Localization of phosphatidylinositol 4‐phosphate 5‐kinase (PIP5K) α, β, γ in the three major salivary glands in situ of mice and their response to β‐adrenoceptor stimulation

Abstract

Phosphatidylinositol 4‐phosphate 5‐kinase (PIP5K), which is composed of three isozymes (α, β and γ), catalyzes the production of phosphatidylinositol bisphosphate (PIP2). This phospholipid functions in membrane trafficking, as an anchor for actin cytoskeletons and as a regulator of intramembranous channels/transporters. It is also a precursor of such second messengers as diacylglycerol, inositol triphosphate and phosphatidylinositol (3,4,5)‐triphosphate. In the present study, the expression and localization of endogenous PIP5Ks were examined in the three major salivary glands of young adult mice in situ. In western blotting of normal control glands, immunoreactive bands for individual PIP5Ks were detectable, with the highest density in the parotid gland and the weakest density in the submandibular gland. In immuno‐light microscopy under non‐stimulated condition, weak immunoreactivity for PIP5Kα was confined to the apical plasmalemma in parotid, but not sublingual or submandibular, acinar cells. Immunoreactivity for PIP5Kβ was weak to moderate and confined to ductal cells but not acinar cells, whereas that for PIP5Kγ was selectively and intensely detected in myoepithelial cells but not acinar cells, and it was weak in ductal cells in the three glands. In western blot of the parotid gland stimulated by isoproterenol, a β‐adrenoceptor agonist, no changes were seen in the intensity of immunoreactive bands for any of the PIP5Ks. In contrast, in immuno‐light microscopy, the apical immunoreactivity for PIP5Kα in parotid acinar cells was transiently and distinctly increased after the stimulation. The increased immunoreactivity was ultrastructurally localized on most apical microvilli and along contiguous plasma membrane, where membranous invaginations of various shapes and small vesicles were frequently found. It was thus suggested that PIP5Kα is involved in post‐exocytotic membrane dynamics via microvillous membranes. The present finding further suggests that each of the three isoforms of PIP5K functions through its product PIP2 discretely in different cells of the glands to regulate saliva secretion.

Introduction

Phosphatidylinositol 4‐phosphate 5‐kinase (hereafter abbreviated as PIP5K) catalyzes the production of phosphatidylinositol (4,5)‐bisphosphate (hereafter simply abbreviated as PIP2). There are three isozymes, α, β and γ, of PIP5K following the mouse nomenclature (Kanaho et al. 2007). Its product PIP2 is a substrate for phospholipase C (PLC) which produces inositol (1,4,5)‐trisphosphate (IP₃) and diacylglycerol (DAG), and for phosphatidylinositol 3‐kinase (PI3K), which produces phosphatidylinositol (3,4,5)‐trisphosphate (PIP₃) (Oude Weernink et al. 2004). In addition to its role as the precursor for these 3‐second messengers in the phosphoinositide (PI) signal cascade, PIP2 itself also plays multiple roles influencing a variety of cellular processes: (1) as a ‘lipid anchor’ that attaches actin cytoskeleton to the plasma membrane, (2) it is involved in processes that require the membrane trafficking via endo‐ and exocytosis, and (3) it stabilizes or activates many intramembranous bioactive proteins, such as ion channels and transporters (Di Paolo & De Camilli, 2006).

To execute these various roles in the membrane dynamics, PIP2 is generally supposed not to be randomly distributed in cells, but to be organized in confined regions of cells, mainly the plasma membrane. The intracellular localization of PIP5K isoforms likely contributes to the organization of diverse structural and functional PIP2 pools in cells. Studies of cultured cells transfected with respective cDNAs for PIP5K isoforms have shown that exogenous PIP5Kα is localized to the plasma membrane and the Golgi complex as well as the intranuclear structures known as nuclear speckles (Payrastre et al. 1992; Mellman et al. 2008; Van den Bout & Divecha, 2009). In contrast, PIP5Kβ has been localized to the plasma membrane and vesicles in the perinuclear region of the cell (Doughman et al. 2003), and PIP5Kγ has been found on focal adhesions (Di Paolo et al. 2002; Ling et al. 2002). However, to understand more clearly the diverse functional significance of individual PIP5Ks and PIP2, it is important to obtain information on the localization of endogenous PIP5Ks in cells in situ.

In the present study, we have attempted to examine the localization of immunoreactivity for endogenous PIP5K isoforms in the three major salivary gland cells of mice in situ. The salivary glands were selected because the glandular cells have a distinct cell polarity, the apical plasma membranes performing regulatory exocytosis and endocytosis, and the basolateral and apical membranes being equipped differentially with various channels, transporters and receptors. All these properties and functions are likely to be intimately related to the roles which PIP2 plays, as summarized above. No studies have so far been reported on the immunohistochemical localization of PIP5Ks in any salivary gland cells in situ. To stimulate the glands, isoproterenol, a β‐adrenoceptor agonist, was the first selection because stimulation with β‐adrenoceptor agonists causes a rapid elevation in intracellular cAMP and enhances the secretion of salivary amylase via exocytosis, a major phenomenon controlled by PIP2, as described above.

