p53 status correlates with the differential expression of the DNA mismatch repair protein MSH2 in non-small cell lung carcinoma

Authors

  • George Xinarianos,

    1. Molecular Oncology Unit, Roy Castle International Centre for Lung Cancer Research, Liverpool, United Kingdom
    2. Molecular Genetics and Oncology Group, Department of Clinical Dental Sciences, University of Liverpool, Liverpool, United Kingdom
    Search for more papers by this author
  • Triantafillos Liloglou,

    1. Molecular Oncology Unit, Roy Castle International Centre for Lung Cancer Research, Liverpool, United Kingdom
    2. Molecular Genetics and Oncology Group, Department of Clinical Dental Sciences, University of Liverpool, Liverpool, United Kingdom
    Search for more papers by this author
  • Wendy Prime,

    1. Molecular Oncology Unit, Roy Castle International Centre for Lung Cancer Research, Liverpool, United Kingdom
    2. Department of Pathology, Medical School, University of Liverpool, Liverpool, United Kingdom
    Search for more papers by this author
  • George Sourvinos,

    1. Laboratory of Virology, Medical School, University of Crete, Heraklion, Crete, Greece
    Search for more papers by this author
  • Andreas Karachristos,

    1. Laboratory of Virology, Medical School, University of Crete, Heraklion, Crete, Greece
    Search for more papers by this author
  • John R. Gosney,

    1. Department of Pathology, Medical School, University of Liverpool, Liverpool, United Kingdom
    Search for more papers by this author
  • Demetrios A. Spandidos,

    1. Laboratory of Virology, Medical School, University of Crete, Heraklion, Crete, Greece
    Search for more papers by this author
  • John K. Field

    Corresponding author
    1. Molecular Oncology Unit, Roy Castle International Centre for Lung Cancer Research, Liverpool, United Kingdom
    2. Molecular Genetics and Oncology Group, Department of Clinical Dental Sciences, University of Liverpool, Liverpool, United Kingdom
    • Roy Castle International Centre for Lung Cancer Research, 200 London Road, Liverpool, Merseyside L3 9TA, UK
    Search for more papers by this author
    • Fax: +44-151-794 8989


Abstract

We examined the p53 status of 108 NSCLCs compared to the expression of MLH1 and MSH2 proteins. p53 overexpression was demonstrated by IHC in 64% of patients examined, whereas p53 mutations were detected in 43%. Twenty-two percent of mutations were located outside of the hot-spot (exons 5–8) area. p53 mutations and overexpression were more frequent in SCCL (57% and 73%, respectively) than in lung adenocarcinomas (22% and 50%, respectively). In NSCLC-carrying wild-type p53, increased expression of MSH2 correlated with p53 overexpression (p = 0.018). In addition, in SCCL, p53 mutations correlated with reduced MSH2 expression (p = 0.019). These data suggest a relationship between p53 and MSH2. While there is evidence for p53 being a transcriptional activator of MSH2, the hypothesis that MSH2 acts as a DNA-damage signaller triggering p53 overexpression needs to be clarified in future studies. © 2002 Wiley-Liss, Inc.

The p53 gene codes for a 53 kDa nuclear phosphoprotein, which holds a key position in the complex network controlling genome stability, the cell cycle and apoptosis.1 Inactivation of the p53 gene is common in human tumours,2, 3, 4, 5 including NSCLC.6, 7, 8, 9, 10 A possible role of p53 inactivation as a prognostic marker in lung cancer has been suggested.8, 10 The majority of the p53 mutations described have been detected within exons 5–8, which encode 88% of the DNA-binding domain of the protein.5 However, there is considerable evidence suggesting that p53 mutations and their biologic effects may have been underestimated since the whole coding region (exons 2–11) of the p53 gene has not been extensively investigated. It has been demonstrated that p53 mutations outside of exons 5–8 account for approximately 20% of the total number of mutations.11, 12, 13, 14, 15

