|Home | About | Journals | Submit | Contact Us | Français|
Cancer cells have alterations in chromatin organization, mostly a reduction in heterochromatin. How this process occurs during transformation and if it participates in the maintenance of a cancerous phenotype is not well understood. Here, using a transformation-inducible cell line, we analyzed the changes that occur in heterochromatin during transformation to a cancerous phenotype. After transformation, there is a reduction in heterochromatin bodies and a nuclear reorganization of HP1α. These occurrences correlate with reductions in H3K9me3 and H3K27me3 levels and with some of the enzymes that introduce these modifications. At the same time, there are increases in the KDM4B and KDM6A/UTX demethylases and an enhancement in the transcription of pericentromeric DNA that correlate with the reduction of H3K9me3 and the recruitment of KDM4B to these elements. The depletion of KDM4B and KDM6A/UTX has a more deleterious effect in transformed cells than in their progenitors, suggesting an important role for these enzymes in the survival of cancerous cells. These results provide new insights into heterochromatin dynamics during transformation to a cancerous phenotype as well as some of the participating mechanisms.
Constitutive heterochromatin, the most condensed form of chromatin, is present in human cells, particularly in the pericentromeric and subtelomeric regions as well as in repetitive elements along the genome. This kind of heterochromatin is enriched in nucleosomes that contain Histone 3 trimethylated at lysine 9 (H3K9me3), which is recognized by Heterochromatic Protein 1 proteins (HP1α, β and γ) 1. Tri-methylation of H3K9 is conducted by Suppressors of Variegation 3-9 Homologs 1 and 2 (SUV3-9H-1 and SUV3-9H-2) enzymes 2. Conversely, facultative heterochromatin is present in genes that are not expressed in particular moments of development or in specific cell types 3. Tri-methylation of Lysine 27 of H3 (H3K27me3) is a histone mark in genes that is characteristic of this type of heterochromatin. This modification is introduced by the action of the Polycomb Repressive Complex 2 (PRC2), whose methyl transferase activity is present in its Ezh1 or Ezh2 enzymes 4, 5. These modifications are reversible since histone demethylases that remove these specific marks have been identified. H3K9me3 can be demethylated by the KDM4A, KDM4B and KDM4C enzymes, and these enzymes also demethylate H3K36 and H1K26 6. H3K27me3 can be demethylated by the action of the UTX/KDM6A and KDM6B demethylases 7. These enzymes are α-ketoglutarate-dependent dioxygenases that belong to the Jumonji family of demethylases 6, 8.
It is well documented that the chromatin in cancer cells has different features than in normal differentiated cells 9. In general, it is less condensed and has broad changes in DNA methylation, H3K9me2, H3K9me3, H3K27me3, and histone acetylation 9. This is linked to changes in the levels of methylation-modifying enzymes. For example, there is evidence suggesting that UTX/KDM6A may act as a tumor suppressor in T-cell acute lymphoplastic leukemia, but the demethylation activity of UTX/KDM6A is required for tumor maintenance through the activation of the NOTCH pathway and Rb dependent tumors 10-12. It is also known that the loss of heterochromatin in cancer cells, in particular pericentromeric and telomeric chromatin, causes an increase in genome instability that favors the generation of aneuploidy, thus increasing the diversity of tumor cells 13,14. Furthermore, there are several reports that show that the KDM4A, KDM4B and KDM4C demethylases are overexpressed in different kinds of cancer and are required for maintaining the cancerous phenotype 15.
