Immunoelectron-Microscopic Studies of Endoplasmic Reticulum-Golgi Relationships in the Intracellular Transport Process of Lipoprotein Particles in Rat Hepatocytes

Home , Endoplasmic reticulum, Intracellular transport

J. Cell Sci. 39, 273-290 (1979) 273 Printed in Great Britain © Company of Biologists Limited

IMMUNOELECTRON-MICROSCOPIC STUDIES OF ENDOPLASMIC RETICULUM-GOLGI RELATIONSHIPS IN THE INTRACELLULAR TRANSPORT PROCESS OF LIPOPROTEIN PARTICLES IN RAT HEPATOCYTES

SHIRO MATSUURA AND YUTAKA TASHIRO* Department of Pliysiology, Kansai Medical University, 1 Fumizonocho, Moriguchi-shi, Osaka 570, Japan

SUMMARY Endoplasmic reticulum (ER)-Golgi relationships in the intracellular transport process of secretory proteins in rat hepatocytes have been studied using lipoprotein particles as a marker for the secretory protein and cytochrome P-450 as a marker enzyme for the ER membranes. Ferritin immunoelectron-microscopic observation revealed that, while almot-t all the microsomal vesicles derived from ER membranes are heavily labelled with ferritin anti-cytochrome P-450 antibody conjugates, labelling of the small peripheral vesicles containing lipoprotein particles, the stacks of Golgi saccules, especially the outermost saccule which is sometimes fenestrated, condensing vacuoles in the trans-Golgi region and the secretion droplets of lipoprotein were scanty and at the control level. Such a characteristic pattern of labelling was especially evident when these structures were prepared from phenobarbital-treated rats. These findings indicate that the membranes of the small peripheral vesicles do not contain cytochrome P-450 and that the cytochrome is probably not transferred to Golgi saccules in the transport process of lipoprotein from ER to Golgi. It is suggested, therefore, that the small peripheral vesicles are formed by budding of the special regions of ER membrane where microsomal marker proteins such as cytochrome P-450 are excluded and the membrane proteins destined to the Golgi complexes are clustered. It is also shown that lysosomal membranes are not labelled with the anti P-450 antibody conjugates.

INTRODUCTION It has been generally accepted that the pathway followed by secretory proteins leads from rough endoplasmic reticulum (ERj to the transitional elements of this system, then to the small peripheral vesicles (Golgi transporting vesicles) on the cis side of Golgi complexes and finally either to condensing vacuoles or to Golgi stacks (Palade, 1975; Jamieson & Palade, 1977). In order to make such a transport process possible, membrane should flow from ER to Golgi complexes (Palade, 1975; Jamieson & Palade, 1977). Recent biochemical evidence indicates, however, that there is no mixing among either the lipid (Keenan & Morre, 1970; Meldolesi, Jamieson & Palade, 1971) or the • Mailing address: Dr Yutaka Tashiro, Department of Physiology, Kansai Medical University, 1 Fumizonocho, Moriguchi-shi, Osaka 570, Japan. 274 S. Matsuura and Y. Tasliiro protein (Bergeron, Ehrenreich, Siekevitz & Palade, 1973; Van Golde, Fleischer, Fleischer, Azzi & Chance, 1971) components of the membranes of the 2 compartments in the pancreas and in the liver. This imposes stringent limitations on membrane interactions since it suggests that lateral diffusion of components is prevented at the time when the membrane of the 2 compartments establishes continuity. In order to clarify the molecular mechanisms involved in the ER-Golgi relationships, we have investigated the transport process from ER to Golgi of lipoprotein particles in rat hepatocytes, since the particles are easily identified by electron microscopy, while cytochrome P-450, a marker enzyme for the ER membranes, can be localized by ferritin immunoelectron-microscopy (Matsuura, Fujii-Kuriyama & Tashiro, 1978). In addition, lipoprotein particles in hepatocytes have been exten- sively studied biochemically as well as electron microscopically by a number of investigators, such as Stein & Stein (1967), Jones, Ruderman & Herrera (1967), Claude (1970), Ehrenreich, Bergeron, Siekevitz & Palade (1973) and Glaumann, Bergstrand & Ericsson (1975) and Bergeron, Borts & Cruz (1978). In the previous paper (Matsuura et al. 1978) we have shown that, although almost all of the microsomal vesicles derived from ER membranes are markedly labelled, Golgi saccules and condensing vacuoles in trans-Golgi regions are hardly labelled with the ferritin antibody conjugates. An important question is whether or not the membranes of the small peripheral vesicles carrying lipoprotein particles have the cytochrome. The most direct approach to this problem would be to localize the antigens in ultrathin sections of the cells. Experimental approaches using frozen thin sectioning as reported by Painter, Toku- yasu & Singer (1973), Tokuyasu (1973), Sasaki & Tashiro (1976) as well as using embedding in albumin as reported by McLean & Singer (1970) and Kraehenbuhl, Racine & Jamieson (1977) have been tried without success. In the latter procedure the antigenicity of the cytochrome appears to be lost during embedding in concentrated albumin by polymerization with glutaraldehyde. We have, therefore, used cell organelles prepared by cell fractionation techniques for localization of antigens. The present immunoelectron-microscopic observations revealed that the small peripheral vesicles containing lipoprotein particles are not labelled with the ferritin antibody conjugates. This means that, when the transfer vesicles are formed on ER membrane, cytochrome P-450 molecules are excluded in advance from those regions of the ER membrane where the vesicles are to be formed by budding mechanisms. The physiological significance of such a finding with regard to the membrane flow accompanying the intracellular transport process of secretory proteins is discussed.

