J Clin Pathol: Mol Pathol 1997;50:317-321

Double immunostaining for p53 and molecularchaperone hsp72/73 in gastric carcinoma

Miguel A Villaseca, Ivan Roa, Juan C Araya, Juan C Roa, Placido Flores

AbstractAims-To examine the relation betweenthe expression of p53 protein and thechaperone heat shock protein (hsp)72/73in a population at high risk for gastriccarcinoma, using single and double im-munohistochemistry, and to compare theexpression ofthese two proteins with clin-icopathological features.Methods-Monoclonal antibodies wereused to investigate the expression of p53protein and hsp72/73 in 46 human gastriccarcinomas. A double immunohisto-chemical technique was used in cases thatshowed p53/hsp72/73 coexpression.Results-p53 immunoreactivity waspresent in 11 tumours (24%), and hsp72/73immunostaining was observed in 22 cases(48%). p53 expression was observed asnuclear staining in tumoral cells. hsp72/73expression was demonstrated mainly ascytoplasmic staining, but six tumours alsoshowed focal weak nuclear staining. Sevencases showed p53 and hsp72/73 coexpres-sion with immunoreactivity for both pro-teins in the same neoplastic cells, three ofthem with focal areas of nuclear co-expression. p53 expression was seen morefrequently in cases that showed a highintensity (+++) of hsp72/73 staining. Nosignificant association was observed be-tween the expression of the two proteinsand clinicopathological features.Conclusions-More than half of our casesmay have some impairment in p53 proteingrowth suppressive function, as a result ofp53 gene alterations or complex forma-tion. The positive correlation between p53expression and intensity of hsp72173 sup-ports the postulate of a p53 regulatingfinction for the chaperone hsp72/73. Ahigh intensity of hsp72/73 immunohisto-chemical staining could be used as an indi-rect marker ofp53 gene abnormalities.( Clin Pathol: Mol Pathol 1997;S0:317-321)

Keywords: p53; heat shock protein 72/73; gastric cancer

Gastric cancer is one of the most prevalentmalignancies in the world, but there is substan-tial variation in the incidence and mortality rateof the disease among different regions orcountries.1 2 Chile is one of the countries withthe highest reported frequencies.' 3 There isincreasing evidence that cancers arise as theresult of an accumulation of genetic alterations

that interfere with normal control of cellgrowth and differentiation. Tumour suppressorgenes are normal cellular genes, which wheninactivated lead to a dysregulation of both thecell cycle and apoptosis. These disturbancesmay facilitate the development of neoplasia.4Cancer cells differ from normal cells in manyimportant characteristics including loss ofdifferentiation, increased invasiveness, anddecreased drug sensitivity. There is evidencethat conversion of normal epithelial cells tocancerous cells is a multistep process thatrequires the accumulation of multiple geneabnormalities affecting DNA repair genes,oncogenes, and tumour suppressor genes.56Gastric cancer is no exception and it displaysmultiple gene alterations involving oncogenes,growth factor or cytokine genes, cell cycleregulatory genes, tumour suppressor genes,and cell adhesion molecule genes. These genealterations may lead to genetic instability.7The p53 tumour suppressor gene is located

on the short arm of human chromosome 17 atposition 17p 13.1. This gene is composed of 11exons, the first of which is non-coding, and islocalised 8-10 kb away from exons 2-11 89The product of p53 gene is a 375 amino acid(with a molecular mass of - 53 kDa) nuclearphosphoprotein, which was first identified as acellular protein in 1979 because it formed atight complex with the SV40 large T antigen.'0The p53 protein was found in very low quanti-ties in normal cells, but large quantities of p53(5-100-fold) could be detected in transformedcells in culture and in human tumours. Loss ofnormal p53 function could be achieved in avariety ofways including genetic changes in thep53 gene (such as mutation, deletion, or struc-tural rearrangements), formation of proteincomplexes with viral oncoproteins (such asSV40 large T antigen, adenovirus E1B, andpapillomavirus E6), and binding to cellularproteins (such as MDM2)." p53 mutations arenow recognised as being one of the most com-mon cancer related genetic changes seen at thegene level.6 Mutant p53 protein is thought toact by formation of pseudohomodimers ofwild-type and mutated p53 which abrogate thegrowth suppressive activity of the normalprotein.'2 A further loss of proliferative controlmay occur when the wild-type allele is deletedfrom the cells." However, genetic alterations ofthe p53 gene are not the only mechanism bywhich the p53 protein can be stabilised. Thenormal protein can be stabilised by the actionof viral and cellular gene products." There isevidence that mutated and probably wild-type

