This article is mentioned in:

Introduction

Breast cancer is the most frequent cancer and the most common cause of cancer-related death in women (1). In Brazil alone, ∼50,000 new cases of breast cancer are expected per year, with an estimated risk of 52 cases for every 100,000 women (2). Several studies have suggested that different genetic alterations are linked to different histological types of breast cancer, such as estrogen receptor expression/lobular type, c-ErbB2 expression/invasive type, among others (3,4). Recently, molecular studies have shown that breast cancer development involves not only alterations in oncogenes and tumor suppressor genes, but also epigenetic alterations in genes related to cellular growth, survival, motility and differentiation (5). Disturbance of epigenetic regulation of gene expression through methylation might contribute to tumorigenesis. In this context, altered histone modifying enzymes such as methyltransferases may play important roles in the process of carcinogenesis.

The mixed lineage leukemia (MLL) family is characterized by a conserved catalytic SET domain, responsible for the methyltransferase activity and comprises five genes [MLL (ALL-1), MLL2, MLL3, MLL4 and MLL5].

The mixed lineage leukemia gene (MLL or ALL-1, HRX and Htrx1) (Gene ID: 4297) is located in human chromosome 11q23 and is frequently involved in chromosomal translocations in leukemia (6,7). MLL seems to function as a mammalian counterpart of Drosophila trx, which maintains the expression of multiple Hox genes during development. Mll-deficiency in mice results in failure to express Hox genes and in embryonic lethality. Heterozygosity for disrupted MLL, by homologous recombination in mouse embryonic stem cells, leads to retarded growth, hematopoietic abnormalities and skeleton malformations (8).

The MLL2 gene (Gene ID: 8085) is located on human chromosome 12q13.12 and encodes a protein with domains similar to MLL. MLL2 was shown to be associated with Pax7 (paired-box transcription factor), forming the Wdr5-Ash2L-MLL2 histone methyltransferase (HMT) complex that directs tri-methylation of histone H3 lysine 4, when bound to Myf5. These chromatin modifications stimulate transcriptional activation of target genes to regulate entry into the myogenic developmental program (9). Another study, based on muscle-specific myogenin gene, suggested that transcriptional activators Mef2d and Six4 mediate recruitment of trithorax group proteins Ash2L/MLL2 and UTX to MyoD-bound promoters to overcome the polycomb-mediated repression of muscle genes, allowing transcriptional activation of muscle-specific genes (10). MLL and MLL2 genes show highly similar exon/intron structures. The proteins encoded by these genes are ubiquitously expressed among adult human tissues and although they share similar domains, they seem to act in a different manner. MLL2 could not compensate for the heterozygosity of the disrupted MLL, by homologous recombination in mice (8), which indicates that their functions seem to be at least partially non-overlapping.

MLL3 (Gene ID: 58508) maps to chromosome 7q36.1, a region usually deleted in hematological malignancies (11). MLL4 (Gene ID: 9757) is located on chromosome 19q13.1. MLL3 and MLL4 are found as part of a complex named ASCOM (for ASC-2 COMplex), that acts as a p53 co-activator. This complex is required for the expression of endogenous p53-target genes in response to the DNA damage (12). Deletions or truncations in Mll, Mll2 and Mll3 in mice show different phenotypes, suggesting that MLL protein functions are not redundant (8,13).

MLL5 (Gene ID: 55904) is located in human chromosome 7q22.1. The nuclear protein encoded shows single PHD and SET domains and lacks other conserved sequence elements, like its Drosophila homolog CG9007. MLL5 shows ubiquitous tissue expression and is highly conserved. It is the latest addition to the mammalian Trithorax/MLL gene family and recent studies show that the MLL5 may function in a different manner, without methyltranferase activity, which is consistent with its different sequence of the SET domain compared to the rest of the MLL family of H3K4 methyltransferases (14). It was suggested that MLL5 would have an indirect mechanism regulating expression of histone-modifying enzymes, such as LSD1 and SET7/9 (15). MLL5 was shown to promote myogenic differentiation through positive regulation of the pro-myogenic genes Pax7, Myf5 and myogenin, which are impaired in MLL5 knockdown, leading to serious differentiation defects (15). Mll5 deficiency in mice leads to growth retardation, male infertility and defects in multiple hematopoietic lineages (16). Therefore, the defects found could be associated with deregulation of cell cycle control.

