Compartmentalization of Pancreatic Secretory Zymogen Granules as Revealed by Low-Voltage Transmission Electron Microscopy

Abstract

Low-voltage (5-kV) transmission electron microscopy revealed a novel aspect of the pancreatic acinar cell secretory granules not previously detected by conventional (80-kV) transmission electron microscopy. Examination of ultra-thin (30-nm) sections of non-osmicated, stain-free pancreatic tissue sections by low-voltage electron microscopy revealed the existence of granules with non-homogeneous matrix and sub-compartments having circular or oval profiles of different electron densities and sizes. Such partition is completely masked when observing tissues after postfixation with osmium tetroxide by low-voltage transmission electron microscopy at 5 kV and/or when thicker sections (70 nm) are examined at 80 kV. This morphological partition reflects an internal compartmentalization of the granule content that was previously predicted by morphological, physiological, and biochemical means. It corresponds to the segregation of the different secretory proteins inside the granule as demonstrated by high-resolution immunocytochemistry and reflects a well-organized aggregation of the secretory proteins at the time of granule formation in the trans-Golgi. Such partition of the granule matrix undergoes changes under experimental conditions known to alter the secretory process such as stimulation of secretion or diabetes.

Examination of tissues and cells at high magnification employing electron microscopy tools had a remarkable impact on cell biology. Indeed, we are today familiar with most of the compartments that make up the cell architecture, whereas cytochemistry and cell fractionation are revealing the chemical components of those compartments. On the basis of these approaches, we are able to identify and understand the structural-functional aspects of the cellular machinery. To yield meaningful data, sample preparation procedures for microscopy are required to preserve tissues and cells as close as possible to their pristine state. It should be noted that only recently significant microscopical observations became possible on living cells, but those are still only restricted to light microscopy domains, limiting the observation to individual cells at relatively low resolution levels. Aside from this particular case, we are still confronted with the need to fix tissues and cells to arrest their metabolism and prevent their degradation. Historically, two different fixation approaches were developed: One, cryo-fixation, is based on the physics of crystallization, whereas the other is based on chemical interactions of cell content with external agents. The less intrusive cryo-fixation procedure is routinely applied in light microscopy. Still, as far as electron microscopy is concerned, it is a fairly delicate and cumbersome technique and therefore less attractive. Furthermore, after cryo-fixation is carried out, it still remains necessary to proceed with further chemical manipulations of the tissues to prepare them for the observation under the electron beam. The more conventional and widely used techniques for electron microscopy involve a sequence of chemical fixations and postfixations, followed by dehydration and embedding in resins, with ultimate counterstaining of the tissue sections with heavy metal compounds. Since the early days of electron microscopy, scientists were and still are concerned with the effects all these chemical interactions may have on the integrity of the tissues and cell components. Over the years, improved protocols as well as new microscopy techniques have been introduced, and the accord in the observations generated by the variety of methods has strengthened our confidence in the accumulated morphological data, which today provide the base of our knowledge in cell biology. However, the motivation behind creating and developing new, less intrusive techniques remains strong, as not all morphology mysteries can be counted as completely resolved.

A new approach in electron microscopy was introduced by Delong using low-voltage transmission electron microscopy (TEM; Delong et al. 1994; Coufalova and Delong 2000; Drummy et al. 2004; Bendayan et al. 2008; Wu et al. 2010). Classical electron microscopes accelerate their electrons up to 120 kV, whereas high-voltage electron microscopy reaches up to 1000 to 1500 kV, and both have been in use for quite some time. However, little attention has been given until now to the advantages of examining tissues and cells with electrons accelerated at much lower voltages, as low as 5 to 10 kV. Of particular interest is the considerable increase in the contrast of acquired images. In order for the low-energy electrons to penetrate the tissue section components and to generate meaningful images, tissue preparation routines have to be modified. The changes in working conditions, procedures, and protocols of sample preparation for low-voltage TEM relate to eliminating the use of heavy metals in the fixation and counterstaining steps and making ultra-thin tissue sections, preferably not thicker than 40 nm.