Materials and methods

Plasmid construction

For bacterial expression vectors, the C‐terminal 79 amino acids of mouse PIP5Kα (GenBank accession number D86176) and 67 amino acids of mouse PIP5Kβ (accession number D86177), which correspond to human PIP5Kβ and PIP5Kα, respectively, were amplified by PCR using a mouse brain cDNA library. The following primers were used, in which the 5′ ends of sense and antisense primers were supplemented with SalI and NotI restriction sites (underlined), respectively: sense, 5′‐GTCGACCGACCTGGTACCCAGCACTCCATCAT‐3′; anti‐sense, 5′‐ GCGGCCGCTTATAAATAGACGTCCAGCACAGAG‐3′ for PIP5Kα, and sense, GTCGACGACCACCAAGGCGGAAGTGGAGCCAG;anti‐sense, 5′‐GCGGCCGCTCAGTGGGTGAACTCTGACTCTGC‐3′ for PIP5Kβ. After the PCR fragments were subcloned into the pGEM®‐T Easy vectors (Promega; Madison, WI, USA), the inserts digested with SalI and NotI restriction enzymes were subcloned into pGEX‐4T2 (GE Healthcare Bio‐Sciences, Piscataway, NJ, USA) and modified pMAL‐c2 (New England Biolabs; Beverly, MA, USA).

For mammalian expression vectors for PIP5Ks, the coding regions of mouse PIP5Kα, PIP5Kβ, and PIP5Kγ_v2 (also known as γ661 or γa) (accession number AB006916) were obtained by PCR using a brain cDNA library. The following primers were used, in which the 5′ ends of sense primers were supplemented with a Sal I restriction site (underlined), respectively: sense, 5′‐GTCGACCATGTCGTCAACTGCTGAAAATGGAG‐3′; anti‐sense, 5′‐TTATAAATAGACGTCCAGCACAGAG‐3′ for PIP5Kα; sense, 5′‐GTCGACCATGGCGTCCGCCTCCTCAGGGCCAG3′; anti‐sense, 5′‐TCAGTGGGTGAACTCTGACTCTGC‐3′ for PIP5Kβ, and sense, 5′‐GTCGACCATGGAGCTAGAGGTGCCGGACGAGG‐3′; anti‐sense, 5′‐TTATGTGTCGCTCTCGCCGTCGGAG‐3′ for PIP5Kγ.

The PCR fragments were subcloned into pGEM®‐T Easy, and the inserts digested with SalI were subcloned into the XhoI site of pEGFP‐C2 (Clontech Laboratories; Palo Alto, CA, USA).

Antibody production

The C‐terminal 79 amino acids of mouse PIP5Kα and 67‐amino acids of mouse PIP5Kβ were bacterially expressed as fusion proteins of glutathione‐S‐transferase (GST) or maltose‐binding protein (MBP), and purified using glutathione‐Sepharose 4B (GE Healthcare Bio‐Sciences) or amylose‐resin (New England Biolabs, UK), respectively. The GST fusion proteins were emulsified with Freund's adjuvant and subcutaneously injected into guinea pigs five times at 2‐week intervals. The antibodies were subsequently affinity‐purified from the sera using the respective MBP fusion proteins. The generation and specificity of antibody for PIP5Kγ have been reported by one of the present authors (H.S.; Hara et al. 2013).

Animals and tissues preparation

Male ICR mice were purchased from the National Laboratory Animal Center (NLAC), Bangkok, Thailand. Thirty male mice at postnatal age 8 weeks were divided into two groups: one for analysis under normal conditions (six mice), the other for time‐course analyses (six mice each at stages of 15, 30, 60, 120 min post‐injection) under extrinsic stimulation by single‐shot intraperitoneal injection of isoproterenol (IPR, 15 mg kg^–1 bodyweight in physiological saline; Sigma Aldrich, Welwyn Garden City, UK‎). This amount of IPR has been shown to be appropriate for experimental stimulation (Tachow et al. 2017; Thoungseabyoun et al. 2017). The three major salivary glands on the left side of each mouse were used individually for western blotting analysis and those on the right side for immuno‐histochemical analysis. All mice were given free access to foods and water until the morning of the experimental day. All procedures were conducted in accordance with Guidelines for the Care and Use of Laboratory Animals at Khon Kaen University. This study was approved by the Animal Ethics Committee of the Khon Kaen University, based on the Ethics of Animal Experimentation of the National Research Council of Thailand (Reference No. ACUC KKU 4/2559).

