Methionine-Dependence Phenotype in the de novo Pathway in BRCA1 and BRCA2 Mutation Carriers with and without Breast Cancer

Methionine is an amino acid, which can be obtained exogenously via the diet and/or endogenously, by either methylation of homocysteine or transamination of 4-methylthio-2-oxobutanoic acid, through the de novo and salvage pathways, respectively (1). Evidence from two studies suggests that the majority of methionine produced through the salvage pathway is transformed to s-adenosylmethionine (SAM), whereas methionine synthesized through the de novo pathway by homocysteine methylation is predominantly used for protein synthesis (2, 3). SAM, in its turn, is required for the formation of spermidine and spermine, which have been suggested to affect cell proliferation (4), and is the primary methyl donor for DNA methylation, which regulates gene expression.

Methionine dependency is the term used to describe the phenotype of decreased ability of cells to proliferate when methionine is replaced by one of its immediate precursors, such as homocysteine, when other methyl donors, such as folate, vitamin B₁₂, and choline are present in adequate amounts. Methionine-dependency phenotype (MDP) has been frequently observed in human tumor cell lines (5, 6); however, human lymphocytes and lymphoid cell lines have also been shown to exhibit varying degrees of methionine-dependency (7, 8). When methionine is replaced by homocysteine in the culture medium, methionine-dependent cells exhibit decreased nuclear division and/or cell growth, reduced levels of SAM, and increased levels of s-adenosylhomocysteine compared with methionine-independent cells (6-9). Concomitantly, a reduced SAM/s-adenosylhomocysteine ratio induces global DNA hypomethylation and specific gene promoter CpG islands hypermethylation, both features of cancer cells that can alter chromatin structure and gene expression and thought to precede mutational and chromosomal abnormalities that are involved in cancer progression (10, 11). A study of methionine dependence in skin fibroblasts of patients with hereditary colon cancer showed that the MDP was expressed in normal somatic tissue several years before the clinical manifestation of cancer, suggesting that the MDP in these cells was an inherited trait possibly related to increased risk of oncogenic transformation and/or tumor growth (12). MDP has been observed in both breast cancer cell lines and tumors (5, 6).

BRCA1 (13) and BRCA2 (14) germline mutations are linked to 5% to 10% of all breast cancer cases diagnosed. BRCA1 and BRCA2 code for proteins of importance to genome stability as both have been shown to be involved in several cellular mechanisms such as homologous recombinational repair of DNA strand breaks, transcriptional regulation, cell cycle control, and/or mitotic spindle formation (15, 16); however, not all BRCA1 and BRCA2 mutation carriers develop breast cancer (∼85% of women carrying a BRCA1 or BRCA2 mutation will develop breast cancer by age 70 years) and other inherited traits may modify their cancer risk (17). Therefore, the primary aim of this study was to test the hypothesis that MDP in peripheral blood lymphocytes may be an inherited or acquired trait associated with an increased risk for developing breast cancer in BRCA1 and BRCA2 germline mutation carriers. A secondary aim of the study was to obtain preliminary data on the possible influence of common polymorphisms in the methionine de novo synthesis pathway in genes such as MTHFR (C677T, A1298C, and A1793G), MTR (A2756G), and MTRR (A66G and C524T) on MDP and breast cancer risk. MTHFR converts 5,10-methylene-tetrahydrofolate to 5-methyl-tetrahydrofolate, the methyl donor required for methionine synthesis by methylation of homocysteine; MTR is the vitamin B₁₂-dependent enzyme required for the synthesis of methionine from homocysteine, and 5-methyl and MTRR function is required to maintain vitamin B₁₂ in its reduced active form. Homozygosity for the C677T or A1298C mutation of the gene encoding MTHFR causes a reduction in its enzyme activity, resulting in limited methyl-group availability for the conversion of homocysteine to methionine (18, 19), which subsequently may affect exogenous methionine requirement. It is not known whether a mutation in A1793G of the gene encoding MTHFR or in any of the previously mentioned mutations of MTRR (A66G and C524T) and MTR (A2756G) significantly alters capacity to generate methionine from homocysteine.

