Losses of expression of the antigens A, Lea and Lex and over-expression of Ley in carcinomas and HG-SIL of the uterine cervix

Background The glycosylation of a great number of molecules, glyco-protein or glycolipids, has been of interest for decades. Objective To compare the expressive patterns of the isoantigenic determinants of histo-blood groups ABH and Lewis in squamous and simple epithelium and in precursors and cancers of the cervix. Methods A total of 36 lesions and neoplasms (10 LG-SIL, 16 HG-SIL and 10 invasive carcinomas) have been studied with immunohistochemical techniques, using monoclonal antibodies (MoAb BG1 to BG8) for precursor chains, blood-group ABH and Lewis group Lea, Leb, Lex, and Ley, and four types of lectins. In addition, we have studied the expression of p53 protein and PCNA, establishing the rate of proliferation of each lesion. Using PCR techniques, we have also detected part of the intron of the E6 gene of HPV-16. Results In the invasive cervical carcinomas, we observed a loss of expression of the Lex antigen (p < 0.01). With regard to the progression of the different lesions studied, we found alterations in the patterns of expression of the antigens of the ABH and Lewis blood groups. There was a tendency towards a loss of expression and heterogeneous patterns in the more advanced lesions, as well as over-expression of the Ley antigens. With PCNA, we established a proliferative rate which tended to be greater in relation to the progression of the cervix neoplasms. Conclusion These results indicate that there is a relation between the losses of histo-blood groups and the progression of the squamous intraepithelial lesions.


Background
The glycosylation of a great number of molecules, glycoprotein or glycolipids, has been of interest for decades. Increasing evidence highlights the fact that the carbohydrates located in the cellular surface and in the extracelu-lar matrix play an important role in cellular recognition and in the organization of tissues and organs, and that the aberrant expression of carbohydrates is closely related to certain diseases and to the progression of neoplasms. Historically, the functional role of carbohydrates in cellular recognition has been the most clearly demonstrable evidence in the development process and in cellular differentiation [1].
Most of the patterns of peripheral/terminal glycosylation is polymorphic in humans and constitute what is known as histo-blood groups [2], among which we know the systems ABO, secretor, Lewis, TTn and P. These antigen determinants have been identified in most human epithelia [3,4].
Furthermore, in some neoplasms there are frequent alterations of the cellular surface glycoconjugates that, in general, have been classified in: (A) incomplete synthesis, with or without accumulation of precursors, (B) synthesis of neoglycolipids and (C) organizational changes of the glycolipids in cellular membranes [5].
In addition, several molecular biology techniques arise as powerful tools that could provide complementary and valuable information to immunohistochemistry and to the traditional morphologic diagnosis. In this work we have combined a immunohistochemical study for the expression of PCNA (a known proliferative marker-PC-10 clone) and p53 (DO7 clone) with results of amplification of part of the E6 intron of the HPV-16.
PCNA is a nuclear protein of 36-kd cofactor of the polimerase delta detected in interfase cells (G1, S and G2) [23,24]. Mutations of p53 are the most frecuently reported in human neoplasms [25]. It has been suggested that p53 intervenes in the control of the cellular growth in the critical point among the phase G1 and S playing an important role of "guardian of the genoma". p53 could prevent the progression of cells through the cell cycle when DNA damage has occurred. Thus allows DNA to repair [26]. With regard to the influence of p53 upon several promoters it is considered that p53 could downregulate the PCNA RNAm [27]. Its contingent lost of function would have as consequence an increase of the cellular proliferation. Some authors also have observed that p53 would regulate the transcription of p21 protein which is known that can inhibit directly the PCNA and other cyclins impeding the synthesis of DNA in case of cellular damage [28,29]. Finally it is known that a HPV protein coded in the E6 gene could block and promote the degradation of the wild type of p53 [30,31]. Consequently the degradation of p53 could decrease p21 releasing its inhibition on PCNA.
We consider that the cervical intraepithelial neoplasms and carcinomas of the uterine cervix represent an optimum model for the study of the phenotypic alterations in the progression towards malignancy although, up until this moment they have been insufficiently explored. For this, reason, we have studied a total of 36 lesions that covered a spectrum from cervical metaplasias with and without dysplasia and/or condyloma to the invasive carcinoma, focusing our attention on the possible alterations of their patterns of glycosylation.
In addition, we used specific monoclonal antibodies NC-p53-DO7 and NCL-PCNA (Novocastra Laboratories Ltd., Newcastle upon Tuyne, UK). Titration of both antibodies without antigen retrieval was carried out. The best results were obtained at 1/100. Then an alkaline phosphatase method chromogen kit (Fast RedTM) was used. The IHC results were interpreted from each case, evaluating the percentage of reacting cells, and the intensity of the staining on a arbitrary scale from 0 (no staining) to 3 (strong staining).
The results of the immunohistochemical study were correlated with the amplification or not of the HPV-16 (part of the intron E6).