Western blotting analysis

Mouse brain and HeLa cells transfected with pEGFP‐C2 encoding individual PIP5Ks were homogenized in a lysis buffer composed of 20 mm Tris‐HCl (pH 8.5), 20 mm KCl, 10 mm EDTA (pH 8.0), 250 mm sucrose and 1× complete protease inhibitor cocktail (Pierce, Thermo Scientific; Waltham, MA, USA). The lysates were then electrophoretically separated on 7.5% sodium dodecyl sulfate (SDS)‐polyacrylamide gels and transferred onto polyvinyl difluoride membrane (PVDF‐PLUS, Micron Separations Inc., Westborough, MA, USA). The blots were then subjected to western blotting with individual antibodies for PIP5K isoforms and EGFP to confirm the specificity of the individual PIP5K antibodies.

Mice under normal and stimulated conditions at the four post‐injection time‐stages were sacrificed using deep anesthesia with thiopental sodium at 60 mg kg^–1 bodyweight (Unique Pharmaceutical Laboratories; Ankleshwar, India); subsequently, each of the three major glands and brains were individually extirpated and immediately put into liquid nitrogen. Each of the frozen specimens were then homogenized in the same lysis buffer as used for the brain and HeLa cells described above, and further processed as follows: after centrifugation at 3000 g for 10 min, the supernatants were collected and the protein concentration measured. Total protein 40 μg from each lysate was individually boiled for 5 min in 2× SDS sample buffer and subjected to SDS/12.5% PAGE electrophoresis and transferred to a polyvinylidene difluoride (PVDF) membrane (Merck KGaA; Darmstadt, Germany). After blocking the nonspecific binding sites with 5% skim milk (w/v)/Tris‐buffered saline (TBS)/0.3% Tween‐20, the membranes were incubated overnight at 4 °C with each of the three PIP5K isozyme antibodies, then with guinea pig IgG (0.1 μg mL^–1) for the α and β isozymes and with rabbit IgG (0.1 μg mL^–1) for the γ isozyme, in 5% skim milk (w/v)/TBS/0.1% Tween‐20. The membranes were subsequently treated with peroxidase‐conjugated (HRP) anti‐guinea pig IgG for the α and β, and HRP anti‐rabbit IgG for the γ (both from Vector Laboratories; Burlingame, CA, USA) diluted 1 : 2500 for 1 h at room temperature. The immunoreactive proteins were then visualized using the ECL prime western blotting substrate (GE Healthcare; Buckinghamshire, UK). Anti‐goat β actin (Santa Cruz Biotechnology; Dallas, TX, USA) was used as a control. The intensities of all immunoreactive bands were quantified using NIH image j, and the relative values of PIP5Kα, β and γ were normalized against β‐actin intensities.

Immunohistochemistry

The same individual mice, after removal of the left salivary glands, were immediately perfused with 10 mL 4% paraformaldehyde/phosphate buffer (PB). Each of the right major salivary glands was then removed, further postfixed with the same fixative overnight, and then were dipped into 30% sucrose/phosphate‐buffered saline (PBS) for cryo‐protection. Cryo‐sections of 20 μm thickness were made and incubated with 0.3%H₂O₂/methanol for 10 min to inhibit intrinsic peroxidase activity, then 10% normal goat serum/PBS for 30 min to prevent nonspecific antibody binding. Sections were incubated at room temperature overnight with the PIP5Kα, β and γ antibodies (1 μg mL^–1). The sections were incubated for 1 h at room temperature with biotinylated anti‐guinea pig IgG (Vector Laboratories) for the α and β forms, and biotinylated anti‐rabbit IgG secondary antibody (Abcam, Cambridge, MA, USA) for the γ form. For visualization of the antigen–antibody reaction sites with diaminobenzidine (DAB) reaction as the marker, the sections were then treated with ABC kit (Vector Laboratories). When qualitative comparison of the immunoreactivity was necessary between specimens under normal conditions and those of the post‐injection time course for isoproterenol, sections of normal specimens and all stimulated specimens were mounted on the same glass slides and processed for the immunoreaction under almost identical reaction conditions.

In immuno‐DAB electron microscopy, some of the sections were postfixed with 1% OsO₄/PB and embedded in Epon after en bloc staining with 1% uranyl acetate. Ultrathin sections were observed under a JEM1010 transmission electron microscopy (Jeol; Tokyo, Japan).

To provide controls for the immunohistochemistry work, the individual antibodies were pre‐absorbed with corresponding synthetic antigens (100 μg mL^–1), and sections were incubated with the absorption solution for 1 h at room temperature and subsequently treated in the same way as the regular immunoreaction described above.