Approval for this study was obtained from the Human Experimental Ethics committees of Commonwealth Scientific and Industrial Research Organisation Health Sciences and Nutrition, University of South Australia and Children's, Youth and Women's Health Service in Adelaide, South Australia.

Subjects were recruited by mail from a database of breast cancer families who had previously undergone BRCA1 or BRCA2 gene mutation testing at Familial Cancer Unit of the South Australian Clinical Genetics Service, which provides a state-wide program for the population of 1.5 million and is the only such provider of genetic counseling and testing for familial cancer in South Australia. The women were referred to the Familial Cancer Unit because of a personal and familial history suggestive of familial breast cancer. Testing of the BRCA1 and BRCA2 genes was offered to women who fell within the “high genetic risk” category as defined by the National Health and Medical Research Council (20). Once a mutation has been identified in an affected person, genetic counseling and testing is offered to first- and second-degree relatives of the know carrier in the family, obligate carriers in the family, and presumed carriers in the family. In practice, ∼40% of the at-risk relatives identified by the South Australian Clinical Genetics Service seek genetic counseling and testing (21). We have not identified any characteristics, which distinguish who seek genetic counseling and testing versus those who do not. In total, 140 female BRCA carriers and noncarriers, ages >18 y and residing in metropolitan South Australia, were approached, on the basis that they had had presymptomatic genetic testing at the Familial Cancer Unit after the identification of an unequivocally pathogenic mutation in a family member. In addition, noncarriers were only approached if they were unaffected by any cancer of any type. We only approached those who lived in metropolitan South Australia because samples had to be processed within 4 h of collection. Sixty-six female subjects gave consent to be part of the study and were recruited into the following groups: (a) controls (n = 24), (b) mutation carriers without breast cancer (n = 20), and (c) mutation carriers with breast cancer (n = 22).

The average age of those consenting to be part of the study was 52 y, whereas the average age of those who declined to be part of the study was 50 y. In addition, among those consenting, the proband's (i.e., affected person) Manchester score (a measurement to predict the likelihood of identifying a BRCA1 or BRCA2 mutation; ref. 22) was 49, whereas the score was 43 among those who denied participation in the study. Testing is offered if the Manchester score is 15 or more, indicating that all study participants came from “high-density” breast cancer kindred. There was no evidence of systematic bias in the subject groups.

Controls consisted of noncarrier relatives of BRCA1 or BRCA2 mutation carriers. Before their presymptomatic genetic test, all members of the family were at 25% to 50% risk of inheriting the family's mutation. Once testing was completed and they were shown to not carry the family's mutation, they were categorized as being non-carrier controls for the purpose of this study.

At the time of the study, all BRCA1 and BRCA2 mutations in the carriers were considered pathogenic and to place a woman at high risk of developing breast cancer. The description and distribution of the BRCA1 and BRCA2 mutations among carriers with and without breast cancer have been published previously (23). Before presymptomatic testing, individuals not affected by breast cancer were at 25% to 50% risk of having the pathogenic mutation, which had been documented in another relative. They were only tested for that mutation and were not screened for mutations elsewhere in the BRCA1 or BRCA2 genes. The frequency of BRCA mutations in the general population is low (∼1:1,000); therefore, screening for other BRCA mutations was not pursued in non-carriers of the family's mutation.

All participants with breast cancer completed chemotherapy and/or radiotherapy >6 mo before blood sample collection with the exception of one participant who completed treatment one month before blood collection.

