Single-molecule study reveals Hmo1, not Hho1, promotes chromatin assembly in budding yeast

ABSTRACT Linker histone H1 plays a crucial role in various biological processes, including nucleosome stabilization, high-order chromatin structure organization, gene expression, and epigenetic regulation in eukaryotic cells. Unlike higher eukaryotes, little about the linker histone in Saccharomyces cerevisiae is known. Hho1 and Hmo1 are two long-standing controversial histone H1 candidates in budding yeast. In this study, we directly observed at the single-molecule level that Hmo1, but not Hho1, is involved in chromatin assembly in the yeast nucleoplasmic extracts (YNPE), which can replicate the physiological condition of the yeast nucleus. The presence of Hmo1 facilitates the assembly of nucleosomes on DNA in YNPE, as revealed by single-molecule force spectroscopy. Further single-molecule analysis showed that the lysine-rich C-terminal domain (CTD) of Hmo1 is essential for the function of chromatin compaction, while the second globular domain at the C-terminus of Hho1 impairs its ability. In addition, Hmo1, but not Hho1, forms condensates with double-stranded DNA via reversible phase separation. The phosphorylation fluctuation of Hmo1 coincides with metazoan H1 during the cell cycle. Our data suggest that Hmo1, but not Hho1, possesses some functionality similar to that of linker histone in Saccharomyces cerevisiae, even though some properties of Hmo1 differ from those of a canonical linker histone H1. Our study provides clues for the linker histone H1 in budding yeast and provides insights into the evolution and diversity of histone H1 across eukaryotes. IMPORTANCE There has been a long-standing debate regarding the identity of linker histone H1 in budding yeast. To address this issue, we utilized YNPE, which accurately replicate the physiological conditions in yeast nuclei, in combination with total internal reflection fluorescence microscopy and magnetic tweezers. Our findings demonstrated that Hmo1, rather than Hho1, is responsible for chromatin assembly in budding yeast. Additionally, we found that Hmo1 shares certain characteristics with histone H1, including phase separation and phosphorylation fluctuations throughout the cell cycle. Furthermore, we discovered that the lysine-rich domain of Hho1 is buried by its second globular domain at the C-terminus, resulting in the loss of function that is similar to histone H1. Our study provides compelling evidence to suggest that Hmo1 shares linker histone H1 function in budding yeast and contributes to our understanding of the evolution of linker histone H1 across eukaryotes.

In this study, we utilized the YNPE to simulate the physiological conditions in yeast nuclei, combined with single-molecule approaches, to explore the functions of Hho1 and Hmo1 in chromatin assembly. We demonstrated that Hmo1, but not Hho1, involves in chromatin assembly in YNPE. The lysine-rich tail of Hmo1 is essential for its function in chromatin assembly. Moreover, we showed that Hmo1 is similar to metazoan histone H1 in terms of phase-separation property and phosphorylation during the cell cycle.

Hmo1, not Hho1, affects chromatin assembly in YNPE
YNPE, which contains all soluble proteins in the yeast nucleus, has been reported to support nucleosome formation and chromatin assembly under a simulated physiological condition (56,57). One end of fluorescently stained λDNA was tethered on the surface of the flow cell via biotin-streptavidin interaction. With the Poiseuille flow of blocking buffer at a constant rate of 50 µL/min, the average extended length of the λDNA molecules was 11.86 ± 2.22 µm (n = 300), which was about 72% of the theoretical λDNA length (~16.49 µm) ( Fig. 1A and B top panel; Fig. S1A). After the original YNPE (protein concentration was around 4.93 mg/mL) extracted from the wild-type yeast strains was introduced at a rate of 50 µL/min, DNA molecules retracted sharply and eventually shrunk into quasi-spherical dots in less than 10 seconds, which represented the biological process of chromatin assembly as described before (52,53,56). To ensure that the compression rate of the λDNA was within a suitable range for easy observa tion and improved the signal-to-noise ratio, we tested 1:200 and 1:2,000 dilutions of Research Article mBio wild-type YNPE (protein concentration of 25 µg/mL and 2.5 µg/mL) (Fig. S1B). Using the 1:2,000 diluted wild-type YNPE (protein concentration of 2.5 µg/mL), the λDNA molecule retracted from the initial length to a dot (<1.5 µm in diameter) in around 400 seconds ( Fig. 1B and C; Fig. S1B). In subsequent single-molecule experiments, all YNPEs were adjusted to the protein concentration of 2.5 µg/mL. To investigate the effects of Hmo1 and Hho1 on chromatin assembly, we constructed hmo1Δ and hho1Δ strains, respectively. The DNA compaction behavior was monitored in YNPE prepared from these two strains. As shown in Fig. 1B and C, after immersion in the Hho1-deficient YNPE, λDNA shrunk quickly and eventually compacted into a dot, the same as what happened in the wild-type YNPE. The packing ratio (the initial DNA length divided by the packaged DNA length) was over 10. On the contrary, in the Hmo1-deficient YNPE, λDNA was incompletely compressed, only retracting to 70% of its original length with a packing ratio of less than 1.5, similar to what happened in the H1.8 depletion in the Xenopus egg extracts (55). Likewise, we observed the DNA compaction process in Hho1&Hmo1-deficient YNPE. DNA was packed in the Hho1&Hmo1-deficient YNPE with similar kinetics to the Hmo1-deficient YNPE ( Fig. 1B and C). The results strongly suggested that Hmo1, but not Hho1, is essential for the chromatin assembly in YNPE.