Inactivation of the DNA MMR machinery is associated with the pathogenesis and predisposition of certain malignancies through a mutator phenotype.16 A possible role of hMLH1 and hMSH2 overexpression in the induction of apoptosis has also been suggested.17 We have demonstrated that reduced expression of these 2 genes is a frequent event in NSCLC.18 It has also been demonstrated that LOH at the DNA MMR loci is a frequent genetic event in lung cancer.19, 20

p53 levels increase in response to DNA damage, and this is mainly accomplished through posttranslational modifications.21 The DNA-PK, ATM and ATR genes are among the known signallers of DNA damage to p53. However, little is known about the relation of MMR genes and p53. The latter can bind in vitro to the promoter region elements of hMSH222 and synergise with c-Jun in regulating MHS2 expression.23 Further support for an MSH2 expression regulatory role of p53 comes from an association between p53 mutations and MSH2 downregulation in human neoplasms.24, 25

We investigated the status of p53 in NSCLC cases from a population in northwest England (Merseyside). In addition, we examined the possible relations of p53 with 2 DNA MMR genes, MLH1 and MSH2, to address questions regarding possible crosstalk between cell-cycle regulation and MMR DNA repair.

Abbreviations

APES, 3′aminopropyl triethoxysilane; HA, heteroduplex analysis; IHC, immunohistochemistry; MAb, monoclonal antibody; MMR, mismatch repair; NSCLC, non-small cell lung carcinoma; SCCL, squamous cell carcinoma of the lung; SSCP, single-strand conformation polymorphism

MATERIAL AND METHODS

Patients and tissue samples

Lung tumour tissue samples were obtained from 108 patients, 41 males and 67 females, who underwent surgery in the Cardiothoracic Centre of Broadgreen, Liverpool (Merseyside, UK). Patient ages ranged between 41 and 95 years (median = 63). The histology of the specimens included in this investigation was as follows: 36 adenocarcinomas, 58 SCCs, 7 adenosquamous carcinomas, 5 large cell carcinomas and 2 unclassified NSCLCs. Smoking history (daily consumption, current status) was available for 108 individuals: 58 current smokers, 12 recently stopped smokers (1–4 years prior to presentation), 30 former smokers (≥5 years prior to presentation) and 8 nonsmokers. Complete data for calculating total smoking exposure were available for 98 smokers. Total smoking exposure is expressed in pack-years:

equation image

The patients' pack-years ranged from 16 to 177 (median = 77).

DNA extraction and PCR amplification

Frozen tumour tissue specimens were available from 108 individuals. Five 10 μm sections of each sample were microdissected to ensure the presence of >75% tumour cells. DNA was extracted from microdissected sections as previously described.20

PCRs (25 μl) were performed, containing 100 ng of genomic DNA, 200 μM of each dNTP, 50 mM KCl, 10 mM TRIS-HCl (pH 8.8), 0.1% Triton X-100, 1.5 mM MgCl2, 6 pM of each primer and 0.6 U BIOPRO polymerase (Bioline, London, UK). Samples were subjected to 35–37 cycles of amplification. Amplification parameters and primers are available upon request.

p53 SSCP and HA

SSCP and HA were primarily used to screen for p53 mutations. SSCP and HA analyses were performed as previously described.26 For all specimens demonstrating altered SSCP/HA electrophoretic patterns, an independent PCR of genomic DNA was performed with subsequent sequencing analysis.

Sequencing

PCR products were used for subsequent sequencing after cleaning with the QIAquick PCR Purification Kit (Qiagen, West Sussex, UK). Sequencing was performed using the ABI PRISM BigDye Terminator Cycle Sequencing Kit and analysis on a 377 sequencer (Applied Biosystems, Warrington, UK) using the supplier's protocol.

IHC analysis of p53

Protein expression was demonstrated by IHC using a modified avidin-biotin complex method. Formalin-fixed paraffin process tissues were available from 105 individuals. They were sectioned at 4 μm thickness, mounted on APES-coated slides and dried at 37°C overnight. After deparaffinization, slides were incubated in the primary antibody buffer (5% goat serum in PBS) for 20 min. The DO-7 MAb against p53 (Serotec, Bicester, UK) was diluted 1:20 in the primary antibody buffer and incubated for 1 hr at room temperature. The primary antibody was visualised and the signal developed as previously described.21 Normal mouse IgG replaced the primary antibody as a negative control. p53 nuclear staining was scored as normal (absent or weak staining) or abnormal (strong staining) following previously described criteria.27, 28

Statistical analysis

Fisher's exact and χ2 (Pearson's correlation) tests were employed to comparatively analyse the molecular and clinicopathologic data tables. Student's t-test was performed to investigate the relation of molecular data to continuous parameters such as age and pack-years. Analysis was performed using the SPSS (Chicago, IL) 10.0 software for Windows.

