Cellular Localization of Gene Expression for Progranulin

Abstract

Granulins, also called epithelins, are 6-kD peptides with growth modulatory effects on a variety of cells. The granulin/epithelin precursor supports tumorigenesis in appropriate cell models and is the only growth factor able to overcome the cell cycle block that occurs in murine fibroblasts after deletion of a functional IGF-1 receptor. However, little is known of the role of granulin/epithelin gene products in vivo. To understand the physiological role of granulins it is essential to know the cell types and conditions in which it is expressed. We examined granulin/epithelin gene expression in adult rodents by in situ hybridization. The granulin/epithelin precursor is constitutively expressed in a number of epithelia, particularly in the skin, GI tract, and reproductive system. Other epithelia express the gene less strongly. Progranulin is expressed in immune cells in vivo and in specific neurons in the brain, including Purkinje cells, pyramidal cells of the hippocampus, and some cerebral cortical neurons. Little expression was detected in muscle cell, connective tissue, or endothelium. Cumulatively, these results define the basal gene expression of a new growth factor system and suggest that the progranulin/epithelin gene is multifunctional, with important constitutive roles in epithelial homeostasis, reproductive, immunological, and neuronal function.

Granulins (grns) (Bateman et al. 1990), also called epithelins (epis) (Shoyab et al. 1990), were initially identified as peptides of approximately 6-kD, some of which can modulate the growth of cells in tissue culture (reviewed in Bateman and Bennett 1998). They are rich in cystine and possess a unique structurally defined motif of six disulfide bonds (Hrabal et al. 1996). All known mammalian grns are generated from a common precursor, progranulin, which consists of 7.5 sequentially arranged grn/epi motifs arranged in tandem (Bhandari et al. 1992, 1993; Plowman et al. 1992; Baba et al. 1993) and which is potently mitogenic in cell culture (Zhou et al. 1993; Zhang and Serrero 1998; Xu et al 1998; He and Bateman 1999). Progranulin stimulates proliferation of many epithelial cells, promotes their anchorage-independent growth and, when overexpressed, confers epithelial tumorigenicity (He and Bateman 1999). In cell culture, the rate of epithelial proliferation is proportional to the level of progranulin gene expression (He and Bateman 1999). Consistent with these results, progranulin has recently been shown to modulate the normal development of early embryonic epithelia (Diaz–Cueto et al. 2000). Many epithelial tissues and carcinomas contain a 25-kD protein called epithelial transforming growth factor (TGFe), which has some structural similarities to progranulin (Parnell et al. 1992). Like progranulin, it is also a potent mitogen for many epithelial and mesenchymal cell lines (Halper and Moses 1983, 1987; Brown and Halper 1990; Parnell et al. 1990; Dunnington et al. 1990a, b) Nonepithelial cells also respond mitogenically to progranulin. PC cells, which are highly tumorigenic murine mesenchymal teratoma cells (Serrero et al. 1991), require progranulin (called PCDGF in this context) to sustain their growth (Zhou et al. 1993), and antisense targeting of progranulin/PCDGF dramatically lowers their tumorigenicity (Zhang and Serrero 1998). The growth of nontransformed embryo fibroblasts typically requires insulin-like growth factor (IGF)-I or IGF-II and at least one other growth factor, such as EGF, PDGF, or FGF (Baserga et al. 1997). The central role of the IGF-I receptor in this pathway has been demonstrated using embryonic fibroblasts from mice whose IGF-I receptor has been deleted (Coppola et al. 1994; Sell et al. 1994). These cells, called R⁻, no longer grow in response to IGFs. Less predictably, in the absence of a functional IGF-I receptor proliferation in response to all other classic growth factors is precluded (Coppola et al 1994; De Angelis et al. 1995), and several intracellular oncogenes are blocked in their ability to transform the cells (Sell et al. 1994). Progranulin, in contrast, potently stimulates cell cycle progression in R⁻ embryo fibroblasts independently of an intact IGF-I signaling system and is the only growth factor yet shown to circumvent the mitotic block imposed on R⁻ cells (Xu et al. 1998). Progranulin does this by activating the MAP Kinase and PI-3 Kinase pathways of R⁻ cells independently of IGF-I receptors or the IRS-1 adaptor protein (Zanocco–Marani et al. 1999).