Although it is evident that heterochromatin undergoes dramatic changes during the generation of a cancer cell, how these alterations occur during the transformation to this phenotype is not well understood. A model is needed to study chromatin changes wherein the transformation to a cancerous phenotype occurs through incubation with an inducer in a cell line that it is immortalized but not cancerous; the MCF10-Er-Src line is such a model 16. In this model, a transformed phenotype is generated by the transient expression of Src, which induces the epigenetic switch from an immortalized human breast cell line to a stable and highly malignant transformed cell line 16. In fact this cell line has been acceptable used as a model of oncogenesis in a variety of studies 16-19. Using this system, we analyzed the broad heterochromatin changes that occur during cell transformation to a cancerous phenotype. After transformation, there is a reduction in heterochromatin centers as well as changes in HP1α distribution. These changes correlate with a reduction in the global levels of H3K9me3 and H3K27me3, an increase in the KDM4B and KDM6A/UTX histone demethylases and a decrease in the total levels of SUVAR39H1 as well as the components of the PRC2 complex. Furthermore, pericentromeric sequences increase transcription, correlating with a reduction in H3K9me3 and the recruitment of the KDM4B histone demethylase in these sequences. The depletion of these demethylases compromises the transformed cells. These results indicate that after the induction of a transformed phenotype, there is a reduction in heterochromatin that correlates with changes in the expression of histone modifying enzymes, allowing an increase in the transcription of heterochromatic sequences.
The cell line MCF10A ER-Src was kindly donated by Dr. Kevin Struhl (Harvard Medical School). Treatment of such cells with tamoxifen (TAM) rapidly induces Src and morphological transformation is observed within 24-36 hours 16, 20. MCF-10A ER-Src cells were cultured as indicated in 16. Cells were treated with 1μM tamoxifen when 80% confluent for transformation. MDA-MB-231 and MCF7 breast cancer cell lines were grow in DMEM, 10% FBS and pen/step.
Nuclear extracts were prepared as described in 21. Cells were washed twice with PBS1X and then resuspended and incubated for 15' in cold lysis buffer A plus inhibitors, followed by adding 10% Triton X100, cytoplasmic lysates were cleared by centrifugation. The pellet was resuspended and incubated for 15' in Nuclear Lysis Buffer (NLB), and then a spin down at high speed was necessary for recovering the nuclear proteins 21. The proteins were quantitated and separated in SDS-polyacrylamide gels and transferred to nitrocellulose membranes, following overnight blocking with 10-20% non-fat milk at 4ºC. Membranes were incubated with the primary antibody at different dilutions at room temperature for 2 h and then incubated with a horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature and detected by Chemiluminescence (Thermoscientific). Antibodies used were: HP1α (ab77256-abcam), H3K9me3 (07-442-millipore), H3K9me2 (ab1220-abcam), H3K9ac (ab4441-abcam), H3K27me3 (ab6002-abcam), KDM4A (PA5-25057-thermo scientific), KDM4B (ab27531-abcam), KDM4C (PA5-23065-thermo-scientific), KDM6A/UTX (ABE409-millipore), EZH2 (MABE362-millipore), SUZ12 (sc-46264-Santa Cruz) H3 (ab1791-abcam), Actin (JLA20-C-DSHB), Tubulin (E7-C- DSHB).
ChIP assays were performed as described in ChIP assay kit (Upstate Biotech). For each condition 2 mg of soluble chromatin fragments of 250-500 bp in length were incubated with 50 μg of Dynabeads for 2-4h at 4º C in rotation. The dynabeads were discarded and the pre-cleared chromatin was incubated with 2 μg of anti-H3K9me3 or anti-KDM4B antibodies 22 and rabbit anti-IgG antibody for 16 h at 4 °C. To recover the DNA-protein complexes, 25 μg of Dynabeads were added (previously blocked). Reverse cross-linking was performed by incubation at 65°C for 12h, followed by 4ºC incubation for 10 min. Samples were treated with RNase A for 1 h at 37 °C proteinase K was added and incubated for 2 h at 42 °C. DNA fragments were purified and resuspended in water.
40-50% confluent cells were cultured on Nunc-Lab-TekII Chamber Slides. The immunofluorescences were performed according to the manufacturer's instructions (http://www.abcam.com/protocols/immunocytochemistry-immunofluorescence-protocol). Cells were incubated overnight with the corresponding antibodies. The chambers were washed six times with PBS1X and incubated with Alexa Fluor 488 or Alexa Fluor 568-conjugated secondary antibodies. Nuclei were stained with Hoechst (dilution 1:20000) for 10 min. Samples were examined with Confocal Olympus FV1000 multi-photonic microscope.
