Histopathology 1991, 18, 1-10
ADONIS
030901 679 10001 2
Ultrastructural localization of blood group antigen A in normal and neoplastic urothelium C.LIMAS, B.CUTLER* & P.LANGEt Research Service, V A Medical Center and the Departments of *Laboratory Medicine and j-Urologic Surgery, University of Minnesota School of Medicine, Minneapolis, USA Date of submission 2 1 March 1990 Accepted for publication 12 July 1990
L I M A S C . , C U T L E R B . & LANGE P .
(1991) Histopathology 18, 1-10
Ultrastructural localization of blood group antigen A in normal and neoplastic urothelium The subcellular distribution of the blood group antigen A in the transitional epithelium of the urinary tract and its neoplastic growths was studied using transmission immuno-electronmicroscopy. Sixty-five tissue specimens from 50 blood group A1 patients were processed according to an immunogold procedure which was optimized for preservation of both antigen and ultrastructure. The reactions were stronger in the glycocalyx of the luminal surfaces and at the interdigitating cytoplasmic processes of the cells. In the intracellular compartment the reactions were associated with tubulovesicular membrane-bound structures and with the Golgi complexes. Secretory products, intra- or extracellular, were also positive. The greatest variability was noted in the cell surface reactions, which were positive in 88% of normal but only 41% of neoplastic urothelial specimens. An inverse correlation was found between malignant behaviour and cell surface, but not intracellular, reactions. We conclude that, in transitional cell carcinomas, there is a quantitative defect in the processing of substance A which affects predominantly the cell surface component and may involve either the transport-insertion steps, the plasma membrane-associated glycosyltransferases or internalization of blood group antigen A. Keywords: blood group antigen A, urothelium, transitional cell carcinoma
Introduction The presence of blood group antigeas A, B, H and Lewis a and b in the normal transitional epithelium of the urinary tract is well-documented' ,'. The localization of the A, B and H antigens specifically on the cell surface has been demonstrated using suspensions of intact urothelial cell^^,^ and immunohistochemistry 5 . In particular, the blood group antigen A was shown by scanning electronmicroscopy to localize on the luminal surface of the urothelium'. The oligosaccharide determinant of antigen A is present on the carbohydrate chains of the glycoproteins and glycolipidsi which form the glycocalyx or cell coat, but is also found on the glycoproteins of many secretions. A great deal of the glycosylation is accomplished in Address for correspondence: Dr C.Limas. VA Medical Center, One Veterans Drive, Minneapolis. MN 5541 7 , USA.
the Golgi apparatus by the consecutive, step-wise action of highly specialized enzymesx-l0. The end-products are transported to their appropriate destinations, e.g. plasma membrane or secretory granules' l . However, terminal addition of sugar residues may occur after the insertion of the molecules into the plasma membrane by the action of plasma membrane-associated glycosyltransfera~es'~.' 3. The localization of the A determinant and the relevant specific transferase in subcellular organelles and the plasma membrane of intestinal absorptive cells has been shown by Roth and coworker~'~. Many biologically important macromolecules, such as receptors for hormones, growth factors, viruses etc.. are glycosylated and inserted into the plasma membrane, probably following post-translational pathways similar to those of blood group substances. Abnormalities in the expression of blood group antigens may thus reflect a 1
2
C.Limas, B.Cutler and P.Lange
biosynthetic defect which could affect multiple molecules essential for the regulation of cell growth and differentiation. In pathological situations, abnormal distribution of blood group A substance could be an indicator of impaired glycosylation and/or disrupted intracellular transport of macromolecules. Specifically, the marked reduction of blood group A antigen, associated with malignant urothelial tumours, could be used to investigate a potentially fundamental deviation in the biosynthesis of glycoconjugates with bearing on invasive and metastatic cell growth. With this in mind, we undertook an extensive study of the subcellular distribution of blood group A substance in urothelial cells in the normal and neoplastic state, hoping to gain some insight into the phenomenon of blood group antigen 'loss'in transitional cell carcinomas.
Materials and methods Tissue specimens were obtained by biopsy or radical surgery from non-neoplastic ( 35 specimens) and neoplastic ( 3 0 specimens) urothelium derived from 5 0 patients. Among these were 1 5 patients who had both neoplastic and non-neoplastic urothelium examined. Specimens were obtained fresh and each was divided into portions to be processed for light and electronmicroscopy. Initially, feasibility studies were conducted in order to determine the optimal conditions for fixation, embedding and staining. The preservation of antigen was monitored, using as controls fresh-frozen tissue sections from the same specimen, and the ultrastructural detail was compared with that obtained in portions of the same specimen processed according to well-established routine methodology. The routine procedure consisted of fixation in a mixture of 2% glutaraldehyde and 2% paraformaldehyde, post-fixation in 1%0s04 and polybed 812 (Polysciences, PA, USA) embedding. In the preliminary studies we utilized various fixatives and embedding medial5-Iy,such as LR white, Lowicryl K4M and polybed 812. An optimal combination of ultrastructural detail and antigenic reactivity was obtained with the method described below, which was then used to carry out the study. The portion of the specimen to be processed for immuno-electronmicroscopy was fixed in a mixture of 3%) paraformaldehyde. 0.2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 2 h at 4OC and washed in three changes of phosphate-buffered saline (PBS). Some of the tissue blocks were post-fixed in 1% 0s04 for 1 h on ice and some were processed without osmication. Following dehydration through ascending ethanol solutions, the blocks were embedded in polybed 812. Sections