Although progranulin, progranulin-derived peptides, and related proteins such as TGFe are clearly capable of regulating cell growth, less is known of their roles in vivo. Most established cell lines of epithelial origin express the progranulin gene strongly (Bhandari et al. 1992; Plowman et al. 1992). It is also abundant in many cell lines of hematopoietic origin (Bhandari et al. 1992), in some fibroblastic cell lines (Bhandari et al. 1992; Plowman et al. 1992) and as a glycoprotein called acrogranin in guinea pig acrosomal granules (Anakwe and Gerton 1990; Baba et al. 1993). Northern analysis of tissues indicates that it is highly expressed in the spleen, placenta, ovaries, epididymis, and adrenal gland. Moderate expression was observed in the gastrointestinal tract, lung, and kidney. Expression was low in skeletal muscle, heart, and brain (Bhandari et al. 1993). In some instances, progranulin gene expression is regulated by sex steroids. In estrogen-responsive breast cell lines, progranulin gene expression was upregulated by estrogens (Lu and Serrero 1999). Androgens potently induced progranulin in the ventromedial and arcuate nuclei of newborn female mice, whereas progranulin expression was high and constitutive in the same regions of newborn male brains (Suzuki et al. 1998), suggesting a role for progranulin products in phenotypic sex determination of the hypothalamus. Given the proliferative effects of progranulin on diverse cells in vitro (Zhou et al. 1993; Xu et al. 1998; He and Bateman 1999) and its ability to activate mitogenic signal transduction cascades (Zanocco–Marani et al. 1999), it is essential to establish the cellular localization and conditions of progranulin gene expression in vivo so that the physiological context of progranulin action can be investigated. Here we report the cellular distribution of the grn/epi precursor mRNA in healthy adult rodent tissue.

Materials and Methods

Isolation of RNA and Northern Blotting Analysis

Around 2 × 10⁶ cells were harvested and washed with ice-cold PBS, pH 7.4, Total cellular RNA was isolated by the guanidium thiocyanate method (Chomczynski and Sacchi 1987). Fifteen μg total RNA was denatured with glyoxal, separated in a 1% agarose gel (1% agarose in 10 mM NaHPO₄, pH 7.0), transferred to nylon membrane, and hybridized with a ³²P-labeled granulin cDNA probe as described (Bhandari et al. 1993). Unless stated otherwise, all cell lines were of human origin and were obtained from ATCC (Manassas, VA). The cell lines used were: A549 (lung carcinoma), A431 (vulval epidermoid carcinoma), CaSki-1 (cervical epidermoid carcinoma), Calu-6 (anaplastic carcinoma), CHO (Chinese hamster ovary), SW-13 (adrenal cortical adenocarcinoma), MDCK (canine kidney epithelia), HepG2 (hepatoma), NHEK (normal human epidermal keratinocytes; gift of Dr. R. Kremer, Calcium Laboratory, Royal Victoria Hospital), primary culture of bronchial epithelium (gift of Dr. A. Giad, Montreal General Hospital), L6 (rat skeletal myoblasts), SKUT-1 (uterine mixed mesodermal tumor), SK-LMS-1 (vulval leiomyosarcoma), Swei (B-lymphoma), CEM-CM3 (T-cell lymphoblastic leukemia), HL60 (promyelocytic leukemia), U937 (histiocytic lymphoma), K562 (erythroleukemia), AML (primary acute myelogenous leukemia; gift of Dr. C. Shustik, Royal Victoria Hospital), CML (primary cells from chronic myelogenous leukemia; gift of Dr. C. Shustik), NIH-3T3 (murine embryonic fibroblasts), COS-7 (transformed monkey kidney fibroblasts), Malme-3M (melanoma), PC-12 (rat adrenal gland pheochromocytoma), and SK-N-DZ (human brain neuroblastoma) cells.