Pathology Unit, Schoolof Medicine,University ofLaFrontera, PO Box54-D, Temuco, ChileM A VillasecaI RoaJ C ArayaJ C Roa

Department ofSurgery, School ofMedicine, Universityof La FronteraP Flores

Correspondence to:Dr Villaseca.

Accepted for publication23 September 1997

317

on October 1, 2021 by guest. P

rotected by copyright.http://m

p.bmj.com

/M

ol Path: first published as 10.1136/m

p.50.6.317 on 1 Decem

ber 1997. Dow

nloaded from

Villaseca, Roa, Araya, Roa, Flores

Figure 1 Nuclear p53 immunoreactivity in a moderately differentiatedgastric adenocarcinoma. Streptavidin-biotin peroxidase reaction withhaematoxylin counterstaining. (Original magnification x 400.)

p53 protein may form complexes vheat shock proteins (hsp), suchisoforms. 4Heat shock proteins are produced

exposure to stressful conditions suc]perature shock, chemicals, and otherenvironmental stress.'5 16 They act aschaperones that bind to denaturated opriately folded proteins, assisting thenon-covalent assembly, although thecomponents of the functional;structures."7 hsp7O concentrates in nucheat shock and returns to the cytopl;the shock is removed. Although hspmembers bind to mutant p53 in tracells, it has been proposed alsoformation of complexes with wild-typ(tein may regulate or interfere with its iThis work evaluates the expressic

protein and hsp72/73 in a high risk pfor gastric carcinoma, using single ar

Figure 3 Weak nuclear hsp72173 immunostaining in a moderately differentiatAadenocarcinoma. Streptavidin-biotin peroxidase reaction with haematoxylincounterstaining. No significant association was observed between the expressionproteins and clinicopathological parameters such as tumour location, histologicalhistological grade, Lauren's type, and nodal status (table 1). However, cardial ttmore frequently positive for hsp72173 than distal tumours (p = 0. 08), mainly innegative subgroup. In addition, hsp72173 was observed more frequently in male(p = 0. 06) and in early gastric carcinomas (p < 0.15). (Original magnificatio

Figure 2 Cytoplasmic hsp72173 immunoreactivity in a neoplastic gastricepithelium. Streptavidin-biotin peroxidase reaction with haematoxylincounterstaining. (Original magnification x 1000.)

vith some immunohistochemistry, and compares it withas hsp7O clinicopathological features.

after cell Methodsh as tem- A random sample of 46 surgically resected gas-r forms of tric carcinoma patients was selected from ourmolecular registry of tumoural diseases. Representative,r inappro- inclusions of formalin fixed, paraffin wax!ir correct embedded tissue were obtained from they are not archives of the department of pathology,assembled Temuco's Regional Hospital, Temuco, Chile.clei during The sample included five early gastric cancersasm when and 41 advanced gastric carcinomas. The)70 family mean age of the patients was 62 years (rangensformed 34-89) and the sex ratio was 1.4:1 (male:fe-that the mal). Surgical specimens were processed ac-ep53 pro- cording to the general rules of the Japanesefunction."4 Gastroenterology Society,'8 and the tumoursrn of p53 were classified according to WHO histological'opulation criteria'9 and Lauren's histological types.20nd double