To test the MDP hypothesis peripheral blood lymphocytes from BRCA1 and BRCA2 germ line mutation carriers, with or without breast cancer, and noncarrier relatives free of cancer were cultured for 9 d in medium containing either methionine or homocysteine and their relative growth under these conditions was measured. Lymphocytes were isolated from heparinized blood samples using Ficoll-Paque gradients (Amersham Biosciences) and stored in liquid nitrogen in cryovials containing a solution consisting of 90% fetal bovine serum (ThermoFisher) and 10% DMSO (Sigma). Cryovials were removed from the liquid nitrogen and thawed at 37°C, upon commencement of the experiments. When thawed, lymphocytes were washed twice in RPMI 1640 without l-glutamine, l-cystine, and l-methionine (Sigma) to which the following supplements were added 0.25 mmol/L l-cystine/L (Sigma), 2 mmol/L l-glutamine (Sigma), 1 mmol/L sodium pyruvate (ThermoFisher), 10% fetal bovine serum (Sigma), and 1% penicillin (5,000 IU/mL)/streptomycin (5 mg/mL) solution (ThermoFisher). Cells were then resuspended in 1.0 mL of medium, and cell number and viability were determined using a Coulter Counter (Coulter Electronics) and Trypan Blue (Sigma) exclusion, respectively, to calculate the volume of cell suspension required to set up the cultures at 0.5 × 10⁶ viable cells/mL. For each volunteer, 2 cultures were set up in 10 mL sterile conical tubes (Technoplas) for both Met⁺Hcy⁻ and Met⁻Hcy⁺ conditions, containing 0.5 × 10⁶ viable cells in 900 μL medium each. Cultures were then added to 100 μL of medium containing either 1 mmol/L l-methionine (Sigma) or 2 mmol/L d,l-homocysteine (Sigma), to achieve final culture concentrations of 0.1 mmol/L l-methionine (Met⁺Hcy⁻) or 0.2 mmol/L d,l-homocysteine (Met⁻Hcy⁺), respectively. Cells were stimulated to divide by adding phytohemagglutinin (30 μg/mL; Murex Biotech) and cultured for 9 d at 37°C in a humidified incubator with 5% CO₂. From day 1 until day 8, 100 μL medium was removed from the cultures and replaced by 100 μL of the appropriate medium containing 2 units of Interleukin-2 (Roche Diagnostics), every 24 h. On day 7 and day 9, cell number and viability were assessed using a Coulter Counter and trypan blue exclusion to determine the viable cell number in both types of medium. Viable cell number on day 9 was used to calculate the Methionine-Dependence Index (MDI) by dividing viable cell number in Met⁺Hcy⁻ by viable cell number in Met⁻Hcy⁺ medium. The reproducibility of duplicate MDI measurements was determined by estimating the coefficient of variation, which was 15.8% ± 1.3% (mean ± SE for 64 separate assays done in duplicate). A high MDI indicates a decreased ability of the cells to grow in the absence of methionine and therefore an increased methionine dependence, whereas cells with a low MDI are less dependent on exogenous supply of methionine.

DNA was isolated from lysed blood cells followed by Proteinase K (Sigma) treatment, salt extraction, and ethanol precipitation. MTHFR C677T, MTHFR A1298C, MTHFR A1793G, and MTR A2756G mutations were determined by PCR as described by Frosst et al. (19), Weisberg et al. (24), Wakutani et al. (25), and Van der Put et al. (26), respectively, whereas the MTRR A66G and MTRR C524T mutations were detected by PCR as reported by Janosíková et al. (27). Primers and restriction endonucleases were obtained from GeneWorks and New England Biolabs, respectively, whereas the deoxynucleotide triphosphates and Taq polymerase were acquired from Roche Diagnostics. Genotyping was done in duplicate to confirm validity of results. If the duplicates were not concordant, the test was repeated until a clear consistent result was obtained. Only 2% of duplicates were not concordant.

