Analysis of Histone H2BGFP Retention in Mouse Submandibular Gland Reveals Actively Dividing Stem Cell Populations

Abstract

The purpose of this study was to use histone 2B–green fluorescent protein (H2BGFP) pulse-chase experiments to provide a broad view of population dynamics in salivary gland and to identify the quiescent stem cells that had previously been suggested to reside in the gland. Two transgenic mouse models in which inducible H2BGFP expression was regulated either by keratin (K)14 promoter or by a ubiquitous promoter were generated. The level of fluorescent label in the submandibular gland induced by a pulse of H2BGFP expression was monitored over a period of 18 weeks of chase. Efficient targeting of H2BGFP label to the relatively undifferentiated ductal cells by K14 promoter did not identify a quiescent population of stem cells. Ubiquitous targeting of all ductal cells identified label-retaining cells but these cells were mapped to the more differentiating ductal compartments. Furthermore, they did not display the major characteristics of quiescent stem cells including the undifferentiated phenotype, mobilization in response to injury, and the clonogenicity in culture. Quantitative assessment of H2BGFP loss in various ductal compartments and short-term lineage tracing of K14⁺ ductal cells were consistent with the presence of actively dividing pools of stem/progenitor cells in the intercalated ducts and the basal layer of excretory ducts functioning independently during homeostasis.

Introduction

Secretion of saliva by salivary glands (SGs) is essential for oral health. Currently, there is a strong interest in gene- and cell-based therapies to rescue SG function following irreversible damage caused by various conditions including radiation therapy of head and neck cancers, autoimmune diseases, cytotoxic insults, and aging [1,2]. However, developing effective therapeutic strategies requires a clear understanding of the cellular mechanism of renewal and regeneration in SG. The submandibular gland (SMG) has been extensively used as a classical model of major SG, and it is composed of three differentiating epithelial tissues including, acini, ducts, and myoepithelial cells (MECs). Acini, the main secretory units, secrete saliva into a ductal system formed by the intercalated ducts, granular ducts, striated ducts, and finally, excretory ducts [3]. Classical cell kinetic studies imply that both acinar and ductal cells replicate and apoptose, and therefore must be periodically replaced by progenitor cells [4–6].

More recently, several biomarkers have been used to identify SG stem/progenitor cells. A progenitor cell population expressing Ascl3 transcription factor was identified in the SG ducts [7]; however, specific ablation of these cells did not compromise gland function. This suggests either contribution from other progenitors or a different mechanism for normal maintenance and regeneration of the gland [8]. Keratin (K)14 and/or K5 have also been shown to mark progenitor cells during embryonic gland development [9,10]; however, whether K14/K5-positive cells in adult gland include stem/progenitor cells has not been determined [11]. In addition to these markers, several antigens commonly identified in stem/progenitor cells in many organs, such as cKit, Sca-1, CD133, CD44, CD24, and CD49f, have been shown to be present in SMG [12,13]. Progenitor cells expressing cKit have the most robust regenerative capacity and transplantation of as low as 300 salisphere-derived cKit⁺ cells have been shown to rescue secretory function of SMG in a mouse model of radiation-induced injury [13–15]. However, the exact location and the contribution of this population to cell renewal during homeostasis remain to be defined. Currently, although there is substantial evidence indicating the presence of several stem/progenitor populations within the adult SMG, the contribution of these populations to gland maintenance and repair and the relationship between them remains unclear [2,11].

Adult stem cells are defined as relatively undifferentiated cells that have the ability to self-renew and to generate progeny that are fated to differentiate into at least one differentiated lineage [16]. Both active and quiescent stem cells have been identified in various mammalian tissues, and they may coexist in adjoining locations in some of these tissues [17]. Currently, the prevailing model of SG renewal assumes that primitive stem cells are located within the major excretory ducts supplying a pool of progenitor cells that are distributed in the striated and intercalated ducts [2]. In a hierarchical model of tissue renewal, stem cells are functionally defined as slow-cycling cells when compared to their progeny [16,18]. In rats, 5-bromo-2′-deoxyuridine (BrdU) pulse labeling of the SMG during postnatal growth (11–14 days) followed by 8 weeks of chase have identified a small number (1.2% of parenchymal cells) of label-retaining cells (LRCs) that appear to be distributed sporadically in various compartments. The relatively undifferentiated LRCs that were localized to the intercalated and excretory ducts were thought to be the putative stem cells, although the identity of these cells was not determined [19].

