Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Mar 23;36(12):1707-1720.
doi: 10.1038/onc.2016.337. Epub 2016 Oct 3.

Identification of a Cancer Stem Cell-Specific Function for the Histone Deacetylases, HDAC1 and HDAC7, in Breast and Ovarian Cancer

Affiliations
Free PMC article

Identification of a Cancer Stem Cell-Specific Function for the Histone Deacetylases, HDAC1 and HDAC7, in Breast and Ovarian Cancer

A E Witt et al. Oncogene. .
Free PMC article

Abstract

Tumours are comprised of a highly heterogeneous population of cells, of which only a small subset of stem-like cells possess the ability to regenerate tumours in vivo. These cancer stem cells (CSCs) represent a significant clinical challenge as they are resistant to conventional cancer therapies and play essential roles in metastasis and tumour relapse. Despite this realization and great interest in CSCs, it has been difficult to develop CSC-targeted treatments due to our limited understanding of CSC biology. Here, we present evidence that specific histone deacetylases (HDACs) play essential roles in the CSC phenotype. Utilizing a novel CSC model, we discovered that the HDACs, HDAC1 and HDAC7, are specifically over-expressed in CSCs when compared to non-stem-tumour-cells (nsTCs). Furthermore, we determine that HDAC1 and HDAC7 are necessary to maintain CSCs, and that over-expression of HDAC7 is sufficient to augment the CSC phenotype. We also demonstrate that clinically available HDAC inhibitors (HDACi) targeting HDAC1 and HDAC7 can be used to preferentially target CSCs. These results provide actionable insights that can be rapidly translated into CSC-specific therapies.

Conflict of interest statement

Tan A. Ince discloses pending patent intellectual proprietary interest as the inventor of BMI and OCMI media.