Results

Antibody production specificity

As for the specificity of the newly synthesized antibodies, a single band of approximately 65 kDa was clearly detected for PIP5Kα and β in the brain lysate in western blotting analysis. In addition, each of the antibodies detected only its corresponding PIPK isozyme conjugated with EGFP (Fig. 1). This excludes the possibility of cross‐reactivity of each antibody with the other two PIP5K isoforms. The generation and specificity of anti‐PIP5Kγ antibody have been reported previously by one of the authors (H.S.; Hara et al. 2013).

Figure 1.

Western blots of lysates of brain and HeLa cells expressing the respective EGFP‐PIP5K isoforms, individual anti‐PIP5Kα and PIP5Kβ antibodies detected a single band (about 65 kDa) without any cross‐reactivity to other isoforms. The positions and sizes (kDa) of molecular weight markers are indicated on the lefthand side.

Protein expression of PIP5Ks in three major salivary glands

Western blots of the three major glands under normal conditions revealed single bands of the same sizes (about 65 kDa) as observed in the brain for PIP5Kβ and γ in the three glands. Band density was intense in the parotid gland, weaker in the sublingual gland and much weaker in the submandibular gland. In contrast, although a single band of the same size as observed in the brain for PIP5Kα was detected in the parotid and sublingual glands, with a higher density in the former, no such bands were discerned in the submandibular gland (Fig. 2).

Figure 2.

Western blots to demonstrate expression of PIP5Kα, β and γ in lysates of parotid (PG), sublingual (SLG) and submandibular (SMG) glands as well as brains of mice. The expression of β‐actin in individual specimens was used as control, as shown at the bottom of the figure. Note lack of the PIP5Kα‐band in SMG in contrast to an intensely positive band in PG and a weaker band in SLG. The density of the bands for the β and γ isozymes are greatest in PG, moderate in SLG and weakest in SMG.

Principal locations of individual PIP5Ks in three major salivary glands

In immuno‐light microscopy of normal parotid glands, weak immunoreactivity for PIP5Kα was apparent as forms of thin lines along the apical plasmalemma in many, but not all, acini (Fig. 3a). In addition to acini, weak to moderate immunoreactivity was seen diffusely throughout the cytoplasm of all ductal cells in the intercalated, striated and excretory ducts. In contrast, no significant immunoreactivity was seen in acini of the sublingual and submandibular glands or in any of the submandibular ducts, but weak to moderate immunoreactivity was seen diffusely within ductal cells throughout the sublingual ducts (Fig. 3b,c).

Figure 3.

(a‐e) Immuno‐light micrographs of PG (a), SLG (b) and SMG (c) for PIP5Kα under normal conditions, and of PG at 30 min post‐injection of isoproterenol (d). Note the remarkable increase in the density of PIP5Kα‐immunoreactive apical plasmalemma (arrows in d) of isoproterenol‐stimulated acinar cells (A) as compared with apical plasmalemma (arrows in a) of acinar (A) cells in normal control PG. Also note no differences in the density of diffuse immunoreactivity in PG striated ductal (SD) (a,d) and intercalated ductal (ICD) cells (a) between the two conditions. In contrast to normal PG acinar cells, lack of significant PIP5Kα‐immunoreactivity is evident in acini of both SLG and SMG (A in b,c) under normal conditions, with weak and diffuse immunoreactivity in SLG ductal cells (b). No significant immunoreactivity was discerned in any portions of stimulated PG in the antigen‐preabsorption control (e). GCT, submandibular granular convoluted tubule. Scale bars: 20 μm.

The immunoreactivity for PIP5Kβ was moderate and diffuse in the cytoplasm of intercalated and striated ductal cells of the three major salivary glands, weak in granular convoluted tubular cells of the submandibular gland, and faint to negligible in acini and intraglandular vessel walls in the three glands (Fig. 4a‐c). Immunoreactivity for PIP5Kγ was weak and diffuse in the cytoplasm of intercalated and striated ductal cells of the three glands and granular convoluted tubular cells of the submandibular gland (Fig. 5a‐c). In contrast, moderate to intense immunoreactivity for the γ isozyme was found in small cells with radiating and tapering processes surrounding intercalated ducts and acini of the three glands and granular convoluted tubules of the submandibular gland (Fig. 5a‐c). In addition, moderate to intense immunoreactivity for the γ isozyme was found in vascular wall cells of intraglandular arterioles and venules in the three glands (Fig. 4a).

Figure 4.