were cut at 1 pm thickness, stained with toluidine blue and examined under a Zeiss light microscope. Illtrathin sections (60 nm) were obtained from selected areas, placed on nickel grids and etched in a 3% solution of saturated sodium methoxide in ethanol for 3 min. The sections were then rehydrated and rinsed in PBS. Nonspecific antibody binding was blocked with horse serum for 2 0 min. Incubation with a monoclonal anti-A antibody (Dako, CA, USA) diluted 1: 10 was carried out for 1 h. The grids were then thoroughly washed in PBS and incubated on drops of biotinylated anti-mouse IgM (Vector Laboratories, CA, USA). Following washings in PBS. they were transferred to a colloidal gold-avidin (E-Y Laboratories, CA, USA) preparation, incubated for 2 h and thoroughly washed in PBS. All incubations were carried out in a moisture chamber at room temperature. Finally, the sections were lightly stained with lead citrate and uranyl acetate and examined with a Zeiss electronmicroscope. As negative controls we used: ( a ) sections from each block incubated with non-immune serum and ( b ) sections from blood group 0 patients incubated with anti-A. The red blood cells of all patients were tested for A and B, as well as Lewis a and b blood group antigens using antisera from Accugenics CA, USA and Cooper Biomedical, PA, USA according to the manufacturer's instructions. In addition, the red blood cells were tested for agglutination with the lectin of Dolichos biflorus (Vector), and those positive were classified as A,. The red blood cells of all patients included in this study were agglutinated with anti-A and anti-Leb sera and with the dolichos lectin. All specimens were examined by conventional light microscopy after staining with haematoxylin and eosin and, in some cases, with alcian blue and periodic acidSchiff. Conventional ultrastructural studies were performed using 60-70 nm sections from post-osmicated blocks stained with lead citrate and uranyl acetate according to established methods. Fresh frozen and paraffin embedded tissues were processed for immunohistology as previously described2". Briefly, following thorough deparaffinization and inactivation of endogenous peroxidases, 4 pm thick sections were incubated with a monoclonal antibody (Dako)against the blood group A determinant (synthetic trisaccharide) for 1 h at room temperature. They were then washed in buffer and incubated with biotinylated anti-mouse IgM (Vector) and finally with the avidinperoxidase complex (Vector). Positive reactions were visualized using diaminobenzidine (Sigma, MO. 1JSA) and HzOz as substrate for the peroxidase. Haematoxylin was used as counterstain to identify the various tissue structures.
Ultrastructure uf uruthelial blood group antigen A
3
Figure 2. Normal urothelium: positive staining for blood group A antigen appears to localire mostly on the cell surface. The cytoplasm of the cells is also focally positive. I’eraffin processed: Vectastain. x 2 50.
Figure 1. Osmicated. epoxy resin-embedded tissue, stained for routine electronmicroscopy, shows normal urothelium with prominent glycocalyx. x 1 3 100.
Results By light microscopy, the non-neoplastic mucosae utilized in this study were either unremarkable or had minor focal lymphocytic infiltrates. They showed no atypia and displayed full cellular differentiation from basal to superficial layers. Of the transitional cell neoplasms, 2 0 were papillary non-invasive, four had superficial and 10 muscular layer invasion. Well-preserved areas without necrosis were selected for immunological studies. Ultrastructurally the most superficial (luminal) cell layer of normal urothelium is characterized by a distinct fuzzy coat (Figure 1).junctional complexes and subluminal vesicular formations. In transitional cell carcinomas this layer showed great variations ranging from normal to very atypical, lacking surface differentiation or showing squamous or secretory features. The deeper layers of normal urothelium consisted of irregularly shaped cells with interdigitating cytoplasmic processes and sparse desmosomes, while the neoplastic cells were extremely variable, showing a wide spectrum of differentiation or absence of it.
Figure 3 . Transitional cell carcinoma. grade 2. the cells of which show variable reactivity for blood group A antigen: some show distinct paranuclear staining while others are completely negative. The reaction conditions were the same as in Figure 3 . Paraffin processed: Vectastain. x 1 0 0 0 .
IMMUNOHISTOCHEMISTRY
The blood group antigen A was detectable in fresh frozen and paraffin processed sections of non-neoplastic urothelium in all cases, but the level of reactivity varied
Figure 4. a. Sections from the same specimen as in Figure 1 were etched and processed with the anti-A imniunogold. There is localixation of gold along the luminal surface. a n d on microvesicular membranes. b Section lrom an unosniicatcd. epoxy resin-embedded, block of the same biopsy shows greater density of gold in the glycocalyx. but the visualization of membranes is reduced. x 9600.
Figure 5. a Normal intermediate cell layer with positive staining for blood group A antigen particularly along cytoplasmic processes. b Localization of gold particles on a tubulovesicular complex (broad arrows) and the interdigitating cell surfaces of normal cells (small arrows). a Unosmicated: etched. b Osmicated: etched block. a x 7500. b x 9000.
Ultrastructure of urotheliul blood group untigen A
from strong, uniform reactions to weak and patchy. As we have previously reported in detail', the use of fixatives and solvents in the processing of tissues may result in modification of reactivity, and this is further discussed below. Distinction of cell surface from cytoplasmic staining was best accomplished by controlling the intensity of staining through antibody dilution and
Figure 6. Immunogold staining for blood group antigen A in transitional cell carcinomas: examples of the discrepancy between the staining of intracytoplasmic vesicular structures and cell surface. a Tissue was epon-embedded, osmicated, and etched. b Tissue was epon-embedded, unosmicated and etched. Notice the poor visualization of membranes in the absence of osmication. a x 8500. gold particles 8.5 nm. b x 5500, gold particles 12.8 nm.
5
timing of the colour development step. As shown in Figure 2 , there was often an impression of cell surface staining which, however, could not be differentiated from peripheral cytoplasmic reactions. Van Brunn nests and the secretory products within their lumina were consistently positive. In transitional cell carcinomas there was a marked
6
C.Lirnas, B.Cutler and P.Lange
Figure 7. Transitional cell carcinoma: a routinely stained section from an osmicated. epoxy resin-embedded block showing intracellular lumina containing secretory products: b immunogold staining for blood group antigen A on etched sections from the same block showing that both the secretions and the luminal surfaces are intensely positive. a x 6000. b x 7700.