Figures

Figure 1
Figure 1
BPLER and HMLER cells differ in their in vivo and in vitro tumor initiating cell (TIC) frequency, and in CSC surface marker expression. (a) In serial dilution xenograft assays BPLER lines display 2–4 orders-of-magnitude (102–104) greater in vivo TIC frequency than paired HMLER cell lines. BPLER2/HMLER2, BPLER3/HMLER3, BPLER4/HMLER4 isogenic cell lines were derived from normal human breast primary cells isolated from three different donors as previously described (See Supplementary Figure 2a for further details). (b) BPLER lines (blue bars) demonstrate greater capacity for sphere formation than HMLER cells (red bars) in 3D sphere formation assays used as an in vitro measurement of CSC self-renewal. Data presented as a mean +/− s.d. of sphere counts from triplicate wells (P<0.05). Results are representative of at least three independent experiments. (c) BPLER lines express higher levels of the CSC markers CD326 (EpCAM/ESA), CD166 (ALCAM), and BMI-1* than HMLER. Additionally, BPLER lines express the CSC-specific CD44 isoform (CD44v-250 kDa), while HMLER express the standard CD44 isoform (CD44s-80 kDa). Western blot of whole-cell lysates. β-Actin represents loading control. *BMI-1-matched β-actin in Supplementary Fig ure 3f. (d) FACS-enriched BPLER CSCs, sorted for high expression of individual CSC markers (CD44, CD166, CD326), or a combination of all three markers, demonstrate enhanced mammosphere formation when compared to BPLER with low CSC-marker expression (black bars, CSC marker high; white bars, CSC marker low). Data presented as a mean +/− s.d. of sphere counts performed in triplicate (P<0.05). Results are representative of at least three independent experiments.
Figure 2
Figure 2
CSC-Like BPLER cells are associated with high HDAC1 and HDAC7 expression and sensitivity to pan-HDAC inhibitors. (a) The pan-HDAC inhibitor TSA preferentially inhibits BPLER proliferation (blue line), compared to HMLER (red line). The results are representative of at least three independent experiments, presented as a percentage of vehicle treated control, the error bars represent standard deviation of the mean (P<0.005). (b) Short-term (24 h) pretreatment with TSA (0.35 μM), preferentially inhibits BPLER sphere formation (3D) in drug-free medium with no effect on 2D proliferation in either BPLER or HMLER. In contrast, pretreatment with Taxol (50 nM) and 5-Fluorouracil (1.0 μM), preferentially inhibit 2D proliferation as compared to 3D sphere formation. The number of viable colonies from triplicate wells were determined by 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-phenyl-2H-tetrazolium chloride (INT) staining. Results are representative of at least three independent experiments presented as percentage of vehicle treated control (P<0.05). BPLER: 2D proliferation (white bars with blue outline) vs 3D sphere formation (blue bars). HMLER: 2D proliferation (white bars with red outline) vs 3D sphere formation (red bars). The error bars represent standard deviation of the mean. (c) BPLER cell lines express higher levels of HDAC1 and HDAC7 proteins compared to matched HMLER lines. Western blot of whole-cell lysates. β-Actin represents loading control. (d) Heatmap of the mRNA expression profile of HDAC1-11 does not reveal any consistent differences between BPLER and HMLER lines (red, increased expression; green, decreased expression). (e) Treatment of BPLER cells for 48 h with TSA (0.35 μM) leads to downregulation of HDAC1, HDAC7, CD44 and CD166 protein expression in BPLER cells. Western blot of whole-cell lysates. β-Actin represents loading control. (f) Double immunofluorescence staining of BPLER cells simultaneously with HDAC7 and CD44 antibodies demonstrate that HDAC7 and CD44 are co-expressed. DAPI (blue), HDAC7 (red) CD44 (green). Scale bar 25 μM.
Figure 3
Figure 3