Scoring and statistical analysis
For each case we tabulate the findings evaluating the positive or negative expression, their homogeneity or heterogeneity of the stains, the presence of isolated positive cells, the stain of the cilia of the secretor cells, the stain of the mucus, and the location of the stain in the epithelium or in the thickness of the lesion. The results of the immunohistochemical study were correlated with the blood group of the patients, which was obtained from the clinical charts.
In order to evaluate correlations between the staining results and clinicopathological variables, the chi-square test and the exact test of Fisher were applied at a significant level of 5%.

DNA extraction and PCR
5-10 μm paraffin sections of each case were cut and placed in Eppendorf tubes of 1.5 ml. To avoid contamination, we used disposal knives and the microtome was washed with absolute ethanol between each block. We also followed recognized measures in order to prevent false positives. Sections of formalin-fixed and paraffinembedded tumor specimens were used to extract DNA following the protocol described by Shibata et al. (1988) [32,33].
We dewased in two applications with xylene and absolute ethanol with a volume of 500 or 1000 μL, depending of the size of the tissue. Between each application, we centrifuged at 12,000 g for 5 min. The tisular sediment was resuspended in a buffer solution 1/50 of proteinase K (50 mM tris pH 8.5, 1 mM EDTA, Tween 20 at 0.5%) and the samples were incubated at 37° between 24 to 72 hours, depending on the cases. DNA was purified by phenol:cloroform followed by ethanol precipitation at -20°o vernight (1/10 volume of 3 M Na-acetate pH 5.2 and 2.5 volumes of pure ethanol with subsequent centrifugation at full speed for 30 min at 4°C in a bench-top centrifuge). 125 ngm of nucleic acid was used for each 50 μl PCR reaction with Taq DNA Polymerase (Perkin-Elmer Hispania). Each PCR reaction contained primers for the intron E6 of the HPV-16 or Beta-globin. Beta-globin was used as a control gene. The primers had the following sequence: E6 HPV-16 H1+5'ATTAGTGAGTATAGACATTA3'; H2 -5'GGCTTTTGACAGTTAATACA3' and B-GLOBIN +5'GGTTGGCCAATCTACTCCCAGG3'; -5'GCTCACTCAGTGTGGCAAAG3'. These primers generated intro E6 and Beta-globin products of 109 and 536 base pairs (bp), respectively. DNA for the HPV-16 was amplified for 35 cycles on the thermal cycler Perkin-Elmer Cetus Intruments, Techne. (Table. 1) The 50 μl DNA amplification reactions for HPV-16 con- Before adding the enzyme, we did a hot start at 95°C for 10 min., followed by 35 cycles: 45 sec. at 96°C for denaturalization, 45 sec at 55°C for anneling, and 50 sec. at 72°C for extension, with a step of final extension at 75°C for 10 min. A second PCR amplification was performed in the same conditions with 1 μl of the previous product as a template. The final product was detected in a 3% agarose gel electrophoresis stained with ethidium bromide. Each PCR reaction and the following hybridization contained negative and positive controls. The last one was a CIN-III, positive for HPV-16 by in situ hybridization and PCR amplification of the L1 region.