Ultrastructure of urotheliul blood group antigen A
variation in the intensity and pattern of reaction from case to case, ranging from uniformly strong positive to completely negative. Total absence of reactivity from neoplastic cells was observed in the face of uniformly positive reactions in the normal urothelium from the same patient. Variability was also observed in different areas of the same neoplasm: some cells expressed the A antigen on both surface and cytoplasm, while others showed strong, discrete cytoplasmic reactions in the absence of cell surface staining (Figure 3). Intracellular and extracellular secretory material which stained with alcian blue at pH 2.5. and was assumed to be mucin. reacted consistently with anti-A. IMMUNOGOLD
Treatment with osmium enhanced the visualization of membranes, but suppressed the antigenic reactivity of the tissue. Greater reactivity (greater density of gold particles) was obtained with blocks embedded in LRWhite or Lowicryl K4M but these embedding media gave poor preservation of membranes and organelles. An optimal combination of antigenic and structural preservation was achieved when unosmicated tissues were
Figure 8. Endothelial cells with positive staining for blood group antigen A. Unosrnicated: etched processed with irnrnunogold. x 21 50.
7
embedded in polybed 812 and the sections etched lightly as described in the methods section. At the ultrastructural level, the luminal cell surface of the normal urothelium showed the greatest density of gold particles. These were clearly localized externally to the plasma membrane, within the zone corresponding to the fuzzy coat (Figure 4). A lesser density of particles was observed on the lateral surfaces of adjoining cells. Gold particles also outlined the inner aspect of the subluminal intracellular vesicles which were best visualized in osmicated sections. The normal intermediate cells showed two distinct and consistent sites of gold labelling: the cell surface and paranuclear tubulovesicular complexes. At the cell surface, the gold particles concentrated along the cytoplasmic processes and intracellularly in the paranuclear formations of Golgi (Figure 5). The basal cells showed positive reactions on all surfaces, including that abutting on the basement membrane. In 38% of the cases the normal cells showed uniformly strong gold labelling, in 50% uniformly weak and in 12% focally trace labelling. The neoplastic cells showed great variability in the intensity and distribution of the immunogold reactions.
In about 60% of transitional cell carcinomas the reactions were either negative or only focally positive. Many tumour cells had a n apparent dissociation between intracellular and cell surface density of gold particles, the latter being much reduced (Figure 6). In carcinomas which showed glandular differentiation, a great concentration of particles was noted over secretions. Even invasive tumours showed distinct localization of gold particles over secretory products and along luminal formation (Figure 7). Twelve of 18 transitional cell carcinomas which showed no cell surface reactivity, and two of the 12 which retained good cell surface reactivity, had an aggressive course. The endothelial cells were consistently positive and could serve as built-in controls (Figure 8), while the red blood cells gave only weak and inconsistent reactions.
Discussion Most information about the biosynthesis of the blood group antigens has been obtained using biochemical approaches but, more recently, the availability of immunohistological methods made possible the study of the glycosylation process in intact cells and opened a new era of research on this subjectY-l3 The prevailing theories about the biosynthesis of the blood group antigens implicate a stepwise glycosylation by genetically controlled enzymes7. The blood group substances found in secretions are proteins which undergo glycosylation in the Golgi apparatus, are transported to the secretory granules and are exteriorized into luminal structures of tissues and organs. The cell surface-associated blood group substances, on the other hand, are integral plasma membrane glycoproteins and glycolipids. which may be processed through the Golgi but transported to the cell surface where they may undergo further glycosylation by plasma membrane associated enzymes. Roth et uJ.I4 demonstrated by immuno-electronmicroscopy that the blood group antigen A is found in the Golgi brush border and mucus of intestinal epithelium. These investigators also demonstrated that the relevant enzyme, cl-l,3-~-acetylgalactosaminyl transferase, is present in the Golgi, multivesicular bodies and at the brush border of intestinal absorptive cells. This suggests that glycosylation can indeed occur after insertion of the macromolecules into the plasma membrane. Furthermore, it has been proposed that some glycosyltransferases of the cell surface can interact with the extracellular milieu2' and could regulate cell migration and recognition functions. Since it is possible to trace the biosynthetic pathway by studying the subcellular distribution of blood group substances in tissue sections, we thought that this approach could be used to
define which steps are impaired when the expression of the blood group antigens is not normal. The blood group antigen A appears to be missing or to be suppressed in several types of neoplastic cells, including the transitional epithelium of the urinary tract' '-?. The normal urothelium has no secretory function (at least in the conventional sense) and its architectural design is simple enough to permit easy orientation for the study of cell layers varying in the degree of differentiation. In order to avoid artifacts due to inconsistencies in the preservation of the blood group antigen A, which depends on the donor's genetic background, we included only specimens derived from patients with the blood group A , , Leh phenotype. We have previously shown that the epithelial and endothelial cells in these individuals retain the A antigen even following exposure to fixatives and solvents commonly used for histologic preparations'. We do not know why 12%of morphologically normal mucosae showed only trace localization of gold particles. We suspect that this is due to some methodological factor(s) because the frozen and paraffin processed blocks from these biopsies gave positive reactions with the same antibody. In this study, we demonstrated the presence of blood group A determinants both on the cell surface and in intracellular organelles. On the cell surface, the greatest density of antigenic sites coincides with the glycocalyx. where most of the glycosylated sites of plasma membrane integral molecules reside. Many of these molecules are part of the cell recognition and interaction system and a similar function may prove germane to blood group substance A. Also, the relatively high density of blood group A substance along the interdigitating cytoplasmic processes of the epithelium may be signiticant in facilitating cell-to-cell interactions. Furthermore. some of the apparently intracellular formations which carry the blood group A substance, such as the subluniinal vesicles of the epithelium and the pinocytotic vesicles of the endothelium, are indeed plasma membranederived and show localization of the gold particles on their inner surface (which corresponds to the external surface of the plasma membrane). The Golgi apparatus apparently plays a central role in the glycosylation of blood group A substance. including the terminal sugar residue addition to the carbohydrate side chains. Whether in urothelial cells the Golgi is the only site of such biochemical events or there is additional glycosyltransferase activity in the plasma membrane remains to be clarified, as was done for the intestinal epithelium by Roth and co-workers'4. These investigators were able to demonstrate by immuno-electronmicroscopy the localization of cl- 1,3-~-acetylgalactosaminyl
