The mechanism of secretion of the milk fat globule

Despite the considerable amount of research on the chemical composition of the milk fat globule membrane (MFGM) (see Brunner, 1969, for a review), few high-resolution morphological studies of its origin and fate have been carried out since Bargmann & Knoop (1959) first suggested that the MFGM was identical with the plasma membrane of the secretory cell. A recent study (Keenan, Morré, Olson, Yunghans & Patton, 1970) showed that the plasma membrane fraction of bovine mammary secretory tissue was almost identical in chemical composition with the MFGM fraction isolated from milk. The morphology of the 2 fractions was not identical, however, and Keenan et al. suggested that some change of state occurred on secretion of the MFG. It has been reported that a small, though significant percentage of MFG’s in the milk of various species include a cytoplasmic crescent, and that this, at least, is bounded by a unit membrane (Wooding, Peaker & Linzell, 1970). The present study demonstrates at high resolution the origin of the large majority of MFG’s which do not contain an obvious cytoplasmic crescent and the mechanisms of the release of these MFG’s from the secretory tissue.

Mammary tissue from goats, guinea-pigs and a Jersey cow was either excised and cut into small cubes in the glutaraldehyde fixative or fixed initially by perfusion via the dorsal aorta (guinea-pig) or mammary artery (goat). All the processing was carried out at room temperature. The fixation was in 4% glutaraldehyde in 0·1 M phosphate buffer, pH 7·2, containing 2% sucrose, for 45 min. The tissue was washed briefly in buffer, postfixed first in 1 % osmium tetroxide in 0·1 M veronal buffer, pH 7·2, for 30 min, then in 5 % aqueous uranyl acetate for 2 h followed by ethanol dehydration and embedding in Araldite. Sections were cut on an LKB Ultrotome, stained with uranyl acetate and lead hydroxide and observed in a AEI EM6B microscope operated at 60 kV.

The fine structure of mammary tissue at the height of lactation from cow, goat or guinea-pig has been found to be almost identical. The micrographs in this paper are from goat and cow, but the description would apply equally well to the guinea-pig.

After the processing for electron microscopy used in this study, fat is represented by electron-transparent (light) areas on the micrographs.

Fat droplets are seen throughout the cytoplasm of a secretory cell (Figs. 1 and 3-5). They are not bounded by a membrane as are the Golgi vesicles in which the protein granule component of the milk secretion is produced (Fig. 6). The cytoplasm of the secretory cell is packed with rough endoplasmic reticulum (ER) and free ribosomes (Figs. 1, 5). Thus, wherever a fat droplet is found it will be in a very close relationship to rough ER cisternae, but droplets have not been seen within the latter. The smallest droplets are found near the basal plasmalemma, which is considerably infolded in all 3 animals (Figs. 1, 2, 4). The droplets usually have a zone free of rough ER around them which contains scattered small vesicles (Figs. 3, 4). Such vesicles frequently have a short bristle coat on their cytoplasmic surface (Figs. 3, 4) and are not arranged in any definite association with the droplet. They are also found in the apical cytoplasm.

The largest fat droplets are found towards the apex of the cell, although lipid droplets of all sizes are found at this level (Figs. 5, 15, 24). The droplets are closely surrounded by numerous vesicles, some containing protein granules derived from the Golgi apparatus, some smaller, with a clear content bounded by a membrane approximately 9–10 nm wide, as are the Golgi vesicles (Figs. 6, 7, 12). The membrane of the rough ER is significantly thinner - in the range 6–7 nm. The bristlecoat vesicles have not been observed in such a close relationship with the lipid droplets.

The apical cytoplasm contains many vesicles, most of which are separate. Those very close to the fat droplets may show an alteration of contour where their boundary is common with that of the fat droplet (Figs. 5, 6, 11, 14). High magnification of such common boundaries (Figs. 7, 12) shows that the unit membrane of the vesicle bears a narrow zone of material toward the fat droplet and is identical with the membrane around the freshly secreted globule (Fig. 21).