Total RNA was extracted with TRIZOL. cDNA synthesis was performed in a reaction mix containing 1 μg of total RNA, oligo-dT, random primer and M-MLV Reverse Transcriptase (Invitrogen). The DNA was used to perform PCRs reactions using the following primers (SATa, SATIII and mcBox) reported in 14.
siRNA-silencing was performed according to Dharmacon instructions. The cells were seeded in a 6 well plates (60% confluence) and incubated for 12h in starvation medium and then independently the siRNA and the DharmaFECT reagent 1 were diluted in serum free medium and incubated for 5 min at room temperature, after that time those were mixed and incubated once more for 25 min at room temperature and finally added to the cells. The siRNA was used in a 25 and 50nM final concentration and incubated at 37ºC in 5% CO2 for 48-72h, ready to be analysed by flow cytometry for apoptosis and viability assays.
The intracelular protein staining was performed according to the protocols previously described by 23. The Viability and apoptosis assays were performed according to the protocols previously described in Affymetrix cat.#65-0865 and Biolegend cat.#640918 respectively. Cells were harvested and stained using Anexin pacific blue to labelled apoptotic cells. The cells were stained with viability solution to label dead cells. Briefly, the cells were wash twice and resuspended in fixable viability dye for 30 min at 4ºC protected from light, then the dye was removed with flow staining buffer, later the cells were resuspend in Annexin V binding buffer containing Annexin V and incubate for 15 min at RT (25°C) in the dark; the cells were washed with binding buffer and then fixed with PFA 2% at 37ºC for 10min, finally the cells were washed newly with FACS juice and stored at 4oC until the analysis. The cells and the data were analysed with FACSCanto II flow cytometer, and FLOWJO software respectively.
The MCF10-Er-Src line can be induced to transform to a malignant phenotype through the activation of Src by incubation with tamoxifen 16. Using this model, we determined the changes that occur in heterochromatin after transformation. The first approximation was to analyze the effect on heterochromatic centers. Ethidium bromide staining of transformed and non-transformed cells shows that there is a reduction in the number of heterochromatic centers in cells induced with tamoxifen after 36-48 hrs (Fig. (Fig.1A1A and B). We observed between 2-5 heterochromatic centers in the non-transformed cells and a reduction to 1-2 centers in the transformed cells (Fig. (Fig.1B).1B). This result suggests that there is a rapid rearrangement and reduction in heterochromatin when the cancerous phenotype is induced in MCF10-Er-Src cells.
Since HP1α is located in constitutive heterochromatin, we decided to determine if there are changes in the levels or distribution of HP1α changes in the levels or distribution of in transformed cells. HP1α can be identified in heterochromatic centers in non-transformed cells; however, there is a change in the nuclear location of HP1α in the tamoxifen treated cells (Fig. (Fig.1C).1C). Some of the HP1α is located in the nuclear periphery, and some HP1α remains in the heterochromatic bodies (Fig. (Fig.1B).1B). These changes in HP1α distribution occur without affecting HP1α levels (Fig. (Fig.1C).1C). Thus, the generation of a transformed phenotype in MCF10-Er-Src cells correlates with board changes in heterochromatin organization, suggesting a reduction in constitutive heterochromatin.