Here, we have employed fluorescent pulse-chase experiments using a tetracycline (tet)-regulated histone H2BGFP system [20,21] to obtain new insights into the stem/progenitor cell dynamics in the SMG, and to identify the potential quiescent stem cells. The fluorescent label in H2BGFP fusion protein is very stable and can be monitored over a long period of time [22]. Upon termination of H2BGFP expression, the label content is reduced by half as the labeled cell divides, until it is undetectable [20]. Therefore, with time, the number of initial H2BGFP-labeled cells will decline with only the nondividing or slowly dividing cells retaining the label [21,23]. To analyze H2BGFP label retention in SG, we generated two strains of bi-transgenic mice expressing high levels of tet-inducible H2BGFP under the control of either the K14 or a ubiquitous promoter. Pulse-chase studies did not identify a slowly dividing SG stem cell population in either mouse model. Quantitative assessments of H2BGFP dilution and the replicative behavior of cells in various ductal compartments were consistent with the presence of active stem/progenitor cell populations in the intercalated and excretory ducts.

Materials and Methods

Animals

Transgenic K14-rtTA (Stock# 008099), Rosa26-rtTA (Stock# 006965), and tetO-H2BGFP (Stock# 005104) were purchased from Jackson Laboratory (Bar Harbor, ME). Bi-transgenic mice were generated by crossing tetO-H2BGFP with either K14-rtTA or Rosa26-rtTA mice. H2BGFP in bi-transgenic embryos was induced by feeding pregnant mice with doxycycline (Dox) (1 mg/g pellet; BioServ, Flemington, NJ) starting 7 days after mating until pups were 2 weeks of age or older. Pulse-labeled mice were placed on a normal diet to suppress H2BGFP expression for the duration of the chase. For consistency in labeling, large cohorts of bi-transgenic mice were generated and pulsed, one mouse from each litter was euthanized at 2 weeks of age; SMG was dissociated to single cells and analyzed by flow cytometry to determine the efficiency and intensity of labeling in each litter. In some animals, gland extirpation was used as a model of injury to induce compensatory proliferation in the SMG. Pulse-labeled mice at 12 weeks of chase were anesthetized, an incision was made to expose the SMG, and one gland was removed to induce compensatory proliferation in the contralateral gland. Mice were injected with BrdU (50 μg/g body weight; Sigma, St. Louis, MO) for 3 consecutive days and the remaining gland was removed 6 h after the last injection, and processed for immunofluorescent analysis. Animals were housed under standard conditions and all animal experiments were performed in accordance with institutional guidelines set forth by the State University of New York.

Histological preparation and immunofluorescent analysis

For histological analysis, the labeled SMGs were dissected, fixed in 4% paraformaldehyde at 4°C for 60 min, soaked in 30% sucrose, and cryopreserved. For detection of green fluorescent protein (GFP) in tissue sections, cryosections were dried, rehydrated, and mounted in Vectashield^® mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlingame, CA). For immunofluorescent analysis, cryosections were immunostained using antibodies against cell-specific markers including cKit (1:100; eBioscience, San Diego, CA), Aqp5 (1:100; EMD Millipore, Billerica, MA), α-SMA (1:100; Abcam, Cambridge, MA), K19 (1:10, TROMA-III, DHSB; University of Iowa), and K14 (1:1,000; Covance, Berkeley, CA) to identify various compartments in the gland. Bound antibodies were detected with Alexa-594-conjugated secondary antibodies (Molecular Probes, Eugene, OR). Slides were mounted with media containing DAPI, examined by a Nikon E800 fluorescent microscope, and images were obtained using NIS-Elements (Nikon Instrument, Inc., Melville, NY). To quantify the GFP-labeling index, the number of GFP⁺ nuclei and DAPI-stained nuclei in each compartment were counted and the percentage of GFP-labeled cells in each compartment was calculated. A total of 400 images were used with between 500 and 1,000 nuclei counted for each compartment.

BrdU labeling and analysis

To determine the BrdU-labeling index in various compartments, adult male mice at 8- or 14-weeks of age (n=3/group) were injected with BrdU (50 μg/g body weight; Sigma) and 6 h later, the SMG was removed and processed as described above. Sections were co-immunostained with anti-BrdU (Accurate Chemical and Scientific Corp., Westbury, NY) and antibodies to cell-specific markers listed above. Sequential staining was done when antibodies to cell-specific markers were made in the same species as anti-BrdU (RatIgG). Both immunofluorescent staining and anatomical features were used to identify different ductal compartments. DAPI was used to counterstain the nuclei and determine the total number of nuclei within each compartment. In addition, BrdU immunofluorescent labeling was compared to that of DAPI to ensure nuclear localization and the proportion of BrdU⁺ nuclei to total number of nuclei in each compartment was determined by analysis of a total of 700 randomized 400×images.

