Contributions of intercalated duct cells to the normal parenchyma of submandibular glands of adult rats

Authors

  • Yan-Gao Man,

    1. Department of Anatomy, College of Medicine, Howard University, Washington, DC 20059
    Current affiliation:
    1. Armed Forces Institute of Pathology, Department of Gynecology and Breast Pathology, 6825 16th. St. NW, Washington, D.C. 23060
    Search for more papers by this author
  • William D. Ball,

    Corresponding author
    1. Department of Anatomy, College of Medicine, Howard University, Washington, DC 20059
    • Department of Anatomy, College of Medicine, Howard University, 520 W St., NW, Washington, DC 20059
    Search for more papers by this author
  • Luigi Marchetti,

    1. Department of Comparative Morphological and Biochemical Sciences, University of Camerino, I-62032 Camerino, Italy
    Search for more papers by this author
  • Arthur R. Hand

    1. Department of Pediatric Dentistry, University of Connecticut Health Center, Farmington, Connecticut 06030
    Search for more papers by this author

  • The content of this manuscript is part of a dissertation submitted by Dr. Yan-gao Man in November, 1997, to the Faculty of the Graduate School of Arts and Sciences of Howard University in partial fulfillment of the requirements for the degree of Doctor of Philosophy.

Abstract

The parenchyma of the submandibular gland in the adult male rat is self-renewing, with most newly formed acinar and granular duct cells believed to differentiate from the rapidly proliferating intercalated duct (ID) compartment. Since the ID cells are phenotypically diverse, based on their different expression of perinatal secretory proteins, we systemically injected tritiated thymidine for 24 hours, and followed the pattern of thymidine distribution in cells by autoradiography and immunocytochemistry of defined cellular phenotypes over a 1-month chase period. Proliferating cells were found within all parenchymal cell compartments; they were most numerous in ID, and primarily in those cells lacking immunoreactivity for the perinatal proteins SMG-B1, -C, and -D. The labeling index (LI) of the ID cells reached a peak at 7 days postinjection, and then decreased over the next 3 weeks. Concurrently, the LI increased significantly in those cells at the junctions of ID with both acini and granular ducts, and also within these larger parenchymal elements. We conclude that the ID cells not reactive for perinatal proteins proliferate to expand the ID compartment, and that ID cells at the ends of the ducts differentiate into both acinar and granular duct cells. Our data provide no evidence for the differentiation of ID cells into cells of striated ducts (SD); however, the small number of excretory duct (ED) profiles seen in our preparations showed extremely high LI (>25%), suggesting that more extensive data might reveal a precursor role for the ED in replacement of SD cells. In addition to the stepwise passage of cells from ID to other parenchymal elements at their junctions, the reported occurrence of occasional clusters of B1-positive acini (BAC) among the typical B1-negative acini had suggested an alternate pathway, in which entire segments of newly expanded ID might develop directly into a recapitulated perinatal stage of B1-reactive cell, pursuant to becoming mature acinar cells. Consistent with this suggestion, the BAC had a fourfold greater LI than typical adult acini; moreover, when analyzed by electron microscopic immunocytochemistry, they appeared similar to the novel perinatal Type III cells both ultrastructurally and in their pattern of B1-immunogold labeling. In contrast, the less common acini showing a sublingual gland phenotype had no significant difference in LI from typical acinar cells. Overall, our results emphasize the importance of the nonimmunoreactive ID cells in normal cellular replacement, and the possibility that ID can undergo en bloc differentiation into replacement acini as well as incremental addition of single cells at the boundaries of ID with acini and with granular ducts. Anat Rec 263:202–214, 2001. © 2001 Wiley-Liss, Inc.

Intercalated ducts (ID) are the smallest elements of the salivary gland parenchyma; they drain secretory products from the acini into higher order ID before these pass into larger ducts, and, ultimately, into the major excretory ducts that discharge their contents into the oral cavity (Young and van Lennep, 1978). The walls of the ID consist of a single layer of low cuboidal cells, which have been reported to constitute 20% of the parenchymal cell population (Chang, 1974). At the ultrastructural level, ID cells have been shown to have less prominent rough endoplasmic reticulum and Golgi apparatus, and fewer secretory granules than do acinar cells (Tandler et al., 1998). The current study was prompted by two prominent features of the ID cell population: The first is the existence of multiple cellular phenotypes marked by novel combinations of cell-restricted secretory proteins (Man et al., 1995); the second is the apparent role of ID cells as stem cells for the normal expansion and replenishment of adult parenchymal cell phenotypes (Denny et al., 1997; Denny and Denny, 1999).

