cells, but it would be consistent with the steady increase in the size of the liver itself, as outlined by Battle (1942).

Observations do not agree with Hayes & Hollett (1940) about the time of appearance of glycogen in the liver. In their Table I is given a first estimation of 0 ·01 mg. liver glycogen 6 days after hatching, but staining reactions are strong in stage V of the present series and very similar in appearance to those in later stages, where it is agreed by them that glycogen is present.

Now stage V was reached at least 8 days before hatching, and since the times between fertilization and hatching are about equal for the two investigations, the position of stage V in developmental time would be much the same in both cases and 14 days before the date given by Hayes & Hollett (1940).

The source from which the liver glycogen has its origin may be considered. The possibility that it came from ingested food may be ruled out because, according to O’Brien & Moult, the oral plate does not break through until the embryos are 9 ·0 mm. long and the oesophagus is not capable of passing food until they are 25 mm. long. In short, there could be ho feeding, in the ordinary sense, until the last stage examined in this series.

Hayes & Hollett (1940) have demonstrated that glucose is present from the time of fertilization until the yolk-sac has disappeared and that it is certainly present in both yolk-sac and embryo from 12 days before hatching until 76 days after hatching. Although there is synthesis of glucose from fertilization until the disappearance of the yolk-sac, this is accompanied by a falling away in the total quantity, particularly in the yolk. They suggest that as the small embryo uses up glucose, it is replaced by simple diffusion from the yolk.

It may be that glucose is able to diffuse in this way through the capsule of splanchnic mesoderm which separates the liver from the adjacent yolk, and that from this source glycogen is formed in the liver cells, the energy for such transformation being derived from the glucose. The present investigation can throw no light upon the presence of such a mechanism.

A more obvious connexion between liver and yolk has been recorded in the sea bass (Serranus atrarius) by Wilson (1889), in the perch (Perea fluviatilis) by Chevey (1924) and in Salmonidae by Portmann & Metzer (1929).

As larval development proceeds, the outline of the liver next to the yolk becomes blurred, the dividing layer breaks down and liver cells in this region begin a direct absorption of yolk material.

In Salmonidae Portmann & Metzner found that in water at 7 −10 ° C. there was no direct connexion in early stages, but that 30 days after hatching complete contact had been established between yolk and liver. In Fig. 4 of their paper liver cells are shown ingesting yolk material.

Examination of sections of Salmo salar shows that there is no suggestion of direct contact between liver cells and yolk, except in the last stage examined. Even here, 43 days after hatching, with temperatures 7 −12 ° C., there is strong encapsulation of the liver, except in one localized area.

Here the splanchnopleura is very much attenuated and the adjoining yolk more vacuolated, but there is no suggestion of yolk absorption by liver cells as depicted by Portmann & Metzner (1929).

During the greater part of the development under review, therefore, there is no histological evidence that liver cells receive material, such as glycogen, by direct absorption from the yolk.

In teleost larvae the yolk, in fact, is isolated from the digestive tube, except under circumstances as mentioned above, a condition which is unlike that of vertebrates in general. Another characteristic of teleostean development is the rich supply of blood vessels which anastomose and form a complete network about the yolk-sac. It has generally been assumed, therefore, that food material is conveyed from the yolk to the embryo by means of these blood vessels. Statements to this effect have been made, for instance, by Jenkinson (1909) and by Ryder (1882). The latter claims to have observed yolk being absorbed into the circulatory system, but the: evidence does not seem to be conclusive.

These blood vessels begin to develop early, before the head fold of the embryo is formed. Movement of the heart is indicated in the salmon before stage III is reached. By the time the liver diverticula arise, the blood system of the organ is connected directly with the yolk circulation which is venous throughout embryonic life, even though in later stages, before complete absorption of the sac, a direct portal vein is formed between liver and heart.

The system may well be a means of absorbing food material from the yolk and distributing it to the embryonic tissue, particularly the liver. In transverse section, however, the veins are seen to be superficial in position. Between them and the yolk, endoderm, as well as mesoderm, has passed out and enveloped the yolk. It is only through these layers of cells that any yolk absorption by the sac circulation can take place.