SG cell preparation and fluorescent-activated cell sorting analysis

SMGs were dissected, dissociated by mechanical and enzymatic digestion as described previously with some modifications [24]. Glands were digested using a Hanks' balanced salt solution (HBSS) containing 1% bovine serum albumin (BSA), 0.025% collagenase, 0.04% hylaronidase, 1 U/mL dispase, and 5 mM CaCl₂ for 1 h at 37°C. After washing, cell pellets were resuspended in ice-cold RBC lysis buffer (0.82% NH₄Cl, 5 mM KCl titrated to pH 7.4) and incubated for 10 min on ice. Recovered cells were trypsinized in 0.25% trypsin-EDTA buffer for 1 min. After neutralization, cells were filtered through a 40 μm nylon mesh to obtain single cell suspensions. For fluorescent-activated cell sorting (FACS) analysis, cells were labeled with APC-conjugated antibody to integrin-α6 (1:500; eBioscience) at 4°C for 30 min, washed with HBSS—0.5% BSA, and analyzed on FACSCaliber (BD Bioscience, San Jose, CA). For cell sorting, cells were resuspended in HBSS containing 2% fetal bovine serum. Cell sorting was performed on FACSAria-III with FACSDiva software (BD Bioscience). Sorted cells were plated at 100 cells/cm² in the presence of gamma-irradiated 3T3 fibroblasts using keratinocyte medium described by Wu et al. [25] and grown for 2 weeks. Colonies were visualized and counted after staining with 1% Rhodamine B (Sigma).

Results

Uniform loss of H2BGFP label from K14-expressing ductal cells

To label the relatively undifferentiated ductal cells, transgenic mice expressing reverse tet-sensitive transactivator under the control of K14 promoter (K14-rtTA) [26] were crossed with tetO-H2BGFP mice [20]. Although K14 expression in mature gland is restricted to MECs and the basal layer of excretory ducts [27,28], during embryonic development, K14 is expressed in a multipotent progenitor population [10] (Supplementary Fig. S1A; Supplementary Data are available online at www.liebertpub.com/scd). To determine the efficiency of K14-rtTA driver to target undifferentiated ductal cells, H2BGFP label was induced by Dox in bi-transgenic K14rtTA:tetO-H2BGFP (K14-H2BGFP) mice throughout SMG development to 2 weeks of age, before differentiation of granular ducts [3]. Analysis of GFP distribution by fluorescent microscopy showed Dox-induced expression of H2BGFP in both small (presumptive ducts, Fig. 1B, arrows) and large ducts (Fig. 1B-ED), similar to that observed for endogenous K14 (Fig. 1D and Supplementary Fig. S1B). Immunofluorescent staining with antibodies to cell-specific markers including K14, K19, and α smooth muscle actin (SMA), verified colocalization of H2BGFP with endogenous K14 in MECs and in the undifferentiated ductal cells (K19⁻ cells) in the presumptive intercalated ducts and excretory ducts (Fig. 1D). Longer pulse labeling resulted in a more heterogeneous labeling in ductal cells due to suppression of K14 expression coinciding with differentiation of the granular ducts (data not shown).

FIG. 1.

Uniform loss of H2BGFP-label from keratin (K)14⁺ ductal progenitor cells. (A) Schematic representation of pulse-chase strategy used to identify label-retaining cells (LRCs) in K14-H2BGFP mice. (B) Fluorescent images of the submandibular gland (SMG) at 2 weeks of age showing induction of H2BGFP expression (green) by doxycycline (+Dox). In young mice K14 driven H2BGFP is detected in both large [excretory duct (ED)] and small ducts (arrows). Sections from control (−Dox) mice were stained with 4′,6-diamidino-2-phenylindole (DAPI; blue) to visualize nuclei. (C) Fluorescent-activated cell sorting (FACS) analysis of dissociated SMG cells immediately after pulse (T0) or after 1 week (T1), 6 weeks (T6), and 12 weeks (T12) of chase showing dilution of label over time. The percentage of green fluorescent protein (GFP)-labeled cells in indicated in the histogram. (D, E) Images of SMG sections obtained immediately after pulse or after 6 weeks of chase stained with antibodies to K14 [undifferentiated ducts and myoepithelial cells (MECs)], K19 (differentiated ductal cells), α smooth muscle actin (SMA; MECs), and α6-integrin (pan epithelial cells). Bound antibodies were detected using 594-Alexa-conjugated secondary antibodies (red). DAPI was used to counterstain nuclei. Insets show a second representative image of ducts. In (E) arrowheads denote GFP-labeled cells in the luminal layer of excretory ducts. SD and GD denote striated duct and granular duct, respectively. Exposure time for GFP was set at 0.1 s for images captures at T0 and 1 s for T6. Scale bar=100 μm in (B) and 50 μm in (D, E). Color images available online at www.liebertpub.com/scd