Since the levels of heterochromatic centers are reduced and the nuclear distribution of HP1α changes after transformation, we next analyzed the histone marks H3K9me3, H3K9me2 and H3K27me3. H3K9me3 and H3K9me2 are recognized by HP1α, and H3K27me3 is a mark present in genes that are not expressed. Fig. Fig.2A2A shows immunostains of non-transformed and transformed cells. H3K9me3 and H3K27me3 can be detected forming foci in the nuclei of normal cells 24-26. In the case of H3K27me3, a predominant focus near the nuclear membrane can be visualized that corresponds to the inactive X chromosome. Intriguingly, a reduction in the levels of H3K9me3 and H3K27me3 is observed in transformed cells (Fig. (Fig.2A).2A). The H3K27me3 signal of the inactive X chromosome is also reduced. The levels of H3K9me2 are not reduced in transformed cells (Fig. (Fig.2A).2A). Conversely, significant changes in the levels of H3K9Ac are not detected (Fig. (Fig.2A).2A). The reduction in the levels of H3K9me3 and H3K27me3 was confirmed by Western blot analysis using nuclear protein from non-transformed and transformed cells (Fig. (Fig.2B).2B). These results suggest that the decrease in heterochromatin is linked to a reduction of both facultative and constitutive histone marks.
We have observed that a reduction in global heterochromatin in transformed cells correlates with a decrease in H3K9me3 and H3K27me3 levels. Therefore, we determined the effect of transformation on the histone methyl-transferases and demethylases that modulate these modifications. For H3K9me3, we determined the total levels of the stress-inducible KDM4B demethylase as well as the levels of the SUV39H1 methyl-transferase. In the transformed cells, there is an important increase in the levels of KDM4B compared with the non-transformed cells (Fig. (Fig.3A).3A). In addition, Western blot analysis confirmed the KDM4B immunostaining data but showed a significant reduction in the levels of other H3K9me3 demethylases (Fig. (Fig.3A).3A). Conversely, the levels of SUV39H1 decrease in the transformed cells (Fig. (Fig.3A).3A). Thus, the reduction in the H3K9me3 level correlates with an increase of KDM4B and with the decrease of one of the methyl-transferases that incorporates methyl groups into H3K9.
To investigate H3K27me3 demethylases and methyl-transferases, we analyzed the levels of UTX/KDM6A and KDM6B as well as the levels of ENZh2 and SUZ12, which are part of PRC2. Fig. Fig.3B3B shows that the levels of UTX/KDM6A are increased, but that the levels of KDM6B are reduced in the transformed cells. Interestingly, the levels of the PRC2 components decrease in the cancerous cell nuclei (Fig. (Fig.3C).3C). Collectively, these results indicate that in transformed cells, there are changes in the levels of enzymes that affect the methylation state of the K9 and K27 residues of H3. In particular, the data suggest that there is a reduction in the levels of SUVAR39H1 and an increase of KDM4B, which may explain the reduction in the total levels of H3K9me3 in the transformed cells. Conversely, a reduction in the levels of PRC2 and an increase in UTX/KDM6A may be linked to the decrease in the levels of H3K27me3. Other factors that participate in these modifications also undergo changes during transformation; however, these factors do not correlate with the global changes observed in H3K9me3 and H3K27me3.
A characteristic of pericentromeric heterochromatin is that it is enriched in H3K9me3. These DNA satellite sequences are only transcribed at low levels in normal cells, but an increase in their transcription has been observed in cancer cells (14). Since we observed a decrease of this mark in the transformed cells, we decided to investigate whether there is also an enhancement in the transcription of Sat a, SatIII and mcBox satellite sequences 27, 28. We observed an increase in the transcript levels of all these elements in the transformed cells (Fig. (Fig.4A).4A). These results suggest that the reduction in the level of H3K9me3 in heterochromatic centers increases the transcription rate of constitutive heterochromatin in transformed cells.
To further analyze this possibility, we performed ChIP experiments to test if there is a reduction in H3K9me3 levels in these satellite sequences as well as an increase in the occupancy of KDM4B that may be removing this repressive mark. Fig. Fig.4B4B shows that in non-transformed cells, the H3K9me3 mark is present in the three satellite sequences and, as expected, it is not present in the GADPH gene. Interestingly, in transformed cells, this H3 modification is not detected, at least no more than the mock antibody. Intriguingly, when we analyzed the non-transformed cells, the presence of KDM4B in the satellite sequences is not detected; however, KDM4B is recruited to these repetitive elements after transformation induction (Fig. (Fig.4B)4B) Collectively, these experiments suggest that there is a recruitment of KDM4B to the satellite sequences that demethylate H3K9, favoring the transcription of these repetitive elements in transformed cells.
