| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Regular Articles |
From the Department of Pathology,*
Rush Medical College,
Chicago, Illinois; the German Cancer Research
Center,
Heidelberg, Germany; and the
Departamento de Especialidades
Medico-Quirurgicas,
Universidad del Pais
Vasco, Bilbao, Spain
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
As part of the investigation that led to the recognition that the molecule bound by monoclonal antibody C6 corresponded to Nup88, a polyclonal antiserum directed to the pertinent recombinant protein was generated. By immunohistochemistry, this antiserum was shown to recognize several human tumor cell lines as well as ovarian carcinomas in tissue sections; parallel results were obtained by immunoblot analysis.3 We now report that this antiserum immunostains richly and extensively conventional histological sections of 214 samples representing a wide spectrum of malignant tumors including carcinomas and sarcomas of diverse sites, lineages, and differentiation as well as some mesotheliomas, gliomas, melanomas, and lymphoreticular neoplasms. Certain fetal tissues were similarly stained. Notably, benign neoplasms and reparative processes showed only focal and less intense reactions whereas normal adult cells reacted only sporadically and weakly. Immunoblots of selected samples showed parallel results. Our findings indicate that this molecule most probably corresponds to Nup88. Its significant overexpression and its exceedingly wide distribution across a wide spectrum of cancers and precancerous dysplasias raise the possibility of using it as a broadly based generic histodiagnostic marker of malignant transformation.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Cases were selected on the basis of known diagnoses from the files
of the Rush-Presbyterian-St. Luke’s Medical Center, Chicago, and the
Hospital of the Faculty de Medicine, Bilbao, Spain. A total of 214
malignant tumors were examined; benign tumors, hyperplasias, and
related conditions were also included (see Table 1
). Most of these cases had been
extensively studied and characterized in previous
studies.9-13
Surgical procedures were performed with due
consent, and were based on widely accepted therapeutic and/or
diagnostic protocols. Autopsy samples from adult and fetal cases were
obtained from Rush-Presbyterian-St. Luke’s Medical Center; autopsies
were performed based on duly obtained consent. The anonymity of the
patients was protected in all cases.
|
Two paraffin blocks per case were selected; diagnoses were confirmed on conventional hematoxylin and eosin-stained sections by two independent observers (VEG and AO). All tissues had been fixed in formalin, conventionally processed, and embedded in paraffin. Sections for immunostaining were cut at 4 µm, set on coated slides, and placed on a warmer at 60°C for 1 hour; subsequently, they were deparaffinized in xylene and rehydrated in graded alcohols. No pretreatment with microwaves or enzymes before exposure to the primary antiserum was applied. Details on the preparation and characteristics of the antiserum have been recently published.3 Staining was accomplished by the avidin-biotin-peroxidase method as originally outlined by Hsu et al14 ; commercial reagents were used (DAKO, Carpinteria, CA). Best results were obtained when the antiserum was applied overnight in a humid chamber at 4°C at a concentration of 1/500; the diluent was that provided in a commercial kit (Ventana Medical Systems, Tucson, AZ). Binding sites were visualized with 3,3'-diaminobenzidine (Aldrich Chemicals, Danvers, MA); in the case of melanocytic lesions, alkaline phosphatase (DAKO) was used as chromogen. All sections were counterstained with hematoxylin to improve nuclear visualization. As negative controls, slides were similarly processed but omitting the primary antiserum. Staining intensity was rated as weak, moderate, or strong; in heterogeneous cases the rating of the predominant pattern prevailed. The extent of the reaction was defined by the percentage of reactive cells and graded as negative (0) to 1+ to 5+ as previously described.15
Immunoblots
For immunoblot analysis, freshly obtained samples were placed in vials containing precooled isopentane, and snap-frozen in liquid nitrogen. All samples were kept in a deep freezer at -80°C until used. Tumors were homogenized in 1% sodium dodecyl sulfate-0.14 mol/L ß-mercaptoethanol and then centrifuged over a Qiashredded (Qiagen, Germany) column to shear the DNA. The protein content of the samples was calculated with the DC protein assay system (Bio-Rad, Richmond, CA) and 15 µg of proteins were separated on 10% polyacrylamide gels. Proteins were electrotransferred to polyvinylidene difluoride membranes, blocked with nonfat milk in phosphate-buffered saline, and incubated with our polyclonal antiserum at a dilution of 1:500. After washing, the membranes were reacted with a horseradish peroxidase-labeled goat anti-rabbit, washed again, and the reacting bands were revealed with the Amersham enhanced chemoluminescence system (Amersham, Buckinghamshire, UK).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
All malignant neoplasms stained convincingly; most reactions were
moderate to strong and 3+ to 5+ in extent. In situ
carcinomas were invariably stained as were, albeit less so, dysplastic
lesions that did not reach the level of carcinoma in situ.
