Somatic alterations of the serine/threonine kinase LKB1 gene in squamous cell (SCC) and large cell (LCC) lung carcinoma.

Somatic LKB1 serine/threonine kinase alterations are rare in sporadic cancers, with the exception lung adenocarcinoma, but no mutations in squamous cell or large cell primary carcinoma were discovered. We screened the LKB1 gene in 129 primary nonsmall cell lung carcinomas, adjacent healthy lung tissue, and control blood samples. Forty-five percent of nonsmall cell lung tumors harbored either intron or exon alterations. We identified R86G, F354L, Y272Y and three polymorphisms: 290+36G/T, 386+156G/T, and 862+145C/T (novel). R86G (novel) and F354L mutations were found in six squamous cell carcinomas and three large cell cancer carcinomas, but not in the adjacent healthy tissue or controls samples. The F354L mutation was found in advanced squamous cell carcinomas with elevated COX-2 expression, rare P53, and no K-RAS mutation. Results indicate that the LKB1 gene is changed in a certain proportion of nonsmall cell lung tumors, predominately in advanced squamous lung carcinoma. Inactivation of the gene takes place via the C-terminal domain and could be related to mechanisms influencing tumor initiation, differentiation, and metastasis.

Keywords: LCC; Mutation detection; K-RAS; DHPLC; COX-2; SCC

LKB1 maps to the chromosomal region 19p13. It is a putative tumor suppressor gene, mainly known for its role in the cancer predisposed disorder Peutz-Jeghers (PJS) syndrome, which causes an increased incidence of several cancers, including lung adenocarcinomas ([1], [2]). Human LKB1 encodes a 433 amino acid protein, approximately of 48 kDa ([1]). It consists of a main catalytic kinase domain (residues 50–337), similar to the SNF1/AMP—activated protein kinase family, and a putative carboxyl—terminal regulatory domain with a CAAX box, a known consensus sequence for prenylation ([3]). Two potential nuclear localization signals have been located between amino acids 38–43 and 81–84 ([3]). LKB1 has been implicated in different pathways, such as P53- dependant apoptosis, P21- mediated cell grow arrest, VEGF- signalling pathway, and ATM response to radiation ([4], [5], [6]). LKB1 activity may also be regulated by phosphorylation and farnesylation and consecutively plays a role in regulating cellular stability ([7]). Based on human lung cancer cell lines studies and LKB1- deficient mouse, it was proposed that LKB1 might represent a critical barrier to pulmonary tumourigenesis ([8]). LKB1 deficient tumors demonstrated enhanced metastasis and differentiation, features not seen in P53/P16/ARF deficient backgrounds ([8]). On the other hand, LKB1 reconstitution in human NSCLC cell lines lacking functional P16, ARF, and P53 showed antitumor effects ([8]).

Somatic mutations were described for the first time in 1998 in colorectal adenomas ([9]). In 1999 Avizienyte et al. published results of the mutational analysis done on 129 different tumors and 14 cell lines, revealing only three different mutations ([10]). The analysis of 28 lung tumors resulted in only one change in one adenocarcinoma sample ([10]). Somatic LKB1 alterations in tumors are rarely reported, with an exception of nonsmall cell lung cancer, where inactivation of LKB1 is a common event ([11]). It is reported that approximately one-third of adenocarcinoma cell lines and sporadic primary lung tumors have mutations in the LKB1 gene ([11], [12], [13], [14]). The sum of discovered mutations in lung cancer of sporadic origin was published last in 2007, revealing there are only 5 amino acid substitutions out of 51 known mutations, and all of the published mutations are located in the kinase domain of the protein ([11]). There were no mutations found in squamous and large cell primary carcinomas, and the number of lung tumors screened for alterations in LKB1 did not exceed 91 ([11]). Mutation F354L, which was also detected in our study, was first published as a somatic mutation in 1998, in colorectal adenocarcinoma ([9]). Amos et al. found the same mutation in Peutz-Jeghers syndrome male patient, who undertook polypectomy early in his teens ([15]). The same mutation was found in extended family with affected relatives, and the change co-segregated perfectly with the disease ([15]). The coexistence of K-RAS and/or P53 mutations with LKB1 mutations has been shown in several lung tumors, mostly adenocarcinomas ([12], [14], [16]). LKB1 can suppress COX-2 induction and cellular invasion through the PEA3 transcriptional factor in lung cancer, unlocking the possibility for LKB1 to regulate COX-2, the deregulation of which is ubiquitous not only in lung but in all human cancers ([17]).