CSCs isolated from standard breast and ovarian cancer cell lines express higher protein levels of HDAC1 and HDAC7, and have increased HDAC enzymatic activity. (a) The higher sphere forming capacity of CD44high/CD166high CSCs (black bars) compared to CD44low/CD166low nsTC cells (white bars) is observed in multiple breast (SUM159, MDA-231, and MCF7) and ovarian (SKOV3 and OV90) cell lines. The data are presented as the mean of sphere counts from triplicate wells, the error bars represent standard deviation of the mean (P<0.01). The results are representative of at least three independent experiments. (b) The higher HDAC enzyme activity is observed CD44high/CD166high CSCs (black bars) compared to CD44low/CD166low nsTCs (white bars) in multiple breast (SUM159, MDA-231, and MCF7) and ovarian (SKOV3 and OV90) and colon (HT29) cancer cell lines. Results are representative of at least three independent experiments. The error bars represent standard deviation of the mean (P<0.01). (c) HDAC1 and HDAC7 protein expression is significantly higher in CD44high/CD166high CSCs (+/+) from multiple breast (BPLER, MDA-MB-231, MCF7, SUM159) and ovarian (SKOV3) cancer cell lines compared to nsTC (−/−, CD44low/CD166low). Merged image of two Western blots run with whole-cell lysates, gel#1(BPLER, MDA-MB-231, MCF7, SUM159) and gel#2(SKOV3). β-Actin represents loading control. (d) HDAC7high (red) and CD44high (green) co-expression in ovarian cancer cell line SKOV3, demonstrated with double IF staining. Scale bar 25 μM. (e) Double IF staining identifies co-expression of CSC-markers (CD44, CD326, CD166, ALDH1, CD29) with HDAC7 in standard breast cell lines (MDA-MB-231, MCF7, SUM159), and primary human ovarian (OCI-P5x and OCI-E1p) and breast cancer cells (BCI-1009 and BCI-1133). (Y) represents consistent positive correlation of HDAC7 and CSC marker expression in surveyed lines. (y) represents positive, but inconsistent correlation. (N) represents lack of correlation. (n/a), not analyzed. (n/e) CSC marker not expressed in cell line. IF staining was repeated a minimum of two times for each line. See Supplementary Figure 10 for additional markers and cell lines.
Figure 4
Figure 4
Confirmation of CSC sensitivity to pan-HDAC inhibitors in other model systems. (a) The black lines within the red and green regions indicate different CMAP experiments that display significant upregulation (red) or downregulation (green) of the 154 BPLER lethality genes upon treatment with the indicated drugs. Black lines in the gray area indicate CMAP experiments with no significant variation. (b) The percent viability of BPLER and HMLER cells that were treated with TSA, Vorinostat, Loperamide and Triamterene at the indicated doses. A vehicle-treated control was used to estimate relative percent cell viability for each treatment. BPLER (blue line), HMLER cells (red line). The error bars represent standard deviation of the mean (P<0.005). (c) HDAC1, HDAC7 and CD166 proteins are expressed significantly higher in the CD44high/CD24low+ CSCs (CD24L) compared to CD44high/CD24Neg ns-TS (CD24N) MDA-MB-231 cells. Western blot of whole-cell lysates. β-Actin represents loading control. (d) The mRNA expression heatmap of CD44high/CD24low+ CSCs (CD24L) compared to CD44high/CD24Neg nsTC (CD24N) MDA-MB-231 cells, with high (red) and low (green) expression. The difference in expression for any HDAC family members between the two populations was less than 1.1 fold. (e) The sensitivity of CD24-negative (CD24N) vs CD24-low (CD24L) MDA-MB-231 cells to 60 epigenetic compounds was measured. The upward sloping curve marked with blue stars indicate 12 of drugs that preferentially inhibit the proliferation CD44high/CD24low+ MDA-MB-231 CSCs. including Trichostatin (a1), Apicidin (a6), Scriptaid (a8), Vorinistat (a12), M-344 (b8), Fluoro-SAHA (b12), Oxamflatin (c10), BML-281 (d6), Rocilinostat (e5), CUDC-907 (e6), CUDC-101 (e7). Dimethyl sulfoxide controls are shown in the last two rows. The red star identifies one compound (GSK126) that selectively killed CD24neg cells.
Figure 5
Figure 5
Knockdown of HDAC1 or HDAC7 alters the CSC phenotype in breast and ovarian cancer cell lines. (a) The knockdown of HDAC1 with shRNA reduces HDAC1, CD44 and CD166 protein levels in breast (MDA-MB-468/MCF7) and ovarian (SKOV3) cancer cell lines compared to a scramble shRNA control. Western blot of whole-cell lysates. β-Actin represents loading control. (b) HDAC1 knock-down with shRNA decreases 3D sphere formation (dark blue bars) more significantly than 2D proliferation (light blue bars) as compared a scramble shRNA-expressing control (white bars) in breast (MDA-MB-468, and MCF7) and ovarian (SKOV3) carcinoma cell lines. The data are presented as a percentage of the scramble shRNA controls. The error bars represent standard deviation of the mean from three replicates (*P<0.05)(**P<0.005). The results are representative of at least three independent experiments. Similar results were observed with additional HDAC1 shRNA constructs and in additional cell lines (Supplementary