After its elaboration in the secretory endpieces [that can consist wholly of serous, mucous, or seromucous cells, or of a combination of any two of the foregoing (Tandler, 1993a)], the primary saliva passes through a series of excurrent ducts where it is progressively modified. The most prominent of the intralobular ducts are the striated ducts (SDs), so called because, in optical microscopy, vertical ‘striations’ mark their basal cytoplasm (Tandler, 1993a). Such ducts are found mainly in the parotid and submandibular glands. The sublingual gland only rarely includes distinct SDs (Braus, 1914), although scattered patches of intralobular duct wall may display a striated duct-like configuration (Riva et al., 1988). A similar situation obtains in certain minor salivary glands of some species (e.g., Tandler et al., 1970, 1994). Transmission (Tandler, 1963, 1993b) and scanning (Riva et al., 1993) electron microscopy have revealed that the basal striations visible by light microscopy are due to highly compartmentalized basal cytoplasm of the principal cells that houses rows of vertically oriented, usually rod-like mitochondria; the basal compartments are delimited by multiply folded plasmalemma. The basal striations seen in optical microscopy actually are the result of rows of mitochondria alternating with folded plasma membrane (Figs. 1 and 2). This configuration applies to the parotid and submandibular glands of virtually all studied mammals, qualitative minor differences between species may exist in the degree of folding of the basal membranes, the height of the basal compartments, and the length and number of the basal mitochondria, but the basic structural theme is maintained throughout most of the 300 mammalian species examined so far.
We interpret the conservation of specific cell types as an indication that they serve important functions. The pancreas and liver are necessary for life, and the basic histological structure of these two organs, as well as the cytology of pancreatic acinar cells and of hepatocytes, seem to have changed little during evolution, as they are very similar from fish to mammals (Willmer, 1970). The need to fashion highly efficient means for breakdown of foodstuffs, e.g., pancreatic enzymes, and for their processing in metabolic pathways (hepatocyte function) must be under intense purifying selection that conserves basic cytoarchitecture across the spectrum of vertebrate species (Patt and Patt, 1969). In strong contrast to conserved cells in the liver and pancreas, salivary glands and their secretory cells are exceedingly variable in mammals. Indeed, based upon comparisons of nearly 300 species, it is clear that no single simple description of salivary glands applies equally well to all kinds of mammals (Phillips, 1996). Such remarkable variation might be interpreted as a lack of evolutionary restraint, implying that salivary glands perhaps have less immediate significance than do the liver and pancreas. In fact, however, such a conclusion cannot be drawn and would be inconsistent with available data. The conservation of histology and cytology seen in organs such as liver and pancreas possibly is indicative of conservation in gene expression and regulation. The implication is that such organs provide fundamental ‘maintenance’ and are not free to vary. The lack of conservation in salivary glands is significant because it parallels evolutionary lineages, i.e., it is not random or disorderly, often correlates with speciation events (the cornerstone of evolution), and generally reflects what is known about the physiological parameters imposed by various nutritional strategies of mammals (Phillips, 1996; Phillips and Tandler, 1987, 1996). The available molecular data are consistent with the ultrastructural variations; lineages of mammals often exhibit their own patterns of gene acquisition and regulation (Phillips, 1996).
GENERAL FUNCTIONS OF SALIVA
Although all salivary glands drain into the mouth, they show considerable anatomic variability in terms of their size and location. It would be impossible to describe these parameters for all examined mammals, but diagrams of representative species are available in the publications of DiSanto (1960), Shackleford and Wilborn (1968), Phillips et al. (1977), and Tandler and Riva (1986). Salivary glands are considered to be part of the digestive system because of their association with the oral cavity and because in many instances they elaborate digestive enzymes. Moreover, it now is firmly established that salivary secretions are critically important for defense of the soft tissues of the oral cavity as well as of the teeth (Edgar, 1992; Tabak, 1995; Wolf et al., 1990) and the upper alimentary tract (Bodner, 1991; Kinoshita et al., 1999; Kongara and Soffer, 1999; Mathison et al., 1997; Olsen et al., 1984; Sarosiek and McCallum, 1995a,b; Scheving et al., 1979; Taichman et al., 1998). Furthermore, it has been suggested that they also have been adapted in various species to function as secretors of pheromones for sexual signaling or territorial and kinship marking, in maintaining the integument of newborn and infant offspring, in aggressive defense mechanisms (including secretion of venoms), in passive defense mechanisms (including neutralization of toxins in foods), and in the therapeutic effects of wound licking. The near ubiquity of SDs in salivary glands indicates that they play an essential role and, as this review will show, there also is considerable variation in their structure, implying that they might serve functions other than only modifying the electrolyte content of saliva. These additional functions seem to be mediated by secretion of organic substances. Under some circumstances, the term ‘secretion’ is considered to include the release of electrolytes and other inorganic molecules. In the present review, we use the term to apply specifically to organic secretion.
The luminal surface of SDs varies interspecifically. This surface usually bears some short, scattered microvilli, but in principal and accessory submandibular glands of vampire bats the microvilli are so numerous as to resemble a renal brush border (Tandler et al., 1990b). In certain species of fruit-eating bats, the luminal surfaces bear leaflike extensions of the apical plasma membrane that are studded on their cytoplasmic aspect with portasomes (Tandler et al., 1989). The latter are small, dense, repeating structures that have been postulated to be involved with electrolyte transport, to harbor an ATPase, and to act as a proton pump (Harvey et al., 1991; Zhuang et al., 1999). These modifications of the apical duct surface seem to be loosely correlated with salivary pH (Nagato et al., 1998; Phillips, 2000; Tandler and Phillips, 1998).