Our previous studies had shown that during prenatal and early postnatal development of the rat submandibular gland, the immature acini were characterized by two distinct types of secretory cells: Type I cells expressed a prominent secretory protein of apparent molecular weight 89 kDa (Protein C); Type III cells expressed a group of related proteins in the molecular weight range 16–28 kDa that we originally referred to as the B1-immunoreactive proteins (B1-IP); still another major secretory protein was expressed in both of the neonatal cell types (Protein D, 175 kDa) (Ball and Redman, 1984; Ball et al., 1988a,b, 1991). The B1-IP include the products of two distinct genes: The rat Psp gene has 70.6% homology with the Psp gene previously described in the mouse parotid gland (Ball et al., 1993; Madsen and Hjorth, 1985; Mirels and Ball, 1992; Mirels et al., 1993; Shaw and Schibler, 1986); the second BI-IP gene of the rat, which we have identified as Smgb, is not expressed in the mouse. In the rat, variant isoforms of Smgb gene products appear to be the result of differential glycosylation and/or posttranslational proteolysis of a single protein (Mirels et al., 1998). In the second postnatal week, the Type III cells downregulate their expression of the Psp and Smgb genes as they differentiate into seromucous acinar cells, which express adult secretory proteins, including mucin and glx-rich proteins (GRP) (Moreira et al., 1989, 1991). Between the first and third weeks postnatally, Type I cells appear to move into the intercalated ducts, where they constitute a substantial cell population, until most disappear between 25 and 30 days postpartum (Hayashi et al., 2000; Hecht et al., 2000). Scattered cells similar to Type I cells in morphology and Protein C immunoreactivity remain in the ID and in juxta-acinar locations throughout adulthood (Ball et al., 1988b; Moreira et al., 1990). Other, distinctly different cells in the ID express the Psp and/or Smgb genes, which are characteristic of perinatal Type III cells, although no developmental continuity of these ID cells with the perinatal Type III cells is apparent (Ball et al., 1988a). Also present in the adult gland are occasional clusters of acini that do express Psp and/or Smgb, but not Protein C, and that are negative for the typical adult marker proteins (Man et al., 1995). We speculated that these might be replacement acini, in transition from ID cells to mature adult acini. In the adult ID, then, one sees a situation parallel to that seen in the perinatal gland, with cells expressing high levels of Protein C being completely distinct from those expressing the Psp- and Smgb-encoded protein products. Also reminiscent of the perinatal period is the expression of Protein D in cells that express products of Psp/Smgb, and also in cells that express Protein C. In both cases, Protein D is present in only a small fraction of the C-reactive or the B1-reactive cells. Finally, still another population of cells in the adult ID are negative for all of the perinatal proteins (Man et al., 1995). It was the goal of this study to determine the degree of involvement of ID cells in cellular replacement and/or expansion, to identify which of the phenotypically identifiable ID cells are the progenitors of the newly generated cell populations, and to investigate the potential role of the B1-positive acini in the replenishment and/or expansion of the different parenchymal cell populations.

MATERIALS AND METHODS

Materials

Antibodies to proteins B1, C, and D were produced and characterized as previously reported (Ball et al., 1988a,b, 1991). Chemicals for electron microscopic tissue fixation and embedding were purchased from Electron Microscopy Sciences (Fort Washington, PA) and Ladd Research Industries (Burlington, VT). Goat anti-rabbit gold complex was obtained from Amersham Life Sciences Products (Arlington Hts., IL); other immunological reagents were obtained from Organon-Teknika-Cappel (Westchester, PA). Light microscopic immunostaining reagents, including BioStain super ABC kits, alkaline phosphatase chromogen kit, and peroxidase chromogen kit, were purchased from Biomeda Corp. (Foster City, CA). Methyl-3H-thymidine (3H-TdR) was purchased from Amersham Pharmacia Biotech, Inc (Piscataway, NJ). Autoradiography emulsion NTB2, Developer D19, and Fixer were purchased from Kodak (Rochester, NY). Gelatin was obtained from BioRad (Richmond, CA), and Hank's balanced salt solution (HBSS) from Flow Laboratories (McLean, VA). Other chemicals and solvents were reagent grade and were purchased from J.T. Baker, Inc. (Phillipsburg, NJ), Fisher Scientific (Pittsburgh, PA), and Sigma Chemical Co. (St. Louis, MO). All solutions were made up in deionized, glass-distilled water.

Experimental Animals and 3H-TdR Injection

A total of 30, 2-month-old male Sprague-Dawley rats, were maintained as previously described (Ball et al.,1988a). All procedures involving animals were approved by the Institutional Animal Care and Use Committee of Howard University. Animals were numbered with permanent ink on the ear, and each rat was intraperitoneally injected with 3H-TdR at a Specific Activity of 351 Ci/mg and a total inoculum of 0.5 μCi/g body weight, four times at 6-hr intervals.