Droplets bulging further into the lumen of the alveolus (Figs. 9, 11, 14) are separated from that lumen by the plasmalemma plus a layer of material 10–20 nm thick. This material often has localized thickenings (Figs. 10, 11) and is continuous with and similar in appearance to the cytoplasm which usually forms an abrupt thickening at the edge of this layer (Figs. 9, 11, 14). This forms a fold of cytoplasm which may contain one or two vesicles (Figs. 9, 11). The vesicles which were present between the fat droplet and the plasmalemma do not seem to accumulate in these cytoplasmic folds. At the base of the fold, where the MFGM is forming, profiles are seen which could be an active inpocketing of the plasmalemma or the peripheral vesicle of the droplet fusing with the plasmalemma (Fig. 13). Sharing the cytoplasmic boundary of most of the droplets which bulge into the lumen of the alveolus there are vesicles and flattened sacs which often contain casein granules (Figs. 11, 14, 15). At the edges of the alveolar boundary of the emerging lipid droplet deep clefts are sometimes seen running down beside the droplet. These clefts are lined by plasmalemma and often contain casein granules (Fig. 14). On the globule side of the flattened sacs and clefts the unit membrane bears the material characteristic of the MFGM (Figs. 12, 13). This material remains remarkably constant in width as the droplet emerges into the alveolus (Figs. 17–19). An occasional local thickening is produced which may contain particles identical with the ribosomes in the cytoplasm (Figs. 17, 18), but the cytoplasmic content is minimal. Fat droplets may emerge side by side, in which case the plasmalemma plus cytoplasmic layer is separate for each (Fig. 22). Profiles have also been observed of one droplet immediately underneath the emerging one (Fig. 23), and ‘figure of eight’ globules have been observed on occasions in expressed milk, suggesting that the 2 droplets have been released together.

Micrographs with lipid droplets very close to the apical plasmalemma occasionally show an abrupt inpocketing of the plasmalemma down to the surface of the droplet, where the plasmalemma forms a structure identical with the MFGM. Casein granules are frequently found in these inpocketings (Fig. 8).

An individual droplet close to the secretory cell may be connected by a cytoplasmic stalk of varying width (Figs. 16-18). This may contain mitochondria or rough ER, but usually only a few ribosomes and membrane-bounded vesicles. Close examination of the edges of the stalk under the plasmalemma has not demonstrated any organized structure which might account for the pinching off of the droplet (Figs. 17, 18). The stalk is usually central but profiles sometimes show an eccentric location (Figs. 16, 17), and sometimes there is an appreciable amount of cytoplasm on the fat droplet side of the constriction in the bridge to the cytoplasm (Figs. 16, 18, 20).

The range of droplet sizes in the alveolus and expressed milk is the same as that seen within the tissue. No evidence for fusion of fat droplets to form a preferred size of droplet for secretion has been observed: different sizes are secreted simultaneously from the same cell (Fig. 24).

Free fat droplets in the lumen of the alveolus immediately adjacent to the secretory epithelium have a continuous plasmalemma and a thin, 10–20 nm layer of material bounding the fat (Figs. 21, 24). The plasmalemma plus this layer thus constitute the milk fat globule membrane immediately after release from the secretory cells (Fig. 21). It must be remembered that those apparently free globules adjacent to the secretory cells might well be connected to cells out of the plane of section. However, sectioning at different levels has shown that in the case, for example, of free globules, such as in Fig. 24, the majority are not so connected.

In such free globules the MFGM is indistinguishable from the plasmalemma plus cytoplasm of the emerging droplet. No sudden change in structure has been observed on release from the secretory cell (compare Figs. 20 and 21).