In order to delineate the mutational spectrum of the LKB1, P53 and K-RAS genes in histological subtypes of nonsmall cell lung cancer (ADC = adenocarcinoma, LCC = large cell carcinoma, SCC = squamous carcinoma) denaturing high-performance liquid chromatography (DHPLC) screening of all the three genes was performed on 129 tumor samples. COX-2 mRNA expression profiles were also established. For establishing the status of the LKB1 gene in noncancerous tissue, we tested adjacent healthy lung tissue from the same patients and 100 healthy controls samples.

Our study included 129 primary nonsmall cell lung carcinoma tumors (51 adenocarcinoma, 67 squamous cell carcinoma, and 11 large cell carcinoma) and adjacent healthy lung tissue from surgical patients, to determine if the alterations are somatic or germline. The snap frozen biopsies were histologically evaluated and patient survival and clinical data obtained. The group consisted of 107 (83%) male and 22 (17%) female patients, with ages ranging from 42 to 83 years and mean age 65.4 years (and SD ± 9.1 years). For evaluating the frequencies of found alterations in a healthy population, 100 blood donors were screened. The control group consisted of 79 males (79%) and 21 (21%) females, with mean age 39.8 (and SD ± 7.1 years).

DNA was extracted from snap frozen lung cancer tumors and corresponding normal tissue using NucPrepTM chemistry (Applied Biosystems, Foster city, CA, USA) on ABI 6100 (Applied Biosystems, Foster city, CA, USA). DNA was extracted from the peripheral blood mononuclear cells of healthy blood donors using QIAamp DNA Blood Midi KitTM (QIAGEN, Germany).

The LKB1 mutational status was analyzed with DHPLC (Transgenomic, USA). Suitable newly designed primers appropriate for DHPLC detection were chosen with an oligo-designer (IDT® SCITools Oligo Design and Analysis), and the conditions for DHPLC mutation screening were default according to WavemakerTM (Transgenomic, USA). The set of primers used is described in Table 1. The tumor, nontumor, and blood DNAs that were analyzed were mixed with corresponding control DNA, with defined normal genotype, to prevent the possibility of overlooking homozygous alterations. Samples with confirmed distinct elution chromatograms were subsequently sequenced to determine the alterations.

Table 1 Primers used for genomic DHPLC analysis, oligonucleotide sequences of PCR primers, PCR product lengths, and annealing temperatures

DNA extracted from peripheral blood mononuclear cells of 100 blood donors was used as a control group for the comparison of previously found alterations in the LKB1 gene in lung tumors, so that it could be determined whether the alterations are related to lung cancer or are commonly present in the population. With DHPLC we analyzed exon 2 (290+36G/T, R86G), exon 3 (386+156G/T), exon 8 (Y272Y, 862+145C/T), and exon 10 (F354L) in the control group.

DHPLC temperatures used for profiling were chosen according to the WavemakerTM program to screen the whole coding and imminent intron region and were

• UTR+EX1+EX2: 62°C, 67.4°C, 69.4°C, and 70°C
• EX2 (UTR): 56°C, 63.4°C, 65.4°C, and 68°C
• EX2: 62°C, 63.2°C, and 64.2°C
• EX3: 60.5°C
• EX4+5: 64.5°C, 67.7°C, and 68°C
• EX6+7: 63.5°C, 65.5°C, and 66°C
• EX8: 64°C, 65°C
• EX9: 64.4°C, 66.8°C
• EX10: 64.4°C, 65°C
• EX11: 63°C, 65.4°C, 66.4°C, and 68°C
• 11(UTR): 54°C, 65°C, and 66.8°C.

The concentrations of WAVE Optimized buffersTM A, B, D and WAVE syringe solutionTM were default for mutation detection, as were the flow and time of elution, calculated with the Wavemaker programTM.

For DHPLC of the P53 gene, primers and conditions were used as described for esophageal cancer ([18]). For DHPLC screening of K-RAS mutational hotspots in exons one and two, primers were used as already described by Konig et al. ([19]) and a column temperature calculated with the Wavemaker programTM.