Although there is little direct evidence, it is assumed (largely on the basis of the resemblance of the SDs to the proximal and distal convoluted tubules of the kidney) that these ducts engage in electrolyte resorption. Using tritium-labeled ouabain, Bundgaard et al. (1977) found that the basal membranes of SD cells in the cat submandibular gland were heavily labeled, and concluded that the SDs ‘play a dominant role in the ductal reabsorption of sodium and that the transport is probably mediated by a (Na+ + K+)-activated ATPase.’ Further, Dinudom et al. (1993, 1994) consider granular convoluted tubules (GCTs) in rodent submandibular glands to be a localized variant of SDs and, having physiologically measured electrolyte transport in the former, extrapolated this function to the bona fide SDs. Thus, the rendering of saliva to a hypotonic state is considered to be a major function of SDs.
A peculiarity of SDs in many species is the presence of a second type of tall cell, the so-called ‘dark cell,’ in addition to the conventional principal cells (Phillips et al., 1977) (Fig. 1). In the rat submandibular gland, they constitute ∼4.8% of the duct wall (Sato and Miyoshi, 1998). The dark cells usually are narrow and contain an electron-dense cytoplasm in which the basic organelles are sparse. In the light microscope, they appear as slender, intensely stained (dark) cells. The frequency of such dark cells is increased in preparations of glands that show signs of suboptimal preservation, and it has been argued that they are fixation artifacts (Tandler, 1993b). Because they have been observed in many species and because their abundance can vary with physiological status (Hosoi et al., 1992; Maruyama et al., 1993), they may be a distinct SD cell type or they may represent the conventional SD cell in a specific functional stage that is particularly sensitive to fixation conditions. The latter notion is supported by the finding of Knauf et al. (1983) that the number of dark cells in the main excretory duct of the rat submandibular gland is related to metabolic acidosis and alkalosis. Garrett (1965) observed that a small number of SD cells become ‘completely black’ after fixation with a mixture of osmium tetroxide and sodium iodide, and suggested that their appearance might be dependent on metabolic state. The relationship between Garrett's stain-blackened cells and the more frequently observed dark cells is not clear. In any case, because reports on the dark cells are only sporadic and incidental, they will not be addressed further in this review.
GRANULAR CONVOLUTED TUBULES
In the submandibular gland of many [but not all, e.g., chipmunks (Tamias striatus), antelope squirrels (Citellus tereticaudus), and guinea pig (Cavia porcellus) (Flon et al., 1970) rodent species, another intralobular duct segment, the GCT, precedes the SD. The development, structure, and function of these ducts have been reviewed in extenso by Gresik (1994). Although these ducts have some basal striations, their outstanding structural feature is the presence of serous-like granules in the apical cytoplasm, especially in males. So pervasive are these ducts that unwary workers new to salivary gland research who take the human submandibular gland as the glandular exemplar sometimes refer to rodent submandibular glands as mixed glands consisting of mucous and serous secretory elements. (Modern terminology reserves the terms ‘mucous’ and ‘serous’ strictly for endpieces, almost all rodent submandibular glands actually consist in part of seromucous endpieces and of ducts with serous-like granules.) The GCT granules contain a variety of enzymes and vasoactive and growth factors and most likely play important roles in the life of rodents (Barka, 1980; Mori et al., 1992; Murphy et al., 1980).
It currently is not generally appreciated that in those mammalian species that lack GCTs the SDs in either the submandibular or parotid glands or both frequently have a secretory capacity other than that concerned with electrolyte homeostasis. Older histologists were well aware of this property of SDs, and for many years these ducts often were referred to in the literature as secretory ducts. Babkin (1950) lists more than a dozen light microscopy articles that describe secretory phenomena in SDs. Because the secretory products of SDs might have significance in the physiology of nonrodent mammals (and in rodents as well), we have surveyed such ducts in a large number of species that we ourselves have examined and on information that we have culled from the literature. Much of the previously unpublished data presented in this review is based on specimens collected by one of us (CJP) and field-fixed by immersion in Phillips's (1985a) modification of the triple aldehyde-DMSO mixture of Kalt and Tandler (1971) or a modification (Forman and Phillips, 1988) of Karnovsky's (1965) fixative. Other specimens were fixed in the laboratory (BT) with either triple aldehyde-DMSO or half-strength Karnovsky's fixative applied either by vascular perfusion or by immersion. After salivary gland tissues were obtained from animals in the field, the mammal specimens were prepared according to museum research protocols, and voucher specimens matching the tissue samples were deposited in the mammalogy research collections at the Carnegie Museum of Natural History, the Nebraska State Museum (University of Nebraska-Lincoln), or the Texas Tech University Museum (Natural Science Research Laboratory). The studies by other authors cited by us encompass a wide variety of fixation and preparative techniques for salivary gland specimens. The reader is referred to the individual articles for these protocols.
There are several good reasons for extensive interspecies reviews of structure and function of cells that are presumed to be homologous. In overview, the coevolution of salivary gland secretory cells and their products probably was an important part of the adaptive radiation of mammals (Phillips, 1996). Comparative studies thus enable us to gain insight into the morphological and functional diversity within mammalian cell types that might have played a significant role in the history of mammals. A review of data from a substantial number of species offers an opportunity to use a phylogenetic framework for (a) making direct comparisons in the context of evolutionary history, and (b) testing hypotheses about function or biological role(s) of a cell type. In the present case, we have generated enough data to ask if there are patterns that can be explained by historical factor, in other words, are similarities among species correlated with their phylogenetic histories, or does some other consideration such as diet explain similarities and differences?
It also might be meaningful if clusters of similarity among cells reflected contemporary molecular genetic-based hypotheses about the relationships among orders of mammals (Pumo et al., 1998). An example of how data can emerge from interspecies comparisons can be seen from our previous review of intercalated duct cells in major salivary glands (Tandler et al., 1998). In this instance, we showed that in 55 species of bats our qualitative assessment of secretory product was correlated with diet, especially with feeding on hard-bodied (chitinous) insects. Based upon immunohistochemical data, we hypothesized that this pattern reflected the presence of lysozyme, which can act as a chitinase at low pH (Phillips et al., 1998).
