Abstract
IGFBP-rP1/mac25 is a recently described member of the insulin-like growth factor binding protein (IGFBP) family. It has structural homology to the other members of the IGFBP family but has a lower affinity for insulin-like growth factors (IGFs). In previous studies using RNA blot hybridization, it was shown that the expression of IGFBP-rP1/mac25 was ubiquitous in normal human tissues. In this report we show by immunohistochemistry that the expression of IGFBP-rP1/mac25 is actually restricted to certain organs and specific cell types. We used an antibody raised against a decapeptide of the C-terminal part of the protein that recognizes a ≍37-kD protein under reduced conditions. The immunohistochemistry performed on normal human tissues showed a ubiquitous intense staining of peripheral nerves and a variable degree of positive staining in smooth muscle cells, including those from blood vessel walls, gut, bladder, and prostate. Cilia from the respiratory system, epididymis, and fallopian tube showed intense immunoreactivity. Most endothelial cells showed some positivity, whereas fat cells, plasma cells, and lymphocytes were negative. There was specific expression limited to certain cell types in the kidney, adrenal gland, and skeletal muscle, indicating a possible specialized function of IGFBP-rP1/mac25 in these organs. We further noted an opposite pattern of staining in the lining epithelium of breast (typically positive) and prostate glands (largely negative). The specific localization of IGFBP-rP1/mac25 as described implies a function of the protein. However, its regulation within the IGF axis or a possible direct action of IGFBP-rP1/mac25 remains to be demonstrated.
T
Four groups independently identified proteins that have been determined to be IGFBP-rP1/mac25. One of these groups cloned the mac25 cDNA from normal leptomeningial and mammary epithelial cells; its expression was found to be decreased in the corresponding tumor cells (Murphy et al. 1993; Swisshelm et al. 1995). The protein was shown to be able to bind IGFs (Oh et al. 1996). During the same period, two other proteins were purified that have been determined to be the same as the protein encoded by mac25. First, tumor adhesion factor (TAF) was isolated from diploid fibroblasts (Akaogi et al. 1996), and the prostacyclin-stimulating factor (PSF) was purified from human bladder carcinoma cells (Yamauchi et al. 1994). Finally, T1A12 was identified by subtractive cDNA cloning using RNAs from a normal breast epithelial cell line Hs578Bst and the breast cancer cell line Hs578T. The product of this cDNA was shown to bind IGFs with low affinity. A polyclonal antibody was raised against a decapeptide located within the C-terminal sequence of T1A12. Immunoreactivity was demonstrated in normal mammary epithelium but not in breast adenocarcinomas, leading to the hypothesis that IGFBP-rP1/mac25 might have a tumor suppressor-like function (Burger et al. 1998).
Using the above polyclonal antibody, we have analyzed the distribution of IGFBP-rP1/mac25 in normal human tissues to better characterize its cellular localization. Although IGFBP-rP1/mac25 mRNA has been found in almost all tissues by RNA blot hybridization (Oh et al. 1996), in this study we have demonstrated that the protein is not expressed ubiquitously and, furthermore, that there is variable quantity of expression among normal cells in these tissues. The characterization of the normal distribution of IGFBP-rP1/mac25 provides evidence that the protein has a highly restricted pattern of expression within a given tissue that allows the analysis of changes in the protein's expression in corresponding neoplasms.
Materials and Methods
Cell Lines and Culture
Cell lines were purchased from ATCC (Rockville, MD). Hs578T was cultured in DMEM (GIBCO; Grand Island, NY) supplemented with 10% FBS and 10 μg/ml insulin. MCF-7 cells were cultured in DMEM supplemented with 10% FBS.
Western Blotting
Conditioned media (CM) of MCF-7 and Hs578T were obtained by treating 80% confluent cells in DMEM/F12 without phenol red or serum, and were collected after 48 hr. Ultrafiltration was performed using Centriprep 3 devices (Amicon; Beverly, MA) to obtain an 18.75-fold concentrated CM (concentration followed by desalting in 10 mM Tris, pH 7.4). Protein content was measured by a Bradford colorimetric assay (BioRad; Hercules, CA). Forty μg of protein was loaded under reducing conditions onto a 15% SDS-PAGE gel and transferred to nitrocellulose (MSI; Westborough, MA). The membranes were blocked with 5% nonfat dry milk in TBST (20 mM Tris, 150 mM NaCl, 0.05% Tween-20) overnight at 4C. The membranes were rinsed four times in TBST and incubated in a 1:1500 dilution of primary antibody directed against IGFBP-rP1/mac25 diluted in 5% nonfat dry skim milk in TBST for 1 hr at room temperature (RT).