IMMUNOHISTOCHEMISTRYp53 and hsp72/73 expression was detectedusing commercially available monoclonal anti-bodies. Immunohistochemical studies wereperformed by means of streptavidin-biotinmethodology. Briefly, 4 gm thick paraffin waxsections were dewaxed and subjected to micro-wave antigen retrieval by immersion in citratebuffer (pH 6.0) at 800 W. Endogenous peroxi-dase was blocked with 3% hydrogen peroxidefor 10 minutes and sections were then treatedwith 20% normal goat serum for 20 minutes.Antigen immunoreactivity was detected byincubation with primary antibodies directedagainst p53 at 1/100 dilution (clone D01;Oncogene Science, New York, USA), andhsp72/73 at 1/100 dilution (clone W27; Onco-gene Science), for one hour at room tempera-ture, according to the manufacturer's instruc-tions. Sections were then washed andincubated with a biotinylated secondary anti-

edgastric body (Multilink; Biogenex, San Ramon, Cali-of the two fornia, USA). After incubation withtype, streptavidin-biotin peroxidase complex (Bio-umours were genex) for 20 minutes, sections were incubatedi the p53patients in 0.05% 3,3-diaminobenzidine tetrahydro-mnx 1000.) chloride (DAB) (Sigma Chemicals, St Louis,

318

on October 1, 2021 by guest. P

rotected by copyright.http://m

p.bmj.com

/M

ol Path: first published as 10.1136/m

p.50.6.317 on 1 Decem

ber 1997. Dow

nloaded from

Immunostainingforp53 and hsp72173 in gastric carcinoma

_

1.

A%

Figure 4 pS31hsp72173 double immunostaining in a moderatelydifferentiated gastric adenocarcinoma. Note the nuclear p53immunoreactivity (brown) and the cytoplasmic and nuclear hsp72173immunostaining (red). Streptavidin-biotin peroxidase reaction (brown)and streptavidin-biotin alkaline phosphatase reaction (red) withhaematoxylin counterstaining. (Original magnification x 1000.)

Missouri, USA). Finally, the sectcounterstained weakly with haeiPositive controls were sections from 1cinoma known to express p53 oncophsp72/73. Negative controls were inincubation and development of sectonly normal blocking serum withoition in primary antibody.

DOUBLE IMMUNOHISTOCHEMISTRYAll cases that showed both p53 andimmunostaining were subjected timmunostaining by means of a sequcble labelling technique, using twodetection systems. Briefly, sections w(first for p53 protein, using the strbiotin peroxidase detection system, asation with DAB (brown) as descril

Table 1 Clinicopathologicalfeatures, p53 andimmunoreactivity in human gastric carcinoma

Clinicopathologicalfeatures p53 hs

SexMale 7/27 (26) 1Female 4/19 (21)

Age< 50 years 0/7¢ 50 years 11/39 (28) 2

Tumour locationCardial third 4/17 (24) 1Medial third 3/9 (33)Antral third 4/20 (20)

Wall stageEarly carcinoma 2/5 (40)Advanced carcinoma 9/41 (22) 1

Histological typeTubular adenocarcinoma 9/33 (27) 1Signet ring cell carcinoma 1/9 (11)Mucinous adenocarcinoma 0/2Undifferentiated carcinoma 1/2 (50)

Histological gradeWell/moderately 2/13 (15)Poor 9/33 (27) 1

Lauren typeIntestinal 5/21 (24) 1Diffuse 6/25 (24) 1

Lymph node statusNegative 2/10 (20)Positive 8/34 (24) 1

Values are n (%). *p = 0.06.