MATERIALS AND METHODS Immunochemical procedures Preparation of the monospecific antibody to phenobarbital-induced cytochrome P-450 has been described previously (Matsuura, Fujii-Kuriyama & Tashiro, 1979). Ferritin and the antibody were coupled according to the procedure of Kishida, Olsen, Berg & Prockop (1975) using glutaraldehyde as a coupling agent, and the ferritin anti P-450 antibody conjugates with the Intracellular transport of lipoprotein 275 molar ratio of immunoglobulin G (IgG) to ferritin of approximately unity were isolated by gel filtration on Bio-gel ArjM as described previously (Matsuura et al. 1978).

Preparation of Golgi and microsome fractions Male non-starved rats (Sprague-Dawley) weighing ~2oo g were used, since the recovery of the Golgi apparatus was higher in these than in starved animals (Glaumann et al. 1975). Golgi fraction was prepared either from untreated or from phenobarbital-treated rats. In the latter case, rats were given a single daily intraperitoneal injection of phenobarbital (10 mg per 100 g body weight) for 4 days and were sacrificed without fasting. The procedures reported by Hino, Asano, Sato & Shimidzu (1978) were used for the preparation of Golgi fraction. Total, smooth and rough microsome fractions were prepared as described previously (Matsuura et al. 1978).

Labelling of t/ie Golgi and microsome fractions The direct ferritin antibody method was used exclusively. The cell fractions were incubated for 30 min at 4 °C either with the antibody conjugates or with the control conjugates as described previously (Matsuura et al. 1978). The concentration of the specific antibodies in the antibody conjugates was adjusted to at least ~3- to 5-fold molar excess over the biochemically calculated number of cytochrome P-450 molecules in each fraction.

Electron microscopy The incubated materials were washed by sucrose density gradient centrifugation (Matsuura et al. 1978). The membrane fractions were resuspended and centrifuged at 10000 g for 10 min, and the thin pellets were fixed with glutaraldehyde—osmium tetroxide mixture, dehydrated and embedded in Epon and sectioned as described previously (Matsuura et al. 1979). Samples of liver tissue from normal rats were fixed in 1 % OsO4 in distilled water, pH 6 (Claude, 1970; Ehrenreich et al. 1973), dehydrated, embedded and sectioned as above. The thin sections were observed with a Hitachi HU-12 electron microscope.

RESULTS In situ localization of lipoprotein particles in rat hepatocytes In rat hepatocytes, the Golgi complexes are less orderly constructed and less clearly polarized than in other protein-secreting cells, as described by Ehrenreich et al. (1973). The stacks of cisternae (saccules) and the vacuoles located on the trans side have the usual Golgi features and are accordingly easily recognized, as shown in Fig. 1. Transitional elements of the type regularly found on the opposite or cis side of the stacks in other secretory cells, such as pancreatic exocrine cells, are, however, rarely encountered: their place is usually taken by irregularly distributed smooth ER elements. Fig. 1 also shows that in the Golgi apparatus, the peripheral dilated portion of the cisternae (star) and the trans Golgi vacuoles (condensing vacuoles) are loaded with a number of electron-opaque particles, 30—80 nm in diameter. The occurrence of such particles in the Golgi elements of mammalian hepatocytes has been described repeatedly and they have been identified as very low density lipoprotein (VLDL) particles or their precursor particles (Jones et al. 1967; Hamilton, Regen, Gray & LeQuire, 1967; Chapman, Mills & Taylaur, 1972). In addition to those found in Golgi complexes, similar particles are present singly or in pairs in the smooth ER in tubular configuration, as well as in the extended S. Matsuura and Y. Tashiro

t i '