(a‐d) Immuno‐light micrographs of PG (a), SLG (b) and SMG (c) for PIP5Kβ under normal conditions. Note moderate and diffuse immunoreaction in the cytoplasm of epithelial cells of striated ducts (SD) and weak reaction in cells of granular convoluted tubules (GCT) in contrast to the absence of reaction in acini (A) of the three glands. No significant immunoreactivity is discerned in blood vessels (v) or in any portions of the parotid gland in the antigen‐preabsorption control (d). ICD, intercalated duct. Scale bars: 20 μm.

Figure 5.

(a‐d) Immuno‐light micrographs of PG (a), SLG (b) and SMG (c) for PIP5Kγ under normal conditions. Note distinct immunoreactivity in myoepithelial cells characterized by radiating processes (white arrowheads), which surround the intercalated ducts (ICD) contiguous to acini (A) in PG (a), the acini themselves (A) in SLG (b) and the acini (A), as well as partially in the granular convoluted tubules (GCT) (black arrowhead) in SMG (c). The ductal epithelial cells in the three glands are weakly and diffusely immunopositive. Also note distinct immunoreactivity in vascular wall cells (v) of intraglandular arterioles (a). No significant immunoreactivity is discerned in any portion of the PG in the antigen‐preabsorption control (d). Scale bars: 20 μm.

Control experiments using antigen‐preabsorption at light microscopic levels did not demonstrate significant immunoreactivities for any of the PIP5Ks isoforms in any portion of the three glands under normal conditions or following stimulation (Figs 3e, 4d and 5d).

Responsiveness of PIP5Ks to β‐adrenoceptor stimulation

Positive immunoreaction for PIP5Kα was only seen in the parotid gland. The acini are known to be regulated by β‐adrenoceptor stimulation (Proctor, 2016) and to be rich in β‐adrenoceptors (Nezu et al. 2005). Therefore, description of immunohistochemical features at the four post‐injection stages after stimulation by isoproterenol will be confined to the parotid gland. At 15 min post‐injection, apical immunoreactivity for the α isozyme was seen in almost all acinar cells, with a more distinct appearance than in normal specimens, and often resembling double lines contouring the expanded acinar lumens. At 30 and 60 min post‐injection, the apical immunoreactivity often appeared distinctly dense, thick and often with double lines contouring the expanded acinar lumens in almost all acini (Fig. 3d). At 120 min post‐injection, the apical immunoreactivity seemed to be weaker in most acini, although its intensity was still higher than in the pre‐injection normal state. In contrast, the immunoreactivity for the α isozyme in ductal cells (Fig. 3d) and for the β and γ isozymes (data not shown) in the gland did not change relative to control at any of the post‐injection stages examined. Western blot analysis of stimulated parotid gland lysates did not demonstrate any significant changes in the intensity of immunoreaction for the α, β and γ isozymes at any post‐injection time stage (Fig. 6).

Figure 6.

Western blots demonstrating expression of PIP5Ks in mouse PG under isoproterenol‐stimulated conditions at the four post‐injection stages. The expression of β‐actin in individual specimens is shown at the bottom of the figure as internal control. Note lack of change in the expression intensity of all three PIP5Ks over time post‐stimulation.

In immunoelectron microscopy for PIP5Kα of normal parotid acinar cells (Fig. 7), the apical membrane surfaces mostly had a smooth contour with invaginations, and infrequently small vesicles. Short microvilli were relatively sparse and occurred irregularly. Most short microvilli appeared in groups of several microvilli and sometimes as a mixture with only one or two longer microvilli. Immunoreactive material was deposited on a few acinar microvilli and, to a limited extent, on their contiguous plasma membranes and subjacent cytoplasm. The limiting membranes of secretory granules were generally free of immunoreactive material. No immunoreactive material was associated with portions of the apical plasma membranes to which the secretory granule membranes were directly attached/confluent. In addition, no immunoreactive material was deposited beyond the epithelial junctional specializations onto the lateral plasma membranes that had longer microvilli.

Figure 7.

Immuno‐DAB‐electron micrographs of apical domains of normal PG acinar cells. PIP5Kα‐immunoreactive material was associated with a few apical microvilli (black arrowheads), and, to a limited extent, their contiguous plasma membranes and subjacent cytoplasm. Note lack of immunoreactive material on the limiting membranes of most secretory granules (g) and also along portions of apical plasma membranes in contact/confluent with secretory granules (asterisks). Epithelial junction (tj) partition of immunoreactive apical membranes of acinar cells from their immunonegative lateral membranes with longer microvilli (arrow) and apposed to intercellular spaces (IS). L, apical lumen of acini; Nu, nucleus. Scale bar: 800 nm.

