Inflammatory factor receptor Toll‐like receptor 4 controls telomeres through heterochromatin protein 1 isoforms in liver cancer stem cell

Abstract Toll‐like receptor 4 (TLR4) which acts as a receptor for lipopolysaccharide (LPS) has been reported to be involved in carcinogenesis. However, the regulatory mechanism of it has not been elucidated. Herein, we demonstrate that TLR4 promotes the malignant growth of liver cancer stem cells. Mechanistically, TLR4 promotes the expression of histone‐lysine N‐methyltransferase (SUV39 h2) and increases the formation of trimethyl histone H3 lysine 9‐heterochromatin protein 1‐telomere repeat binding factor 2 (H3K9me3‐HP1‐TRF2) complex at the telomeric locus under mediation by long non coding RNA urothelial cancer‐associated 1 (CUDR). At the telomeric locus, this complex promotes binding of POT1, pPOT1, Exo1, pExo1, SNM1B and pSNM1B but prevents binding of CST/AAF to telomere, thus controlling telomere and maintaining telomere length. Furthermore, TLR4 enhances interaction between HP1α and DNA methyltransferase (DNMT3b), which limits RNA polymerase II deposition on the telomeric repeat‐containing RNA (TERRA) promoter region and its elongation, thus inhibiting transcription of TERRA. Ultimately, TLR4 enhances the telomerase activity by reducing the interplay between telomerase reverse transcriptase catalytic subunit (TERT) and TERRA. More importantly, our results reveal that tri‐complexes of HP1 isoforms (α, β and γ) are required for the oncogenic action of TLR4. This study elucidates a novel protection mechanism of TLR4 in liver cancer stem cells and suggests that TLR4 can be used as a novel therapeutic target for liver cancer.

decompaction and demethylation of the proximal TLR4 promoter. 3 In addition, TLR4 signalling via NANOG cooperates with Signal Transducers and Activators of Transcription 3 (STAT3) to promote formation of tumour-initiating stem-like cells in livers. 4 It also suggests that TLR4 drives breast cancer cell growth differentially depending on the presence of tumour suppressor P53. 5 Three members of the human heterochromatin protein 1 (HP1) family (HP1a, HP1b and HP1c) are involved in chromatin packing and epigenetic gene regulation. 6 Emerging evidence has shown that HP1a plays a unique biological role in breast cancer-related processes and particularly in epigenetic control mechanisms involved in aberrant cell proliferation and metastasis. 7 a, b and c proteins of HP1 family selectively bind to methylated lysine 9 of histone H3 via their chromo-domains. Also, HP1c recognition of lysine 9 in the histone H3 tail in different nucleosome structures plays a role in reading the histone code. 8 Notably, both HP1a and Argonaute 1 (AGO1) are involved in chromatin-related splicing regulation. 9 Moreover, HP1 regulates alternative splicing in a methylation-dependent manner by recruiting splicing factors to its methylated form. 10 Telomere length and telomerase activity normalize after several rounds of passaging, which is consistent with the ability of Pim-1 (PIM kinases, a family of Ser/Thr kinases) to transiently increase mitosis. 11 Telomere repeat binding factors 1 (TRF1) and 2 (TRF2) binding to telomeres are modulated by nucleosomal organization. 12 The stability of mammalian telomeres depends on TRF2, which prevents inappropriate repair and checkpoint activation. 13 Upon telomere shortening or telomere uncapping induced by loss of TRF2, telomeres elicit a DNA damage response leading to cellular senescence. 14 The human telomerase RNA component (hTR) activates the DNA-dependent protein kinase to phosphorylate heterogeneous nuclear ribonucleoprotein. 15 In particular, long non coding RNA urothelial cancer-associated 1 (lncRNA CUDR) promotes liver cancer stem cell growth through up-regulating telomerase reverse transcriptase catalytic subunit (TERT) and C-Myc. 16 Long non coding telomeric repeat-containing RNA (TERRA) is involved in telomere maintenance in a telomerase-dependent and a telomerase-independent manner during replicative senescence and cancer. 17 TERRA participates in the regulation of telomere length, telomerase activity and heterochromatinization. 18 Some research shows that telomeres are protected from hyper-resection through the repression of the Ataxia-telangiectasia-mutated (ATM) and ATR kinases by TRF2 and tripeptidyl peptidase 1-bound telomeric DNA binding proteins 1a/b (TPP1-bound POT1a/b), respectively. 19 Moreover, Exo1 extensively resects both telomere ends, generating transient long 3 0 overhangs in S phase/G2 phase checkpoint. CST/AAF, a DNA pola primase accessory factor, binds POT1b and shortens the extended overhangs produced by Exo1, likely through fill-in synthesis. 20,21 Furthermore, genetic variants in telomere maintenance genes are associated with genomic instability, cancer risk and cancer metastasis. [22][23][24] In this study, we attempted to elucidate TLR4 functions during the malignant growth of liver cancer stem cells. Specifically, we investigated whether TLR4 promotes the malignant proliferation and growth of liver cancer stem cells in vitro and in vivo, and investigated its potential role in the malignant transformation of liver stem cells by analysing the cascade of TLR4-HP1 (a, b and c)-telomere signalling.