Generation of Digoxigenin-UTP-labeled Progranulin RNA Probe

A 238-bp fragment corresponding to nucleotides 748–1002 of full-length rat progranulin cDNA was blunt end-ligated to an expression vector BlueScript KII (Invitrogen, Carlsbad, CA). The 5′ end of the fragment is adjacent to an Xba restriction site with an upstream T7 promoter; a Xho restriction site is located downstream of the 3′ end of the fragment with a T3 promoter further down on the line. Sense or anti-sense digoxigenin-UTP-labeled RNA probes were generated by reverse-transcribing the 238-bp progranulin cDNA fragment using either T7 (sense) or T3 (antisense) RNA polymerase after digestion with either Xho (sense) or Xba (anti-sense). The reverse-transcription was performed using a commercially available Dig-UTP labeling and detection kit (Boehringer–Mannheim; Indianapolis, IN) according to the manufacturer's instructions. The probe was checked by performing Northern blotting hybridization with mouse kidney total RNA. In brief, two identical membranes containing 15 μg mouse kidney total RNA were hybridized to 100 ng/ml sense or antisense probe and washed. Signals on the membranes were detected as described (Genius System User's guide for membrane hybridization, version 3.0; Boehringer–Mannheim).

In Situ Hybridization

Paraffin-embedded sections were deparaffinized in xylene (Fisher; Montreal, PQ, Canada) for 4 min, then dehydrated in 100% ethanol followed by progressive hydration in 95%, 70%, and 50% ethanol. The sections were then postfixed in prechilled 4% paraformaldehyde (Fisher), pH 7.4, for 10 min, and washed in 0.5 × SSC for 5 min. Permeabilization of the tissues was achieved by incubating the sections with 3.5 μg/ml proteinase K in 100 mM Tris-HCl, 50 mM EDTA, pH 8.0, at 37C for 15 min. The slides were fixed again in 4% paraformaldehyde for 10 min and rinsed thoroughly in PBS (6 min) and 0.5 × SSC (10 min). The slides were then pre-hybridized in hybridization solution [5 × SSC, 5 × Denhardt's solution, 50% deionized formamide (Fisher), and 250 μg/ml tRNA] at 42C. Three hours later the sections were cleaned with a lint-free tissue and hybridized with 75 ng Dig-UTP labeled progranulin RNA probe in hybridization solution for 18 hr at 42C. After washing, the slides were incubated with conjugated Dig antibody (Boehringer–Mannheim) and the reaction products were visualized according to the manufacturer's instructions. All experiments were conducted using parallel antisense and sense probes, and in most cases no nonspecific hybridization was observed (this is shown only for the testis sections; see Figure 3). Any experiments showing significant nonspecific hybridization with the sense probe were discarded. Sections from two species, rat and mouse, were analyzed. The numbers of different animals used for each tissue were as follows: testis and epididymis; n = 4; female reproductive organs; n = 3; mammary gland; n = 4; skin; n = 4; gastrointestinal tract; n = 5; urinary system; n = 5; lung; n = 5; liver; n = 4; skeletal muscle; n = 4; heart; n = 5; spleen; n = 5; brain; n = 3. Each tissue was serially sectioned and at least three sections per sample were examined. The species shown in the figure is identified in the legends.

Purification of the Recombinant Progranulin

Recombinant progranulin was produced as described previously (He and Bateman 1999). Briefly, COS-7 cells were transiently transfected with a full-length progranulin construct in pcDNA3 or with the pcDNA3 vector alone (mock-transfected). [³⁵S]-Cysteine was used to metabolically label protein production. After 48 hr the serum-free conditioned medium was collected and then concentrated using a Centricon-30 spin column (Millipore; Bedford, MA). The concentrated CM was then fractionated using C-4 reversed-phase HPLC as described (He and Bateman 1999). A single transfection-specific [³⁵S]-cysteine-labeled component was detected (not shown) with an apparent molecular weight of 80 kD as assessed by SDS-PAGE and autoradiography (He and Bateman 1999). The purity of the progranulin was further confirmed by SDS-PAGE and silver staining (not shown). This protein exhibited colony-forming activity on SW-13 adrenal carcinoma cells, and no comparable activity was detected in the mock transfectants (not shown). N-terminal gas-phase microsequence analysis has confirmed the identity of this protein as intact human progranulin (He and Bateman 1999).