In early stages after fertilization the yolk is homogeneous, but when stage IV is reached, there are indications that the more superficial part of the yolk is undergoing changes, which are reflected by a difference in appearance after fixation.

Notably in stage V this ‘rind’ has assumed an open, almost granular texture (Pl. 3, fig. 9). Thus modified the yolk reacts to staining in a different manner. It stains much more deeply with eosin and very intensely with aniline blue. It is equally well marked off from the rest of the yolk by a lighter blue belt in the general tissue stain that accompanies Bauer’s reaction.

Another indication that the globulin in the outer portion of the yolk is undergoing change is the formation of inlets or rifts, so described because that is how they appear in sections and as such are figured in the plates. They probably represent a system of tunnels formed in the yolk periphery and may be numerous, but rarely occur in yolk adjacent to the embryo.

Yolk inlets are present after fixation of larvae by Bouin-Allen or picro-dioxan. They are present, also, in larvae fixed in formalin and subsequently sectioned by freezing microtome, but are less conspicuous. These latter sections also show rounded fat globules enveloped in the yolk.

In larvae of stages I-IV fixed for glycogen, fat globules have disappeared, but rounded vacuoles mark their position and show an early distribution over the periphery of the yolk, with a subsequent massing immediately below the periblast underlying the developing embryo.

In later stages inlets and fat vacuoles may be present in close proximity, and they look quite distinctive when seen in sections. Moreover, the fat vacuoles begin to concentrate under the embryo, and few are left round the yolk-sac when the inlets are increasing rapidly to a maximum in stage VIII.

The presence of these inlets near the yolk-sac cells in stages IV-XII may be taken as an indication that absorption by these cells is taking place as the yolk diminishes in size. This absorption is accompanied by the deposition of glycogen.

From stage III onwards there are three distinct cellular layers which make up the yolk-sac envelope. First an outer epidermis of typical stratified squamous cells from which, after hatching, gland cells extrude.

Then there is an intermediate layer which takes the form of a dermis. It consists of an inner stratum of fibro-elastic tissue, and an outer one with interlacing bundles, mainly parallel to the outer surface of the envelope. It is this layer, immediately below the epidermis, which contains the blood vessels of the yolk-sac venous system. In parts of the sac, the dermis may almost disappear, as represented in stages III and V (Pl. 3, figs.. 8, 9).

The inmost tissue consists of comparatively large and sometimes quite irregularly shaped cells which may form compact layers. The width of this cellular band varies a good deal and may be reduced to one or two cell layers by the intrusion of fat vacuoles.

It is not known if they are entirely endodermal in origin, and so they will be referred to simply as yolk-sac cells.

Stage III (Pl. 3, fig. 8). In stage II the anterior and lateral margins of the blastoderm have grown over the yolk and practically covered it. The germ layers in the embryo have already differentiated and the ectoderm is continuous with the yolk-sac envelope as layers of elongated cells with the walls hardly delimited, except at the outer edge lying near the chorion.

The rest of the sac lining in fig. 8 consists of loosely knit cells, some of them large, which present a foam-like structure in which it is not possible to distinguish mesoderm from endoderm cells. This tissue forms a thin layer for the most part, but in places it widens and projects into the yolk, when it has the deposition represented in the figure.

There is evidence of glycogen. It is deposited between the larger inner cells, and the epidermis and is present also in certain epidermal cells. The inner mass of cells has small amounts of glycogen which apparently lie both inside and between the cells.

Stage IV (Pl. 2, fig. 3). Cells on the edge of the yolk are now more uniform in appearance, and certain of them contain stained granules of glycogen which are concentrated towards the centre of the cell. There are also small deposits of glycogen in yolk near to these yolk-sac cells.

Stage V (Pl. 3, fig. 9). There is glycogen in certain of the yolk-sac cells. A good deal of cell proliferation is taking place next to the yolk, and there are heavily stained granular streaks of glycogen which have the appearance, in section, of being imbedded in the yolk or, more usually, lying where there are inlets or rifts in the yolk. Reference has already been made to the disturbed superficial area of the yolk at this stage.

In addition there may be deposits of glycogen between the yolk-sac cells and the epidermis, in regions where the dermis may have almost disappeared (as in fig. 9).