Protocols for PCRs were chosen for polymerase and primer specificity. Changes and elution shifts were analyzed with an ABI prism 310 Genetic Analyzer (Applied Biosystems, USA).

Expression of the COX-2 mRNA in tumor samples relative to their normaladjacent tissues was measured using quantitativereal-time PCR (RT-PCR) based on the TaqMan® fluorescence methodology. A ready mixture of probes and primers, specific for COX-2 gene expression, was used (Assay-on-DemandTM, Hs0015133_m1, Applied Biosystems, Foster City, USA). Human 18S rRNA (TaqMan predeveloped assayreagents for gene expression, Applied BioSystems, Foster City, USA) was used asthe endogenous control. Reactions were performed with a TaqManUniversal PCR Master Mix (Applied Biosystems, Foster City, USA) in a 25 μl reaction volume. All reactions were performed in triplicate and included a negative control. Quantification was performed using the ABI Prism 7900 sequence detection system (Applied Biosystems).

The relative quantification of mRNA levels of the target gene (quantity of transcripts of the target in tumor samples relative to normal tissues) was determined using the ΔΔCT (ΔCT tumor − ΔCT normal) method. If ΔΔCTwas significantly (2σ) higher or lower than zero, the expression was considered to be significantly over or underexpressed.

Magnitudes and directions of the associations between variables (ADC, SCC, and LCC tumors with and without mutations in LKB1, P53 and K-RAS, corresponding healthy tissue and blood) were determined using Fisher's exact test with two-sided p-value. A value of p<.05 was considered statistically significant. All statistical analyses were performed using SPSS ver.14 (SPSS Inc., Illinois). Spearman's Rho correlation was used to evaluate COX-2 mRNA expression and LKB1 status. A value of p<.05 was considered significant.

Forty-five percent of the tested nonsmall cell lung cancer tumors have alterations in either the intron or exon region of the LKB1 gene. 72.7% of the large cell lung tumors, 43.3% of squamous cell lung tumors, and 41.2% of lung adenocarcinomas have either LKB1 intron or exon alterations (Table 2). Most represented were two polymorphisms, occurring in 16.3% (290+36G/T) and 17.1% (386+156G/T), and a F354L mutation, which occurred in 6.2% of all analyzed tumors (Table 2). We found a new mutation in large cell carcinoma (CGA256GGA), changing arginine into glycine at position 86, which is the main kinase domain, the only location described as affected by mutations in lung cancer ([11]) (Figures 1 and 2). The normal tissue was used as a reference, showing no distinct pattern and subsequent sequencing, normal genotype at R86 (Figures 1 and 3). This is also the first time the mutation TTC1062TTG (F354L) has been discovered in NSCLC. We did not detect any F354L mutations in adjacent noncancerous lung tissue (Figure 5) or in healthy population group, therefore it is more likely that F354L is a tumor specific variation than polymorphism. F354L mutation was found in 6 out of 67 (9%) of squamous cell tumors and in 2 out of 11 (18%) large cell cancer tumors (Table 2). Using Fisher's exact test with a two-sided p-value, we found that the F354L mutation is statistically significant for large cell cancer tumors and squamous cell cancer tumors (p<.03). The F354L mutation was not discovered in any of the ADC tumors so it is specific for SCC and LCC tumors. In adenocarcinoma we detected somatic alteration Y272Y, which was absent in corresponding healthy lung tissue and in the control cohort (Figure 4). That was also the only exon alteration found in adenocarcinomas, although the frequencies of intron alterations were similar to those of squamous lung cancer. Polymorphisms 290+36G/T, 862+145C/T, and 386+156G/T were often detected but their frequencies were not significant for any type of the NSCLC (Fisher's exact test, p>.05). Analysis of the corresponding healthy lung tissue revealed that polymorphisms 290+36G/T, 386+156G/T, and 862+145C/T are germline (Table 2). Frequencies of the most frequent polymorphisms in tumors were slightly higher in healthy population, 290+36G/T (25.0%) and 386+156G/T (18.8%), but not significant (Fisher's exact test, p>.05, Table 2). In one patient with adenocarcinoma and two with squamous carcinomas we discovered novel polymorphism 862+145C/T (Figures 6 and 7 and Table 2). The polymorphism was absent in control group (Table 2). Analyzing corresponding normal lung tissue and DNA from control cohort showed no changes in LKB1 exons regions (Table 2). No deletions or insertions, detectable with the techniques we used were found, nor any frame shift or stop mutations.