The infoldings of the basal plasmalemma seem considerably greater and more regular in cow and goat than in the rat and mouse (Bargmann & Welsch, 1969). This increase in surface area is presumably correlated with the rapid rate of passage of nutrients across the plasmalemma into the cell. The smallest fat droplets are usually found near the base of the cell, but no indication of smooth (Kurosumi, Kobayashi & Baba, 1968) or rough endoplasmic reticulum specifically associated with the droplets has been observed. The origin of the small bristle-coat vesicles frequently found in the neighbouihood of the droplets is obscure. They are very similar to pinocytotic vesicles and could be involved in synthesis or transport of the fat. The majority of the triglyceride of the droplet has been shown to be formed from fatty acids from the circulation (West, Annison & Linzell, 1967) and synthesized in the mammary gland by enzymes associated with the rough endoplasmic reticulum (Stein & Stein, 1967). How or why the nucleation of a fat droplet occurs is not known.

The use of epoxy resin for embedding results in a far smaller extraction of lipid during processing than does the use of methacrylate resins (Cope & Williams, 1968). Epoxy resins also preserve the fine-structural detail far more accurately, with much less distortion. The fight areas in sections embedded in epoxy resin are therefore considered to be a more accurate representation of the fat droplets in vivo than the distorted dark areas produced after methacrylate embedding (Bargmann, Fleischauer & Knoop, 1961).

With electron micrographs it is impossible to be certain of a suggested time course of events. If it is reasonable to assume that the extrusion of a fat droplet is essentially identical in mechanism in each case, then a hypothetical sequence can be assembled from the micrographs presented here.

As the fat droplet increases in size the numerous membrane-bounded vesicles in the apical cytoplasm come up against its surface. Cytoplasmic streaming and Brownian movement should cause separation of the 2 surfaces. It is suggested that long-range London-Van der Waals forces plus electrostatic forces act to prevent this, producing the specific association illustrated on the micrographs. In the area of association the vesicle membrane and its layer of material continuous with the cytoplasm are identical to the freshly secreted MFGM. Thus the MFGM appears to be partially formed when the lipid droplet is still within the cytoplasm.

Most of the vesicles in the apical cytoplasm may be tentatively identified as having originated from the Golgi region by the width of their bounding membrane, significantly greater than that of the endoplasmic reticulum. Such vesicles frequently also contain casein granules and bear no ribosomes, thus aiding identification. The Golgi vesicles may be of considerable size yet very few have a common contour with other vesicles; such association seems to be restricted to that formed with the fat droplets. Fat droplets do not always have associated vesicles, but in the majority of sections observed this is the usual finding. Lack of peripheral vesicles on any one section does not mean they are not present at different levels on the same droplet. The fusion of Golgi vesicles with the apical plasmalemma and release of their contents has been observed many times (see Wellings, 1969, for a review). If the fusing Golgi vesicle has a common contour with a fat droplet, then that common contour becomes the first part of the MFGM of the droplet. The crater produced in the plasmalemma by the usual Golgi vesicle fusion is smoothed out, presumably by the elasticity of the plasmalemma and the viscosity of the cytoplasm. Further fusion of lipid-associated vesicles with the plasmalemma continues the process. The small cytoplasmic blebs occasionally found in the MFGM would represent cytoplasm originally between 2 vesicles which had been isolated from the rest of the cytoplasm by their fusion. The lipid droplet would then gradually emerge from the cytoplasm as successive vesicles fused with the plasmalemma.