To identify the potential quiescent stem cells that may be established within the K14⁺ ductal progenitor population during development, a cohort of bi-transgenic mice (n=12) were pulsed to 2 weeks after birth and were then placed on a regular diet to suppress H2BGFP expression (Fig. 1A). SMGs were harvested either immediately (T0), or after 1 week (T1), 6 weeks (T6), or 12 weeks (T12) of chase (n=3/group). For each mouse, one gland was processed for histology and the other was dissociated for FACS to monitor loss of H2BGFP label. FACS analysis of SMG at T0 showed a broad fluorescent peak indicating the heterogeneity of the labeled cell population, which includes both ductal cells and MECs (Fig. 1C, D). After suppression of H2BGFP expression, a progressive but nonuniform loss of label was noted with about 20% of labeled cells retaining significant levels of fluorescence after 6 weeks of chase (Fig. 1C). Immunofluorescent analysis of tissue sections revealed the identity of the majority of brightly labeled cells as MECs (Fig. 1E-SMA). The slow loss of H2BGFP in MECs is reflective of the low proliferative activity of MECs after birth (Fig. 4) [28]. Contrary to the MECs, H2BGFP label in the K14⁺ ductal cells was almost uniformly lost and appeared to be shifted to more differentiated cells in the luminal layer of the excretory ducts (Fig. 1E-arrowheads, also Fig. 5). In addition, a small number of GFP-labeled ductal cells were detected in other K19-expressing compartments including granular and striated ducts (Fig. 1E-K19 panel). Overall, the loss of label from K14⁺ ductal cells was inconsistent with the establishment of a quiescent stem cell population by the K14-progenitor cells during pre- and postnatal development.

FIG. 4.

Dynamics of H2BGFP-labeled cell loss in various gland compartments. (A, B) Quantitative assessment of labeled cells in various SMG compartments before (T0) or after 6, 12, or 18 weeks of chase. Fluorescent images obtained from Rosa-H2BGFP and K14-H2BGFP mice were used for analysis of ductal and MECs, respectively. At T0, presumptive IDs are considered as precursors for both ID and GD. Graphs show the GFP-labeling index (ratio of GFP/DAPI) for each compartment and values are expressed as mean±SEM using a minimum of 15 images (400×)/compartment/mouse (n=3). (B) The linear plot of the graph shown in (A) indicating the labeled cell dilution for each compartment in adult mice (8–20 weeks of age). (C) BrdU-labeling index of various gland compartments in 8- or 14-week-old male mice after 6 h of BrdU-pulse. SMG sections were co-immunostained with antibodies to BrdU and to markers specific for each gland compartments. Values are expressed as mean±SEM in each compartment using a minimum of 15 images (400×)/compartment/mouse (n=2/age). The number of total nuclei counted for each compartment ranged from 520 (for ED) to 3,600 (for GD/SD). No BrdU staining observed in the SDs or the luminal layer of excretory duct (ED-K19⁺).

FIG. 5.

The fate of K14⁺ progenitor cells in excretory ducts. (A, B) K14-H2BGFP mice were labeled to 2 weeks of age (T0) and chased for 2 weeks (T2) or 6 weeks (T6). (A) Representative images of the excretory ducts stained with antibodies specific to either K14 or K19 (inset) is shown. Antibodies are shown in red. Inset is included to show the transition from K14⁺ to K19⁺ cells within 2 weeks of chase. At T0, arrowheads note the lack of GFP-labeling in luminal cells. In T2, arrows points to the GFP-labeled luminal cells. DAPI was used to stain nuclei blue. Scale bar=50 μm. (B) Quantitative assessment of GFP-labeled cells in the K14⁺ basal layer (ED-K14) and K19⁺ luminal layer (ED-K19) at indicated periods of chase showing the transition of the GFP label from K14- to K19-expressing cells. Values are expressed as mean±SEM (n=4 glands). Color images available online at www.liebertpub.com/scd

Ubiquitous expression of H2BGFP in the SMG reveals the location of label-retaining ductal cells

Multiple progenitor populations have been reported to exist in the developing gland [11]. To determine the dynamics of all potential stem/progenitor cells in the gland, we used the ubiquitously expressed Rosa26-rtTA to drive tetO-H2BGFP expression in transgenic mice. The Rosa26 locus is active in most cells of the mouse [29] and is expected to label all cells in the SMG. In addition, previous pulse-chase studies in Rosa26rtTA:tetO-H2BGFP (Rosa-H2BGFP) mice have identified GFP-LRCs in tissues where slow-cycling stem cells are known to exist, including the hair follicle and intestinal crypt [30]. Bi-transgenic Rosa-H2BGFP mice were subjected to pulse-chase experiments as described above (Fig. 1A). FACS analysis of labeled SMG at T0 showed H2BGFP labeling in more than 70% of cells that included both parenchymal (Integrin-α6⁺) and nonparenchymal (Integrin-α6⁻) cells (Fig. 2A–C). Immunofluorescent analysis showed uniform and intense labeling of all ductal cells regardless of their location or differentiation state (Fig. 2D; K14 and K19 staining). Acini and MECs were also labeled, however, the label intensity was too low in acinar cells (Fig. 2D-Aqp5, arrow), and was heterogeneous in MECs (Fig. 2D-SMA), therefore, they were excluded from our analysis.