Stage VI (Pl. 2, fig. 4). In this stage, taken shortly after hatching, the inlets stretching from the peripheral cells are more pronounced and glycogen may occur near the cellular regions at their bases. There are many such concentrations, however, that do not correspond to an inlet. The concentrations all lie near to yolk-sac cells, and although they may be separated from them by yolk, they are usually found in restricted areas which are bounded on one side by yolk-sac cells.

There are also small amounts of glycogen in the dermal layer.

Stage VII (Pl. 3, fig. 10). In sections the specific stains for glycogen appear as bands of colour in the yolk-sac cells. The colour is concentrated towards the centre of the cells in such areas, but there is also a general diffuse staining which almost obscures the cell outlines.

In addition there may be intensive staining between cells and yolk ; such deposits do not usually lie at the bases of the inlets and may be adjacent to clusters of small yolk-sac cells. In the latter case the cells are separated from the main glycogen deposit by an area of attenuated yolk, which contains open-spaced granules and, nearer to the glycogen, they are closer spaced. All the granules stain deeply for glycogen but are not entirely composed of it, since they are still present as unstained bodies after the glycogen has been removed.

In other cases the yolk-sac cells which border such an area are of normal size, or the area, with its granules, may be pinched off into the yolk. (These gradations are shown in fig. 10.)

Stage VIII (Pl. 2, fig. 5). The inlets near the edge of the yolk are now so numerous that they may join up and run into each other. Some of them show a staining reaction for glycogen at the inner (i.e. deeper) end.

There are also oval or pear-shaped areas in the yolk, which may be easily distinguished in section. They lie on the edge of the yolk-sac cells or may actually intrude into them. These areas are similar to those mentioned in stage VII ; they may be partly empty but always contain glycogen either diffused or as granules. Some of the areas are practically imbedded in the sac cells. In such cases they may not be much bigger than the cells themselves and are completely filled with stained material after iodine or Best’s carmine.

There is also a considerable amount of glycogen in the yolk-sac cells. The staining is very dense in the cells, or there may be stained granules scattered just inside the membrane.

Stage IX (Pl. 3, fig. it). Although in stage IX the inlets are very much decreased in numbers, the general disposition of glycogen is similar to that of stage VIII. The boundaries of glycogen containing areas on the edge of the yolk-sac cells are partly formed by the cells themselves. These, in most cases, show no diminution in size, but in cases where they are smaller, it is possible for yolk to intrude between the cells and the area.

Further, any line of demarcation may disappear so that apparently nothing is left in the yolk but a streak of glycogen.

There are, also, large amounts of glycogen in the yolk-sac cells, and in the dermis along the inner epidermal cells.

Stage X (Pl. 2, fig. 6). Deposits of glycogen are less evenly arranged. They are wider and deeper than in the above stage, but there are fewer of them. In addition, there is a more scattered distribution along the border of cells and yolk as well as in the dermis.

Stage XI (Pl. 3, fig. 12). Granules which show a glycogen-staining reaction in areas adjacent to yolk-sac cells are comparatively large in size. They are not large enough, however, to be confused with certain small yolk-sac cells that occur near these granular deposits. Deposits of glycogen are found in the yolk-sac cells, as depicted in fig. 12, and there is also a general distribution of glycogen in the dermis round the blood vessels, when these are present.

Stage XII (Pl. 4, fig. 13). The distributing of glycogen is much the same as in stage XI.

In brief, the presence of glycogen in the yolk or on the edge of the yolk-sac cells can only be reviewed here as spatial relationships revealed by the histological preparations.

In stages IV and V (Pl. 2, fig. 3 and Pl. 3, fig. 9) such glycogen appears in the open texture of the more superficial part of the yolk, but removed somewhat in space from the yolk-sac cells, so that it may have arisen independently of the activities of these cells.

There is known to be a supply of glucose in the yolk (Hayes & Hollett (1940)), and it is possible that the necessary mechanism for transformation may also be there.

In stage VI and onwards, however, deposits are always near the boundary between yolk and yolk-sac cells, and therefore more obviously bound up with the latter. In these later stages there are areas of the yolk lying in contact with the cells, which may appear, in extreme cases, as spaces containing only glycogen, but which more usually contain granular bodies of different sizes.

