Abstract
Basophil invasion, i.e., invasion of basophilic corticotrophs from the residual intermediate lobe into the posterior lobe of the human pituitary gland, is believed to be a physiological phenomenon. This study evaluated the distribution of CPE, CPD, CPZ, α-MSH, ACTH, and Ki-67 immunoreactivity between human anterior pituitary and basophil invasion of the neurohypophysis. Mild to moderate immunoreactivities for CPE and CPZ were distributed relatively uniformly in the majority of the anterior pituitary cells and basophil invasion. In contrast, only corticotrophs exhibited intense CPD immunoreactivity. Basophil invasion showed similar immunoreactivities for α-MSH, ACTH, CPE, and CPZ as corticotrophs in the anterior pituitary, except for CPD, which was detected much less frequently. In the posterior lobe, CPE, CPD, and CPZ were present within the Herring bodies. Although no MIB-1 immunoreactivity was identified in anterior pituitary cells, limited MIB-1 labeling was detected in basophil invasion in five of ten cases. Highly selective expression of CPD in corticotrophs suggests that CPD plays a particularly important role in prohormone (POMC) processing in corticotrophs, with minimal or no significant roles in non-corticotrophs. Evidence that corticotrophs in basophil invasion are undergoing proliferation and are also phenotypically different from their counterpart in the anterior pituitary has further raised the possibility of some neoplastic potential.
Keywords
Many neuropeptides, peptide hormones, and neurotransmitters in a variety of species are commonly synthesized as larger precursors that are further processed post-translationally and targeted to either constitutive or regulated pathways (Mains et al. 1990). Prohormones are cleaved by endoproteases or endopeptidases, such as prohormone convertase 1 (PC1), prohormone 2 (PC2), or furin, at basic amino acid-containing sites. Removal of basic residues by carboxypeptidases generates active peptides. Carboxypeptidase E (CPE) was initially believed to be the only intracellular carboxypeptidase responsible for that role.
CPE is involved in the biosynthesis of most neurotransmitters and peptide hormones. It is localized to secretory vesicles in cell processes and is the only carboxypeptidase detected in mature secretory vesicles. It cleaves arginine or lysine residues from the carboxyl terminus of peptides with no endopeptidase activity (Fricker and Snyder 1982, 1983). High levels of CPE protein and mRNA have been detected in CNS, including pituitary gland (Lynch et al. 1990; MacCumber et al. 1990; Manser et al. 1990). More recently, the finding that Cpe(fat)/Cpe(fat) mice, which lack CPE activity, still exhibit limited neuroendocrine peptide processing prompted a search for other members of the CPE gene family and led to the discovery of carboxypeptidase D (CPD), carboxypeptidase Z (CPZ), and several others (Naggert et al. 1995; Fricker et al. 1996).
CPD, a novel carboxypeptidase with CPE-like activity, has been detected in a variety of tissues, with highest levels in pituitary, brain and, adrenals (Song and Fricker 1995, 1996; Tan et al. 1997; Xin et al. 1997; Ishikawa et al. 1998). It is a predominantly membrane-bound trans-Golgi network (TGN) protein processing enzyme that transits the secretory pathway within TGN and immature vesicles (Varlamov and Fricker 1998; Varlamov et al. 1999a,b). It is believed that CPD carries out redundant processing functions of CPE under certain circumstances (Dong et al. 1999). Studies have shown much higher levels of CPD than CPE in human placenta, suggesting that CPD plays a more important role than CPE in prohormone processing in placenta (Reznik et al. 1998).
CPZ, another novel carboxypeptidase, was cloned in human in 1997 and in rat in 1998. Its mRNA has been detected in placenta, brain, lung, thymus, and kidney by Northern blotting analysis and localized in leptomeningeal cells by in situ hybridization. CPZ cleaves substrates with C-terminal arginine residues. It is proposed that CPZ may play a select role in the processing of extracellular peptides or proteins (Song and Fricker 1997; Xin et al. 1998).
A regional differential expression of CPE has been described in rat pituitary, in which a higher level of CPE was observed in the intermediate lobe than in the anterior lobe (Lynch et al. 1990). However, no detailed descriptions of regional and cellular localization of CPs have been documented in human pituitary, particularly if there is any differential expression between the anterior pituitary and basophil invasion.