In the present study, we have thus examined pancreatic tissue ultra-thin sections with the Delong low-voltage electron microscope at 5 kV. The omission of the postfixation with osmium tetroxide, the absence of counterstaining with heavy metals, the use of ultra-thin sections, and the accelerating voltage of only 5 kV have resulted in the generation of images that by some features differ from our conventional knowledge. Indeed, the secretory granules of the pancreatic acinar cells in particular display an aspect that can easily be interpreted as reflecting an internal compartmentalization. The compartmentalization of acinar cell secretory granules is not a novel concept. Indeed, by physiological, biochemical, and morphological approaches, such a compartmentalization was previously predicted (Ermak and Rothman 1978; Bendayan et al. 1980; Rothman et al. 1989; Gingras and Bendayan 1994). The fact that we can now view it directly by electron microscopy is because masking effects introduced by the impregnation with heavy metals and the examination of thick sections are avoided.

We demonstrate in the present study that secretory granules of pancreatic acinar cells are compartmentalized. This compartmentalization, revealed by low-voltage electron microscopy, reflects the segregation of the different proteins within the granule, a fact supported by high-resolution immunocytochemistry.

Materials and Methods

Pancreatic tissue from Sprague Dawley rats was fixed by immersion with 1% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 1 hr at room temperature. Tissues were then processed for embedding with and without osmium tetroxide postfixation. When performed, the postfixation with 1% osmium tetroxide was carried out at 4C for 1 hr. Tissues were dehydrated in graded ethanol and propylene oxide and embedded in Epon according to standard procedures. For examination at low voltage, ultra-thin (30–40 nm) tissue sections were prepared, mounted on naked nickel grids, and examined using the Delong LVEM5 (Soquelec Ltd., Montreal, QC, Canada) low-voltage TEM at 5 kV, without counterstaining with heavy metals. Pancreatic tissues from seven different animals, processed at different times, were examined by low-voltage TEM. For the immunogold, thin (70–80 nm) tissue sections were mounted on Parlodion-carbon-coated nickel grids and processed for the demonstration of pancreatic amylase using the anti-amylase antibody and the protein A–gold complex as described in detail previously (Bendayan 1995). These tissue sections were examined with a conventional transmission electron microscope at 80 kV.

In addition to pancreatic tissue from control animals, we examined tissues from animals under experimental conditions. The first experimental situation consisted of an acute stimulation of secretion, which was performed by a single intraperitoneal (IP) injection of carbamyl β-methylcholine chloride (carbachol) (Sigma-Aldrich; Oakville, Ontario, Canada) at a final concentration of 12 mg/kg body weight. The animals were anesthetized with urethan, and the pancreatic tissue was sampled 3 hr after induction of secretion (Bendayan et al. 1985). The second situation corresponded to pancreatic tissue from experimental diabetic animals. Hyperglycemia was induced by a single intravenous (IV) injection of streptozotocin (Boehringer Mannheim, QC, Canada) 50 mg/kg body weight dissolved in 10 mM sodium citrate, pH 4.5. The study was performed on the pancreatic tissue of animals kept hyperglycemic for 3 months (Gregoire and Bendayan 1986; Bendayan and Gregoire 1987). Tissues were fixed with glutaraldehyde and processed for embedding in Epon without postfixation with osmium tetroxide. Ultra-thin sections (30–40 nm thick) were performed and examined with the Delong LVEM5 as described above. These two experimental conditions were chosen because they introduce major changes in the secretory granules as described in detail previously (Bendayan et al. 1985; Bendayan and Gregoire 1987; Bendayan and Levy 1988).

