Heterochromatin protein 1 alpha (HP1α) undergoes a monomer to dimer transition that opens and compacts live cell genome architecture

Abstract

Our understanding of heterochromatin nanostructure and its capacity to mediate gene silencing in a living cell has been prevented by the diffraction limit of optical microscopy. Thus, here to overcome this technical hurdle, and directly measure the nucleosome arrangement that underpins this dense chromatin state, we coupled fluorescence lifetime imaging microscopy (FLIM) of Förster resonance energy transfer (FRET) between histones core to the nucleosome, with molecular editing of heterochromatin protein 1 alpha (HP1α). Intriguingly, this super-resolved readout of nanoscale chromatin structure, alongside fluorescence fluctuation spectroscopy (FFS) and FLIM-FRET analysis of HP1α protein-protein interaction, revealed nucleosome arrangement to be differentially regulated by HP1α oligomeric state. Specifically, we found HP1α monomers to impart a previously undescribed global nucleosome spacing throughout genome architecture that is mediated by trimethylation on lysine 9 of histone H3 (H3K9me3) and locally reduced upon HP1α dimerisation. Collectively, these results demonstrate HP1α to impart a dual action on chromatin that increases the dynamic range of nucleosome proximity. We anticipate that this finding will have important implications for our understanding of how live cell heterochromatin structure regulates genome function.

Graphical Abstract

Figure 1.

The majority of HP1α exists in a monomeric state in live cells. (A) Intensity images of the eGFP signal throughout a selected HeLa^{eGFP-HP1α+KD} versus HeLa^{eGFP-HP1αI165E+KD} cell (A, top) and the region of interest (ROI) from which each NB data acquisition was recorded (A, bottom). Scale bars 2 μm. (B) Intensity versus brightness scatterplot of the eGFP-HP1α NB data acquisition presented in (A) with the calibrated brightness windows superimposed (Supplementary Figure 6a–c). (C) Quantification of the fractional contribution of HP1α monomer (teal), dimer (green) and oligomer (red) in the NB data acquisitions presented in (A). (D) Brightness maps of the NB data acquisitions presented in (A) pseudo-coloured according to the brightness windows defined in (B) spatially map HP1α monomer (teal), dimer (green) and oligomer (red) localisation. (E–G) Quantification of the fraction of pixels containing monomer, dimer, and oligomer in HeLa^{eGFP-HP1α+KD} versus HeLa^{eGFP-HP1αI165E+KD} (N = 14 cells, two biological replicates). (H) Intensity image of the eGFP signal throughout a selected HeLa^{eGFP-HP1α+KD} cell (H, left) and representative ROIs from which single point FCS acquisitions were recorded (i.e. foci versus nucleoplasm) (H, right). (I) ACF profiles that result from temporal correlation of fluctuations in eGFP-HP1α fluorescence intensity within ROIs containing foci versus nucleoplasm that exhibit an amplitude at τ = 0 (i.e. G(0)) that is inversely proportional to the local number of molecules present (I, left) and enables quantification of the average eGFP-HP1α concentration in these two distinct environments (I, right) (N = 9 measurements, 4–5 cells, one biological replicate). (J) Quantification of the fraction of pixels containing eGFP-HP1α dimer and oligomer within ROIs containing foci versus nucleoplasm in the data underpinning (E–G). (K) Quantification of the nuclear wide concentration weighted fraction of eGFP-HP1α self-association in the data underpinning (E–G). The box and whisker plots in (E-G), (I) and (J) show the minimum, maximum, sample mean: * P < 0.05**** P < 0.0001, un-paired t-test.

Figure 2.

HP1α dimers induce sub-micron heterogeneity throughout a homogenous chromatin architecture that is maintained by HP1α monomers. (A) Intensity images of the Hoechst 33342 signal (i.e. DNA density) throughout the nuclei of HeLa (left), Hela^KD (middle), and HeLa^{HP1αI165E+KD} (right). Scale bar 10 μm. (B) Coefficient of variation (CV) analysis quantifies the heterogeneity in DNA density present throughout a Hoechst 33342 intensity image via calculation of the pixel intensity variance across the whole nucleus normalised to the pixel intensity mean (CV_index). (C) Quantification of the CV_index across multiple HeLa, HeLa^KD and HeLa^{HP1αI165E+KD} nuclei (N ≥ 30 cells, two biological replicates). (D) Hi-C quantification of the frequency of short to long range chromatin interactions (0.1–100 Mb) throughout two replicates of Hela^KD and HeLa^{HP1αI165E+KD} (P_sample) relative to HeLa (P_HeLa1). (E) DNA fragmentation product (2 μg in each case) from MNase digestion of HeLa, HeLa^KD, and HeLa^{HP1αI165E+KD} chromatin for an increasing amount of time (0, 2, 6 and 12 min at 0.5-unit MNase) (E, top panel), alongside quantification of the mean fluorescence intensity of the digested DNA fragments under the 12 min condition (E, bottom panel). The box and whisker plots in (C) show the minimum, maximum, sample mean: ****P< 0.0001, one-way ANOVA.

Figure 3.