The intermediate lobe of adult human pituitary gland is vestigial and poorly defined. It consists of colloid-like follicles and a thin layer of mainly basophilic cells interposed between the posterior and anterior lobes. In contrast, however, the intermediate lobe in other mammals, such as rats, is well formed, and cells in the intermediate lobe are reactive for α-MSH in addition to ACTH, whereas anterior lobe cells are non-immunoreactive to α-MSH (Dube et al. 1978; Coates et al. 1986). Basophil invasion. an enigmatic phenomenon in which ACTH-immunoreactive cells in the residual intermediate lobe extend into the posterior lobe, has been observed in many normal individuals. It occurs as early as fetal life and increases in intensity and frequency with age. Kuebber et al (1990) have shown the presence of occasional cellular atypia and mitotic activity in basophil invasion, suggesting that basophil invasion may be responsible for extremely rare pituitary adenomas in the posterior lobe.
Our preliminary experiments revealed differential expression of CPD between corticotrophs in the anterior pituitary and basophil invasion, with lower levels of CPD in basophil invasion. In this study, to determine phenotypic differences between corticotrophs in the anterior pituitary and basophil invasion, we localized and compared expression patterns of three carboxypeptidases (CPD, CPE, and CPZ), α-MSH, and ACTH. To evaluate the possible neoplastic potential of basophil invasion, we also evaluated these tissues with a monoclonal antibody (MIB-1) against Ki-67, a nuclear protein expressed in G1, S, G2, and mitosis of the cell cycle.
Materials and Methods
Human Tissue
Fifteen randomly selected formalin-fixed, paraffin-embedded pituitary glands were retrieved from archival autopsy material from 1997 to 1999 at the Vanderbilt University Medical Center. They were divided into three groups of five on the basis of the degree of basophil invasion, as described by Kuebber et al. with minor modifications: (a) without basophil invasion; (b) with mild basophil invasion; or (c) with marked basophil invasion (Kuebber et al. 1990). Ages of patients ranged from 35 to 84 years. Review of the medical records revealed no significant clinical endocrinopathies. The postmortem intervals ranged from 5 to 25 hr. Six-μm-thick paraffin sections were stained for routine H&E and evaluated immunohistochemically for CPE, CPD, CPZ, ACTH, α-MSH, and MIB-1.
Antibodies
Antibodies against CPE, CPD, and CPZ were gifts from Dr. Lloyd Fricker (Albert Einstein College of Medicine; Bronx, NY). Two rabbit polyclonal antisera, one raised against the C-terminal and the other raised against the N-terminal portion of the CPE and CPD enzymes, were initially used. Because they produced similar results, the anti-C-terminal antibody to CPD and anti-N-terminal antibody to CPE were used for this study. A single rabbit polyclonal antibody raised against CPZ was also employed. The specificity of these antisera has been characterized previously by Western blotting analysis (Varlamov et al. 1996; Song and Fricker 1996; Xin et al. 1997, 1998) and immunohistochemistry (Reznik et al. 1998). Polyclonal antisera against ACTH, α-MSH and Ki-67 (MIB-1) were purchased commercially (see Table 1).
Immunohistochemistry
Immunostaining was performed on a Ventana ES automated immunostainer (Ventana; Tucson, AZ). Optimal conditions were determined separately for each antibody. After trials of different pretreatments for each antibody, the most appropriate conditions for each antibody were chosen. The antibody sources, dilutions and pretreatment protocols are listed in Table 1. Epitope retrieval was achieved using either heat-induced epitope retrieval (HIER) or protease treatment. For HIER, deparaffinized slides were incubated in 1 × citrate buffer (BioGenex; San Ramon, CA) for 25 min in a preheated steamer and then cooled in same buffer for 15 min at room temperature. Protease treatment was performed by incubating slides in Protease I-Ventana solution at ±40C for 4 min on a Ventana ES automated immunostainer. Protease treatment was used for the double staining of CPD and ACTH. The antigen-antibody complexes were visualized using an avidin-biotin complex detection system with 3,3′-diaminobenzidine as the chromogen. No significant immunoreactivity for each antibody was observed when the primary antibodies were substituted with non-immune rabbit serum or when the primary antibodies were omitted. For analysis of immunoreactivity, only moderate or intense immunostaining of more than 50% cells was accepted as positive.
Antibodies and experimental conditions a
aHIER, heat-induced epitope retrieval; Protease, protease pretreatment.