Preparation of Tissue Sections for Light Microscopic Immunocytochemical and Autoradiographic Analysis

One hour after each injection (i.e., 1, 7, 13, and 19 hr after the first injection), and at 3, 5, 7, 14, 21, and 28 days after the first 3H-TdR injection, three rats were sacrificed, and the left SMG of each rat was removed, cut in half, and fixed in 4% paraformaldehyde overnight at 47°C. Fixed tissues were briefly washed with water, dehydrated with ascending concentrations of ethanol, cleared with xylene, and embedded in paraffin. Serial 4–5-μm sections were placed on polylysine-coated slides, heated at 60°C for 2 hr, deparaffinized with xylene, and washed with ethanol and water. The first and last sections from each animal were stained with hematoxylin and eosin for morphological assessment. The rest of the sections were immunostained with anti-B1 (1.1 μg/ml), anti-C (1.4 μg/ml}, a mixture of anti-B1 and anti-C, or anti-D IgG (20 μg/ml), as previously described (Man et al., 1995), using 3 to 5 tissue sections from each animal for each antibody. Briefly, sections were treated with 0.3% H2O2 in methanol for 30 min to block endogenous peroxidase activity, and with 10% normal serum for 30 min to eliminate nonspecific binding. Sections were incubated overnight at 4°C with the primary antibody or control solution. Negative controls included substitution of preimmune IgG for the primary antibody or omission of the primary or secondary antibody from the staining sequence. Controls all showed no reactivity. After the primary antibody incubation, sections were washed sequentially with PBS, avidin-peroxidase or avidin-alkaline phosphatase solution, and with chromogen, and dried at room temperature for 10–20 min. Kodak nuclear track emulsion was liquefied at 40–45°C in a darkroom, air bubbles were removed, and the immunostained sections were coated by dipping into the emulsion for a few seconds. To evaluate the possible obscuring of 3H-TdR labeling by the chromogen, deparaffinized tissue sections from each animal were directly coated with Kodak emulsion without immunostaining. The slides were drained and air-dried for 20–30 min, placed in a light-proof box containing Drierite, and stored at 4°C for 3 to 4 weeks. The slides were developed with Kodak developers Dektol or D19 for 2–3 min, fixed with Kodak fixer for 3–5 min, and washed with running tap water for 10–20 min. Sections were lightly counterstained with hematoxylin, washed with tap water, dried at room temperature, and coverslips were mounted with 3–5% gelatin mounting medium (liquefied at 40–45°C). In serial sections, control and immunostained preparations showed identical thymidine-labeling patterns, indicating that the immunochemical staining did not interfere with detection of the tritium label.

Electron-Microscopic (EM) Immunostaining of B1-Positive Acini

Glands from 7 adult rats were minced with fine scissors, briefly rinsed in HBSS, and fixed in 4% buffered paraformaldehyde overnight at 4°C. After fixation, tissues were washed in 0.1 M phosphate buffer (pH 7.4), dehydrated in increasing concentrations of ethanol, substituted with propylene oxide, and embedded in Epon 812. Sections of 1 μm thickness were immunostained with anti-B1 antibody to identify the tissue blocks containing B1-positive acini. Thin sections of selected areas containing B1-positive acini were immunolabeled for EM observation as previously described for Protein B1 (Moreira et al., 1991; Hand, 1995).

Assessment of 3H-TdR Labeling Indices Among Parenchymal Cells

To evaluate the 3H-TdR labeling, the slides were covered with a thin sheet of transparent film with 1 mm2 squares marked with black lines, to define the boundary of a given field. Each tissue section was divided into three equal parts; in each part, cells of each phenotype in 3–5 randomly selected fields were observed using a 40× objective and 10× ocular. The identification of 3H-TdR labeled cells was based on previously published criteria, with a labeled cell generally being defined when at least five silver grains were seen over its nucleus (Cleaver, 1967; Leblond et al., 1959). An exception was Figure 15, in which three grains were used to define a labeled nucleus because of considerably lower background in this preparation.

Cell Counting and Statistical Analysis

In order to determine the labeling index of the different parenchymal cell types, a total of 1,000–3,000 cells of each phenotype were counted for each animal, and the number of labeled and total cells in the three animals of each group were summed. The labeling index (LI) of each cell type was obtained by dividing the sum of labeled cells by the sum of cells counted and multiplying by 100. The differences in LI among different cellular phenotypes (see Figs. 17,18) were analyzed by ANOVA, and by a 2-tailed Pearson correlation analysis, using the Statistical Package for the Social Sciences (Version 10.0) 1999. For assessing the possible transformation of ID cells into acinar (AC) and/or granular duct (GD) cells, all the ID-AC and ID-GD junctions encountered in longitudinal sections of these structures were dotted on the slide with a black marker, and the labeling indices of the cells within three cells to each side of the junction were statistically compared at different times after 3H-TdR injection. The increases in labeled cells at these junctions (Figs. 19,20) were analyzed by a 2-tailed Pearson correlation analysis, as above. Since the B1-positive acini (BAC) were not seen in every section, and the anomalous mucous acini (AMA) were found in only 10% of the animals, the numbers of labeled and unlabeled cells in each of these structures in different animals at different time-points postinjection were pooled (although the number in each animal was separately counted). A similar number of typical adult acinar cells was counted in each of the animals containing BAC and AMA, for the purpose of statistical comparison. The differences between the sample means in the above comparisons were tested for significance using the Student's t-test, since the numbers are normally distributed and on a continuous scale (Hassard, 1991; Sokal and Rohlf, 1987).