RESULTS

Mutational analysis of the p53 gene in NSCLC

Forty-seven mutations were detected in 46 of the 108 (43%) samples examined (Table I). To validate the sensitivity of SSCP and HA primary screening results, 15 of the SSCP/HA-negative tumour samples were chosen at random and sequenced for all exons, and no further mutations were revealed.

Table I. p53 Mutations in NSCLC
NumberDiagnosisExon (codon)Codon changeaa changep53 IHCPack-yearsT statusN statusAge (years)
  1. AdenoCa, adenocarcinoma; AdenoSq, adenosquamous; LCCL, large cell carcinoma of the lung; SCCL: squamous cell carcinoma of the lung; ND, not done; NA, not available.

L145SCCL10 (337)CGC-CCCArg-ProAbnormal562077
L151SCCL10 (364)GCT-CCTAla-ProAbnormal552070
L127SCCL4 (58)CCA-ACAPro-ThrNormal402272
L052LCCL5 (158)g delFrameshiftNormal522167
L003SCCL5 (136)CAA-TAAGln-StopNormal1042267
L041SCCL5 (158)CGC-CCCArg-ProAbnormal512065
L086AdenoCa5 (159)CGC-CTGArg-LeuAbnormal1232067
  5 (163)TAC-TGCTyr-Cys     
L179SCCL6 (188)TCT-TTTSer-PheAbnormal353257
L112SCCL5 (165)CAG-TAGGln-StopNormalNA2074
L172SCCL5 (179)CAT-CGTHis-ArgAbnormal942166
L055SCCL5 (163)TAC-TGCTyr-CysAbnormal702068
L129SCCL5 (181)CGC-CCCArg-ProAbnormalNA2174
L143AdenoCa5 (184)GAT-AATHis-AsnAbnormal352156
L159AdenoCa6 (218)GAA-TAAGlu-StopNormal322048
L152SCCL6 (188)TCT-TTTSer-PheAbnormal352056
L014SCCL6 (195)ATC-TTCIle-PheAbnormal1102057
L093SCCL6 (204)GAG-TAGGlu-StopAbnormal1742076
L124AdenoCa6 (216)GTG-ATGVal-MetAbnormal942065
L173SCCL6 (218)GTG-GAGVal-GluAbnormalNA2063
L021AdenoSq7 (229–235)19 bp delTruncationNormal392059
L087AdenoCa7 (239)AAC-GACAsn-AspAbnormal342072
L035SCCL7 (243)g delFrameshiftNormal602148
L034AdenoCa7 (245)GGC-TGCGly-CysAbnormal1122073
L027SCCL7 (248)CGG-TGGArg-TrpAbnormal1472157
L025SCCL7 (248)CGG-CAGArg-GlnAbnormal903169
L061LCCL7 (249)AGG-TGGArg-TrpAbnormal1402173
L028LCCL7 (258)GAA-AAAGlu-LysAbnormal1082072
L141SCCL7 (259)GAC-GTCGlu-ValAbnormalNA2071
L080SCCL8 (261)ATG-TTGMet-LeuAbnormal743170
L163SCCL8 (266)GGA-GTAGly-ValND722054
L043AdenoCa8 (273)CGT-CATArg-HisAbnormal982065
L049AdenoCa8 (273)CGT-CTTArg-LeuAbnormal1564159
L024SCCL8 (275)TGT-TTTCys-PheAbnormal1232067
L057SCCL8 (277)TGT-TTTCys-PheAbnormal422069
L146SCCL8 (278)c delFrameshiftAbnormal582079
L107SCCL8 (282)CGG-TGGArg-TrpAbnormalNA2264
L092SCCL8 (285)GAG-GTGGlu-ValAbnormal1122073
L044SCCL8 (294)GAG-TAGGlu-StopNormal922064
L019AdenoSq8 (297)3bp insHis-Gln-SerNormal202065
L161SCCL9 (307)GCA-TCAAla-SerNormal612073
L048SCCL9 (319)AAG-TAGLys-StopAbnormal922068
L164SCCL9 (325)GGA-TGAGly-StopNormal02164
L012SCCL9 (326)GAA-TAAGlu-StopAbnormal682068
L144SCCL9 (331)CAG-TAGGln-StopAbnormal622061
L154SCCLIntron 2G-ASplicingND932272
L007SCCLIntron 4G-CSplicingNormal702255