Results

Examination of rat pancreatic acinar tissue by low-voltage TEM yields images displaying acinar cells having a general aspect similar to the one generated by conventional (80-kV) TEM. The tissue not being osmicated and not counterstained by uranyl acetate or lead citrate has a contrast that is, however, different from the conventional one. At low magnification (Fig. 1), we can clearly identify large numbers of secretory zymogen granules in the apical region of the cells, whereas the basolateral region contains the rough endoplasmic reticulum and some mitochondria. Limits of the cells are easily identified because the extracellular space appears more contrasted than the cytoplasm (Fig. 1). At higher magnification, some structures yield images that are somehow different from what we find usually by conventional electron microscopy. In the extracellular space, the bundles of collagen fibers with their typical striations with periodicity exhibit a strong brightness (Fig. 2). Within the cytoplasm, although the membranes of the rough endoplasmic reticulum are not contrasted, the compartment is easily identifiable because the ribosomal particles aligning the endoplasmic reticulum membranes appear very bright (Fig. 3). Intercellular spaces, at the lateral sides, and the space between the apical membrane and the luminal content are well contrasted, revealing the classical interdigitations and microvilli of the acinar cells (Fig. 4). The secretory granules, although displaying their typical size and shape, reveal a particular pattern not seen by conventional microscopy. Indeed, the secretory granules display a rather granular non-homogeneous matrix, and obvious partitions having circular or oval profiles are seen within their content. These profiles have different electron densities and sizes. Some are centrally located, whereas others are at the edge of the granules. Some granules display as many as five or six of these profiles, whereas others have much less or even none (Fig. 4). Some of these circular profiles appear very bright, but others blend within the gray granular background of the granule matrix. At even higher magnification (Fig. 5), the heterogeneous pattern of the granules is evident. Besides the granular matrix, some circular or semi-circular profiles having different electron densities are easily seen within some of the granules (Fig. 5).

Figure 1.

Low magnification of an acinar cell illustrating large areas of the cell with abundant rough endoplasmic reticulum (RER), some mitochondria (m), and numerous secretory zymogen granules (g).

Figure 2.

Higher magnification of collagen fibers (cf) in the extra-cellular space. These fibers appear quite bright and reveal their characteristic striations.

Figure 3.

Higher magnification of the rough endoplasmic reticulum (RER). The ribosomal particles are bright and delineate the membranes of the endoplasmic reticulum. m, mitochondria.

Figure 4.

High magnification of the apical region of an acinar cell. The secretory zymogen granules (g) show some heterogeneity within their matrix. The lateral extracellular space between two cells as well as the space between the content of the acinar lumen (L) appear very bright, delineating the lateral interdigitations (id) as well as the apical microvilli (mv).

Figure 5.

High magnification of the secretory granules (g). The compartmentalization of some of the granules is obvious, with circular or oval profiles being brighter than the rest of the granular matrix. Not all granules display such round profiles, but all appear to have a quite heterogeneous matrix. m, mitochondria; RER, rough endoplasmic reticulum.

On the other hand, secretory granules of pancreatic endocrine cells (Fig. 6) present in the same ultra-thin tissue sections did not display any particular matrix heterogeneity when examined by low-voltage transmission microscopy. The morphology of these granules generated by low-voltage microscopy is similar to the classical one. The granules display their typical halo around their core but without any additional pattern (Fig. 6).

Figure 6.

The secretory granules (g) of an endocrine islet cell appear quite homogeneous. They display their classical halo between the granule content and the granule limiting membrane.

When osmicated pancreatic ultra-thin (30-nm) tissue sections were examined by low-voltage TEM, the overall morphology resembled the one obtained by conventional TEM. Limiting membranes are well delineated, and the secretory granules in particular did not present any obvious partition, with the granule matrix being more or less dense and homogeneous (Fig. 7).

Figure 7.

Rat pancreatic tissue fixed with 1% glutaraldehyde post-fixed with 1% osmium tetroxide and embedded in Epon. The ultra-thin section (30 nm) was not counterstained. Examination with the Delong low-voltage transmission microscope (5 kV). The secretory zymogen granules (g) appear surrounded by a well-contrasted limiting membrane (arrow), and the granule matrix appears quite dense and homogeneous.