To demonstrate specificity, the rabbit IgG affinity-purified primary antibody (Burger et al. 1998) was preabsorbed with the C-terminal decapeptide LSKEDAGEYE that was used to raise the antibody, for 1 hr at RT (antibody:peptide molar: molar ratio of 1:4). Additional controls were performed with the same conditions using peptide only or primary antibody only. After incubation, the membranes were rinsed five times in TBST and incubated in the same buffer as before, containing a 1:1500 dilution of a donkey anti-rabbit IgG linked to horseradish peroxidase (Amersham; Arlington Heights, IL) for 1 hr at RT. The membranes were rinsed seven times in TBST. Detection of immunoreactive protein was performed by chemiluminescence (ECL kit; Amersham).
Immunohistochemistry
Zinc-formalin-fixed, paraffin-embedded tissues of normal human specimens obtained from surgical pathology samples were sectioned at 5-μm thickness, deparaffinized in xylene, rehydrated in a graded ethanol series, rinsed briefly in PBS, and incubated for 15 min at RT in 10% normal goat serum (Jackson Immunoresearch Laboratories; West Grove, PA). This was followed by treatment with the avidin-biotin blocking kit (Vector; Burlingame CA). After a brief rinse, the sections were incubated overnight at 4C with the T1A12 antibody (1:1500 dilution in 5% normal goat serum, 1% BSA, 0.5% FSG in PBS). After several washes in PBS, the slides were incubated for 30 min at RT with a 1:100 dilution of a biotinylated goat anti-rabbit IgG (Multilink; BioGenex, San Ramon, CA). Endogenous peroxidase was quenched by treating the tissues for 30 min in 0.3% hydrogen peroxide. Slides were then incubated in streptavidin-peroxidase diluted 1:60 for 30 min at RT (BioGenex). Peroxidase activity was visualized by a 5-min incubation in diaminobenzidine and hydrogen peroxide (0.01%). The sections were rinsed in water, counterstained with hematoxylin, dehydrated, and mounted.
For the blocking experiments, the primary antibody was first incubated for 1 hr at RT with the peptide (antibody:peptide molar:molar ratio of 1:4). The staining procedure was then performed as described above.
Results
Antibody Characterization
By Western blotting, the polyclonal antibody raised against the decapeptide was able to recognize a single band in Hs578T CM having an apparent Mr of ≍37 kD under reduced conditions that was not present in MCF-7 CM, in accordance with what was described previously. This band was competed out when the primary antibody was preabsorbed with the decapeptide (molar ratio antibody:peptide of 1:4) (Figure 1C).
For immunohistochemistry experiments, the primary antibody was preabsorbed following the same procedure with the decapeptide (1:4 ratio) and used to stain a normal breast specimen. The immunoreactive signal observed in epithelium, smooth muscle cells, and nerves (Figure 1A) was lost when the antibody was preabsorbed with the decapeptide (Figure 1B).
IGFBP-rP1/mac25 Immunostaining
All peripheral nerves stained strongly (Figure 2), but the cytoplasm of ganglion cells was variably positive (some cells were moderately positive and others were negative). Some of the supporting cells in peripheral nerves showed nuclear positivity in addition to intense cytoplasmic immunoreactivity. Virtually all smooth muscle cells from assorted sites, such as blood vessel walls, gut, and bladder (Figure 2J), showed moderate positivity, whereas prostate smooth muscle stroma was less intense. Cilia from all sites, including respiratory system, fallopian tube (Figures 2B and 2C), and epididymis (not shown), showed intense staining. Most endothelial cells were positive, with the exception of those in the brain. Fat cells, plasma cells, and lymphocytes (with rare exceptions) lacked staining.