r 'S

Fig. i. Lipoprotein particles in hepatocytes of non-starved rat. A Golgi complex appears in the vicinity of the cell surface. Lipoprotein particles mark trans-Golgi vacuoles (v) and a peripheral dilated rim (asterisk) of the Golgi saccules (s). Similar particles are also found singly or in pairs in the smooth ER (er) in tubular configuration and rarely in cis-Golgi vesicles (arrowheads). The arrow marks a lipoprotein particle in a small pit which is probably in the process of formation of small peripheral vesicles by budding of ER membrane, x 68000. Fig. 2. Total microsome fraction from rat liver incubated with ferritin anti P-450 antibody conjugates. Lipoprotein particles are found in the intracisternal space of both smooth (arrowheads) and rough microsomes (arrows), x 62500. Intracellular transport of Upoprotein 277 portion of the smooth ER connected with the distal ends of rough ER and also, although rarely, in the rough ER. But nothing like the accumulation of the particles seen in Golgi vacuoles is encountered in the ER. These findings are essentially in agreement with the observations by Jones et al. (1967), Claude (1970), Ehremeich et al. (1973) and Glaumann et al. (1975). The electron-microscopic observations suggest that the lipoprotein particles are transported from ER to Golgi complex via small peripheral vesicles formed by budding of smooth ER membranes (Fig. 1, arrow). This suggestion is also supported by the existence of small peripheral vesicles containing a lipoprotein particle in the cis Golgi region (Fig. 1, arrowheads).

Lipoprotein particles in the microsome fraction Fig. 2 shows the total microsome fraction incubated with the ferritin antibody conjugates. Some of the microsomal vesicles contain electron-opaque lipoprotein particles, 20-80 nm in diameter, as reported by Glaumann et al, (1975) and Matsuura et al. (1978). The particles appear in various microsomal vesicles. One is in rough microsomes which are studded simultaneously with ribosomes and ferritin particles as indicated by arrows, and the other is in smooth microsomes which are labelled only with ferritin (arrowheads). They are usually round or oval and variable in diameter. In the smooth microsome fraction, the lipoprotein particles appear in the conven- tional smooth microsomal vesicles labelled with ferritin as shown in Fig. 3 (vesicle nos. 7-j). They are also found occasionally in the smooth vesicles which are not labelled with ferritin particles as shown in Fig. 3 (vesicle nos. 4, 5) and Fig. 4. Fig. 4 is rather exceptional, because all the vesicles containing lipoprotein (nos. 7-5) are not labelled with ferritin particles. A number of these non-labelled vesicles were observed and their average diameter was calculated to be 122 + 25 nm (Table 1). They contain an electron-opaque lipoprotein particle, ~ 60 nm in diameter and sometimes flocculent materials were found in the gap between the particle and the membrane (See Fig. 4, vesicle nos. 2, j, 4). Occasionally the outer surfaces of the vesicles are coated with fine fibrous materials as shown in the inset to Fig. 4 (arrowhead). It is also to be noted here that the membranes of the unlabelled vesicles containing lipoprotein particles appear to be more fragile than those of the labelled microsomal vesicles, because occasionally broken vesicles are found, as shown in Figs. 3 and 4 (arrowheads).

Lipoprotein particles in isolated Golgi complexes We prepared a Golgi fraction according to the procedure of Hino et al. (1978). In this procedure rat liver cells were disrupted simply by straining through a ioo-mesh stainless steel sieve using a bamboo spatula. The structural organization of the Golgi complexes thus obtained was well preserved as reported previously (Hino et al. 1978), even after incubation with the antibody conjugates and washing by sucrose density gradient centrifugation (Matsuura et al. 1978). As shown in Figs. 5 and 6, the Golgi fraction contains a number of Golgi complexes 5. Matsuura and Y. Taskiro

Figs. 3, 4. Smooth microsome fraction incubated with ferritin anti P-450 antibody conjugates. The smooth microsomes containing lipoprotein particles (cored vesicles) are numbered. The cored vesicles nos. 1-3 in Fig. 3 are labelled with ferritin, while nos. 4 and 5 in Fig. 3 and all the cored vesicles in Fig. 4 (nos. 1-5) are not. The arrowheads in Figs. 3 and 4 mark the broken vesicles which are loaded with the lipoprotein particles. Inset to Fig. 4 shows a coated vesicle which contains a lipoprotein particle but is not labelled with ferritin. x 62500; inset, x 75000. Intracetlular transport of lipoprotein 279 which are usually composed of stacks of 3 or 4 cisternae (saccules), clearly showing a concave or forming (cis) and convex or mature (trans) face. The outermost cis-Golgi cisterna is sometimes fenestrated (Figs. 5, 6) and, in extreme cases, appears to be composed of small vesicles connected by narrow channels (Fig. 6, arrow). Occasion- ally the outermost fenestrated saccule contains lipoprotein particles, as shown in Figs. 6, 7. The flattened saccules, however, usually do not contain lipoprotein particles except at the dilated rims, where sometimes a number of lipoprotein particles are packed as illustrated in Fig. 5 (asterisk). It is also shown that the last trans-Golgi saccule is sometimes fenestrated as shown in Fig. 8 and in continuity with the vacuoles containing the particles (Figs. 5 and 8, large arrows).