Ultrnstriictiire of urothi~linlblood group antigen A
transferase in both the Golgi and the brush border of absorptive intestinal cells. In neoplastic cells, the overall subcellular localization of blood group A determinants was similar to that of normal cells. However, deviations from the normal were noted which point to a deficiency of cell surface glycosylated molecules. There was a pronounced variability in the cell surface reactions of transitional cell carcinomas. both between cases and within the same tumour, and the frequency of negative reactions correlated with the malignant potential. Also, the apparent discrepancy between persistent intracellular and markedly diminished or absent cell-surface reactions that was frequently noted in transitional cell carcinomas suggests that there is a neoplasia-associated defect in the late phase of substance A biosynthesis. Stellner et aLz2have reported a suppression of specific glycosyltransferase activities in adenocarcinomas, but we do not know whether this is also the case for transitional cell carcinomas. We have previously reported that the reduction of blood group A reactivity in transitional cell carcinomas often occurs in the face of an almost reciprocal increase in antigen H, as measured by their reactions with the Ulex europeus lectin4. This also points to a deficiency in terminal Nacetylgalactosaminyl residues. The present study confirms the existence of such a defect and, in addition. points out that this affects particularly the cell surface molecules. Several mechanisms may be responsible for the defective glycosylation of blood group substance A: e.g. selective suppression of plasma membrane glycosyltransferases; impaired transport and insertion of glycosylated molecules into the cell surface: or enhanced internalization of blood group substance. Further studies are needed to define the precise mechanism( s) involved. Intracellular reactivity and that of secretory products was preserved even in invasive transitional cell carcinomas and, therefore. only defects in the pathway of cell surface blood group A substance expression seem to be relevant to malignant behaviour. Unfortunately, the methodology used for evaluation of blood group antigen reactivity by light microscopy. i.e. red cell adherence and immunohistochemistry on tissue sections, has not permitted clear-cut distinction of cell surface reactions in the presence of positive cytoplasmic staining, and most reports dealing with the prognostic significance of blood group antigen expression in tumours do not address this issue. The results of this study confirm the fact that abnormalities in the glycosylation of blood group A substance occur with high frequency in malignant urothelial neoplasms and, in addition, point out that a suppression of the cell surface component is particularly relevant to the aggressive potential and should be specifically
9
analysed. We believe that such a n analysis is worthwhile because it will advance our knowledge not only of the process of malignant transformation of the urothelium but also of the function of cell surface glycoconjugates.
Acknowledgements This work was supported in part by a grant (CA 3 323a) from the National Cancer Institute, National Institutes of Health, Bethesda, MI), USA.
References I . Linias C . I m g e P. A.B.H antigen detectability in normal and neoplastic urothelium: influence of methodologic factors. Carrcw 19x2: 49: 247h-24X4. 2. Ihnas C. Imige P. Lewis antigens in normal and neoplastic urothelium. h i . /. l’citl~nl. 19x5: 121: 176-183. 3. Kay HEM. Wallace DM. A and B antigens of tumors arising from urinary epithelium: JNCI 1961: 26; 1349-1365. 4. Iimas C . [.anye 1’. Altered reactivity for A.B.H antigens in transitional cell caricinomas of the urinary bladder: a study of the niechanisms involved. C m c w 1980: 46: 1 366-1 373. 5 . Coon IS. Weinstein KS. Detection of A.B,H tissue isoantigens by imiiiunoperoxidase methods in nornmal and neoplastic urothelium: coinparison with the erythrocyte adherence method. A m . /. Clifi. Pntlinl. 1 9 X 1 : 76: 1 h 3- 1 7 1 . 6 . DeHarven E, He S.Hanna W. Bootsma G. Connolly JG. Phenotypically heterogeneous deletion of the A.B.H antigen from the transformed bladder urothelium. A scanning electron microscope study. /. Subrtiicrosc~.C!jtol. 19x7: 19: h 39-649. 7. Watkins WM. I%ochemical genetics of blood group antigens: retrospect and prospect. HiocIicwt. Sot. Trcirts. 19X 7 : i 5: h20-624. X. Ilunphy WG, Kothman YE. Compartmental organization of the Golgi stack. C d l 19x5: 42: 1 3-1 1. 9. Roth J , 13erger EG. Inimiinocytocheinical localization of galatosyltransferase in HeLa cells: Codistribution with thiamine pyrophosphase in trans-Golgi cisternae. /. Cell, H i d . 1982: 93: 2 1 3-229. 0 , Kornfeld K. Kornfeld S. Assembly of asparagine-linked oligosaccharides. Afttiit. K w . Rioclrcw. 1985: 54: 6 3 1-664. I , Rothman JE. The coinpartmental organization of the Colgi apparatus. Sci. A t r i c v . 19x5: 253: 74-89. 2. Shaper NL. Mann PI.. Shaper JH. Cell surface galatosyltransferase: immunochemical localimtion. /. Cell. Hiodtcw. 1985: 28: 229239. 1 3 , Pestaloui PM. Hess M. Herger EC. Ii~imunohistochemical evidence for cell surface and Golgi localization of galactosyltransferase in human stomach. jejunum. liver and pancreas. /. Histocltm. C!jfoc/irwt. 1982: 3 0 : 1146-1 152. 14. Roth J. Greenwell P. Watkins WM. Subcellular distribution of the products of the blood group A gene. Biochem. SOC.Trans. 198 7: 1 5: 599-601. 15. Newman G K . Hohot J A . Modern acrylics for post-embedding immunostaining techniques. /. Hisfoc~lie~fi. C!jtochcrrt. 19x7: 35: 971-98 1. 1 6 . Lane BP. Europa L)L. I)ifferential staining of ultrathin sections of epoxy-embedded tissues for light microscopy. /. Hisfoclicw. Cgtodim. 1965: 1 3 : 579-585. 17. Rodning CB. I
ADONIS
030901 679 10001 2
Ultrastructural localization of blood group antigen A in normal and neoplastic urothelium C.LIMAS, B.CUTLER* & P.LANGEt Research Service, V A Medical Center and the Departments of *Laboratory Medicine and j-Urologic Surgery, University of Minnesota School of Medicine, Minneapolis, USA Date of submission 2 1 March 1990 Accepted for publication 12 July 1990
L I M A S C . , C U T L E R B . & LANGE P .