| Co-immunoprecipitation (IP)
Co-immunoprecipitation was carried out according to methodology as previously described. 25 Briefly, cell lysates were incubated with 2 lg antibody or normal mouse/rabbit IgG under rotation for 4 hours at 4°C. Then, the immunoprecipitates were incubated with 30 lL protein G/A-plus agarose beads under rotation overnight at 4°C. The precipitates were washed with beads solution for five times, and then, the precipitates were resuspended in 60 lL 2 9 SDS-PAGE sample loading buffer. Western blotting was then performed.

| Chromatin immunoprecipitation (ChIP) assay
Chromatin immunoprecipitation was carried out according to methodology as previously described. 26 Briefly, cells were crosslinked with 1% (v/v) formaldehyde (Sigma-Aldrich) for 10 minutes at ZHENG ET AL.
| 3247 room temperature. Chromatin extracts were immunoprecipitated with specific antibody on protein-A/G-sepharose beads. After washing and de-cross-linking, the ChIP DNA was detected by PCR.

| Telomere length assay
Telomere length assay using Telo TAGGG PCR ELISA plus kit was performed according to manufacturer's instructions (Roche). A standard curve was established by dilution of known quantities of a synthesized 84-mer oligonucleotide containing only TTAGGG repeats.

| Methylation analysis
Methylated DNA Immunoprecipitation (MeDIP)-Dot blot-Western blotting was performed with anti-5-methylcytosine (5-mC) and methylation analysis by MspI plus BamHI digestion.   Figure 1Aa, CD133, CD44, CD24 and EpCAM were expressed in hLCSCs. However, these were not expressed in non-hLCSCs. We detected the expression of TLR4, MD2 and CD14 in non-transfected hLCSCs and non-hLCSCs by Western blotting. The results showed that the expression of TLR4 in hLCSCs was significantly higher than that in non-hLCSCs, and the expression of MD2 or CD14 in hLCSCs was lower than that in non-hLCSCs ( Figure S1).

| Xenograft transplantation in vivo
Then, we established four stable hLCSC lines transfected with pCMV6-AC-GFP (GFP ctrl group), pCMV6-AC-GFP-TLR4 (TLR4 group), pGFP-V-RS (RNAi ctrl group) and pGFP-V-RS-TLR4 (TLR4i group), respectively. As shown in Figure 1Ab, compared with GFP ctrl group, TLR4 expression was significantly enhanced in TLR4 group. However, TLR4 expression was significantly reduced in TLR4i group compared with RNAi ctrl group. Furthermore, the expression of TLR4 on cell surface was significantly increased in TLR4 overexpressing hLCSC and was significantly decreased in TLR4 knockeddown hLCSC compared to control ( Figure S2). Figure 1B, compared with growth of hLCSCs in GFP ctrl group (P < .01), that in TLR4 group was significantly increased; however, compared with growth of hLCSCs in RNAi ctrl group, that in TLR4i group was significantly decreased. In addition, F I G U R E 1 TLR4 accelerates hLCSCs growth in vitro. A, (a) Western blotting analysis of expression of CD133, CD44, CD24 and EpCAM in hLCSCs and non-hLCSCs. (b) Western blotting analysis of TLR4 expression in four hLCSC lines. b-Actin was used as internal control. B, Cell growth assay using CCK8. C, Soft-agar colony formation assay. D, S phase cells assay using BrdU. E, Cell sphere formation ability. Each value was presented as mean AE standard error of the mean (SEM). mean AE SEM. **P < .01; *P < .05. For all Western blotting, we repeated the experiments for three times. We measured grey value of the bands for quantification. Each value was presented as mean AE standard error of the mean (SEM) (Student's t test) compared with soft-agar colony formation of hLCSCs in GFP ctrl group (P < .01), that in TLR4 group was significantly increased; however, compared with soft-agar colony formation of hLCSCs in RNAi ctrl group (P < .01), that in TLR4i group was significantly decreased ( Figure 1C). Furthermore, compared with the proportion of BrdUpositive cells in GFP ctrl group, that in TLR4 group was significantly increased; however, compared with the proportion of BrdU-positive cells in RNAi ctrl group (P < .01), that in TLR4i group was significantly decreased ( Figure 1D). Strikingly, sphere formation rate of hLCSCs was significantly higher in TLR4 group than in the GFP ctrl group (P < .01), whereas that was lower in TLR4i group than in the RNAi ctrl group (P < .01) ( Figure 1E). Moreover, our results showed that excessive TLR4 significantly increased the interaction between TLR4 and MD2 (a TLR4 ligand) or TLR4 and MyD88 (a TLR4 dimer ligand) ( Figure S3A). Excessive TLR4 significantly promoted the colony formation ability of liver cancer stem cell ( Figure S3B). These results suggest that TLR4 promotes the proliferation of liver cancer stem cells in vitro.