The results of the positive immunostaining are given in Table 1. Representative examples are shown in Figure 2. Endometrium (Figure 2A) displayed prominent cytoplasmic staining of both proliferative and secretory phase epithelium. The stromal cells were largely negative. In the fallopian tube there was moderate cytoplasmic immunoreactivity of the epithelial lining and of the smooth muscle cells in the wall (Figure 2B). Respiratory bronchial epithelium (Figure 2C) was positive but the alveolar lining cells and pigmented macrophages were negative. In the kidney (Figure 2D), the glomeruli were largely negative, and the epithelia of distal tubules stained stronger than those of proximal tubules. The cells of the adrenal medulla showed some variable reactivity; the cells of the zona reticularis and the zona glomerulosa showed more intense staining than those of the zona fasciculata (Figure 2E). Skeletal muscle (Figure 2F) displayed discrete positivity in some fiber bundles, whereas others were negative. An opposite pattern of immunoreactivity was observed in breast epithelium (Figure 2G) compared to prostate epithelium (Figure 2H). In both organs there was diffuse cytoplasmic staining of smooth muscle cells and strong staining of nerves. The lining epithelium of breast lobules and ducts was immunoreactive, whereas the secretory cells of the prostate were chiefly negative. In the central nervous system, staining was observed in the neuropil with positivity of astrocytes and oligodendrocytes and negativity of neurons (Figure 2I). Immunoreactivity was intense in smooth muscle cells and transitional epithelium of the urinary bladder (Figure 2J). In contrast to the skeletal muscle fiber bundles, there was no mosaic pattern in the bladder smooth muscle tissue. Finally, epithelial cells of the gastrointestinal tract and hepatobiliary system showed weak to moderate immunoreactivity.
Specificity of T1A12 antibody. Immunohistochemistry was performed on two consecutive slides of a normal human breast specimen using T1A12 antibody (
Discussion
In this study, we analyzed systematically by immunohistochemistry the distribution of IGFBP-rP1/mac25 in normal human tissue specimens. We used a polyclonal antibody directed against the product of T1A12, a gene identified by a subtractive cDNA cloning strategy using RNAs from the normal breast cell line Hs578Bst and the tumor cell line Hs578T (Burger et al. 1998).
Mac25 has been shown to be expressed in all human tissues using a cDNA probe and RNA blot analysis (Oh et al. 1996). By immunohistochemistry, our data indicate that IGFBP-rP1/mac25 localization was not universal; rather, its distribution in each of the organs was restricted to certain cell types. Immunostaining was absent or weak in several tissues, including the ovary, placenta, and prostate secretory epithelium, despite high levels of mRNA (Oh et al. 1996). Staining of peripheral nerves in all tissue specimens was always intense, thereby serving as a suitable internal positive control for immunoreactivity. Furthermore, the high level of IGFBP-rP1/mac25 production in nerves and often in muscle is a likely explanation for the ubiquitous mRNA expression found in human tissue samples by Northern blotting (Oh et al. 1996). In addition, mac25 was cloned from leptomeningial cells (Murphy et al. 1993), and in this study we observed strong staining of glial cells in brain (astrocytes and oligodendrocytes) and spinal cord. A previous report suggested that IGF-I is involved in an autocrine proliferation loop in the growth of astrocytes (Han et al. 1992b). Saneto et al. (1988) showed that IGF-I also regulates the differentiated function of oligodendrocytes by increasing the synthesis of myelin basic protein. As shown for the other IGFBPs in the CNS, the immunostaining of IGFs is co-localized with staining for IGFBPs, suggesting a mechanism to locally increase the amount and bioavailability of IGFs (D'Ercole et al. 1996).
Immunohistochemical detection of IGFBP-rP1 in normal human tissues
aSmooth muscle cells and nerves are positive at all sites. Most endothelial cells are positive, with the exception of those in brain and spinal cord.
Some organs were noted to have compartmentalized staining for IGFBP-rP1/mac25. In the adrenal gland, staining was more intense in the cortex (glomerulosa ≍ reticularis > fasciculata) than in the medulla, which displayed variable reactivity. Han et al. (1992a) showed that IGF-II gene expression was localized in the steroidogenic cells of the developing adrenal gland of the ovine fetus, suggesting that IGFs may have an autocrine/paracrine regulatory role in the development of steroidogenic cells. It has also been shown in the adrenal that IGFBP-1 and −3 expression is lost in adult vs fetal tissues (Ilvesmäki et al. 1993).
Representative examples of IGFBP-rP-1 immunostaining in normal tissues. Smooth muscle cells and peripheral nerves (arrowheads,
The staining intensity of IGFBP-rP1/mac25 in the renal tubule epithelium was variable and was consistent with the fact that different parts of the nephron express different IGFBPs (Feld and Hirschberg 1996). IGFBP-rP1/mac25 was localized to a greater degree in the distal than in the proximal tubules. Because the intensity of IGFBP immunoreactivity in the renal system is variable, this may indicate that IGFBP-rP1/mac25 plays a particular role in renal physiology.