Sequential double labeling using horseradish peroxidase for CPD (orange) and alkaline phosphatase for ACTH (red) was performed to determine co-localization of ACTH and CPD. The immunostaining for CPD was performed on a Ventana ES automated immunostainer followed by manual immunohistochemistry for ACTH. The reverse orders of enzymes, chromogens, and staining sequences were also tried, and similar immunostaining patterns were obtained.
Serial sections were also immunostained for CPE or CPD with adjacent sections immunolabeled for anterior pituitary hormones by following the above-mentioned single-labeling protocols to determine cell types that express CPD and CPE.
Results
Expression of CPs
Only select anterior pituitary cells (10–20%), either scattered or in clusters, were strongly immunoreactive for CPD (Figures 1A and 1B). Most labeled cells were found in clusters in the central mucoid wedge corresponding to the distribution pattern of corticotrophs; some were also in the lateral wings (Figure 1H). Some epithelial cells of vestigial follicular structures in the intermediate lobe also exhibited CPD immunoreactivity. Almost all CPD-positive cells were basophils. The majority of them also contained one or multiple clear cytoplasmic vacuoles or the “enigmatic body,” a characteristic feature of corticotrophs (Figure 1B). In contrast, mild to moderate diffuse intracytoplasmic immunoreactivities for CPE (Figure 1C) and CPZ (Figure 1D) were detected in the majority of the anterior pituitary cells. Their staining intensities were relatively uniform among labeled cells, with scattered cells stained slightly more strongly than others for CPE (Figure 1C). No significant immunoreactivities for the three CPs were observed in endothelial and folliculostellate cells (Figures 1A-1D). In the posterior lobe, mild to moderate immunoreactivities for all CPs were present in the Herring bodies (Figure 1E).
There were no significant differences in staining pattern and intensity for CPE and CPZ between cells in the anterior pituitary and basophil invasion. However, CPD-immunoreactive cells were absent in basophil invasion in all five cases with mild basophil invasion and in three of five cases with marked basophil invasion. In the remaining two cases with marked basophil invasion, a few cells, either in clusters or in a scattered pattern, exhibited CPD immunoreactivity, but the majority of cells were negative (Figure 1F).
Localization of CPD
Both sequential dual immunohistochemistry and serial sections immunolabeled for ACTH (red) or CPD (orange) were employed to determine the cell types that selectively express CPD. When serial sections immunostained for ACTH (Figure 1G) or CPD (Figure 1H) were compared and matched, an almost identical distribution pattern for ACTH- and CPD-positive cells was observed although, at high magnification, it was apparent that slightly more cells labeled for ACTH (Figures 1G and 1H). Most labeled cells were located in the mucoid wedge, with scattered cells at the lateral wings. Sequential double labeling for CPD (orange) and ACTH (red) revealed co-localization of ACTH and CPD, with very rare cells labeled for CPD or ACTH alone. Almost all cells that were strongly labeled with CPD were also immunoreactive for ACTH. CPD stains usually exhibited a perinuclear pattern, whereas the ACTH stains were commonly peripheral surrounding CPD labeling. Occasional cells exhibited reverse staining patterns. Most cells that were immunoreactive for both CPD and ACTH also contained enigmatic bodies and were largely located at the mucoid wedge. A few cells were labeled with CPD or ACTH alone (Figures 1J and 1K). Co-localization of CPD and ACTH was also observed in the intermediate lobe and basophil invasion, although the majority of cells in basophil invasion exhibited ACTH labeling alone. Similar staining patterns were observed when the staining sequences and colors for ACTH and CPD were reversed (not shown).
(
Localization of CPE
Although the majority of cells in the anterior pituitary cells expressed relatively low levels of CPE, scattered cells (approximately 10%) were stained more strongly than others (Figure 1C). To determine cell types that expressed higher levels of CPE, serial sections immunostained for CPE and six anterior pituitary hormones were compared. Some co-localizations were noted between CPE- and LH-labeled cells (Figure 1I), with scattered cells immunoreactive for both CPE and FSH. No significant co-localizations between CPE and GH, ACTH, PRL, or TSH were detected.
Immunoreactivities for Ki-67 in Basophil Invasion
Using human tonsil and pituitary adenoma tissue as positive controls, we evaluated Ki-67 expression with an MIB-1 monoclonal antibody in pituitaries with or without basophil invasion. Although scattered cells with MIB-1 immunoreactivity were observed in the anterior pituitary, they were readily recognized as white blood cells on the basis of their morphology, including small size and often multilobulated nuclei and intravascular location. No significant MIB-1 immunoreactivity (<1/10 high-power fields) was detected in the anterior pituitary cells. Focal MIB-1 labeling in basophil invasion was observed in two of five cases with mild basophil invasion and in three of five cases with marked basophil invasion (Figure 1L). Rare MIB-1 labeling was noted in the epithelium of follicular structures in the intermediate lobe (Figure 1L, inset). However, no cytological atypia or mitoses were observed in basophil invasion, even in cases with marked basophil invasion and focal MIB-1 labeling.