Nothing is known of the time course of the extrusion process. If it is dependent upon the number of suitably associated Golgi vesicles it may well be discontinuous, with the droplet remaining partly extruded until vesicles come into the correct position to continue the process. This could explain the range of size found in secreted fat droplets (Brunner, 1965) and the fact that droplets of very different sizes may be secreted simultaneously from the same cell. Profiles showing lipid droplets practically touching yet not fusing on secretion (Fig. 22) suggest that coalescence of droplets is unusual. It is more normal to find contiguous droplets separated by peripheral vesicles (Fig. 23). The forces causing fusion of these vesicles with the plasmalemma would cause the separate secretion of each droplet. These forces do not explain the remarkable uniformity in width of the layer of material on the fat droplet side of the MFGM. Such a layer has been reported by Bargmann et al. (1961) and by Helminen & Ericsson (1968), but not illustrated in any detail. Bargmann et al. used methacrylate embedding and found a light gap between outer membrane and the electron-dense fat droplet. Thus the material is unlikely to be an outer shell of lipid or phospholipid adsorbed by the plasmalemma from the fat droplet. In the cytoplasm where a Golgi vesicle is associated with a fat droplet or at the edge of a fat droplet emerging into the alveolus the cytoplasm is continuous with and identical to the material, just as it is in the localized blebs of cytoplasm occasionally found on the MFGM. The very uniform thickness of the layer suggests that some short-range forces are holding the material to the plasmalemma and/or fat droplet. Patton & Fowkes (1967) have suggested that the attractive London-Van der Waals forces between the lipid phases of plasmalemma and fat droplet would be sufficient to cause expulsion of a water phase (or cytoplasmic solution) from between the 2 surfaces. The plasmalemma would thus automatically adsorb to the fat droplet, which would be ‘pulled ‘out of the cytoplasm by the attraction between the 2 surfaces. To initiate this process Patton & Fowkes calculated that the two surfaces need to come within about 2 nm of one another, since the London forces are only effective at very short distances.

The plasmalemma and lipid of the droplet always appear to be separated by the 10-20 nm zone of dense material. Thus it seems unlikely that sufficient force could be generated to be the sole cause of extrusion of the droplet. The calculation assumes that electrostatic surface charges play no role in the attraction. As yet there is insufficient evidence that this assumption is justified. However, there seem to be sufficient attractive forces between fat droplet and membrane (be it Golgi vesicle or plasmalemma) to cause a strong specific association at a certain minimum distance apart. This would explain the association between fat droplet and Golgi vesicles or plasmalemma. The evidence presented in this paper suggests that the progressive fusion of the droplet-associated Golgi vesicles with the plasmalemma allied with cytoplasmic viscosity would provide one major force for droplet secretion. The adsorption of the plasmalemma to the fat droplet may play an equal part in the secretory process, but the nature of the zone of dense material between plasmalemma and droplet has to be established before any definite conclusion can be reached.

The morphological evidence (this paper; Bargmann et al. 1961; Helminen & Ericsson, 1968) indicates a cytoplasmic derivation of the dense material but the difficulty of establishing the exact boundary of the droplet means that a significant amount of fat may also be present in vivo.

The milk fat globule membrane just after secretion thus consists of a continuous unit membrane plus a zone of dense material separating it from the fat globule itself. The product of secretion therefore is bounded by a cell structure unrelated in origin to the secretory product, the fat droplet. It seems useful to distinguish this from secretion in which only the secretory product is released from the cell. Kurosumi (1961) has suggested an exact definition of various types of secretion, related to the way in which the secretory product is released from the cell. Using his scheme the release of non-membrane-bounded casein granules by exopinocytosis is merocrine and the packaging of the fat within a unit membrane sac would be apocrine. Bargmann et al. (1961) consider that the teim apocrine should be dropped, largely because of its historical association with the idea of a general decapitation of a cell rather than a specific process of fat secretion. The specific nature of the secretion of fat is now generally accepted and thus retention of the term apocrine to indicate a constant loss of part of the secretory cell structure with the secretory product is considered justifiable.

I am grateful to Dr A. D. Bangham, Dr J. L. Linzell and Dr M. Peaker for helpful discussions, and to Drs Linzell and Peaker for the surgery involved in the perfusion fixation. Mrs K. Cooper provided invaluable technical assistance.

 
• b

  basement membrane of secretory epithelium

 
• d

  fat droplet

 
• e

  arrays of rough endoplasmic reticulum

 
• g

  Golgi vesicle

 
• gb

  Golgi body

 
• I

  lumen of secretory alveolus

 
• m

  mitochondrion

 
• my

  myoepithelial cell

 
• n

  nucleus

 
• p

  infolded basal plasmalemma of secretory cell