Htz1 interacts with Nap1 and multiple karyopherins in cytosol
It is generally assumed that Htz1 forms an obligate heterodimer with H2B shortly after synthesis. As Htz1 synthesis occurs throughout the cell cycle, this suggests a model where Htz1/H2B is imported separately, both spatially and temporally, from H2A/H2B, demonstrating that the import of the two heterodimers is likely distinctly regulated. To understand how the nuclear transport of histone variant Htz1 is coordinated with its deposition into chromatin, we immunopurified Htz1 from yeast cytosol to identify interacting proteins. A strain containing an allele of HTZ1 with a C-terminal protein A (PrA) tag was grown to logarithmic phase and Htz1-PrA and associated proteins were purified on IgG sepharose from the post-ribosomal cytosolic fraction. Bound protein was eluted with 1 M MgCl2, enzymatically digested, and the complex mixture was submitted for analysis by liquid chromatography/tandem mass spectrometry. Among the interacting proteins, four karyopherins (Kaps) were identified; Kap114, Kap123, Kap95, and Kap108 (Sxm1). These Kaps were previously shown to transport canonical H2A and H2B into the nucleus, with a specific role identified for Kap114. Histone chaperone Nap1 was also identified. Nap1 acts as an import cofactor for H2A and H2B and has been shown to be a chaperone for Htz1. Chz1 is also a histone chaperone specific for Htz1, but surprisingly we did not find any in our cytosolic interacting protein fraction.
To verify the Kap interactions by Western blot, we immunoprecipitated Htz1-PrA from a strains expressing Kap114-Myc, Kap123-Myc, Kap95-Myc and Kap108-Myc. Bound proteins were eluted with a step gradient of MgCl2 and the Kaps were detected with α-Myc antibodies, confirming the mass spectrometry results ( and data not shown). As a negative control we determined that Kap120-Myc did not co-elute with Htz1-PrA (). We also tested whether Nap1 was co-precipitated with Htz1-PrA, and determined that it eluted in the same fractions as Kap114 (). In the reciprocal experiment, Htz1 was co-immunoprecipitated from cytosol with Nap1-PrA (). Additionally, we tested for the presence of Chz1 co-eluting with Htz1-PrA using anti-Chz1 antibodies. We were unable to detect any co-eluting Chz1 by Western blot although we did detect Chz1 in the unbound fraction (). In contrast, Nap1 co-elution was clearly evident in the same experiment. These results suggested that Htz1 interacted with several Kaps representing multiple import pathways, and that Nap1 and Htz1 associated in the cytoplasm prior to nuclear import. The Kaps we observed were similar to those we reported for H2A and H2B and raised the possibility that H2B mediated the import of the Htz1/H2B heterodimer.
Figure 1 Htz1 associates with Kap114, Kap123, and Nap1 in yeast cytosol. (A) Htz1-PrA and associated proteins were purified from cytosol prepared from a strain expressing Kap114-13myc, eluted with a MgCl2 step gradient, separated by SDS-PAGE and analyzed by Western (more ...)
The N-terminus of Htz1 contains an NLS
We therefore addressed whether Htz1 contains its own active NLS. All histones contain a histone fold domain that consists of three α-helices (α1, α2 and α3), as well as the unstructured N terminal domain. In addition some histones contain additional alpha helical domains N-terminal to α1 (αN), and C-terminal to α-3 (α-C). All four core histones, H2A, H2B, H3 and H4, have been shown to contain an NLS in their N-terminus (31
). To test the Htz1 N-terminus for NLS activity, we expressed Htz1 residues 1–53 fused to two GFP moieties (Htz11–53
) in wild type cells. This fragment of Htz1 corresponds to the fragment of H2A that contains full NLS activity, and encodes the flexible N terminus, αN and α1 domains (32
). Upon induction of GFP reporter expression, the Htz11–53
fusion protein was localized to the nucleus, indicating that Htz1 contained an NLS in its N-terminus (). We then expressed Htz1 residues 1–24 (containing the unstructured N terminus) or 24–53 (containing the αN and α1 domain) fused to GFP2
to determine if the entire 1–53 segment is necessary to direct the reporter into the nucleus. The Htz11–24
fusion was distributed throughout the nucleus and cytoplasm, indicating the extreme N-terminus was not sufficient for NLS activity (). Htz124–53
was localized to the nucleus and cytoplasm, but the nucleus was discernable in most cells, indicating that Htz1 residues 24–53 contained partial NLS activity. Our results indicated that Htz1 contains an NLS in its N-terminus, and that the entire 1–53 sequence is required for full NLS activity. We wanted to determine whether Htz1 contained an NLS in the C terminal part of the protein, therefore we expressed Htz132–134
, (deleting the unstructured N terminus and the αN domain) and Htz152–134
. (deleting the unstructured N terminus, the αN domain and the α1 domain). The Htz132–134
was nuclear, and Htz152–134
was expressed very poorly, but in expressing cells was visible in both nuclear and cytoplasmic compartments, with some nuclear acccumulation. This result suggested that there maybe second weaker NLS in Htz1, or that some nuclear import was mediated by the H2B NLS, via dimerization with H2B.
