Mode of SUV420H2 heterochromatin localization through multiple HP1 binding motifs in the heterochromatic targeting module

Abstract

Constitutive heterochromatin is transcriptionally repressed and densely packed chromatin, typically harboring histone H3 Lys9 trimethylation (H3K9me3) and heterochromatin protein 1 (HP1). SUV420H2, a histone H4 Lys20 methyltransferase, is recruited to heterochromatin by binding to HP1 through its Heterochromatic Targeting Module (HTM). Here, we have identified three HP1 binding motifs within the HTM. Both the full‐length HTM and its N‐terminal region (HTM‐N), which contains the first and second motifs, stabilized HP1 on heterochromatin. The intervening region between the first and second HP1 binding motifs in HTM‐N was also crucial for HP1 binding. In contrast, the C‐terminal region of HTM (HTM‐C), containing the third motif, destabilized HP1 on chromatin. An HTM V374D mutant, featuring a Val374 to Asp substitution in the second HP1 binding motif, localizes to heterochromatin without affecting HP1 stability. These data suggest that the second HP1 binding motif in the SUV420H2 HTM is critical for locking HP1 on H3K9me3‐enriched heterochromatin. HTM V374D, tagged with a fluorescent protein, can serve as a live‐cell probe to visualize HP1‐bound heterochromatin.

SUV420H2 heterochromatic targeting module (HTM) contains three HP1 binding motifs and their cooperative binding to HP1 stabilizes, or locks, HP1 on heterochromatin. An HTM mutant that does not affect HP1 dynamics can be useful for visualizing HP1‐bound heterochromatin in living cells.

1. INTRODUCTION

Heterochromatin is a form of densely packed chromatin in which transcription is typically suppressed. Constitutive heterochromatin harboring histone H3 Lys9 trimethylation (H3K9me3) is often formed on DNA repeat elements. Heterochromatin protein 1 (HP1) binds to both H3K9me2 and H3K9me3 through its chromodomain (CD) (Bannister et al., 2001; Lachner et al., 2001; Nielsen et al., 2002). In mammalian cells, there are three subtypes of HP1, namely HP1α, HP1β, and HP1γ (Jones et al., 2000). HP1α and HP1β are predominantly found in pericentromeric heterochromatin, also known as chromocenters, in mouse cells, whereas HP1γ is generally localized in euchromatin (Minc et al., 2000; Nielsen et al., 2001). The dimerization of HP1 through its chromoshadow domain (CSD) is implicated in chromatin compaction, as it facilitates the bridging of two nucleosomes within H3K9me3‐rich chromatin fibers (Hiragami‐Hamada et al., 2016; Machida et al., 2018) and provides a binding platform for effector proteins that contain PxVxL motif (Nozawa et al., 2010; Smothers & Henikoff, 2000; Yan et al., 2018). Some HP1‐binding proteins, such as Sentrin/SUMO‐specific protease SENP7 and a subunit of the Origin Recognition Complex (ORC), harbor multiple PxVxL motifs (Maison et al., 2012; Prasanth et al., 2004; Prasanth et al., 2010) and stabilize HP1 on heterochromatin through multivalent interactions, a process referred to as “HP1 locking” (Romeo et al., 2015). Furthermore, HP1 is also thought to contribute to heterochromatin compaction through phase separation (Larson et al., 2017; Qin et al., 2021; Sanulli et al., 2019; Strom et al., 2017).

In addition to H3K9me3, H4K20me3 is also enriched in constitutive heterochromatin, in a manner dependent on H3K9me3 and HP1 (Hahn et al., 2013; Schotta et al., 2004). Four histone methyltransferases—SUV420H1, SUV420H2, SMYD5, and SMYD3—have been reported to mediate the H4K20me3 modification (Foreman et al., 2011; Kidder et al., 2017; Schotta et al., 2004; Stender et al., 2012). Among these, SUV420H2 is thought to be the primary enzyme responsible for H4K20me3, as SUV420H2 knock‐out (KO) results in an almost complete loss of this modification (Schotta et al., 2008). Although Suv4‐20h2 KO mice develop normally (Schotta et al., 2008), cells lacking SUV420H2 show reduced heterochromatin compaction, chromocenter fragmentation, and abnormalities in mitotic spindle separation during mitosis (Hahn et al., 2013). The minimal effect of Suv4‐20h2 KO in development could be due to its low expression during the embryonic development stage and/or a potential compensatory effect by Suv4‐20h1, which possesses a SET domain highly similar to that of SUV420H2 (Schotta et al., 2008). In vitro studies have indeed demonstrated that the SET domain of SUV420H1 is capable of mediating both H4K20me2 and me3 modifications (Schotta et al., 2004). While the loss of SUV420H2 does not significantly affect mouse development, ectopic expression of SUV420H2 in pre‐implantation embryos disrupts embryogenesis, indicating defects in the S‐phase progression (Eid et al., 2016). Artificial enrichment of H4K20me3 at centromeres restricts the localization and function of AuroraB, leading to an increase in errors during chromosome separation in mitosis (Herlihy et al., 2021). Hence, controlled H4K20me3 localization appears to be important in regulating the genome function.

The mechanism by which H4K20me3 and SUV420H2 localize to pericentromeric heterochromatin, or chromocenters, in mouse cells has been investigated. SUV420H2 contains a region known as the Heterochromatic Targeting Module (HTM) at its C‐terminal region, which binds to HP1 on H3K9me3‐enriched chromatin (Hahn et al., 2013; Schotta et al., 2004). Interestingly, a photobleaching assay indicated that HTM tagged with the green fluorescent protein (GFP) binds to chromatin more stable than HP1‐GFP (Hahn et al., 2013; Souza et al., 2009). Such different behavior can be attributed to the multiple HP1 binding modules in the HTM (Souza et al., 2009). When HTM is divided into two fragments, each independently binds to HP1 and accumulates at chromocenters (Hahn et al., 2013). However, the detailed mechanism of how the SUV420H2 HTM binds to HP1 remains elusive.

In this study, we analyzed the localization and HP1 binding of various deletion and amino acid substitution mutants of the SUV420H2 HTM. We identified the three HP1 binding motifs whose spatial configuration was crucial for efficient heterochromatin targeting and HP1 interaction when overexpressed. One of the HTM mutants which harbors two independent HP1 binding motifs could serve as a live‐cell probe for detecting HP1 on heterochromatin.

2. RESULTS

2.1. SUV420H2 harbors three HP1‐binding motifs within its heterochromatic targeting module (HTM)

Previous studies have indicated that two regions within the HTM of SUV420H2 bind to HP1 (Hahn et al., 2013), yet the specific binding motifs have not been defined. To identify the HP1 binding regions within the HTM, we constructed various truncated mutants and transiently expressed them in mouse A9 cells (Figure S1). The full HTM tagged with superfolder GFP (HTM‐sfGFP) localized to pericentromeric heterochromatin, or chromocenters, highlighted by H3K9me3 and Hoechst 33342 staining (Figure S1A,B), as reported previously (Hahn et al., 2013). In cells expressing HaloTag‐tagged KDM4D, an H3K9me2/3 demethylase (Hayashi‐Takanaka et al., 2020), HTM‐sfGFP was diffused throughout the nucleoplasm (Figure S1C), suggesting that the heterochromatin targeting of HTM is dependent on H3K9me2/3 to which HP1 binds. Both the N‐ and C‐terminal regions of HTM, amino acid (aa) 347–380 and 381–435, respectively, could target heterochromatin (Hahn et al., 2013), albeit more weakly than the full HTM (Figure S1D). We thus constructed deletion mutants to determine the minimal essential element required for chromocenter enrichment in each N‐ or C‐terminal region (Figure S1D). Within the HTM N‐terminal region, aa 352–380 exhibited chromocenter localization (heterochromatin enrichment ratio 2.59 vs. 2.67 for aa 347–380), whereas a further N‐terminal deletion (aa 356–380) distributed homogenously in the nucleoplasm, and a further C‐terminal deletion (aa 352–374) was poorly enriched at chromocenters (heterochromatin enrichment ratio 1.65) (Figure S1D). Regarding to the C‐terminal region, aa 389–410 still localized to chromocenters (heterochromatin enrichment ratio 1.73 vs. 1.87 for aa 381–435). A further C‐terminal deletion mutant (aa 381–404) distributed homogenously throughout the nucleoplasm, and a further N‐terminal deletion mutant (391–410) showed reduced accumulation at chromocenters (heterochromatin enrichment ratio 1.39). Since aa 352–380 and 389–410 retained the similar chromocenters targeting efficiency to their respective original fragments, we selected these for further point mutation studies and designated them as the HTM‐N and HTM‐C, respectively.