Figures 5b–d). (c) HDAC1 knockdown decreases tumour size in SUM159 breast cancer xenografts expressing two different stable HDAC1-shRNAs (shH1#1 and shH1#2) as compared to control scramble shRNA (shCntrl). Mean xenograft size measured from mice injected with 10 000 cells and plotted over time weeks) (P<0.05). (d) HDAC1 knockdown decreases tumour frequency in MDA-MB-231 xenografts expressing stable HDAC1-shRNA as compared to a scramble shRNA control line. TIC frequency calculated by limiting dilution analysis using http://bioinf.wehi.edu.au/software/elda/. TIC frequency of control cells (1.05 × 10−5), HDAC1 shRNA#1 (4.42 × 10−5) (P<0.005) and HDAC1 shRNA#2 (4.62 × 10−5) (P<0.005). (e) The knockdown of HDAC7 with shRNA reduces HDAC7, CD44 and CD166 protein levels compared to a scramble shRNA control in breast (MDA-MB-468, and MCF7) and ovarian (SKOV3) carcinoma cell line within 72 h western blot of whole-cell lysates. β-Actin represents loading control. (f) HDAC7 knockdown with shRNA decreases 3D sphere formation (dark green bars) more significantly than 2D proliferation (light green bars) as compared to a scramble shRNA-expressing control (white bars) in breast (MDA-MB-468, and MCF7) and ovarian (SKOV3) carcinoma cell lines. The results from one shRNA construct is shown, similar results were observed with additional HDAC7 shRNA constructs and in additional cell lines (Supplementary Figure 5). The cells are counted after trypan blue (2D) or INT staining (3D), and the results are presented as a percentage of the scramble shRNA control. The error bars represent standard deviation of the mean from three replicates (*P<0.05) (**P<0.01). The results are representative of at least three independent experiments.
Figure 6
Figure 6
Overexpression of HDAC7 alters the CSC phenotype in breast and ovarian cancer cell lines. (a) HDAC7 over-expression (H7) increases CD44 and CD166 protein expression in MCF7 and SUM159, and CD44v(*) in SUM159 and HCC1937, compared to control cells expressing empty vector (EV). Western blot of whole-cell lysates. β-Actin represents loading control. (b) HDAC7 overexpression increases 3D sphere formation (dark green bars) with minimal effect on 2D proliferation (light green bars), as compared to an EV-expressing control (white bars), in breast (MCF7/HCC1937) and ovarian (CaOV3) cell lines. 2D growth assays were counted after trypan blue staining to assess viable cells counts. 3D sphere assays were counted after INT staining. The data is presented as a percentage of the EV-expressing control. The error bars represent standard deviation of the mean from triplicates (*P<0.05) (**P<0.01). The results are representative of at least three independent experiments. See Supplementary Figure 16 for additional cell lines. (c) Image of a representative well corresponding to the counts of 3D spheres in panel b. These images illustrate that both the number and size of the spheres are increased in MCF7 and HCC1937 cells over-expressing HDAC7-GFP as compared to an EV-expressing control. (d) HDAC7 overexpression increases TIC frequency in MDA-MB-231 xenografts expressing HDAC7-CMV as compared to EV-control. TIC frequency calculated by limiting dilution analysis using http://bioinf.wehi.edu.au/software/elda/. TIC frequency of control cells (1.14 × 10−5) and HDAC7 over-expressing cells (H7-CMV) (5.56 × 10−4)(P<0.03).
Figure 7
Figure 7
Isoform-specific HDACis that inhibit HDAC1 and HDAC7 can be used to selectively target CSCs. (a) The HDAC class I specific drugs that also downregulate HDAC7 (MS275 and MGCD) significantly inhibit BPLER proliferation (blue bar), with minimal effect on HMLERs (red bar). The class II specific HDACi MC1568 and MC1575 also preferentially inhibit BPLER proliferation compared to matched HMLER lines. In contrast, the HDACi that do not target HDAC1 or HDAC7 (Droxinostat and Apicidin) preferentially inhibit HMLER. The cells were treated with MS275 (1 μM), MGCD0103 (1 μM), MC1568 (1 μM), MC1575 (1 μM), Apicidin (0.1 μM) and Droxinostat (5 μM). Similar results were observed with two additional matched BPLER/HMLER pairs, in at least three independent experiments. The viable cells were assessed with trypan blue staining and counted after four days of treatment. Inhibition in proliferation is represented as a percentage of vehicle treated control (white bar). The error bars represent the standard deviation of the mean of triplicate samples (P<0.005). (b) Isoform-specific