Some studies have shown that MLLs are critical co-activators in estrogen (E2)-dependent regulation of E2-responsive genes, through LXXLL motifs (NR boxes) (17,18). The suppression of MLL2 decreased expression of estrogen receptor (ER) target genes (18). MLL3 and MLL4 coordinate with ERs and act in transcriptional regulation of HOXC10 in the presence of estrogen (19). MLL3 and MLL4, as members of the ASCOM complex, play essential roles in gene activation mediated through nuclear receptors, which bind hormone responsive elements (12). The activity of the MLL family in hormone-mediated gene activation and signaling could be helpful to understand its involvement in breast carcinogenesis.

Here we present an analysis of the patterns of MLL family gene expression in seven breast cancer cells lines, twelve human normal tissues and eight breast tumors and their adjacent normal breast tissues. Our results may help to provide a better understanding of the role of the MLL genes in breast carcinogenesis.

Materials and methods

Cell lines and culture

Seven breast cancer lines were used on this study: HCC-1954, MCF-7, CAMA-1, SKBR-3, MDA-MB-231, MDA-MB-436 and MDA-MB-468. The cell lines HCC-1954, MCF-7 and MDA-MB-231, obtained from the ATCC, were cultured in our laboratory in recommended media and checked periodically for mycoplasma contamination. The main characteristics of the cell lines are summarized in Table I (20). Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM, Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (Sigma-Aldrich) and penicillin/streptomycin (100 U/ml; Life Technologies, Carlsbad, CA, USA) at 37°C and 5% CO2. Total RNA from these cells was extracted and used for expression analysis as described below. Additionally, total RNA from the remainder cells were kindly provided by the Ludwig Institute for Cancer Research, São Paulo, and were included in the gene expression analysis.

Table I. Characteristics of the breast cancer cell lines.

Table I.

Characteristics of the breast cancer cell lines.

Cell lines	Origin	ER	PR	Cerb-B2	Tumor type	Characteristics
HCC-1954	Primary breast cancer	−	−	+	DC	Ductal carcinoma with no lymph node metastases
MCF-7	Pleural effusion	+	+	*	IDC	Invasive ductal carcinoma; weakly invasive
CAMA-1	Pleural effusion	+	−	*	AC	Metastatic adenocarcinoma
SKBR-3/AU565	Pleural effusion	−	−	+	AC	Metastatic adenocarcinoma
MDA-MB-231	Pleural effusion	−	−	*	AC	Metastatic adenocarcinoma; highly invasive
MDA-MB-436	Pleural effusion	−	−	*	IDC	Metastatic adenocarcinoma; highly pleomorphic
MDA-MB-468	Pleural effusion	−	−	*	AC	Metastatic adenocarcinoma

{ label (or @symbol) needed for fn[@id='tfn1-ijo-43-02-0653'] } ER, estrogen receptor; PR, progesterone receptor; Cerb-B2, epidermal growth factor receptor 2 overexpression; DC, ductal carcinoma; IDC, invasive ductal carcinoma; AC, adenocarcinoma;

* Unavailable data.

Normal tissues

A commercial panel (OriGene Technologies, Rockville, MD, USA) of total RNA from twelve human normal tissues from different organs (lung, small intestine (SI), brain, colon, kidney, muscle, liver, spleen, heart, testis, stomach and placenta) was used in this study. Total RNA of each normal tissue was used to verify the expression levels of the genes studied.

Tumor specimens

The sample collection protocol was approved by the Research Ethics Committee of the Faculty of Health Sciences, University of Brasília, Brazil, based on resolution 196/96 of the National Health Council/Brazilian Ministry of Health, project no. 025/09. Written informed consent was obtained from all participants. The samples were collected from surgically removed breast tissue from female patients diagnosed with breast cancer at the Hospital of the University of Brasilia. A total of 14 samples (6 normal and 8 tumor samples) was used. The identification of tumor tissue from the removed samples was based on the histopathological examination and all types of breast cancer were included in our analyses. Samples were selected based on tumor content (minimally 80% tumor) as determined by microscopic pathological analysis. A sample of normal tissue was collected from each patient whenever possible. The clinical and histopathological characteristics of the patients are summarized in Table IV.

