To confirm and extend our previous finding of nucleosome-induced condensation of the H1 CTD, we used FRET to determine the distances between labels located at either end of the CTD and when the downstream fluorophore was located at a more interior position. To this end the double H1 mutants G101C/K195C and G101C/T173C were modified with a 50/50 mixture of Cy3-maleimide and Cy5-maleimiade (A) and fluorescence emission spectra measured in the absence or presence of nucleosomes reconstituted on a 207-bp DNA fragment, with excitation at 515 or 610
nm (see ‘Materials and Methods’ section). In accord with previous work (8
), Cy5 fluorescence emission (max ~667
nm) was low when either of the two Cy3/Cy5-labeled H1s were irradiated at 515
nm in the absence of nucleosomes (B and C), consistent with the domain being disordered. The calculated FRET efficiencies of the double-labeled proteins in the absence of nucleosomes (D) combined with an R0
for the Cy3/Cy5 FRET pair of 5.4
nm (see ‘Materials and Methods’ section) indicates a distance between fluorophores attached to G101C and K195C of 8
nm, consistent with the predicted end-to-end distance of an ~100 aa residue worm-like chain of ~9.0
). Likewise, the calculated distance between fluorophores attached to G101C and T173C is ~6.2
nm, consistent with the predicted end-to-end distance of a ~73
aa residue disordered domain of ~6.6
Figure 1. The H1 CTD is intrinsically disordered and condenses upon binding to nucleosomes. (A) Schematics of H1 G0101C/K195C (left) and H1 G101C/T173C (right) modified with Cy3 and Cy5 (red and blue). N, G and C denote the N-terminal, globular and C-terminal domains, (more ...)
We observed significant increases in FRET upon nucleosome binding for both Cy3/Cy5-labeled G101C/K195C and G101C/T173C; emission from Cy3 (~560
nm) was significantly reduced and emission from Cy5 was increased in both cases. Titrations of nucleosomes into the double-labeled H1s show that FRET efficiency increased with nucleosome concentration until reaching a constant value when the ratio of H1: nucleosome was 1:1 or higher (D). Thus the maximum FRET efficiency on such plots corresponds to saturated nucleosome binding by H1 in our bulk assays and is used to calculate the distances between labeled sites within the nucleosome-bound form of H1. Interestingly, Cy3/Cy5-labeled G101C/K195C exhibited a greater maximal efficiency in the nucleosome-bound H1 compared to that obtained with Cy3/Cy5-labeled G101C/T173C. The calculated FRET efficiencies of the double H1 mutants indicate a distance between G101C and K195C of 4.2
nm while the distance between G101C and T173C is 4.9
nm, consistent with extensive condensation of the CTD domain.
H1 binds strongly and cooperatively to naked DNA and can induce aggregation of samples when present in stoichiometric excess over nucleosomes (20
). In such complexes, H1s are expected to be located in close proximity; thus there is a possibility that some fraction of the FRET signal in our nucleosome experiments is due to inter-molecular energy transfer. To further characterize the H1 binding in our assays and to ensure that our assays were reporting intra-molecular FRET, a 50:50 mixture of Cy3- and Cy5-only labeled H1s were incubated with nucleosomes. Importantly, these samples exhibited only very low FRET in the absence of nucleosomes and did not exhibit significant increases in efficiencies upon nucleosome binding (D). These results indicate that the FRET signal observed with the double-labeled protein in the presence of nucleosomes is solely due to intra-molecular energy transfer.
H1s exhibit rapid mobility about the nucleus and encounter DNA in nucleosome free regions, linker DNA regions and DNA surfaces outside of the defined binding site at the nucleosome dyad (21
). Therefore it is important to determine whether H1 binding to double-stranded DNA induces CTD folding similar to that observed when binding to nucleosomes. We found that binding of Cy3/Cy5-labeled H1 G101C/K195C to 207-bp DNA fragments resulted in significant FRET (A). Cooperative binding of H1 to naked DNA forms oligomeric ‘tramtrack’ structures and ultimately large rod-like structures (24
). Such structures would position several H1 molecules in close proximity, increasing opportunity for inter-molecular resonance energy transfer (C). Indeed, titration of DNA fragments into a 1:1 mixture of Cy3-only and Cy5-only labeled H1 G101C/K195C results in significant intermolecular FRET (B). Thus, in contrast to nucleosomes, at least a portion of the FRET observed when Cy3/Cy5-labeled H1 G101C/K195C binds to naked 207-bp DNA fragments is due to inter-molecular energy transfer.
Figure 2. Binding of Cy3/Cy5-labeled H1 G101C/K195C to naked DNA results in both intra- and inter-molecular FRET. (A) Binding of Cy3/Cy5 labeled H1 G101C/K195C to naked DNA induces FRET. The protein was incubated alone (black trace) or with increasing amounts of (more ...)
Since the FRET observed when Cy3/Cy5-labeled H1 G101C/K195C binds to naked DNA fragments is at least partially due to inter-molecular energy transfer, we wished to quantify the intra-molecular FRET component, if any, to determine whether DNA binding induces CTD condensation. We devised a strategy in which the inter-molecular FRET due to close apposition of neighboring H1 molecules is eliminated by diluting the labeled H1 in the sample with unlabeled protein. We hypothesized that as the ratio of unlabeled:labeled molecules increases, the probability of two labeled molecules being close enough in space to allow for inter-molecular FRET will decrease (D). Indeed, in experiments with a 1:1 mix of Cy3- and Cy5-only H1 G101C/K195C, as the ratio of unlabeled H1: labeled H1 increases, the observed (intermolecular only) FRET decreases until reaching background values (E). These data demonstrate that energy transfer between labeled H1s can be decreased and finally be diluted out completely using this method.