The ultrastructural evidence for secretory capacity for organic components of SD cells runs the gamut from zero to extensive depending on gland and species. Table 1 is a compendium of the frequency of occurrence of secretory granules and vesicles and their size (discrepancies in size reported for the same species by different authors may be the result of differences in granule dimensions according to their location within SDs, i.e., proximal, distal, or central, or to differences in technique) in SD principal cells, mainly in the submandibular and parotid glands, in a variety of species. SD cells in some glands and species are devoid of any sign of secretory activity; the human sublingual gland is one example (Riva et al., 1988); the same observation holds true for the sublingual gland in many other species. In certain species, cells within a given SD can vary considerably in the number of secretory granules, this could be an indication of different types of cells under the rubric of SD cells, or more likely is the result of asynchronous secretory cycles within neighboring cells or duct segments.
Table 1. Secretory granules in striated duct cells of mammalian major salivary glands
Generic and specific names are based on Wilson and Reeder (1993). In a few instances, we have corrected or updated generic or specific names from the original published nomenclature. In published work other than our own, we are unable to confirm the identifications. AP, accessory parotid gland; ASM, accessory submandibular gland; P, parotid gland; PP, principal parotid gland; PSM, principal submandibular gland; RL, retrolingual gland; SG, unidentified major salivary glands; SL, sublingual gland; SM, submandibular gland; V, clear vesicles; irr. V, irregular vesicles (dimensions not included); obl. V, oblong vesicles (dimensions not included). 0, none; +, few; ++, intermediate; +++, many; dist, distal; med, medial; prox, proximal; NA, not available.
Granule dimensions are not supplied by referenced authors—approximate diameters calculated from their published micrographs.
Observation based on light microscopy of semithin sections.
Observation based on paraffin sections.
Discrepancies in abundance or measurements within a given category may be due to measurements being taken at different sites or, in some cases, to strain differences in the animals tested. Unpublished, personal observations by the various authors of this review.
The apical cytoplasm of SD cells in salivary glands of many species contains numerous small, ‘empty’-appearing vesicles (Fig. 3A–C) [although in the rat submandibular gland they may contain fine filaments (Sato and Miyoshi,1998)]; a variety of functions has been suggested for them. By using tritium-labeled fucose, (Hand 1979, 1987) demonstrated that at least some of these vesicles are involved in the transport of polysaccharides to the glycocalyx coating the apical surface of the duct cells and of membrane components to be incorporated into the apical plasmalemma. These SD vesicles appear to increase in number in starved rats (Scott and Pease, 1964). Interestingly, the apical cytoplasm of some SD cells in the rat parotid glands contains very small granules of moderate density as well as empty-appearing vesicles. Using immunocytochemistry, Yamamoto-Hino et al. (1998) have shown that apical vesicles in certain SD cells are positive for inositol 1,4,5-trisphosphate receptor (type 2), which is related to cellular calcium oscillation. Riva et al. (1976) speculated that the relatively sparse, empty vesicles in the apical cytoplasm of human submandibular SDs are involved in uptake of unspecified components from the luminal saliva. In the latter case, the apparent emptiness of the vesicles may be the result of inadequate fixation, because other studies have shown the presence of a smattering of small dense vesicles (and no empty ones) in the same locus (Tandler, 1978, 1987). In support of the notion that at least some of the apical vesicles in SD cells may be involved in uptake rather than in secretion is the finding by Webster et al. (1994) that these structures are immunochemically labeled for transferrin receptor and rab4, both considered to be markers of early endosomes and for receptor-mediated endocytosis. In the rat parotid gland, small endocytic vesicles in SD cells take up retrogradely-infused native and cationized ferritin (Coleman and Hand, 1987), horseradish peroxidase (Hand et al., 1987), and native and glycosylated bovine serum albumin (Lotti and Hand, 1989). In the parotid glands of rats made diabetic by administration of streptozotocin, the SD cells were shown by immunocytochemistry to have internalized proteins synthesized by the acinar cells (Lotti and Hand, 1988).
That clear vesicles in SD cells are involved in transcytosis also must be considered (Matthews and Lemon, 1981). Low levels of epidermal growth factor (EGF) in human saliva have been documented in numerous reports (Christensen et al., 1996; Dagogo-Jack et al., 1985; Hormia et al., 1993; Ino et al., 1993). Although Thesleff et al. (1988) and Ino et al. (1993) reported that parotid saliva is the richest source of EGF in human beings, Mogi et al. (1995) found that it contains only one-third the amount detected in submandibular-sublingual saliva; the latter group also detected low levels of transforming growth factor-α (TGFα) in saliva from all three glands. Nevertheless, the cellular site of synthesis of salivary EGF in man remains controversial. EGF has been localized immunocytochemically in human submandibular and parotid glands in both the secretory granules of the serous acinar cells and in small vesicles in the apical and basal cytoplasm of SD cells (Lantini and Cossu, 1998). Previous immunocytochemical studies of human salivary glands at the light microscopic level reported that EGF is in the serous acini (Christensen et al., 1996; Heitz et al., 1978; Kasselberg et al., 1985; Poulsen et al., 1986) or in intercalated ducts and SDs (Elder et al., 1978; Ino et al., 1993; Sato et al., 1985; Tsukitani et al., 1987). The findings of Lantini and Cossu (1998) offer a resolution of these conflicting data: EGF may be synthesized by serous acinar cells, released into the primary saliva, and then delivered by transcytosis across SD cells to the interstitium and ultimately into the circulation. Proof of the site of synthesis of human salivary EGF would require demonstration by means of in situ hybridization of the sites of expression of the mRNA for preproEGF. Other nonelectrolytes are reported to be transported across the SD epithelium in the rabbit submandibular gland (Case et al., 1985), possibly in vesicles.