Table 1. Morphological analyses of the smooth vesicles containing lipoprotein particles in smooth microsome fraction*

Diameter, nm UI1IUUII1 VC9IUC9 UUlllalllll ^ Vesicles lipoprotein particles Lipoprotein particles Labelled with ferritin (81 %) 128 ±35 56 ±14 (n = 75) Non-labelled (19 %) 122 ±25 57 ±13 (n = 18) * The number of smooth vesicles observed was 1274, in which 93 vesicles were loaded with lipoprotein particles (7"3 %)• Most of them (81 %) were labelled with ferritin, while the rest (19 %) were not.

Occasionally a large condensing vacuole containing varying numbers of lipoprotein particles was found embraced by the last trans-Golgi cisterna (Fig. 6). In the Golgi fraction there are also a number of secretion droplets - mature secretory granules - which contain a number of closely packed lipoprotein particles (Figs. 5-8, large arrowheads). No stack of cisternae, including the lateral dilated rims and the outermost saccule which sometimes appears fenestrated or vesiculated, nor any condensing vacuole in the trans-Golgi region, was labelled more heavily with ferritin than were the controls. Occasionally, a few ferritin particles were found between the Golgi stacks. They are probably the conjugate particles simply trapped between the saccules, because similar trapping of particles was also observed in control specimens (not shown). The surface membranes of the secretion droplets which contain a number of lipoprotein particles were also hardly labelled with ferritin. In marked contrast, smooth and rough microsomal vesicles which are found in the Golgi fraction are heavily labelled (Figs. 5, 6, 8, small arrowheads). Thus they serve as internal markers. In addition to the Golgi saccules, condensing vacuoles and secretion droplets, there are small vesicles, 100-120 nm in diameter, which contain lipoprotein particles, as indicated by small arrows in Figs. 5, 7, 8. Most of these cored type vesicles (84%) were not labelled when incubated with the antibody conjugates (Table 2). The average S. Matsuura and Y. Tashiro Intracellular transport of lipoprotein

Figs. 5-8. Rat liver Golgi fractions incubated with the ferritin antibody conjugates. The Golgi complexes were cut cross-sectionally in Figs. 5 and 6, and obliquely in Figs. 7 and 8. The lipoprotein particles are found in the fenestrated first stack (Fig. 6, arrow), dilated outer rims of the flattened Golgi stacks (Fig. 5, asterisk), in the vacuoles connected with the last trans-Golgi stacks by a narrow channel (Figs. 5, 8, large arrows). In the secretion droplets, a number of the lipoprotein particles are packaged in a condensed state as shown in Figs. 5-8 (large arrowheads). The lipoprotein particles are also found in small smooth vesicle as indicated by small arrows in Figs, s, 7 and 8. Small microsomal vesicles probably derived from ER membranes are heavily labelled with ferritin (small arrowheads), while all the Golgi components and small peripheral vesicles containing lipoprotein particles are hardly labelled with ferritin. x 62 500.

Table 2. Morphological analyses of the smooth vesicles containing lipoprotein particles in Golgi fraction*

Diameter, nm Smooth vesicles containing lipoprotein particles Vesicles Lipoprotein particles Labelled with ferritin (16%) IIO±2S 60 ± 14 (« = 16) Non-labelled (84 %) 99 ±18 (n = 86) • The number of smooth vesicles loaded with lipoprotein particles were 102, only 16 % of which were labelled with ferritin. 282 S. Matsuura and Y. Tashiro Intracellular transport of Upoprotein

Figs. 9, 10. Serial sections of cross-sectioned (Fig. 9) and tangentially sectioned (Fig. 10) Golgi complexes. Fig. 9 shows that the upper rim of the first Golgi stack is dilated and contains lipoprotein particles in the first 3 sections (Fig. 9A-C, arrowheads), while the lower rim of the same stack contains the particles in the last two sections (Fig. 9D, E, arrowheads). Fig. 10 shows that a number of the Golgi vacuoles loaded with lipoprotein particles are connected with narrow channels to the last fenestrated Golgi stack. These electron micrographs strongly suggest that the secretion droplets are formed by the detachment of these vacuoles. x 62500. diameter of the lipoprotein particles in the unlabelled and labelled vesicles was the same (~6o nm).