FIG. 2.

Identification of LRCs after uniform labeling of all ductal cells. (A–C) Bi-transgenic Rosa-rtTA:tetO-H2BGFP (Rosa-H2BGFP) mice were maintained on doxycycline-containing diet to 2 weeks of age when SMG was analyzed by either fluorescent microscopy (A) or FACS (B, C). In (B), dissociate SMG cells were stained for integrin-α6 to verify efficient labeling of parenchymal cells. (C) A representative histogram showing labeling efficiency of more than 70% (n=3). (D) Immunofluorescent images of SMG from Rosa-H2BGFP at 2 weeks of age stained with the antibodies indicated in the figure (red) showing uniform and high intensity labeling (green) of ductal cells (K14 and K19 panels) when compared with acinar (Aqp5 staining) or MECs (SMA staining). Arrow points to dim GFP in acinar cells (AC). Arrowheads in the same image denote MECs, which appear as brightly labeled cells surrounding acini. Images are exposure matched. (E, G) Rosa-H2BGFP mice labeled to 2 weeks of age were chased for 18 weeks (T18) and sections prepared from SMG at 6 weeks intervals were stained with ductal-specific markers including K14, K19, and cKit (red staining) to map the location of GFP-LRCs to the EDs (E), SDs (F), and intercalated ducts (IDs; G), respectively. DAPI was used to counterstain nuclei (blue). By 12 weeks of chase high fluorescent label is only detected in the SD (F) and to a lower extent in the luminal layer of ED (arrows in E). Boxed areas in (E) are higher magnifications of ED to show the location of GFP-labeled cells. Arrows in (G) point to dim fluorescent label detected in some but not all ID. Exposure for GFP was set at 0.025 s for images taken at T0 and at 0.25 for T6 and 1 s for all other time points. Scale bar=100 μm in (E) and 50 μm in (D, F, and G). Color images available online at www.liebertpub.com/scd

Immunofluorescent staining for ductal-specific markers including K14 (the basal cells in excretory ducts), K19 (granular ducts, striated ducts, and the luminal cells in excretory ducts), and cKit (intercalated ducts), in combination with the anatomical features of ducts were used to map the location of GFP-labeled ductal cells during the chase period (Fig. 2E–G). The majority of ductal cells retained detectable levels of GFP after 6 weeks of chase, although the label intensity was significantly reduced (exposure times for images captured for T0 and T6 were 0.025 and 0.25 s, respectively). By 12 weeks of chase (T12) however, labeled cells with bright GFP were predominantly localized to the striated ducts (Fig. 2F, G) and to the luminal cells in the excretory ducts (Fig. 2E-arrows). Neither the intercalated ducts nor the basal layer of excretory ducts contained brightly labeled cells (Fig. 2E–G), although a number of dimly labeled cells were detected in some intercalated ducts (Fig. 2G-arrows).

The lack of label retention in the excretory and intercalated ducts, where SG stem/progenitor cells are thought to reside, could be related to the activation of stem cells during the rapid growth and expansion of the SMG in the growing mice. To test this possibility, Rosa-H2BGFP mice were pulsed to adulthood (8 weeks of age) and H2BGFP dilution was monitored during 18 weeks of chase (Supplementary Fig. S2). Even during homeostasis and despite the slow turnover of SG in adult mice, GFP-labeled parenchymal cells were again confined to the striated ducts and the luminal layer of excretory ducts (Supplementary Fig. S2). These results indicated that the dynamics of stem/progenitor cells in the SMG remain consistent between the young and adult mice. Localization of GFP-LRCs to differentiated compartments, however, suggested that LRCs may not include stem cells.

The proliferative potential of label-retaining ductal cells

Quiescent stem cells in various tissues are mobilized to proliferate in response to injury [31]. Unilateral gland extirpation is known to induce compensatory proliferation in all compartments of the contralateral gland [19,32]. To determine whether the GFP-LRCs we have identified in the SMG proliferate in response to injury, Rosa-H2BGFP mice were pulsed and one gland was surgically removed after 12 weeks of chase (Fig. 3A). BrdU incorporation in the remaining gland was analyzed by immunofluorescent staining at 3 days postextirpation when the total mitotic index reaches the maximum [19]. As shown in Figure 3B, although BrdU⁺ cells were readily detected in various compartments including acini and intercalated ducts (arrows), analysis of more than 500 GFP-labeled nuclei failed to show colocalization of GFP and BrdU (red fluorescence). Moreover, FACS analysis of dissociated SMG cells isolated from the extirpated or the remaining contralateral gland 5 days later showed comparable number and intensity of GFP-labeled cells, confirming the lack of cell division in GFP-LRCs (Supplementary Fig. S3). These data indicated that H2BGFP-labeled cells do not contribute to compensatory proliferation in response to unilateral extirpation. More severe injury induced by duct ligation/de-ligation, however, resulted in the loss of all labeled parenchymal cells and, therefore could not be used to assess proliferative potential of GFP-LRCs (data not shown).