In addition to contributing their own secretory products to saliva, SD cells add immunoglobulins, principally IgA, to ductal saliva by transcytosis (Corthesy and Kraehenbuhl, 1999). SD cells synthesize the polymeric immunoglobulin receptor (pIgR) that is incorporated into the basolateral plasmalemma where it functions as a receptor for the J-chain of dimeric IgA synthesized and released by plasma cells derived from the mucosa-associated lymphoid tissue (MALT). After binding of the J-chain of dimeric IgA to pIgR, this complex is brought into the SD cell in endocytic vesicles that traverse the cell to bind to and fuse with the apical cell membrane where secretory IgA (composed of a J-chain bound to two IgA molecules, and a cleaved portion of the pIgR known as secretory component) is released into the duct lumen (Corthesy and Kraehenbuhl, 1999). Both IgA and secretory component have been localized histochemically to the basolateral membrane and to the apical cytoplasm of SD cells (Korsrud and Brandtzaeg, 1982). At least some of the clear vesicles observed by electron microscopy in SD cells must be involved in the transcytosis of these immunoglobulins from the interstitium to the lumen.
Other secretory products associated with SDs are the vasoactive substances, kallikreins. These have been detected in the SDs of many species, including pig (Dietl et al., 1978), rat (MacDonald et al., 1996; Ørstavik et al., 1979, 1982; Simson et al., 1979; Simson and Chao, 1994), mouse (Kurabuchi et al., 1999; Penschow and Coghlan, 1993), cat, guinea pig, dog (Maranda et al., 1978; Schachter et al., 1980, 1983), sheep (Trahair and Ryan, 1989), monkeys (Yahiro and Miyoshi, 1996), and man (Garrett et al., 1984; Kimura and Moriya, 1984; Ørstavik et al., 1980; Wolf et al., 1998). Molecular biological analyses have established that the kallikrein enzymes belong to multigene families in mouse (Mason et al., 1983; van Leeuwen et al., 1986), rat (Scicli et al., 1993; Wines et al., 1989), and human beings (Berg et al., 1992; Clements, 1994). Although not yet analyzed, this relationship probably holds true for other species as well. Individual members of these families are expressed in various tissues, but all or almost all of them are expressed in the submandibular glands of rats and mice (McDonald et al., 1996; Scicli et al., 1993).
True tissue kallikrein (mK1, rK1, hK1) is the family member expressed in SD cells (Kurabuchi et al., 1999, and included references). It is localized in small apical secretory granules of moderate density (Fig. 4) (Schachter et al., 1983) or clear vacuoles (Penschow and Coghlan, 1993) in these cells or in empty-appearing vesicles (Yahiro and Miyoshi, 1996). Because it appears in saliva (Jensen et al., 1991; Jenzano et al., 1992; Lövgren et al., 1999), it probably is released by a merocrine process at the luminal membrane; Yahiro and Miyoshi (1996) present immunocytochemical evidence in support of this idea. Garrett et al. (1995) argue that kallikrein also can be released from clear vesicles across the basolateral membranes, thus giving it access to the circulation, a view supported by Penschow and Coghlan (1993).
Irrespective of the function of clear vesicles, these structures occur in the SD cells of many mammalian species (Table 1). Because in some species these clear vesicles are either oblong or irregular in form, their dimensions are omitted from Table 1; if the vesicles are spherical, their diameter is included.
The granule content of SD cells varies from species to species and from gland to gland. In some glands, the granules are so numerous that in semithin sections they form a distinct, heavily-stained subluminal band (Pinkstaff et al., 1982). In other species, the granules are both less abundant and small, so that at the light microscope level the apical cytoplasm lacks obvious secretion. For example, SD cells in the human parotid and submandibular glands examined by light microscopy appear ‘empty,’ but by electron microscopy they are seen to contain a few tiny dense granules (Fig. 5A). In many species, especially bats, secretory granules are abundant in the apical cytoplasm (Fig. 5B). In certain bats (for example Sturnira lilium and Hipposideros larvatus), SD cells with granules may alternate with cells that are devoid of granules (B. Tandler, unpublished observations). In a few species such as the African mole-rat, Tachoryctes splendens, SD cells contain a mixture of granules and vesicles (Fig. 3B). In the parotid gland of the beaver, Castor canadensis, SD cells with numerous dense granules often lie next to SD cells that lack granules, but that contain an abundance of clear vesicles. SD granules are dense, generally structureless, and usually measure 0.1–0.2 μm; in the SDs of any given species, the granules exhibit very little variation in size. One exception is in the parotid gland of the velvety free-tailed bat, Molossus molossus, in which most SD granules are about 0.2 μm in diameter, but scattered among these typical granules may be an occasional granule measuring as much as 6.0 μm, and even a few that exceed the duct cell nuclei in diameter. Other bats in which an occasional parotid SD granule is much larger than its companions include the yellow epauletted bat, Sturnira lilium, where a granule measuring 0.7 μm sometimes is present among the typical ∼0.2 μm granules, and the pipistrelle, Pipistrellus coromandra, in which SD granules are 0.12 μm, with a few that are 0.4 μm.
GIANT SECRETORY GRANULES
The most exceptional known example is present in the SDs of the slow loris, Nyctecebus coucang, where in both the parotid and submandibular glands virtually every SD cell contains at least one or several giant granules that may exceed 14 μm in diameter (Tandler et al., 1996) (Fig. 6A). These outsize granules are not homogeneously dense, but may contain aggregates of short, lucent laminae or lucent vesicles. Smaller granules may lie alongside the giant ones. The giant granules often have bundles of filaments associated with their surface membrane. These filaments, which are the thickness of actin filaments, may be necessary to pull the giant granules to the cell surface so that they can be exocytosed. The excretory duct cells in both the submandibular and parotid glands of the slow loris also contain giant granules in their supranuclear cytoplasm (Tandler et al., 1996).