The fact that the levels of UTX/KDM6A and KDM4B increase after transformation and that in the case of KDM4B this increase correlates with its occupancy in repetitive DNA sequences in which its transcription levels are enhanced opens the question of whether these demethylases are required for the maintenance of transformed cells. To resolve this question, we transfected transformed and non-transformed cells with dsRNA against the UTX/KDM6A and KDM4B transcripts. The levels of UTX/KDM6A and KDM4B as well as the viability of the cells were analyzed after 48 and 72 hrs. At these two time points, there is a clear reduction in the two demethylases in both non-transformed and transformed cells transfected with the dsRNAs (Fig. (Fig.5A5A and and5C).5C). Interestingly, the reduction of both enzymes significantly reduced the viability of the transformed cells compared with the non-transformed cells (Fig. (Fig.5B5B and and5D).5D). This reduction in viability correlates with an increase in apoptosis in the transformed cells (Fig. (Fig.5B5B and and5D)5D) and suggests that the transformed cells require high levels of UTX/KDM6A and KDM4B to survive.
It is known that the global configuration of chromatin is altered in cancer cells, which is observed as a preferential reduction in the levels of heterochromatin. For instance, recent analysis has found the loss of heterochromatin in large extensions of the genome in cancer cells 29, 30. In addition, mutations in several factors involved in heterochromatin maintenance have been related to the cancerous phenotype 28, 29. Interestingly, there are examples in which mutations that favor the formation or the decondensation of heterochromatin favor some types of tumors and suppress the cancerous phenotype in others 31-35, 36. However, many of the studies analyzing the changes in heterochromatin in cancer cells are not based on comparative studies between the cancerous cell and its progenitor. In this work, we used the MCF10-Er-Src cell line to compare the changes that occur in heterochromatin during the generation of a transformed phenotype. In fact we corroborate the advantage to perform comparative experiments between a malignant cell and its direct precursor when we compare the levels of KDM4B as well as the transcript levels of the Sat a, SatIII and mcBox satellite sequences in other breast cancer lines. In particular we analyzed the MCF7 cell line that is positive for the Estrogen Receptor (ER) and the triple negative MDA-MB-231 together with the non-cancerous cell line MFC10A. Intriguingly, when we analyzed the levels of KDM4B, we found that also in these lines the KDM4B protein levels are higher than in the MCF10A non-cancerous line (Fig. S1). However, when we determined the Sat a, SatIII and mcBox transcript levels, we found that only Sat a and SatIII in the MDA-MB-231 and in the transformed MCF10A-ER-SRc cell lines are increased, if compared with the MCF10A line (Fig. (Fig.1S).1S). From these experiments it is clear that comparative experiments between cell lines that have different origin are hard to be conclusive, since in the absence of the precursor cells for each cell line, the levels of KDM4B and of the satellite sequences before the transformation of these cells are unknown. However, our comparative results using the MCF10A-Er-Src line after and before transformation indicates that there are severe alterations in heterochromatin in the transformed cells, such as the reduction of heterochromatin centers and a partial delocation of HP1α.