Hmo1 directly participates in chromatin assembly in YNPE
Hmo1 has been identified as a transcription factor in S. cerevisiae (58). Knocking out HMO1 affects the expression of a range of genes. Therefore, it remained the possibility that the effect of Hmo1 on chromatin assembly in YNPE was due to its regulatory effect on gene expression rather than direct involvement in chromatin assembly. To rule out this possibility, the purified protein Hmo1 was supplemented into the Hmo1-deficient YNPE. When purified Hmo1 was added to the YNPE hmo1Δ at a final concentration of 8.2 nM, the λDNA molecule resumed further retraction and finally contracted into a dot with kinetics similar to the wild-type YNPE [ Fig. 2A, YNPE hmo1Δ + Hmo1 (light blue curve), YNPE wt (dark blue curve)]. Moreover, the DNA retraction rate increased with the increasing concentration of Hmo1 (Fig. S2). When introducing 8.2 nM human linker histone H1.4 to the YNPE hmo1Δ , the compaction of λDNA was recovered in a faster kinetic manner compared to Hmo1 ( Fig. 2A, gray curve). Given that both Hmo1 and H1.4 have been reported to have the ability to bend and bridge dsDNA in vitro (47,59), we further investigated the effect of bacterial HU protein (derived from Listeria monocytogenes) on DNA compaction in YNPE hmo1Δ . Like Hmo1 and H1.4, HU can bridge and loop dsDNA in vitro, as demonstrated by the single-molecule magnetic tweezers experiments (Fig. S6). However, LmHU apparently does not partake in the chromatin assembly of the yeast cells. As shown by the orange curve in Fig. 2B, even adding 0.1 µM purified HU protein to YNPE hmo1Δ failed to induce further DNA compaction. This finding excluded the possibility that further DNA compaction in YNPE hmo1Δ was only due to the supplemented protein's capacity to loop DNA or alter DNA elasticity. Likewise, supplementation with 8.2 nM purified protein Hho1 to YNPE hmo1Δ did not restore the DNA compression ( Fig. 2A, red curve).
Furthermore, when the purified proteins Hmo1, Hho1, H1.4, and HU were introduced into the flow cell alone rather than with YNPE, they all showed the capability to compress λDNA, but with different DNA compaction dynamics from adding these proteins in YNPE hmo1Δ . When 8.2 nM of purified Hmo1 was directly pumped into the flow cell, the DNA molecule was compressed rapidly with faster dynamics than that observed in YNPE hmo1Δ supplemented with 8.2 nM Hmo1 (Fig. 2B, blue curve, and Fig. 2C). Interest ingly, when introducing 8.2 nM of purified Hho1 directly into the flow cell, the DNA molecule was retracted to a dot within 60 seconds, although adding Hho1 in YNPE hmo1Δ to the final concentration of 8.2 nM showed no effect on DNA compaction for a long time (compare Fig. 2A and B, red curves, and Fig. 2C). These observations provided additional evidence that the DNA compaction observed in YNPE hmo1Δ by adding Hmo1 was not due to the simple protein-DNA interaction, protein aggregation, or DNA elasticity change. The chromatin assembly in YNPE seems to be a tightly regulated process similar to what occurred in vivo, and Hmo1 directly participates in this process.