Figure 5 p531hsp72173 double immunostaining in lymphatic tumourpermeation. Note nuclear p53 (brown)1hsp72173 (red) coexpression(arrows). Streptavidin-biotin peroxidase reaction (brown) andstreptavidin-biotin alkaline phosphatase reaction (red) with haematoxylincounterstaining. (Original magnification x 1000.)

ions were ously. Sections were then washed and stainedmatoxylin. for hsp72/73 using the streptavidin-biotin alka-breast car- line phosphatase detection system and visuali-rotein and sation with fast red (red).cluded by All immunostained slides were analysed andions using scored in a blind fashion by two differentut incuba- observers. Staining intensity was graded ac-

cording to the numbers of positive cells as fol-lows: fewer than 5% of tumour cells positive(-); 5-25% of tumour cells positive (+);26-50% of tumour cells positive (++); greater

I hsp72/73 than 50% of tumour cells positive (+++). Sta-to double tistical analysis was undertaken using theential dou- Student's t test, Fisher's exact test, and

different Pearson's correlation test.

ere stainedeptavidin-nd visuali- Resultsbed previ- p53 immunoreactivity was present in 11 (24%)

tumours and hsp72/73 immunostaining was

observed in 22 (48%) cases (table 1). p53Ihsp72173 expression was seen as nuclear staining in

tumoural cells (fig 1). Nuclear positivity was notseen in the non-cancerous mucosa adjacent to

p72173 the carcinoma. hsp72/73 expression was dem-onstrated mainly as a fine granular cytoplasmic

6/27 (59)* staining pattern in neoplastic cells (fig 2). Six6/19 (32) tumours (13%) also showed focal weak nuclear2/7 (29) staining (fig 3). All cases that showed nuclear

t0/39 (51) hsp72/73 expression were also p53 positive..1/17 (65) Cytoplasmic hsp72/73 was also detected in4/9 (44) non-neoplastic mucosa adjacent to the tu-

7/20 (35) mours, mainly in intestinal metaplasia areas.

4/5 (80) Seven cases (15%) showed p53 and.8/41 (44) hsp72/73 coexpression with immunoreactivity5/33 (45) for both proteins in the same neoplastic cell,5/9 (56)1/2 (50)1/2 (50) Table 2 hsp72173 immunostaining intensity according to

p53 expression in human gastric carcinoma8/13 (62)4/33 (42)

[1/21 (52)1/25 (44)

4/10 (40)17/34 (50)

hsp72173

p53 (-) (+) (++) (+++)

Negative 20 (83) 5 (83) 9 (82) 1 (20)Positive 4 (17) 1 (17) 2 (18) 4 (80)

Values are n (%).p < 0.02, Pearson's correlation value: r = +0.31.

319

on October 1, 2021 by guest. P

rotected by copyright.http://m

p.bmj.com

/M

ol Path: first published as 10.1136/m

p.50.6.317 on 1 Decem

ber 1997. Dow

nloaded from

Villaseca, Roa, Araya, Roa, Flores

three of them (7%) with focal areas of nuclearcoexpression (figs 4 and 5). An associationbetween intensity of hsp72/73 and p53 expres-sion was seen (table 2). Eighty per cent of casesshowing high (+++) hsp72/73 intensity werep53 positive and only 17% of tumours thatshowed a lower intensity or were negative forhsp72/73 were p53 positive (p < 0.02; Pear-son's correlation value: r = +0.31). A total of26 cases (57%) showed expression of p53and/or hsp72/73.

DiscussionSeveral gene alterations have been found to beimportant in the upper digestive tract includingthose affecting multiple oncogenes, tumoursuppressor genes, cell cycle regulator genes,and DNA repair genes.2' 22 Altered expressionand amplification of the c-met gene, inactiva-tion of the p53 gene, and abnormal CD44transcripts are common events in gastriccarcinoma.2' Tahara has demonstrated thatgene changes differ in relation to the two histo-logical types, because intestinal or differenti-ated types and diffuse or undifferentiated typesof carcinoma may develop by different geneticpathways.7 21 Genetic alterations of the p53gene are found frequently in both types of gas-tric carcinomas. These alterations lead to p53protein overexpression and allow the formationof complexes with some heat shock proteins,such as hsp7O isoforms." 14