Two-tailed Pearson correlations were used to analyze the potential confounding effect of age and/or body mass index (BMI) on viable cell growth and MDI. One-way ANOVA, followed by Tukey's post hoc multiple comparison test, was used to compare the age, BMI, and MDI of the controls, carriers without breast cancer and carriers with breast cancer, and the MDI in relation to selected polymorphisms in folate metabolism genes. In addition, viable cell number in both types of medium were compared using analysis of covariance (ANCOVA), with age and BMI included as covariates to control for their potential confounding effect. Assumptions relating to ANOVA and ANCOVA (i.e., normal distribution and homogeneity of variance) were examined and met. The paired t test (two tailed) was used to compare the effect of methionine replacement by homocysteine on viable cell number, and the χ² test was used to determine a possible association between genotype and BRCA1 and BRCA2 carrier and breast cancer status. Fisher Exact test was used to calculate odds ratios based on allele frequency. All data are expressed as mean ± SE. Significance was accepted at a P value of <0.05. All statistical analyses were done using SPSS 14.0 (SPSS). The choice of statistical tests was based on published recommendations (28).

The mean age of the 3 study groups (56.4 ± 2.44 years, 47.0 ± 3.12 years, and 51.0 ± 2.37 years for controls, BRCA carriers without breast cancer, and BRCA carriers with breast cancer, respectively) and BMI (27.6 ± 0.85 kg/m², 26.7 ± 0.92 kg/m², and 29.0 ± 1.32 kg/m² for controls, BRCA carriers without breast cancer, and BRCA carriers with breast cancer, respectively) did not differ significantly from each other, except for the age of the controls, which was significantly increased relative to BRCA carriers without breast cancer (P = 0.045).

Plasma folate (nmol/L) in non-carrier controls, BRCA carriers without breast cancer, and BRCA carriers with breast cancer was 12.6 ± 0.5, 11.8 ± 0.8, and 12.4 ± 0.9, respectively, plasma vitamin B₁₂ (pmol/L) was 295 ± 54, 345 ± 64, and 249 ± 17, respectively, and plasma homocysteine (μmol/L) was 7.7 ± 0.5, 7.1 ± 0.4, and 7.9 ± 0.3, respectively. The study groups did not differ significantly.

Results in Fig. 1 show that (a) viable cell growth of peripheral blood lymphocytes is greater in medium containing Met⁺Hcy⁻ relative to Met⁻Hcy⁺ (P < 0.001), and (b) growth was still on the increase on day 9 relative to day 7. These patterns of cell growth were also observed in preliminary experiments (data not shown), which also showed no evidence of viable cell growth in the absence of both methionine and homocysteine.

Viable cell number on day 9 was significantly negatively correlated with age and BMI (Table 1). Consequently, analyses were done using ANCOVA with age and BMI as covariates to control for their potential confounding effect. Viable cell number in both types of medium did not differ significantly between BRCA1 and BRCA2 mutation carriers with or without breast cancer and noncarrier controls on day 9 (Table 2). In addition, BRCA1 and BRCA2 carriers with or without breast cancer and noncarriers controls displayed a significantly increased viable cell number in Met⁺Hcy⁻ relative to Met⁻Hcy⁺ (P < 0.001; Table 2).

The MDI did not differ significantly among BRCA1 and BRCA2 carriers with or without breast cancer and noncarrier controls (Table 2); however, there was a trend for a lower MDI in carriers without breast cancer relative to the two other groups.

Table 3 shows the frequency of selected polymorphisms in methionine metabolism genes. The frequency distribution of wild-type homozygotes, heterozygotes, and variant allele homozygotes did not differ significantly among BRCA1 and BRCA2 mutation carriers with or without breast cancer and noncarrier controls for the MTRR A66G, MTRR C524T, MTHFR C677T, MTHFR A1298C, and MTHFR A1793G polymorphisms; however, for MTR A2756G, there was an increased frequency of G variant allele carriers in the BRCA mutation carriers with breast cancer relative to the noncarrier control group (P = 0.009) but no difference for other group comparisons.

In BRCA1 and BRCA2 mutation carriers, the presence of the MTR A2756G mutation (i.e., AG or GG) and MTHFR C677T mutation (i.e., CT or TT) was associated with increased risk in breast cancer with an odds ratio of 3.2 (P = 0.16; 95% confidence interval, 0.76-13.9) and 3.9 (P = 0.09; 95% confidence interval, 0.93-16.3), respectively.

