Human cancer cell lines were cultured according to the instructions from ATCC and were maintained under a fully humidified atmosphere of 95% air and 5% CO₂ at 37°C. Cancer cells were grown to 70% confluence in plates, the cells were transfected with siRNA or GFP-tagged vectors using Lipofectamine 2000 reagent, according to manufacturer’s protocol (Invitrogen Life Technologies, Carlsbad, CA, USA). After incubation for 48 h, mRNAs and proteins were extracted from the cells. The sequences of siRNAs targeting HP1 genes were: HP1α, 5′-GUUCCAGUCCUCUCUCAAAGC-3′ and 5′-GCUUUG AGAGAGGACUGGAAC-3′; HP1β, 5′-GACUCCAGUGGA GAGCUCAUG-3′ and 5′-CAUGAGCUCUCCACUGGA GUC-3′; HP1γ, 5′-AUUCUUCAGGCUCUGCCUC-3′ and 5′-GAGGCAGAGCCUGAAGAAU-3′ (HP1α HP1β, and HP1γ primers were described previously) (23).

Immunoblotting was performed as previously described (24). Cell lysates were prepared using lysis buffer supplemented with protease and phosphatase inhibitors (Thermo Scientific Pierce, Rockford, IL, USA). Protein concentrations were quantified according to the BCA Protein Assay kit (Thermo Scientific Pierce, Rockford, IL, USA).

Total RNA was extracted using easy-Blue reagent (Intron Biotechnology, Seoul, Korea) and 1 μg of RNA with oligo dT primers was subjected to reverse transcription using the ImProm-II™ Reverse Transcription system (Promega Corporation, Madison, WI, USA). cDNA was amplified using Super Premix Sapphire PCR master mix (Mbiotech, Inc., Seoul, Korea). qRT-PCR was performed with the KAPA™ SYBR^® FAST qPCR (Kapa Biosystems, Inc., Wilmington, MA, USA) using CFX96™ or Chromo4™ real-time PCR Detector (Bio-Rad, Hercules, CA, USA). Relative levels of mRNA were normalized to the values of GAPDH mRNA for each reaction. Primer sequences used for PCR were: MMP2, 5′-ACCAGCTGGCCTAGTGATGATG-3′ and 5′-GGCTT CCGCATGGTCTCGATG-3′; MT1-MMP, 5′-GGAATAAC CAAGTGATGGATGG-3′ and 5′-TTGTTTCCACGGAAGA AGTAGG-3′; tissue inhibitor of metallopeptidase (TIMP)2, 5′-GCGGTCAGTGAGAAGGAAGTGG-3′ and 5′-CTTGCA CTCGCAGCCCATCTG-3′; GAPDH, 5′-GAGTCAACGGAT TTGGTCGT-3′ and 5′-TTGATTTTGGAGGGATCTCG-3′.

Human cancer cells grown in 6-well plates were scratched with a yellow pipette tip to form a thin wound. Microscope images were observed immediately and 48 h after the scratch.

To investigate the potential role of HP1 in human cancer, we first examined the expression levels of the three HP1 proteins in diverse human cancer cell lines, which originated from eight different organs. The cell lines we used are indicated in Table I. HP1α and HP1β were more highly expressed in MCF7, a non-invasive breast cancer cell line, than the two invasive breast cancer cell lines (Fig. 1A). HP1γ was uniquely repressed and histone H3 lysine 9 methylation (H3K9me3) was dramatically increased in only MDA-MB-231, invasive breast cancer cells. Colon cancer cells are sorted according to the Dukes grade. Among the five colon cancer cell lines, only SW-480 is classified as Dukes grade B, which has invasion through the bowel wall, but does not involve lymph nodes. The expression levels of HP1α, β and γ were all high in SW-480 (Fig. 1B). The other colon cancer cell lines, which are classified as Dukes grade C, involving invasion of the lymph nodes, expressed high levels of all three HP1 proteins. In IMR90, a normal lung cell line, the three HP1 proteins and H3K9me3 were all suppressed compared to the two lung cancer cell lines (Fig. 1C). On the other hand, the HP1 proteins and H3K9me3 were more highly expressed in 293T/17, normal kidney cell line, than ACHN, renal cell adenocarcinoma (Fig. 1D). These results suggest that HP1 proteins are related to tumor progression in lung, but tumor suppression in kidney. In the liver cancer cell lines, HP1α and HP1γ were more highly expressed in HepG2 while HP1β and H3K9me3 were more highly expressed in SK-HEP1 (Fig. 1E). In the ovarian cancer cell lines, the three HP1 proteins showed similar expression patterns, but H3K9me3 showed an opposite pattern (Fig. 1F). In leukemia and osteosarcoma cell lines, HP1γ and H3K9me3 were expressed in the same pattern (Fig. 1G and H). Taken together, these results show that the expression patterns of the HP1 proteins vary in different cancer types and degree of invasiveness (Table I).