ACTH and α-MSH Expression in Corticotrophs
Some cells in the anterior lobe, most cells in the intermediate lobe, and almost all cells in basophil invasion were immunoreactive for both ACTH and α-MSH, with variable staining intensities observed. There was no significant difference in intensities for ACTH and α-MSH between corticotrophs in the anterior pituitary and basophil invasion.
Discussion
When first isolated and cloned, CPE was believed to be the only carboxypeptidase responsible for removal of C-terminal basic residues in prohormone processing. More recently, however, several novel members of the carboxypeptidase gene family, including CPD, CPZ, and others, have been isolated, suggesting that CPE may be the major but not the only intracellular processing enzyme. Other enzymes, such as CPD, may also contribute significantly in prohormone processing, especially when CPE activity is deficient. In this study we demonstrated a highly selective expression of CPD in human corticotrophs and relatively uniform expression of CPZ in the majority of anterior pituitary cells, supporting this notion.
Diffuse CPE immunoreactivity in the majority of anterior pituitary cells compared to selective expression of CPD in corticotrophs would explain recent observations by Song and Fricker (1995) of a much greater abundance of CPE than CPD in the rat pituitary by Western blotting analysis. The finding of higher levels of CPE in gonadotrophs relative to other CPE-labeled anterior pituitary cells is essentially consistent with studies by Lynch et al. (1990). In their studies, immunoreactive CPE in rat anterior pituitary was localized in select cells that showed ultrastructural features of gonadotrophs. However, high expression of CPE in melanotrophs in the intermediate lobe is not observed in the present study. This may be related to a much less developed and vestigial intermediate lobe in the human pituitary.
Highly selective expression of CPD by corticotrophs in the anterior pituitary is consistent with earlier findings. Using Northern blotting analysis and in situ hybridization Xin et al. (1997) have demonstrated higher levels of mRNA for CPD in the intermediate lobe than the anterior lobe, with scattered labeling in the anterior lobe in rats. Although not pointed out by the authors, their findings suggest that CPD mRNA was selectively expressed by some anterior pituitary cells and by melanotrophs in the intermediate lobe. The intermediate lobes in rats are well developed and contain exclusively a special type of corticotroph or melanotroph. POMC processing in melanotrophs is more extensive than that in corticotrophs in the anterior lobe, with more small peptides including α-MSH, β-MSH, and LP generated (Bertagna 1994). Therefore, one would expect to see much higher levels of CPD in the intermediate lobe than in the anterior lobe. In contrast, corticotrophs in the anterior pituitary are scattered, with less POMC processing than their counterparts in the intermediate lobe. The scattered labeling of CPD mRNA in the anterior lobe demonstrated in that study may represent selective expression of CPD mRNA in corticotrophs. In the present study, co-localization of CPD and ACTH was demonstrated by sequential dual immunolabeling and serial section techniques. However, the regional selective expression of CPD in the intermediate lobe, as described by Xin et al. (1997), was not observed in this study, which, again, may be related to a much less developed and vestigial intermediate lobe in human pituitary.
Differential expression of CPD and CPE among different tissues has been demonstrated by Reznik et al. (1998) in the placenta, where a much higher level expression of CPD than CPE was detected, suggesting a more important role for CPD than CPE in prohormone processing in placenta. To the best of our knowledge, this is the first report demonstrating selective expression of CPD by a particular cell type in an individual tissue. Therefore, among all carboxypeptidases, CPD may play a particularly important role in POMC processing in corticotrophs without a significant role in prohormone processing in non-corticotrophs. CPE and CPZ are probably more critical in prohormone processing in non-corticotrophs, and CPE may play a role in prohormone processing in gonadotrophs. A reduced level of CPD in basophil invasion suggests low prohormone processing activity in the corticotrophs in basophil invasion.