Table 4 displays the MDI of selected polymorphisms in methionine metabolism genes. MTR, MTRR, and MTHFR polymorphisms did not significantly affect MDI, except for MTHFR C677T, which was borderline significant (P = 0.050). When comparing MTHFR 677CC homozygotes with carriers of at least one T allele (i.e., CT or TT), MDI of T allele carriers was significantly increased (3.9 ± 0.4) relative to CC homozygotes (3.0 ± 0.2; P = 0.017).

The study of MDP in familial cancer is important for two reasons: (a) MDP may reflect important differences in folate/methionine metabolism, which is a known risk factor for several cancers (29, 30), and (b) knowledge of MDP in normal somatic and cancer cells is essential to evaluate the feasibility of using methionine restriction as a strategy to selectively restrict cancer growth.

To the best of our knowledge, this is the first time that MDP has been studied in peripheral blood lymphocytes of BRCA1 and BRCA2 carriers with or without breast cancer. The relevance of determining whether MDP is a risk factor for breast cancer relates to the pressing need to identify who of the BRCA1 and BRCA2 carriers is at highest risk as this has implications on critical personal decisions relating to radical preventive surgery, i.e. mastectomy. Excessive proliferation of cells and altered methylation status is a common risk factor for a great variety of cancers and it is therefore plausible that a single potential risk factor that can affect both of these processes (such as MDP) might be relevant to the risk and etiology of both BRCA1 and BRCA2 breast cancers and, thus, justifies combining BRCA1 and BRCA2 carriers in the analysis.

Viable cell number in Met⁺Hcy⁻ and Met⁻Hcy⁺ medium as well as MDI did not differ significantly among carriers with or without breast cancer and controls, suggesting no effect of MDP on breast cancer risk in BRCA1 and BRCA2 germline mutation carriers. Of the three comparisons, that between BRCA1 and BRCA2, carriers with and without cancer is the most important because it addresses the key question of who is most at risk among those with a predisposition to breast cancer. The comparison with non-carriers without breast cancer may be informative if carriers without breast cancer show a unique characteristic that differs from the general population. In this regard, carriers without breast cancer tended to have the lowest MDI, which supports the hypothesis that MDI may be associated with their resistance to the disease. The trend for a 16% lower MDI in carriers without breast cancer relative to non-carrier controls and carriers with breast cancer provides a hint that lower MDI might be associated with reduced cancer risk in BRCA carriers, but the study was not sufficiently powered to verify this. Based on the observed SD of the data obtained, the study was adequately powered to detect an effect size difference of 40% at 80% power P < 0.05 for the MDI.

Furthermore, these results need to be interpreted with caution because methionine dependency is a feature that is not solely dependent on homocysteine metabolism but can also be due to limited supply of the other precursors of methionine, i.e. 4-methylthio-2-oxobutanoic acid and its precursor 5-methylthioadenosine (31). 5-Methylthioadenosine is formed when dcSAM transfers its aminopropyl moiety to putrescine and spermidine to synthesize spermidine and spermine, respectively. SAM carboxylase activity, which converts SAM into dcSAM, is increased in methionine-dependent cells when methionine is restricted to maintain polyamine levels (32, 33). Cells unable to synthesise methionine from 5-methylthioadenosine or homocysteine would be expected to exhibit reduced cell growth when methionine is restricted; however, because methionine dependency in the salvage pathway was not specifically assessed in the current study, it cannot be excluded that cells in this study, exhibiting MDP were not able to salvage methionine from 5-methylthioadenosine. In future studies, it will be necessary to assess growth not only in Met⁻Hcy⁺ medium but also in medium deficient in methionine to which 5-methylthioadenosine is added to test the functionality of both the de novo and salvage pathway.