Table 2 Frequencies and types of somatic mutations, alteration, and polymorphisms found in the LKB1 gene in NSCLC tumors, corresponding healthy tissue from patients and blood from control group

962 ADC = adenocarcinoma samples; LCC = large cell carcinoma samples; SCC = squamous carcinoma samples; Corresponding lung tissue = normal tissue taken from the patients without histological changes; refSNP ID = reference SNP identification number in NCBI reference assembly; *as control for frequencies of the discovered alterations we used DNA samples from 100 blood donors; **although Y272Y is described as polymorphism under Rs9282859, considering we detected alteration only in tumors, we defined it as acquired somatic alteration. All found alterations we discovered in tumors and adjacent healthy tissue and blood were heterozygous.

Graph: Figure 1 DHPLC chromatogram report on the PB/97-01 tumor (LCC), showing heterozygosis for missense mutation CGA256GGA, R86G (temp. of DHPLC analysis 62°C). Upper chromatogram report of the normal adjacent lung tissue (PB/97-01) shows lack of pattern, specific for R86G.

Graph: Figure 2 The result of sequencing of the PB/97-01 (LCC) tumor, heterozygous for CGA256GGA, R86G mutation, reverse sequence.

Graph: Figure 3 The result of sequencing of the PB/97-01 normal lung tissue, showing no heterozygosis at R86 position (CGA256CGA), reverse sequence.

Graph: Figure 4 DHPLC chromatogram report of the representative PB/99-10 tumor (ADC) showing heterozygosis for silent acquired somatic alteration TAC816TAT, Y272Y but not for adjacent healthy tissue (sample PB/99-10 normal lung tissue; temp. of DHPLC analysis 64°C).

Graph: Figure 5 DHPLC chromatogram report of the representative PB/98-06 tumor (SCC) showing heterozygosis for TTC1062TTG mutation, F354L. The upper chromatogram corresponding to adjacent healthy lung tissue (PB/98-06) shows lack of pattern significant for F354L (temp. of DHPLC analysis 64.4°C).

Graph: Figure 6 DHPLC chromatogram report of the representative PB/99-45 tumor (SCC) showing heterozygosis for novel polymorphism 862+145C/T, discovered in tumor and adjacent healthy tissue. As a wild type control for the polymorphism we used PB/97-01 (LCC), without 862+145C/T, (temp. of DHPLC analysis 64.4°C).

Graph: Figure 7 The result of sequencing of the PB/99-45 tumor (SCC), heterozygous for novel polymorphism 862+145C/T.

P53 was mutated in 26.5% ADC, 27.3% LCC, and 55.6% SCC. The frequency of P53 mutation was significant for SCC tumors regarding to ADC (Fisher's exact test, p<.02). K-RAS mutations were found in 27.0% ADC, 18.2% LCC, and in 2.2% SCC; all were previously described at codon hotspots (codons: 12 (77.0%), 13 (15.3%), and 61 (7.7%)) and all were missense. Fisher's exact test showed a difference in frequency of K-RAS mutations between ADC and SCC (p<.03).

Elevated expression of COX-2 was observed in 60.0% of tested NSCLC (61.8% in ADC, 27.3% in LCC, and 66.7% in SCC) Average overall ΔΔCT for COX-2 in NSCLC was 5.3 (SD ± 0.6), 8.9 (SD ± 0.9) in ADC, 2.3(SD ± 0.1) in LCC, and 3.0 (SD ± 0.2) in SCC. Expression of COX-2 was not significant for either type of NSCLC or for patients with LKB1 alterations (Spearman's Rho statistics, p>.05). Exceptions are patients with F354L mutation, where all of the tested SCC with the mutation showed elevated levels of COX-2 mRNA (Table 3). For informative statistical evaluation of the relation between COX-2 elevated mRNA expression and F354L mutation we would need more F354L mutated tumors.