FIG. 3.

GFP-LRCs in SMG do not represent quiescent stem cells. (A) Experimental schematics for induction of compensatory proliferation in SMG of Rosa-H2BGFP mice by extirpation. (B) Representative images of the contralateral gland (n=3) stained for 5-bromo-2′-deoxyuridine (BrdU) demonstrating the lack of BrdU incorporation (red) in LRCs (green). Arrow points to GFP⁻ (inset) proliferating cells in an ID. Scale bar=50 μm. (C–F) Proliferative potential of LRCs in culture was assessed using a colony formation assay. Rosa-H2BGFP mice were labeled and chased for 12 weeks, SMGs were removed, dissociated to single cells, and stained with PE-integrin α6 and two populations of α6⁺GFP⁺ (LRCs) and α6⁺GFP⁻ cells were sorted by FACS (C). (D) Images of sorted cell populations spotted on a slide and stained for K19 (red) and DAPI (blue) to show purity of sorted cell populations. (E, F) Representative images of colonies formed from 1,000 nonsorted or sorted cell populations plated at 100 cells/cm² on fibroblast feeders and grown for 14 days. Colonies (pink) were visualized by staining with 1% Rhodamine B. Graph shows the number of colonies formed per 1,000 cells. Values are expressed as mean±SEM (n=9 from three mice). Color images available online at www.liebertpub.com/scd

Clonogenicity in culture is a well-established method used to assess the proliferative potential of quiescent stem cells [33]. In the mouse SMG, the colony-forming ability and salisphere formation by dissociated cells are attributed to stem/progenitor cells [14,34]. Consistent with previous studies, our attempts to establish salispheres from single cell suspensions were not successful [14] (data not shown). Therefore, we used colony formation to assess the proliferative potential of GFP-LRCs in culture. SMGs were dissected from Rosa-H2BGFP mice after 12 weeks of chase, dissociated, stained with an antibody to integrin-α6, and FACS was used to isolate α6⁺GFP⁺ and α6⁺GFP⁻ cell populations (Fig. 3C). Cytospin and immunofluorescent analysis of the sorted populations verified the purity of the GFP-labeled population and the expression of K19 in the majority of α6⁺GFP⁺ cells (Fig. 3D). When sorted cells were cultured at clonal density and grown for 2 weeks, α6⁺GFP⁺ failed to form colonies while a substantial number of colonies formed from the sorted α6⁺GFP⁻ and the nonsorted cells (Fig. 3E, F). Overall, functional analysis of GFP-LRCs in SMG using both in vivo and ex vivo assays indicated that GFP-LRCs do not include cells with a high proliferative potential and therefore may not represent quiescent stem cells.

Proliferation dynamics of ductal cell subpopulations in the SMG

The pattern of H2BGFP loss in the SMG was consistent with the presence of actively cycling stem/progenitor and a diverse cellular behavior in the adult gland. The efficient labeling of ductal compartments in Rosa-H2BGFP (Fig. 2) and that of MECs in K14-H2BGFP (Fig. 1), and the ability to monitor GFP for 18 weeks provided an opportunity to obtain a comprehensive view of the dynamics of different subpopulations in the SMG through monitoring of GFP loss in various compartments. Using immunofluorescent images obtained in the studies described in Figures 1 and 2, the number of GFP-labeled cells within each compartment was quantified. GFP-labeling indices (% GFP⁺ cells irrespective of GFP levels) for various compartments were determined at T0, T6, T12, and T18 and these data were used to plot a “cell dilution” curve from T6 to T18, during homeostasis (Fig. 4A, B). Analysis of the slopes of these curves identified the basal layer of the excretory duct (slope_T6–T18=−4.5) and, to a lesser extent, the intercalated duct (slope_T6–T18=−3.7) as the most dynamic compartments in the gland, consistent with the presence of progenitor populations in these two compartments [2,6,35,36]. The rate of GFP loss was substantially lower but remarkably similar for the granular ducts, the luminal layer of the excretory ducts, and the MECs (slope_T6–T18=−2). Interestingly, there was no significant decline in the number of labeled cells in the striated ducts (slope_T6–T18=−0.53), indicating the absence or very low level of cellular turnover that is often observed in “static” tissues [37].