Hmo1 promotes the transition of nucleosomal arrays to a more compact conformation
Next, we employed TIRFM to visualize the fluorescent nucleosomes bound to the individual λDNA molecules in Hmo1-deficient YNPE. In budding yeast, the core histone H2B is encoded by two genes, HTB1 and HTB2. We tagged these two genes with 3×FLAG tags at the C-terminus on their native chromosomal loci in the hmo1Δ strain to construct the hmo1Δ-H2B-3FLAG strain and extract its YNPE (Fig. S3A). On injection of the YNPE hmo1Δ-H2B-3FLAG into the flow cell, the length of λDNA molecules reduced to 70% of the original length and remained unchanged, in line with what we observed before. We then labeled the nucleosomes in situ with fluorescent anti-FLAG antibod ies and visualized them on TIRFM. By alternate excitation with 532 nm and 640 nm lasers, the fluorescent signals of nucleosomes on DNA were imaged. As the fluid flows, the fluorescent nucleosome can be observed to fluctuate along with the dsDNA. We observed one to four fluorescent signals on a single DNA molecule ( Fig. 2D; Fig. S3B). The majority of DNA contained one or two fluorescent signals (accounting for 64% and Research Article mBio 22%, respectively), while only a small percentage of DNA (10% and 4%) contained three or four fluorescent signals. Unfortunately, due to resolution limitations, we could not determine the exact number of nucleosomes in each fluorescent signal. Time courses confirmed that ≥99% of the fluorescent nucleosomes remained stably bound to the DNA for up to 20 minutes without dissociating or moving along the DNA. Furthermore, most fluorescent signals remained stably bound to DNA after washing with 0.5 to 1 M NaCl, ruling out the possibility of nonspecific bound florescent H2B (Fig. S3C). Subsequently, we supplemented YNPE hmo1Δ-H2B-3FLAG with 8.2 nM Hmo1. As shown in Fig. 2D, λDNA was further compressed into a dot where the fluorescence of DNA and nucleosomes were co-localized. These results suggested that Hmo1 assisted the nucleosomal array in condensing into a more compact structure (Fig. 2D). In other words, the observations implied that the nucleosomes could not further fold into a higher compact form in the absence of Hmo1. It is well known that facilitating nucleosomes to compact into a higher-order structure is an exact function of linker histone H1 in metazoan (60).