The staining of some groups of skeletal muscle fibers was positive. whereas others were negative. This could relate to differences in fast- vs slow-twitch muscle fiber bundles, although additional muscle immunohistochemistry would need to be performed. A study performed on rabbit slow and fast skeletal muscle myoblasts showed that there were no constitutive differences in the components of the IGF system between these two fiber types (Barjot et al. 1996); however, IGFBP-rP1/mac25 was not examined in this particular study. IGFBPs, although not linked to fiber physiology, clearly play a role in myogenesis. IGFBP-4 is involved in myoblast proliferation, and IGFBP-5 is important in differentiation (Florini et al. 1996). IGFBP-rP1/mac25 is expressed more by myoblasts than by differentiated myotubes (Damon et al. 1997), and its expression in myoblasts partially inhibits differentiation (Haugk et al. 2000). However, the significance of differential IGFBP-rP1/mac25 expression in adult skeletal muscle fiber bundles that are nonproliferative is unclear and warrants additional investigation.
In the digestive system, the smooth muscle cells and nerves stained positively. There was some staining of the surface epithelium but not of the mucous glands. There exist only a few studies of IGFBP expression in adult digestive tract. Caco-2 intestinal epithelial cells change their IGFBP secretion profile with their differentiation status (Oguchi et al. 1994). During fetal development, the mucosal epithelia of the stomach and intestine are stained positively for IGFBP-1 and −3 (Braulke et al. 1996). The significance of these findings is unclear at present.
In addition to the intense staining of cilia, the respiratory bronchial epithelium was uniformly positive. The IGF axis has been believed to play a major role in the development of the rat lung because all IGFBPs show a temporally specific pattern of expression (Retsch-Bogart et al. 1996). Although IGFBP-rP1/mac25 is present in the adult bronchi, no data are available on its role in lung development and differentiation.
In endocrine glands, regulation of the IGF axis by other peptide hormones has been shown. The increase of IGFBP expression in the thyroid gland is accompanied by a decrease in thyroid function (Eggo et al. 1996). In the pancreas, there was a specific temporal expression of IGFBPs during fetal life and after birth (Hogg et al. 1994). We found that the islet cells were positively stained for IGFBP-rP1/mac25 in the adult pancreas. Whether or not IGFBP-rP1 is involved in insulin resistance in different pathological conditions remains to be determined.
Intense staining was observed in cilia of fallopian tubes, cervix, endometrium, bronchus, and epididymis. This implies an association of IGFBP-rP1/mac25 with motile structures, because sperm tails also stained intensely. It is possible that IGFBP-rP1/mac25 is critical in motility of cell structures either by regulating the availability of IGFs or by some IGF-independent mechanism.
The data concerning the role of IGFBPs in the female reproductive tract are complex. In the fallopian tube, IGFBP-1 is the predominant binding protein exceeding that of IGFBP-4, −3, and −2 (Pfeifer and Chegini 1994). An investigation of the IGF axis in serum from women throughout their menstrual cycle revealed no modification of the circulating levels of IGF-I, -II, IGFBP-1, and IGFBP-3, suggesting that these factors are unlikely to play an endocrine role in cyclic ovarian follicle development (Van Dessel et al. 1996). On the other hand, an analysis of the follicular fluid supports an autocrine or paracrine role for the IGF system in the local regulation of ovarian function (Van Dessel et al. 1996). We found no IGFBP-rP1/mac25 immunoreactivity in the ovary except in blood vessels, suggesting that IGFBP-rP1/mac25 does not play a major role in ovarian physiology. There is clearly menstrual cycle-dependent expression of the IGF axis in the endometrium, in which IGF-I and IGFBP-5 predominate in the proliferative phase, whereas IGF-II and the other IGFBPs are predominant in the secretory phase (Zhou et al. 1994). We found that IGFBP-rP1/mac25 was present in the secretory phase as well as in the proliferative glandular epithelium, whereas in both cases the stromal cells were largely negative. IGFBP-1 has been found to be produced by decidualized stromal cells and may function to limit the effects of trophoblast-derived IGF-II, playing a crucial role in the regulation of trophoblast implantation (Giudice 1997). In the placenta, in situ hybridization results show that IGFBP-2 may be an important regulator of IGF-II action (Zhou and Bondy 1992). We found that IGFBP-rP1/mac25 is present at a low level in amnionic and chorionic cells, whereas trophoblasts were negative. Whether IGFBP-rP1/mac25 participates in the availability of IGFs during fetal development is unknown.
