Background
MLL2 (MLL) [Swiss-Prot: Q9UMN6] is a member of the MLL/trx family of proteins. It contains several evolutionarily conserved domains [
1] including AT hooks at the N-terminus, cluster of PHD (plant homeodomain) zinc fingers associated with a bromodomain, and a SET (
s uppressor of variegation,
e nhancer of zeste,
t rithorax) domain at the C-terminus [
1]. The full length MLL2 (MLL2
FL) is an uncleaved precursor protein with a predicted molecular weight of ~290 kD. MLL2
FL precursor protein undergoes post-translational proteolytic maturation, which is critical to its normal biological activity [
2]. The enzyme responsible for MLL2 cleavage is taspase 1, and its consensus cleavage site (D/GVDD) is at a.a. 2063 [
2]. Proteolytic cleavage generates a large N-terminus fragment with a predicted molecular weight of 215 kD, and a smaller C-terminus fragment which separates at ~75 kD in a denaturing gel. The cleaved fragments subsequently associate to generate a stable, functional, noncovalent heterodimeric complex [
2].
The SET domain of MLL2 possesses histone H3 lysine 4 (H3K4) methyltransferase activity, and is an important component of the multi-protein complex involved in epigenetic gene regulation and embryonic development [
3‐
5]. For example,
in vitro, MLL2 complex has been shown to associate with Pax7, a transcription factor, and activate myogenic genes through H3 K4 methylation [
4].
In vivo, Mll2 is shown to be required for normal embryonic development in mice [
5‐
7]. A survey of the literature shows that several proteins with a primary function in epigenetic regulation and/or embryonic development are often aberrantly expressed in cancer. This finding is related to the observation that embryonic development and tumorigenesis share several common pathways [
8]. Furthermore, proteins with chromatin remodeling motifs, such as PHD zinc fingers and SET domains, are often aberrantly expressed in tumors [
9‐
11]. Considering all these features of MLL2, along with its significant structural similarity to MLL, we suspected that the
MLL2 gene or its product may be altered in cancer, similar to it's paralog MLL, which is directly linked to haematopoietic tumorigenesis [
12]. A literature survey, however, found only one published report describing
MLL2 amplification through complex chromosomal rearrangements and duplications in human cancer cell lines [
13]. Querying ONCOMINE, a publicly available source of gene expression data sets in cancers [
14], we identified a few studies which listed
MLL2 as one of the deregulated genes in some cancers-including melanoma, bladder and lung carcinomas-when compared to the corresponding normal tissues [
14]. Subsequently, tissue microarray based preliminary screening in our laboratory also indicated that MLL2 may be disrupted in certain cancers. We, therefore, decided to investigate MLL2 expression in breast and colon cancer cell lines, and then substantiated our findings in archived formalin fixed paraffin embedded (FFPE) tumor tissues from patients with confirmed diagnoses of breast and colon cancers.
In order to study MLL2 in breast cancer cells, we selected a panel of six breast epithelial cell lines representing non-tumor breast epithelial derived cell lines (184A1 and MCF 10A) [
15], weakly invasive breast tumor cell lines (T47D and MCF 7) [
15,
16] and highly invasive breast tumor cell lines (MDA-MB-157 and MDA-MB-231) [
16]. For investigating MLL2 in colon cancer cells, we selected three cell lines derived from well-differentiated colon carcinomas (HT29, DLD-1 and Ls174T) [
17‐
19] and three from poorly differentiated colon carcinomas (Lovo, Colo 205 and SW 480) [
18,
20,
21]. We then substantiated our observations in cell lines by investigating MLL2 levels in breast and colon cancer tissues. Here we report that MLL2 expression is disrupted in invasive tumor cell lines and invasive carcinomas.
Discussion
In the present study we observed that: (a) MLL2, primarily regarded as a nuclear protein with nuclear localization signals, displayed significant cytoplasmic localization in both normal and malignant cells, (b) mammary and colonic cell lines derived from highly invasive tumors exhibited altered sub-cellular distribution and proteolytic processing of MLL2 compared to non-tumor/less-invasive-tumor cell lines, and (c) MLL2 is overexpressed in breast and colon tumors tissues compared to the corresponding normal adjacent tissues.