A previous study has indicated that the full HTM (aa 347–435) binds to HP1 proteins through CSD (Souza et al., 2009), but the exact amino acids that are required for HP1 binding were not defined. To identify HP1 binding motifs within the HTM‐N and HTM‐C regions, we aligned amino acid sequences from different species, as essential amino acids are often evolutionally covered (Figure S2). In the HTM‐N region, two PxVxL‐like motifs were found at aa 352–356 (ARVSL in humans) and 372–376 (ALVAL in humans). These motifs were moderately conserved across amniota, including Homo sapiens, Mus musculus, Phascolarctos cinereus, Ornithorhynchus anatinus, Alligator mississippiensis, Chrysemys picta, Anolis carolinensis, and Gallus gallus (Figure S2A,B). In the HTM‐C region, a motif combining PxVxL‐ and PxxVxL‐like sequences (Maeda & Tachibana, 2022; Liu et al., 2017) was found at aa 399–406 (PYVVRVDL in humans), which was conserved only among mammals (Figure S2A,B). To confirm that these HP1 binding motifs are necessary for chromocenter targeting, we engineered amino acid substitution mutants by replacing Val to Asp (Figure 1) because it has been reported that a V to D substitution in the PxVxL disrupts the hydrophobic interaction with HP1 (Vassallo & Tanese, 2002; Liu et al., 2017). The V354D mutant of the HTM‐N distributed homogenously throughout the cells (Figure 1a,b), suggesting that this mutation abolished the interaction between the HTM‐N and HP1. Conversely, the V374D mutant still accumulated in the chromocenters, albeit the enrichment ratio was substantially lower compared to the wild‐type HTM‐N (1.59 vs. 2.59) (Figure 1b,c). Thus, the PxVxL‐like motif at aa 352–356 (ARVSL) alone is likely to bind to HP1 and the motif at aa 372–376 (ALVAL) assists in a more stable binding. Regarding the HTM‐C, mutations V402D, V404D, and the double V402D‐V404D mutations eliminated chromocenter localization (Figure 1d), suggesting that both Val residues within the HTM‐C are crucial for HP1 binding. A full HTM mutant containing the triple mutation V354D‐V402D‐V404D did not accumulate at chromocenters (Figure 1e), indicating that there are no additional regions that could mediate heterochromatin targeting within SUV420H2 HTM. Moreover, the heterochromatin enrichment ratios of the HTM mutants harboring single V354D and the double V402D‐V404D mutations (2.05 and 2.95, respectively) were comparable to those of the HTM‐C (1.73) and HTM‐N (2.59), respectively (Figure 1c,e), supporting the view that the SUV420H2 HTM binds to HP1 through three HP1 binding motifs.

FIGURE 1.

Human SUV420H2 contains three HP1 binding sites within HTM. (a) Schematic representation of SUV420H2 domains, including the amino acid sequence of HTM. The N‐ and C‐terminal heterochromatic targeting regions (aa 352–380 and 389–410) are labeled as HTM‐N and HTM‐C, respectively. Val residues in HP1 binding motifs are indicated in red. (b–e) Analysis of the impact of Val to Asp mutations in HP1‐binding PxVxL‐like motifs on heterochromatin localization. sfGFP‐tagged HTM‐N (b), HTM‐C (d), and full HTM (e), along with their mutants, were expressed in A9 cells. Heterochromatin accumulation of each sfGFP‐tagged protein is represented as heterochromatin enrichment ratio relative to the nucleus (c). (b, d, and e) Single confocal sections of typical nuclei. (c) Boxplots. Center lines represent the medians; box limits indicate the 25th and 75th percentiles; whiskers extend 1.5 times the interquartile range from the 25th to 75th percentiles; × denote the means; and individual data points are shown as dots. Median values and number of analyzed nuclei are indicated above and to the right of box plots, respectively. Scale bars: 10 μm.

To investigate the interaction between HP1 and the HTM‐N and HTM‐C regions, as well as the impact of mutations within HP1‐binding motifs, we carried out co‐immunoprecipitation assays using cell extracts prepared from HeLa cells that express sfGFP, HTM‐N‐sfGFP, HTM‐C‐sfGFP, and their corresponding Val to Asp mutants, namely HTM‐N V354D and HTM‐C V402D‐V404D (Figure 2a,b). During preparation of cell lysates using a high salt buffer containing 1 M NaCl, HP1 proteins, especially HP1β, were less efficiently extracted from cells expressing HTM‐N‐sfGFP, along with the HTM‐N‐sfGFP itself (Figure 2a, lanes 5 and 14), compared to those expressing sfGFP and HTM‐N V354D‐sfGFP (Figure 2a, lanes 4 and 13, and 6 and 15). This suggests that the expression of HTM‐N may stabilize the association of HP1 with chromatin (a point that will be discussed later). Following immunoprecipitation using the cell extracts with anti‐GFP magnetic beads, the recovery of the sfGFP fusion proteins and HP1 subtypes was assessed by immunoblotting. All three HP1 subtypes, HP1α, HP1β, and HP1γ, were co‐immunoprecipitated with both HTM‐N‐sfGFP (Figure 2a, lane 11) and HTM‐C‐sfGFP (Figure 2b, lane 8), but not with their corresponding mutants (V354D and V402‐V404D) nor with sfGFP alone (Figure 2a, lane 10 and 12, and 2B, lane 7 and 9). Despite the HTM‐N V354D mutant retaining a PxVxL‐like motif at aa 372–376, this motif on its own does not appear to suffice to form a stable complex with HP1. These results, aligning with microscopic data to evaluate heterochromatin targeting, reinforced the conclusion that the SUV420H2 HTM is directed to chromocenters via its interaction with HP1 CSD (Souza et al., 2009).

FIGURE 2.

HTM‐N and HTM‐C interaction with HP1 through their PxVxL‐like motifs. (a) Analysis of HeLa cells expressing sfGFP, HTM‐N‐sfGFP, and HTM‐N V354D‐sfGFP (in which Val354 is substituted with Asp). Cells were treated with a buffer containing 0.1% Triton X‐100 and 1 M NaCl to extract most chromatin‐binding proteins (whole cell fractions; lanes 1–3). Post‐centrifugation, the supernatants (1 M NaCl extract; lanes 4–6) were separated from the pellet (1 M NaCl pellet; lanes 13–15) and utilized as inputs for immunoprecipitation using anti‐GFP beads. The unbound fractions (Unbound; lanes 7–9) and the immunoprecipitated materials (IP; 40× concentrated compared to the input; lanes 10–12) were processed. Protein samples were then separated on SDS‐polyacrylamide gels, transferred to membranes, and probed with the indicated antibodies. (b) Extracts from HeLa cells expressing sfGFP, HTM‐C‐sfGFP, and HTM‐C V402D‐V404D‐sfGFP were prepared, immunoprecipitated, and immunoblotted, as described in (a). The full blots are shown in Supplementary Figure S8.

2.2. Zinc finger‐like motif in HTM‐N is crucial for targeting to heterochromatin

In HTM‐N, three amino acids, H357, C362, and C366, positioned between the two PxVxL‐like motifs, were highly conserved across different species (Figure S2B). To examine whether these conserved amino acid residues are essential for heterochromatin targeting, we expressed HTM‐N‐sfGFP H357A, C362V, and C366V single mutants in A9 cells (Figure 3a). Each of the three mutants distributed throughout the nucleoplasm and cytoplasm (Figure 3a), implying that each of these His and Cys residues is crucial for heterochromatin targeting probably via HP1 binding. Given the importance of His and Cys, we postulated that this region might adopt a structure akin to a zinc finger domain with four Cys/His residues (Neuhaus, 2022) to properly orient the two PxVxL‐like motifs, thereby bridging two HP1 CSD binding sites and stabilizing the interaction. To evaluate this hypothesis, we constructed a mutant that included an insertion of a flexible linker comprising a 3x(GGGGS) sequence and then expressed it in A9 cells (Figure 3b). Although this mutant localized to chromocenters, its enrichment was lower compared to HTM‐N, similar to the mutant lacking the second PXVXL‐like motif (Figure 3b). Thus, in addition to the critical functions of the His and Cys residues in HP1 binding, a specific configuration of the intervening region appears to assist in cooperative binding of the two HP1 binding motifs to stabilize the interaction.

FIGURE 3.

Functions of the linker region between two HP1 binding motifs in HTM‐N. (a) The evolutionally conserved His and Cys residues are critical for HTM‐N heterochromatin targeting. Single amino acid substitution mutants of HTM‐N (H357A, C362V, and C366V) were expressed in A9 cells and their localization was analyzed using confocal microscopy. (b) The heterochromatic accumulation of HTM‐N is weakened by the insertion of a flexible linker between the two HP1 binding motifs and the deletion of the second motif. Single confocal sections of typical nuclei are shown, along with quantitative data as in Figure 1. The averages of enrichment ratios and the numbers of analyzed nuclei are indicated above and below box plots, respectively. p‐values by Tukey–Kamer test are indicated. Scale bars: 10 μm.