Observation of isolated Golgi complexes by serial sectioning In view of the importance of the structure of the Golgi complexes in the intracellular transport of lipoprotein particles, we have tried to make out its 3-dimensional structure by studying serial sections. Figs. 9 and 10 show cross-sectional and tan- gential profiles of the Golgi stacks, respectively. Several ambiguous points have been clarified by this method. Golgi saccules which are apparently not loaded with lipoprotein particles in one section, sometimes contain them in adjacent sections, usually at the rims of the saccules as indicated by the arrowheads in Fig. 9. It was also revealed that the trans-Golgi vacuoles which are apparently 284 Matsuura and Y. Tashiro Intracellular transport of lipoprotein 285 free and contain a number of lipoprotein particles in one section are frequently continuous with the last trans-Golgi saccules in some other sections. They probably correspond to the condensing vacuoles and the secretion droplets may be formed by detachment of the dilated portions of the Golgi saccules or Golgi vacuoles (Fig. 10). Immunoelectron-microscopic observation also revealed that the labelling with ferritin of the Golgi stacks and vacuoles and the secretion droplets are scanty and at the control level.

Labelling of Golgi fraction prepared from phenobarbital-treated rat liver As reported previously, liver microsomes prepared from phenobarbital-treated rats are almost completely covered by ferritin when incubated with the ferritin antibody conjugates (Matsuura et al. 1978, 1979). We have, therefore, examined the Golgi fraction prepared by Hino's procedure from phenobarbital-treated rats. As shown in Fig. 11, the Golgi bodies are heavily contaminated with smooth ER elements which have markedly proliferated following the phenobarbital treatment. It is evident that, while microsomal vesicles are heavily labelled with ferritin, Golgi saccules and trans-Golgi vacuoles are hardly labelled. Fig. 12 shows a Golgi body densely loaded with a number of lipoprotein particles. As in the previous case, Golgi saccules and Golgi vacuoles were hardly labelled with ferritin. It is to be noted in Fig. 11 that there are a number of small vesicles which are not labelled with ferritin. Although we need serial sections to determine whether or not these vesicles are connected with the Golgi stacks, it is very probable that at least some of them correspond to the small peripheral vesicles which are transporting secretory proteins from ER to Golgi bodies.

Labelling of lysosomes Golgi fraction prepared according to the procedure of Hino et al. (1978) occasionally contain lysosomes. As shown in Fig. 13, ferritin labelling of the outer surface of the organelles was scanty and at the control level.

Figs. 11, 12. Rat liver Golgi fractions prepared from phenobarbital-treated animals. In these photographs, all the microsomal vesicles are heavily labelled with ferritin, while the Golgi components such as Golgi transporting vesicles, Golgi stacks and Golgi vacuoles are hardly labelled. The Golgi complex in Fig. 12 is heavily loaded with lipoprotein particles, while that in Fig. 11 is empty, x 62500. Fig. 13. Rat liver lysosomes observed in the Golgi fraction. The Golgi fraction isolated by the procedure of Hino et al. (1978) is contaminated with various cell organelles such as microsomes and lysosomes. It is apparent that the former are heavily labelled with ferritin (arrowheads), while the latter are not. x 62500.