Cell Growth Assay

To compare the effect of progranulin on selected cell lines, 1.5 × 10⁴ cells were seeded in 24-well plates (Corning Co-star; Cambridge, MA). After 24 hr the medium was replaced with serum-free DMEM for 24 hr. Before addition of the re combinant progranulin, the cells were washed twice in 1 × PBS (pH 7.4) and then incubated with increasing concentrations of recombinant progranulin for 2 days. The cells were trypsinized and counted in a hemocytometer.

Results

Northern Blot Analysis

Total RNAs from several cell lines of various origins were probed for the expression of the progranulin gene (Figure 1A). All epithelial cell lines except the hepatoma line HepG2 expressed the progranulin transcript at high levels. Progranulin transcripts were also abundant in primary cultures of human bronchiolar epithelia and normal human keratinocytes. Myeloid- and lymphoid-derived cell lines also expressed the progranulin transcript, as did leukocytes from patients with acute and chronic myelogenous leukemia. Progranulin transcripts were detected in two leiomyosarcoma-derived lines (SK-UT-1 and SK-LMS-1) but not in undifferentiated L6 rat myoblasts. Neither of the neuron-derived cells lines (PC-12 and SK-N-DZ) expressed progranulin mRNA (not shown). The anti-sense riboprobe used for in situ hybridization is specific for progranulin transcripts because on Northern blotting analysis of murine kidney mRNA it hybridized to only one mRNA, which had a size of 2.4 kb as predicted for the rodent progranulin transcript (Figure 1B). No hybridization was observed with the sense riboprobe.

Figure 1.

(A) Northern blotting analysis of progranulin mRNA in cells from different origins in culture (the origin of the cells is given in Materials and Methods). Note that this is a summary of several experiments. (B) To evaluate the specificity of the riboprobe used for in situ hybridization, the antisense and sense probes were hybridized to murine kidney mRNA. No signals were detected for the sense probe.

Normal Tissue Distribution

Testis and Epididymis. In the seminiferous tubule, the cells that line the basement membrane (mainly primary spermatogonia) expressed progranulin mRNA, as did the cells between one to three layers deeper within the tubule (Figure 2A). More mature spermatocytes showed little or no expression of progranulin mRNA. In some tubules, Sertoli cells also expressed the progranulin gene. Leydig cells and other cells in the interstitium rarely expressed progranulin. No vascular elements in the testis expressed progranulin. The sense control probe detected no hybridization (Figure 2C). The epididymis showed intense expression of the progranulin gene, which was confined to the epithelial layer (Figure 2B).

Female Reproductive Organs. A section of uterus in the proliferative phase is shown. Glandular epithelium showed pronounced expression of the progranulin gene (Figure 2D). Stromal cells in the proliferative zone expressed the progranulin gene. There was a marked boundary in the basal endometrium which exhibited much lower levels of progranulin hybridization. The smooth muscle cells of the myometrium expressed elevated levels of progranulin mRNA. In the oviducts, only the epithelia lining the oviduct expressed the progranulin mRNA (Figure 2E). The oviductal smooth muscle and vasculature did not express the progranulin gene. In the ovary, hybridization for the progranulin riboprobes was highest in follicular epithelial cells. Hybridization was weak in the corpus luteum and negligible in thecal cells. Progranulin transcripts were also present in oocytes (Figure 2F).

Mammary Gland. Hybridization of the progranulin riboprobe in virginal rat mammary gland was detected exclusively in the glandular epithelium (Figure 2G).

Skin. Epidermal keratinocytes stained strongly for granulin mRNA (Figure 3A). The outer hair follicle hybridized with the progranulin antisense riboprobes (Figure 3B). Cross-sections of the hair follicle showed that the dermal papillae did not express progranulin mRNA. All the epithelial cells at the hair root strongly expressed the progranulin gene. The progranulin gene was expressed in exocrine glands (Figure 3C), although the staining was less intense than that seen in the keratinocytes. Apart from follicles and glandular structures, little if any granulin mRNA was detected in dermal layer fibroblasts or in blood vessels.

Gastrointestinal Tract. Progranulin mRNA appeared to be predominantly located in epithelial cells in the gastrointestinal tract. Squamous epithelium in the esophagus and esophageal/gastric junction stained strongly for progranulin mRNA (Figure 3D). In the small intestine, progranulin was expressed by enterocytes in the deep crypts, with the mRNA levels progressively fading higher in the villus (Figure 3E). A similar pattern was observed in the colon (not shown). Smooth muscle and vasculature showed no progranulin mRNA. However, neuronal ganglia were strongly positive for progranulin mRNA (Figure 3E and insert, white arrow). Lymphocytes in the gut-associated lymphoid tissue hybridized strongly for progranulin mRNA (Figure 3F).