RESULTS

Labeling Indices (LI) of the Different Parenchymal Cell Types

3H-TdR-labeled cells were seen throughout the gland, although more labeled cells appeared to be at the periphery than in the center. Labeled cells were seen in all parenchymal cell types, and the number of labeled cells varied considerably among cell types and among animals of the same group. 3H-TdR-labeled acinar cells (AC) were generally singly distributed during the first 3 days. At 5 to 7 days, about one-third of the labeled cells were seen in pairs, and clusters of more than two labeled cells were occasionally seen (Fig. 1), suggesting that the acinar cells had divided. Labeled granular duct (GD) cells were also generally distributed singly during the first 5 days. By 7 days, about one-third of labeled GD cells were in pairs, and clusters of more than two labeled cells were also seen, suggesting that they had arisen by division (Fig. 2). The same general pattern was seen in striated ducts (SD); by 7 days after injection, pairs or larger clusters of labeled SD cells were detectable. Figure 3 shows a group of three labeled cells in a longitudinal section of a striated duct. Morphologically distinct excretory ducts (ED) were encountered in sections from only 10% (3 of 30) of the animals, because the plane of section usually missed these larger, less frequent structures. Observations on these limited samples, however, revealed that in these ED the cells had a very high labeling index. Figure 4 shows a cross-sectional profile of an ED at 5 days after 3H-TdR injection, in which 23 of 84 (27%) ED cells displayed 3H-TdR labeling. This was a higher percentage than seen in any other parenchymal structure, including the intercalated ducts (see below). The small sample size, however, precluded a statistical comparison.

Figure 1.

A cluster of 3H-TdR-labeled acinar (AC) cells at seven days postinjection. Reacted with anti-B1/C IgG. Arrows identify 3H-TdR-labeled acinar cells. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 2.

Thymidine-labeled granular duct (GD) cells at 7 days. Reacted with anti-B1/C IgG. Three labeled GD cells are identified with arrows. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 3.

Labeled striated duct (SD) cells at 7 days. Reacted with anti-B1/C IgG. Arrows identify three labeled SD nuclei. Arrowheads identify two labeled stromal cells. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 4.

Labeled excretory duct (ED) cells at 5 days. Stained with hematoxylin. Many cells in an excretory duct (star) have thymidine-labeled nuclei. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 5.

Intercalated duct (ID) at one day. Reacted with anti-B1/C. One ID cell is heavily labeled with tritiated thymidine (arrow) and is not immunoreactive. Many ID cells are immunoreactive (red-staining) but not thymidine-labeled, and these connect to acini (stars). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 6.

Intercalated ducts at 7 days. Reacted with anti-B1/C. A cross-sectional profile of an ID (arrow) contains four thymidine-labeled, nonimmunoreactive cells. Arrowheads identify acini, and stars mark granular ducts. Immunoreactive ID cells (red-staining) are not thymidine-labeled. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Localization of 3H-TdR-Labeled ID Cells

After the four thymidine injections, labeled ID cells were detectable throughout the gland, and the ID phenotypes were distinguished on the basis of their pattern of expression of the several perinatal proteins (Man et al., 1995). In most subsequent preparations (e.g., Figures 5–8), the population of nonimmunoreactive cells was detected by their lack of reactivity with a mixture of antibodies to Protein C and to Protein B1. 3H-Thymidine-labeled ID cells were generally singly distributed, and often were seen toward the middle of the ID (Fig. 5). Five to seven days after injection, about half of the labeled cells were in pairs, and clusters of labeled cells were more frequently seen in the ID than in AC, GD, and SD cell populations. These clusters were generally located at the periphery of the gland, often at the tips of lobules. In Figure 6, a section at 7 days postinjection shows a cross-sectioned profile of an ID in which 4 of 5 cells show 3H-TdR labeling but no immunoreactivity, while none of the adjacent GD, AC or immunostained ID cells are thymidine-labeled. Figure 7 also shows a section at 7 days, in which most cells in a cross-sectioned profile of an ID are 3H-thymidine-labeled, but none are immunoreactive. In two other ID profiles, a total of 10 cells are immunoreactive; one reactive ID cell near an ID-AC junction is also tritium-labeled, while two profiles of GD and all acini in the field show neither immunoreactivity nor 3H-TdR labeling. Figure 8 shows a section of a rat at 5 days after injection in which no immunoreactive ID cells show 3H-TdR incorporation, while five unreactive ID cells in the same field display thymidine labeling.

Figure 7.

Intercalated ducts at 7 days. Reacted with anti-B1/C. A nonimmunoreactive cross-sectional profile of an ID (large arrow) contains four labeled nuclei. Another ID in longitudinal section has many immunoreactive cells, one of which has thymidine labeling (small arrow). Granular ducts are marked with stars. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 8.