Table IV. Clinical and histopathological characteristics of the 8 breast cancer patients.

Table IV.

Clinical and histopathological characteristics of the 8 breast cancer patients.

Samples	Breast affected	Age	Histological type	Histological grade	Stage	Neoadjuvant chemotherapy	C-erb-B2	ER	PR
1	Left	58	DCI	High	T3N2AM0	Yes	+	−	−
2	Left	54	DCI	Intermediate	T4DN3M0	Yes	−	+	+
3	Left	64	DCI	Intermediate	YPT4PN1M0	Yes	−	−	−
4	Right	43	DCI	High	T2N1M0	Yes	*	*	*
5	Left	56	DCI	High	T2N1AM0	Yes	−	−	−
6	Right	60	DCI	High	T2N3M0	Yes	−	+	+
7	Right	61	LCI	Intermediate	T2N0M0	No	+	+	+
8	Right	32	DCI	Low	T2N0M0	Yes	+	+	+

{ label (or @symbol) needed for fn[@id='tfn5-ijo-43-02-0653'] } DCI, ductal carcinoma in situ; LCI, lobular carcinoma in situ; Cerb-B2, epidermal growth factor receptor 2; ER, estrogen receptor; PR, progesterone receptor.

* Unavailable data.

RNA extraction and cDNA synthesis

Total RNA was isolated using TRIzol LS reagent, according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA, USA). Single-stranded complementary DNA was generated from total RNA obtained from cell lines, normal tissues and clinical samples with reverse transcriptase and random primers, using the Applied Biosystems High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Carlsbad, CA, USA).

Quantitative PCR (qPCR)

Reactions of qPCR were performed on a StepOnePlus™ Real-Time PCR System (Applied Biosystems) using TaqMan Universal PCR MasterMix and TaqMan Gene Expression Assays (Hs01061596_m1 for MLL, Hs00231606_m1 for MLL2, Hs00419011_m1 for MLL3, Hs00207065_m1 for MLL4 and Hs00218773_m1 for MLL5, Applied Biosystems), according to the manufacturer’s instructions. Briefly, 2 μl of each sample template was added to a 10 μl final reaction volume per well. Amplification conditions were as follows: 2 min at 50°C and 10 min at 95°C on holding stage and then 40 cycles of 15 sec at 95°C and 1 min at 60°C. Gene expression values are reported as ratios between the amplification levels of genes of interest and that of endogenous control gene (Hs99999903_m1 for β-actin, Applied Biosystems), which provides a normalization factor for the amount of RNA isolated from a specimen. This was subsequently normalized with the ratio obtained in control samples (relative expression level). Each assay was performed in triplicate for each sample.

qPCR data analysis

To determine the relative quantification (RQ) of gene expression, the data were analyzed using the comparative quantification Ct method (ΔΔCt) (Applied Biosystems). Briefly, the mean of Ct (cycle threshold) values of the replicates were calculated and normalized by subtracting the Ct value of the co-amplified endogenous control gene to yield a ΔCt value. The ΔCt of a random control calibrator sample (1X sample) was subtracted from the ΔCt of the other samples to yield a ΔΔCt value. The amount of target gene, normalized to an endogenous reference and relative to a calibrator was converted into relative quantification by the formula: 2−ΔΔCT. For the cell lines, normal tissues and clinical samples used, the MLL genes expression was normalized using the β-actin as an endogenous control. Expression level was considered altered when augmented or diminished ≥2X compared to the control calibrator sample, chosen for each analysis. To check the statistical significance of the relative quantification of the genes, the non-parametric Mann-Whitney U test was performed for the pools of normal (n=6) and tumor (n=8) samples. The one-sample Wilcoxon signed-rank test was used for the expression results of the MLL genes family in seven cancer cell lines, to test if the samples means significantly differed from the hypothesized value (sample calibrator relative quantification: 1) The statistical tests were considered significant when the P-value was <0.05 (CI 95%). Calculations were performed using the software SPSS version 20 (SPSS Inc., Chicago, IL, USA). Additionally, we searched for MLL gene expression profiles in breast cancer datasets, available in the database Oncomine, to compare with our most significant findings.