19 CEL39 286 S. Matsuura and Y. Tashiro

DISCUSSION It has been generally accepted that the ER is the principal site of lipoprotein synthesis in liver cells (Stein & Stein, 1967; Hamilton et al. 1967; Jones et al. 1967; Mahley et al. 1969; Claude, 1970; Glaumann et al. 1975). The assembly of the apolipoproteins and lipid moieties into the lipoprotein particles which are presumed to be precursors of hepatic very low density lipoprotein (VLDL), begins within the channels of the rough ER and continues in the smooth ER. After con- jugation with lipid moieties, the lipoprotein particles are transported to the Golgi complexes, where some remodelling such as terminal glycosylation of the apolipoprotein, concentration and packaging into secretion droplets may take place, and finally the secretion droplets are transported to the apical region of the cell and dis- charge their VLDL particles into the space of Disse. Since lipoprotein particles are easily identified by electron microscopy, we used them as a marker for analysis of the transport process from ER 50 Golgi in rat hepatocytes. The present electron-microscopic observations of rat hepatocytes have shown that lipoprotein particles are found in the intracisternal cavities of rough and smooth ER, in the small peripheral vesicles on the cis-Golgi region, as well as in the dilated rims of Golgi stacks. This finding does not prove but is consistent with the view that lipoprotein particles are transported from ER to Golgi by small peripheral vesicles. On the surface of the forming face of Golgi complexes, these vesicles may fuse with the outermost saccules, or they may fuse to each other, forming a new stack. The lipoprotein particles in small peripheral vesicles are thus transported into the saccules and are presumably translocated rapidly to the outer rims of the saccules to be stored there, forming dilated rims containing lipoprotein particles. An important question is whether or not the small peripheral vesicles, and the first stack of Golgi cisternae, which are presumably formed by the fusion of the small peripheral vesicles, contain cytochrome P-450. Electron-microscopic observation of the smooth microsome fraction incubated with the anti P-450 antibody conjugates revealed that about 80% of the smooth vesicles containing lipoprotein particles are labelled with the antibody conjugates, while the rest (~2O%) are not. Our interpretation for the former, labelled vesicles is that they are derived from smooth ER in the tubular configuration. In pancreatic exocrine cells, secretory proteins are transported directly from transi- tion elements of ER where they are packaged into small peripheral vesicles to be transported either to Golgi stacks or to trans-Golgi condensing vacuoles (Jamieson & Palade, 1977; Palade, 1975). In hepatocytes, however, lipoprotein particles are first transported from rough to smooth ER (Jones et al. 1967; Claude, 1970; Glaumann et al. 1975), where they are probably packed into the small peripheral vesicles by budding of smooth ER membranes. When hepatocytes are homogenized, smooth ER may fragment into the self-sealing small vesicles (Palade & Siekevitz, 1956). Since smooth ER in the hepatocyte is usually tubular in configuration and lipoprotein particles are sometimes found singly or in rows in the narrow tubules, it is very probable that cored type vesicles containing a single lipoprotein particle are formed by Intracellular transport of lipoprolein 287 self-sealing of the fragmented smooth ER upon homogenization. Since their membrane is ER in origin, it is quite natural that they are labelled with the ferritin anti P-450 antibody conjugates. The problem is the existence of the small-cored type vesicles in the smooth microsome fraction which are not labelled with the antibody conjugates. They could not be derived from the ER membranes, because practically all the microsomal vesicles derived from ER are labelled with the conjugate (Matsuura et al. 1978, 1979). They could not be derived from secretion droplets either, because the latter are much larger than the small vesicles and contain a number of the lipoprotein particles in an extremely condensed state. The most probable interpretation may be that they correspond to the small peripheral vesicles or Golgi transporting vesicles which are carrying the lipoprotein particles from ER to Golgi. This interpretation is supported by the fact that similar small vesicles containing lipoprotein particles are commonly found in the Golgi fraction, where most of them (84 %) are not labelled with the antibody conjugate (Table 2). The present immunoelectron-microscopic observations of the Golgi fraction showed that all parts of the Golgi complex, including the outermost Golgi saccules and the peripheral dilated rims of the Golgi stacks, are hardly labelled with the ferritin antibody conjugates. It is concluded, therefore, that the membranes of the small peripheral vesicles per se do not contain cytochrome P-450 and therefore the cytochrome is hardly transferred either to the Golgi vesicles or Golgi membranes when the lipoprotein particles are transported from ER to Golgi complexes. These findings are consistent with the pre- assembling hypothesis proposed in the previous paper (Matsuura et al. 1978) for intracellular transport from ER to Golgi complex. The hypothesis assumes that on a special region of ER membrane where small peripheral vesicles are to be formed by a budding mechanism, microsomal marker proteins such as cytochrome P-450 are excluded in advance, while the proteins destined to form Golgi membrane are pre- assembled there. This hypothesis is also supported by the fact that cytochrome P-450 molecules are non-randomly distributed on ER membrane, sometimes forming clusters or patches (Matsuura et al. 1978, 1979), because such non-random distribution may allow the formation of such special regions on ER membranes. It is to be noted here that this hypothesis is not always inconsistent with the possibility proposed by Palade (1975) that some of the Golgi membrane is transported back to ER membrane by forming small empty vesicles on the Golgi membranes. All that is needed is that the membranes of the shuttle vesicles do not contain cytochrome P-450. Whether the membrane flow at this step is unidirectional or bidirec- tional remains to be solved by future work. There is clear biochemical and electron-microscopic evidence for the participation of the hepatocyte Golgi apparatus in the intracellular transport of lipoprotein particles. Recently Bergeron, Borst & Cruz (1978) demonstrated the participation of the liver Golgi apparatus in the secretion of most of the serum proteins and provided docu- mentation for a sequential progression of secretory protein through the cis and trans components of the Golgi apparatus. 19-2 288 S. Matsuura and Y. Tashiro