At 30 and 60 min post‐injection of isoproterenol, only a few secretory granules remained in the apical cell domains, in accordance with a previous ultrastructural study (Simson, 1969). The surfaces of apical membranes were largely wavy/irregular in contour, enclosing expanded lumens, and were frequently associated with microvilli or invaginations and subjacent vesicles, resulting in a smaller smooth surface than seen in normal apical membranes. The immunoreactive material was densely deposited on a large number of apical microvilli and their contiguous plasma membranes, and . in their subjacent cytoplasm to the depth of approximately 100–200 nm (Fig. 8a‐c). Immunoreactive microvilli with a larger diameter and length than normal were frequently seen. Membranous invaginations and small vesicles were frequently associated with the immunoreactive deposits. Portions of plasma membranes with smooth contours were largely free of immunoreactive materials (Fig. 8a).

Figure 8.

(a‐c) Immuno‐DAB‐electron micrographs of apical domains of parotid acinar cells at 30 min post‐injection of isoproterenol. Note that electron‐dense immunoreactive material is more extensively deposited along apical plasma membranes of wavy contour, especially on much more numerous microvilli than in the normal control shown in Fig. 7. Membranous invaginations and small vesicles (white arrows) are also immunoreactive. Inset (in part label a) shows that immunoreaction deposits never occur beyond epithelial junction (tj) partitioning apical membrane domains enclosing lumen (L) from lateral ones (*). Underlining with a white solid line indicates apical plasma membranous portions of smooth contours which lack immunoreactive material and are associated with immunonegative microvilli (m). Further note that most immunoreactive microvilli are larger in size than control (white arrowheads). mi: mitochondria. Scale bars: 500 nm (8a, 8a inset): and 200 nm (b,c).

In immune‐electron microscopy for PIP5Kβ, the immunoreactive materials were deposited rather diffusely throughout the cytoplasm of ductal cells in the three glands (data not shown). Conversely, in immuno‐electron microscopy for PIP5Kγ in the three glands under normal conditions, the intensely immunoreactive cells and cytoplasmic processes were directly apposed to immunonegative acinar cells of the glands, with an intercellular space of 50 nm in width, and were characterized by the occurrence of thin bundles of filaments, indicating that the cells were myoepithelial cells (Fig. 9a,b). The immunoreactive materials were deposited in association with the filament bundles. In addition, vascular immunoreactive cells were confirmed to be medial smooth muscle cells of arterioles in which the immunoreactive materials were deposited in association with myofilament bundles (Fig. 10).

Figure 9.

(a,b) Immuno‐DAB electron micrographs for PIP5Kγ in normal PG (a) and SMG (b). Note immunoreactive cellular processes (My) rich in myofibrils and in direct contact with glandular cells without basal lamina intervening, which are characteristics of myoepithelial cells. Acinar cells (A) characterized by specific secretory granules (ag) in (a) and granular convoluted tubular (GCT) cells characterized by specific granules (g) in (b). White asterisks indicate the immunoreaction products in association with filament bundles of My. Scale bars: 1 μm.

Figure 10.

Immuno‐DAB electron micrograph for PIP5Kγ of interlobular arteriole in PG. Note immunoreaction products (white asterisks) in association with myofilament bundles in medial smooth muscle cells (m). E, endothelial cell; IS, interstitial space; Lv, vascular lumen. Scal bars: 1 μm.

Discussion

The distinct, though weak, localization of endogenous PIP5Kα on the apical plasma membranes in acinar cells in situ of normal parotid glands is a notable finding of the present study. This was not discerned in the sublingual or submandibular acinar cells. The densities of individual western‐blot bands specific for the three PIP5Ks varied among the three glands, with the highest found in the parotid gland. These findings strongly suggest differences among the three glands in the apical membrane dynamics in which PIPKα is involved, indicating that the three salivary glands are discrete not only in histology but also in secretory activity, as recently noted (Isola & Lilliu, 2015; Lilliu et al. 2015).

Although this is the first report of PIP5Kα activity confined to the acinar cells in salivary glands, several studies have reported the apical plasmalemmal localization of endogenous PIP5Kα and β in cultured renal collecting tubular cells, another polarized epithelial cell type similar to the glandular acinar cells (Weixel et al. 2007; Szalinski et al. 2013). In addition, the localization at the plasmalemma of exogenously expressed PIP5K isoforms has also been reported in MDCK and CCD cells of kidney origin (Bairstow et al. 2006; Guerriero et al. 2006; Weixel et al. 2007; Szalinski et al. 2013). This suggests a similar functional activity of PIP5K at the apical plasmalemma of both salivary acinar cells and renal tubular cells.