| TLR4 accelerates growth of hLCSCs in vivo
To further explore the effect of TLR4 on hLCSCs in vivo, the four stable hLCSCs lines were injected subcutaneously into athymic BALB/c mice, respectively. As shown in Figure 2A, | 3249 appeared later in TLR4i group than in RNAi ctrl group (P < .01) (Figure 2C). Furthermore, xenograft tumour differentiation was poorer in TLR4 group than in GFP ctrl group, whereas that was better in TLR4i group than in RNAi ctrl group ( Figure 2D, upper pictures).
Strikingly, the percentage of proliferating cell nuclear antigen (PCNA)-positive cells from xenograft tumours was significantly higher in TLR4 group than in GFP ctrl group (P < .01), whereas that was significantly lower in TLR4i group than in RNAi ctrl group (P < .01) ( Figure 2D, lower pictures; Figure 2E). These results demonstrate that TLR4 accelerates malignant growth of liver cancer stem cells in vivo.

| TLR4 enhances the interplay between HP1 isoforms and H3K9me3
To investigate the possible mechanism of action of TLR4, we first studied whether TLR4 influenced histone H3 modification in hLCSCs. We analysed the NF-kB responsive element (5 0 -AGTT GAGGGGACTTTCCCAGGC-3 0 ) in all the promoters investigated in this study and only found that two NF-kB-responsive elements is in CUDR promoter region. As shown in Figure 3A, TLR4 overexpression enhanced the binding of NF-jB to the non coding RNA CUDR promoter region. As shown in Figure S4, the luciferase activity of F I G U R E 3 TLR4 increases interplay between HP1 isoforms and H3K9me3 via CUDR. A, Chromatin Immunoprecipitation (ChIP) with anti-NF-jB followed by PCR with CUDR promoter primers. IgG ChIP served as negative control. B, RT-PCR analysis of CUDR mRNA. b-Actin served as internal control. C, RNA Immunoprecipitation (RIP) with anti-SUV39 h1 followed by RT-PCR with CUDR promoter primers. IgG RIP served as the negative control. D, Co-immunoprecipitation (Co-IP) with anti-SUV39 h2 followed by Western blotting with antihistone. IgG IP served as the negative control. Western blotting with anti-SUV39 h2 served as INPUT. E, Co-IP with anti-SUV39 h2 followed by Western blotting with antihistone H3. IgG IP served as negative control. Western blotting with anti-SUV39 h2 served as INPUT. F, Western blotting with anti-HP1a, anti-HP1b, anti-HP1c, anti-H3K9me3, anti-SUV39 h2. b-Actin served as an internal control. G, Western blotting with anti-H3K9me3 (four hLCSC lines with CUDR being depleted). b-Actin was the internal control. H, Co-IP with anti-H3K9me3 followed by Western blotting with anti-HP1a, anti-HP1b and anti-HP1c. IgG IP served as negative control. Western blotting with anti-HP1a, anti-HP1b, anti-HP1c served as INPUT. I, Co-IP with anti-H3K9me3 followed by Western blotting with anti-HP1a. IgG IP was used as negative control. Western blotting with anti-HP1a as INPUT. Each value was presented as mean AE standard error of the mean (SEM). Mean AE SEM. **P < .01; *P < .05. For all Western blotting, we repeated the experiments for three times. We measured grey value of the bands for quantification. Each value was presented as mean AE standard error of the mean (SEM) (Student's t test) CUDR promoter in hLCSCs was higher in TLR4 group than in GFP ctrl group, whereas that was lower in TLR4i group than in RNAi ctrl group. As shown in Figure 3B, compared with GFP ctrl group, TLR4 group enhanced CUDR expression in hLCSCs; however, compared with RNAi ctrl group, TLR4i group decreased that. As shown in Figure 3C,D, compared with GFP ctrl group, TLR4 group enhanced   interplay between histone-lysine N-methyltransferase (SUV39 h2) and CUDR and interplay between SUV39 h2 and histone H3; however, compared with RNAi ctrl group, TLR4i group decreased those.
CUDR knockdown can fully abrogate the action of excessive TLR4 involved in enhancing interplay between SUV39 h2 and histone H3 ( Figure 3E). In addition, as shown in Figure 3F, compared with GFP ctrl group, TLR4 group increased formation of trimethyl histone H3 lysine 9 (H3K9me3); however, compared with RNAi ctrl group, TLR4i group decreased that. It should be noticed that neither excessive TLR4 nor TLR4 knockdown can alter the expression of HP1a, HP1b, HP1c, SUV39 h2 in the hLCSCs; especially, CUDR depletion drastically abolished the action of excessive TLR4 involved in increasing formation of H3K9me3 in hLCSCs ( Figure 3G). Compared with GFP ctrl group, TLR4 group increased the interplay between HP1 (HP1a, HP1b and HP1c) and H3K9me3; however, compared with RNAi ctrl group, TLR4i group decreased that ( Figure 3H). Furthermore, as shown in Figure 3I, CUDR depletion can drastically abrogate the TLR4 action involved in increasing interplay between HP1a and H3K9me3. These results suggest TLR4 not only enhances formation of H3K9me3, but also enhances the interplay between HP1 (HP1a, HP1b and HP1c) and H3K9me3 depending on long non coding RNA CUDR. 3.4 | TLR4 controls telomere length through