Results

MLL genes are downregulated in breast cancer cell lines

The results of the MLL gene expression analysis in seven cancer cell lines using the comparative quantification method ΔΔCT are shown in Fig. 1. All genes showed expression levels diminished in comparison to the normal breast sample calibrator (1X sample) (P<0.05). The MLL2 and MLL5 were the genes with the lowest expression level in all cell lines, with the exception of HCC-1954. The HCC-1954 lineage is the least aggressive among the breast cancer cell lines studied, with no lymph node metastases (Table I). Although all MLL genes were also downregulated in HCC-1954 cells in relation to the normal breast calibrator control, this cell line showed the least reduced expression among the breast cancer cell lines.

Figure 1. Expression levels of the MLL genes in 7 breast cancer cell lines, using the ΔΔCt method. Sample calibrator, normal breast clinical sample (relative quantification = 1). All genes showed expression levels below the calibrator sample (P<0.05).

MLL genes are differentially expressed among normal tissues

In order to determine the MLL genes expression pattern in normal human tissues, quantitative real-time polymerase chain reaction (qPCR) was performed using a commercial panel of total RNA from normal tissues obtained from OriGene®. Expression analysis using the comparative quantification method ΔΔCT was performed. The kidney tissue, which showed a median expression level, was chosen as a sample calibrator (1X sample) and a normal breast clinical sample was included for the relative quantification. Relative quantification for the normal tissues can be seen in Fig. 2 and in Table II. These results evidence a heterogeneous tissue expression for the MLL family genes, suggesting that MLL proteins functions are not redundant, but acting in a tissue-specific manner.

Figure 2. Expression levels of the MLL genes in 12 normal human tissues and in a normal breast clinical sample. Random sample calibrator, normal kidney (relative quantification = 1).

Table II. Relative quantification of MLL genes in normal human tissues.

Table II.

Relative quantification of MLL genes in normal human tissues.

	MLL	MLL2	MLL3	MLL4	MLL5
Kidney	1	1	1	1	1
Lung	2	2	2	1	2
Small intestine	1	3	3	3	5
Brain	5	3	3	1	3
Colon	1	2	1	3	1
Muscle	9	10	8	3	6
Liver	8	1	163	1	20
Spleen	1	1	1	2	1
Heart	1	2	50	1	6
Testis	3	2	2	5	1
Stomach	2	1	23	1	6
Placenta	14	1	12	1	28
Breast	1	1	1	3	7

[i] Sample calibrator, kidney (1X sample). Bold, upregulated expression in fold changes; underlined, downregulated expression in fold changes.

MLL2 is significantly downregulated in clinical samples

The paired expression analysis between normal and tumor samples from the same patient can be seen in Fig. 3 and the relative quantification values in Table III. The quantification analysis for the pools of normal (n=6) and tumor (n=8) clinical samples showed a decreased expression level of all genes in the tumors when compared to the pool of normal samples (Fig. 4). The expression of MLL was reduced in 5 of the 8 (62.5%) tumor samples and remained the same in the 3 other (37.5%) samples. The expression of MLL2 was reduced in all the 8 tumor samples (100%). The expression of MLL3 was reduced in 4 of the 8 (50%) tumor samples, remained the same in the 3 other (37.5%) samples and was increased in 1 sample (12.5%). The expression of MLL4 was reduced in 5 of the 8 (62.5%) tumor samples, remained the same in the other 2 (25%) samples and was increased in 1 sample (12.5%). For the MLL5, the expression was reduced in 6 of the 8 samples (75%) and remained the same in 2 (25%). Although this observation reveals a significantly suppressed expression of MLL2 and MLL5 in the majority of tumors, the small number of cases does not allow strong associations with the clinical and pathological characteristics of the patients (Table IV).

Figure 3. Relative quantification of the MLL genes by qPCR using the comparative quantification method ΔΔCT in 6 paired clinical samples. Sample calibrator, normal, non-tumor tissue from the same donor patient (relative quantification = 1).