In the male genital tract, in contrast to other IGFBPs (−2, −3, −4, and −5) (Lin et al. 1993; Zhou and Bondy 1993), IGFBP-rP1/mac25 was not present in Leydig cells. On the other hand, Sertoli cells stained positively for IGFBP-rP1/mac25. Analogous to the staining of glial cells, these data suggest a paracrine role for IGFBP-rP1/mac25 in cells that support others.
In breast and prostate, the involvement of the IGF axis in normal and pathological growth has been demonstrated in several studies (for review see Cohen et al. 1994; Oh 1998; Rasmussen and Cullen 1998). Interestingly, the staining pattern of IGFBP-rP1/mac25 is opposite in the lining epithelium of breast and prostate. In both cases, smooth muscle cells and nerves were stained but the epithelium was immunopositive in the breast and was largely negative in prostate secretory cells. The significance of this result needs to be addressed by studying the differential regulation of IGFBP-rP1/mac25 by male and female hormones in vivo. Furthermore, the role of IGFBP-rP1/mac25 in the development of prostate adenocarcinoma compared to breast adenocarcinoma needs to be examined. Results from our laboratory suggest that IGFBP-rP1/mac25 is increased in prostate carcinoma (Degeorges et al. 1999), whereas breast cancer epithelium tends to lose IGFBP-rP1/mac25 staining (Burger et al. 1998).
Perhaps the function of IGFBP-rP is exclusive of IGF. Other members of the IGFBP-rP1 family are induced in response to serum or other growth factors, which places them in the immediate-early response gene category (O'Brien et al. 1990; Bradham et al. 1991). IGFBP-rP1 has the ability to stimulate prostacyclin production in vascular endothelial cells (Yamauchi et al. 1994), thereby controlling vascular permeability. Furthermore, IGFBP-rP1 accumulates in small blood vessels from tumor tissue (Akaogi et al. 1996), providing a mechanistic link to tumor survival by increasing vessel size and permeability. Recently, Girard et al. (1999) showed that IGFBP-rP1 is a marker of high endothelial venules in an area of cell junctions that might be involved in the control of lymphocyte emigration. Therefore, IGFBP-rPs may be directly involved in the regulation of cell growth, differentiation, or migration in an IGF-independent manner.
The role of IGFBP-rP1 in tumorigenesis is complex. In breast cancer, the loss of IGFBP-rP1 expression was correlated with LOH at chromosome 4q12-13 (Burger et al. 1998). The continued expression of IGFBP-rP1/mac25 was suggested to be indicative of a better prognosis. However, a longitudinal study of patient survival has not been performed to support this theory. In prostate cancer, genetic abnormalities involving chromosome 4 are rare (Dong et al. 1997) and do not occur in the LNCaP lineage (Hyytinen et al. 1997). Our immunostaining indicates that only rarely do normal prostate epithelial cells express IGFBP-rP1, and then only basal cells that are the proliferative compartment of the normal gland. Despite this, Sprenger et al. (1999) were able to suppress the growth of a tumorigenic SV40 lg T-antigen-transformed prostate cell line, M12, by transfecting the IGFBP-rP1 cDNA. It may well be that the expression of a particular IGFBP-rP in the context of the cellular tumor suppressor milieu may determine whether or not a given IGFBP-rP1 is a tumor suppressor or a growth factor. This is implied by the inverse correlation of novH: wt1 expression in Wilms' tumors (Martinerie et al. 1994,1996). There are other examples of dichotomous behavior of proteins given a particular cell background. The most notable example is TGF-β1, which can be either growth promoting, growth suppressive, or pro-apoptotic (Barrack 1997; Taipale et al. 1998; and references therein). Which effect is observed depends both on the tumor:stromal microenvironment and the TGF-β receptor status of the carcinoma.
Finally, despite the lack of evidence to date, IGFBP-rP1/mac25 might undergo proteolysis, as shown for IGFBP-3, for which proteolytic fragments can have IGF-independent actions through specific receptors (Oh et al. 1993; Leal et al. 1997). Obviously, many of the biological roles of IGFBP-rP1/mac25 remain to be elucidated, including an investigation of its proteolysis. We have determined the specific localization in normal human tissues, which provides a framework for further studies that examine its actions.
Footnotes
Acknowledgements
Supported in part by the Canadian Breast Cancer Research Initiative (AS).
We would like to thank John Sanders and Angelika Burger for fruitful discussions concerning immunohistochemistry. Figure plates were prepared in the University of Virginia Information Technology Center-Academic Computing Health Sciences.