We then used the dilution strategy to isolate any intra-molecular FRET caused by folding of H1CTD upon DNA binding. Total FRET was measured for samples containing Cy3- and Cy5-labeled H1 G101C/K195C in the absence and presence of DNA, with increasing ratios of unlabeled: labeled protein. The total observed FRET efficiency in the absence of unlabeled protein was 0.8, significantly higher than that observed for the mix of homogeneously labeled proteins, suggesting an intra-molecular contribution to total FRET. Moreover, upon increasing the ratio of unlabeled: labeled protein, total FRET efficiency decreased, as observed with the mixture of Cy3-H1 and Cy5-H1, until reaching a constant value of 0.33 at 20- to 30-fold dilution (E). As we have established that at high dilutions, all inter-molecular FRET has been reduced to 0, this value represents the intrinsic amount of intra-molecular FRET for the H1 CTD within the H1–DNA complex.
Since we observed significant inter-molecular FRET, we next wondered whether the specific labeled sites on the CTD occupied distinct average distances from each other in the cooperatively formed tramtrack structures. To this end, we determined FRET for combinations of singly labeled H1s upon DNA binding (). Our results indicate that the labeled sites in all three combinations are within FRET distance from one another in H1–DNA complex and thus all contribute to the observed inter-molecular FRET. Interestingly, since FRET efficiency as indicated by emission by Cy5 (λ max 677
nm, red arrows) is inversely related to fluorophore distance, the shortest distance appears to be between G101C’s on adjacent H1s (A). In contrast, distances between G101C-K195C and K195C-K195C are somewhat greater (B and C). These results suggest that H1s are at least partially arranged in a regular spatial relationship within the tramtrack/rod-like structures.
Figure 3. Intermolecular FRET between combinations of labeled sites in H1 upon binding to naked DNA. Pairs of the single substitution mutants H1 G101C and H1 K195C labeled with either Cy3 or Cy5 were mixed in a 1:1 ratio and emission spectra recorded before (black (more ...)
We noticed in our fluorescence studies different extents of fluorophore quenching upon H1-binding nucleosomes and DNA fragments. Since quenching is dependent upon specific aspects of the fluorophore environment, including proximity to DNA, we further explored the extent of quenching at individual sites in H1 upon nucleosome binding. To simplify the analysis in these experiments we focused on Cy5-labeled proteins since this fluorophore appeared somewhat more susceptible to quenching in our experiments than Cy3 and both exhibited the same relative changes (results not shown). Interestingly, upon binding of Cy5-labeled H1 G101C or H1 K195C to nucleosomes, much greater quenching was observed when the fluorophore was attached to position 195 in H1 compared to position 101 (A and B). A plot of the extent of quenching observed upon titration of H1 G101C-Cy5 (C) with nucleosomes shows a peak of quenching at lower nucleosome-H1 ratios that was reduced when the H1-nucleosome ratio was 1:1 or greater. We hypothesize that at low nucleosome: H1 ratios, H1 binds nucleosomes at the canonical dyad site but also interacts with nucleosomes in non-specific modes, which contribute more greatly to quenching. At stoichiometric ratios of nucleosomes:H1, where the vast majority of H1 is bound at the dyad site, quenching of the fluorophore at 195 remains high (~0.6) while quenching at 101 is greatly reduced (~0.2). This indicates that the environment surrounding residue 195 is distinct from that surrounding residue 101 when H1 is bound to the nucleosome.
Figure 4. Nucleosome-induced fluorophore quenching is dependent on attachment site. (A) Emission spectra of Cy5-modified H1 G101C were recorded in the absence and presence of increasing concentrations of nucleosomes, as indicated. (B) As in A except Cy5-H1 K195C (more ...)
To further investigate the nature of this difference, we examined quenching due to DNA binding by H1 at residues 101 and 195. H1 binds in both cooperative and non-cooperative modes to naked DNA. At the salt concentrations used in our binding studies (50
mM NaCl), the cooperative binding mode is favored, while in buffers of lower ionic strength, a non-cooperative mode is favored (24
). Moreover, at high DNA concentrations non-cooperative, distributive binding is expected to increase. We observed that incubation of H1 G101C-Cy5 in the absence or presence of increasing amounts of the 207-bp DNA fragment in 50
mM NaCl resulted in higher fluorescence quenching compared to the same protein incubated with DNA in 20
mM NaCl (A). Similarly, when the fluorophore is attached to residue 195, quenching is much higher over the entire range of DNA concentrations in 50
mM NaCl, and compared to that in 20
mM NaCl, with the latter reaching a final value of 0.2 (B). This behavior is consistent with the cooperative binding mode being favored at 50
mM NaCl and at lower DNA concentrations, leading to fluorescence quenching of ~0.5. In contrast, the non-cooperative mode is favored at high DNA concentrations and 20
mM NaCl, and yields a quenching factor of ~0.2.
Figure 5. Quenching due to H1 binding naked DNA. (A) Quenching of Cy5-H1 G101C fluorescence depends on binding mode. The protein was incubated with increasing amounts of 207bp naked DNA fragment and the fraction of Cy5 fluorescence quenched determined (more ...)