As is the case in many rodents, the submandibular gland of the house shrew, Suncus murinus, has intralobular ducts that produce very large granules; the asymmetric distribution of light and dark matrical components gives these granules an ‘op-art’ appearance. Although such ducts for the most part lack basal striations, they have been labeled as striated (van Lennep and Dryden, 1985). Actually, they are analogous to the GCTs of rodents; in his light microscopical description of the submandibular gland of another species of shrew, Blarina brevicauda, Pearson (1950) clearly illustrates the direct continuity of the granular duct segment with the SD (labeled ‘secretory duct’ in his diagram). The bona fide SDs (that possess the usual basal striations) of the house shrew are virtually devoid of any secretory material whatsoever (Fig. 6B).
SD granules generally can be described as “serous-like” [cf. Young and van Lennep (1978) and Tandler and Phillips (1993) for a discussion on ultrastructural identification of serous granules]. In most mammals, the SD granules are homogeneously dense after fixation in an aldehyde-osmium sequence, although staining intensity may vary within narrow limits from one species to another (Fig. 7A,B) SD granules in the submandibular gland of the European hedgehog, Erinaceus europaeus, display an anomalous reaction to fixatives. Normally, secretory granules are better preserved by an aldehyde-osmium fixation sequence than by osmium alone. In the case of the hedgehog, however, osmium fixation by itself yields dense granules, whereas initial fixation in aldehydes followed by osmium leads to empty-appearing granules (Tandler and MacCallum, 1974). SD granules in the parotid gland of the hedgehog are not quite so sensitive to the modalities of fixation; although fixation in osmium alone yields dense granules, initial fixation in aldehydes does not produce granules that appear to be totally empty, but ones with a moderately dense content (B. Tandler, unpublished observations)
The 0.5 μm granules in the SDs of the parotid gland of the little brown bat, Myotis lucifugus, appear to be homogeneously dense at lower magnifications (Fig. 5B), but at high magnification it becomes evident that they have a paracrystalline content with a periodicity of 8 nm (Tandler and Cohan, 1984) (Fig. 7C). These crystalline granules clearly are very different from the crystalloid-containing lysosomes that have been described in the apical cytoplasm of rat parotid gland SD cells (Tandler et al., 1979). Granules similar to those in the little brown bat are present in the salivary glands of the mongoose, Herpestes edwardsi (Schramm et al., 1979). Although the SD granules in the little brown bat and in the mongoose have a crystalline substructure, they lack the angular silhouettes that characterize true crystalloids. In various mustelids, however, the ductular secretory material is in the form of membrane-bounded rhomboidal or prismatic crystalloids (Fig. 8A). In the North American mink, Mustela vison, rhomboidal crystalloids display a periodicity of 6 nm (Tandler, 1983; 1991) and in the Siberian weasel their periodicity is ∼5.2 nm (calculated from micrographs by Matsunaga, 1992). Although a repeat pattern has not been resolved in the secretory products of the SDs in the ferret (Jacob and Poddar, 1987a,b; Shono, 1997), the shape of these products strongly suggests that many have a paracrystalline core. Interestingly, in each of the aforementioned mustelids, such crystalloids are present in equal abundance in both the submandibular and parotid glands of both males and females (Jacob and Poddar, 1987a,b; Matsunaga, 1992; Tandler, 1983, 1991). In their study of salivary glands in the Old World marten, another mustelid, König and Masuko (1998) refer only to granules in the SD of the parotid gland (the submandibular gland is not mentioned in this context). Inspection of their Figure 5 reveals the presence of several angular bodies intermingled with granules in the apical cytoplasm of the SD cells that might be crystalloids, but no periodicity is apparent.
RODS AND TUBULES
SD cells in cats contain many secretory granules (Fig. 4). In contrast, the same cells in dogs lack granules; instead, they contain numerous membrane-bounded rods of moderate density, some of which are oriented perpendicular to the luminal plasma membrane (Nagato and Tandler, 1986a; Suzuki et al., 1975). These appear to be attached to the surface membrane by short filaments. The opposite (free) ends of the rods often have a terminal expansion. A histochemical study of dog parotid gland found the apical cytoplasm of SD cells to be strongly positive for glycoconjugates (Pedini et al., 1994); presumably this material is situated within the apical tubules. Similar structures have been observed in the submandibular SDs of the red bat, Lasiurus borealis (Fig. 8B), and of De Winton's long-eared bat, Laephotus wintoni. Apical tubules have been reported to occur in cells from a variety of other organs (see listing by Sagara et al., 1997), where they have variously been described as either secretory or as endocytic in nature, but their true function still is uncertain. In the raccoon dog, Nyctereutes procyonoides, parotid SD cells contain both dense rods that lack obvious connection to the luminal membrane and small, dense granules.
Whatever the nature of SD secretory products, the cytological pathway for their synthesis and packaging remains untraced. SD cells usually contain no more than a few scattered cisternae of rough endoplasmic reticulum (RER), totally lacking the orderly arrays of RER that typify secretory cells such as those in serous endpieces. A small Golgi apparatus frequently surmounts the nucleus; occasionally, it is positioned lateral to the nucleus. Condensing vacuoles are conspicuous by their absence. A similar situation obtains in the highly secretory GCT cells of the rat submandibular gland. It is only after pharmacological depletion of secretory granules from GCT cells that it was possible to track the production of new granules (Cutler and Chaudhry, 1973; Murphy et al., 1980; Thomopoulos et al., 1996). At first, there is a rapid evolution of RER and development of a more prominent Golgi apparatus, followed by the appearance of prospective granules. Thus, GCT cells follow the classic pathway of granule secretion. To confirm that the same sequence of events occurs in SD cells, it will be necessary to find a pharmacological agent that leads to discharge of secretory products from these cells so that the subsequent restitution of their secretory complement can be seen [the commonly used sialogogue, carbamylcholine, evokes only a loss of SD cell glycogen (Fava-De-Moraes et al., 1967)]. Until that time, the cytological events leading to production of secretory granules in SD cells will remain shrouded in mystery.