Single-molecule force spectroscopy suggests that Hmo1 facilitates nucleo some assembly in YNPE
To further investigate the role of Hmo1 in chromatin assembly in YNPE, we adopted single-molecule magnetic tweezers experiments. A 1,491-bp dsDNA substrate was designed and labeled with three digoxigenin molecules and biotin, and then attached to the flow cell surface by one end. YNPE wt , YNPE hho1Δ , and YNPE hmo1Δ diluted in blocking buffer were separately incubated with dsDNA for about 20 minutes at room temperature at extremely low force (<0.1 pN) to reconstitute chromatin (Fig. 3A). After the magnetic beads were attached, we stretched the chromatin formed in various YNPE with a continuously increasing force to track the conformational changes. During the stepwise disassembly events, we observed the disruption length of approximately 25 nm in both YNPE wt (Fig. 3B, red curve) and YNPE hho1Δ (Fig. 3B, orange curve) in a single disruption event ( Fig. 3B and C). The jump of 25 nm (corresponding to ~73.5 base pairs) is a typical event during nucleosome disassembly under mechanical tension, which indicates the unraveling of one DNA wrap from the histone octamer (61)(62)(63)(64)(65). In the force-extension curves of YNPE wt and YNPE hho1Δ , a jump of integer multiples of 25 nm, such as 50 nm (~147 bp) or 75 nm (~220.5 bp) or 100 nm (~294 bp), occasionally occurred in one single disruption event, indicating the simultaneous rupture of several nucleosomal DNA wraps ( Fig. 3B; Fig. S4A). Up to 3-4 nucleosomes were observed assembled on 1,500-bp dsDNA by YNPE wt or YNPE hho1Δ , which is approximately half of the nucleosomes that can theoretically be accommodated on the 1,500-bp dsDNA (~up to 9 nucleosomes). The distribution of forces involved in the observed jumps was very broad (Fig. S4B). In contrast, no such extension jumps were observed in YNPE hmo1Δ (Fig.  3B, blue curve) throughout the entire stretching process. However, its force-extension curve deviated from the curve of bare DNA (Fig. 3B, blue curve and violet curve). This observed phenomenon could be attributed to no canonically assembled nucleosomes on the 1,500-bp dsDNA in Hmo1-deficient YNPE. Despite this, various proteins, including histones, might electrostatically interact with the dsDNA, causing the resulting curve ( Fig.  3B, blue curve) to deviate from that of bare dsDNA (Fig. 3B, violet curve). When the force exceeds 30 pN, all the force-extension curves approached that of naked 1,500-bp dsDNA and that of dsDNA fitted by the worm-like chain (WLC) model (Fig. 3B, violet curve and black dashed curve), implying that all dsDNA unwrapped the histones.
To further verify the role of Hmo1 in chromatin assembly, we performed the magnetic tweezers experiments in YNPE hmo1Δ by supplementing with purified Hmo1 and H1.4, respectively. As shown in Fig. 3B; Fig. S5, supplementing Hmo1 or H1.4 in YNPE hmo1Δ resulted in force-extension curves resembling that of wild-type YNPE. The jumps representing the disruption of nucleosome wraps were clearly observed in forceextension curves (Fig. 3B, olive curve, and Fig. S5, dark yellow curve), implying the supplementation of Hmo1 or H1.4 effectively restored the nucleosome assembly in YNPE hmo1Δ , comparable to what occurred in the wild-type YNPE. In a single disruption event, the disruption length in YNPE was usually 25 nm (~73.5 bp) or multiples thereof, which indicated the rupture of the nucleosomal DNA wraps. However, the jumps formed by the single purified protein were variable and much larger in size (200 to 300 nm, corresponding to ~588-882 bp) and usually occurred only once at low forces ( Fig. S4 and S6). The control experiments revealed distinct differences between unraveling the nucleosome formed by YNPE and disrupting the DNA loop formed by purified protein. This is consistent with our previous observations in the single-molecule fluorescence Research Article mBio experiments (Fig. 2). In conclusion, combined with the observations on TIRFM before, it suggested that Hmo1 can facilitate nucleosome assembly in the YNPE system.

Lysine-rich extension is essential for the chromatin assembly function of Hmo1
For human linker histone H1, it has been reported that the lysine-rich CTD plays an essential role in chromatin binding and condensation in vivo (20,66,67). Amino acid sequence analysis showed that both human H1.4 and Hmo1 possess a lysine-rich CTD.
On the contrary, the lysine-rich region of Hho1 is sandwiched between its two globu lar domains (Fig. 4A). To determine the possible function of the lysine-rich region in chromatin assembly, Hmo1-AB, a truncated mutant of Hmo1 which lacked the lysinerich CTD, was expressed and purified. Meanwhile, we also expressed and purified a Hho1 mutant, Hho1(1-176), in which the second globular domain was deleted, and the lysine-rich region turned to be the C-terminus (Fig. 4A). Unlike Hmo1, when 8.2 nM Hmo1-AB was introduced, chromatin assembly was not restored in YNPE hmo1Δ . The length of the λDNA molecules remained at 70% of the original length within 300 seconds, which was similar to the results of YNPE hmo1Δ and YNPE hmo1Δ + Hho1 (Fig.  4B). Interestingly, when 8.2 nM Hho1(1-176) was supplemented in YNPE hmo1Δ , the λDNA molecules were compacted from 70% of the original length to a dot at a much faster rate (about 10 seconds) (Fig. 4B). The results indicated that the lysine-rich CTD is required for the chromatin assembly function of Hmo1. For Hho1, it seems that the second globular domain buries the lysine-rich region so as to hinder its function as linker histone H1.