2.3. Overexpression of HTM‐sfGFP facilitates HP1β accumulation in H3K9me3‐enriched heterochromatin

HP1 at chromocenters can be maintained and immobilized by “HP1 locking” with two HP1 binding modules within a protein, like SENP7 (Romeo et al., 2015). SUV420H2 HTM‐N was implied to cause HP1 locking because HP1 exhibited resistance to extraction, as demonstrated by immunoblotting (Figure 2a). To investigate the impact of HTM expression on HP1 at the cellular level, immunofluorescence signals of HP1β in wild‐type and HTM‐sfGFP‐expressing cells were compared. For this purpose, we used human HeLa cells because HP1 is highly enriched in chromocenters in mouse cells and a further enrichment by HP1 locking may not be easily detected. To directly compare cells without and with HTM‐sfGFP expression in the same microscopic field, HeLa cells stably expressing HTM‐sfGFP were co‐plated with the parental cells and stained with antibodies specific to HP1β and H3K9me3. HP1β was more pronouncedly concentrated in cells expressing HTM‐sfGFP compared to wild‐type cells (Figures 4a and S3), without affecting H3K9me3 levels (Figure S3B,C). Quantitative measurements of the enrichment ratios of HP1β signals in H3K9me3‐enriched heterochromatin (in the top 10% highest H3K9me3 intensity pixels over the whole nucleus) revealed that the enrichment ratio was substantially higher in cells expressing HTM‐sfGFP (Figure 4b). This result is consistent with the immunoblotting data showing increased levels of the total and extraction‐resistant HP1β in cells expressing HTM‐N‐sfGFP (Figure 2a). A weaker HP1‐locking effect was observed when HTM‐N, which has two PxVxL‐like motifs, was expressed (Figure 4a,b). Conversely, the expression of HTM‐C slightly decreased HP1 accumulation (Figure 4a,b). Note that the HTM‐C and HTM V374D, which is described below, were diffusely distributed throughout the nucleus in paraformaldehyde‐fixed cells, likely because their binding to HP1 is not very stable and cross‐linkable Lys residues are limited. We also examined the effect of an HTM mutant that harbors a Val to Asp substitution in the second PxVxL‐like motif, necessary for stable HP1 binding in conjunction with the first PxVxL‐like motif and the zinc‐finger‐like motif. Although this mutant, HTM V374D, retained two HP1 binding motifs (aa 352–356 and 399–408), no HP1 locking effect was observed (Figure 4a,b). These results thus suggest that the second HP1 binding motif at 372–376 (ALVAL) plays a role in HP1 locking and merely having two HP1 binding sites is insufficient for HP1 locking. Thus, proper configuration of multiple binding motifs appears to be crucial in HP1 locking.

FIGURE 4.

Expression of HTM‐sfGFP and HTM‐N‐sfGFP facilitates HP1β accumulation in H3K9me3‐enriched heterochromatin. (a and b) Effects of SUV420H2 HTM expression on HP1β accumulation in H3K9me3‐enriched heterochromatin. HeLa cells expressing HTM, HTM‐N, HTM‐C, and HTM V374D, each tagged with sfGFP, were stained with specific antibodies directed against H3K9me3 and HP1β. DNA was counterstained with Hoechst 33342. (a) Single confocal sections of typical nuclei. See Figure S4 for larger fields of view containing multiple nuclei. (b) Quantification of HP1β on H3K9me3‐marked heterochromatin. Using images like those shown in (a), the average HP1β intensities in the top 10% of H3K9me3 highest intensity pixels were measured and normalized to the mean intensities in entire nuclei. HP1β enrichments in H3K9me3‐enriched heterochromatin in individual nuclei from a single experiment are shown as dot and box plots on the left. The numbers of nuclei (N) analyzed are indicated above the box plots. Mean values of individual experiments in biological triplicates are shown on the right. Between 17 and 37 nuclei were analyzed for each mutant in single experiments. p‐values obtained from Student's t‐test (paired, two‐tailed) are indicated. (c and d) Co‐expression of sfCherry‐HP1α with HTM, HTM‐N, HTM‐C, and HTM V374D, each tagged with sfGFP. (c) Representative images of sfCherry‐HP1α in live cells when co‐expressed with HTM and its mutants are displayed. (d) Fluorescence recovery after photobleaching. During time‐lapse imaging, an area containing sfCherry‐HP1α focus (yellow circle) was bleached (top panels). The graphs below show the relative fluorescence intensities of sfCherry‐HP1α in the bleached area, normalized by those before bleaching (mean ± s.e.m.). The total number of cells (N) analyzed from two independent experiments are shown. Scale bars: 10 μm.

We next investigated whether the effect on HP1β induced by overexpression of HTM‐N‐sfGFP is also observed for HP1α and HP1γ. HeLa cells stably expressing HTM‐N‐sfGFP were co‐plated with the parental cells and then stained with antibodies against HP1α, HP1β, and HP1γ. The intensity of each HP1 subtype was higher in cells expressing HTM‐N‐sfGFP compared to the control parental cells while HP1β showed the highest intensity increase (Figure S4), which is in good agreement with the immunoblotting data (Figure 2).

Finally, we analyzed the effects of HP1 locking using fluorescence recovery after photobleaching (FRAP) (Hahn et al., 2013; Romeo et al., 2015; Souza et al., 2009) using sfCherry‐HP1α. When sfCherry‐HP1α was co‐expressed with the full HTM in A9 cells, sfCherry‐HP1α appeared to be more concentrated at heterochromatin compared to cells without the full HTM (Figure 4c) and the fluorescence recovery rate was drastically decreased (Figure 4d). The recovery of sfCherry‐HP1α was also slowed when co‐expressed with HTM‐N‐sfGFP, albeit the effect was less pronounced than with the full HTM‐sfGFP (Figure 4d). In contrast, when HTM‐C‐sfGFP was expressed, the recovery of sfCherry‐HP1α was slightly accelerated (Figure 4d), which aligns with the effects of the expression of a single PxVxL module in SENP7 (Romeo et al., 2015). The recovery rate of sfCherry‐HP1α was not affected by HTM V374D‐sfGFP expression. These data are in complete agreement with the immunofluorescence results described above.

2.4. Depletion of SUV420H1 and SUV420H2 did not affect HP1 signals

To investigate whether endogenous SUV420H2 affects HP1 locking, as observed for SENP7 and ORC proteins (Maison et al., 2012; Prasanth et al., 2004; Prasanth et al., 2010; Romeo et al., 2015), we established SUV420H1/H2 double knockout (DKO) HeLa cells, in which H4K20me3 was diminished (Figures 5a and S5). HP1β signals in H3K9me3‐enriched heterochromatin in DKO cells appeared slightly decreased compared to parental cells, although the changes were not always significant in DKO clones (Figures 5a,b and S6A). When HTM‐sfGFP was overexpressed in DKO cells, HP1β heterochromatin accumulation was increased but in a lesser extent than the parental HeLa cells (Figures 5c,d and S6B). These results, in one hand, suggest that the endogenous SUV420H2 contributes subtly to HP1 locking probably because its endogenous expression level is not very high. On the other hand, as the increase of HP1β accumulation in H3K9me3‐enriched heterochromatin in DKO cells was limited compared to parental cells (Figure 5d), other functions of SUV420H2 than the HTM‐mediated locking, such as adding H4K20me3 may enhance HP1 heterochromatin accumulation.

FIGURE 5.

Endogenous SUV420H2 subtly contributes to HP1β locking. (a and b) Slight decrease of HP1β enrichments in H3K9me3‐enriched heterochromatin in SUV420H1/H2 double knockout (DKO) HeLa cells. Parental and three DKO HeLa cell lines were stained with antibodies specific to H4K20me3, H3K9me3, and HP1β. (a) Single confocal images of typical nuclei are shown. See Figure S5A for larger fields of view containing multiple nuclei. (b) Box plots. Average HP1β intensities in the top 10% of H3K9me3 highest intensity pixels were measured and normalized using the mean intensities in entire nuclei. HP1β enrichment in H3K9me3‐enriched heterochromatin in individual nuclei from a single experiment is represented as dot and box plots on the left, with the numbers of nuclei (N) analyzed indicated above. Mean values of individual experiments in biological quadruplicates are shown on the right. p‐values obtained from Dunnett's test are indicated. Between 83 and 136 nuclei were analyzed for each cell line in single experiment. (c and d) The effect of HTM‐sfGFP on HP1β accumulation in H3K9me3‐enriched heterochromatin in DKO cells. Parental HeLa and DKO (clone B6) cells transfected with HTM‐sfGFP were stained with antibodies specific to H3K9me3 and HP1β. (c) Single confocal images of typical nuclei. See Figure S5B for larger fields of view containing multiple nuclei. (d) Box plots as described in (b). Experiments were performed in triplicate with 52–156 cells analyzed in each line in single experiments. p‐values obtained from Tukey–Kramer test are indicated. Scale bars: 10 μm.