Benign tumors showed variably extensive and generally weak
immunoreactivity. Salient points are summarized in Table 1
.
Gastric carcinomas were extensively and strongly stained. The contrast
between the surrounding or overlying mostly negative mucosa and the
convincingly reactive tumor was evident (Figures 1, a and b)
. Reactions were similar in
tumors showing variably differentiated glands, solid clusters, or
linitis plastica single-cell pattern. One case showed intestinal
metaplasia with some dilated glands as well as in situ
carcinoma with the latter reacting strongly (Figure 1c)
. The
immunostaining was characteristically granular; the granules were
rather large as compared with the delicate dots seen in synaptophysin
reactions.16,17
In a minority of cells, a perinuclear
localization was noted but the distribution was predominantly
cytoplasmic (see below). The cell membrane did not stain.
|
. Two carcinomas of mixed exocrine and neuroendocrine
phenotypes and two pure neuroendocrine carcinomas were similarly
stained (Figure 1f)
. In polyps not associated with carcinoma, villous
adenomas stained stronger and more extensively than their tubular
counterparts; foci of intramucosal carcinoma stained convincingly.
Hepatocellular carcinomas were strongly reactive. One of these
developed in the background of advanced cirrhosis; in this case,
occasional cells in proliferating bile ductules in cirrhotic septa were
weakly stained (Figure 1g)
whereas the adjacent carcinoma stained
diffusely and strongly (Figure 1h)
, and dysplastic nodules reacted
moderately. Pancreatic carcinomas of variable glandular differentiation
showed consistent, moderate to strong reactions (Figure 1i)
; two
neuroendocrine carcinomas were similarly stained. Regardless of tumor
phenotype, nonneoplastic exocrine and neuroendocrine cells in the
vicinity of, or entrapped within, tumors were occasionally, focally
reactive.
All infiltrating breast carcinomas of ductal (Figure 1j)
or lobular
types reacted convincingly, whereas ducts nearby, or entrapped within
them, reacted weakly and sporadically or not at all. Mucinous (colloid)
carcinomas showed strong staining of the scanty malignant cells whereas
the dominant mucous pools did not react. In several carcinomas, the
strongest staining was noted in the peripheral, invasive edge of the
tumors. Fibrocystic changes in the vicinity of carcinomas including
cysts, ductal hyperplasia, adenosis, papillomas, and apocrine
metaplasia, showed weak to moderate, focal reactions, whereas similar
changes not associated with carcinomas ranged from focally reactive to
entirely negative (Figure 1k)
. Atypical ductal hyperplasia and all
variants of ductal carcinoma in situ stained convincingly
(Figure 1l)
, fibroadenomas reacted weakly and focally or not at all.
Pulmonary squamous, adeno-, bronchioloalveolar, large-cell, and
neuroendocrine carcinomas reacted richly and extensively (Figure 1, m and n)
; in well-differentiated squamous carcinomas, the peripheral
growing edge of the clusters reacted strongly whereas the central cells
approaching the squamous pearls were less reactive. Hyperplastic
bronchi in the vicinity or within tumors were often stained. Bronchial
carcinoids stained moderately and often extensively (Figure 1o)
.
Ovarian carcinomas including serous, mucinous, clear-cell, and
endometrioid types were convincingly stained; borderline tumors stained
moderately whereas their invasive (Figure 2a)
and particularly their high-grade
counterparts were strongly reactive. Benign cystadenomas showed uneven,
generally weak reactions. In a benign teratoma, the basal cells of the
skin and adnexa, the respiratory epithelium and cartilaginous cells
showed moderate staining. Endometrial carcinomas of all types reacted
strongly (Figure 2b)
. Several of these cases were associated with
endometrial hyperplasia; and, although the carcinomas stained
convincingly, the bland variants of hyperplasia did not (Figure 2c)
.