Amylase, one of the secretory proteins packaged in the pancreatic zymogen secretory granules of the acinar cells, was revealed by applying the immunogold approach. Examination of these immunogold-labeled tissue sections was carried out with a conventional microscope at 80 kV. The gold particles confirmed the presence of amylase within the granules and revealed the fact that its distribution within the granule is not homogeneous (Fig. 8). Some granules display a rather uniformly distributed gold labeling, whereas others definitely show some heterogeneity; areas of the granules are intensely labeled, whereas others are rather devoid of labeling. It is interesting to note that some of the non-labeled areas are quite circular in shape. This was previously demonstrated for amylase as well as for several other pancreatic secretory proteins (Bendayan et al. 1980; Gingras and Bendayan 1994). Because the secretory granules contain more than 20 different proteins, the patchy labeling was interpreted as reflecting a compartmentalization of the proteins inside the granules. It should be mentioned that it was not possible to perform immunogold on ultra-thin tissue sections and examine them by low voltage because these ultra-thin sections (30–40 nm), besides their thinness, have to be mounted on naked grids and are thus too fragile to support the various and long incubations with antibodies and the multiple washing steps required in the immunogold labeling protocol.

Figure 8.

(A, B) Rat pancreatic tissue fixed with 1% glutaraldehyde, no osmium postfixation, and embedding in Epon. The thin sections (70 nm) were labeled for amylase, applying the protein A–gold technique with the anti-amylase antibody. The sections were counterstained before examination with a classical 80-kV transmission electron microscope. The labeling by 15-nm gold particles is intense at the level of the granules. Although the labeling appears to be distributed quite homogeneously in some of the granules, in others, the labeling is clearly not homogeneously distributed, displaying areas devoid of gold particles. These areas are rather circular in shape.

In addition to these studies, we examined pancreatic tissues from two different experimental conditions known to alter the secretory granules and their content. Under a condition of stimulated secretion, the number of granules decreases significantly while the Golgi apparatus expands (Bendayan et al. 1985). Upon examination with the low-voltage electron microscope, the granules display a regular appearance, but their matrices exhibit a granular appearance with some areas brighter than others (Fig. 9A). In many granules, areas very irregular in shape were significantly darker—an appearance that was quite different from that of granules under a control condition.

Figure 9.

Pancreatic acinar cells under experimental conditions. The pancreatic tissue was sampled 3 hr after induction of stimulated secretion (A) or 3 months after induction of hyperglycemia (B). Tissues were fixed with glutaraldehyde and embedded in Epon without osmium postfixation. Examination by low-voltage transmission electron microscopy reveals very heterogeneous matrixes at the level of the granules with areas that are particularly bright and others that are rather dark. Some are circular in shape, but others are very irregular. The appearance of the granules, however, differs from those of the control condition.

In long-term hyperglycemic conditions (Fig. 9B), we previously reported that the enzymatic content of the granules changes drastically with very low levels of amylase and increases in lipase (Gregoire and Bendayan 1986; Bendayan and Levy 1988). By low-voltage TEM, the secretory granules display again a granular matrix with bright and dark areas, although in this case the dark areas either circular or very irregular in shape appear to be predominant. Again, in this case, the granule matrix was revealed by low-voltage transmission microscopy, but its appearance differs from that under the control condition.

Discussion

Observation of pancreatic cells by low-voltage TEM has revealed that the secretory zymogen granules display internal substructures reflecting the existence of sub-compartments within the granules. Although such a conclusion has been anticipated for several years, no definite morphological evidence has been provided until now. To discuss the findings of this study, we will first briefly consider the particular advantages of image generation in low-voltage TEM as compared to imaging by conventional electron microscopy with beam energies of typically 80 kV and higher. The compartmentalization of the secretory granules reported here will then be considered, in reference to cell biology aspects of protein secretion.