MLL2 is primarily regarded as a nuclear protein. However, we observed both nuclear and cytoplasmic MLL2 in cell lines and tissue sections from breast and colon. In the breast epithelial cell lines, MLL2 specific bands representing the cleaved (75 kD) protein were observed in the cytoplasmic fractions of not only the tumor cell lines but also the non-tumor cell lines, 184A1 and MCF10a (Fig
1A); while bands representing the full-length, unprocessed MLL2 (290 kD) were present in the highly invasive tumor cell lines alone, and which may be related to oncogenic activity. The nuclear-cytoplasmic localization pattern of MLL2 was also evident in the benign sections of both breast and colonic tissues, but it was not possible to delineate the cleaved from the uncleaved MLL2 as the antibody used in this study cannot differentiate one from the other
in situ (details of antibody specificity are described in Methods section). Considering this cytoplasmic presence of MLL2 in the non-tumor cell lines and non-tumor tissue sections, it is possible that MLL2 may have a yet unidentified function in the cytoplasm, besides its role in epigenetic regulation in the nucleus [
22,
23]. However, a notably increased cytoplasmic presence in the cancer cell lines and cancer tissues, together with the presence of MLL2
FL could be a tumor-related anomaly resulting from overexpression or increased protein/RNA stability.
Our study of the breast and colonic cell lines revealed two notable trends related to the proteolytic processing of MLL2. First, we observed a gradient in the intensities of the 75 kD signal in the nuclear fractions of both breast and colonic cell lines (Fig.
1A &
1C). In the breast epithelial cell lines, the intensities of the 75 kD nuclear MLL2, indicative of normal proteolytic processing, showed a decreasing trend with increasing malignancy from MCF10A to MDA-MB-231 except for MCF7. Of these cell lines, 184A1 and MCF10A are non-tumorigenic, MCF7 [
16] and T47D are tumorigenic [
15] but weakly invasive [
24], while MDA-MB-157 and MDA-MB-231 [
16] are highly invasive [
15,
24,
25]. The non-tumor cell line, 184A1, did not fit this trend and this discrepancy could be attributed to an overall lower level of MLL2 in 184A1, and/or the difference in proliferation rate--184A1 cells are reported to have a relatively lower proliferation rate compared to MCF10A [
25]. In the colonic epithelial cell lines a similar intensity gradient was noted for the 75 kD nuclear MLL2 fragment. Of these cell lines, HT29 [
17], DLD-1 [
18] and Ls174T [
19] are derived from well-differentiated tumors while, Lovo [
18], Colo205 [
20] and SW480 [
21] are derived from poorly differentiated tumors; and the 75 kD nuclear MLL2 was least in the Colo205 and SW480 cell lines.
The second notable trend was the presence of an additional band of 290 kD size (corresponding to the uncleaved precursor MLL2
FL) in the more invasive/poorly differentiated cell lines. More importantly, the MLL2
FL (290 kD) signal intensity increased in the nuclear and cytoplasmic fractions as the 75 kD nuclear signal decreased (Fig.
1A &
1C). This uncleaved MLL2
FL observed in the cytoplasmic and nuclear fractions of the advanced tumor cell lines could be a consequence of insufficient endogenous taspase 1 required for processing the excess MLL2. Since proteolytic cleavage could also be impaired as a consequence of non-cleavable mutations in MLL2
FL or the absence/decrease in the endogenous taspase 1 that cleaves MLL2 [
2], we looked for mutations in the
MLL2 sequence coding for taspase 1 cleavage site, and also examined taspase 1 protein levels in the cell lines. Sequence analysis of the coding region for taspase 1 cleavage sites in cell lines carrying MLL2
FL did not reveal any alterations in the cleavage site encoding sequences. In addition, we found that the taspase 1 protein levels did not vary in parallel with the presence or absence of MLL2
FL. Taspase 1 protein levels were consistent across the six breast tissue cell lines, irrespective of the presence of MLL2
FL. Although we did not observe consistent levels of taspase 1 across the colonic cell lines, we did observe that the presence or absence of MLL2
FL and its levels, failed to correlate with taspase 1 level. That is, higher levels of taspase 1 did not correspondingly correlate with decreased levels of MLL2
FL or its total absence. These results suggest that the presence of MLL2
FL in the invasive cell lines is not a consequence of diminished levels of taspase 1 or a mutated cleavage site in the
MLL2. Further investigation is required, to determine the cause and consequences of MLL2
FL in the nucleus, which might be related to the shift in the nuclear-cytoplasmic localization of MLL2 in the invasive cell lines. Whatever may be the cause, our results suggest that the presence of precursor MLL2
FL is associated with a higher degree of malignancy.