These areas are large compared with the cells and have no direct connexion with the yolk channels or rifts, already mentioned, which occupy a more interior position in the yolk; nor are they to be confused with fat vacuoles that occur in the same vicinity.

The outline of the area which borders the cells is made up of the walls of these cells and is, therefore, somewhat irregular, a condition which is in marked contrast with the outline in contact with the yolk. Examination of this latter outline, in unstained sections, suggests that the immediate yolk is being pressed inwards. It is against this bow-shaped compression that the greater part of the glycogen lying in the area is laid down.

In addition, there is usually in such an area an open network of material as well as the numerous granular bodies.

The larger of these lie close to the yolk-sac cells. Since the cells which border an area may be smaller than usual, it is occasionally not easy to distinguish granular body from cell on size alone, but they can be differentiated because of their different reactions to various stains.

After the use of Mallory’s connective tissue stain the yolk-sac cells are always a deep orange colour, whereas the contents of the areas and the yolk edge are blue as, incidentally, are the dermis and epidermis.

With a combination of eosin and haematoxylin, the yolk-sac cells take both stains, but the granules only the latter.

These distinctions between yolk-sac cells and the contents of the areas are marked even in those cases where the areas are practically surrounded by the cells, as they may be in stage IX.

The general plan presented by these areas, therefore, is that of the larger granules nearest to the yolk-sac cells, smaller granules farther away and finally a line of what appears to be a dense deposit of glycogen abutting on the yolk.

It might be considered, therefore, that the formation of glycogen, in the later stages of development under review, was proceeding from the yolk-sac cells, or at any rate was being activated by them. Certainly there is no parallel case of the laying down of glycogen in yolk lying near to the embryo. In addition, it may be noted that glycogen is contained in all three tissues of the yolk envelope. Deposits in the yolk-sac cells may be heavy, and to a lesser degree it is found in the dermis and the epidermis or at the common boundary of these two layers.

The present observations demonstrate that in the early development of salmon, before the gastrula stage, there is glycogen in the blastoderm cells, in the perivitelline fluid and small amounts in the periblast.

There is a parallel to this distribution in both amphibian and avian development. Woerdemann (1933) describes the blastula of Axolotl with its accumulation of glycogen in many of the cells at the animal pole, with small quantities, or none, to be found elsewhere, except in the perivitelline fluid, which latter storage place has also been recorded for the frog (Needham).

Similarly, Jacobson (1938) records considerable amounts of glycogen, in almost every cell of the blastoderm, in the blastula stage of Aves.

There is also loss of glycogen in the invaginating cells of the gastrula stage in all three groups. There may also be similarity in the time or manner in which glycogen is laid down during the formation of the muscular system and the development of the liver.

It has already been described how, in the salmon (stage III), glycogen is present as isolated granules or streaks during the laying down of the myoblasts in anticipation of the formation of striped muscle fibre. There is like anticipation in the frog and the chick.

Konopacki (1924) records that in the frog glycogen occurs as irregular masses or grains in the myomere cells. During the elongation of the myoblasts and formation of myofibrils, glycogen increases and is localized along the fibrils as small grains or irregular masses in non-differentiated sarcoplasm.

In the chick, according to Allen (1919), glycogen is present in the myotomes 56 hr. after fertilization. Jacobson (1938) states of the chick that when the somites appear glycogen occurs only in the dorsomedial part which probably represents the presumptive myotome tissue.

Differences in the disposition and concentration of glycogen in early and later stages of salmon liver have already been noted. These apparently have their parallel in the chick, for Dalton (1937) has recorded that glycogen is first homogeneous and then in later livers more widely distributed and granular. The disposition of liver glycogen in his Fig. 4 (8-day embryo) and Fig. 3 (17-day embryo) is similar to that in stage VI (Pl. 4, fig. 17) and stage XII (Pl. 4, fig. 18) of the salmon respectively.

In spite of such similarities, however, there are wide divergencies in the carbohydrate metabolism, including glycogen, of the three types. Evidence gathered by Needham (1931, 2) shows that in the developing chick glucose is strongly attacked and glycogen left intact, whereas in Amphibia during development while there is only 7% loss in total carbohydrates, 41% of the original glycogen disappears.