The main advantage of conducting TEM observation of biological tissues with the LVEM5 microscope is the enhanced level of contrast achieved in the images of unstained tissues at beam energies below 10 keV. Any conventional TEM technique nowadays employs an accelerating voltage of 80 to 120 kV or higher, and modern electron microscopes are generally not designed for TEM imaging with beam energies below 40 keV. Increasing accelerating voltage is a prerequisite to achieve higher spatial resolutions. At the same time, it has an adverse effect on contrast in the biological tissues, as there is very little difference in the scattering of high-energy electrons from most cellular components. That leads to a very low contrast when imaging a tissue “as it is” unless a contrasting agent is added. Consequently, staining tissues with heavy metal compounds in pre- or postembedding stages became established and routine steps in sample preparation before TEM observation. However, the use of stains has its disadvantages for the correct interpretation of fine cellular morphology as this can be modified or shadowed by chemical reactions with stain components. To avoid staining steps before TEM imaging, significantly lower incident beam energy should be employed to increase contrast. It was shown that for practical purposes of transmission electron microscopy in organic materials, accelerating voltages below 10 kV are required. In this case, the variations in scattering “slow” incident electrons by different tissue features are sufficient to produce adequate contrast without any stain while maintaining a satisfactory resolution level.

Present and previous investigations of unstained biological tissues employing the LVEM5 transmission electron microscope at 5 kV confirmed that cell ultrastructure is well contrasted on TEM images, allowing resolutions to be achieved down to 2 nm, provided the sections are thin enough (ideally below 30 nm). In most cases, the dense areas of the tissue will appear dark on the final image, as they scatter away more electrons, and the low-density areas will appear lighter. Both phase contrast and multiple scattering phenomena may modify the relative brightness of some organelles. There is a natural divergence in appearance of the same tissue when stained or stain free. In the stained tissue, the darkest areas are invariably those where the stain is preferentially retained. In the unstained tissue, as observed at 5 kV in the LVEM5, the same areas, being stain free, may appear lighter and offer enough details for further resolving their internal structure.

Indeed, as seen on the micrographs from the stain-free sections in the present investigation, the contrast of electron microscope images in the absence of all stain is still adequate to allow visualization of cellular ultrastructure. Some of the features are dark, similar to their appearance in stained tissue, due to a natural high content in proteins. Not all elements show the same level of contrast as stained: Most membranes—plasma membranes, as well as intracellular membranes—are less contrasted and not well delineated. This may be related to the section thickness. On the other hand, multiple scattering effects result in a lighter appearance such as for the ribosomes and collagen fibers despite their protein-enriched content.

As illustrated in Figures 1 to 5, images of completely non-stained tissue reveal substructures within certain cell compartments, such as the secretory granules. Those substructures do not appear on the images of stained tissues, both in conventional and in low-voltage TEM. As seen from Figure 7, showing the osmicated tissue imaged at 5 kV, the osmium postfixation, although improving contrast on membranes and other elements, does not reveal any substructural elements within the granules, which appear as homogeneous, electron-dense vesicles. Differences between images generated by low-voltage microscopy and those acquired with tissues postfixed with osmium tetroxide seem to indicate that impregnation with heavy metals does in fact mask subtle substructures present in some cellular compartments. This is not surprising because absorption and binding of osmium by the tissue are not selective enough to differentiate all the morphological features in the tissue. In such case, osmium will decorate the area non-selectively and mask the intrinsic structural details by its overall high electron density. In thicker sections (70–80 nm), prepared for conventional TEM, the masking effect is even more pronounced due to large amounts of osmium present throughout the thicker section. On the other hand, TEM imaging at 5 kV results in an adequate contrast level to discriminate between the areas of very close density; a difference of 0.07 g/cm³ in density is enough to produce a 5% gap in contrast (Drummy et al. 2004). As staining is avoided, the interpretation of electron microscope images produced by the low-voltage technique relies on the image-generating procedure and ultra-thin section stability, rather than on the chemical modification of the tissue. It should be noted that all the ultra-thin sections observed at 5 kV exhibited remarkable structural stability under the beam with very little degradation after long exposures. Thus, TEM images of non-osmicated, non-counterstained tissues obtained by low-voltage microscopy must be less inclined to exhibit chemical artifactual changes and shadowing effects. At the same time, more real intrinsic substructures, usually masked by the staining agents, can be exposed. Any new morphological aspects revealed within cell compartments in the absence of stain are likely to reflect those in the living cell. For the secretory granules, subtle substructures were revealed, and we propose that they do correspond to an internal compartmentalization—a fact that we and others have repeatedly predicted considering results obtained previously employing a number of experimental approaches. In addition, the study of secretory granules of tissues from animals under experimental conditions shows that such internal partition is not unalterable. When the experimental conditions dictated changes in the content of the granules, the generated image of the granule was modified. Although the partition was still present, the appearance of the granule was different, which speaks in favor of a highly dynamic or versatile phenomenon.