Kidney. Progranulin mRNA was confined to epithelial cells in the kidney. There was weak hybridization in the proximal and distal convoluted tubules of the cortex and in some collecting ducts of the medulla (Figure 3G). No hybridization was observed in the glomerulus or any vascular elements. The strongest hybridization occurred in the transitional epithelium of the ureter, which approached that of skin and intestinal crypt epithelium in intensity.

Lung. Progranulin mRNA was localized in alveolar epithelium only after prolonged incubation with the color reagents, and was confined only to sporadic epithelial cells (Figure 3H). Higher levels of progranulin mRNA were observed in the bronchiolar epithelium (Figure 3I, black arrow), but the intensity of hybridization was considerably lower than in keratinocytes or enterocytes. In the lung, lymphoid cells exhibited the highest levels of progranulin transcript (Figure 3I, open arrow). Progranulin mRNA was absent from cells of the lung vasculature.

Figure 2.

Progranulin gene expression in the male (A–C) and female (D–F) reproductive system and the virginal rat mammary gland (G). Seminiferous tubules of the murine testis (A, × 175). In the epididymis (B, × 175), the epithelia stained strongly. Negative staining of the testis after hybridization with the sense progranulin probe (C, × 175). In the murine uterus (D, × 175) progranulin gene expression was observed in the glandular epithelium (black arrow), stroma (open arrow), and weakly in the myometrial smooth muscle (∗). In rat oviduct, only epithelial cells exhibit progranulin expression (E, × 175). In rat ovary (F, × 340), some follicular cells and oocytes have progranulin expression. In the virginal rat mammary gland (G, × 175), progranulin expression is seen in the glandular epithelium (black arrow).

Liver. The liver was essentially devoid of cells expressing the progranulin mRNA (not shown).

Skeletal Muscle and Heart. Neither skeletal muscle nor cardiac muscle expressed the progranulin gene (not shown).

Spleen. In the spleen (Figure 4A and 4B), expression of progranulin mRNA occurred predominantly in lymphoid cells on the outer edges of the periarteriolar lymphoid sheaths (white pulp). Cells in the red pulp stained sporadically.

Brain. Progranulin transcripts were prominent in neurons within the superficial lamina of the cerebral cortex (Figure 5A). Expression was high in the Purkinje cells of the cerebellum (Figure 5B), and in scattered cells in the molecular layer, but less strongly. Expression was absent from granule cells of the internal granular layer. Neurons of the cerebellar roof nuclei showed intense hybridization with the progranulin probe, which was comparable to that of the Purkinje cells (not shown). Progranulin gene expression was high in the hippocampus and was localized to the pyramidal cells and the granule cells. In the posterior portion of the hippocampus, all the granule cells stained evenly (not shown). However, in the mid-hippocampus (Figure 5C and 5D), the granule cells stained more intensely in the Ammon's horn (black arrow) than those in the dentate gyrus (open arrow). Nonneuronal components of the brain, such as ependymal cells and glia, did not express progranulin transcripts.

Cell Proliferation

Eight cell lines were selected and tested for their proliferative response to progranulin (Figure 6). Two epithelial lines previously shown to respond to progranulin (He and Bateman 1999) were included for comparative purposes (A549, human lung adenocarcinoma; MDCK, a nontransformed canine renal epithelial cell type). To test for the proliferative effects of progranulin in hematopoietic-derived cells, we used U937 (histiocytic lymphoma), and HL-60 (promyelocytic leukemia). To test for the effects of progranulin in proliferative smooth muscle-like cells, we used SK-UT-1 (a uterine leiomyosarcoma) and SK-LMS-1 (a vulval leiomyosarcoma). To test for the proliferative effects of progranulin in neuron-derived cells, we used PC-12 (rat adrenal gland pheochromocytoma) and SK-N-DZ (human brain neuroblastoma). Both epithelial cell lines proliferated in response to progranulin, as did the mesen-chyme-derived leiomyosarcoma SK-LMS-1 and neuronal PC-12 cells, but not SK-UT-1, SK-N-DZ, HL-60, and U937.