Hmo1 possesses phase separation properties when mixed with dsDNA
Linker histone H1 regulates the dynamic organization of chromatin by promoting liquid-liquid phase separation (LLPS), and its highly disordered lysine-rich CTD is essential for this process (18,19,33,68). Since Hmo1 possesses a similarly long lysine-rich CTD which was predicted to be disordered (Fig. S7A), we examined whether Hmo1 also induced LLPS. Similar to human linker histone H1.4, Hmo1 alone did not form condensates. When mixed with 300-bp dsDNA, the Hmo1-DNA mixture readily formed many conden sates. Unlike the spherical droplets formed by H1.4-DNA, the condensates formed by

Research Article mBio
Hmo1-DNA were irregular and gel-like. The condensates were dissolved in the pres ence of high salt concentration, showing a reversible character ( Fig. 5A; Fig. S7B). The phenomenon is similar to the coarse condensates formed by H1-polynucleosomes (33). In contrast, Hho1 did not form a condensate even when mixed with dsDNA (Fig. 5A).
To determine the mobility of Hmo1 inside the irregular condensates, we monitored the fluorescence recovery after photobleaching (FRAP) (69). In the Hmo1-DNA condensate, significant fluorescence recovery was observed shortly after photobleaching, indicating that Hmo1 was mobile within the condensates (Fig. 5B). However, unlike the droplets formed by H1.4-DNA, which disappeared with the increase of the DNA length, the Hmo1-DNA condensates appeared nearly unaffected by DNA length (Fig. S7B). Meanwhile, we tested the two mutants, Hmo1-AB and Hho1(1-176), for phase separation. Hmo1-AB, which lacked the disordered C-terminus, failed to form conden sates in the presence of dsDNA. However, Hho1(1-176) readily undergone LLPS when mixed with 300-bp dsDNA, similar to the observations in the H1.4-DNA mixture. FRAP experiments showed a highly dynamic diffusion of Hho1(1-176) within the spherical droplet ( Fig. 5A and B). Interestingly, as shown in Fig. S7B, the morphology of the Hho1(1-176)-DNA condensates gradually changed from the initial spherical droplets to large gels as the DNA length increased. After adding a certain amount of DNase, the large gel-like condensates reverted to droplets again (Fig. S7B). Meanwhile, we used the FAM-labeled dsDNA to examine whether DNA was dynamically diffused in the conden sates. As shown in Fig. 5B, the bleached DNA did not undergo recovery in the droplet, suggesting that the dsDNA might have been arrested in it.

Fluctuation in Hmo1 phosphorylation coincides with linker histone H1 during the cell cycle
Linker histone H1 is phosphorylated in a cell cycle-dependent manner and correlates with the open or closed states of chromatin. The extent of H1 phosphorylation progressively increases as the cell progresses from G1 to mitosis and then sharply decreases. Low phosphorylation levels of H1 in the S-phase are associated with an open chromatin conformation that facilitates replication and transcription events. In M-phase, high phosphorylation levels of H1 promote the entry of condensins (7,15,70). Given that many properties of Hmo1 are similar to those of H1, we sought to detect the phosphorylation of Hmo1 during the cell cycle.
The yeast cells were released from the synchronization at the G1 phase and harves ted cells every 10 minutes to extract the whole cell proteins. The cell cycle phase of harvested cells was determined by flow cytometry analysis (Fig. S8). As shown in Fig.  6A and B, the extent of Hmo1 phosphorylation was variable during the cell cycle. At different periods of the cell, Hmo1 was usually present as a mixture of unphosphorylated and phosphorylated forms. The phosphorylation level of Hmo1 was lowest in G1 phase and increased during the S and G2 phases. The highest phosphorylation level was determined at the end of the G2 phase and mitosis when the expression of HMO1 reached the highest level ( Fig. 6A and B). Accordingly, the phosphorylation extent of Hmo1 at different cell cycle stages is consistent with that of linker histone H1. This observation is in agreement with the hypothesis that Hmo1 might function as a linker histone to regulate chromatin structure in S. cerevisiae (22,23).

DISCUSSION
The identity of the histone H1 protein in budding yeast, whether it is Hho1 or Hmo1, has been long debated. In this study, we utilized the single-molecule approach to directly investigate the chromatin assembly in an in vitro biochemical system (YNPE), which can accurately simulate the physiological conditions in the nucleus. The observations revealed that Hmo1, rather than Hho1, plays a crucial role in promoting DNA compaction and facilitating chromatin assembly. Furthermore, we employed magnetic tweezers to demonstrate that Hmo1 can facilitate the formation of nucleosomes in YNPE from the mechanical aspect. Same as metazoan histone H1, Hmo1 exhibits phase separation properties when mixed with dsDNA. Moreover, the extent of Hmo1 phosphorylation at various cell cycle stages is similar to that of linker histone H1. These findings have implications that Hmo1 might take some functions of histone H1 in budding yeast.