Disse

Fig. 14. Schematic view of the synthesis, intracellular transport and secretion of the lipoprotein particles as suggested by the present investigation, rer, ser, rough and smooth endoplasmic reticulum, respectively; gv, Golgi transporting vesicles or small peripheral vesicles, gb, Golgi body; c, t, cis and trans faces of the Golgi body, respectively; sd, secretion droplets; Disse, space of Disse. Processes occurring in the numbered positions are described below. 1. Biosynthesis and initial glycosylation of apolipoprotein and assembling of the precursor particles of lipoprotein molecules (#) in the intracisternal space of rough ER. The ER membranes, which are studded with ribosomes (4), contain microsomal enzymes such as cytochrome P-450 (•) and also the newly synthesized membrane proteins (*) which are destined to Golgi bodies. 2. Intracisternal transport of lipoprotein particles (#) from rough to smooth ER and maturation of the particles in smooth ER. 3. Formation of Golgi transporting vesicles containing a lipoprotein particle by budding of a special region of the smooth ER membrane where membrane proteins destined for Golgi bodies (•) are clustered and microsomal marker enzymes such as cytochrome P-450 (•) are excluded in advance. 4. Intracellular transport of the small peripheral vesicles containing lipoprotein particles from smooth ER to Golgi. 5. Formation of the outermost Golgi saccule by fusion of the small peripheral vesicles. 6. Intracisternal transfer of the particles from the central flat regions to the peripheral rims of the saccules. 7. Progressive translocation of the Golgi saccules from cis to trans position by formation of the new stack at the cis or forming face of the Golgi bodies. 8. Formation of condensing vacuoles by protrusion of the dilated rims of the Golgi stacks, where lipoprotein particles are packed together. 9. Condensing vacuoles are detached from the Golgi stacks, and the secretion droplets thus formed are transported from Golgi region to apical cytoplasm. 10. Secretion of lipoprotein particles to the space of Disse. Intracellular transport of lipoprotein 289 From their kinetic data (Bergeron et al. 1978) and our present immunoelectron- microscopic observations, the intracellular transport process of lipoprotein particles could be schematically illustrated as shown in Fig. 14. Small peripheral vesicles or Golgi transporting vesicles, loaded with a lipoprotein particle, bud off from the special regions of the smooth ER as discussed above, and ferry their contents to the outermost or first cis-Golgi saccule by membrane fusion. The lipoprotein particles are rapidly translocated to the rims of the saccules, so that distended rims filled with the particles are formed. The saccules move successively from the forming (cis) face to the maturing (trans) face, as a new stack is formed at the forming face of the complex and in the last trans-Golgi saccule, the distended rims filled with the particles may pro- trude more, and now appear as Golgi vacuoles. Simultaneously the particles will be condensed markedly within the Golgi vacuoles. At first the Golgi vacuoles are connected by narrow channels to the last trans-Golgi saccules but they are gradually detached and the secretion droplets thus formed are sent to the apical region of the cell to be secreted to the space of Disse. As the Golgi membrane is lost by detachment of the Golgi vacuoles, the last trans-Golgi saccule will become concave, fenestrated, and finally disappear. Recently the participation of GERL (Golgi-associated Endoplasmic Reticulum from which Lysosomes apparently form) in the secretory process of VLDL has been suggested by Novikoff & Yan (1978 a, b). However, whether GERL is a distinct organelle into which secretory protein is transferred directly from the ER and thence packaged into secretion droplets, or whether GERL receives secretory protein only after transport through the Golgi apparatus is currently in dispute (Bergeron et al. 1978; Hand & Oliver, 1977; Jamieson & Palade, 1977). We only point out here that, so far no trans-Golgi membrane - the membrane structure which is in continuity with the innermost trans-Golgi saccules - was labelled with the ferritin anti P-450 antibody conjugate. This problem has lo be more thoroughly studied in future.

We thank Dr Y. Fujii-Kuriyama for the preparation of antibody to cytochrome P-450 and Miss Keiko Miki for assistance with the manuscript. This work was supported by a Grant-in- Aid for Scientific Research from the Ministry of Education, Science and Culture, Japan.

REFERENCES BERGERON, J. J. M., BORTS, D. & Cnuz, J. (1978). Passage of serum-destined proteins through the Golgi apparatus of rat liver. An examination of heavy and light Golgi fractions. .7. Cell Biol. 76, 87-97. BERCERON, J. J. M., EHRENREICH, J. H., SIEKEVITZ, P. & PALADE, G. E. (1973). Golgi fractions prepared from rat liver homogenates. II. Biochemical characterization. J. Cell Biol. 59, 73-88. CHAPMAN, M. J., MILLS, G. L.&TAYLAUR, C. E. (1972). Lipoprotein particles from the Golgi apparatus of guinea-pig liver. Biochem. J. 128, 779-787. CLAUDE, A. (1970). Growth and differentiation of cytoplasmic membranes in the course of lipoprotein granule synthesis in the hepatic cell. I. Elaboration of elements of the Golgi complex.J. Cell Biol. 47, 745-766. EHRENREICH, J. H., BERGERON, J. J. K., SIEKEVITZ, P. & PALADE, G. E. (1973). Golgi fractions prepared from rat liver homogenates. I. Isolation procedure and morphological characterization. J. Cell Biol. 59,45-72. 290 S. Matsuura and Y. Tashiro