Ion and fluid secretion is a likely target for PIP5Kα and PIP2 because it is a vital function of all epithelia, including these two cell types for survival of underlying tissues, and especially because it is critical for saliva formation in the acinar cells. Various ion‐transporting proteins, such as a chloride channel termed TMEM16/Ano1 and a Na⁺/H⁺ exchanger named NHE3, are localized at the apical plasmalemma of both salivary and renal cells (Park et al. 1999; Wade et al. 2003; Farial et al. 2014; Janga & Oh, 2014). Both TMEM16 and NHE3 are among the membrane molecules regulated by PIP2 (Suh & Hille, 2005; Pritchard et al. 2014; Hille et al. 2015). In addition, a water channel aquaporin and TRPC (transient receptor potential cation channels) are localized on the apical plasmalemma of both kidney and salivary gland cells (Nielsen et al. 2002; Goel et al. 2006; Sugimoto et al. 2013; Ambudkar, 2014). The regulation of aquaporin by PIP2 occurs only in plant cells, but has not yet been demonstrated in mammalian cells (Ma et al. 2015), and TRPC is included in the group of channels regulated by PIP2 (Suh & Hille, 2005). It is thus possible that PIP5K and its product PIP2 are involved in regulation of some ion‐transporting proteins in the apical membranes of the glandular acinar cells.

Next, the possible involvement of PIP5Kα and PIP2 in regulation of membrane trafficking including exocytosis and endocytosis, together with regulation of actin cytoskeletons should be considered: the apical plasma membrane is the site of regulatory exocytosis of salivary proteins. Although there have already been numerous studies analyzing the involvement of PIP5K and PIP2 in exocytosis, most have been done in neuroendocrine cells and neurons (Hay et al. 1995; Holz et al. 2000; Micheva et al. 2001; Aikawa & Martin, 2003; Bai et al. 2004; Zheng et al. 2004; Wang & Hilgemann, 2008; Wen et al. 2011). We found a significant increase in the immunoreactivity for PIP5Kα at the apical membrane of acinar cells after stimulation by isoproterenol, indicating the likely involvement of PIP5K and PIP2 in exocrine secretion via exocytosis.

The increase in the intensity of the immunohistochemical reaction in the acinar apical domain seen in the present immuno‐light microscopy is regarded as valid because both normal and stimulated specimens were mounted and immuno‐processed on the same glass slide. The observations were made by two independent histologists in addition to the present authors, and no significant differences in the immunoreactivity intensity of the ductal cells were discerned between the two specimens. In contrast, no significant difference in the intensity of immunoblot bands was seen between normal and stimulated specimens during the time‐course of post‐injection. This discrepant immunoblot finding, together with the changes in immunohistochemical reaction in relatively short post‐injection time‐courses (increasing to a peak between 30 and 60 min post‐injection and decreasing at 120 min post‐injection), leads us to conclude that the increase in the immunoreaction intensity along the apical plasma membranes is not due to stimulation‐induced synthesis enhancement of PIP5Kα but to its intracellular shift to the apical plasma membranes from intracellular sites, including the entire cytoplasm, whose immunoreaction under normal conditions was at a level below the immunohistochemical detection threshold. The Golgi apparatus seems to be one of the sites supplying PIP5Kα for the shift because it has already been described to contain PIP5K activity by one (H.K.) of the present authors (Jones et al. 2000) and by others in cells in vitro (Godi et al. 1999).

The present ultrastructural analysis clarified that the apical weak immunoreaction in normal acinar cells in immuno‐light microscopy corresponds to the confinement of its localization to a few microvilli and, to a limited extent, their contiguous plasma membranes and subjacent cytoplasm. This finding, together with a lack of immunoreaction at sites in close apposition or contact between the secretory granule membranes and the apical plasma membranes, suggests that the direct involvement of PIP5Kα in exocytosis itself is less likely than its involvement in post‐exocytic processes. This suggestion is supported by the peak of immunoreaction increase at 30~60 min post‐stimulation in immuno‐light microscopy, which corresponds to the period of post‐peak exocytosis reported by Simson (1969) where, similar to the present findings, acinar cells were reported to be almost devoid of secretory granules.

On the other hand, the present immuno‐electron microscopy of stimulated acinar cells revealed that the increase in immunoreactivity was almost all due to immunoreactive microvilli and contiguous plasma membranes including membrane invaginations and subjacent small vesicles. These findings suggest that PIP5Kα and PIP2 are involved in post‐exocytotic membrane traffickings that may be governed by microvilli.