2.5. HTM V374D‐sfGFP can be used for visualizing endogenous HP1 on H3K9me3‐enriched heterochromatin

The HTM V374D fragment, which harbors two PxVxL‐like motifs for HP1 binding, localized to constitutive heterochromatin in living cells without exhibiting HP1 locking and unlocking effects. Therefore, we thought that HTM V374D could be utilized as a live‐cell probe for visualizing endogenous HP1 on heterochromatin. While GFP‐tagged HP1 proteins have been used for labeling heterochromatin and live‐cell tracking, their localization to heterochromatin in human cells is less distinct compared to their pronounced highlighting of chromocenters in mouse cells. We compared the localization of HTM‐sfGFP, HTM V374D‐sfGFP, HP1α‐sfGFP (C‐terminus GFP version), and sfGFP‐HP1α (N‐terminus GFP version) in mouse A9 and human HeLa cells, which also expressed HaloTag‐tagged histone H2B (H2B‐Halo). In A9 cells, all four sfGFP‐tagged proteins accumulated at Hoechst‐dense chromocenters. Both HTM‐sfGFP and HTM V374D‐sfGFP apparently highlighted chromocenters more clearly than HP1α‐sfGFP and sfGFP‐HP1α (Figure S7A,B). In HeLa cells, which show less distinctive heterochromatin than mouse cells, the differences among the four proteins were more pronounced (Figures 6a,b and S7C). HTM‐sfGFP and HTM V374D‐sfGFP were enriched in H2B‐Halo‐dense regions, whereas HP1α‐sfGFP and sfGFP‐HP1α were more homogenously distributed throughout the nucleus (Figure 6a). The image contrast, or coefficient of variation, which measures the concentration of sfGFP‐tagged proteins in heterochromatin over the entire nucleus (e.g., Hilbert et al., 2021), was highest for HTM‐sfGFP (Figure 6b), consistent with its stable binding to heterochromatin involving HP1 locking. HTM V374D‐sfGFP was concentrated in heterochromatin less than HTM‐sfGFP but higher than HP1α‐sfGFP and sfGFP‐HP1α, albeit not always statistically significant (Figure 6b). The addition of a bulky tag at either the N‐ or C‐terminus could interfere with HP1α localization. Many studies have preferred N‐terminus GFP‐fusion proteins because C‐terminus fusion could affect CSD function (Hayakawa et al., 2003; Nozawa et al., 2010; Romeo et al., 2015). However, a previous study showed a quite diffused distribution of GFP‐HP1α (Nozawa et al., 2010), probably because the N‐terminus fusion close to CD could affect H3K9me3 binding. The nucleoplasmic background of sfGFP‐HP1α was indeed more intense than that of HP1α‐sfGFP (Figure 6a,b), although a heterochromatic concentration was still visible. The fusion partner protein and the linker amino acids may affect the binding ability of N‐terminus‐tagged HP1α. In addition, sfGFP‐HP1α, and HP1α‐sfGFP to a lesser extent, often formed tiny bright foci devoid of H2B‐Halo, which likely represent PML bodies (Figure 6a, arrows) (Everett et al., 1999; Hayakawa et al., 2003; Lehming et al., 1998; Seeler et al., 1998). Since HP1 interacts with SP100, a constitutive component of PML bodies through its CSD (Hayakawa et al., 2003), GFP‐tagged HP1 can also target PML bodies via such interactions. As HTM V374D binds to the HP1 CSD, HP1 molecules that are localized to specific compartments via CSD could not be recognized by HTM V374D.

FIGURE 6.

Visualizing HP1‐bound heterochromatin in living cells using HTM V374D‐sfGFP. (a and b) Distribution of HTM‐sfGFP, HTM V374D‐sfGFP, sfGFP‐HP1α, HP1α‐sfGFP in interphase nuclei of HeLa cells. H2B‐Halo was co‐expressed to visualize global distribution of chromatin. (a) Confocal images. Arrows indicate the accumulation of sfGFP‐tagged HP1 in small foci devoid of H2B‐Halo. (b) The contrast (coefficient of variation) of sfGFP‐tagged proteins in HeLa cells. Coefficient variations of the pixel intensities in single nuclei from a single experiment are shown in box plots on the left. The numbers of analyzed nuclei are indicated above the plots. Mean values from individual experiments in biological triplicates are shown on the right. p‐values obtained from Dunnett's test are indicated. (c) Distribution of HTM‐sfGFP, HTM V374D‐sfGFP, HP1α‐sfGFP, sfGFP‐HP1α in mitotic cells, with H2B‐Halo co‐expressed to visualize chromosomes. (d) Distribution of HTM‐sfGFP and HTM V374D‐sfGFP with co‐expressed sfCherry‐HP1α in mitotic cells. High‐power views of boxed area are shown in insets. (e) Dynamics of HP1‐enriched heterochromatin during DNA replication in the late S phase. Yellow arrowheads indicate the same foci, with high‐power views shown in insets. Scale bars: 10 μm.

During mitosis, HP1α is known to dissociate from condensed chromosomes due to Aurora B kinase‐mediated histone H3 Ser10 phosphorylation, which inhibits HP1 binding to the neighboring K9 di‐/tri‐methylation (Fischle et al., 2005; Hirota et al., 2005), and it redistributes to be concentrated at centromere regions through its binding to the chromosome passenger complex via CSD (Ainsztein et al., 1998; Hayakawa et al., 2003; Minc et al., 1999; Nozawa et al., 2010). Both sfGFP‐HP1α and HP1α‐sfGFP were indeed concentrated at centromere‐like dots in mitotic chromosomes (Figure 6c). Unexpectedly, however, HTM‐sfGFP was localized at the surface of condensed chromosomes during mitosis (Figure 6c). When sfCherry‐HP1α and HTM‐sfGFP were co‐expressed, both proteins were also observed at the periphery of mitotic chromosomes, whereas most were diffused into the cytoplasm, and sfCherry‐HP1α no longer accumulated at centromeres (Figure 6d). This observation suggests that HP1 locking by HTM can block the access of Aurora B kinase to histone H3 tail during the G2 and M phases. Since Aurora B kinase is localized to the inner centromere and its activity gradually decreases toward the outer regions (Fuller et al., 2008), HTM‐bound HP1α might remain at the far periphery of chromosomes. Thus, HTM‐sfGFP, which locks HP1, disturbs normal HP1 dynamics and causes mislocalization. By contrast, HTM V374D‐sfGFP dissociated from mitotic chromosomes and sfCherry‐HP1α was localized to centromeres (Figure 6d). Thus, HTM V374D‐sfGFP can be useful to specifically visualize HP1‐bound heterochromatin during interphase without affecting HP1 dynamics.

To demonstrate the use of HTM V374D‐sfGFP as a live‐cell heterochromatin marker, we tracked the dynamics of HTM V374D‐sfGFP in chromocenters during the S‐phase progression in A9 cells, together with H2B‐Halo and mCherry‐PCNA to visualize chromatin and DNA replication foci (Chagin et al., 2016; Leonhardt et al., 2000), respectively (Figure 6e and Movies 1 and 2). Previous studies using fixed cells have reported that DNA replication in heterochromatin begins at the periphery of HP1‐rich domains, with replicated DNA moving toward the domain centers (Quivy et al., 2004). However, there are no reports observing the dynamics of HP1‐rich domains during replication through live imaging. We observed mCherry‐PCNA accumulating around domains enriched in HTM V374D‐sfGFP—and so chromatin‐bound HP1—during the late S phase. A detailed view revealed that mCherry‐PCNA first accumulated at the periphery of a chromocenter and then expanded toward the center. During this process, the HTM‐V374D‐sfGFP focus became fuzzy and transformed into a donut‐like shape (Figure 6e, 03:30–05:00; Movie 2). As the mCherry‐PCNA signal decreased upon the completion of DNA replication, the HTM‐V374D‐sfGFP focus reformed into an oval shape (Figure 6e, 05:00–8:00); Movie 2. This observation suggests that HP1 may temporarily dissociate from heterochromatin during DNA replication, and/or heterochromatin may become decondensed during this process (Chagin et al., 2019; Leonhardt et al., 2000).