Also interesting was that the luminal aspect of several tumors stained
less intensely than the deep, invasive aspect. In several cases,
segments of fallopian tubes included in these samples showed focal,
moderate reactions, as did cells comprising small follicular cysts and
corpora lutea.
|
whereas nonneoplastic ducts and
acini entrapped within them stained weakly and sporadically or not at
all. In the case of poorly differentiated and hypernephroid carcinomas,
minute clusters or single reactive cells were readily detected in the
stroma and, around and/or within nerves (Figure 2e)
. Foci of low- and
high-grade prostatic intraepithelial neoplasia were consistently
stained, particularly the latter. Benign hyperplastic and atrophic
glands were either negative (Figure 2f)
or showed rare positive cells.
All renal carcinomas examined were of clear cell type and all reacted
strongly (Figure 2g)
; some reactive tubules in the vicinity stained as
well but weakly. Two adrenocortical adenomas showed weak but fairly
extensive reactions. Mesenchymal Tumors
Fibrosarcomas, malignant fibrous histiocytomas (Figure 2h)
, and
Kaposi sarcomas stained moderately to strongly, and a single
leiomyosarcoma reacted strongly in both primary and metastatic sites
(Figure 2i)
. Dermatofibrosarcoma protuberans, benign giant cell tumors,
atypical fibroxanthomas, and angiolipomas showed focal, moderate to
weak reactions. Two (uterine) leiomyomas stained rather extensively but
weakly.
Miscellaneous Tumors and Fetal Tissues
Diffuse large cell lymphomas and a lymphoblastic lymphoma stained
strongly and extensively (Figure 2j)
. In several cases of Hodgkin’s
disease, the Reed-Sternberg as well as the lacunar cells stained
convincingly whereas the associated, nonneoplastic leukocytes did not.
Malignant mesotheliomas stained strongly (Figure 2k)
; similar reactions
were seen in epithelial, sarcomatoid, and biphasic variants. In
contrast, samples from two benign cystic mesotheliomas (multiple
peritoneal inclusion cysts) were consistently negative (Figure 2l)
.
Several small samples of glioblastoma multiforme stained strongly and
extensively (Figure 2m)
. In situ malignant melanomas showed
convincing reactions of the dysplastic melanocytes at the base of the
epidermis, and of single cells migrating upwards in the epidermis.
Invasive malignant melanomas of various sites showed intense and
diffuse reactions (Figure 2n)
irrespective of architecture or melanin
content.
Fetal samples examined ranged from 20 to 24 weeks of gestational age;
tissues included lung, heart, liver, kidney, and bowel. Reactions were
particularly convincing in developing bronchi and primitive air spaces
(Figure 2o)
as well as in the crypts of the intestinal mucosa.
Immunoblots
The results of these experiments are shown in Figure 3
. With a single exception, all samples
of malignant tumors analyzed, showed a convincing increase in the
amount of Nup88. All colon carcinomas (A1–A5) showed reactive bands
that were far stronger than the controls; the comparatively weak band
noted in A2 may reflect this tumor’s extensive necrosis. In the case
of lung tumors, the respectable band of the bronchial carcinoid (B1)
corresponds in fact to a rather intensely reactive tumor (Figure 1o)
.
Notice also the intense B2 and B4 bands (carcinomas) and compare with
the meager B3 band that represents a carcinoma treated with radio- and
chemotherapy before removal resulting in extensive tumor necrosis. The
strong B5 band pertains to a leiomyosarcoma metastatic to the lung
(Figure 2i)
, and the relatively weak but distinct B6 band corresponds
to bronchi in the vicinity of squamous carcinoma showing hyperplastic
and metaplastic changes.
|
)
show delicate bands, and the normal breast is virtually negative (C7).