RELEASE OF SECRETORY GRANULES
Even the mode of liberation of secretory granules from SD cells is unclear. In the ontological descendants of SDs, namely, the GCTs, spontaneous exocytosis rarely is encountered (see, for example, Toyoshima and Tandler, 1991a), although it seems that in house mice, aggressive behavior can trigger release of GCT secretory product (Aloe et al., 1986). Although in all probability granule discharge in SD cells is by a merocrine process, another mechanism has been advanced. In 1969, Takano proposed that, in a variety of mammals, prominent blebs (Fig. 9) on the surface of parotid SD cells are liberated from the cells by possible closure of a ring of fibrils at their base; he christened these fibrils the “separating zone.” Messelt (1982) and Messelt and Dahl (1982), who examined harbor seal and rat submandibular glands, considered the blebs to constitute a form of apocrine secretion. Pinkstaff (1980) provides a listing of salivary glands in which apical blebs have been noted. In addition, such blebs have been depicted in SDs of salivary glands of mouse (Parks, 1962), rat (Han, 1967; Messelt, 1982), guinea pig (Reinecke, 1967), Japanese wood mouse (Odajima, 1981a), Chinese hamster (Suzuki et al., 1981b), Djungarian hamster (Suzuki et al., 1983), Syrian hamster (Chaudhry et al., 1986), guinea pig (Reinecke, 1967), pipistrelle (Azzali and Grandi, 1996), goat (Suzuki et al., 1981c; Suzuki and Otsuka, 1976; Takano et al., 1977), Kobe mole (Mineda and Kameyama, 1980), mongoose (Schramm et al., 1979), harbor seal (Messelt, 1982), cow (Suzuki et al., 1981a), house shrew (Tandler and Phillips, unpublished observations), and Indian pipistrelle (Fig. 9).
The status of these apical blebs is open to question. Proponents of their reality consider them to be an indication of apocrine secretion. It must be pointed out, however, that no one, to the authors' knowledge, has ever shown a detached mass of cytoplasm with included secretory material in the SD lumen, a sine qua non for apocrine secretion. As early as 1965, Tamarin and Sreebny noted that ‘ballooning’ of GCT cells in the rat submandibular gland is influenced by the fixative and that small blebs can appear on the SD cells when a hypo-osmolar fixative is employed. Moreover, the degree of blebbing is dramatically enhanced in the submandibular glands of guinea pigs by administration of antidiuretic hormone (Planel et al., 1966), which strongly (Baïsset and Montastruc, 1960; Junqueira et al., 1967) or slightly (Holmes, 1964) inhibits salivation. Because virtually all reports of SD blebs are based on glandular fixation by immersion, and because such structures are rarely encountered in glands fixed by vascular perfusion, we believe that in the vast majority of cases the blebs are fixation artifacts based on, in the words of Tandler and Erlandson (1976), ‘an idiosyncrasy of hydration in SD cells.’ Messelt and Dahl (1983) are of the opinion that the blebs are normal morphological features of rat submandibular gland SDs because exposure of the gland to X-rays erases the ability of these cells to form blebs. An alternative explanation is that irradiation simply puts out of commission those physiological functions dealing with hydro-osmotic phenomena in SD cells that are subverted by delayed, i.e., immersion, fixation.
In the parotid gland of the common vampire bat, the cells of the duct that follows the intercalated duct lack the usual basal membrane infoldings; they seem to have surrendered their canonical function of electrolyte transport in favor of organic secretion (Tandler et al., 1997c). The supranuclear cytoplasm of these cells is crammed with homogeneously dense particles measuring 0.5–1.5 μm, whereas the basal cytoplasm is occupied by a great deal of RER and lacks obvious striations (Fig. 10). Their histological position between intercalated and excretory ducts suggests that these ducts are homologues of SDs. The possibility that such cells represent a completely new kind of cell that is developmentally and evolutionarily unrelated to SD cells cannot be excluded.
In a few rare instances, the SDs are so unusual that their identification also can be affirmed only by virtue of their continuity with intercalated ducts. In the accessory parotid gland of the free-tailed bat, Tadarida thersites, the SDs have combined with the very short intercalated ducts to form macroducts, i.e., structures with lumina that are over 100 μm in diameter, that for the most part lack basal striations (Tandler et al., 1998). Nonetheless, the principal cells of these macroducts acquire a subluminal row of dense granules as they approach the union of the macroducts with the intralobular (proximal) portion of excretory ducts.
In the parotid and submandibular glands of the leaf-chinned bat, Mormoops blainevilli, the homologues of the SDs are twisted into a glomerulus-like skein. Their constituent cells, which lack basal striations, are characterized by deep, microvillus-lined intracellular canaliculi (Toyoshima et al., 1988). But despite their peculiar morphology, these cells contain a few apical secretory granules.
Unlike the parotid and submandibular glands, both of which are major salivary glands, most minor salivary glands in almost all studied mammals lack SDs altogether. In a few cases, particularly in human beings, some of these glands (as was mentioned previously) have excretory-like ducts that contain patches of cells with SD configuration. This is true in human labial (Tandler et al., 1970) and anterior lingual (Tandler et al., 1994) glands. In both cases, the SD cells are devoid of any sign of organic secretion.