3. DISCUSSION

3.1. HP1 binding of SUV420H2 HTM

In this study, we identified three HP1 binding motifs within the HTM that are required for efficient heterochromatin localization of SUV420H2. The first and second motifs can function cooperatively, assisted by the zinc‐finger‐like motif harboring one His and two Cys residues. The full HTM containing the all three motifs, as well as the 29‐aa HTM‐N peptide containing two PxVxL‐like motifs, possesses HP1 locking activity. This stabilizes HP1 on heterochromatin through bivalent binding, as observed in SENP7, Orc3 and PRR14, all of which have multiple HP1 binding motifs (Kiseleva et al., 2023; Maison et al., 2012; Prasanth et al., 2004; Prasanth et al., 2010; Romeo et al., 2015). The 22‐aa HTM‐C peptide, which harbors the third motif with an overlapped PxVxL and PxxVxL stretch, also targets heterochromatin and slightly destabilizes the HP1‐chromatin interaction. This is likely due to unlocking, as demonstrated by SENP7's single HP1 binding motifs accelerating HP1 dissociation (Romeo et al., 2015). Hence, the stability of HP1 chromatin binding is likely controlled by the balance between different effector proteins harboring single and multiple cooperative HP1 binding modules. The full HTM, containing all three motifs, exhibits the highest HP1 locking effect. However, mutation in the second HP1 binding motif (V374D) abolishes this effect, while the other two motifs remain. Therefore, HP1 locking efficiency depends on the configuration of multiple HP1 binding sites.

Given the strong HP1 locking function in SUV420H2 HTM, a decrease of HP1 locking in SUV420H1/H2 DKO cells could be expected. However, HP1 levels at heterochromatin in DKO cells exhibited only a subtle decrease compared to those in wild‐type HeLa cells. The expression level of SUV420H2 in HeLa cells may be too low to induce substantial HP1 locking, and the multiple HP1 binding sites may simply increase its heterochromatin targeting efficiency for H4K20me3 deposition in an HP1‐dependent manner. In contrast, H4K20me3 deposited by SUV420H2 appears to enhance the capacity of the HP1 locking, likely through recruiting proteins such as LRWD1/ORCA, which binds to ORC, and DNA methyltransferase 1 (Ren et al., 2021; Vermeulen et al., 2010). A subtle HP1 locking induced by SUV420H2 and/or H4K20me3 may also induce nano‐compaction, detectable through energy transfer between two histones (Dupont et al., 2023). It has been reported that HP1β is functionally associated with SUV420H2 and H4K20me3 (Bosch‐Presegué et al., 2017), which may be consistent with our finding that the HP1 locking effect induced by HTM‐N expression was more pronounced in HP1β compared to HP1α and HP1γ. The detailed molecular mechanism determining how HTM binds to different HP1 subtypes through amino acids surrounding the PxVxL motifs (Canzio et al., 2014) remains to be elucidated.

3.2. Visualizing HP1‐bound heterochromatin using HTM V374D

As HTM V374D does not induce HP1 locking nor unlocking, like HTM or HTM‐C, this fragment can be applicable for visualizing heterochromatin marked by HP1. Although HP1 could be directly visualized by tagging with the fluorescent proteins, GFP‐tagged HP1 does not always exhibit heterochromatin localization in human cells, whereas it is concentrated in chromocenters in mouse cells (Figures 6 and S7) (Dialynas et al., 2007), suggesting that the bulky protein tag may interfere with the binding to H3K9me3 and/or effector proteins. The binding of HTM V374D probe to HP1 CSD could also block the binding of effector proteins. However, under low expression levels, potential blocking effects may be minimal because the dynamics of HP1 remains unchanged when HTM V374D is expressed, whereas the expression of HTM stabilizes HP1 chromatin binding and causes HP1 mislocalization during mitosis. Indeed, we have demonstrated that HTM V374D can be used for tracking DNA replication at chromocenters during the mid to late S phase, in combination with mCherry‐PCNA. We observed that HTM V374D probe became less concentrated, either by dissociation or massive decondensation (Chagin et al., 2019; Leonhardt et al., 2000), when replication occurs from the surface of chromocenters. To monitor H3K9me3 in living cells, several probes based on HP1 CD have been developed (Sánchez et al., 2019; Sasaki et al., 2022; Villaseñor et al., 2020). The binding of these HP1 CD‐based probes to H3K9me3 can be influenced by H3S10 phosphorylation, similar to HTM V374D, which dissociates from mitotic chromosomes. The HTM V374D probe, as reported here, offers an additional option for tracking HP1‐bound H3K9me3‐rich heterochromatin in living cells.

4. EXPERIMENTAL PROCEDURES

4.1. Cells and transfection

HeLa (obtained from Peter R. Cook at Oxford University; Pombo et al., 1999), A9 (obtained from Nobuo Takagi at Hokkaido University), and NIH3T3 cells (obtained from Nobuo Takagi at Hokkaido University) were grown in Dulbecco's modified Eagle's medium (DMEM), high‐glucose (Nacalai Tesque) containing 10% fetal bovine serum (FBS) (Gibco, Thermo Fisher Scientific) and 1% L‐glutamine–penicillin–streptomycin solution (GPS; Sigma‐Aldrich) at 37°C in a 5% CO₂ atmosphere. For transfection, FuGENE HD (Promega) was used according to the manufacturer's instructions. Briefly, 2 μg DNA was mixed with 6 μL of FuGENE HD in 100 μL of Opti‐MEM (Thermo Fisher Scientific) and incubated at RT for 10 min before being added to HeLa cells grown in 35‐mm glass‐bottomed dishes (AGC Technology Solutions) at 40%–70% confluency. To obtain stably expressing cells, 1.8 μg PB533‐ or PB510‐based PiggyBac plasmid (System Biosciences) and 0.2 μg transposase expression vector (System Biosciences) were used, and cells were selected in 1 mg/mL G418 (Nacalai Tesque). Cells stably expressing sfGFP fusion proteins at similar levels among different mutants were collected using a cell sorter (SH800; Sony).

4.2. Plasmid construction

The HaloTag‐tagged KDM4D expression vector (FHC06842; Promega), the PB533‐based H2B‐Halo expression vector (Uchino et al., 2022), and the PB533‐based mCherry‐PCNA expression vector (Uchino et al., 2022) were reported previously. For transient expression in mammalian cells, psfGFP‐N1 (Addgene 54,737) and psfCherry‐C1 (Kono et al., 2022) were used. To construct SUV420H2 deletion mutants, the HaloTag‐tagged SUV420H2 expression vector (FHC26822; Promega) was used as a template for PCR amplification of the target regions. The primers used for PCR in this study are listed in Supplementary Table S2. The amplified DNA fragments were then cloned into psfGFP‐N1 linearized by EcoRI and BamHI digestion, using the In‐Fusion® HD Cloning Kit (Z9648N; TaKaRa). To construct psfCherry‐HP1α, HP1α coding region was amplified by PCR using the GFP‐HP1α expression vector provided by C. Obuse (Nozawa et al., 2010) as a template. The amplified DNA fragment was cloned into psfCherry‐C1 linearized by EcoRI digestion. To introduce point mutations, inverse PCR was conducted (Uchino et al., 2022). Some coding sequences with amino acid substitutions were amplified using a plasmid that already harbored another mutation. For establishing stably expressing cells, PiggyBac system was used. The fragment coding HTM mutants‐sfGFP in psfGFP‐N1 was exercised by EcoRI and NotI digestion and ligated into the PB533 vector digested with EcoRI and NotI. The psfGFP‐352‐(GGGGS) × 3–380 was constructed by insertion of a flexible linker using primers (pSUV420H2[352‐]_sfGFP_InF_s, pSUV420H2[−380]_sfGFP_InF_as, LINK.AMP5T, LINK.AMP3T, 370‐GSLINK_s, 369‐GSLINK_as) as described previously (Sato et al., 2013). The nucleotide sequence of all the inserts was verified.

4.3. Live cell microscopy

For Figures 1, 3, S1C, and S1D, A9 cells (Figures 1, 3, and S1D) or those stably expressing HTM‐sfGFP (Figure S1C) were plated in a 24‐well glass‐bottom plate (AGC Technology Solutions) at a cell density of 0.6–1.2 × 10⁵ cells/well. The following day, cells were transfected with plasmids to express sfGFP‐tagged SUV420H2 mutants (Figures 1, 3, and S1D) or Halo‐KDM4D (Figure S1C). One day post‐transfection, the medium was replaced with FluoroBrite (Thermo Fisher Scientific) containing 10% FBS and 1% GPS. To detect Halo‐KDM4D, cells were incubated with HaloTag TMR Ligand (Promega) at a final concentration of 100 nM for 30 min incubation, before replacing the medium. The cell culture plates were placed on a heated stage (Tokai Hit) at 37°C under 5% CO₂ regulated by a CO₂ control system (Tokken) on a confocal microscope (FV‐1000; Olympus) operated by built‐in software (Fluoview ver. 4.2) with a PlanSApo 60× (NA 1.40) oil‐immersion objective lens. Images were acquired using the line‐sequential imaging mode (2.0% 488‐nm laser transmission; 1024 × 1024 pixels; pinhole 200 μm; 2× zoom for HeLa cells or 3× zoom for A9 cells; 2‐line Kalman filtration) with a 405/488/543/633 dichromic mirror and a 505–525 emission filter. Image analysis was performed using the NIS‐elements Analysis software ver. 5.1 (Nikon). After background subtraction, the chromocenters and the entire nucleus were both selected by Magic Wand tool to measure the fluorescence intensities in the selected areas. Heterochromatin enrichment ratios were calculated by dividing the mean intensity of chromocenters by that of the nucleus.