In the ovarian samples, the meager band noted in D1 (benign
cystadenoma) is in stark contrast with the broad bands seen in the
carcinomas (D2–D4). Except for problems related to limited sampling,
we cannot explain the rather weak band of the endometrioid carcinoma
(D5) that by immunohistochemistry was moderately stained. Densitometric quantification of the blots showed an increased expression in carcinomas between 1.5 and 5 times greater than the normal controls. Still, it should be considered that these figures might well represent a significant underestimation for they do not take into account the considerable amount of stroma that many of these tumors had. More detailed quantitative analyses should be performed based on larger series, and subsequent to appropriate microdissection.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
With regard to epithelial cancers, Nup88 enhancement was seen not only across a broad spectrum of sites but through all major differentiation lines. Thus, lung carcinomas with squamous, glandular, and neuroendocrine features were strongly positive. In the gastrointestinal tract and pancreas adeno- and neuroendocrine carcinomas were similarly reactive, whereas in the ovary, serous, mucinous, endometrioid, and clear-cell carcinomas reacted as well. Moreover, Nup88 overexpression seems to be independent of presumed histogenesis as it was found in carcinomas from diverse organs derived from all embryonal layers. In addition, as noted earlier,3 we also found that in sites such as the ovary, high-grade carcinomas seemed to react more vigorously than their low-grade and borderline counterparts. Furthermore, in some breast and endometrial carcinomas, we noted that the invasive periphery of the tumors stained more strongly than the center. These findings suggest a spatial and temporal relationship between Nup88 overexpression and tumor expansion, and parallel previous observations wherein certain extracellular matrix proteins, eg, tenascin and cellular fibronectins are strongly enhanced in the periphery of carcinomas reflecting areas of active remodeling.10,18
Nup88 enhancement was clearly evident in severe epithelial dysplasias and in situ carcinomas of the colon, stomach, breast, and prostate; this overexpression was noted irrespective of the presence or absence of an overt synchronous cancer. In contrast, proliferative but benign variants of conditions such as fibrocystic disease of the breast, endometrial hyperplasia, tubular adenomas of the colon, and glandular prostatic hyperplasia showed no significant Nup88 enhancement. Similar contrasts were noted in mesenchymal and other miscellaneous malignancies, eg, strong and extensive reactions in a leiomyosarcoma and in malignant mesotheliomas contrasting with weak and sporadic or absent staining in their benign counterparts. Notably, certain active reparative processes, eg, bile duct proliferation in cirrhosis, and proliferating renal tubules in the vicinity of carcinoma associated with pyelonephritis showed modest but convincing Nup88 reactions. Again, such areas are known to undergo active remodeling as reflected by the enhancement of pertinent matrix molecules.10,18-21 These findings indicate that enhanced Nup88 reflects a selective cellular proliferation that is most often, but not exclusively, associated with the malignant or premalignant phenotype.
Some of the above findings of Nup88 overexpression in malignancy coupled with its detection in some fetal tissues suggest parallels between it and a number of oncodevelopmental marker molecules including carcinoembryonic antigen and some related substances. However, the latter molecules are for the most part cell membrane-associated glycoproteins known to have or suspected of having cell-cell adhesive functions.22,23 In addition, they are selectively expressed in some epithelia but not in others, and are rare in nonepithelial tissues. Moreover, the punctiform, perinuclear, and cytoplasmic localization of Nup88 differs substantially from that of the above molecules. These observations added to those that Nup88 is also significantly enhanced in malignancies as diverse as carcinomas, some sarcomas and lymphomas, mesotheliomas, melanomas, and gliomas point to radical differences between Nup88 and oncodevelopmental markers currently used.