It is known that HP1α binds chromatin, recognizing H3K9me2 or H3K9me3 1, 9. We observed that the levels of H3K9me3 were reduced after transformation, suggesting that the delocation of HP1α from the heterochromatin bodies may be due to the reduction in H3K9me3 levels. In addition, the reduction in H3K9me3 in the transformed cells also correlates with an increase of the H3K9me3 demethylase KDM4B. The levels of the KDM4B enzyme rise significantly 36-48 hrs after the induction of transformation in the MCF10a-Er-Src cells, showing a rapid response during transformation to a cancerous phenotype. Contrary to KDM4B, the levels of KDM4A and KDM4C, which also demethylate H3K9me3 and H39me36, are reduced in the cancerous cells (Fig. (Fig.3A).3A). These results show that these three demethylases are differentially regulated during acquisition of the transformed phenotype. This suggests that different factors may associate with these enzymes for performance of their functions and that the reduction in H3K9me3 levels may be controlled in part by KDM4B. However, we cannot disregard the possibility that the reduction in H3K9me3 levels may also be caused by an effect on the H3K9 methyl transferase SUV39H1, since the levels of this enzyme were also affected. Based on these results, we suggest that the combination of an increase of KDM4B and a reduction of SUV39H1 cause an important reduction in the global levels of tri-methylated H3K9 in transformed cells.
Repetitive DNA elements are organized as heterochromatin. Although these elements are transcribed, the loss of heterochromatin increases their transcription, generating genome instability 30, 35, 36. In addition, it has been reported that mutations in the BRCA1 gene linked to breast cancer induce heterochromatin decondensation in repetitive DNA, increasing its transcription and promoting the generation of cancer 14. It has also been reported that there is an expansion of SatII sequences mediated by an increase in transcription of this repetitive element in colon cancer cells 37. Intriguingly, we found that the expression of the Sat a, SatIII and mcBox satellite sequences is increased in the transformed cells and that this increase correlates with a decrease in the H3K9me3 mark in these sequences. Furthermore, the occupancy of KDM4B is increased in these elements after transformation, suggesting that KDM4B is recruited to these DNA sequences, demethylating H3K9me3 and favoring the transcription of these sequences. Based on these observations, it could be important in future experiments to determine if the increase in the presence of transcripts of these repetitive elements contribute to generation of the cancerous phenotype. Additionally, it will be important to determine what recruits KDM4B to these sequences in the transformed cells as well as the occupancy of SUV39H1 in these regions. These observations agree with recent reports that show that KDM4B may participate in the generation of cancer 38, 39.
We also observed a global reduction in the tri-methylation of H3K27. The decrease in this histone mark correlates with an enrichment of the UTX/KDM6A demethylase, although there is also a reduction of KDM6B. Intriguingly, the levels of EzH2 and Suz12, components of PRC2, also decrease in the transformed cells. As mentioned before, there is evidence that some tumors are linked to an increase in PRC2 activities but that others are associated with a reduction of PRC2 40. These results suggest that the global reduction in H3K27me3 occurs with an increase in the UTX/KDM6A demethylase and a decrease in the levels of PRC2 in the MCF10A-Er-Src transformed cells, which may be related to the expression of genes that are necessary for the induction and maintenance of the transformed phenotype.
The high levels of KDM4B and UTX/KDM6A in the transformed cells seem to be necessary for viability of the transformed MCF10A-Er-Src cells since the reduction of both enzymes is more deleterious in these cells than in the progenitor cells. These results suggest that the decondensation of heterochromatic regions is required for the survival of cancerous cells. However, at this point, we do not know if the relaxation of heterochromatin is maintained in mamospheres derived from these cells or in these cells when they are introduced into animals that allow the development of tumors 16. Although it is not possible to generalize that the relaxation of the heterochromatin is required for the acquisition of the cancerous phenotype in other cancer models, our results strongly suggest that dramatic changes in heterochromatin configuration occur soon after or during transformation and that these changes are important for the survival of cancer cells. In addition, these studies allow further exploration of the mechanisms that cause alterations to heterochromatin in the early stages of transformation, for instance to determine how KDM4B is recruited to heterochromatin. Furthermore, we have identified important candidates that participate in the heterochromatin dynamics during transformation that are suitable for global genomic analysis before and after the acquisition of a cancerous phenotype.
This work was supported by grants from CONACyT 219673, CONACyT 250588 and PAPIIT/UNAM IN200315. We thank Arturo Pimentel, Andres Saralegui, Chris Wood and LNMA for the advice in the use of the confocal microscopes.