Different experiments have in the past indicated and/or predicted that the pancreatic enzymes are segregated within the secretory granules and that some kind of compartmentalization must exist inside the granule matrix. We have revealed here such a compartmentalization of the granules, as its morphology was generated by low-voltage electron microscopy of unstained tissues and supported by the immunocytochemical localization of a pancreatic enzyme. The detection of this compartmentalization through imaging appears to be affected by the tissue-processing protocol. Indeed, when postfixation of the tissues with osmium is carried out, the generated image of the granule no longer exhibits internal subcompartments, with the granule appearing quite homogeneous. However, the distribution of the secretory proteins within such a granule remains segregated (Bendayan et al. 1980; Gingras and Bendayan 1994). Those results, together with the ones presented herein, confirm the fact that osmium fixation has a dramatic masking effect on the appearance of the internal architecture of the granules. Furthermore, the fact that double immunolabeling has demonstrated that some pancreatic enzymes occupy different areas in the same granule (Bendayan et al. 1980; Gingras and Bendayan 1994) establishes that secretory proteins do not form a homogeneous mixture within the granules but are rather segregated—which by cell biology standards makes perfect sense. Such segregation must start at the level of the trans-Golgi at the time of proteins packaging into condensing vacuoles or immature secretory granules. Heterogeneity in the granule content was previously detected by high-resolution X-ray analysis (Rothman et al. 1989; Goncz and Rothman 1992). The granule content not being a homogeneous mixture of pancreatic enzymes is further supported by the existence of a matrix of a charged polymer network (Nanavati and Fernandez 1993). Rothman proposed the existence of a reticular meshwork within the granule that could reflect an organized aggregation of the secretory proteins (Ermark and Rothman 1978). Developing this concept, we previously demonstrated the presence of molecular chaperones along the rough endoplasmic reticulum/Golgi-granule secretory pathway and particularly within the granules themselves (Velez-Granell et al. 1994; LeGall and Bendayan 1996). The presence of chaperones, such as Cpn10 and Cpn60, as well as other proteins, such as the structural GP-2 protein (Jacob et al. 1992) within the granule matrix, speaks in favor of an organized aggregation of secretory proteins within the granules prior to their secretion. The present observation of partition and compartmentalization of the granule matrix represents new evidence supporting the concept that the secretory granule is a well-organized cellular compartment carrying a large number of segregated secretory proteins and participating in a well-planned and controlled discharge into the acinar lumen.

Once we acknowledge the fact that the content of the granules is segregated within its internal subcompartments, a number of data and theories advanced in the past can then be approached with a new perspective. Indeed, Rothman as well as others have shown that secretion of pancreatic enzymes can occur in a non-parallel fashion, so that the nature of the pancreatic juice does not correspond to a homogeneous mixture of all pancreatic enzymes. Selective discharge of certain proteins does occur under particular conditions of stimulated secretion (Adelson and Rothman 1975; Rothman 1976, 1981; Dagorn, 1978; Adelson and Miller 1985; Sommer et al. 1985; Rothman et al, 1991; de Dios et al. 1999). On the other hand, Anderson, Jena, and colleagues (Anderson 2004; Jena 2004, 2009) have introduced the concept of a discharge mechanism different from the classical exocytotic event (in which the entire granule fuses in one single event with the plasma membrane). In the proposed concept, small sequential membrane openings, the porosomes, are responsible for the discharge. If we assume that a subcompartment of the granule rich in a particular enzyme is the one preferentially interacting with the plasma membrane releasing its content during transient porosomal events, then the pancreatic juice at that stage will be enriched in that particular enzyme. Advances in imaging techniques allow revealing new morphological details in the cell architecture. These new observations then come in agreement with other results generated by different approaches, reinforcing old or new concepts in cell biology.