(1991) Histopathology 18, 1-10
Ultrastructural localization of blood group antigen A in normal and neoplastic urothelium The subcellular distribution of the blood group antigen A in the transitional epithelium of the urinary tract and its neoplastic growths was studied using transmission immuno-electronmicroscopy. Sixty-five tissue specimens from 50 blood group A1 patients were processed according to an immunogold procedure which was optimized for preservation of both antigen and ultrastructure. The reactions were stronger in the glycocalyx of the luminal surfaces and at the interdigitating cytoplasmic processes of the cells. In the intracellular compartment the reactions were associated with tubulovesicular membrane-bound structures and with the Golgi complexes. Secretory products, intra- or extracellular, were also positive. The greatest variability was noted in the cell surface reactions, which were positive in 88% of normal but only 41% of neoplastic urothelial specimens. An inverse correlation was found between malignant behaviour and cell surface, but not intracellular, reactions. We conclude that, in transitional cell carcinomas, there is a quantitative defect in the processing of substance A which affects predominantly the cell surface component and may involve either the transport-insertion steps, the plasma membrane-associated glycosyltransferases or internalization of blood group antigen A. Keywords: blood group antigen A, urothelium, transitional cell carcinoma
Introduction The presence of blood group antigeas A, B, H and Lewis a and b in the normal transitional epithelium of the urinary tract is well-documented' ,'. The localization of the A, B and H antigens specifically on the cell surface has been demonstrated using suspensions of intact urothelial cell^^,^ and immunohistochemistry 5 . In particular, the blood group antigen A was shown by scanning electronmicroscopy to localize on the luminal surface of the urothelium'. The oligosaccharide determinant of antigen A is present on the carbohydrate chains of the glycoproteins and glycolipidsi which form the glycocalyx or cell coat, but is also found on the glycoproteins of many secretions. A great deal of the glycosylation is accomplished in Address for correspondence: Dr C.Limas. VA Medical Center, One Veterans Drive, Minneapolis. MN 5541 7 , USA.
the Golgi apparatus by the consecutive, step-wise action of highly specialized enzymesx-l0. The end-products are transported to their appropriate destinations, e.g. plasma membrane or secretory granules' l . However, terminal addition of sugar residues may occur after the insertion of the molecules into the plasma membrane by the action of plasma membrane-associated glycosyltransfera~es'~.' 3. The localization of the A determinant and the relevant specific transferase in subcellular organelles and the plasma membrane of intestinal absorptive cells has been shown by Roth and coworker~'~. Many biologically important macromolecules, such as receptors for hormones, growth factors, viruses etc.. are glycosylated and inserted into the plasma membrane, probably following post-translational pathways similar to those of blood group substances. Abnormalities in the expression of blood group antigens may thus reflect a 1
2
C.Limas, B.Cutler and P.Lange
biosynthetic defect which could affect multiple molecules essential for the regulation of cell growth and differentiation. In pathological situations, abnormal distribution of blood group A substance could be an indicator of impaired glycosylation and/or disrupted intracellular transport of macromolecules. Specifically, the marked reduction of blood group A antigen, associated with malignant urothelial tumours, could be used to investigate a potentially fundamental deviation in the biosynthesis of glycoconjugates with bearing on invasive and metastatic cell growth. With this in mind, we undertook an extensive study of the subcellular distribution of blood group A substance in urothelial cells in the normal and neoplastic state, hoping to gain some insight into the phenomenon of blood group antigen 'loss'in transitional cell carcinomas.
Materials and methods Tissue specimens were obtained by biopsy or radical surgery from non-neoplastic ( 35 specimens) and neoplastic ( 3 0 specimens) urothelium derived from 5 0 patients. Among these were 1 5 patients who had both neoplastic and non-neoplastic urothelium examined. Specimens were obtained fresh and each was divided into portions to be processed for light and electronmicroscopy. Initially, feasibility studies were conducted in order to determine the optimal conditions for fixation, embedding and staining. The preservation of antigen was monitored, using as controls fresh-frozen tissue sections from the same specimen, and the ultrastructural detail was compared with that obtained in portions of the same specimen processed according to well-established routine methodology. The routine procedure consisted of fixation in a mixture of 2% glutaraldehyde and 2% paraformaldehyde, post-fixation in 1%0s04 and polybed 812 (Polysciences, PA, USA) embedding. In the preliminary studies we utilized various fixatives and embedding medial5-Iy,such as LR white, Lowicryl K4M and polybed 812. An optimal combination of ultrastructural detail and antigenic reactivity was obtained with the method described below, which was then used to carry out the study. The portion of the specimen to be processed for immuno-electronmicroscopy was fixed in a mixture of 3%) paraformaldehyde. 0.2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 2 h at 4OC and washed in three changes of phosphate-buffered saline (PBS). Some of the tissue blocks were post-fixed in 1% 0s04 for 1 h on ice and some were processed without osmication. Following dehydration through ascending ethanol solutions, the blocks were embedded in polybed 812. Sections