According to an earlier report [
2], proteolytic processing of MLL2
FL is crucial to its stability, sub-nuclear localization, and methyltransferase activity. Impaired proteolytic maturation could result in significant changes in the normal epigenetic regulatory activities of MLL2. It has been shown
in vitro that MLL2 forms a multiprotein complex with Wdr5-Ash2L [
26] and associates with proteins like Pax7 and NF-E2 to direct histone lysine methylation at specific gene loci [
3,
4]. MLL2 specific histone methylation complex is also known to associate with the tumor suppressor protein, menin, and mediate histone methylation at
Hoxc8 locus [
27]. Given the critical role for MLL2 in histone methylation activities, we believe that proteolytically immature and/or inappropriately expressed MLL2 may fail to effectively associate with the other members of the histone-methyltransferase complex, which in turn can adversely affect its role in epigenetic gene regulation.
In our analysis of the breast and colonic cell lines by reverse transcription (RT)-PCR, we also observed that MLL2 RNA levels were highest in the invasive tumor cell lines and least in the non-tumor/less-invasive-tumor cell lines (Fig.
1B &
1D). This trend in the MLL2 RNA levels was consistent with our observation of the overall increase in protein levels in the invasive tumor cell lines and in the tumor tissues. Since a real-time measurement was not performed on these cell lines it is not known if the increased levels of RNA resulted from an increased rate of transcription. The observed gradient in RNA levels in the breast and colon cell lines could also be due to differences in RNA stability. Though the RT-PCR results are not strictly quantitative, the results do indicate that MLL2 RNA levels are more abundant in the highly invasive/less differentiated cell lines.
Finally, our study on tissue sections from breast and colon cancer patients revealed that immunohistochemically detected MLL2 is significantly increased in tumor tissues. In breast tumors from patient samples, cytoplasmic MLL2 was significantly overexpressed as compared to normal adjacent tissues, and in colon tumors both cytoplasmic and nuclear MLL2 were significantly overexpressed when compared to adjacent benign mucosa. Since immunohistochemical signals can often arise from non-specific antibody reactions, we evaluated the specificity of antibody reactivity using a blocking peptide, a part of which represented the epitope recognized by anti-MLL2 antibody (detailed in Methods section). The peptide blocks the ability of the antibody to bind to its antigen. These experiments confirmed a high specificity of the antibody to the MLL2 antigen (Fig.
3D).
Analysis of MLL2 expression data with clinicopathological variables revealed a small correlation between MLL2 overexpression and early tumor stages (breast and colon) and absence of lymph node involvement (colon). However, the number of cases in each category was too small and the association too tentative to draw a substantive conclusion at this juncture. Despite a lack of correlation with established clinicopathological variables, the elevated levels of MLL2 protein in both breast and colon cancer was significant. These results are in line with our observation of increased MLL2 protein and RNA levels in the cell lines. However, due to the lack of an appropriate antibody to distinguish the cleaved MLL2 from uncleaved MLL2
FL by immunohistochemistry, we cannot comment on the composition of the overexpressed protein as to whether it constituted more of the 75 kD fragment or the 290 kD MLL2
FL. The elevated levels of MLL2 in the tumor tissues could be the result of overexpression, genomic amplification, increased RNA and/or protein stability, or, at least in part, due to alterations in protein processing. For example, genomic amplification of
MLL2 (through complex chromosomal rearrangements or chromosomal duplications) resulting in four
MLL2 copies has been previously reported in one of the breast cancer cell line, MDA-MB-157, used in this study [
13].
Whichever may be the cause, deregulated expression of MLL2 and/or defective proteolytic processing may adversely impact MLL2 mediated histone methylation activities, and in turn, disrupt downstream target genes potentially involved in cell cycle or cell proliferation activities. Consequently, aberrantly expressed MLL2 driven epigenetic regulation may contribute to tumor growth and/or progression. If such is the case, deregulation of MLL2 may be a more generic feature in tumorigenesis rather than an event specific to a particular tumor type, as is indicated by our findings in both breast and colon tumors.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
TGN conceived the study design, conducted in-vitro studies on cell lines, participated in immunohistochemical analysis and drafted the manuscript. BVK conducted histopathological reading of slides and their interpretations. CES performed the immunohistochemical staining and statistical analysis. MBB participated in the in-vitro experiments on cell lines. NG participated in primer design, optimization of immunohistochemical experiments and helped to draft the manuscript. AP participated in optimizing RT-PCR reactions. JSR provided critical comments for improving the manuscript. KTF was involved in study design, revising the manuscript critically for important intellectual content, and gave final approval of the version to be published. All authors read and approved the final manuscript.