Table 3 Gender, type of lung cancer, pTNM pathological classification, status of P53 and K-RAS gene and COX-2 expression in the tumors with F354L, Y272Y, and R86G alterations

963 NT = not tested; T = primary tumor; N = regional lymph nodes; M = distant metastasis; LCC = large cell carcinoma; SCC = squamous cell carcinoma; ADC = adenocarcinoma. COX-2 expression values are referred to the relative quantification of mRNA levels of the target gene(quantity of transcripts of the COX-2 mRNA in tumor samples relative to normal tissues), determined using the ΔΔCT (ΔCT tumor − ΔCT normal) method. All of the tumors with F354 or R86G mutation were T2 classification and all but two (one large cell carcinoma and one squamous cell carcinoma) with regional lymph nodes involvement. COX-2 mRNA levels were elevated in all the six tested samples, all squamous cell carcinomas.

Using Fisher's exact test for analyzing significance among patients with polymorphisms found and P53 mutations, K-RAS mutations and COX-2 overexpression, we found no correlation between polymorphisms (290+36G/T, 386+156G/T, or 862+145C/T) and the two variables. The F354L mutation is significantly collated with tumors staging T2, and regional node infiltration (Fisher's exact test, p<.05, Table 3). F354L was also gender specific, all patients with the mutation being male (Fisher's exact test, p<.05). The group of patients with the Y272Y alteration was too small for any statistical analysis but all of the patients with the somatic mutation were female with ADC staging T1 or 2, N1 and without metastasis (Table 3). Using Fisher's exact test, we could not relate patients with mutations and age at diagnosis, survival or tumor location (p>.05).

There is already substantial evidence that LKB1 is a tumor suppressor gene involved in the development of lung adenocarcinoma ([12], [13], [16], [20]). Bronchioloalveolar carcinoma (BAC) of mucinous type was reported in a 22-year-old male PJS patient with a germline frame shift insertion of the LKB1 gene ([21]). Acquired mutations in LKB1 were mostly observed in adenocarcinomas of the lung, in which up to 50% inactivating mutations were identified ([12], [13], [16], [20]). Larger gene deletions and frame shift or nonsense mutations were reported in lung cancer cell lines ([13]), however a comparison of the mutational status in different NSCLC types is still lacking. We found only heterozygous alterations, and the possibility of overlooking homozygous changes was abolished by mixing the samples with DNA with known (normal) genotype for the DHPLC analysis. To the best of our knowledge, this is the first study reporting mutations in primary squamous cell and large cell carcinomas. The F354L mutation, which had previously been identified in one colorectal sporadic tumor and in a family with PJS syndrome, was found in our study in six squamous cell carcinoma and two large cell carcinomas ([9], [15]). Mutation was found also in many affected relatives of the PJS patient and the change co-segregated perfectly with the disease ([15]). F354L was also found in Korean population (6%) and in 0.2% of Finns population ([22], [23]), but in our study we have not detected F354L in healthy adjacent lung tissue or in healthy population cohort, therefore is more likely to be a tumor specific variation. We also found three polymorphisms, one of them (862+145C/T) published for the first time and found only in ADC and SCC, but they do not show any relation with NSCLC type, mutations in P53 and K-RAS or COX-2 mRNA elevated expression (Fisher's exact test, p>.05). 862 +145C/T was not detected in healthy control group, and looks like it is NSCLC related polymorphism or very rare polymorphism but it still needs to be determined on a larger sample size. Frequencies of found alterations are slightly different in large cell carcinomas, but that could be the case of small number representation in this study. All of the polymorphisms were found also in adjacent healthy lung tissue so they are germline. Frequencies in normal population control for most represented polymorphisms are slightly higher but not statistically relevant (Fisher's exact test, p>.05). There was no correlation between survival of the patients, age at diagnosis, tumor location, gender of the patients, COX-2 expression and mutations in K-RAS, or P53 and polymorphisms found in LKB1 gene. R86G, novel mutation, was found in large cell carcinoma, but was absent from adjacent healthy tissue and from healthy population cohort. Mutation R86G alters amino acid in the kinase domain of the protein, where missense mutations abolish the ability of LKB1 to autophosphorylate or to phosphorylate AMP-activated protein kinase, which is direct LKB1 substrate ([24], [25], [26]). In three adenocarcinomas we found Y272Y somatic alteration, which was absent in corresponding normal tissue. The Y272Y is listed as polymorphism (Rs9282859 in NCBI reference assembly), but in our case the alteration was present only in three adenocarcinomas. Y272Y was absent from other types of NSCLC, healthy lung tissue, and population control (Table 2), which define the alteration as NSCLC type specific acquired somatic alteration. The F354L mutation was found mostly in advanced squamous cell carcinomas with regional lymph nodes involvement, in all tested cases elevated COX-2 expression, rare P53 gene mutation, and no mutation in the K-RAS gene. With our study we discovered that the mutation F354L is significantly related to SCC and also to LCC. The F354L mutation lies adjacent to the C-terminal noncatalytic domain of LKB1, which has many sites for posttranslational modifications, and mutations in this region are often found in tumors inactivating the LKB1 gene ([1], [2], [3], [15]). This observation accords with the fact that LKB1 modulates lung cancer differentiation and metastasis ([4], [5], [6], [27]). In all of the tested cases (six SCC tumors with F354L mutation), elevated COX-2 expression was also detected, although no correlation between COX-2 mRNA expression and tumors without the LKB1 mutation was observed. Relation of elevated COX-2 mRNA levels and LKB1 mutations is in accordance with already published observations ([17], [28]). COX-2 over-expression has been shown in different cancer cell lines with a mutated LKB1 gene ([17]). Reintroduction of wild type LKB1 into cells resulted in reduced COX-2 expression, suggesting a role for LKB1 in the regulation of COX-2 ([17]). There is also a correlation between expression patterns of the LKB1 relative to COX-2 in polyps and carcinomas from patients with Peutz-Jeghers syndrome ([28]). The coexistence of LKB1 mutations, P53 in K-RAS mutations was infrequent. We did not detect any K-RAS/LKB1 double mutation and only 2 mutations in P53 in tumors with F354L (Table 3). The absence of mutations in K-RAS is logical, since K-RAS mutations have been mainly found in lung adenocarcinomas, and the low frequency of P53 mutations might be an indication that LKB1 inactivation in SCC occurs earlier in lung carcinogenesis. Further analyses on more samples are needed to elucidate the pathogenic significance of the interaction of these genes.