Hayes & Hollett (1940) have demonstrated that the developing larva of salmon finishes up with more glucose than when it starts, and they find this to be true also for glycogen, a point that is in agreement with this paper. In the later stages of development glycogen actually increases more rapidly than glucose, so that about 70 days after hatching it is twice the weight per larva.

Nevertheless, so far as the present investigation is concerned, there is no increase of total glycogen in salmon from 44 to 9 days before hatching. There is a certain similarity at this period with the Amphibia, in which, however, glycogen falls from fertilization to hatching, in that the amounts involved in both cases are comparatively small, being 0 ·08 mg. per egg for the earliest stage in the frog, according to Brachet & Needham (1935), and 0 ·06 mg. per egg in the salmon.

Hayes & Hollett (1940) have pointed out that the complete cleavage of the amphibian egg, with its dividing up and sharing out of yolk material, has set up an entirely different set of conditions from those to be found in the meroblastic egg of the teleost. They believe one consequence to be a difference in the form of carbohydrate initially present. In the Amphibia it is glycogen, whereas in the salmon it is free glucose which is more diffusible and therefore better suited to the more complicated task of yolk distribution presented by a meroblastic egg.

This does not preclude the presence of extra-embryonic glycogen in the meroblastic egg, however, for chemical estimations and histological preparations have shown it to be present in both the chick and the salmon during development, and it is practically the only source of glycogen in the former at the beginning of the incubation period. Allen (1919) has shown, by staining methods, that this condition is true for the chick from the earliest stages to 10 days after fertilization. It is concentrated in that part of the yolk-sac nearest to the embryo, the embryonic shield, which corresponds,most nearly, in development, to the yolk-sac tissue of the salmon.

In addition, both the chick and the salmon have a well-developed yolk-sac circulatory system to assist in the mechanism of yolk absorption. Allen concluded from these conditions in the chick that ‘carbohydrates are stored as glycogen on their way from the yolk to the embryonic tissue’.

Concerning the salmon, there are fat globules in the yolk-sac cells as well as glycogen, so that this tissue, which lies between the yolk and the blood system of the yolk-sac, has a storing or regulating function.

Although the storage of glycogen is less concentrated than in liver cells, there may be heavy deposition in or near the yolk-sac cells, and this is accompanied by disturbances in the peripheral part of the original homogeneous yolk.

Further, the presence of concentrated glycogen (and fat) in the myosepta of the developing embryo suggests a connexion between the laying down of this product and the blood system, since connective tissue, in general, receives material from, and gives up waste material to, blood capillaries.

The conclusion, therefore, that as the growing salmon embryo uses up glucose this is replaced by diffusion from the yolk, is based upon an over-simplification of the process that is actually taking place.

It has already been pointed out that the yolk mass in the Salmonidae is closely linked to the liver by its vascular supply.

This condition is shown clearly by Portmann & Metzner (1929) in their Fig. la. Blood from the liver traverses the yolk-sac and then runs to the heart. In the trout it is only 40 days after hatching that there is direct venous connexion between liver and heart.

In the development of certain animals, the embryonic liver does not store up glycogen in the early stages, so that this function is taken on by other tissues which may be extra-embryonic. In general, as soon as it is possible for liver cells to store glycogen, this wider accumulation no longer takes place ; it has only been a transitory phase.

What, then, is the position in the salmon embryo? According to Hayes & Hollett (1940), small quantities of liver glycogen were estimated upon the 6th day after hatching, although, in the text, the liver is not accepted by them as a storage organ until 30 −40 days after hatching. In their discussion (p; 61) these workers draw attention to the sudden appearance of glycogen in the liver some 40 days after hatching.

Neglecting, as they do, their own figures indicating traces of liver glycogen, then the stage in our histochemical series corresponding to its appearance is the last one, stage XII. There is, indeed, a large quantity of glycogen displayed in the liver at this stage, as well as in the muscles, but, in addition, sections and estimations show that glycogen is still present in the yolk-sac cells and in yolk adjacent to them.

Liver glycogen has not been estimated separately in the present investigation, but if the earliest figures given by Hayes & Hollett (1940) in their table have any significance, then this overlapping of the glycogenic functions of liver and yolk-sac may be pushed as far forward as stage VII, in which there is clear histochemical indications of glycogen in both areas. After this stage the amount of glycogen in the yolk-sac actually increases.