DNA blot transfer and nonradiactive hybridization detection
For each sample, we added 120 μl of NaOH 0.5 M to 30 μl of the first amplification product and vortex. After 10 min. at room temperature, we added 150 μl of HN 4 -acetate 2 M and it was placed in ice. All this volume was transferred by a filtration process to a nylon membrane embedded in HN 4 -acetate 1 M (Boheringer Mannheim #1209299). After blotting we allowed for drying. The DNA was cross-linked on the blot membrane on a UV transilluminator (320 nm for approx. 40 sec.). Subsequently, we performed membrane hybridization with digoxigenin-11-dUTP-labeled DNA probe specific for the HPV-16 product amplified, followed by chemiluminiscence detection (DIG DNA Labeling and Detection Kit nonradiactive, Boehringer Mannheim). The probe for intron E6 of the HPV-16 had the following sequence: p1: 5'atggaacaacattagaacagcaatacaacaaaccgttgtg3'. (Table 1) The cross-linked membranes were placed in a dry hybridization tube and prehybridized in a 5 ml hybridization solution (5 × SSC; 0.1% N-lauryl sarcosine, Na-salt; 0.02% SDS; blocking regeant) during 30 min. at 42°C. We replaced the hybrization solution with a new 4 ml hybridization solution preheated at 42°C and added 1 ml of 2 pmol of digoxigenin-11-dUTP labeled DNA probe previously denatured at 95°C for 5 min. The tubes were placed in a hybridization for over 16   After withdrawing the excess AMPPD, we placed the membranes in a clean hybridization bag and incubated them at 37°C for 15 min. Finally a film was exposed with the membranes at room temperature for few seconds.
We also dehybridized the membranes and attempted a second hybridization at 49°C. The results of this second hybridization were similar to the first one.
The autoradiographies were scanned and studied by densitometry to generate a curve of grays for each signal. With the control cases we determined a cut-off line to establish semiquantitatively the positivity of the test (Figure 1).

Results
The results are summarized in Tables 2 through 6 and in Figure 2. The eight antigens were expressed in the cytoplasmatic membrane in the case of the squamous epithelium, and in the cytoplasm/mucus of the pseudoglandular lumen, in the case of the simple epithelium. (Table 2, Table 3, Table 4, Table 5, Table 6).
The antigenic expression of the vessel endothelium was useful as an internal control. Generally, we observed conformity among such expressions and the blood group of the patients that already we knew. As in our previous studies, we found a stronger staining and a higher number of positively reacting cells were observed in both normal and neoplastic tissues, stained with monoclonal antibodies [14].
As expected, we obtained polymorphic results, given the expressive variations of the ABH blood group according to the phenotype of each individual. In addition, within each group we found individual variations in relation to the epithelial expression in accordance with their secretor status. On the other hand, the results for the expression of the Lewis antigen group were less polymorphic. In table 6 we have classified the cases according to their secretor character and the expression of the determinant antigens Le a and Le b .

Normal tissues
The squamous exocervical epithelium as well as the simple endocervical epithelium expressed the ABH and Le antigens in the exocervix at the level of the cytoplasmic membrane and in the endocervix in the cytoplasm and mucus of the pseudoglands. Both the squamous and simple epithelium expressed the ABH and Lewis antigens. In the exocervix, the expression was in the membrane of the cytoplasm and, in the endocervix, in the cytoplasm and mucus of the pseudoglands.
PCR -Dot-blotting Figure 1 PCR -Dot-blotting. Results of the amplification from intro E6 of the HPV-16 and its specific hibridization.

Areas of Mature Metaplasia
20 areas of metaplasia were available for the study in the cervix of 6 and 14 patients bearing LG-SIL and HG-SIL respectively. Three cases expressed precursor chains T1-CH and four cases the antigen H. In one of these cases we observed a heterogeneous expression of both antigens. All the patients belonging to blood group A and AB expressed A antigens with the same pattern in their normal tissues and the areas of metaplasias. The same results were seen with Lewis antigens. 11 of 16 metaplasias were positive for Le a and Le b , 11 of 14 expressed Le x and 10 of 13 Le y . In short, we have not found differences in the glycosylation patterns when comparing the normal exocervical epithelium with the areas of mature metaplasia.

LG-SIL
We have not observed relevant differences in the exocervix among the patterns of expression in comparison with the LG-SIL for the ABO and Lewis blood group antigens.
Precursor chains T1-CH and antigen H were not expressed by immunohistochemistry, coinciding with the absence  of expression in the squamous epithelium. We found only one of 10 cases with occasional expression fields for H1. A and B antigenic determinants, when expressed in the LG-SIL, coincided with the blood groups and the pattern of expression seen in the rest of the epithelium.
When the basal stratum was hyperplasic, this was invariably negative for the eight antigens with displacement of the antigenic expression to the most superficial layers. (Figure 4)