Figure 2 Htz1 contains an NLS in its amino terminus (A) Wild type (WT) yeast containing the indicated plasmid, were induced to express Htz1 residues 1–53, 1–24, 24–53, 32–134 or 52–134 fused to GFP2 and visualized by fluorescent (more ...)
To test this we expressed our GFP constructs in a strain (JHY200) that bears deletions of both copies of each of the four core histones, and growth is supported by plasmid pQQ18 that encodes one copy of each of the core histones. The genomic copy of HTZ1 was deleted in this strain and Htz1-GFP2, Htz132–134-GFP2, and Htz152–134-GFP2, were expressed from plasmids. A mutation was created in HTB1, the gene encoding H2B, whereby amino acids 1–32, containing the H2B NLS, were deleted in the context of pQQ18. Htz1-GFP2 and Htz132–134-GFP2 were nuclear in strains with and without the H2B NLS (). Htz152–134-GFP2 was expressed at very low levels in this strain background. In cells with intact H2B, Htz152–134-GFP2 appeared to be both cytoplasmic and nuclear, however, in a subset of cells (<10%) distinct nuclear accumulation of GFP was observed (). In contrast, nuclear accumulation was not apparent in cells lacking the H2B NLS, and GFP was equilibrated through the cell (). However, Htz152–134-GFP2 was very poorly expressed in these cells, and cytoplasmic and nuclear aggregates were also visible. These data are consistent with the finding that the major Htz1 NLS is located in the N terminal half of the protein and the α1 domain (between residue 32 and 52) is particularly important for this activity.
Acetylation of Htz1 is not necessary for import or growth
We have shown that mutation of key acetylation sites in H3 and H4 perturbs NLS function, and loss of positive charge, mimicking acetylation, reduces nuclear import (34
). Htz 1 can be acetylated on four lysine residues located in the N-terminus (8
). We mutated the four acetylated lysines within the Htz1 NLS to glutamine to mimic acetylation, and to arginine, which mimics the positively charged unacetylated lysine. We expressed the Htz1 NLS mutants in the context of our Htz11–53
reporter and observed the localization of GFP. The Htz11–53
reporter was localized throughout the cell while the Htz11–53
reporter was localized primarily to the nucleus similar to wild type (). These results suggested that positive charge rather than acetylation of these lysines promotes import. In the context of full length Htz1-GFP, mutation of these residues had no effect on localization (data not shown). To prevent co-import via the H2B NLS we expressed full length htz1 K3,8,10,14Q-GFP2
in strains expressing H2B Δ1–32. The observed nuclear signal demonstrated that H2B Δ1–32 /Htz1 K3,8,10,14Q dimers were able to effectively access the nucleus (). This suggests that in the context of the full length protein the htz1 K3,8,10,14Q
mutation does not abrogate NLS activity.