Although to different extents, the three HP1 proteins were highly expressed in colon cancer cell lines. To determine whether HP1 and MMP2 are involved in colorectal cancer, we first analyzed the expression pattern of MMP2 and HP1 proteins in six colon cancer cell lines. MMP2, which degrades the basic component of cellular membrane, is regarded as a leading molecule in the metastatic process of cancer development (17,19). Inactive pro-MMP2 is 70 kDa and active-MMP2, derived from cleavage of pro-MMP2 is smaller (35–50 kDa). Among the members of HP1 family, HP1β showed a reverse expression pattern with active-MMP2 (Fig. 2A). Using ImageJ program, we quantified the expression level of active-MMP2 and HP1s displayed in Fig. 2A and assessed the correlation between the values. We confirmed a strong correlation between active-MMP2 and HP1β through the coefficient of determination (R²=0.5202) (Fig. 2B, middle). HP1α showed a weak correlation with active-MMP2, as R² is 0.1066 (Fig. 2B, left) and HP1γ had no correlation with active-MMP2 (Fig. 2B, right). These data suggest the possibility of a biological relationship between HP1β and MMP2.

To investigate whether HP1β has influence on the MMP2 protein level, we transfected GFP-empty vector or GFP-tagged HP1 vectors into colon cancer cell line, HCT-15, which showed a weak HP1β signal. Overexpression of HP1β in HCT-15 cells resulted in reduction of pro-MMP2 as well as active-MMP2 (Fig. 3A). In OVCAR3, an ovarian cancer cell line, overexpressing HP1β caused a reduction of pro-MMP2, but not active-MMP2 (Fig. 3B). Similarly, only the pro-MMP2 level in HeLa cells was decreased depending on the amounts of ectopically expressed HP1β (Fig. 3C). We cannot exclude the possibility that, because the high basal expressions of HP1β is enough to saturate the active-MMP2 level, exogenously expressed HP1β results in no significant change in the active-MMP2 level in OVCAR3 and HeLa cells (Fig. 3E and F). Therefore, we next examined the MMP2 level in HP1β-depleted cells. In HT-29, a colon cancer cell line, KD of HP1β promoted both pro- and active-MMP2 (Fig. 3D). Increase of pro-MMP2 and active-MMP2 levels following HP1β KD, was observed in OVCAR3 and HeLa cells (Fig. 3G and H). These effects were also observed in HP1α but to a lesser extent (Fig. 3A–D and G). Thus, these results suggest that HP1β is a major regulator of MMP2 expression and activation and HP1α is a minor regulator.

To examine whether HP1β regulates MMP2 expression at the transcription level, we measured the mRNA level of MMP2 in HP1β-deleted cells. The mRNA level of the MMP2 gene increased 1.5-fold, following KD of HP1β in HCT-15 cells (Fig. 4A). However, considering that HCT-15 cells express very low basal level of HP1β, this is regarded as a significant change. In OVCAR3 and HeLa cells, which express a high basal level of HP1β, the mRNA level of MMP2 increased >5-fold following KD of HP1β (Fig. 4B and C).

The activation of MMP2 is triggered by the cleavage of pro-MMP2 by MT1-MMP, which is influenced by the level of TIMP. TIMP2 binds to the active site of MT1-MMP and this complex acts as a receptor for pro-MMP2. Then, the free MT1-MMP cleaves the propeptide of pro-MMP2, leading to the intermediate stage. Through further autocatalytic proteolysis, fully active MMP2 is generated (Fig. 5A) (25). However, when the TIMP2 level is high enough to saturate all MT1-MMP, cleavage of pro-MMP2 is impossible preventing MMP2 activation (Fig. 5B). We found that active-MMP2 as well as pro-MMP2 was changed by HP1β (Figs. 3 and 4). Thus, we examined whether the mRNA levels of MT1-MMP and TIMP2 are regulated by HP1β. The mRNA levels of MMP2 and MT1-MMP decreased following overexpression of HP1β in HCT-15 and OVCAR3 cells, while the level of TIMP2 did not significantly change (Fig. 5C and D). Consistent with this result, KD of HP1β promotes the mRNA level of MMP2 and MT1-MMP in both cell lines (Fig. 5E and F).

Recent studies of breast cancer cells revealed an inverse correlation between HP1α expression and tumor cell invasiveness (13,14). We therefore investigated whether tumor cell motility is regulated by all three HP1 protein or in an HP1 isoform-specific manner. To do this, we assessed the migratory ability of HeLa cells transfected with GFP-tagged vectors or siRNA following scratch with a yellow tip. Overexpression of HP1β markedly decreased the migration rate of HeLa cells (Fig. 6A), while KD of HP1β promoted migration of HeLa cells (Fig. 6B). Interestingly, overexpression or KD of HP1α resulted in the similar, but lesser effects, on the migration rate compared to HP1β. In contrast, modulating the expression of HP1γ did not affect the migration. Thus, it suggests that HP1β and HP1α, but not HP1γ, are responsible for cancer metastasis.