were cut at 1 pm thickness, stained with toluidine blue and examined under a Zeiss light microscope. Illtrathin sections (60 nm) were obtained from selected areas, placed on nickel grids and etched in a 3% solution of saturated sodium methoxide in ethanol for 3 min. The sections were then rehydrated and rinsed in PBS. Nonspecific antibody binding was blocked with horse serum for 2 0 min. Incubation with a monoclonal anti-A antibody (Dako, CA, USA) diluted 1: 10 was carried out for 1 h. The grids were then thoroughly washed in PBS and incubated on drops of biotinylated anti-mouse IgM (Vector Laboratories, CA, USA). Following washings in PBS. they were transferred to a colloidal gold-avidin (E-Y Laboratories, CA, USA) preparation, incubated for 2 h and thoroughly washed in PBS. All incubations were carried out in a moisture chamber at room temperature. Finally, the sections were lightly stained with lead citrate and uranyl acetate and examined with a Zeiss electronmicroscope. As negative controls we used: ( a ) sections from each block incubated with non-immune serum and ( b ) sections from blood group 0 patients incubated with anti-A. The red blood cells of all patients were tested for A and B, as well as Lewis a and b blood group antigens using antisera from Accugenics CA, USA and Cooper Biomedical, PA, USA according to the manufacturer's instructions. In addition, the red blood cells were tested for agglutination with the lectin of Dolichos biflorus (Vector), and those positive were classified as A,. The red blood cells of all patients included in this study were agglutinated with anti-A and anti-Leb sera and with the dolichos lectin. All specimens were examined by conventional light microscopy after staining with haematoxylin and eosin and, in some cases, with alcian blue and periodic acidSchiff. Conventional ultrastructural studies were performed using 60-70 nm sections from post-osmicated blocks stained with lead citrate and uranyl acetate according to established methods. Fresh frozen and paraffin embedded tissues were processed for immunohistology as previously described2". Briefly, following thorough deparaffinization and inactivation of endogenous peroxidases, 4 pm thick sections were incubated with a monoclonal antibody (Dako)against the blood group A determinant (synthetic trisaccharide) for 1 h at room temperature. They were then washed in buffer and incubated with biotinylated anti-mouse IgM (Vector) and finally with the avidinperoxidase complex (Vector). Positive reactions were visualized using diaminobenzidine (Sigma, MO. 1JSA) and HzOz as substrate for the peroxidase. Haematoxylin was used as counterstain to identify the various tissue structures.
Ultrastructure uf uruthelial blood group antigen A
3
Figure 2. Normal urothelium: positive staining for blood group A antigen appears to localire mostly on the cell surface. The cytoplasm of the cells is also focally positive. I’eraffin processed: Vectastain. x 2 50.
Figure 1. Osmicated. epoxy resin-embedded tissue, stained for routine electronmicroscopy, shows normal urothelium with prominent glycocalyx. x 1 3 100.
Results By light microscopy, the non-neoplastic mucosae utilized in this study were either unremarkable or had minor focal lymphocytic infiltrates. They showed no atypia and displayed full cellular differentiation from basal to superficial layers. Of the transitional cell neoplasms, 2 0 were papillary non-invasive, four had superficial and 10 muscular layer invasion. Well-preserved areas without necrosis were selected for immunological studies. Ultrastructurally the most superficial (luminal) cell layer of normal urothelium is characterized by a distinct fuzzy coat (Figure 1).junctional complexes and subluminal vesicular formations. In transitional cell carcinomas this layer showed great variations ranging from normal to very atypical, lacking surface differentiation or showing squamous or secretory features. The deeper layers of normal urothelium consisted of irregularly shaped cells with interdigitating cytoplasmic processes and sparse desmosomes, while the neoplastic cells were extremely variable, showing a wide spectrum of differentiation or absence of it.
Figure 3 . Transitional cell carcinoma. grade 2. the cells of which show variable reactivity for blood group A antigen: some show distinct paranuclear staining while others are completely negative. The reaction conditions were the same as in Figure 3 . Paraffin processed: Vectastain. x 1 0 0 0 .
IMMUNOHISTOCHEMISTRY
The blood group antigen A was detectable in fresh frozen and paraffin processed sections of non-neoplastic urothelium in all cases, but the level of reactivity varied
Figure 4. a. Sections from the same specimen as in Figure 1 were etched and processed with the anti-A imniunogold. There is localixation of gold along the luminal surface. a n d on microvesicular membranes. b Section lrom an unosniicatcd. epoxy resin-embedded, block of the same biopsy shows greater density of gold in the glycocalyx. but the visualization of membranes is reduced. x 9600.
Figure 5. a Normal intermediate cell layer with positive staining for blood group A antigen particularly along cytoplasmic processes. b Localization of gold particles on a tubulovesicular complex (broad arrows) and the interdigitating cell surfaces of normal cells (small arrows). a Unosmicated: etched. b Osmicated: etched block. a x 7500. b x 9000.
Ultrastructure of urotheliul blood group untigen A
from strong, uniform reactions to weak and patchy. As we have previously reported in detail', the use of fixatives and solvents in the processing of tissues may result in modification of reactivity, and this is further discussed below. Distinction of cell surface from cytoplasmic staining was best accomplished by controlling the intensity of staining through antibody dilution and
Figure 6. Immunogold staining for blood group antigen A in transitional cell carcinomas: examples of the discrepancy between the staining of intracytoplasmic vesicular structures and cell surface. a Tissue was epon-embedded, osmicated, and etched. b Tissue was epon-embedded, unosmicated and etched. Notice the poor visualization of membranes in the absence of osmication. a x 8500. gold particles 8.5 nm. b x 5500, gold particles 12.8 nm.
5
timing of the colour development step. As shown in Figure 2 , there was often an impression of cell surface staining which, however, could not be differentiated from peripheral cytoplasmic reactions. Van Brunn nests and the secretory products within their lumina were consistently positive. In transitional cell carcinomas there was a marked
6
C.Lirnas, B.Cutler and P.Lange
Figure 7. Transitional cell carcinoma: a routinely stained section from an osmicated. epoxy resin-embedded block showing intracellular lumina containing secretory products: b immunogold staining for blood group antigen A on etched sections from the same block showing that both the secretions and the luminal surfaces are intensely positive. a x 6000. b x 7700.