Discussion

Progranulin gene products promote tumorigenesis in experimental models (Zhou et al. 1993; Zhang and Serrero 1998; He and Bateman 1999) and are uniquely able to circumvent the requirement for a functional IGF-I receptor in the proliferation of untransformed murine embryonic fibroblasts (Xu et al. 1998), but their significance and roles in physiological systems are not well understood. It has recently been shown that progranulin is needed for epithelial development in early embryos (Diaz–Cueto et al. 2000). Here we have defined the basal gene expression of progranulin in adult rats and mice and find that it is prominent in certain epithelial, immune and neuronal cells in vivo. Comparable expression patterns were observed both between species and between individuals of the same species. If, as proposed elsewhere from cell culture studies (He and Bateman 1999), the level of epithelial expression of the progranulin transcript regulates epithelial mitogenesis, we would predict that highly proliferative epithelia should express elevated levels of the progranulin gene in vivo. In general, this is observed, because rapidly cycling epithelium, such as that in the skin and gastrointestinal tract, exhibits greater hybridization with the progranulin riboprobes than mitotically less active epithelia, such as the pulmonary alveolar cells or the renal tubules. In this regard, the small intestine and colon are particularly instructive because proliferating enterocytes are located deep within the crypt and are segregated from terminally differentiating cells at higher levels within the villus (Figure 3E). In the small intestine and colon, the most intense progranulin-gene expression occurs in the deep crypts, roughly coincident with the zone of less well-differentiated but highly proliferative cells. Progranulin expression declines rapidly at higher levels within the villus as the cells differentiate and cease to divide. Although progranulin expression in epithelial cells appears highly regulated in vivo, progranulin mRNA was abundant in all tranformed or immortal epithelial cell lines examined except for the HepG2 hepatoma cells, even in cells such as A549 (lung) and MDCK (kidney) that originate from tissues that express the progranulin gene relatively weakly (Figure 1A). This suggests that high progranulin expression in epithelia is associated with a rapidly proliferative phenotype regardless of tissue origin. Clearly, however, some epithelial cells, such as the epididymis and urothelium, that are less mitotically active than the skin or gastrointestinal tract retain high levels of progranulin expression.

Figure 3.

Progranulin gene expression in the skin (A–C), GI tract (D–F), kidney (G), and lung (H, I). In the rat epidermis, extensive staining was observed only in the keratinocytes (A, black arrow, × 340), hair follicles (B, black arrow, × 340) and outer cells of the shaft (B, open arrow, × 340). Exocrine sweat glands in the dermis express progranulin (C, × 340). In the rat GI tract, epithelial cells express progranulin in the esophageal gastric junction (D, × 170) and in the ileum (E, open arrow). Inset in E shows progranulin expression in the enteric ganglia (black arrow, × 680). Progranulin expression was observed in lymph node follicles adjacent to the large intestine (F, × 340). In the rat kidney (G, × 340), the strongest hybridization is in the transitional epithelium of the beginning of the ureter (∗), and is weak in the collecting ducts of the medulla (open arrow). Tubules of the cortex stained with the same intensity as the collecting tubules (not shown). In the rat lung parenchyma, only sporadic epithelial cells showed progranulin expression (H, × 340, black arrow). The bronchial epithelium (I, × 340, black arrow) and lymphoid tissues (I, open arrow) beneath the bronchial mucosa have the strongest progranulin expression in the lung.

Figure 4.

(A, B) Progranulin gene expression in rat spleen. Progranulin expression is noted in lymphoid cells in the outer margin of the white pulp (black arrow); cells in the red pulp stained sporadically. Original magnifications: A × 175; B × 340.

Figure 5.

Progranulin expression in mouse brain. Progranulin expression was detected in neuronal cells, but not glial cells in several brain regions. Cortex (A, × 680); in the cerebellum, expression was strong in Purkinje cells (arrow) (B, × 680); In the mid-hippocampus, granule cell neurons stain throughout the mid-hippocampal region, but more intensely in Ammon's horn (black arrow) than the dentate gyrus (open arrow) (C, × 40); D, × 175.