Basophil invasion is a phenomenon that describes corticotrophs extending or migrating from the intermediate lobe into the posterior lobe of the pituitary. Its functional significance is still not known, although it is believed to be a physiological phenomenon. The hormonal activity of basophil invasion is not known, and no recognizable endocrine abnormalities have been associated. Cells in basophil invasion are believed to be less responsive to the negative feedback of glucocorticoids because they do not accumulate keratin filaments or Crooke's hyaline change in response to glucocorticoid excess. It has also been suggested that cleavage of the POMC molecule may be different in this cell population (Asa 1998). In studies by Kuebber et al. (1990), basophil invasion was found in approximately 62% of routine autopsy series. The invasion was observed as early as fetal life and increased in intensity and frequency with age, without significant sex differences. Because of the presence of occasional cytological atypia and mitotic activity in basophil invasion, they postulated that those cells may be responsible for extremely rare pituitary adenomas in the posterior lobe (Kuebber et al. 1990; Hori et al. 1999).
In this study, basophil invasion showed similar immunoreactivity for CPE and CPZ as cells in the anterior and intermediate lobes, but much lower levels of CPD immunolabeling in basophil invasion. This reduction may be related to a reduced level of POMC processing activity in corticotrophs of basophil invasion. Studies have shown that POMC processing of corticotrophs in the intermediate lobe is different from that of corticotrophs in the anterior lobe in rats. Both PC1 and PC2 were shown to be present in prohormone processing in the intermediate lobe as opposed to PC1 alone in the anterior lobe (Bertagna 1994).
Like CPE and CPZ, immunoreactivities of ACTH and α-MSH in basophil invasion are not significantly different from those in the anterior pituitary. These phenotypic similarities between corticotrophs in the anterior lobe and basophil invasion raise the possibility that at least some cells in basophil invasion are derived from direct extension of cells in the intermediate lobe and may represent a physiological phenomenon. If that were true, one would expect to see a significant reduction of the residual intermediate lobe mass in cases with marked basophil invasion. However, no evidence of reduction of the intermediate mass has been observed, even in cases in which almost the entire posterior lobe is replaced by basophil invasion. Therefore, it is reasonable to suspect that another source(s) may also contribute to basophil invasion. The detection of MIB-1 labeling in more than half of the cases of basophil invasion, including occasional labeling in the intermediate lobe, raises the possibility that some cells in basophil invasion and the intermediate lobe are actively proliferating. Our observations are consistent with observations by Kuebber et al. (1990), who found that approximately 3% of cases with basophil invasion showed a combination of mitoses, multinucleated cells, polymorphism, and hypertrophy of the cytoplasm, in addition to intense proliferation, in basophil invasion. They suggested that those changes might contribute to a better understanding of the origin of silent corticotrophic cell adenomas. The high incidence of MIB-1 labeling in the present study compared to the mitotic rate in basophil invasion described by Kuebber et al. (1990) undoubtedly reflects the high sensitivity of MIB labeling, which identifies cells in G1, G2, S, and M-phases rather than in M-phase alone. Our findings raise the possibility that, in addition to cell migration from the intermediate to the posterior lobe, at least some cells in basophil invasion are derived from dividing cells. Therefore, basophil invasion may represent a combination of physiological migration of cells from the intermediate lobe into the posterior lobe as well as proliferation of cells in the intermediate lobe and basophil invasion. It is conceivable that some neoplastic potential exists in basophil invasion, especially when the MIB-1 index is high.
In summary, the data presented here suggest that CPD, CPE, and CPZ are actively involved in the processing of at least some of the many pituitary hormones and peptides. The selective CPD expression by corticotrophs, with more uniform expression of CPE and CPZ by the majority of the anterior pituitary cells, suggests that CPD plays a particularly important role in peptide processing in the corticotrophs, whereas CPE and CPZ play a more important role in peptide processing in non-corticotrophs. Taken together, the differential cell expression and the previously reported difference in subcellular localization of CPD and CPE further suggest that CPE and CPD are probably involved in the processing of different neuroendocrine peptides and proteins in different pituitary cells. Reduction of CPD in corticotrophs involved with basophil invasion further indicates that corticotrophs at sites of basophil invasion are phenotypically different from their counterparts in the anterior pituitary. The presence of MIB-1 labeling in basophil invasion suggests the possible neoplastic potential of basophil invasion.
Footnotes
Acknowledgements
We would like to thank Dr Lloyd Fricker at Albert Einstein College of Medicine for kind gifts of CPE, CPD, and CPZ antibodies, Ms Rebecca Meitus and Ms Tracy Moss for technical assistance, and Mr Brent Weedman for photographic assistance.