H3K9me3
Given that TLR4 increases formation of H3k9me3, we studied whether TLR4 can alter telomere length via H3K9me3. As shown in Figure 4A, compared with GFP ctrl group, TLR4 group increased interplays between the telomere DNA probe and HP1a, HP1b, HP1c, TRF2, H3K9me3; however, compared with RNAi ctrl group, TLR4i group decreased those. As shown in Figure 4B, compared with GFP ctrl group, TLR4 group increased loadings of HP1a, HP1b, HP1c, TRF2 and H3K9me3 onto the telomere DNA; however, compared with RNAi ctrl group, TLR4i group decreased those. As shown in Figure 4C, compared with GFP ctrl group, TLR4 group increased interplays between the telomere DNA probe and pTOP1, TOP1, pExo1, Exo1, pSNM1b, NM1b, but decreased interplay between CST/AAF and the telomere DNA probe; however, compared with RNAi ctrl group, TLR4i group decreased interplays between the telomere DNA probe and pTOP1, TOP1, pExo1, Exo1, pSNM1b, SNM1b, but increased interplay between CST/AAF and the telomere DNA probe. As shown in Figure 4D,E, compared with GFP ctrl group, TLR4 group increased telomere length; however, compared with RNAi ctrl group, TLR4i group decreased telomere length. It should be noticed that when H3K9me3 was demethylated by pCMV6-AC-GFP-JMJD2A (a demethylase that demethylates trimethyl histone H3 lysine 9) (Figure 4Fa), the action of TLR4 to alter telomere length was fully abrogated (Figure 4Fb). These results suggest that TLR4 increases telomere length depending on H3K9me3.