Figure 6.

Cell lines of epithelial, neuronal, smooth muscle, and myelogenic origin as described in Materials and Methods were incubated with increasing concentrations of purified recombinant progranulin in serum-free medium. Cell growth was determined and expressed as percentage of unstimulated cells (n = 4).

The results reported here support an important role for progranulin gene expression in both male and female reproductive tissues. In the adult testes, the highest levels of progranulin mRNA are associated with immature spermatocytes and spermatogonia but not with more mature spermatozoa (Figure 2A). This is presumably correlated with the synthesis of acrogranin, a high molecular weight progranulin product found within acrosomes (Anakwe and Gerton 1990). The expression is also high in the epididymal epithelium (Figure 2B), but whether the progranulin products act on the epididymal cells or on maturation of intralumenal sperm (Moore and Akhondi 1996) is unknown. In ovaries, a subset of follicles show high levels of expression in the epithelial cells but not in thecal cells (Figure 2F). The expression in the corpus luteum is uniformly low. Because luteal cells and mature spermatocytes are highly differentiated epithelial cells, this result is consistent with a model in which less well-differentiated or highly proliferative, or potentially proliferative, epithelia express elevated levels of the progranulin gene, and fully differentiated but less mitotically active epithelial cells express lower or negligible levels of progranulin mRNA.

Progranulin is a potent mitogen for fibroblasts (Zhou et al. 1993; Xu et al. 1998) and other mesenchymal cell lines such as SK-LMS-1 (Figure 6). These cells also express the progranulin gene (Figure 1).

However, the progranulin transcript was undetectable in most mesenchymal tissues in vivo, such as the connective tissue, adipose cells, skeletal muscle, smooth muscle, cardiac muscle, or vascular elements. Therefore, despite the mitotic responsiveness of mesenchyme to progranulin, intrinsic progranulin gene expression by these cells is not required for their survival and constitutive activity in vivo. The major exception is the female reproductive tract. In the female reproductive system, the epithelial cells of the uterine glands are strongly positive for progranulin gene expression. However, the primitive mesenchyme in the proliferating layer surrounding the glands also expresses the progranulin gene. The connective tissue beneath the proliferating layers shows little progranulin gene expression, but the smooth muscle of the myometrium expresses the progranulin mRNA (Figure 2D). In contrast, the smooth muscle of the oviduct does not express progranulin (Figure 2E). The myometrial smooth muscle is highly proliferative during pregnancy, whereas oviductal and gastrointestinal smooth muscle cells, which do not express appreciable levels of progranulin mRNA, are in general mitotically quiescent. The leiomysarcoma cell lines SK-UT-1 and SK-LMS-1 show relatively high level of expression of the progranulin gene (Figure 1), and progranulin stimulated proliferation in one of two leiomyosarcoma cell lines to the same extent as in the epithelial cell lines A549 and MDCK (Figure 6), supporting a potential mitotic role for progranulin in transformed or proliferating smooth muscle. These results suggest that some nonepithelial somatic cells with an intrinsically high capacity to proliferate may express the progranulin mRNA at levels substantially higher than quiescent mesenchymal cells.

The peripheral immune system shows very strong expression of the progranulin gene in lymphoid tissue of the lung, gut, and spleen. The distribution of progranulin mRNA in immune tissues does not correspond to the expected patterns of immune cell proliferation, implying a nonmitogenic function for progranulin products in the immune system. In the spleen, the progranulin transcript is confined mainly to marginal cells of the periarteriolar lymphoid sheath (PALS) (Figure 4A and 4B). These cells are largely T-lymphocytes. The B-lymphocytic population within the interior of the PALS shows less extensive progranulin staining. Lymphoid aggregates in the lung and gut express progranulin mRNA at high levels (Figure 3I and 3F). Cell lines derived from both B- and T-lymphomas express the progranulin gene in culture (Figure 1A). The cellular specificity of progranulin expression in the peripheral immune system, which is particularly apparent in the spleen, reflects stringent regulation of the progranulin gene in the immune system. Granulins were initially detected in activated inflammatory cells (Bate-man et al. 1990), and the gene is highly expressed in myelogenous leukemic cell lines (HL-60, U937) and in leukocytes from the blood of leukemic patients (Figure 1). Neither HL-60 nor U937 responded mitotically to progranulin. Expression in myelogenous cells and lymphocytes indicates a probable role for progranulin or granulin in the defense system.