The innervation of SDs deserves comment. In many species, these ducts lack apparent hypolemmal innervation. Hypolemmal nerve terminals are naked axons that leave interstitial nerve bundles to penetrate the epithelial basement membrane to end in the intercellular space between effector cells (Garrett, 1982). Such terminals appear to be absent from the SDs of many species, and are present only sparingly in the ducts of the parotid (Cope, 1977) and submandibular (Garrett, 1977) glands of the rabbit, and of the submandibular glands of the squirrel monkey (Cowley and Shackleford, 1970a), horseshoe bat (Mineda, 1977), and cat (Kidd and Garrett, 1979). In contrast, hypolemmal nerve terminals are very abundant in the SDs of certain bat species, especially in members of the genus Myotis (Tandler et al., 1989, 1990). Target cells in salivary glands lack postsynaptic specializations, so the precise cell that is being innervated cannot be determined. But in the SDs of some bats (see compendium by Tandler and Phillips, 1995), mitochondria in the SD cells immediately neighboring a synaptic vesicle-laden varicosity form an intimate relationship with these structures, partially or completely enshrouding the terminals.
Garrett and Kidd (1993) list four categories of nerve function in relation to salivary glands, including water mobilization, protein secretion, induction of synthetic activities, and maintenance of mature cells in a functioning size and state. Although these functions have the greatest applicability to end-piece cells, all of them probably are necessary in those striated ducts that engage in secretory processes. It has been shown that electrically-evoked sympathetic stimulation of the cat submandibular gland leads to 90–95% depletion of kallikrein (Barton et al., 1975). A complementary study found a dramatic loss of secretory granules from the feline SD cells after sympathetic stimulation (Garrett and Kidd, 1975). Sympathetic stimulation results in profound degranulation of SD cells in the rat sublingual gland (Garrett and Anderson, 1991). Nerve impulses upregulate salivary secretion of IgA (Carpenter et al., 1998), which takes place in SDs (vide supra). The profuse innervation of SDs in certain species suggests the need for a rapid response on the part of these ducts, either in terms of secretion or of electrolyte transport, or both.
In those species with sparse terminals, it might be expected that, to facilitate intercellular transmission of nervous stimuli, gap junctions would be abundant, but these structures have been documented by electron microscopy only in the SDs of the parotid glands of the vampire bat (Tandler et al., 1990), of the accessory parotid glands of the free-tailed bat (Tandler et al., 1998) and of the submandibular gland of the African multimammate rodent (Fig. 12). They have not been documented in SDs either by freeze-fracture (Shimono et al., 1980) or by immunocytochemistry (Hirono et al., 1995; Lee et al., 1998). Recently, however, Iwasa and Kondo (1999) have reported the presence of many point contacts (focal gap junctions?) between the plasma membranes lining the basal folds of SDs in the rat submandibular gland; it may be that such structures are abundant in the SDs of other species, but that they are overlooked in the welter of membranes that constitute the basal striations.
An example of ducts with a dual function, i.e., electrolyte homeostasis and organic secretion, is seen in avian salivary glands. The histological organization of such organs is completely different from that of the analogous organs in mammals. For example, the angularis oris salivary glands of the house sparrow consist of a main excretory duct that expands into a large central intraglandular chamber (Nagato and Tandler, 1986c). The epithelium lining this chamber is thrown into deep folds whose lumina open into the chamber. At the closed end of the folds, the cells are pure mucus-secreting cells with all of the cytological trappings of typical secretory cells. Proceeding toward the central chamber, the secretory product is reduced in amount and the cells acquire progressively more mitochondria (Fig. 11). At the same time, the lateral plasma membranes become highly folded and interlocked. Although there is no clear line of demarcation between the purely secretory cells and the mitochondria-rich cells, it seems obvious that the epithelial cells have undergone a physiological differentiation, so that the initial mucus-rich saliva is modified by the mitochondria-rich cells. Physiologically, this is precisely what happens in mammalian salivary glands, where the primary saliva is modified by the SDs. In the sparrow, the quasi-SD cells retain their organic secretory capacity, albeit in reduced form.
SIGNIFICANCE OF STRUCTURE
Comparative data from many representative mammalian species, such as presented in this review, offer opportunities to explore the significance of ultrastructural features. A cursory glance at Table 1 reveals that species, even those within a particular family, can be quite different from one another. Nevertheless, there are some overall patterns that emerge from this data set. For example, the marsupials collectively exhibit relatively little evidence of organic secretion in the parotid and submandibular gland SDs. The species of marsupials included in our sample have diverse diets and are representative of distantly related evolutionary lineages within the Subclass. This means that the similarity in their SDs is not the result of ecological factors, but instead reflects something basic about marsupial salivary glands.
The bats and rodents stand in stark contrast to the marsupials. In each of these orders, the parotid and submandibular gland SDs generally display evidence of abundant secretion (Table 1). In one instance, there is a noteworthy family pattern: in contrast to other members of the Order Carnivora, the mustelidae typically have crystalloid secretory products in SDs of both major salivary glands (Table 1). Once again, this shared feature probably does not relate to diet, which is quiet different among the sampled species. A shared feature such as this certainly suggests something special about mustelid salivary glands, but what this is is unknown for now.
An example of a broad comparison is seen in Table 2, which compares secretion in the parotid and submandibular gland SDs in more than a hundred species. It is apparent that SDs in both major salivary glands exhibit secretory features and that the relative percentages of abundance of secretory granules is remarkably similar (Table 2). For example, in both glands the SDs in about 20% of the examined species showed no evidence of organic secretion. Each of the other qualitative categories also was similar: for instance, secretory granules were abundant in the parotid and submandibular gland SDs in 51% and 43% of species, respectively. Although there is no way of knowing whether or not SDs in the parotid and submandibular glands secrete the same products, these data imply that patterns of secretion are approximately the same in the ducts of both organs. The remarkable similarity of secretory patterns between SDs in the parotid and submandibular glands supports a hypothesis that SDs in different glands in a given species co-evolved rather than specialized. At the same time, a review of the data in Table 1 shows that within a species the SDs are not necessarily the same in the parotid and submandibular glands. In other words, in some species only one gland or the other exhibits much evidence of organic secretion, whereas in other species both glands are about the same.