Intercalated ducts at 5 days. Reacted with anti-B1/C. Several thymidine-labeled cells are nonimmunoreactive (arrows). The red-stained, immunoreactive ID cells are unlabeled. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 9.

3H-TdR-labeled cells at the ID-AC junction, 2 weeks postinjection. Reacted with anti-C IgG. Thymidine-labeled cells (arrows) are seen at junctions with acini (stars). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 10.

Labeled cells at the ID-AC junction after 2 weeks. Reacted with anti-protein D IgG. Four thymidine-labeled ID cells are marked with arrows. Four immunoreactive ID and juxta-acinar cells are unlabeled (arrowheads). In general, acini (stars) are faintly reactive for Protein D. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 11.

Labeled ID cells at 3 weeks. Reacted with anti-B1/C IgG. To the right, a cross-section of a nonimmunoreactive ID (arrow) shows thymidine-labeled cells. A single immunoreactive and thymidine-labeled ID cell is marked with an arrowhead. To the left, several labeled but unreactive cells (arrows) mark the junction of an immunoreactive ID with a GD (star). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 12.

Labeled cells at the ID/GD junction after 3 weeks. Reacted for anti-B1/C IgG. Thymidine label is seen in cells at the junction (arrows) of an immunoreactive ID with a GD. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Labeling Indices (LI) of Phenotypically Different ID Cells

A systematic count confirmed what had appeared true from examples like those in Figures 5–8. The data showed that the cells expressing Protein C and those expressing the products of the Psp and Smgb genes had low labeling indices that were not significantly different (see Fig. 17). Those cells lacking immunoreactivity for perinatal proteins had a LI that was more than 10-fold higher than cells reactive for either Protein C or Psp and Smgb products after one week, and substantially higher at all time points. The peak value at one week then decreased markedly through 4 weeks postinjection.

Measurement of LI in the Different Parenchymal Cell Types Over Four Weeks of Postnatal Development

The overall preponderance of ID cells in the proliferating cell population is quantitatively documented in Figure 18. After a single 3H-TdR injection, 1% of the ID cells showed 3H-TdR incorporation; after multiple injections, the number of labeled cells increased to 2.5% at 19 hr. The labeling index increased and reached a peak of 4% at one week, at which time it was more than 10-fold greater than the LI of AC, GD, or SD. The LI then dropped to less than 2% by 4 weeks. The other cell types, after a single 3H-TdR injection, showed labeling indices of about 0.2% for AC, 0.1% for GD, and less than 0.2% for SD cells; after multiple injections, the indices at 19 hr were 0.23, 0.25, and 0.3%, respectively. During the first week, the labeling indices for these three cell types remained very low, but after one week all labeling indices increased markedly, reaching 1.4% (AC), 1.0% (GD), and 0.8% (SD) at 2 weeks. At 3 and 4 weeks, the LI of the three cell types was not further increased, and there were no major differences in LI among them.

3H-Thymidine Incorporation in Cells at ID-AC Junctions

Figure 9 shows a section at 2 weeks after injection, immunostained with anti-C antibody. Two acini are connected with a long ID, in which three labeled ID cells are seen near the junctions, and two others near the middle of the duct profile. Figure 10 shows a section from a rat sacrificed at 2 weeks after injection and immunostained with anti-D antibody. Our previous study had shown that Protein D is found at high levels only in those ID cells that are also reactive for Protein B1 or Protein C (Man et al., 1995). Protein D reactivity is seen at low levels in the acini, but this is easily distinguishable from the intense signal seen in the reactive subset of ID cells. An acinus is connected with an ID, in which four labeled cells, all devoid of D-reactivity, are seen on one side. Two unlabeled ID cells are strongly reactive for protein D. Two other highly reactive cells in the field appear to be juxta-acinar cells, which often are reactive with the antibodies to neonatal proteins (Man et al., 1995). The quantitation of changes in the LI of ID cells at the ID/AC junction is shown in Figure 19. The LI of cells at ID-AC junctions was 1.2% at 19 hr after injection and 1.7% at one week, then increasing to 2.1% at 2 weeks, 2.3% at 3 weeks, and 2.5% at 4 weeks.

3H-Thymidine Incorporation Into Cells at ID-GD Junctions.

Figure 11 shows a field at 3 weeks with a cluster of three 3H-TdR-labeled cells at the junction of an intercalated duct with a granular duct. This field has about 12 cells that have immunoreactivity with anti-B1 or anti-C antibodies, and only one of these has thymidine label. Figures 12 and 13 show two sections from the same rat at 3 weeks after injection, immunostained with a mixture of anti-B1 and anti-C IgG. In Figure 12, two GD cells at the ID-GD junction show labeling; the other GD cells are unlabeled. The two labeled GD cells are adjacent to two labeled ID cells, which show reactivity to anti-B1/C IgG. In Figure 13, a longitudinal profile of GD has three labeled cells at the ID-GD junction, again adjacent to two labeled cells in the connecting ID. The pattern of 3H-TdR incorporation and the time-related changes among cells at the ID-GD junctions (see Fig. 20) were similar to those of cells at or near ID-AC junctions (see Fig. 19). The LI was 1% at 19 hr, after multiple injections, then increased to 1.5% at 1 week, 2.3% at 2 weeks, 2.5% at 3 weeks, and 2.8% at 4 weeks.