For Figures 6 and S7, A9 and HeLa cells were plated in a 24‐well glass‐bottom plate (AGC Technology Solutions) at a cell density of 0.8 × 10⁵ and 0.6 × 10⁵ cells/well, respectively. The next day, the cells were transfected with plasmids to express HTM‐sfGFP, HTM V374D‐sfGFP, HP1α‐sfGFP, or sfGFP‐HP1α. In some cases, H2B‐Halo and sfCherry‐HP1α expression vectors were co‐transfected. One to three days post‐transfection, cells were incubated in 1 μg/mL Hoechst 33342, or 100 nM HaloTag TMR Ligand (Promega) when H2B‐Halo is expressed, for 30 min. The medium was replaced with FluoroBrite (Thermo Fisher Scientific) containing 10% FBS and 1% GPS, before the cell culture plates were set on to a heated stage at 37°C under 5% CO₂ (Tokai Hit) on a confocal microscope (A1R; Nikon) operated by NIS Elements ver. 5.21.00 (Nikon) with an Apo TIRF 60× (NA 1.49) oil‐immersion objective lens, a 405/488/543/633 dichromic mirror, and 450/50, 525/50, and 595/50 emission filters. For Figure 6d, images were collected using the line‐sequential imaging mode (512× 512 pixels; pinhole 39.59 μm; 6× zoom; 2‐line Kalman filtration) with three laser lines (1.0% 405‐nm, 0.2% 488‐nm, and 0.5% 561‐nm laser transmissions). For others, images were collected using the line‐sequential imaging mode (1024 × 1024 pixels; pinhole 39.59 μm; 2× zoom for HeLa or 3× zoom for A9; 2‐line Kalman filtration) with two laser lines (0.1% 488‐nm and 0.1% 561‐nm laser transmissions for HeLa; 0.8% 405‐nm and 0.2% 488‐nm laser transmission for A9). The intensity variations in nuclei (for Figure 6b) were measured using NIS‐elements Analysis software ver. 5.1 (Nikon). After background subtraction, nuclei were selected using the Magic Wand tool, and the mean fluorescence intensities with standard deviation were measured. Coefficient variations were calculated by dividing the standard deviation by the mean fluorescence intensity in single nuclei.

4.4. Immunofluorescence

Antibodies and staining conditions used in this study are listed in Supplementary Table S2. For Figure S1B, HeLa, NIH3T3, and A9 cells grown on a 24‐well glass‐bottom plate were transfected with expression plasmids of HTM‐sfGFP and its derivatives. One day post‐transfection, cells were fixed with 4% paraformaldehyde (PFA; Electron Microscopy Sciences) in 250 mM HEPES‐NaOH (pH 7.4) for 5 min at room temperature, washed with Dulbecco's phosphate buffered saline, calcium‐ and magnesium‐free (PBS; Fujifilm Wako Chemicals), and permeabilized using 1% Triton X‐100 (Nacalai Tesque) for 20 min with gentle shaking. Cells were then incubated in Blocking‐One P solution (Nacalai Tesque) for 20 min with gentle shaking and then stained with Hoechst 33342 (1 μg/mL) and mouse monoclonal antibody directed against H3K9me3 directly labeled with Cy3 (2 μg/mL; Chandra et al., 2012) in 10% Blocking‐One P in PBS for 3 h at room temperature, before washing with PBS three times. Fluorescence images were collected using a spinning disk confocal microscope system (CSU‐W1; Yokogawa and Ti‐E; Nikon) with a PlanApo VC 100× (NA 1.40) oil‐immersion objective lens, a 405/488/561/640 dichroic mirror, and 450/50, 525/50, 595/50, and 700/75 emission filters, equipped with an electron‐multiplying charge‐coupled device (iXon+; Andor) and 405‐, 488‐, 561‐, and 640‐nm laser lines (LU‐N4; Nikon).

For Figure S5A, parental HeLa and SUV420H1/2 DKO cells were plated on a 24‐well glass‐bottom plate. The following day, cells were fixed with 4% PFA in 250 mM HEPES‐NaOH (pH 7.4) with 0.1% TritonX‐100 for 5 min at room temperature, then permeabilized and blocked as described above, before staining with Hoechst 33342 (1 μg/mL) and fluorescent dye‐labeled mouse monoclonal antibodies specific to H4K20me3 (Cy3), H4K20me2 (Alexa488) and H4K20me1 (Cy5) (4 μg/mL for each antibody) in 10% Blocking‐One P in PBS at 4°C overnight (Hayashi‐Takanaka et al., 2015). Cells were then washed with PBS three times. The samples were set on a confocal microscope (A1R; Nikon) operated by NIS Elements ver. 5.21.00 (Nikon) with a PlanApo λ 100× (NA 1.45) oil‐immersion objective lens, a 405/488/561/640 dichroic mirror, and 450/50, 525/50, 595/50, and 700/75 emission filters. Images were collected using line‐sequential imaging mode (1024 × 1024 pixels; pinhole 255.43 μm; 2× zoom; 2‐line Kalman filtration) with four laser lines (LU‐N4; Nikon; 0.1% 405‐nm laser transmission; 0.1% 488‐nm laser transmission; 0.1% 561‐nm laser transmission; and 0.1% 640‐nm laser transmission). For quantitative analysis, individual nuclei were defined using magic wand tool in NIS Elements ver. 5.30.02 to measure the mean fluorescence intensities in single nuclei.

4.5. HP1 locking assay by immunofluorescence

Cell were plated in a 35‐mm glass‐bottom dish a day before fixation (total 4–5 × 10⁵ cells; Figures 4a, 5a,b, S3A, S4A, and S6A; cells expressing sfGFP‐tagged protein and the parental cells were mixed at a 1:1 ratio) or a day before transfection (2–2.5 × 10⁵ cells; Figures 5c and S6B). One day after cell plating or transfection, the cells were fixed, permeabilized, blocked, and stained with the primary and the secondary antibodies (Supplementary Table S2), as described above. Fluorescence images were acquired using a confocal microscope (A1R; Nikon) operated by NIS Elements ver. 5.21.00 (Nikon) with an Apo TIRF 60× (NA 1.49) oil‐immersion objective lens, a 405/488/561/640 dichroic mirror, and 450/50, 525/50, 595/50, 700/75 emission filters, using the line‐sequential imaging mode (1024 × 1024 pixels; pinhole 26.82 μm; 2× zoom; 2‐line Kalman filtration) with four laser lines (0.3%–0.6% 405‐nm laser transmission; 0.3%–0.4% 488‐nm laser transmission; 0.1%–4% 561‐nm laser transmission; 0.3%–0.8% 640‐nm laser transmission).

For quantitative analysis for Figures 4b and 5b,d, individual nuclei were defined using the Magic Wand tool in NIS Elements ver. 5.30.02 and minimum rectangle areas containing single nuclei were cropped for exporting as TIF files. The intensity in each pixel was measured using “TIFF” package in R (https://www.r-project.org/). To compare the changes in the enrichment ratio of HP1β in the H3K9me3‐enriched domain, the pixels showing the top 10% highest intensity in the H3K9me3 channel were selected and then the mean intensity of HP1β in these H3K9me3‐top10% pixels was divided by that in the whole nucleus for normalization. To compare the changes of HP1α, HP1β and HP1γ signals by HTM‐N‐sfGFP expression, the mean intensity of HP1 signals was divided by the mean value of the mean intensities of HP1 in control cells.

4.6. HP1 locking assay by fluorescence recovery after photobleaching (FRAP)

A9 cells plated in a 24‐well glass bottom plate were co‐transfected with transient expression plasmids of HTM‐sfGFP, HTM‐N‐sfGFP, HTM‐C‐sfGFP, or HTM V374D‐sfGFP, along with the sfCherry‐HP1α expression plasmid. The next day, the medium was replaced with FluoroBrite (Thermo Fisher Scientific) containing 10% FBS and 1% GPS, before a dish was placed on a heated stage (Tokai Hit) at 37°C under 5% CO₂ regulated by a CO₂ control system (Tokken) on a confocal microscope (FV‐1000; Olympus), operated by built‐in software (Fluoview ver. 4.2) with a PlanSApo 60× (NA 1.40) oil‐immersion objective lens. For the FRAP experiment, 20 images were collected (10% 543‐nm laser transmission; 128 × 32 pixels; pinhole 800 μm; 12× zoom). A 1.38 μm diameter circular area was bleached (100% transmission for 458‐, 488‐, 515‐, and 543‐nm laser lines; duration 364.32 ms), followed by the consecutive collection of another 80 images. Fluorescence intensity measurements were performed using NIS Elements ver. 5.30.02. The net intensities of the bleached and unbleached areas were calculated by subtracting the background intensity outside nuclei in each time frame. To calculate relative intensities to the initial intensity of the bleached area, the global loss of fluorescence was normalized by dividing the intensity in bleached area by that in unbleached area, before normalizing to the average intensity of the pre‐bleach images.