A consistent finding in the present study was the predominantly cytoplasmic localization of the overexpressed protein in the involved cells. Our previous studies on several cell lines showed that most of the protein was located at the nuclear membrane with comparatively small amounts in the cytoplasm. Noteworthy in many neoplastic and some nonneoplastic cells is the presence of aggregates of annulate lamellae24 ; these structures are thought to derive from nuclear membranes,25 and show features of the latter, eg, components of nuclear pore complexes including nucleoporins as described in Xenopus oocytes,26 and in rat cells wherein they were visualized as cytoplasmic dots.27 Therefore, we infer that the conspicuous cytoplasmic granules we found may reflect increased complements of annulate lamellae. In this context, it merits mentioning that other oncoproteins may also be aberrantly located, eg, the known nuclear-cytoplasmic mislocation of the BRCA1 gene product in breast carcinomas.28
The polyclonal antiserum used in these experiments allowed us to recognize clearly and consistently a Mr 88,000 band in immunoblots; in addition, other weakly reactive bands were at times detected. The precise nature of these additional reactive bands remains unclear, but it should be stated that our antiserum reacts strongly with the glutathione S-transferase-Nup88 fusion proteins used for the immunization of the rabbits.3 We cannot as yet state whether the aforementioned bands of lower molecular weight reflect degradation products of a single molecule or distinct proteins that share a similar or common epitope. The eventual identification of the epitope recognized by our antiserum should help elucidate the fragments of the proteins recognized by it. Given our current data, some questions may be said to persist as to whether the material recognized by our antiserum corresponds in fact to Nup88. Significantly, in carcinomas of various sites and in one sarcoma, strong and extensive immunostaining of tissue sections was paralleled by similarly strong Mr 88,000 reactive bands in Western blots of the same samples; conversely, in samples of hyperplasias, benign tumors, and normal tissues, weak or undetectable immunoreactions were reflected by weakly reactive bands in the corresponding immunoblots. Moreover, no differences in intensity were noted in the additional reactive bands found in tumors as well as controls thus reinforcing the notion that our antiserum is specific for Nup88. These data strongly suggest that the molecule in question corresponds indeed to Nup88.
In vertebrate cells, the nuclear pore complex is a large macromolecular aggregate29,30 with an estimated molecular mass of 125 Mr31 it includes 50 to 100 proteins termed nucleoporins.32 On the other hand, in yeast, nuclear pore complexes are smaller, have a molecular mass in the range of 66 Mr, and may include 30 to 40 nucleoporins.33,34 Among the known nucleoporins, the rather recently characterized Nup88 in its dynamic subcomplex with the oncogenic nucleoporin CAN/Nup214 seems to play several essential roles; depletion of the complex results in defective import-export processes, and eventual cell-cycle arrest;5 and, in overexpressing cells, it seems that CAN/Nup214 and one of its interacting proteins, ie, Nup88, may function on both aspects of the nuclear pore complex.35 Notably, preliminary experiments in our laboratory showed no increased expression of Can/Nup214 in our tumor samples thus suggesting an uncoupling of the latter from Nup88, at least in some instances. Interesting in this context is the fact that the Nup98 gene seems to be involved in therapy-related leukemias by translocation producing fusion proteins that may act as transcription factors modulating the expression of other genes.36
We can only speculate about the possible role(s) of the overexpressed Nup88 in malignant cells. A possible explanation is that its overexpression is simply the result of increased nucleocytoplasmic transport required to meet the increased demand of proteins by transformed cells. Increased traffic is indeed known to occur in this context, and it merits mention that the diameter of the pore channel seems to be increased in transformed mammalian cells.37 An alternative explanation is that Nup88 may play a role in the formation and maintenance of annulate lamellae as outlined above: but, although this might explain their presence it would still not clarify their function. Irrespective of these speculations, our findings suggest that Nup88 may be a potentially significant marker given its dramatic overexpression in a broad spectrum of malignant tumors of most, if not all, major types. If these results were confirmed, Nup88 might be said to approach an ideal generic marker of transformation readily demonstrable in conventional tissue sections, and possibly also in cytological samples.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Supported by a grant from the Universidad del Pais Vasco (University of the Basque Country), Bilbao, Spain (to N. M.).
Accepted for publication July 27, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  H. Zhang and I. Rosdahl Ultraviolet A and B differently induce intracellular protein expression in human skin melanocytes--a speculation of separate pathways in initiation of melanoma Carcinogenesis, December 1, 2003; 24(12): 1929 - 1934. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Grinstein, M. V. Preciado, P. Gattuso, P. A. Chabay, W. H. Warren, E. De Matteo, and V. E. Gould Demonstration of Epstein-Barr Virus in Carcinomas of Various Sites Cancer Res., September 1, 2002; 62(17): 4876 - 4878. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