Figure 4. Relative quantification of the MLL genes in pools of normal (n=6) and tumor (n=8) clinical samples, compared using the independent Mann-Whitney U test (P<0.05 and CI 95%). The P-values for each gene are shown above the bars.

Table III. Relative quantification of MLL genes in tumor samples compared to their normal counterparts.

Table III.

Relative quantification of MLL genes in tumor samples compared to their normal counterparts.

	MLL	MLL2	MLL3	MLL4	MLL5
1T	6	15	3	2	9
2T	30	22	14	13	14
3T	2	4	1	2	1
4T	6	8	11	14	20
5T	4	1	6	2	6
6T	67	85	34	49	26

[i] Sample calibrator, normal tissue from the same donor patient (1X sample). Bold, upregulated expression in fold changes; underlined, downregulated expression in fold changes.

To overcome our limited clinical breast cancer cohort, we searched for MLL genes expression profile available in the Oncomine microarray database. We found that MLL2 was significantly downregulated in a set of 53 invasive breast carcinomas compared to 6 normal breast tissues (P=8.19E-16) (21). MLL5 was found to be 2-fold downregulated in a set of 7 invasive breast carcinomas compared to 15 normal breast tissues (P<0.005) (22). These data corroborate our findings indicating that the levels of MLL2 and MLL5 are widely downregulated in breast carcinomas (Fig. 5).

Figure 5. MLL2 and MLL5 are downregulated in breast cancer. Gene expression data in Oncomine was analyzed. The thick bars in the boxes are average expression levels and the boxes represent 95% of the samples. The error bars are above or below the boxes and the range of expression levels is enclosed by two dots.

Discussion

Despite the advances that have been made in uncovering potential roles for protein methyltransferases in human cancers, it is still largely unknown how they contribute to carcinogenesis. Deregulated methylation mechanisms in cancer cells may be due to enzyme mutations or altered expression (23,24). Recent studies have shown a relationship between SMYD3, a lysine methyltransferase and breast carcinogenesis (25). SMYD3 modulates the chromatin structure through specific methylation of lysine 4 of histone 3 (H3K4), an epigenetic alteration that leads to uncontrolled cell proliferation (25). Several other lysine methyltransferases have already been shown to be involved in cancer. For example, the gene EZH2 was overexpressed in different cancer tissues, including breast and prostate (26). The SETD2, a potential tumor suppressor gene, was downregulated in breast cancer samples and in high-grade tumors (27). Deregulation of PRDM14 gene has been associated to several human tumors, including an overexpression state in breast cancer cells (28). Although little is known about the exact roles of MLL family genes in breast carcinogenesis, some recent studies have linked alterations in individual MLL genes with cancer progression. The present study provides the first comparative expression analysis of the five MLL family genes in breast cancers. In this study, we investigated the differential expression of MLL genes in breast tumor samples and cell lines, as well as in normal tissues. All MLL genes had a significantly decreased expression in the seven breast cancer cell lines analyzed. It is noteworthy that although these cell lines represent different stages of the disease, presenting different receptors and characteristics of aggressiveness (Table I), they all showed a similar expression profile for all five MLL genes. In the HCC-1954 cell line, which represents a less aggressive breast cancer, with no lymph node metastases, a subtle higher expression for the MLL genes when compared to the other cell lines was found (Fig. 1). This observation suggests that expression of MLL family genes may decrease as disease progresses. Cell lines exhibit substantial genomic, transcriptional and biological heterogeneity found in primary tumors and represents a good system to study primary breast tumor features (20). Indeed, the results obtained in cell lines were in accordance to those obtained in clinical samples.

The analysis of the MLL genes in normal tissues showed highly heterogeneous expression among the different tissues studied. This finding indicates a possible tissue-specific function for each individual gene. The only consensus seems to be the high expression of all MLL genes in skeletal muscle and brain. For instance, MLL2 gene showed a markedly increased expression in skeletal muscle (>10-fold) and heart (>2-fold), which is in agreement with other studies showing the involvement of this gene with myogenic developmental program (9,10). It is of interest that in normal breast tissue, MLL5 has a markedly higher expression among all MLL family genes. The analysis of the primary breast clinical samples clearly pointed to decreased expression of the genes MLL2 and MLL5 in breast cancer. The other MLL genes did not show a significant difference in expression between tumor and normal samples.