Research Article mBio
The more open chromatin conformation may allow the yeast to perform transcriptional events at a fast rate (75,76). Therefore, yeast has long been thought to lack linker histone H1 (77). Hho1 was proposed to be histone H1 of budding yeast based on the amino acid sequence similarity alignment (26). However, the chromatin structure and the sensitivity to nuclease were not altered in the hho1Δ strain (25,34). In this study, we demonstrated the deletion of HHO1 did not affect the chromatin assembly in YNPE (Fig. 1, Fig. 2A, and Fig. 3). It is likely that Hho1 might not function as a linker histone in S. cerevisiae. On the contrary, the observations in single-molecule experiments suggested that Hmo1 is essential for chromatin assembly. Hmo1 is a unique member of the HMGB protein family, which showed the ability to maintain chromatin integrity in genetic studies (22,39,48,78). In the YNPE derived from the hmo1Δ strain, λDNA was only compacted to 70% of its original length and was not further compressed. When supplementing Hmo1 in YNPE hmo1Δ , the compaction of λDNA was restored ( Fig. 1; Fig. 2). A similar phenomenon was observed for supplementing the histone H1.4 but not for bacterial HU protein ( Fig. 2A). In the YNPE hmo1Δ , only a few discrete distributions of fluorescent nucleosomes were observed on λDNA, and no nucleosome was successfully assembled on 1.5 k dsDNA in magnetic tweezers experiments ( Fig. 2D; Fig. 3B). Supplementation of purified Hmo1 rendered the nucleosomal chain into a more compacted structure, and the nucleosome assembled on the 1.5 k dsDNA ( Fig. 2D; Fig. 3B, olive curve). Based on the observations from TIRFM and magnetic tweezers experiments, it is unlikely that this process was simply due to the DNA-looping function of Hmo1 and H1.4 since the DNA-looping protein HU did not restore the further DNA compaction in YNPE hmo1Δ , and there were markable differences between the DNA force-extension curves with or without YNPE (Fig. 2 and 3; Fig. S6). Our data provide evidence that Hmo1, but not Hho1, facilitates chromatin assembly in budding yeast. However, the detailed molecular mechanism of how Hmo1 facilities the chromatin assembly still needs to be discovered. It has been proposed that Hmo1 bound near the dyad axis with its box A; thus, the lysine-rich CTD interacted with linker DNA to form a stem-like structure to prevent DNA bending, similar to linker histone H1 (23). But some studies suggested that Hmo1 strongly bound to the nucleosome and unwound DNA from the core nucleosome to separate linker DNA, especially at a low protein concentra tion (24,79). These observations against the function of Hmo1 acting as histone H1 and explained the function of Hmo1 as an HMGB protein to promote transcription. Interestingly, at a high protein concentration (>3 nM), the presence of Hmo1 significantly compacted nucleosome arrays and shortened nearest neighbor nucleosome spacing (79). In this study, we supplemented the YNPE hmo1Δ with 8.2 nM Hmo1 and observed the restoration of DNA compaction ( Fig. 2A). It was estimated that the average concen tration of Hmo1 was 12.5 µM in yeast nuclei (42,80,81). In the real chromatin assembly process, the local concentration of Hmo1 around the core nucleosome could be even higher. Accordingly, we hypothesized that Hmo1 might bind to the nucleosomes and link adjacent nucleosomes via oligomerization. This hypothesis is supported by the fact that Hmo1 can form higher-order oligomers (82,83) and shows gel-like phase separation properties when mixed with dsDNA (Fig. 5).
We demonstrated that the chromatin assembly function of Hmo1 depended on its lysine-rich CTD (Fig. 4B). Unlike the other HMGB proteins, Hmo1 possesses a unique lysine-rich CTD similar to metazoan histone H1. Hho1 also contains a lysine-rich region, but the lysine-rich region is buried between two globular domains (Fig. 4A). Deletion of the second globular domain exposes the lysine-rich region. Interestingly, similar to histone H1 or Hmo1, the Hho1 mutant (1-176) restored the DNA compaction in YNPE hmo1Δ and showed the phase separation property. It is possible that the budding yeast Hho1 accidentally acquired an extra globular domain by recombination during the evolution, which resulted in the loss of its chromatin assembly function. In this case, the HMGB family protein Hmo1 might perform some of the linker histone functions instead. Certainly, we cannot rule out another possibility that the metazoan histone H1 evolved from the Hho1 by losing the globular domain at the C-terminus.
Currently, accumulating evidence favors the fluid-like model of chromatin structure (18,84). The linker histone H1 has been proven to promote phase separation of chromatin, which is consistent with its role in promoting chromatin condensation in cells (18). Similar to histone H1, Hmo1 showed the phase separation property when incubated with dsDNA. We found that the lysine-rich CTD of Hmo1 was necessary for forming condensates, which further demonstrated the importance of lysine-rich CTD for chromatin assembly. Unlike the spherical droplets formed by human linker histone H1.4, Hmo1 formed gel-like condensates with dsDNA. This phenomenon indicated stronger Hmo1-DNA and Hmo1-Hmo1 interactions (Fig. 5). In addition, Hmo1 was phosphorylated in the same fluctuation as H1 in the cell cycle, suggesting the regulation of Hmo1 activity by the post-translational modifications during the cell cycle (Fig. 6). Collectively, we speculated that Hmo1 could perform some functions as histone H1 to facilitate the chromatin assembly, potentially utilizing its lysine-rich CTD that participates in phase separation.
The established model for linker histone H1's function in packaging nucleosomal array into a higher-order chromatin structure is primarily based on purified proteins or purified chromatin studies in vitro (85)(86)(87)(88). However, the role of H1 in vivo is less certain. YNPE is a robust biochemical system that mimics the physiological conditions in vivo to some extent. This study showed the biochemical characteristics of the purified proteins (H1.4, Hmo1, Hho1, and HU) in vitro differ from those shown in YNPE ( Fig. 2 and 3; Fig.  S5 and S6), suggesting that a protein's contribution to chromatin assembly is strongly regulated by multiple other proteins in vivo, rather than relying solely on its biochemical properties. This may explain some discrepancies in behavior observed between purified Hmo1 (24,79) and Hmo1 in YNPE. In the future, ultra-resolution imaging techniques and highly sensitive mechanical measurements appropriate for use in YNPE are highly desirable to obtain a more comprehensive understanding of the biochemical details of chromatin assembly that occur in vivo.