Figure 13.

Labeled cells at the ID/GD junction after 3 weeks. Reacted for anti-B1/C IgG. Thymidine label is seen in cells at the junction (arrows) of an ID with a GD (star). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 14.

Thymidine labeling in a field that contains a group of B1-positive acini, 2 weeks postinjection. Reacted for anti-B1/C. Four cells within the B1-positive acini show labeling in immunoreactive cells (arrows). Three thymidine-labeled but unreactive cells appear to be in ID (arrowheads). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 15.

A large group of B1-positive acini at the tip of a lobule, 2 weeks postinjection. Reacted for anti-B1/C. Many nuclei are labeled, while labeling intensity is low. Inset: Labeled nuclei at higher magnification (scale bar = 5 μm). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 16.

A group of anomalous mucous acini (AMA), 2 weeks postinjection. Reacted for anti-B1/C. Profiles of AMA (stars) have no labeled cells. Two thymidine-labeled cells (arrows) are adjacent to these acini. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

3H-Thymidine Incorporation in Cells of B1-Positive Acini (BAC)

The number of acinar profiles seen in these clusters was highly variable, from less than 5 to over 30 in a given section. These acini were smaller than typical adult acini and often located at the periphery of the gland. Figure 14 shows a group of B1-positive acini, 2 weeks after injection. Five labeled cells are seen in these acini, while no labeled cells are found in the surrounding typical acini. As previously described, BAC were often located at or near the tips of lobules (Man et al., 1995). Figure 15 shows a large group of B1-positive acini at the tip of a lobule at 2 weeks postinjection. Of the 109 cells visible in this group, 25 (23%) showed 3H-TdR labeling, although the intensity was very low. In order to quantitate the LI of these relatively infrequent structures, the B1-positive acini seen at all times after 3H-TdR injection were pooled and compared with the adjacent typical acinar cells. The statistical comparison of the 3H-TdR labeling indices between the B1-positive and typical acinar cells is summarized in Table 1. The LI among different BAC was highly variable, ranging from 0 to 23%, with an average of 1.9%, compared to 0.4% for typical adult acini.

Table 1. Comparison of 3H-TdR labeling indices between B1-positive acinar cells (BAC) and typical adult acinar cells (TAAC)
Cell typeNumber countedNumber labeledLabeling index (%)P*
  • *

    Student's t-test was used. The B1-positive acinar cells have a significantly higher (P < 0.05) 3H-TdR labeling index than the typical adult acinar cells.

BPAC2,903551.9
TAAC2,900120.4<0.05

Electron Microscopic Immunocytochemistry of BAC

Immunogold cytochemistry was performed on thin sections of blocks in which BAC had been identified by light microscopic immunolabeling, and showed that these cells contained secretory granules that were morphologically and immunocytochemically similar to those of neonatal Type III cells (Moreira et al., 1990, 1991). Figure 21 shows a section of a submandibular gland from a 2-month-old rat, which was labeled with anti-Protein B1 antibody. The two acinar cells contained secretory granules with electron-lucent substructures in a moderately electron-dense matrix; these are morphologically the same as the Type III granules of perinatal glands and they are reactive for protein B1.

3H-Thymidine Incorporation in Cells of Anomalous Mucous Acini (AMA)

In our previous report (Man et al., 1995), we described the occasional occurrence in the adult submandibular gland, of scattered small groups of acini with the histological appearance of sublingual gland mucous cells, rather than of the typical seromucous cells of SMG acini (Young and van Lennep, 1978). Like typical adult acini, they were negative for B1 reactivity, but also failed to react for the normal acinar marker, GRP; instead, they expressed sublingual gland mucin. In this autoradiographic study, they were found in 2 of 30 rats. Figure 16 shows a group of these anomalous mucous acini (AMA), which show no thymidine-labeled cells, while two labeled cells are adjacent to the AMA. It is not clear whether these labeled cells are in ID, or perhaps in serous demilunes, which were seen around AMA as described in our previous report (Man et al., 1995). Table 2 shows the LI of AMA to be not significantly different from that of typical acinar cells.

Figure 17.

Labeling Indices (LI) of phenotypically different ID cells at different times postinjection. There were no statistically significant differences between the mean labeling indices of B1 and C-labeled cells (P = 0.938). The means of the labeling indices of B1 and C-labeled cells were statistically significantly different from the mean of the labeling index of the negative cells (P < 0.001). In Figures 17–20, vertical lines are the range of LI values among the three rats in each sample. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 18.