The loss of GFP-labeled cells could not only be attributed to cell division, but also to cell migration, apoptosis, and H2BGFP degradation. Unlike DNA label, H2BGFP protein degrades and is eventually lost even in nondividing cells [23,30]. To verify a direct correlation between the proliferative behavior of ductal cells and the loss of H2BGFP, mitotic indices of various compartments in the SMG of adult mice were determined. Mice at the age of 8 or 14 weeks (corresponding to T6 and T12) were pulsed with BrdU for 6 h and BrdU-labeled cells were mapped to various compartments using double immunofluorescent staining (Supplementary Fig. S3). As shown in Figure 4C, the BrdU-labeling indices were consistent with the cell dilution curves, identifying the basal cells in the excretory ducts as the most mitotically active cell population in the gland. BrdU-labeled cells were identified in all other compartments, except the striated ducts and the luminal layer of the excretory ducts. This indicated that not all differentiating cells in the SG are capable of proliferation, as reported previously [4,5]. The significantly lower turnover rate of the striated duct when compared with the luminal layer of excretory duct (Fig. 4B) suggested that these two postmitotic compartments are maintained by distinct mechanisms. Overall, the proliferative behavior of ductal cells suggested the presence of mitotically active progenitor populations in the intercalated and excretory ducts, separated by the striated duct that is practically a “static” compartment.

Short-term H2BGFP tracing reveals the excretory duct as a self-renewing compartment

The dynamics of GFP loss and the restriction of proliferation to the basal layer of the excretory duct are suggestive of the presence of an active stem/progenitor population in this compartment. To assess the fate of these cells, K14-H2BGFP mice were labeled and distribution of GFP⁺ cells in the excretory duct was analyzed after 2 and 6 weeks of chase (Fig. 5A). Immunofluorescent analysis of the excretory ducts at completion of pulse (T0) showed restriction of bright green fluorescence to the nuclei of K14⁺ cells (red), which sometimes extended to the suprabasal layer (Fig. 5A-arrowheads). After 2 weeks of chase (T2), however, a significant number of GFP⁺ nuclei were detected in the K14⁻/K19⁺ luminal layer of excretory ducts (Fig. 5A-T2). In subsequent chase periods (T6), GFP-labeled cells significantly diminished in number and were localized almost exclusively to the luminal layer (Fig. 5B-T6). Quantitative analysis of labeled cells in the basal and luminal layers showed a clear shift of H2BGFP from the K14⁺ to the K19⁺ cells (Fig. 5B). This shift in distribution of GFP indicated a precursor/progeny relationship between K14⁺ basal and K19⁺ luminal cells, and a stratified program of differentiation in the excretory duct, similar to that described for the sweat ducts in skin [38]. Therefore, the excretory duct could be considered as a self-renewing tissue in which the continuous loss of differentiated luminal cells is compensated by proliferation of a single active progenitor cell population in the basal layer.

Discussion

The prevailing model for SG assumes the existence of stem and progenitor populations with stem cells localized within the major excretory ducts, supplying the progenitor cells distally within the smaller ducts. In turn, the latter replenish the more differentiated cell types in secretory units (reviewed in Pringle et al. [2]). Our experiments were designed to utilize histone H2BGFP pulse-chase experiments to determine the dynamics of stem/progenitor cells in the mouse SMG and to characterize the putative quiescent stem cell population that may exist in this gland. Despite efficient H2BGFP labeling of ductal cells in two transgenic mouse models, we found no evidence for a slowly dividing stem cell population in ducts or for a hierarchical model of cell renewal in the SMG. Label retention in the gland was restricted to nondividing ductal compartments including the striated ducts and the luminal cells in the excretory ducts. The dynamics of H2BGFP loss from the excretory and intercalated ducts, where the stem/progenitor cells are thought to reside [2,39], is more consistent with the presence of active stem or progenitor populations in these compartments.