In this regard, it should be noted that cells are equipped with membrane reservoirs in the form of cell surface protrusions and pits, and cell surface areas are regulated by the reservoir dynamics which are co‐ordinated with membrane supply from exocytosis and membrane demand from endocytosis, and is also collaborated with assembly and disassembly of F‐actin as one of reservoir scaffolds (Gauthier et al. 2011; Masters et al. 2013). Two types of microvilli are present: reservoir microvilli, represented by motile cells including macrophages, are heterogeneously shaped and sized, and contain relatively disorganized F‐actins with a short turnover of tens of seconds; and other microvilli, represented by stereocilia and brush border microvilli, are homogeneously shaped and sized, and contain highly ordered bundles of F‐actins with long turnover in minutes to hours (Figard et al. 2016). As far as we know, no reports in the literature have clarified which of the two types the acinar apical microvilli belong to. The present ultrastructural analysis showed that the wavy/irregular‐contoured surfaces of apical membranes were frequently associated with microvilli or invaginations and subjacent vesicles, and most apical microvilli were immunoreactive in stimulated acinar cells. In addition, immunoreactive microvilli with a larger diameter and length than normal were frequently seen. These findings suggest that microvilli as well as invaginations increased after the stimulation. The confirmation of this possibility admittedly requires further detailed analyses of total apical microvilli in statistically sufficient numbers of acinar apical lumens of specimens well preserved with immunoreaction damages overcome. This possible interpretation, together with the fact that apical microvilli occur irregularly on normal apical surfaces, leads us to suggest that the acinar apical microvilli represent the reservoir microvilli, and that PIP5Kα is involved in post‐exocytotic membrane transfer to and from the microvilli as a transient membrane reservoir.

Immunoreactivity for PIP5Kα was below detection levels in normal sublingual and submandibular acinar cells, at a lower level in the sublingual gland than in the parotid gland and below detection level in the submandibular gland. This suggests at least quantitative differences among the gland types in regulation mechanisms for membrane trafficking, membrane proteins and/or PI signaling, whichever is regulated by PIP5Kα via PIP2, at the acinar apical plasma membrane. It is well known that acinar secretory proteins differ according to gland type: pure serous secretions in the parotid gland vs. mucous and sero‐mucous secretions in the other two glands. The degree of sympathetic innervation is also known to vary among the three glands (Young & Cook, 1996; Proctor & Carpenter, 2007). Whether these differences are related to the degree of PIP5Kα ‐immunoreactivity on the acinar apical membrane among the three glands remains to be elucidated.

Diffuse localization of PIP5Kβ‐immunoreactivity in the cytoplasm of ductal cells at moderate to weak levels was commonly observed in the three glands. PIP5K is activated by Arf6 (ADP‐ribosylation factor 6; see D'Souza‐Shorey & Chavrier, 2006), and immunoreactivity for Arf6 has been detected in salivary ductal cells of the submandibular gland by the present authors (Tachow et al. 2017). Our preliminary studies also found diffuse intracellular distribution of Arf6‐immunoreactivity in the parotid and sublingual ductal cells (data not shown). This corresponding localization of PIP5Kβ and Arf6 in the ductal cells suggests that the known activation of PIP5K by Arf6 is performed through the β isoform in the salivary ductal cells.

We identified PIP5Kγ ‐immunoreactive cells having radiating processes as myoepithelial cells based on their characteristic shapes, intimate association with acinar and intercalated ductal cells, and ultrastructural features including the presence of myofibrils (Young & Van, 1977; Redman, 1994; Kawabe et al. 2016). It is highly possible that PIP5K and its product, PIP2, are involved in the regulation of contraction of the myoepithelial cells. In support of this, there have been reports that, in airway smooth muscle cells, stimulation of G‐protein‐coupled receptors by contractile agonists activate Src‐kinase, which modulates the cell contractility and Ca²⁺ signaling by affecting the levels of PIP2 (Tolloczko et al. 2002), and that the expression level of PIP5Kγ directly correlates with force development in the smooth muscle cells (Chen et al. 2007). The present finding of PIP5Kγ ‐immunoreactivity in smooth muscle cells of intraglandular arterioles is compatible with this.

The differential localization reported here of the three isoforms of PIP5K individually in acinar, ductal and myoepithelial cells may explain why multiple isoforms of a given functional molecule exist. The same is the case for diacylglycerol kinase, which is two steps downstream of PIP5K in the PI signal cascade, as demonstrated by one (H.K.) of the present authors (Goto & Kondo, 1999; Kobayashi et al. 2007). Individual isoforms of diacylglycerol kinase are expressed/localized differentially in specific intra‐brain regions, specific species of neural cells and even in specific organelles within a cell in vitro.

Taken together, the present findings provide information that can lead further analysis of the involvement of phosphoinositide‐signaling in the regulation of exocrine secretion.

Author contributions

S. Khrongyut contributed to the practical experiment, data analysis and interpretation. A. Rawangwong and A. Pidsaya contributed to the practical experiment and data analysis. H. Sakagami contributed to the synthesis of antibodies, data analysis and interpretation. H. Kondo contributed to the experimental design, interpretation and made critical revisions to the drafted manuscript. W. Hipkaeo contributed to the experimental design, data analysis and interpretation, and drafted the manuscript. All authors read and approved the submitted manuscript.