In the present study, although we found p53expression in only 24% of cases, almost half ofthe tumours showed hsp72/73 expression.Thus, 57% of cases may have some impairmentof p53 protein growth suppressive function, asa result of p53 gene alteration or complex for-mation. In cases that were negative for p53protein and positive for hsp72/73, the chaper-one may have exerted a wild-type p53 blockingeffect that could also have given a proliferativeadvantage to malignant cells. Both the observa-tion that hsp72/73 is expressed in non-neoplastic mucosa (particularly in metaplasticareas) and the trend towards higher chaperoneexpression in early gastric carcinomas suggestthat hsp72/73 induction may be related to earlyevents in the multistep carcinogenic modelproposed by Correa.2' This may represent anearly non-genetic p53 protein dysfunction.Polymorphonuclear leucocyte chemotaxis andactivation by Helicobacter pylori (with their oxi-dative effects) may play a role in the early stagesof gastric carcinogenesis. These are knowncauses of DNA damage and hypotheticallycould induce mutations in replicating gastricepithelial cells.24 To our knowledge, there areno studies on Hpylori status and hsp70 isoformexpression in gastric epithelial lesions. Thesehypotheses must be studied by means ofgenetic analysis of p53 genes in relation to hspexpression analysis, by cell proliferation assess-ment, and by Hpylori status assessment in earlygastric carcinoma and preneoplastic lesions.The main finding of our study is that of the

positive correlation of p53 expression andintensity of hsp72/73 staining, concordant withthe observation that hsp72/73 binds to mutant

p53 protein. Thus, hsp72/73 expression maybe used as an indirect marker of p53 geneabnormalities. This association also suggeststhat the unfavourable prognostic outcomedescribed for hsp expression in some tumours25may be related to the alteration of associatedp53. Thus, hsp chaperones could be a p53dependent prognostic factor. In the literature,Elledge et al found no correlation betweenhsp70 expression and p53 accumulation inbreast carcinoma,26 whereas Chant et al didfind such an association in acute myeloidleukemia.27 Recently, some additional carcino-genic hsp mediated mechanisms have beenpostulated. hsp chaperones may protect certaintumour cells from tumour necrosis factormediated cytotoxicity by interfering with thesignal transduction pathway leading to theactivation of phospholipase A2.2,

In conclusion, the accumulation ofhsp72/73may be related to the early stages of gastric car-cinogenesis. The association of p53 andhsp72/73 accumulation supports the postulateof a p53 regulating function for the chaperonehsp72/73.

This work was supported by grants from the National Fund forScience and Technology, Fondecyt No. 1950643 and 1970874.

1 Yamagata S, Hisamichi S. Epidemiology of cancer of thestomach. WorldJ Surg 1979;3:663-9.

2 Csendes A, Smok G, Medina E, Salgado I, Rivera R, QuitralM. Caracteristicas evolutivas del cancer gastrico 1958-1990. Rev Med Chile 1992;120:36-42.

3 Medina E. Las enfermedades digestivas en Chile: panoramaepidemiol6gico. Rev Med Chile 1988;1 16:282-8.

4 Vogelstein B, Kinzler KW. The multistep nature of cancer.Trends Genet 1993;9:138-41.

5 Vogelstein B, Fearon ER, Hamilton SR, Kern SE, PreisingerAC, Leppert M, et al. Genetic alterations during colorectal-tumor development. NEnglJtMed 1988;319:525-32.

6 Harris CC. 1995 Deichmann lecture-p53 tumor suppres-sor gene: at the crossroads of molecular carcinogenesis,molecular epidemiology and cancer risk assessment. ToxicolLett 1995;82183:1-7.

7 Tahara E. Molecular mechanism of stomach carcinogenesis.J Cancer Res Clin Oncol 1993;119:265-72.

8 McBride 0, Merry D, Givol D. The gene for human p53cellular tumor antigen is located on chromosome 17 shortarm (17p 13). Proc NatlAcad Sci USA 1986;83:130-4.