Although previous BrdU pulse-chase studies in the SMG have identified LRCs that were distributed in various compartments including the intercalated and excretory ducts [19,40], these LRCs may not have represented slowly dividing stem cells. Because BrdU pulse-chase depends upon active cell proliferation at the labeling stage, postnatal administration of BrdU results in labeling of only a fraction of the cells, likely rapidly proliferating cells. Under these conditions, BrdU label retention after a relatively short period of chase (7–10 weeks) could be explained by either the diverse cell kinetics in the gland (Fig. 4C) [5], or by asymmetric cell division in actively dividing progenitors [19]. Our experiments were designed to identify quiescent stem cells, if they existed. A major advantage of the H2BGFP system is the ability to label virtually all cells, irrespective of their proliferative status [30]. By using two transgenic mouse models and continuous pulse labeling during gland development, any potential slowly dividing stem cell population that may be established during development would have been detected [41]. Additionally, by extending the pulse labeling in Rosa-H2BGFP mice to adulthood (8 weeks of age), quiescent stem cells that may have been established after gland maturation would have been detected. It is worth noting that GFP loss in intercalated ducts was not uniform and a number of ductal cells presenting a dim fluorescent signal persisted even after 12–18 weeks of chase (Fig. 2G). These dimly labeled cells, however, were not present in every intercalated duct and could not be localized to a specific niche, as has been described for quiescent stem cells in other tissues [21]. The nonuniform loss of label is more consistent with the regional differences in replication activity of the ductal cells in intercalated ducts and their contribution to adjoining compartments [35,36,39]. Taken together, our data ruled out the existence of a quiescent stem cell population in the mouse SMG ducts. A recent study using K5 promoter and a similar H2BGFP pulse-chase strategy in minor SGs (palatal) has localized a population of quiescent stem cells to the basal layer of the lower excretory ducts and MECs in the acini [42]. Although the contribution of each of these cell populations to normal gland renewal and regeneration needs to be determined, the apparent differences in ductal cell dynamics observed between the palatal and SMGs may be explained by their distinct embryonic origin and significant structural differences [43,44].

Our results provide a broad view of population dynamics in the mouse SMG. Our data expand on the earlier cell kinetic studies identifying the intercalated and excretory ducts as the most dynamic compartments in the SG [5,36], and they reveal a stratified program of differentiation in the excretory ducts. Although the dynamics of intercalated ducts and their contribution to cell renewal in the granular duct and acini have been extensively studied [6,35], little is known about the cell replacement in other compartments. Previously, the high mitotic index of the ductal cells in excretory ducts suggested that these cells may contribute to the replacement of striated duct cells [36]. Our analysis, however, shows that at least in the adult mouse SMG, striated ducts are essentially static structures. We have defined striated ducts as intralobular ducts composed of a single layer of columnar epithelia expressing high levels of K19. Retention of almost all labeled cells in the striated duct during homeostasis is consistent with nondetectable proliferative and apoptotic activities for this compartment [6] (Fig. 4C and Supplementary Fig. 4). Given the presence of a stratified program of differentiation in the excretory ducts revealed in this study, the high proliferative activity in the basal layer (K14⁺) is likely required for the continuous replenishment of terminally differentiated postmitotic luminal (K19⁺) cells. This is consistent with the higher rate of GFP loss in the luminal layer of excretory duct when compared with that in the striated ducts (Fig. 4B).

It has been proposed that the excretory ducts contain a rare population of cKit-expressing stem cells that may give rise to other progenitor cells in the gland [2]. Progenitors expressing cKit in the adult mouse SMG are expanded in salispheres and show a robust regenerative potential in a mouse model of radiation-induced injury [13–15]. Although in these studies cKit expression was found to be restricted to the excretory ducts, subsequent immunofluorescent analyses of human and mouse SGs have localized cKit to both the intercalated and excretory ducts [10] (Fig. 2G). The broad tissue expression of cKit suggests that only a subset of cKit⁺ cells may function as stem cells. In the embryonic mouse SMG, Kit⁺K14⁺ progenitors may give rise to other progenitor populations including K5⁺ ductal progenitors [11]. Interestingly, there is a partial overlap between cKit⁺ and K14⁺ cells in the excretory ducts of adult SMG (data not shown); however, whether these cells are multipotent and can give rise to other SG progenitor cells needs to be investigated. It is worth noting that although K14 marks a multipotent progenitor population during gland morphogenesis (Supplementary Fig. S1) [11], after development, K14⁺ progenitors appear to have more restricted lineage (Fig. 5). This is consistent with recent studies in mammary gland and sweat gland in which multipotent embryonic progenitor cells give rise to unipotent cells after development [38,45]. Given the architecture of the SMG and the presence of a nondividing compartment separating the intercalated and excretory ducts, a lineal relationship between stem/progenitor cells in the excretory duct with those in the intercalated ducts seems unlikely. However, generation of a permanent genetic marker by using the inducible Cre-LoxP system is the most effective method to determine cell lineage of these progenitor cells.

Conclusions

H2BGFP pulse-chase experiments in two mouse models ruled out the coexistence of quiescent and active stem/progenitor populations in the SG. The pattern of H2BGFP loss was consistent with the presence of actively cycling stem/progenitor populations in the excretory and intercalated ducts functioning independently under steady-state conditions. It would be important to further characterize these two progenitor populations and define their roles and potentials during SG regeneration.

Acknowledgments

The authors are grateful to Ninche Alston, Sihanna Rugova, and Troy Lam for technical assistance. This work was supported by a grant to S.G. from NIH (R21DE022959).

Author Disclosure Statement

No competing financial interests exist.