Preparation of the YNPE and single-molecule visualization with total internal reflection fluorescence microscopy (TIRFM)
YNPEs from different yeast strains were prepared as described by Chen et al. (56). The protein concentrations of YNPEs were measured using the Bradford Protein Concentration Assay Kit (Beyotime Inc.). Single-molecule experiments were performed in the flow cell. Flow cell assembly and λDNA biotinylation were modified from our existing procedure (90). Streptavidin of 10 µL (0.5 mg/mL, Sigma-Aldrich) was injected into the flow cell and incubated for 2-3 minutes at room temperature. After the outlet tubing was connected to the aspiration syringe pump (Harvard Instruments, standard injection/extraction PHD ULTRA), the flow cell was flushed with blocking buffer [20 mM Tris-HCl pH 7.5, 50 mM NaCl, 2 mM EDTA, 0.2 mg/mL bovine serum albumin (BSA)] at a rate of 100 µL/min. Next, diluted biotin-λDNA (final concentration: 1.5 nM) contain ing 10 nM Sytox orange (Thermo Fisher Scientific) was pumped into the flow cell at 50 µL/min for approximately 30 seconds. Excess DNA was washed with the blocking buffer containing Sytox dye at 100 µL/min. After immobilizing λDNA molecules at one end in the flow cell, the flow cell was washed with buffer, and YNPE was introduced at the rate of 50 µL/min. The inverted TIRFM [modified from IX-71 Olympus, 100× objective with numerical aperture (N.A.) = 1.49] equipped with an Andor iXON ultra 897 EMCCD was used to acquire the images of DNA molecules under excitation with a 532-nm laser. Videos recording DNA length changes were captured by an electron multiplying charge coupled device (EMCCD) with 0.1-second exposure and 0.1-second interval for 10 minutes. The imaging data were analyzed as previously described (56).

Observation of fluorescently labeled nucleosomes on DNA
The procedure described by Visnapuu and Greene was followed with modifications to label the nucleosomes with fluorescent dyes (91). Briefly, the YNPE hmo1Δ-H2B-3FLAG was diluted with the blocking buffer containing 10 nM Sytox orange and pumped into the flow cell at 50 µL/min until the DNA length no longer changed. The anti-FLAG tag antibody (Abbkine) of 2 µg/mL was pumped into the flow cell at 50 µL/min and incubated for 10 minutes. After washing the unbound antibody with the blocking buffer, the fluorescence of DNA and histone H2B were visualized under the excitation by 532 nm and 640 nm lasers. The co-localization of nucleosome fluorescence and DNA was determined by the fluctuation along with DNA in the fluid flow.