HP1α self-association is critical for nuclear wide chromatin network compaction while the HP1α monomeric subunit imparts a baseline level of de-compaction. (A) Merged intensity images of HeLa (A, left), HeLa^KD (A, middle) and HeLa^{HP1αI165E+KD} (A, right) nuclei expressing histone FRET reporter H2B-eGFP and H2B-mCh. Scale bars 5 μm. (B-C) Lifetime maps (B) of the FLIM data acquisitions presented in (A) pseudo-coloured according to the palette defined in the phasor plot in (C) spatially map open (teal) versus compact chromatin (red). Scale bars 5 μm. The phasor distribution of H2B-eGFP (C) detected throughout the lifetime maps in (B) is superimposed with a theoretical FRET trajectory that enables the efficiency of histone FRET with H2B-mCh (red cursor) to be characterised upon independent determination of the unquenched donor fluorescence lifetime (teal cursor) and cellular autofluorescence (black cursor) (Supplementary Figure S9a–d). (D) Quantification of the fraction of pixels exhibiting FRET (i.e. a compact chromatin state) across multiple HeLa, HeLa^KD and HeLa^{HP1αI165E+KD} nuclei (N ≥ 8 cells, two biological replicates.). (E) Intensity images of a HeLa nucleus co-expressing H2B-eGFP and H2B-mCh (E, left) with immunofluorescence (IF) against endogenous HP1α and histone modification H3K9me3 labelled with Alexa Fluorophore 647 (AF647) (E, middle) and AF405 (E, right) (respectively). Scale bars 5 μm. (F) Lifetime map (left) of the FLIM data acquisition presented in (E) pseudo-coloured according to no FRET (teal) versus FRET (red) alongside masks based on the HP1α-AF647 (F, middle) and H3K9me3-AF405 (F, right) intensity images presented in (E). Scale bars 5 μm. (G-H) Quantification of the fraction of histone FRET (compact chromatin) inside versus outside HP1α-AF647 (G) and H3K9me3-AF405 (H) intensity masks derived from multiple cells (N = 15 cells, two biological replicates). The box and whisker plot in (D) shows the minimum, maximum, sample mean: *P< 0.05, ***P< 0.001, one-way ANOVA. In (G-H): ns P> 0.05, ***P< 0.001, paired t-test.

Figure 4.

HP1α monomers bind chromatin via the SUV39H1 mediated H3K9me3 histone modification. (A) Merged intensity images of eGFP-HP1α_I165E in the absence (A, left) versus presence (A, middle) (FRET experiment) of H2B-mCh in HeLa^{eGFP-HP1αI165E+KD}nuclei and eGFP-HP1α in the presence of H2B-mCh in a HeLa^{eGFP-HP1α+KD} nucleus (FRET experiment) (A, right) (top row) with in each case a selected region of interest (ROI) for FLIM data acquisition (A, bottom row). Scale bars 2 μm. (B-C) Lifetime maps of eGFP-HP1α_I165E versus eGFP-HP1α throughout the FLIM data acquisitions selected in (A) pseudo-coloured according to the palette defined in the phasor plot in (C) spatially map chromatin binding interaction (red pixels). Scale bars 2 μm. The phasor distribution of eGFP-HP1α_I165E detected throughout the lifetime maps in (B) is superimposed with a theoretical FRET trajectory that enables the efficiency of FRET with H2B-mCh (red cursor) to be characterised, given our independent determination of the unquenched donor fluorescence lifetime (teal cursor) (right column in A-B). (D) Quantification of the fraction of pixels exhibiting FRET (i.e. chromatin binding) across multiple HeLa^{eGFP-HP1αI165E+KD} and HeLa^{eGFP-HP1α+KD} nuclei co-expressing H2B-mCh (N ≥ 7 cells, two biological replicates). (E) Merged intensity image of a HeLa^{eGFP-HP1αI165E+KD} nucleus co-expressing SUV39H1-mCh and a selected ROI for cross RICS data acquisition. Scale bar 2 μm. (F) Analysis of the cross RICS data acquisition in (E) results in a cross RICS profile that exhibits an amplitude (G_cc(0,0)) indicative of the fraction of heterocomplex present in the selected ROI. (G) Quantification of the fraction of eGFP-HP1α_I165E in a hetero-complex with the potential binding partners (N = 9 cells, two biological replicates). (H) Quantification of the fraction of eGFP-HP1α mutants in a hetero-complex with the SUV39H1 (N ≥ 9 cells, two biological replicates). (I) Merged intensity images of eGFP-HP1α_I165E (green), H3K9me3 (magenta) and SUV39H1-mCh (red) in HeLa^{HP1αI165E+KD} cells transiently expressing SUV39H1-mCh. Scale bars 10 μm. (J) The intensity plot of SUV39H1-mCh versus H3K9me3 across multiple nuclei (N = 190 cells, one biological replicate). (K) The diffusion coefficient of eGFP-HP1α_I165E in the presence versus absence of SUV39H1-mCh expression (N ≥ 7 cells, two biological replicates). (L) The diffusion coefficient of eGFP-HP1α mutants in the presence of SUV39H1-mCh expression. The box and whisker plots in (D), (G-H) and (K-L) show the minimum, maximum, sample mean: *P< 0.05, ** P< 0.01, ****P< 0.0001, one-way ANOVA.

Figure 5.

Schematic of the HP1α monomer to dimer transition modulating nucleosome proximity. HP1α monomers and dimers (blue) upon binding H3K9me3 modified (red) nucleosomes (yellow), space apart and bridge together nucleosomes to locally regulate DNA template access. This nanoscale organisation in nucleosome proximity is unchanged by HP1α nuclear condensate formation.