The MLL gene had a slightly decreased expression in the tumor samples, compared to the normal ones (P= 0.142). In all the cancer cell lines MLL was downregulated in comparison to the pool of normal samples (P<0.05). Mammalian MLL and MLL2 have a role in long-term maintenance of Hox and other gene expression patterns during development (8,13). Several rearrangements of the MLL gene underlie the majority of infant acute leukemias, as well as of therapy-related leukemias occurring in cancer patients treated with inhibitors of topoisomerase II (29). Higher transcript levels of MLL were associated with patients who remained disease-free compared to those with bone metastasis. (30). Liu et al identified MLL as a key constituent of the DNA damage response pathway, showing that deregulation of the S-phase checkpoint caused by MLL translocations may contribute to the pathogenesis of human MLL leukemias (31).

The comparative analysis for MLL2, revealed a significantly decreased expression of this gene in tumor samples when compared to normal ones (P= 0.043). In the analysis using breast cancer cell lines, MLL2 was the second gene with the most prominent downregulation, after MLL5 (P<0.05). The Oncomine database results corroborated our data, showing that MLL2 was found markedly downregulated in a set of 53 invasive breast carcinomas compared to 6 normal breast tissues (P=8.19E-16) (21). These results are consistent with several other studies on this gene, in different types of cancer. Curiously, Huntsman et al (32) showed that MLL2 is amplified in some solid tumor cell lines (pancreatic and glioblastoma cell lines), with the exception of breast cancer cell lines. Morin et al (33) detected, in lymphoma patients, several somatic mutations distributed across MLL2, with some mutations affecting both MLL2 alleles or leading to loss of heterozygosity (LOH), which is consistent with the complete, or almost complete, loss of MLL2 in the tumor cells studied. They suggested that since the majority of the somatic mutations found in MLL2 were inactivating, MLL2 probably acts as a tumor suppressor of significance in lymphomas. Pasqualucci et al (34) found that the MLL2 gene was the most frequently mutated gene in diffuse large B-cell lymphoma (DLBCL) with most cases of inactivating mutations, generating truncated proteins lacking the entire or part of the C-terminal cluster of conserved domains (including the SET domain). Since most of the cases affected a single allele, they suggested a role for MLL2 as a haploinsufficient tumor suppressor. Dalgliesh et al (35) demonstrated that the MLL2 is frequently mutated in clear cell renal cell carcinoma (ccRCC), with silence, missense and truncating mutations. It was identified as a likely ccRCC cancer gene in statistical analyses. Finally, MLL2 was shown to be involved in several cellular signaling pathways, such as p53 and cAMP, regulating different sets of genes. The link of MLL2 to the p53 pathway corroborates a possible role of MLL2 as tumor suppressor gene. Also, among the categories of genes regulated by MLL2, were those that respond to nuclear hormones, including genes from the retinoic acid signaling (36).

MLL3 did not show a statistical significant differential expression between the tumor and normal samples used in this study (P=0.142), although in Fig. 4, this tendency can be noted. In the seven breast cancer cell lines analyzed (Fig. 1), there is a clear decreased expression of MLL3 (P<0.05). The MLL3 seems to be frequently mutated in many types of human tumors and a recent study detected that 40% of the primary breast tumor samples showed reduced expression of this gene, suggesting its possible role in the development of breast cancer as a tumor suppressor (37). Parsons et al (38) found, that MLL3, as well as MLL2, harbors several inactivating mutations in medulloblastoma that seem to be driver mutations. In the case of MLL2, 67% of the mutations were predicted to truncate the encoded proteins. This observation led to the conclusion that both genes act as tumor suppressors inactivated by mutation in medulloblastoma. MLL3 also showed inactivating mutations in colorectal cancer samples (5,39) and reduced expression in primary breast tumor samples, suggesting a role in cancer development (37).