Of the 46 tumours with mutations, 36 (78%) harboured the mutation within the DNA-binding domain of p53 (exons 5–8), and 10 (22%) mutations were detected in exons outside of 5–8. Sequencing analysis demonstrated 33 missense, 9 nonsense, 3 frameshift and 2 splicing mutations. In particular, the mutational profile found was 4 deletions, 1 insertion and 42 base substitutions (Table I). Base substitutions consisted of 18 transitions (13 GC→AT, 4 of which were at CpG sites; 5 AT→GC) and 24 transversions (13 GC→TA, 4 GC→CG and 7 AT→TA). Four of the 13 GC→AT transitions occurred at CpG dinucleotides (Fig. 1). All mutations were somatic since no mutations were detected in the corresponding normal tissues.

Figure 1.

p53 mutational profile of the patients in our study compared to NSCLC (smokers only) from the IARC TP53 Database. A drop in GC→TA frequency and an increase of the non-CpG GC→AT transitions and AT→TA transversions are profound in our population.

p53 mutations were more frequent in SCCL [33 of 58 (57%)] than adenocarcinomas [8 of 36 (22%)] (Pearson's chi square, p = 0.0008). No significant correlation was found between p53 mutations and smoking status (current, former smokers and nonsmokers), total tobacco exposure (pack-years) and tobacco consumption (cigarettes/day). In addition, no association was found between p53 mutations and tumour size, nodal metastasis (TNM classification), age and gender of the patient.

Expression of p53 in NSCLC

Expression levels of the p53 gene were examined by IHC in 105 NSCLC cases. p53 overexpression was demonstrated in 67 of 105 (64%) cases examined. In particular, it was detected in 18 of 36 (50%) adenocarcinomas and 40 of 55 (73%) SCCLs (p = 0.02). p53 overexpression did not correlate with the overall presence of mutations (p = 0.08); however, it correlated with the type of mutation, being more frequent in specimens harbouring missense (27 of 30) than null (5 of 14) mutations (p = 0.0004). No significant correlation was found between p53 overexpression and smoking history, TNM classification, age and gender of the patient.

p53 in relation to MLH1 and MSH2 differential expression in NSCLC

MLH1 and MSH2 protein expression (IHC) data were available for this set of samples.18 Overall, p53 mutations were detected in 22 of 48 (46%) specimens with normal expression of MLH1 and 22 of 57 (39%) specimens with reduced expression levels of MLH1. In addition, p53 mutations were detected in 22 of 57 (39%) specimens with normal MSH2 expression and 22 of 49 (45%) specimens with reduced MSH2 expression. In SCCL, p53 mutations were more frequently detected in specimens with reduced MSH2 expression (14 of 18) than in specimens with normal MSH2 expression (17 of 38) (p = 0.019).

p53 overexpression was demonstrated in 38 of 56 (68%) specimens with reduced MLH1 expression and 28 of 47 (60%) specimens with normal MLH1 expression. In addition, 27 of 48 (56%) specimens with reduced MSH2 expression and 39 of 55 (71%) specimens with normal MSH2 expression demonstrated overexpression. This trend (p = 0.09) became a significant correlation when wild-type p53 specimens were analysed separately, excluding samples with p53 staining due to DNA mutation. In this case, p53 overexpression staining was found in 11 of 27 samples with reduced MSH2 and in 24 of 35 samples with normal MSH2 expression (p = 0.018) (Fig. 2). No significant correlations were found between p53 mutations, p53 overexpression and simultaneous reduced expression of both MLH1 and MSH2.