Ultrastructure of urotheliul blood group antigen A
variation in the intensity and pattern of reaction from case to case, ranging from uniformly strong positive to completely negative. Total absence of reactivity from neoplastic cells was observed in the face of uniformly positive reactions in the normal urothelium from the same patient. Variability was also observed in different areas of the same neoplasm: some cells expressed the A antigen on both surface and cytoplasm, while others showed strong, discrete cytoplasmic reactions in the absence of cell surface staining (Figure 3). Intracellular and extracellular secretory material which stained with alcian blue at pH 2.5. and was assumed to be mucin. reacted consistently with anti-A. IMMUNOGOLD
Treatment with osmium enhanced the visualization of membranes, but suppressed the antigenic reactivity of the tissue. Greater reactivity (greater density of gold particles) was obtained with blocks embedded in LRWhite or Lowicryl K4M but these embedding media gave poor preservation of membranes and organelles. An optimal combination of antigenic and structural preservation was achieved when unosmicated tissues were
Figure 8. Endothelial cells with positive staining for blood group antigen A. Unosrnicated: etched processed with irnrnunogold. x 21 50.
7
embedded in polybed 812 and the sections etched lightly as described in the methods section. At the ultrastructural level, the luminal cell surface of the normal urothelium showed the greatest density of gold particles. These were clearly localized externally to the plasma membrane, within the zone corresponding to the fuzzy coat (Figure 4). A lesser density of particles was observed on the lateral surfaces of adjoining cells. Gold particles also outlined the inner aspect of the subluminal intracellular vesicles which were best visualized in osmicated sections. The normal intermediate cells showed two distinct and consistent sites of gold labelling: the cell surface and paranuclear tubulovesicular complexes. At the cell surface, the gold particles concentrated along the cytoplasmic processes and intracellularly in the paranuclear formations of Golgi (Figure 5). The basal cells showed positive reactions on all surfaces, including that abutting on the basement membrane. In 38% of the cases the normal cells showed uniformly strong gold labelling, in 50% uniformly weak and in 12% focally trace labelling. The neoplastic cells showed great variability in the intensity and distribution of the immunogold reactions.
In about 60% of transitional cell carcinomas the reactions were either negative or only focally positive. Many tumour cells had a n apparent dissociation between intracellular and cell surface density of gold particles, the latter being much reduced (Figure 6). In carcinomas which showed glandular differentiation, a great concentration of particles was noted over secretions. Even invasive tumours showed distinct localization of gold particles over secretory products and along luminal formation (Figure 7). Twelve of 18 transitional cell carcinomas which showed no cell surface reactivity, and two of the 12 which retained good cell surface reactivity, had an aggressive course. The endothelial cells were consistently positive and could serve as built-in controls (Figure 8), while the red blood cells gave only weak and inconsistent reactions.
Discussion Most information about the biosynthesis of the blood group antigens has been obtained using biochemical approaches but, more recently, the availability of immunohistological methods made possible the study of the glycosylation process in intact cells and opened a new era of research on this subjectY-l3 The prevailing theories about the biosynthesis of the blood group antigens implicate a stepwise glycosylation by genetically controlled enzymes7. The blood group substances found in secretions are proteins which undergo glycosylation in the Golgi apparatus, are transported to the secretory granules and are exteriorized into luminal structures of tissues and organs. The cell surface-associated blood group substances, on the other hand, are integral plasma membrane glycoproteins and glycolipids. which may be processed through the Golgi but transported to the cell surface where they may undergo further glycosylation by plasma membrane associated enzymes. Roth et uJ.I4 demonstrated by immuno-electronmicroscopy that the blood group antigen A is found in the Golgi brush border and mucus of intestinal epithelium. These investigators also demonstrated that the relevant enzyme, cl-l,3-~-acetylgalactosaminyl transferase, is present in the Golgi, multivesicular bodies and at the brush border of intestinal absorptive cells. This suggests that glycosylation can indeed occur after insertion of the macromolecules into the plasma membrane. Furthermore, it has been proposed that some glycosyltransferases of the cell surface can interact with the extracellular milieu2' and could regulate cell migration and recognition functions. Since it is possible to trace the biosynthetic pathway by studying the subcellular distribution of blood group substances in tissue sections, we thought that this approach could be used to
define which steps are impaired when the expression of the blood group antigens is not normal. The blood group antigen A appears to be missing or to be suppressed in several types of neoplastic cells, including the transitional epithelium of the urinary tract' '-?. The normal urothelium has no secretory function (at least in the conventional sense) and its architectural design is simple enough to permit easy orientation for the study of cell layers varying in the degree of differentiation. In order to avoid artifacts due to inconsistencies in the preservation of the blood group antigen A, which depends on the donor's genetic background, we included only specimens derived from patients with the blood group A , , Leh phenotype. We have previously shown that the epithelial and endothelial cells in these individuals retain the A antigen even following exposure to fixatives and solvents commonly used for histologic preparations'. We do not know why 12%of morphologically normal mucosae showed only trace localization of gold particles. We suspect that this is due to some methodological factor(s) because the frozen and paraffin processed blocks from these biopsies gave positive reactions with the same antibody. In this study, we demonstrated the presence of blood group A determinants both on the cell surface and in intracellular organelles. On the cell surface, the greatest density of antigenic sites coincides with the glycocalyx. where most of the glycosylated sites of plasma membrane integral molecules reside. Many of these molecules are part of the cell recognition and interaction system and a similar function may prove germane to blood group substance A. Also, the relatively high density