The Htz1 NLS directly interacts with several karyopherins
An Htz1/H2B dimer contains two distinct NLSs, and therefore the associated Kaps identified by mass spectrometry may associate with either the Htz1 or H2B NLS. To test which Kaps were important for Htz1 mediated import, we performed an import assay using the Htz11–53GFP2 reporter, which should not dimerize with endogenous H2B. We expressed Htz11–53GFP2 in strains containing a deletion or mutation of one of the Kap genes. Deletion of KAP123 resulted in strong mislocalization of the Htz11–53GFP2 reporter to the cytoplasm, but did not cause mislocalization of an Asf1-GFP2 reporter (). Deletion of KAP114 or KAP108/SXM1 had little affect on Htz11–53GFP2 reporter localization when compared to wild type (). Strains containing a mutant allele of KAP95(kap95ts) or SRP1 (srp1–31), also showed no significant mislocalization of the GFP reporter at the restrictive temperature (). As mislocalization of the NLS reporter was not observed in most of the single Kap deletion strains, even those Kaps that co-purified with Htz1-PrA, it suggested that Htz1 import was mediated by several Kaps. The mislocalization observed in the kap123Δ strain suggested that Kap123 is a major import karyopherin for Htz1.
Figure 3 Htz1, Nap1, Kap114 form a co-complex. (A) The Htz1 NLS (residues 1–53) or Asf1 GFP2 reporter was expressed in yeast strains containing mutations of individual KAP genes as indicated and localized by fluorescent microscopy. Coincident Hoechst staining (more ...)
To determine whether the interaction between Htz1 and the Kaps detected in cytosol was direct or mediated through histone H2B, we tested these interactions using recombinant proteins. Htz1 residues 1–53 was expressed with a n N-terminal GST tag and immobilized on glutathione Sepharose beads. The beads were incubated with recombinant MBP tagged Kap114, Kap123, or untagged Kap95, and bound protein was analyzed by SDS-PAGE and Coomassie blue staining. Unlike the MBP-LacZ control protein, each of these Kaps was able to bind directly to the Htz1 NLS (), further supporting their proposed role in Htz1 transport. This suggests that Htz1 binds these Kaps directly.
Htz1, Nap1, and Kap114 form an import co-complex
Previously, we have demonstrated that Nap1 plays a role in H2A/H2B import and promotes association of the histones with Kap114 in the cytoplasm (22
). In this study we used a RanGTP dissociation assay to demonstrate for the existence of an H2A-Nap1-Kap114p co-complex (22
). In the absence of Nap1, RanGTP readily dissociates H2A from Kap114. However, the Nap1-Kap114p complex is insensitive to RanGTP, and we showed that addition of Nap1 to the Kap114-H2A complex, rendered this complex insensitive to RanGTP dissociation. This suggested that Nap1 was able to bridge the histone-Kap interaction (22
). We used this assay again to determine whether Htz1, Kap114 and Nap1 formed a co-complex. Nap1 interacts with the NLS domain of H2A and we tested whether Nap1 directly associated with an Htz1-NLS fusion protein. Recombinant Nap1, but not Chz1, directly interacted with the GST-Htz11–53
(). Recombinant GST-Htz11–53
fusion protein was incubated with or without Nap1. MBP-Kap114 was then added with or without pre-incubation with Gsp1Q71L-GTP, a GTPase deficient yeast Ran mutant. As seen in , the interaction between GST-Htz11–53
and MBP-Kap114 was sensitive to Gsp1Q71L-GTP. In contrast, in the presence of Nap1, the GST-Htz11–53
, MBP-Kap114 interaction was insensitive to Gsp1Q71L-GTP (). This result suggests that Nap1 can serve as a bridge between the Htz1 NLS and Kap114 in an import complex. In a parallel experiment, we showed that the GST-Htz11–53
, MBP-Kap123 interaction was also sensitive to Gsp1Q71L-GTP (). However, in this case addition of Nap1 did not render the GST-Htz11–53
, MBP-Kap123 complex insensitive to Gsp1Q71L-GTP. In fact, Nap1 appeared to compete with MBP-Kap123 for binding to GST-Htz11–53