Labeling indices (LI) of the major parenchymal cell types at different times postinjection. There were no statistically significant differences between the mean labeling indices of AC and GD cells (P = 1.000), AC and SD cells (P = 0.999), and GD and SD cells (P = 1.000) over the time span between 19 hr and 4 weeks. For the same time span, the mean of the labeling index of the ID cells differed significantly from the mean labeling indices of the AC, GD, and SD cells (P < 0.001 for all three cell types). The LI of ID cells reached a maximum at one week, at a level about 10-fold greater than that of the other cell types. From 1 to 4 weeks, there is a statistically significant positive correlation of LI with time (P < 0.01) for AC, GD, and SD cells, and a statistically significant negative correlation of LI with time (P < 0.01) for ID cells. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 19.

Number of labeled ID cells at ID-AC junctions at different times. The bars give the LI for cells located within three cells from the junction. LI of these cells increases significantly from 1.2% at 19 hr to 2.5% at 4 weeks (P < 0.001). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 20.

Number of labeled cells at ID/GD junctions at different times. The bars give the LI for cells located within three cells from the junctions. LI increases significantly from 1.0 % at 19 hr to 2.8% at 4 weeks (P < 0.001). All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Figure 21.

Electron microscopic immunolabeling of B1-positive acini at 2 weeks postinjection. Immunoreactivity with anti-B1 IgG was detected with 10-nm colloidal gold particles conjugated to GAR-IgG. The two acinar cells contain secretory granules that are morphologically similar to those of neonatal Type III cells, and these are marked by the gold particles. All times given in legends for all figures are the days following injection of labeled thymidine. Scale bars = 20 μm except where noted for the insert in Figure 15.

Table 2. Comparison of 3H-TdR labeling indices between anomalous mucous acinar cells (AMAC) and typical adult acinar cells (TAAC)
Cell typeNumber countedNumber labeledLabeling index (%)P*
  • *

    Student's t-test was used. The 3H-TdR labeling index in anomalous mucous acinar cells is not significantly different (P > 0.05) from typical adult acinar cells.

AMAC1,23790.73
TAAC1,24060.48>0.05

DISCUSSION

Normal Cellular Replenishment

The previous findings of other investigators had shown that, in rodent salivary glands, the populations of every parenchymal cell type contain proliferating cells; thus each type is able to undergo some replacement of dying cells and perhaps contribute to the expansion of the cell populations (Chai et al., 1993; Denny et al., 1990a,b; 1993; Denny and Denny, 1999; Zajicek et al., 1985). Consistent with these reports our data showed that within the first day after 3H-TdR injection, all parenchymal structures contained labeled cells and labeled cell clusters, suggesting that all are capable of at least some self-replenishment. For the AC, GD, and SD populations the 3H-TdR labeling indices of these larger parenchymal elements did not change significantly through the first week, suggesting that these labeled cells had not undergone further division into cells of the same type. The data discussed below strongly indicate that the subsequent increase in the LI of AC and GD from 1 to 2 weeks resulted from the differentiation of labeled ID cells.

Progenitor Cells in Intercalated Ducts

Our previous findings had suggested the working hypothesis that the B1-reactive subpopulation of intercalated duct cells might include the precursors of replacement acinar cells, first becoming intermediate cell types in the B1-reactive acini (BAC) of adult glands (Man et al., 1995). The data presented here, however, clearly eliminate this original hypothesis, since the ID cell population unreactive for perinatal proteins contains the actively dividing progenitors of new ID cells as well as of new acinar and granular duct cells. On the other hand, it leaves open the possibility, discussed below, that the BAC originate from the unreactive ID cells, and are intermediates in a pathway leading from ID to AC cells.

Differentiation of ID to AC and GD Cells at the Compartmental Boundaries

The increase in labeled AC and decrease in labeled ID cells between 1 and 2 weeks postinjection suggest the differentiation of ID into AC. The observed increase of labeled cells at the ID-AC boundary as a function of time also was consistent with the model proposing a translocation and differentiation of ID cells. Similar data suggest the differentiation and addition of GD cells at the boundary with ID. Our data are consistent with the conclusion from previous reports that addition of cells to both AC and GD compartments occurs at boundaries with ID in the SMG of the adult rat (Zajicek et al., 1985) and the female mouse (Denny et al., 1990).