HP1a-DNMT3b pathway
To study whether activation of HP1a-H3K9me3 pathway helps TLR4 alter activity of telomerase involved in DNA methyltransferase (DNMT3b), we first analysed the interrelation between HP1a and DNMT3b in hLCSCs. As shown in Figure 5A, there was an interplay between HP1a and DNMT3b in hLCSCs.
Moreover, the formation of HP1a-DNMT3b complex reduced the interplay between DNMT3b and telomere DNA (lncRNA TERRA promoter) in hLCSCs ( Figure 5B). In addition, HP1a inhibited the DNMT3b activity and reduced the methylation on lncRNA TERRA promoter region ( Figure 5C,D). However, these actions were fully abrogated when TLR4 was knocked down ( Figure 5C). Furthermore, the excessive DNMT3b increased the methylation on lncRNA TERRA promoter region. However, these actions were fully abrogated when TLR4 was knocked down ( Figure S5). Moreover, TLR4 knockdown increases the interaction between HP1a and DNMT3b ( Figure S6). It suggests that TLR4 knockdown inhibits the activity of DNMT3b through increasing the interaction between HP1a and DNMT3b. Thus, TLR4 knockdown seems to block the activity of DNMT3b by enhancing the interplay between HP1a and DNMT3b.
Given that HP1a inhibited DNMT3b activity and reduced methylation on TERRA promoter region, which is associated with TLR4, we studied whether TLR4 could alter telomerase activity. As shown in Fig We measured grey value of the bands for quantification. Each value was presented as mean AE standard error of the mean (SEM) (Student's t test) knockdown can alter the expression of TERC ( Figure 6B). Moreover, compared with GFP ctrl group, TLR4 group increased interplay between TERT and TERC, but decreased interplay between TERT and TERRA; however, compared with RNAi ctrl group, TLR4i group decreased interplay between TERT and TERC, but increased interplay between TERT and TERRA ( Figure 6C). Finally, compared with GFP ctrl group, TLR4 group increased telomerase activity in hLCSCs; however, compared with RNAi ctrl group, TLR4i group decreased that (Figure 6D). These results suggest that TLR4 increases telomerase activity via HP1a-DNMT3b pathway.  HP1(a, b, c), and TLR4i group transfected with pcDNA-(HP1a, HP1b and HP1c). Each value was presented as mean AE standard error of the mean (SEM). Mean AE SEM. **P < .01; *P < .05. For all Western blotting, we repeated the experiments for three times. We measured grey value of the bands for quantification. Each value was presented as mean AE standard error of the mean (SEM) (Student's t test) 3.6 | HP1 isoforms (HP1a, HP1b and HP1c) are required for TLR4 oncogenic action Given that TLR4 enhances the interplay between HP1 isoforms (HP1a, HP1b and HP1c), increases the telomere length by HP1-H3K9me3 and increases telomerase activity by HP1-DNMT3b pathway, we studied whether HP1 (HP1a, HP1b and HP1c) could determine the TLR4 oncogenic function. To analyse the formation of tricomplex of HP1a-HP1b-HP1c, we performed the repeat co-immunoprecipitation (IP) experiments, that is, first, HP1a IP and second, HP1b repeat IP with the immunoprecipitates from HP1a IP.
As shown in Figure 7A,B, TLR4 overexpression increased interaction among HP1a, HP1b and HP1c in hLCSCs, whereas TLR4 knockdown decreased that. As shown in Figure S7, we show the specificity of each knockdown reagent (HP1a, HP1b and HP1c).
The expression of HP1a was significantly reduced only in HP1a knockdown group, the expression of HP1b was significantly reduced only in HP1b knockdown group, and the expression of HP1c was significantly reduced only in HP1c knockdown group.
Strikingly, TLR4 overexpression increased loadings of TRF2, POT1, Exo1 and SNM1b on telomere DNA in hLCSCs, whereas TLR4 knockdown decreased those ( Figure 7C). TLR4 overexpression increased loading of H3K9me3 and HP1a on telomere DNA, loading of DNMT3b on TERRA promoter region and interaction between TERT and TERC in hLCSCs, whereas TLR4 knockdown decreased those ( Figure 7D). Furthermore, TLR4 overexpression led to increase in telomerase activity and telomere length; however, these actions were fully abrogated by depletion of HP1a, HP1b, HP1c, HP1(a, b, c) ( Figure 7E and F). Moreover, overexpression of HP1a, HP1b, HP1c did not significantly alter the telomerase activity and telomere length in hLCSCs whose TLR4 was depleted  Finally, the function of TLR4 in liver cancer stem cells should be further explored. We postulate that TLR4 functions may be independent of NF-jB, which is a key transcriptional regulator involved in inflammation, cell proliferation, survival and transformation. In this respect, outstanding questions include the following: (i) What is the mechanism of oncogenic action of TLR4? (ii) How does TLR4 cooperate with HP1? and (iii) Does TLR4 regulate a series of molecular events during the malignant growth of liver cancer stem cells?
Answering these questions will help understand the mechanism underlying the malignant differentiation of liver stem cells. In summary, our data indicate that TLR4 promotes liver cancer stem cells malignant progression by altering telomere length. These results provide insight into a novel link between TLR4 and hepatocarcinogenesis and also have diagnostic and prognostic implications.

ACKNOWLEDG EMENTS
This study was supported by grants from National Natural Science

CONF LICT OF I NTEREST
The authors declare no conflict of interests.