Table 2. Comparative qualitative assessment of organic secretion by SD cells in the parotid and submandibular glandsa
0/+ to 0/+++
++ to +++
0, none; +, few; ++, intermediate, +++, many
The bats offer the best opportunity for using comparative ultrastructural data for assessing whether there is a relationship between diet and secretion by the SDs. There are two reasons for this: (1) there are good genetic and morphological data sets about their evolution, and (2) they show ecomorphological adaptation to various diets (Baker et al., 2001; Phillips, 2000; Phillips et al., 1993; Pumo et al., 1998). Table 3 provides a comparison of secretion in SDs in correspondence to general dietary categories, e.g., frugivory and insectivory. It is clear that there is a major difference between fruit- and nectar-feeding species on the one hand and insectivorous, carnivorous, and sanguivorous species on the other. As mentioned previously, in most (>80%) of the fruit and nectar bats the parotid and submandibular SDs show little evidence of secretion. The significance of this variation in secretory pattern is unknown, but it probably means that that SDs have assumed different roles according to diet. The composition of SD granules in insectivorous and carnivorous bat species is unknown, but the electron-density of such granules suggests that they have a high protein, possibly enzymatic, content. In laboratory rats, granules in the GCT contain protease (Sreebny and Meyer, 1964); because rodent SDs seem to share many properties with SDs in non-rodent species, it is not stretching a point to postulate that some proteases might be present in SD granules in nonrodents.
Table 3. Qualitative comparison of secretion by SDs in submandibular glands in species of neotropical phyllostomid batsa
All examined species (6) of insect-meat-blood-eating bats show evidence of profuse organic secretion in the SD cells, whereas the fruit and nectar-eating species (13 examined) show relatively little or no evidence of secretion in the same cell type. 0, none; +, few; ++, intermediate; +++, many
Insects; meat; blood
In other comparative analyses, we have pointed out that fruit bats confront special physiological challenges, including water and electrolyte balance and pH of saliva (Nagato et al., 1998; Phillips, 2000; Phillips et al., 1993). Many tropical fruits are low in sodium and high in potassium. SD cells in fruit bats are specialized to deal with this exigency (Nagato et al., 1998; Tandler et al., 1997a,b; Tandler and Phillips, 1998). In some, but not all, species of fruit bats, the apical plasmalemma of the SD cells typically form leaf-like extensions that are lined on their cytoplasmic aspect with repeating units that resemble the portasomes that have been reported to occur in a variety of invertebrates (Harvey et al., 1981; Zhuang et al., 1999). As previously mentioned, portasomes have been associated with electrolyte transport, oxidative phosphorylation, and proton-pumping (Harvey et al., 1981; Nagato et al., 1998; Zhuang et al., 1999). In our comparative studies, it seems that portasomes typically occur within only one major or one accessory gland, but not in all in a given species. Another example of how SDs might be mustered to handle unbalanced salt intake is seen in at least two frugivorous bat species, where endpieces in the parotid gland have been modified into a SD-like configuration (Nagato et al., 1998; Tandler et al., 1997a). This same feature also could be related to production of acidic saliva in many fruit bats (Nagato et al., 1998; Phillips, 2000).
Finally, in our sample of specimens it is possible to ask if the SDs are the same in all species in a given genus. This is one way of addressing two questions: (1) Do SDs evolve rapidly enough to exhibit differences correlated with speciation? (2) Do SDs reflect the details of adaptation? The latter question requires a situation in which congeneric species (species that share a genus and thus have a common shared ancestry) exhibit similar or different (divergent) diets. Two examples, again bats, are shown in Table 4. The horseshoe bats of the genus Hipposideros typically feed on insects, but not necessarily on the same kinds. In this genus, it seems that the submandibular gland SDs vary not only interspecifically, but within individual glands. In one species (H. diadema), there was no evidence of secretion, whereas in the other three species some cells had abundant secretory granules (Table 4). It seems possible that in the latter species there are two classes of SD cells, or at least some type of cycle among the cells. In the second sample, we highlight three species of spear-nosed bats, genus Phyllostomus. In this instance, there are dramatic interspecies differences in SD cells. The spear-nosed bats are Neotropical and the species shown (Table 4) display a variety of dietary differences (ranging from feeding primarily on flowers and fruit to being carnivorous at times). Although we cannot match the interspecies SD differences to any specific differences in diet or physiology, the similarity in SD organic secretion between P. hastatus and P. elongatus also is reflected in ecomorphological comparisons by Freeman (2000). Likewise, the distinctiveness of the SD in P. discolor also is consistent with how it differs from the other two species in terms of ecomorphology. The multidimensional ecomorphological analyses by Freeman (2000) are based on dental and cranial morphology. These analyses illustrate how species occupy morphospace and provide a means of interpreting dietary diversity in a morphological context.
Table 4. Quantitative assessment of intrageneric variation in organic secretion by SD cells in the submandibular gland of two genera of batsa
0, none; +, few, ++, intermediate; +++, many
In summary, comparative ultrastructural studies of SDs offer a means of deciphering the possible significance of this salivary gland feature. By comparing mammalian species on the basis of evolutionary history or of diet or of other biological characteristics, we should be able to create testable hypotheses about the role(s) of secretion by SD cells
The authors appreciate the professional assistance of several colleagues who participated in field work or confirmed species identification of voucher specimens, or both. In particular, we thank Duane A. Schlitter, Stephen Williams, and Hugh H. Genoways. Kuniaki Toyoshima participated in some of the studies upon which this review is based. Carol Ayala and Thomas J. Slabe provided technical assistance for some aspects of this work, which was supported in part by a grant from the National Institute of Dental Research (BT, CJP) and Hofstra University HCLAS grants (CJP), and by financial support from the Texas Tech University and Illinois State University Departments of Biological Sciences (CJP).