We cannot totally exclude that exposure to cancer chemotherapeutic agents might have influenced MDI measurements; however, none of the chemotherapeutic agents in breast cancer chemotherapy target the methionine, homocysteine, or 5-methyltetrahydrofolate synthesis reactions, which may influence the MDP measured in this study. In addition, because measurement of the MDI uses one's own cells as controls, any residual effects of chemotherapy on proliferation would have been equivalent in the Met⁺Hcy⁻ and the Met⁻Hcy⁺ cultures and therefore nullified when the MDI is calculated as a growth ratio. Furthermore, the MDI in the non-carrier controls, and the carriers with breast cancer were almost identical, which argues against an effect of chemotherapy.

Our results show that the ability of cells of MTHFR C677T mutant allele carriers to grow in absence of methionine is restricted as indicated by an increased MDI. This finding is consistent with the previously identified reduced MTHFR enzyme activity of the MTHFR C677T variant genotype (19, 24), which results in a reduction of the availability of 5-methyltetrahydrofolate as a methyl donor for the synthesis of methionine from homocysteine. A previous study by our group (8) did not show an association of MTHFR C677T and A1298C or MTR A2756G polymorphisms on nuclear division of peripheral blood lymphocytes in Met⁻Hcy⁺ in lymphocytes from healthy individuals. The absence of an effect of MTHFR C677T in the latter study was probably due to a less rigorous assessment of methionine dependence based on measuring proportion of cells undergoing nuclear division during a 48-h period rather than the actual accumulated viable cell number over a 9-day culture period used in our study. Further research into the influence of the MTHFR C677T polymorphism on methionine dependency is necessary as our study population was small and the concentration of riboflavin, the cofactor for MTHFR, in RPMI 1640 was supraphysiologic and may have minimized the difference in MTHFR activity due to the C677T polymorphism (34). Future studies with medium containing physiologic concentration of cofactors and/or substrates required in methionine metabolism such as 5-methyltetrahydrofolate, riboflavin, choline, and vitamin B₁₂ are recommended to better reflect the in vivo situation.

In our study, BRCA1 and BRCA2 germline mutation carriers with breast cancer exhibited an increased number of MTR A2756G variant allele carriers relative to those without breast cancer and noncarrier controls. To date, several studies have investigated the possible link between MTR A2756G and breast cancer. One of these studies reported a reduced breast cancer risk in heterozygotes and homozygous variant allele carriers relative to common homozygotes (35), whereas the others did not find a significant association between MTR A2756G genotype and breast cancer (36-39). Therefore, our findings of an association of breast cancer with the MTR A2756G polymorphism may be relevant only to BRCA1 and BRCA2 families examined in our study. The plausibility of an association of the MTR A2756G polymorphism with breast cancer risk is supported by the studies of (a) Li et al. (40) who reported a strikingly reduced estrogen hormone receptor concentration in breast tumors homozygous for the MTR 2756G allele and (b) studies that show BRCA1 and BRCA2 breast tumors are characterized by a low estrogen receptor concentration (41, 42). This suggests that further studies into the association of MTR A2756G with breast cancer estrogen receptor phenotype, especially in BRCA1 and BRCA2 germline mutation carriers, may be required to verify a possible causal effect of MTR A2756G polymorphism on estrogen receptor expression.

The apparent association of the MTHFR C677T mutation with breast cancer risk observed in our study is consistent with the observations in large cohort studies in Japan (43), USA (37), Austria (44), Italy (45), Poland (46), China (47), and Australia (48); however, given the small group size and uncertainty due to multiple comparisons, the results from our study of an association of breast cancer risk with MTR A2756G and MTHFR C677T polymorphism can only be considered preliminary.

Taken together, the results of this study suggest a weak trend for increased breast cancer risk in BRCA1 and BRCA2 mutation carriers who exhibit increase methionine dependence and who also carry a MTHFR C677T and/or MTR A2756G variant allele.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C.

Section 1734 solely to indicate this fact. pWe thank the participants who kindly donated their time and their blood samples to enable this research to be completed.