We found no significant difference between normal and tumor sample expression of MLL4 (P=0.282). In the seven breast cancer cell lines (Fig. 1), MLL4 showed decreased expression, although it was the less downregulated gene among the other members of the family (P<0.05). MLL4 is mostly reported as a member of the complex ASCOM (for ASC-2 COMplex), that acts as a p53 co-activator. Nonetheless, it is suggested that MLL3 and MLL4 are found in distinct ASC-2-containing complexes rather than in a common ASC-2 complex, with only partially redundant functions (12).

For the MLL5 gene, our results revealed a clear tendency of diminished expression in the clinical tumor samples (P=0.059). It also showed a significant decreased expression in the cancer cell lines, representing the most downregulated gene among the MLL family for the seven lineages (P<0.05). The Oncomine database results for MLL5 expression in breast cancer also corroborated the tendency of our data, with a 2-fold downregulation of MLL5 in a set of 7 invasive breast carcinomas compared to 15 normal breast tissues (P<0.005) (22). MLL5 gene is the latest addition to the mammalian Trithorax/MLL gene family and recent studies show that the MLL5 may function in a different manner, without histone methyltransferase activity (16). MLL5 lacks DNA-binding motifs of A-T hooks and methyltransferase homology motifs, both common to other MLL protein members. Instead of binding directly to DNA, MLL5 may modulate transcription indirectly via protein-protein interactions through its PHD (plant homeodomain, a nuclear protein-interaction domain) and SET domains (14). Overexpression of MLL5 prevented cell cycle progression pointing to a role of this gene in the cell cycle regulatory network (6). MLL5 seems to form intranuclear protein complexes, similar to other tumor suppressors, that may play a role in chromatin remodeling and cellular growth suppression (6). In cell line transfection studies, overexpression of MLL5 transcripts inhibits cell cycle progression, suggesting a role for MLL5 in tumor suppression (6,16). Moreover, MLL5 was identified as a candidate suppressor gene in 7q22, a chromosomal site frequently deleted in patients with aggressive acute myeloid leukemia and high MLL5 transcript levels seem to be associated with a favorable outcome in this disease (40). Therefore, it is possible that in cancer, the truncated or absent MLL5 protein is unable to form the intranuclear protein complexes to interact with chromatin, allowing the cell cycle to progress.

Although the complete genome screening is essential to unveil the genetic profile of complex diseases, the comparative analysis of a specific group of genes can be valuable to better understand the complexity of cancer mechanisms. In the present study, we analyzed the expression profiles of the MLL family in different breast cancer cell lines and clinical samples. Our results demonstrate a significant decreased expression of the MLL2 and a clear tendency of MLL5 down-regulation in tumor samples, which could indicate a possible role as tumor suppressor genes. Since MLL2 has a role in maintenance of Hox genes expression during development and MLL5 seems to have a role in cell cycle regulation, they may both be involved in the regulation of other genes and pathways (8,13). Furthermore, MLLs seem to have other roles in regulating gene activation beyond their methyltransferase activities, including responses under a hormonal environment (12,17–19). Our results are consistent with several studies on these genes in different cancer types, as discussed above, confirming that our high-throughput approach was reliable enough for validation of the genes expression profiles. Although these results point to an involvement of MLL2 and MLL5 in breast cancer, the number of cases does not enable robust statistical analyses neither associations with the clinical characteristics. Therefore, further studies using a larger clinical breast cancer cohort are required to determine the exact role of these methyltransferases as epigenetic regulators in cancer development.

The correlation between various methyltransferases and breast cancer highlights the importance of this protein family in the progression of this disease. Further study is required to reinforce the importance of the MLL family as a target for breast cancer therapy and will help elucidate the mechanisms involved in the regulation of its activity. Our findings reveal the importance of these genes in breast carcinogenesis and may contribute to the identification of novel strategies to treat breast tumors.

Acknowledgements

This study was supported by grants from CAPES, FAPDF and CNPq. The authors thank Dr Anamaria Camargo Aranha from the Ludwig Institute for Cancer Research, São Paulo, for providing total RNA from several breast cancer cell lines. The authors received financial support from the University of Brasilia UnB-DPP (Decanato de Pesquisa e Pós-graduação).

References