HG-SIL
We have observed an increasing expression of precursor chains T1-CH and H antigen among the HG-SIL compared with the epithelia and lesions described up until now. 2 out of 16 cases (12,5%) expressed T1-CH and 5 out of 16 cases (31%) expressed H antigen. The H antigenic expression in HG-SIL was higher in the case of patients belonging to the O group (4/7 of the O group, 1/ 8 of the A group). In the patients in the A group with HG-SIL, we found undoubtful fields of precursor chain expression.
For the A, B, Le a and Le b antigenic determinants, we found significant changes of expression compared with the normal epithelium and the high-degree lesions.
All the cases with A phenotype expressed this antigenic determinant in high-degree lesions, but with a certain tendency towards a heterogeneous pattern of expression and weaker stains. 5 out of 9 cases (55,5%), positive for BG2 in normal epithelia, lost their expression in areas of HG-SIL, some with sharp losses of expression and without a gradual transition from the positive fields. Concerning the B antigenic determinant, we found four cases that expressed this antigen (4/16 25%). Three of these cases showed heterogeneous patterns. Two were patients from the O group and another patient belonged to the AB group. These four cases also expressed the same antigen in the squamous epithelium and, therefore, we cannot consider them aberrant expressions. Among these cases, we also observed focal losses of antigenic expression. ( Figure  5, Figure 6) 7 out of 11 cases -63%-positive for Le a antigen had uncompleted losses of expression of this antigen ( Figure  7a). Five negative cases for Le a did not express this antigen in the normal epithelia 3 was Le (a-b-) and 2 Le (a-b+) phenotype.

PCNA, p53 and PCR amplification of HPV16 (part of the E6 intron)
The PCNA and p53 antigens, when expressed, were seen in a nuclear pattern. In the normal exocervical epithelium, PCNA was observed to be positive in the deepest layers, corresponding with the basal and the first parabasal row. This expression was intense (++) and concerned 5% of the cellularity (index of proliferation). Only occasionally did we find some positive cells for p53.
In the endocervical epithelium, we found neither an expression for PNCA nor of p53, with the exception of  some isolated cells. We have not seen differences of expression for these antigens between the mature metaplasias and normal epithelia.
7 to 10 SIL-HG cases had an increased expression of PCNA with an average proliferative index of 14,7%. This expression for PCNA was limited to the suprabasal cellular hyperplasia. On the other hand, we have not observed an expression for this antigen in those morphologic changes recognized as related with HPV infection (koilocitos, disqueratosis and atypical mitosis). p53, however was found almost positive in some scattered occasional cells.
Only in one case of invasive carcinoma was p53 found over-expressed with 40% positive cell. In this case, PCNA expression presented a proliferative index of 40%, clearly inferior to the other cases.
Amplifying by PCR and with subsequent hybridization in dot-blot, we identified 7 positive cases for the gene E6 of the HPV-16 of 10 SIL-LG; 7 cases of 15 SIL-HG and 7 out of 10 invasive carcinomas (21/35 lesions).