It seems possible to take this overlapping even further. Reaction by liver tissue to fixation and staining methods is so characteristic that what is acceptable as liver glycogen in the sections of stage XII is equally acceptable as early as stage V, which occurred about 5 days before hatching.

The functioning of cells in the yolk-sac, therefore, overlaps the storage of glycogen in the liver. Like the cells of the latter organ, they are absorbing food material and, at the same time, being supplied by a blood supply which is venous. In both cases a certain amount of glycogen may be primarily necessary to enable them to carry out food absorption in the presence of small amounts of oxygen.

The yolk-sac cells are formed originally from the germ layers which spread and enclose the yolk. When the latter is absorbed they form part of the true embryo again, and in this sense the yolk is inside the embryo throughout the various stages. While they encircle the yolk these tissues function as a ‘supplementary’ rather than a ‘transitory’ liver, linked closely to the true liver by the venous system.

1. Quantitative estimations and histological methods have been used to determine the presence and distribution of glycogen in fertilized salmon eggs and subsequent stages of development.

2. A check upon the occurrence of glycogen in each sample was obtained by the use of amylase and the presence of glucose as a result of this technique confirmed, in certain large samples, by the formation of phenylglucosazone crystals.

3. Results of estimations agree with the general distribution of glycogen as shown in histological sections. They differ from those of Hayes & Hollett (1940) who, using water extraction, found no glycogen in stages corresponding to stages I and II and only recorded it as doubtfully present in the yolk-sac.

4. In stages I and II glycogen is concentrated in the blastoderm and perivitelline space. Later the main sources are the muscles, liver and yolk-sac envelope.

5. Glycogen is present in embryonic muscle tissue when fibrils are being laid down (stage III). Subsequently it occurs in both sarcoplasm and muscle fibres.

6. Eight days before hatching (stage V) there is strong staining reaction for glycogen in liver cells, and it is present in all later stages. This glycogen is not obtained from engulfed food or from direct absorption of yolk by liver cells.

7. There is histological evidence that yolk is taken up by the yolk-sac blood vessels after absorption by the yolk-sac cells and dermis. This absorption is accompanied by the appearance of glycogen in these cells and in yolk lying adjacent to them.

8. An increase in amounts of glycogen in both embryo and yolk-sac coincides with a rapid absorption of fat which takes place about 20 days after hatching.

9. The presence of glycogen in blastoderm cells, before the gastrula stage, is similar to the condition in developing Aves and Amphibia. This is true, also, for its appearance early in the formation of muscle tissue. Other similarities are the presence of glycogen in the perivitelline fluid of salmon and Amphibia and the manner of its distribution in the liver of early and late stages of the salmon and chick.

10. There is extra-embryonic glycogen present in the meroblastic eggs of both the salmon and chick. It is concentrated in the embryonic shield of the latter which corresponds, in development, to the glycogen-carrying envelope of the salmon yolk-sac. Accompanying these deposits in both cases is a well-developed circulatory system to assist in the absorption of yolk.

11. The presence of glycogen in the yolk-sac cells of the salmon refutes a suggestion that the embryo receives glucose only by direct diffusion from the yolk.

12. The liver cells and yolk-sac cells overlap in the function of laying down glycogen during development. The latter, therefore, do not form a ‘transitory’ but rather function as a ‘supplementary’ liver, connected with the liver by a venous system until the cells rejoin the true embryo after complete yolk exhaustion.

I wish to thank Prof. R. A. Morton for allowing me to work in the Department of Biochemistry, University of Liverpool and Miss M. J. Durham for valuable assistance in checking the examination of histological preparations.

The distribution of glycogen is shown by black dots and masses. All figures are drawn from sagittal sections. Fixatives and stains listed were used for the particular section represented in the figure. They were not the only fixatives and stains used in each stage.

The number of days before or after hatching is given for each stage. The intervals differ from those in Text-fig. i, in which the stages have been fitted into the different developmental rate of samples from a later Season.

Figures drawn from low-power projection through A.A. objective. Final Magnification × 8 ·8.

Microscope. Eyepieces. Objective $14$ in. Final Magnification × 270.