4.7. Double knockout of SUV420H1/H2

Lacking access to a highly specific antibody for SUV420H2 knockout validation, we decided to establish SUV420H1/H2 double knockout (DKO) cells. These cells were reported to have almost no H4K20me3 and H4K20me2, and have greater level of H4K20me1 (Schotta et al., 2008). To establish DKO cells of SUV420H1 and SUV420H2, the CRISPR/Cas9 system (pX330, Addgene 42,230; pKN7, Addgene DU70250; and pX459, Addgene 62988) was used. Plasmid construction followed the “Target Sequence Cloning Protocol” provided by Feng Zhang lab (https://www.addgene.org/crispr/zhang/). The gRNAs were designed using the “CRISPR Finder” (Wellcome Sanger Institute Genome Editing; https://wge.stemcell.sanger.ac.uk/find_crisprs), targeting the first exons and SET domains, considering several known splicing isoforms. The sequences of the primers used for plasmid construction are summarized in Supplementary Table S1. HeLa cells were plated in a 6‐well plate at a density of 2.4 × 10⁵ cells/well. The following day, cells were transfected with plasmids for KO (pX330 and pKN7 for the first exons; and pX459 for SET domains) using Lipofectamin 2000 according to the manufacturer's instruction. One and three days post‐transfection, the medium was replaced to DMEM containing 1 μg/mL puromycin. For single cell cloning, puromycin‐resistant cells were seeded at a density of 50–100 cells in a 10 cm dish. After a week, single colonies were picked and transferred to wells in two 96‐well plates; one with a plastic bottom for stock and another with a glass bottom for immunofluorescence. To screen and validate DKO cells, immunofluorescence was performed using the fluorescent dye directly‐labeled mouse monoclonal antibodies against H4K20me3 (Cy3), H4K20me2 (Alexa488), and H4K20me1 (Cy5). DKO cells were generated by three steps. First, the first exons of SUV420H1 and SUV420H2 were simultaneously targeted. As H4K20me2 and H4K20me3 signals were still observed, the SET domain of SUV420H1 was targeted next. In one clone (#20), post‐immunofluorescence screening, H4K20me2 signals were diminished to background level but some H4K20me3 signals remained. The SET domain of SUV420H2 was then targeted. Three clones (B6, C9, and D2) showed no signals for both H4K20me2 and H4K20me3 (Figure S5).

To validate gene disruption by CRISPR/Cas9‐based genome editing, the targeted regions of SUV420H1 and SUV420H2 mRNA were amplified by RT‐PCR and subjected to nucleotide sequencing. The total RNAs were isolated from the parental HeLa, clone #20, B6, C9, and D2 lines (~1 × 10⁵ cells) using TRIzol RNA isolation Reagents (Thermo Fisher Scientific), according to the manufacturer's instruction. RT‐PCR was conducted using the OneStep RT‐PCR Kit (QIAGEN) with primers listed in Supplementary Table S1. After agarose‐gel electrophoresis, PCR products were purified using the QIAquick Gel Extraction kit (Qiagen) and the nucleotide sequence of the purified fragments was directly determined without subcloning.

4.8. Time‐lapse observation of the HP1 dynamics during the late S phase

For Figure 6e, A9 cells stably expressing sfGFP‐HTM‐C, mCherry‐PCNA, and H2B‐Halo, were plated on a 35‐mm glass‐bottom dish (AGC Technology Solutions). The next day, Janelia Fluor 646® HaloTag Ligand® was added to the medium at a final concentration of 100 nM. After a 30‐min incubation, the medium was replaced with FluoroBrite (Thermo Fisher Scientific) containing 10% FBS and 1% GPS. The dish was placed on a spinning disk confocal microscope system (CSU‐W1; Yokogawa and Ti‐E; Nikon) equipped with a PlanApo VC 100× (NA 1.4) oil‐immersion objective lens (with Type F immersion oil; MXA22168; Nikon) and integrated with a culture system (Tokai Hit) maintained at 37°C under 5% CO₂. Fluorescence images were captured using NIS Elements ver.5.11.03 (Nikon), operated with an LDI‐7 Laser Diode Illuminator (Chroma Technologies Japan; 10% 470‐nm laser transmission; 20% 555‐nm laser transmission; 50% 640‐nm laser transmission) was used with a 405/470/555/640 dichroic mirror, a 520/60, a 600/50, and a 690/50 emission filters, and an EM charge‐coupled device (iXon+, Andor; gain 300; exposure time 470 nm: 1 s, 555 nm: 1 s, 640 nm: 500 ms). Nine different focal planes were imaged at 0.5 μm intervals every 5 min.

4.9. Immunoprecipitation and western blotting

HeLa cells expressing the sfGFP fusion proteins (4.4 × 10⁶ cells) were plated on a 10‐cm cell culture dish (Greiner). The next day, cells were collected using a cell lifter (Corning), washed with ice‐cold PBS (Takara), and resuspended in 500 μL of ice‐cold Lysis buffer (1 M NaCl, HEPES‐NaOH [pH 7.4, Nacalai Tesque], 300 mM sucrose [Nacalai Tesque], 0.1% Triton X‐100 [Nacalai Tesque], 1 mM MgCl₂ [Sigma Aldrich], 1 mM EGTA [Nacalai Tesque]). Small aliquots (20 μL) were kept as “Whole cell” fractions for immunoblotting. After centrifugation (20,000×g, 20 min, 4°C), the supernatants (470 μL) were collected and 625 U of Benzonase (Novagen) was added to each sample to digest nucleic acids, thereby eliminating DNA/RNA‐mediated interactions. The pellets were resuspended in the original volumes of Lysis buffer for “Insoluble pellet” fractions. After a 30‐min incubation on ice, the supernatants (450 μL) were transferred to new tubes and mixed with an equal volume (450 μL) of NaCl‐free Dilution Buffer (HEPES‐NaOH [pH 7.4], 300 mM Sucrose, 0.1% Triton X‐100, 1 mM MgCl2, 1 mM EGTA) to lower the salt concentration to 500 mM for immunoprecipitation. Following centrifugation (20,000×g, 20 min, 4°C), the supernatants were collected, and 20 μL aliquots were kept as “IP input” fractions for immunoblotting. The remaining supernatants were mixed with GFP‐Trap magnetic beads (8 μL of slurry per sample; Chromotek, gtma‐20) prewashed with IP Buffer (500 mM NaCl, HEPES‐NaOH [pH 7.4], 300 mM Sucrose, 0.1% Triton X‐100, 1 mM MgCl₂, 1 mM EGTA). The mixtures were incubated at 4°C overnight with rotation. After collecting the beads using a magnetic stand (Thermo Fisher Scientific), supernatants were transferred to a new tube for “Unbound” samples. The beads were washed with ice‐cold IP Buffer and then resuspended in 20 μL of IP Buffer (×44 concentrated compared to the input). To adjust the concentration of “Whole cell” and “Insoluble pellet” fractions to that of “IP input,” an equal volume of NaCl‐free Dilution Buffer was added. Samples for immunoblotting were mixed with 2× Sample‐Loading Buffer (125 mM Tris–HCl, pH 6.8, 20% glycerol [Fujifilm Wako Chemicals], 4% sodium dodecyl sulfate [SDS; Fujifilm Wako Chemicals], 0.01% bromophenol blue [Fujifilm Wako Chemicals], and 10% dithiothreitol [Fujifilm Wako Chemicals]) and heated at 95°C for 10 min. Then, 5 μL of each sample was separated on 15% polyacrylamide gels (SuperSep™ Ace, 17 well pre‐cast; Fujifilm Wako Chemicals) and transferred to FluoroTrans W PVDF Transfer Membranes (Pall; 90 min; 170 mA constant for a 9 cm × 9 cm membrane) using EzFastBlot (Atto) as a transfer buffer. The membranes were blocked with Blocking One (Nacalai Tesque) for 30 min with gentle shaking. After washing with TBST (20 mM Tris–HCl, pH 8.0, 150 mM NaCl, 0.02% Tween 20), the membranes were incubated with the primary antibody, that is, rabbit monoclonal anti‐HP1α (1:10,000; Abcam; ab109028), rabbit monoclonal anti‐HP1β (1:1000; Cell Signaling Technology; D2F2, #8676), rabbit polyclonal anti‐HP1γ (1:1000; Cell Signaling Technology; #2619), and rabbit polyclonal anti‐GFP (1:2000; MBL; No.598), in Can‐GetSignal® Solution 1 (Toyobo) for 2 h at room temperature. After washing the membranes with TBST three times, they were incubated with horseradish peroxide‐conjugated goat anti‐mouse or anti‐rabbit IgG (H + L) (1:10,000; Jackson ImmunoResearch) in Can‐Get‐Signal® Solution 2 (Toyobo) for 1 h at room temperature. The membranes were then washed with TBST three times. Western Lightning® Plus‐ECL reagent (PerkinElmer) was used for chemiluminescence detection using a gel imaging system (LuminoGraph II, Atto).