In earlier studies we reported that the levels of progranulin mRNA in the brain were very low compared to those in other tissues (Bhandari et al. 1993). This was probably misleading, because although the levels are low overall, we show here that the progranulin gene is highly expressed in specific subsets of neuronal cells, notably cortical neurons (Figure 5A), Purkinje cells of the cerebellum (Figure 5B), and granule cells of the hippocampus (Figure 5C and 5D). It is known that the hippocampus is crucial for memory formation and the transformation of short-term to long-term memory (reviewed by Squire 1992; Parkin 1996). Given that progranulin is expressed in the hippocampus, it will be interesting to determine whether progranulin gene products contribute to this process. Nerve cells in the intestinal myenteric and submucosal plexi also express progranulin transcripts (Figure 3E), indicating that the progranulin gene is expressed in both the peripheral and the central nervous system. PC-12 cells, which are often used to study neuron development, show enhanced growth in response to progranulin (Figure 6). This is the first demonstration of a direct biological response to progranulin in a neuron-related cell line. PC-12 cells grow poorly or not at all in response to most growth factors, including epidermal growth factor, platelet-derived growth factor, nerve growth factor, and bombesin (Nielsen and Gammeltoft 1988), but they respond mitotically to IGF-I and IGF-II. Interestingly, progranulin and IGFs activate comparable signal transduction pathways in embryonic fibroblasts (Zarocco-Marani et al. 1999), which may underlie the shared ability of progranulin and IGFs to activate mitosis in PC-12 cells. Progranulin is induced by androgens in ventromedial and arcuate nuclei of the newborn female mouse hypothalamus but are expressed constitutively in these regions of the male brain (Suzuki et al. 1998). In addition to the progranulin gene, several other somatic growth factors are expressed in brain neurons, including PDGF-A chain (Yeh et al. 1991), PDGF-B chain (Sasahara et al. 1991), EGF (Fallon et al. 1984), TGFα (Wilcox and Derynck 1988), IGF-I (Bondy et al. 1992), acidic FGF (Kresse et al. 1995), and basic FGF (Matsuyama et al. 1992). The functional significance of somatic growth factor expression in CNS neurons, including that of progranulin, is not well understood. The highly localized neuronal expression of the progranulin gene reported here, and its inducibility in response to hormones (Suzuki et al. 1998), demonstrate stringent neuron-specific gene regulation. This, together with the progranulin's ability to promote the growth of PC-12 cells, implies possible regulatory functions within the nervous system.

In conclusion, progranulin gene expression is closely associated with epithelial cells in somatic tissues. Progranulin gene expression is prominent in rapidly self-renewing epithelia, such as the skin and gut. In the gut, the expression is greatest in the highly proliferative cells of the deep crypts and becomes negligible in the terminally differentiated cells higher in the villus. Chronically nonproliferating highly differentiated epithelia, such as the lung alveolae and kidney tubules, show much lower levels of progranulin expression, although immortalized or neoplastic cell lines from low progranulin-expressing tissues, such as the kidney (MDCK) or the lung (A549), express the gene highly and respond to the recombinant protein. This is consistent with results showing that overexpression of progranulin in some epithelial cell lines leads to a more proliferative phenotype (He and Bateman 1999). The gene is expressed in a cell-specific manner in both developing spermatocytes and in ovarian follicles, indicating that progranulin is likely to play an important role in reproductive physiology in both sexes. Unstimulated connective tissues, muscle, and endothelia show little or no expression of the gene in the healthy adult animal in vivo. The major exceptions are the endometrial mesenchyme and the smooth muscle of the myometrium. The immune system and the brain provide strong evidence for additional functions for progranulin and its products. In both cases, gene expression is regional, implying stringent mechanisms of gene regulation in these systems. The results presented here therefore provide evidence that progranulin gene expression is tightly regulated in vivo and plays a role in the biology at least of highly proliferative somatic epithelia, the male and female gonads, the immune system, and nerve cells in several regions of the brain.