HDACis that target HDAC1 and/or HDAC7 (MS-275 and MGCD0103) preferentially inhibit 3D sphere formation (black bars) compared to 2D proliferation (grey bars). The breast cancer (BPLER, SUM159, MCF7, ZR751) and ovarian cancer (SKOV3) cell lines were pre-treated for 24 h with MS-275 (1 μM), MGCD0103 (1 μM) or MC1568 (1 μM). The next day, 2D and 3D cultures were established in drug-free medium. In contrast, pre-treatment with Droxinostat (5 μM) has minimal effects on 2D or 3D cell proliferation. The number of viable cells for 2D growth was determined by trypan blue staining. The number of 3D spheres was determined by INT staining. Inhibition in proliferation is represented as a percentage of vehicle treated control (white bar). The error bars represent the standard deviation of the mean of triplicate samples (*P<0.05) (**P<0.01). (c) The level of HDAC7 protein expression is significantly downregulated by short-term (24 h) treatment with MS-275 or MGCD0103 at 1 μM (low) or 2 μM (high) doses, but not with MC1568 or Droxinostat at 1 μM (low) or 5 μM (high) doses in SUM159 cells. Similar results are observed with BPLER, MCF7 and MDA-MB-231 cells (data not shown). Western blots of whole-cell lysates. β-Actin (*) was used as a loading control. (d) MS-275 pretreatment preferentially inhibits 3D sphere formation (black bars) compared to 2D proliferation (grey bars) in a panel of primary human ovarian cancer cell lines, including; OCI-C5x (clear cell), OCI-P7a (papillary serous), OCI-P9a1 (papillary serous) and OCI-P5x (papillary serous). The lines were pre-treated for 24 h with 1 μM MS-275, and plated into 2D and 3D cultures the following day in drug-free medium. The number of viable cells for 2D growth was determined by trypan blue staining. The number of 3D spheres was determined by INT staining. Inhibition in proliferation is represented as a percentage of vehicle treated control (white bar). The error bars represent the standard deviation of the mean of triplicate samples (*P<0.05) (**P<0.01).
Figure 8
Figure 8
MS275 inhibits xenograft tumour growth. The OCI-P5X ovarian cancer cell line that express luciferase was treated either with vehicle (Ctrl) or MS275 at 0.5 μM for 24 h and injected into the right and left flanks of nu/nu Balb/C mice at serial dilutions (106, 105, 104 and 103 cells/site). (a) The tumour growth was monitored with IVIS imaging once a week by intraperitoneal injection of 150 mg/kg D-luciferin 10 min before scanning. The representative image shows tumour specific bioluminescence signal in mice at week 6 after OCI-P5X cell injection. The same scale bar was used between ctrl and MS275 groups at each dilution for comparison. (b) Quantification of bioluminescence intensities in MS275 and control (vehicle) groups 6 weeks after injection. Black dots: Ctrl group. Black circles: MS275. Black line in the middle of the cluster represents the average signal in each group. (c) The number of gross tumours that formed at each injection site in MS275 and control (vehicle) groups; 10 sites of injection (5 mice per group, 2 sites/mouse). (d) The number of tumours spheres that are formed by OCI-P5x explants that were treated with MS275 and control (vehicle). White bar gross tumours that formed at each injection site in MS275 (black bar) and control (white bar). The images show representative wells photographed with regular and fluorescent microscopy (scale bar=2000 μm).

Similar articles

See all similar articles

Cited by 30 articles

See all "Cited by" articles

References

    1. Ablett MP, Singh JK, Clarke RB. Stem cells in breast tumours: are they ready for the clinic? Eur J Cancer [Research Support, Non-US Gov't Review] 2012; 48: 2104–2116. - PubMed
    1. Vincent A, Van Seuningen I. On the epigenetic origin of cancer stem cells. Biochimica et biophysica acta 2012; 1826: 83–88. - PubMed
    1. Seto E, Yoshida M. Erasers of histone acetylation: the histone deacetylase enzymes. Cold Spring Harb Perspect Biol [Review] 2014; 6: a018713. - PMC - PubMed
    1. Kidder BL, Palmer S. HDAC1 regulates pluripotency and lineage specific transcriptional networks in embryonic and trophoblast stem cells. Nucleic Acids Res 2012; 40: 2925–2939. - PMC - PubMed
    1. Liang J, Wan M, Zhang Y, Gu P, Xin H, Jung SY et al. Nanog and Oct4 associate with unique transcriptional repression complexes in embryonic stem cells. Nat Cell Biol 2008; 10: 731–739. - PubMed

Publication types

MeSH terms

Feedback