(). In a control experiment we showed that the complex remained intact when incubated with the nonspecific protein MBP-LacZ (). This suggests that Kap123, Nap1 and Htz1 do not form a co-complex, and similarly to its role with H2A, Nap1 may promote binding of Htz1/H2B to Kap114 (22
Chz1 contains an NLS and interacts with Kap95 in cytosol
It has been argued that Nap1 and Chz1 are redundant chaperones for Htz1. Our Htz1 immunoprecipitation experiments suggest Chz1 did not interact with Htz1 in cytosol. As Chz1 is nuclear we wanted to determine whether Chz1 did indeed enter the nucleus independently of Htz1 and contained its own NLS (19
). Chz1 was identified in a bioinformatics search for yeast nuclear proteins with consensus classical NLS (cNLS) sequences, although the putative Chz1 NLS was not examined experimentally (27
). We tested whether the putative cNLS of Chz1 was functional using a GFP reporter assay. An N-terminal fragment of Chz1 (amino acids 1–80) was fused to GFP2
and expressed in wild type cells. The GFP signal was localized to the nucleus, suggesting that the N-terminus of Chz1 does contain NLS activity (). We then expressed Chz1 residues 22–49, which contain the predicted cNLS (36
), fused to GFP2
and expressed in wild type cells. The GFP signal was predominately localized to the nucleus, indicating that the cNLS is indeed functional (). Chz1 74–128-GFP2
, which contains the Htz1/H2B-interacting “CHZ” domain, was localized throughout the cell (19
) (). We also tested the C-terminus of Chz1 (residues 129–160) for NLS activity, but this reporter was also localized throughout the cell ().
Figure 4 Chz1 contains an NLS and associates directly with Srp1/Kap95 in cytosol. (A) Fragments of Chz1, bearing the indicated mutations, fused to GFP2 were expressed in wild type yeast as indicated and visualized by fluorescence microscopy. Coincident Hoechst (more ...)
To further delineate the NLS we made a series of lysine and arginine to alanine mutations in the context of Chz11–80GFP2. The mutant Chz11–80K36A, K38A, R39A, R40A, R42A-GFP2 was mislocalized to the cytoplasm (). We tested a mutant with fewer changes and showed that Chz11–80K36A, K38A, R39A-GFP2 was also mislocalized suggesting that these are key amino acids in the NLS (). In contrast mutation of one lysine was not sufficient as Chz11–80K36A GFP2 was localized to the nucleus (). These data suggest that Chz1 contains a cNLS in its N-terminus and that amino acids K36, K38 and K39 are critical residues.
The presence of a cNLS within Chz1 suggested import by the classical import pathway, which is mediated by the Srp1(Kap60)/Kap95 heterodimer. To test this we tagged KAP95 with a 13-Myc epitope in the CHZ1-TAP strain. If Chz1 was co-imported with Htz1, we would expect an interaction with Htz1-interacting Kaps, so we also tested Kap114. Chz1-TAP was immunopurified from cytosol and bound protein was analyzed by Western blotting. Kap95-Myc eluted in the 50mM to 1M MgCl2 fractions, consistent with Chz1 being an import cargo of Srp1/Kap95 (). No Kap114-Myc was detected in association with Chz1, suggesting that Chz1 does not use Kap114 for import (). We next tested if the Srp1/Kap95, Chz1 association was direct and we could assemble a Chz1-Srp1/Kap95 import complex in vitro. Recombinant H6-PrA-Chz1 was immobilized and incubated with H6-Srp1 and untagged Kap95. Analysis of bound protein by Coomassie blue staining showed that similar amounts of H6-Srp1 and Kap95 bound to H6-PrA-Chz1 (). Neither Srp1 nor Kap95 bound PrA alone. These results suggest that while Nap1 and Htz1 are co-imported, Chz1 is imported independently by the Srp1/Kap95 pathway.
We tested whether Chz1 (amino acids1–80) was mislocalized in strains bearing mutant alleles of srp1 and kap95 grown at the nonpermissive temperature. We did not see any obvious mislocalization in these strains suggesting that Chz1 may also have additional, secondary Kap-mediated pathways into the nucleus (data not shown).