Single-molecule mechanical assay with magnetic tweezers
DNA substrate with the random sequence was labeled with three digoxigenin mole cules and biotin at each end. The flow chamber was assembled by two coverslips and functionalized with anti-digoxigenin protein. Reference beads were attached to the flow chamber for drift correction. In the magnetic tweezers assay, the same concentration (100 µg/mL) of different types of YNPE was first incubated with double-stranded DNA (dsDNA) for 20 minutes at room temperature to assemble chromatin, then the chroma tin assembled in YNPE was bound to a functionalized coverslip at one end and to a magnetic bead (M280, Invitrogen) at the other end. As for the magnetic tweezers experiments with purified proteins, the same concentration (10 nM) of purified protein (Hmo1 or H1.4 or LmHU) was added to the flow chamber after the dsDNA bound to the magnetic bead and then incubated for 5 minutes. The blank buffer in the experiments referred to the blocking buffer (20 mM Tris-HCl pH 7.5, 50 mM NaCl, 2 mM EDTA, 0.2 mg/mL BSA), which was used as a dilution buffer for all YNPE and purified proteins. The magnetic bead moved along the Z-axis stepwise under the action of the changing external force, which was applied by moving the permanent magnet above the flow chamber at a rate of 62.5 µm/step (>1 pN, constant force with 3 seconds per step). The forces in the magnetic tweezers experiments varied from ~0.2 to ~40 pN. The diffraction ring shape of the magnetic bead was observed with a microscope (100×, N.A. 1.45; Olympus) to monitor the DNA extension (92). The magnetic bead was rotated 50-100 turns clockwise or counter-clockwise with a force of ~1-2 pN to check for the individually tethered dsDNA. After rotation, the magnetic beads with the invariant positions on the z-axis were selected to plot the force-extension curves. The curves fitted by the worm-like chain (WLC) model formula were used as a marker for DNA extension (93).

Phase separation and fluorescence recovery after photobleaching (FRAP) assays
Protein concentrations were diluted to 10 µM with blocking buffer and further fluorescently labeled with the RED-NHS Protein Labeling Kit (Monolith). Protein and dsDNA were mixed at equal concentrations of 10 µM (final nucleic acid to protein ratio of approximately 1, N/P ~ 1) in a centrifuge tube and incubated for 2-3 minutes at room temperature. Then the mixture was gently transferred to a confocal petri dish (35 mm diameter; Biosharp Inc.). Phase separation was observed using a Leica SP8 laser confocal microscope (100×, N.A. 1.4) under the bright field, 633-nm or 488-nm laser. The 300-bp dsDNA substrate was obtained by PCR (primer 5′−3′: atgcaaagcccatatccaatgacac and gcctcttttacctcgagggtgg) from the yeast genome. FAM-labeled and unlabeled 60-bp dsDNA were synthesized by Tsingke Biotechnology. pBR322 and λDNA were purchased from TaKaRa and New England Biolabs. DNAase was purchased from YEASEN. FRAP assays were carried out as previously described by Taylor et al. (69). A self-selected circular region of interest (ROI) within the phase separation assembly was photobleached by laser excitation, and the recovery process was observed for approximately 3 minutes. Fluorescence intensity was normalized and plotted against a function of time.

Phosphorylation detection of Hmo1
The yeast cells were synchronized in the G1 phase by arresting with α factor (5 µg/ mL). The cell cycle phase was determined by microscopic cell morphology and flow cytometry (BD FACS Calibur). After being released from arrest, yeast cells were collected every 10 minutes to extract total proteins. Phosphorylation of Hmo1 was detected by the acrylamide-dependent Phos-tag kit (Wako). The protein samples from 0 to 90 minutes were electrophoresed and detected with the anti-Hmo1 antibody (ab71834) and the anti-GAPDH antibody (AC033).

DATA AVAILABILITY
All relevant data are included in the manuscript and the Supplementary material file.

ADDITIONAL FILES
The following material is available online.
Research Article mBio