Interestingly, no somatic mutations, changing amino acid, were detected in adenocarcinomas. The lack of mutations in adenocarcinoma may show some population differences but, in general, 9% of detected exons alterations in NSCLC types are in accordance with other studies ([12], [13], [14], [20]).

Our findings show that LKB1 F354L mutation (Fisher's exact test, p<.03) is lung cancer type specific alteration and that there could be lung cancer type dependant hotspots (Y272Y only in ADC and F354L in SCC and LCC). We did not find any frame shift or stop mutations (deletions or insertions), which are known to inactivate LKB1 gene mostly in PJS syndrome, lung cancer cell lines, and in adenocarcinoma patients ([1], [3], [13], [14], [15], [16]), so it is possible that the inactivation of SCC and LCC occurs through missense mutations, although with the technique we used for mutational screening it is possible to overlook large intragenic deletions, which are commonly found in lung cancer cell lines ([11]). There is also a relation between the expansion of the tumor and existence of that mutation, because all of the SCC tumors with F354L mutation are advanced (T2 stage) with regional lymph nodes involvement (Fisher's exact test, p<.05). In squamous cell lung cancer samples, the missense mutations were also gender specific and all of the same type, so they may constitute a genetic signature for squamous carcinoma (Fisher's exact test, p<.05). In relation to other clinical data, such as age at diagnosis, survival, tumor location, we could not determine whether the mutation is significantly correlated with any of those variables since for any informative statistics, more patients with the same mutation would be needed.

These results indicate that the LKB1 gene could be inactivated with mutation in a certain proportion of nonsmall cell lung tumors and that this inactivation is probably related to mechanisms influencing tumor initiation, differentiation, and metastasis.

We have demonstrated that LKB1 mutations are not confined only to lung adenocarcinomas and lung cancer cell lines, but also appear in primary squamous cell and large cell carcinomas and alter the C-terminal region with CAAX box, which is consensus sequence for prenylation ([3]).

This work was supported by the Slovenian Research Agency (ARRS) (Program P3-054). This work is part of the Ph.D. Thesis of Mojca Stražišar.