Discussion
We have compared the expression of different isoantigenic determinants of ABH and Lewis blood groups in uterine cervix tissues where we have normal epithelium and a spectrum of preneoplasias and neoplasias, from low-grade lesions (morphologic changes of epithelium related with HPV infection, condylomas and CIN I) to the invasive carcinoma.
These antigens resisted fixation with saline formaline very well. In order to compare the results with the monoclonal antibodies used, we employed different types of lectines: Ulex Europeaus I, Lotus Tetragonolobus, Dolichus Biflorus, and Shopora Japonica Agglutinins [34].
The results of both methods permitted us to observe a better stain definition when we used the MoAbs, although the patterns of stain tended to be similar in both.
This work stands in the fact that the study of cellular glycoconjugates can provide valuable information concerning the events related with neoplastic cellular transformation. Accordingly, much evidence shows that in different neoplasias the phenomena of glycosylation of certain molecules of the cellular membrane show modifications that could play an important role in relation to their malignant behavior [35].
The first works in which we began to observe those alterations date from the beginning of the 1930s [36] (Hir-    colleagues 1968, 1969, with immunofluorescence techniques and mixed cellular reaction agglutination (MCRA), were the first authors to describe that uterine cervix neoplasias showed losses of expression of specific antigens of ABH blood groups. This fact was interpreted as a manifestation of the function of dedifferentiation analogous to morphological dedifferentiation and was related, also from the first studies, to a greater incidence of metastases. The same findings were found in adenocarcinomas of the gastrointestinal tract, ovary, tumors of transitional cells originating in the urinary tract, and in squamous carcinomas of the skin, tongue and larynx [37,17].
Likewise, since then it was recognized that in the sequence: metaplasia, dysplasia and carcinoma in situ of the cervix, a good model is reproduced in which the alterations of the isoantigen ABH expression can be followed.
Several authors have confirmed the findings of Kovarik and Davidsohn in the cervix [18,36,38]. Bonfiglio and colleagues were the first in publishing a work using, in addition to the immunofluorescence technique, an immunoperoxidase technique for the determination of the A and B isoantigens [39].
Mambo observed that 33% of the condyloma acuminatum of the uterine cervix had complete losses of some of the blood group ABH isoantigens, and partial losses in 47% [40]. We could not confirm those findings, although the number of cases in our study is smaller and our objective was not centred on low-grade lesions.
The pattern of expression of the ABH antigens in the middle of the epithelium that we have observed was previously described by Stubbe Teglbjang and colleagues using immunofluorescence. These authors found that the exocervical epithelium had a progressive pattern in the expression of the blood group isoantigens, with expressions in the basal and parabasal cells of N-Acetyllactosamine, of the antigen H in the parabasal and in the cells of the low spinosus stratum, and of the A antigen in cells of the high spinous stratum. They also observed that the premalignant and malignant lesions had irregular decreases in the content of the antigens A and H, with an almost complete loss of them, and precursor chain accumulation [41].
Following a technique using immunohistochemistry with the avidin-biotin complex, we have not been able to observe a clear accumulation of precursor chains although we have found significant results in the loss of expression of the A antigen.
To and colleagues have described that the expression of the blood group antigens, in combination with the depth of the tumoral invasion, could serve as a predictor of sur-vival in cervical carcinoma [42]. In the same direction, Sakamoto and colleagues observed a greater frequency of complete loss of the antigen H in small cell carcinomas, which have also a worse survival to the two years, versus to the squamous carcinomas with a better survival [43].
The sum of this evidence has led to us to study, after our experience with the respiratory mucosa [13,14], the alterations of the glycosylation in different cervical lesions. Our findings are very similar to those from our previous studies, leading us to believe that these alterations could be a phenomenon common in different tumors with different etiological implications.
Our work, unlike its precedents, is complemented with the use of lectins, a tool that has served to control the results by means of immunohistochemistry to discard On the other hand, HPV infection has been strongly related to cancers of the lower female genital tract, penis, larynx, lungs and anus. It is known that the subtypes 16, 18, 45 and 56 in cervical lesions are related to a greater relative risk of developing invasive carcinomas [44,45].
Some papers point out that the percentage of PCNA positive proliferation cells is significantly higher in premalignant and malignant lesions of the uterine cervix than in nonneoplastic lesions [46].
To investigate the possible differential expression of PCNA in negative and positive lessions of the cervix for E6 HPV-16 lesions, we have studied, using immunohisto-chemistry and PCR, different cases from flat condylomas to invasive caricinomas. We also inquired into the overexpression of p53 in order to explain the cases with a possible alteration of this tumor supressor gen.
We did not find a significant expression of p53 as other authors have reported [47]. Only occasional basal or parabasal nucleus could be seen as positive with the DO-7 MoAB clone. However PCNA was expressed more progressively from a lower grade to the higher degrade and invasive carcinomas. Our results are consistent with the observations that the cell proliferating index as detected immunohistochemically using PCNA may be a useful parameter to indicate the degree of CIN. In addition, we found in our study a higher PCNA proliferation index in CIN III and invasive carcinomas of cervix positive for HPV-16 versus the cases that are negative for this type of virus.

Conclusion
To sum up, the results and conclusions of this paper are as follows: (1) In the invasive cervical carcinomas, we observed a loss of expression of the Le x antigen (p < 0.01).
(2) Almost significant differences were observed when comparing the expression of Le a , and blood group A antigens in squamous epithelium and invasive carcinomas (p < 0.1). (3) With regard to the progression of the different lesions studied, we found alterations in the patterns of expression of the antigens of the ABH and Lewis blood groups. There was a tendency towards loss of expression and heterogeneous patterns in the more advanced lesions, and also overexpression of the Ley antigens. Expression of PCNA in a 100% of the cellularity of an invad-ing squamous carcinoma Figure 8 Expression of PCNA in a 100% of the cellularity of an invading squamous carcinoma.