FLEISCHER, S., FLEISCHER, B., AZZI, A. & CHANCE, B. (1971). Cytochrome b5 and P-450 in liver cell fractions. Biochim. biophys. Acta 225, 194-200. GLAUMANN, J., BEKGSTRAND, A. & ERICSSON, J. L. E. (1975). Studies on the synthesis and intracellular transport of lipoprotein particles in rat liver. J. Cell Biol. 64, 356-377. HAMILTON, R. L., REGEN, D. M., GRAY, M. E. & LEQUIRE, V. S. (1967). Lipid transport in liver. 1. Electron microscopic identification of very low density lipoproteins in perfused rat liver. Lab. Invest. 16, 305-319. HAND, A. R. & OLIVER, C. (1977). Relationship between the Golgi apparatus, GERL, and secretory granules in acinar cells of the rat exorbital lacrimal gland. J. Cell Biol. 74, 399-413. HINO, Y., ASANO, A., SATO, R. & SHIMIZU, S. (1978). Biochemical studies in rat liver Golgi apparatus. I. Isolation and preliminary characterization. J. Biochem., Tokyo 83, 909-923. JAMLESON, J. D. & PALADE, G. E. (1977). Production of secretory proteins in animal cells. International Cell Biology, 1976-1977 (ed. B. R. Brinkley & K. R. Porter), pp. 308-317. New York: The Rockefeller University Press. JONES, A. L., RUDERMAN, N. B. & HERRERA, M. G. (1967). Electron microscopic and biochemical study of lipoprotein synthesis in the isolated perfused rat liver. J. Lipid Res. 8, 429-446. KEENAN, T. W. & MORRE, D. J. (1970). Phospholipid class and fatty acid composition of Golgi apparatus isolated from rat liver and comparison with other cell fractions. Biochemistry, N.Y. 9,19-25- KISHIDA, Y., OLSEN, B. R., BERG, A. & PROCKOP, D. J. (1975). Two improved methods for preparing ferritin-protein conjugates for electron microscopy. J. Cell Biol. 64, 331-339. KRAEHENBUHL, J. P., RACINE, L. & JAMIESON, J. D. (1977). Immunocytochemical localization of secretory proteins in bovine pancreatic exocrine cells. J. Cell Biol. 72, 406-423. MAHLEY, R. W., HAMILTON, R. L. & LEQUIRE, V. S. (1969). Characterization of lipoprotein particles isolated from the Golgi apparatus of rat liver. J. Lipid Res. 10, 433-439. MATSUURA, S., FUJII-KURIYAMA, Y. & TASHIRO, Y. (1978). Immunoelectron microscope localization of cytochrome P-450 on microsomes and other membrane structures of rat hepatocytes.J'. Cell Biol. 78, 503-519. MATSUURA, S., FUJII-KURIYAMA, Y. & TASHIRO, Y. (1979). Quantitative immunoelectron- microscopic analyses of the distribution of cytochrome P-450 molecules on rat liver microsomes.J. Cell Sci. 36,4i3~435- MCLEAN, J. D. & SINGER, S. J. (1970). A general method for the specific staining of intracellular antigens with ferritin-antibody conjugates. Proc. natn. Acad. Sci., U.S.A. 65, 122-128. MELDOLESI, J., JAMIESON, J. D. & PALADE, G. E. (1971). Composition of cellular membranes in the pancreas of the guinea-pig. II. Lipids.J'. Cell Biol. 49, 109, 130-149. NOVIKOFF, P. M. & YAN, A. (1978a). Site of lipoprotein particles in normal rat hepatocytes. J. Cell Biol. 76,1-11. NOVIKOFF, P. M. & YAN, A. (19786). The cytochemical demonstration of GERL in rat hepatocytes during lipoprotein mobilization.^. Histochem. Cytochem. 26, 1-13. PAINTER, R. G., TOKUYASU, K. T. & SINGER, S. J. (1973). Immunoferritin localization of intracellular antigens: The use of ultracryotomy to obtain ultrathin sections suitable for direct immunoferritin staining. Proc. natn. Acad. Sci. U.S.A. 70, 1649-1653. PALADE, G. E. (1975). Intracellular aspects of the process of protein synthesis. Science, N.Y. 189, 347-358. PALADE, G. E. & SIEKEVITZ, P. (1956). Liver microsomes. An integrated morphological and biochemical study. J. biophys. biochem. Cytol. 2, 171-200. SASAKI, S. & TASHIRO, Y. (1976). Studies on the posterior silk gland of the silkworm, Bombyx mori. V. Electron microscope localization of fibroin in the posterior silk gland at the later stage of the fifth instar. J. Cell Biol. 70, 648-659. STEIN, O. & STEIN, Y. (1967). Lipid synthesis, intracellular transport, storage, and secretion. I. Electron microscopic radioautographic study of liver after injection of tritiated palmitate or glycerol in fasted and ethanol-treated rats. J. Cell Biol. 33, 319-338. TOKUYASU, K. T. (1973). A technique for ultracryotomy of cell suspensions and tissues. .7. Cell Biol. 57, 551-565- VAN GOLDE, L. M. G., FLEISCHER, B. & FLEISCHER, S. (1971). Some studies on the metabolism of phospholipids in Golgi complex from bovine and rat liver in comparison to other sub- cellular fractions. Biochim. biophys. Acta 249, 318-330.

(Received 3 October 1978 - Revised 16 March 1979)