4.10. Statistical analysis and data visualization

For statistical analysis, Dunnett's, Tukey–Kramer, and Student's t‐tests were performed using the lawstat (for Student's t‐test) and multcomp (for Dunnett's and Tukey–Kramer tests) package in R software (version 4.2.3; https://www.r-project.org/). Line graphs and box plots were drawn using R.

4.11. Amino acid alignment

The amino acid alignments in Figure S2 were aligned using MAFFT version 7 on the web (https://mafft.cbrc.jp/alignment/server/index.html), employing L‐ins‐I method. The accession numbers used were: Homo sapiens, CCDS12922.1; Mus musculus, CCDS20743.2; Phascolarctos cinereus, XP_020859355.1; Ornithorhynchus anatinus, XP_028930021.1; Alligator mississippiensis, XP_014459640.1; Chrysemys picta, XP_023969116.1; Anolis carolinensis, XP_016851995.1; Gallus gallus, XP_040550951.1.

AUTHOR CONTRIBUTIONS

Y.S. and H.K. conceived the study. M.N. and A.A. performed experiments. M.N. analyzed the data. M.N., Y.S., and H.K. wrote the manuscript.

Supporting information

Supplementary Figure S1. Regions required for heterochromatin localization within SUV420H2 heterochromatic target domain (HTM) dependent on H3K9me3. (A) Schematic representation of SUV420H2 domains. (B) Immunofluorescence analysis of cells expressing HTM‐sfGFP. HeLa, NIH3T3, and A9 cells expressing HTM‐sfGFP were fixed, stained with an H3K9me3‐specific antibody, and counterstained with Hoechst 33342. Single confocal sections are shown. (C) Redistribution of HTM‐sfGFP upon H3K9 demethylation induced by Halo‐KDM4D expression in A9 cells. Cells stably expressing HTM‐sfGFP were transfected with Halo‐KDM4D and the next day cells were stained with JF646 HaloTag ligand and Hoechst 33342, and then confocal images were collected. (D) Localization of SUV420H2 HTM sub‐fragments. (left) A schematic illustration of the deletion mutants is shown. (middle) Localization patterns of the mutants tagged with sfGFP. Both the high‐ and low‐power views are indicated. (right) Heterochromatin accumulation ratios of the deletion mutants are presented as box plots, indicating the top and bottom 25% with median (thick bar with values) and average (x). The numbers of analyzed nuclei are indicated on the right. Plots for HTM (347–435), HTM‐N (352–380), and HTM‐C (389–410) are reproductions of those in Figure 1c. Scale bars: 10 μm.

Supplementary Figure S2. Amino acid sequence alignments. The alignments were generated using MAFFT with L‐ins‐I protocol. Black, dark gray, and light gray indicate 100%, >80%, and >60% conservation, respectively. Similar groups are considered when counting conserved amino acids. Green‐ and blue‐shaded boxes highlight the regions of HTM‐N and HTM‐C, respectively. (A) Conserved residues among mammals including Homo sapiens, Mus musculus, Phascolarctos cinereus, and Ornithorhynchus anatinus. (B) Conserved residues among amniota, including Homo sapiens, Mus musculus, Phascolarctos cinereus, Ornithorhynchus anatinus, Alligator mississippiensis, Chrysemys picta, Anolis carolinensis, and Gallus gallus. Red and blue asterisks indicate the conserved amino acid residues in PXVXL‐like and zinc finger‐like motifs, respectively. The HTM‐N region is conserved throughout amniota, whereas the HTM‐C region is conserved only among mammals.

Supplementary Figure S3. Effects of SUV420H2 HTM expression on HP1β accumulation in H3K9me3‐enriched heterochromatin. (A) Views containing multiple nuclei corresponding to Figure 4a. Enlarged images of nuclei boxed in yellow (sfGFP‐tagged protein‐expressed) and white (non‐expressed) are shown in Figure 4a. (B and C) H3K9me3 levels were not altered by the expression of HTM and its mutants. (B) H3K9me3 intensities (arbitrary units; a.u.) in individual nuclei from a single experiment with the number of nuclei (N) analyzed (top) and mean intensities from single experiments in biological triplicates (bottom) are plotted. p‐values obtained from Student's t‐test (paired, two‐tailed) are indicated. (C) Scatter plots representing the relationship between H3K9me3 and HP1β intensities, showing no correlation. The number of analyzed nuclei (N) is indicated on the right. Scale bar: 10 μm.

Supplementary Figure S4. Overexpression of HTM‐N‐sfGFP results in the increased intensity of all HP1 isoforms. HeLa cells expressing HTM‐N‐sfGFP were co‐plated with parental HeLa cells for immunostaining with antibodies specific to HP1α, HP1β, and HP1γ. DNA was counterstained with Hoechst 33342. (A) Single confocal sections are shown. (B) Boxplots display the mean intensities of HP1 signals with the number of analyzed nuclei (N). Scale bar: 10 μm.

Supplementary Figure S5. Validation of SUV420H1/H2 double knockout (DKO) cell lines. Levels of H4K20me1, me2, and me3 in parental and SUV420H1/H2 DKO HeLa cells were analyzed by immunofluorescence using specific antibodies conjugated with different fluorescent dyes. (A) Single confocal sections are shown. (B) Boxplots indicate the mean intensities (arbitrary units; a.u.) in individual nuclei. The number of analyzed nuclei is indicated at the bottom. Compared to parental HeLa cells, both H4K20me2 and H4K20me3 levels were diminished to background levels in DKO clones, while H4K20me1 levels were increased. (C) Validation of genome editing by CRISPR/Cas9. Targeted regions for genome editing (exon 1 and exon 8 in SUV420H1, and exon 1 and exon 4 in SUV420H2) in expressed alleles were amplified by RT‐PCR for direct sequencing. The nucleotide sequences with Sanger sequencing profiles are shown. Clone #20 contained a single‐nucleotide insertion near the start codon (exon 1) and a three‐nucleotide deletion in the catalytic domain (exon 8) in SUV420H1, as well as a large deletion near the start codon (exon 1) in SUV420H4. Clones B6, C9, and D2 had the common mutations found in the parental clone #20, and each had an additional deletion in SUV420H2 catalytic domain (exon 4). Scale bar: 10 μm.

Supplementary Figure S6. Effects of SUV420H1/H2 DKO and HTM expression in DKO cells on HP1β accumulation in H3K9me3‐enriched heterochromatin. (A) Views containing multiple nuclei corresponding to Figure 5a. Enlarged images of nuclei boxed in yellow are shown in Figure 5a. (B) Views containing multiple nuclei corresponding to Figure 5c. Enlarged images of nuclei boxed in yellow (sfGFP‐tagged protein‐expressed) and white (non‐expressed) are shown in Figure 5c. Scale bars: 10 μm.

Supplementary Figure S7. Panels of cells expressing HTM‐sfGFP, HTM V374D‐sfGFP, sfGFP‐HP1α, and HP1α‐sfGFP. (A and B) Mouse A9 cells transiently expressing HTM‐sfGFP, HTM V374D‐sfGFP, sfGFP‐HP1α, and HP1α‐sfGFP. DNA was counterstained with Hoechst 33342. (A) Views containing multiple nuclei. (B) Enlarged views of single nuclei. (C) Views containing multiple HeLa cell nuclei expressing HTM‐sfGFP, HTM V374D‐sfGFP, sfGFP‐HP1α, and HP1α‐sfGFP, along with H2B‐Halo, corresponding to Figure 6a. Enlarged images of nuclei boxed in white are shown in Figure 6a. Scale bars: 10 μm.

Supplementary Figure S8. Whole membranes of western blotting. The entire membranes from western blotting experiments shown in Figure 2a (A) and Figure 2b (B) are displayed. The positions of the size standards are indicated.

Table S1. The list of primers used in this study.

Table S2. The list of antibodies and dyes used in this study.

Movie 1. This movie showcases a cell expressing HTM V374D‐sfGFP, mCherry‐PCNA, and H2B‐Halo. It features single confocal images that were acquired at 5 min intervals.

Movie 2. This movie presents enlarged views of a chromocenter in a cell expressing HTM V374D‐sfGFP, mCherry‐PCNA, and H2B‐Halo. It focuses on detailed views of a chromocenter from cells shown in Movie 1.

Nakao, M. , Sato, Y. , Aizawa, A. , & Kimura, H. (2024). Mode of SUV420H2 heterochromatin localization through multiple HP1 binding motifs in the heterochromatic targeting module. Genes to Cells, 29(5), 361–379. 10.1111/gtc.13109