Potential Role of B1-Positive Acini in Replacement of Acinar Tissue

If, as seems to be the case, the ID cells not expressing perinatal proteins are the progenitors of replacement acinar cells, then the existence of the BAC suggests an alternate means of replacement that is different than the stepwise addition of single, newly differentiated acinar cells described above. The rapidly proliferating ID cells might form larger units that could differentiate en bloc into acinar units, perhaps undergoing morphogenesis and cytodifferentiation similar to that occurring in perinatal development. This could involve a recapitulation of the transient perinatal state of differentiation seen in the perinatal Type III cells, which express the genes Smgb and Psp, encoding Proteins SMGB1/SMGB2 and PSP, respectively (Mirels et al., 1998). In fact, the cells of the BAC have the immunocytochemical specificities of neonatal Type III cells, being reactive for the expressed products of the Smgb and Psp genes and for Protein D, while showing no reactivity for the adult markers, mucin, and glx-rich proteins (Ball et al., 1991; Man et al., 1995; Moreira et al., 1991). The secretion granules of the cells in the BAC have the ultrastructural morphology and the subcellular pattern of anti-B1 immunoreactivity that are seen in the perinatal Type III cells. One would predict that a thorough search would find a range of transitional acini, from those like the ones we have described, that are just beginning to change into mature seromucous acinar cells, to those that have almost completed this transition, having downregulated the Psp and Smgb genes and activated the adult genes encoding mucin and GRP (Moreira et al., 1991). Also, it should be kept in mind that we do not know the detailed timing of events in the sequence from undifferentiated embryonic cell to Type III cell to mature acinar cell, and, thus, how many cells at each stage of development would be found in an adult gland.

Possible En Bloc Conversion of ID Segments into B1-Positive Acini

The possible direct development of BAC from large segments of ID is consistent with a number of older and recent findings. These include:

  • The finding of Auerbach (1964) that epithelial tissue from adult mouse submandibular gland could undergo branching morphogenesis in vitro when recombined with embryonic salivary mesenchyme.

  • The observations of Denny et al. (1990) showing three-dimensional reconstructions of adult mouse submandibular parenchyma with ID ending blindly, in a conformation geometrically well-suited to initiate the formation of a new acinar cluster.

  • The report of Royce et al. (1993) describing the differentiation of cells from a human submandibular ductal cell line into functioning acini capable of synthesizing and secreting amylase (a marker for human but not rat submandibular glands).

  • The description in this study of substantial clusters of BAC at the ends of lobules, as well as elongate profiles of B1-positive cells of a diameter intermediate between ID and AC.

  • The finding in our parallel study of regeneration, 3 days after a partial extirpation of the gland, of numerous small, cordlike or ductlike structures, near the point of surgical incision. These had many thymidine-labeled, but nonimmunoreactive cells (Man, unpublished communication).

  • Lastly, it should be kept in mind that evaluating the presumed long-term fate of rapidly dividing progenitor populations will at some point become complicated by the dilution of a fixed label (i.e., incorporated 3H-thymidine) with ensuing divisions, to the point where the radioactive signal no longer is distinguishable over background. For example, a nucleus with 48 silver grains after four divisions would have 3 grains and thus be scored as unlabeled using the criterion (5 grains) used here. Under these circumstances, then, parenchymal structures derived from the more rapidly dividing cell populations would show no label. For this reason, the LI calculated for the BAC (Table 1) might well underestimate the actual amount of proliferation that the precursor cells had undergone. In future studies, it may be advisable to use a proliferating marker, such as a rodent-specific retrovirus, which can provide permanent marking of cells in S-phase because the marker is replicated during subsequent divisions (Barka and van der Noen, 1996; Bralet et al., 1994).

Replacement of Cells in Striated and Excretory Ducts

Other than the modest self-replenishment reported, our data provide little information on replacement of striated duct cells. It seems unlikely that proliferating cells originating in ID transit through a granular duct phenotype to become striated duct cells. Given the very high labeling index (>25%) seen in the profiles of ED found in this study and in our report of regeneration in the adult SMG, a potential replacement pathway for SD cells could originate in the excretory ducts (ED). The paucity of ED profiles in our tissue samples, however, does not permit evaluation of this possibility from the data at hand. Future studies will include ED samples sufficient to address this question.

Occurrence of the “Anomalous Mucous Acini” (AMA) in the Submandibular Parenchyma

These curious acinar structures, which have the morphological and immunocytochemical features of sublingual acini, were described in our previous report (Man et al., 1995), and their occurrence, here and in our accompanying study of cellular replacement during regeneration (Man, unpublished communication), adds little to our understanding of their possible significance, except to emphasize that they occur with some regularity. Should future experimental studies fortuitously provide circumstances under which their formation is evoked with some predictability, they could become useful as a model for investigating the switch points that control specific salivary cytodifferentiation.

Acknowledgements

We thank Drs. Paul C. Denny, James C. McKenzie, and Robert S. Redman for critically reading the manuscript. We thank Mr. Gregory L. Robertson for his expert assistance in electronic reproduction of the figures. We are indebted to Dr. Michael R. Peterson for performing the statistical analyses used here. Dr. Peterson is Chairman of the Department of Epidemiology, Repository and Research Services at the Armed Forces Institute of Pathology, Washington, DC. This study was supported by NIH grant DE-06635 to W.D. Ball, and by support from the University of Connecticut Health Center to A.R. Hand.

Ancillary