of blood group A substance along the interdigitating cytoplasmic processes of the epithelium may be signiticant in facilitating cell-to-cell interactions. Furthermore. some of the apparently intracellular formations which carry the blood group A substance, such as the subluniinal vesicles of the epithelium and the pinocytotic vesicles of the endothelium, are indeed plasma membranederived and show localization of the gold particles on their inner surface (which corresponds to the external surface of the plasma membrane). The Golgi apparatus apparently plays a central role in the glycosylation of blood group A substance. including the terminal sugar residue addition to the carbohydrate side chains. Whether in urothelial cells the Golgi is the only site of such biochemical events or there is additional glycosyltransferase activity in the plasma membrane remains to be clarified, as was done for the intestinal epithelium by Roth and co-workers'4. These investigators were able to demonstrate by immuno-electronmicroscopy the localization of cl- 1,3-~-acetylgalactosaminyl
Ultrnstriictiire of urothi~linlblood group antigen A
transferase in both the Golgi and the brush border of absorptive intestinal cells. In neoplastic cells, the overall subcellular localization of blood group A determinants was similar to that of normal cells. However, deviations from the normal were noted which point to a deficiency of cell surface glycosylated molecules. There was a pronounced variability in the cell surface reactions of transitional cell carcinomas. both between cases and within the same tumour, and the frequency of negative reactions correlated with the malignant potential. Also, the apparent discrepancy between persistent intracellular and markedly diminished or absent cell-surface reactions that was frequently noted in transitional cell carcinomas suggests that there is a neoplasia-associated defect in the late phase of substance A biosynthesis. Stellner et aLz2have reported a suppression of specific glycosyltransferase activities in adenocarcinomas, but we do not know whether this is also the case for transitional cell carcinomas. We have previously reported that the reduction of blood group A reactivity in transitional cell carcinomas often occurs in the face of an almost reciprocal increase in antigen H, as measured by their reactions with the Ulex europeus lectin4. This also points to a deficiency in terminal Nacetylgalactosaminyl residues. The present study confirms the existence of such a defect and, in addition. points out that this affects particularly the cell surface molecules. Several mechanisms may be responsible for the defective glycosylation of blood group substance A: e.g. selective suppression of plasma membrane glycosyltransferases; impaired transport and insertion of glycosylated molecules into the cell surface: or enhanced internalization of blood group substance. Further studies are needed to define the precise mechanism( s) involved. Intracellular reactivity and that of secretory products was preserved even in invasive transitional cell carcinomas and, therefore. only defects in the pathway of cell surface blood group A substance expression seem to be relevant to malignant behaviour. Unfortunately, the methodology used for evaluation of blood group antigen reactivity by light microscopy. i.e. red cell adherence and immunohistochemistry on tissue sections, has not permitted clear-cut distinction of cell surface reactions in the presence of positive cytoplasmic staining, and most reports dealing with the prognostic significance of blood group antigen expression in tumours do not address this issue. The results of this study confirm the fact that abnormalities in the glycosylation of blood group A substance occur with high frequency in malignant urothelial neoplasms and, in addition, point out that a suppression of the cell surface component is particularly relevant to the aggressive potential and should be specifically
9
analysed. We believe that such a n analysis is worthwhile because it will advance our knowledge not only of the process of malignant transformation of the urothelium but also of the function of cell surface glycoconjugates.
Acknowledgements This work was supported in part by a grant (CA 3 323a) from the National Cancer Institute, National Institutes of Health, Bethesda, MI), USA.
References I . Linias C . I m g e P. A.B.H antigen detectability in normal and neoplastic urothelium: influence of methodologic factors. Carrcw 19x2: 49: 247h-24X4. 2. Ihnas C. Imige P. Lewis antigens in normal and neoplastic urothelium. h i . /. l’citl~nl. 19x5: 121: 176-183. 3. Kay HEM. Wallace DM. A and B antigens of tumors arising from urinary epithelium: JNCI 1961: 26; 1349-1365. 4. Iimas C . [.anye 1’. Altered reactivity for A.B.H antigens in transitional cell caricinomas of the urinary bladder: a study of the niechanisms involved. C m c w 1980: 46: 1 366-1 373. 5 . Coon IS. Weinstein KS. Detection of A.B,H tissue isoantigens by imiiiunoperoxidase methods in nornmal and neoplastic urothelium: coinparison with the erythrocyte adherence method. A m . /. Clifi. Pntlinl. 1 9 X 1 : 76: 1 h 3- 1 7 1 . 6 . DeHarven E, He S.Hanna W. Bootsma G. Connolly JG. Phenotypically heterogeneous deletion of the A.B.H antigen from the transformed bladder urothelium. A scanning electron microscope study. /. Subrtiicrosc~.C!jtol. 19x7: 19: h 39-649. 7. Watkins WM. I%ochemical genetics of blood group antigens: retrospect and prospect. HiocIicwt. Sot. Trcirts. 19X 7 : i 5: h20-624. X. Ilunphy WG, Kothman YE. Compartmental organization of the Golgi stack. C d l 19x5: 42: 1 3-1 1. 9. Roth J , 13erger EG. Inimiinocytocheinical localization of galatosyltransferase in HeLa cells: Codistribution with thiamine pyrophosphase in trans-Golgi cisternae. /. Cell, H i d . 1982: 93: 2 1 3-229. 0 , Kornfeld K. Kornfeld S. Assembly of asparagine-linked oligosaccharides. Afttiit. K w . Rioclrcw. 1985: 54: 6 3 1-664. I , Rothman JE. The coinpartmental organization of the Colgi apparatus. Sci. A t r i c v . 19x5: 253: 74-89. 2. Shaper NL. Mann PI.. Shaper JH. Cell surface galatosyltransferase: immunochemical localimtion. /. Cell. Hiodtcw. 1985: 28: 229239. 1 3 , Pestaloui PM. Hess M. Herger EC. Ii~imunohistochemical evidence for cell surface and Golgi localization of galactosyltransferase in human stomach. jejunum. liver and pancreas. /. Histocltm. C!jfoc/irwt. 1982: 3 0 : 1146-1 152. 14. Roth J. Greenwell P. Watkins WM. Subcellular distribution of the products of the blood group A gene. Biochem. SOC.Trans. 198 7: 1 5: 599-601. 15. Newman G K . Hohot J A . Modern acrylics for post-embedding immunostaining techniques. /. Hisfoc~lie~fi. C!jtochcrrt. 19x7: 35: 971-98 1. 1 6 . Lane BP. Europa L)L. I)ifferential staining of ultrathin sections of epoxy-embedded tissues for light microscopy. /. Hisfoclicw. Cgtodim. 1965: 1 3 : 579-585. 17. Rodning CB. I