9 Miller C, Mohandas T, Wolf D, Prokocimer M, Rotter V,Koeffler PH. Human p53 localized to short arm ofchromosome 17. Nature 1986;319:783-4.

10 Lane DP, Crawford LV. T antigen is bound to a host proteinin SV40-transformed cells. Nature 1979;278:261-3.

11 Chang F, Syrjanen S, Tervahauta A, Syrjanen K. Tumouri-genesis associated with the p53 tumour suppressor gene. BrJ Cancer 1993;68:653-61.

12 Stenger JE, Mayr GA, Mann K, Tegtmeyer P. Formation ofstable p53 homotetramers and multiples of tetramers. MolCarcinogen 1992;5:102-6.

13 Vogelstein B, Kinzler KW. p53 function and dysfunction.Cell 1992;70:523-6.

14 Lane DP, Midgley C, Hupp T. Tumour suppressor genesand molecular chaperones. Philos Trans R Soc Lond Biol1993;339:369-72.

15 Beckmann R, Lovett M, Welch W. Examining the functionand regulation of HSP70 in cells subjected to metabolicstress.J CellBiol 1992;117:1 137-50.

16 Macario A. Heat-shock proteins and molecular chaperones:implications for pathogenesis, diagnostics, and therapeu-tics. IntJ7 Clin Lab Res 1995;25:59-70.

17 Beckmann R, Mizzen L, Welch W. Interaction of Hsp7Owith newly synthesized proteins: implications for proteinfolding and assembly. Science 1990;248:850-4.

18 Japanese Gastroenterology Society. The general rules for thegastric cancer study in surgery andpathology, 12th edn. Tokyo:Japanese Research Society for Gastric Cancer, 1993.

19 Watanabe H, Jass JR, Sobin LH, eds. Histological typing ofoesophageal and gastric tumours, 2nd edn. Geneva: WorldHealth Organization (International HistologicalClassification ofTumours), 1990.

20 Lauren P. The two histological main types of gastriccarcinoma: diffuse and so-called intestinal type carcinoma.An attempt at a histoclinical classification. Acta PatholMicrobiol Scand 1965;64:31-49.

21 Tahara E. Genetic alterations in human gastrointestinalcancers. The application to molecular diagnosis. Cancer1995;75:1410-7.

320

on October 1, 2021 by guest. P

rotected by copyright.http://m

p.bmj.com

/M

ol Path: first published as 10.1136/m

p.50.6.317 on 1 Decem

ber 1997. Dow

nloaded from

Immunostaining for p53 and hsp72173 in gastric carcinoma

22 Wright PA, Quirke P, Attanoos R, Williams GT. Molecularpathology of gastric carcinoma: progress and prospects.Hum Pathol 1992;23:848-59.

23 Correa P. A human model of gastric carcinogenesis. CancerRes 1988;48:3554-60.

24 Correa P. Is gastric carcinoma an infectious disease? N EnglJ'Med 1991;325:1170-1.

25 Ciocca DR, Clark GM, Tandon AK. Heat shock proteinHSP 70 in patients with axillary lymph node-negativebreast cancer: prognostic implications. Natl Cancer Inst1993;85:570-4.

26 Elledge RM, Clark GM, Fuqua SA, Yu YY, Allred DC. p53protein accumulation detected by five different antibodies:relationship to prognosis and heat shock protein 70 inbreast cancer. Cancer Res 1994;54:3752-7.

27 Chant ID, Rose PE, Morris AG. Susceptibility ofAML cellsto in vitro apoptosis correlates with heat shock protein 70(hsp 70) expression. Brj Haematol 1996;93:898-902.

28 Jaattela M. Overexpression ofmajor heat shock protein HSP70 inhibits tumor necrosis factor-induced activation ofphospholipase A2.J Immunol 1993;151 :4286-94.

321

on October 1, 2021 by guest. P

rotected by copyright.http://m

p.bmj.com

/M

ol Path: first published as 10.1136/m

p.50.6.317 on 1 Decem

ber 1997. Dow

nloaded from