Figure 2.

MSH2 expression relative to p53 expression status in samples with wild-type p53. MSH2 expression appears to increase p53 staining in samples.

DISCUSSION

In the present study, we examined the status of p53 (mutations and nuclear expression) in relation to the MLH1 and MSH2 gene expression. We screened exons 2–11, which cover the whole coding region of p53. Mutations were found in 43% of samples. It is of note that 22% of mutations were located outside of exons 5–8, which traditionally comprise the main screening area for this molecule in the majority of previous studies. Our results are in agreement with previous studies reporting 17–20% of p53 mutations outside of exons 5–8 in lung tumours.12, 13, 14 Similar frequencies have been reported in breast and ovarian tumours,11, 12 while a higher frequency (33%) of such “outside” mutations was found in head-and-neck tumours.15 Also, we found no mutations in exons 2, 3 and 11. This is in agreement with previous reports12, 13, 14 indicating that exons 4–10 may comprise a minimal region of required mutational analysis to cover the vast majority of p53 mutations in lung cancer.

Only 8 patients from those with an available smoking history were nonsmokers. This did not allow us to draw statistically significant conclusions concerning the mutation frequency in relation to smoking. However, no difference was found between current and former smokers, which is in agreement with our previous observations in upper respiratory tract tumours8, 26 and indicates that p53 mutations may be acquired during a patient's early smoking history.

The mutational profile in our study is somewhat different from that of smokers in the IARC TP53 Database.5 In particular, the GC→TA transversion frequency drops, while there is an increase of GC→AT transitions at non-CpG sites (i.e., induced) and AT→TA transversions (Fig. 1) (χ2p = 0.016). The lack of a GC→TA transversion preponderance, which is usually reported for smoking populations, further supports our previous suggestion of additional environmental carcinogens in the northwest of England and/or predisposition factors that may also play a role in respiratory tract carcinogenesis.8, 26 A similar profile has been reported in patients from Gdansk, Poland, which is also an industrialised area with an analogous climate.27

p53 overexpression correlated with missense mutations as null and frameshift mutations frequently result in a truncated protein that cannot be detected by IHC. This suggests that IHC detection of p53 is not adequate by itself for identification of p53 abnormalities. However, 56.5% of samples with no detected p53 sequence mutations showed positive immunostaining. In our study, 75% of samples showed p53 aberrations and 46% were mutations. Consequently, both mutational and IHC analyses are required to identify the whole spectrum of p53 aberrations, as has been previously suggested.12

MLH1 and MSH2 expression data were available for the current set of tumours from one of our previous studies.18 Comparative analysis revealed that MSH2 expression correlated (p = 0.018) with p53 staining in samples with wild-type p53, while a trend (p = 0.09) occurred for the overall population. Two hypotheses can be generated from this finding.

The first hypothesis is that p53 overexpression results in MSH2 upregulation as p53 is a transcriptional activator of MSH2.22, 23 This is supported by the correlation of p53 mutations with reduced MSH2 expression in SCCL as well as a similar association shown in adult acute leukaemia.24 However, this suggests that overexpression of p53, which has traditionally been linked to p53 inactivation, indicates nuclear stabilisation of the protein but not necessarily inactivation since stabilised p53 can transactivate MSH2. Further investigations are required to support this argument.

The second hypothesis is that MSH2 may act as a DNA-damage signaller to p53, similarly to ATM and DNA-PK, thus binding and stabilising the protein. Therefore, high levels of MSH2 expression result in p53 stabilisation and correlate with positive p53 immunophenotype in samples carrying wild-type p53. However, in samples carrying a mutant p53, stabilisation of the protein due to the mutation masks the effect of MSH2. Taking both hypotheses together, we may speculate a possible feedback loop between p53 and MSH2; however, additional functional studies are required to confirm and reveal the full nature of this relationship.

Acknowledgements

We are indebted to the clinical staff at the Cardiothoracic Centre (Liverpool, UK) for access to their patients.

Ancillary