Nap1, but not Chz1, maintains the soluble pool of Htz1
Our results predict that Nap1, but not Chz1, is an import cofactor for Htz1. This finding is consistent with the fact that we did not identify Chz1 in the Htz1 interacting proteins from cytosol and Chz1 does not interact with Kap114. As NAP1 is a non-essential gene, we wanted to determine whether in the absence of Nap1, Chz1 could serve as import cofactor and interact with Htz1 in cytosol. We generated cytosol from yeast strains expressing Htz1-FLAG in wild type, nap1Δ, or chz1Δ genetic backgrounds. Htz1-FLAG was immunoprecipitated using anti-FLAG agarose beads and associated proteins were eluted with MgCl2. Purified proteins were separated by SDS-PAGE and visualized with Coomassie blue stain. In the wild type strain, Htz1-FLAG was associated with Nap1, as well as H2B, and other copurifying proteins (). Surprisingly, in the nap1Δ strain we were not able to detect any soluble Htz1-FLAG by Coomassie stain and consequently no other major interacting proteins, including Chz1, were evident (). Parallel experiments using mutant strain chz1Δ were similar to wild type (). To verify that all our strains expressed Htz1-FLAG, we analyzed the cytosolic unbound fractions, the nuclei enriched pellet fractions, and whole cell lysates from each strain for the presence of Htz1-FLAG. In the absence of NAP1, Htz1-FLAG was not visible in the unbound cytosolic fraction as determined by Western blot, in contrast, Chz1 was apparent (). This could suggest that in the nap1Δ strain, cytoplasmic Htz1 is shifted to the nucleus. In the nuclei enriched pellet and whole cell lysate fractions, seemingly equivalent amounts of Htz1-FLAG were visible from all strains suggesting that the total cellular levels were unchanged (). This leaves the cytoplasmic fraction of Htz1 unaccounted for, however, as there is far more Htz1 in the nucleus, complete loss of Htz1 from the cytoplasm would be expected to only lead to a very modest increase in the nucleus which may be difficult to detect.
Figure 5 Nap1 is important to maintain soluble pool of Htz1. (A) Htz1-FLAG was immunopurified from yeast cytosol from the indicated strain backgrounds. Bound protein was eluted with the indicated concentration of MgCl2 and analyzed by SDS-PAGE and Coomassie stain (more ...)
This result further suggested that Chz1 does not associate with Htz1 in cytosol, whilst Nap1 appeared to be important for the presence of a cytoplasmic pool of Htz1. In this case, we predicted that overexpression of Nap1 would increase the cytosolic pool of Htz1 and lead to relocalization of the Htz11–53
reporter. To test this, NAP1
was overexpressed under the control of the yeast GPD1
promoter. Overexpression was confirmed by the presence of long budded cells, a phenotype indicative that Nap1 overexpression () (35
). In this strain the Htz11–53
reporter was partially relocalized to the cytoplasm, suggesting the excess Nap1 was increasing the amount of Htz11–53
in the cytoplasm at steady state (). We observed a similar result when we expressed full length Htz1-GFP2
(). These results were consistent with a role of Nap1 in maintaining a soluble pool of Htz1. To ensure that excess Nap1 did not result in a general transport defect, we also expressed Asf1-GFP2
in this strain and did not observe relocalization of the reporter to the cytoplasm ().
CHZ1 and NAP1 are not functionally redundant
Our data suggests that Nap1 is a transport cofactor for Htz1/H2B and helps maintain a soluble pool of Htz1, and Chz1 does not appear to share these functions. However, both Nap1 and Chz1 have been described to deliver Htz1/H2B heterodimers to the SWR1 complex, indicating functional redundancy (19
). We next wanted to explore if NAP1
display similar genetic interactions indicative of the extent of their functional overlap. Deletion of NAP1
did not lead to reduced fitness on YPD, while deletion of HTZ1
does lead to reduced fitness as exhibited by slow growth on YPD (11
). If HTZ1
and its chaperones belong in a linear functional pathway, then a double mutant strain should have a similar growth phenotype as the htz1
Δ null. We deleted NAP1
in an htz1
Δ null strain and observed the fitness of the double mutants on complete synthetic media (CSM). The htz1
Δ mutant grew more slowly than wild type whereas both the nap1
Δ and chz1
Δ single mutants grew similarly to wild type (). The chz1
Δ double mutant grew similarly to htz1
Δ (). Surprisingly, the nap1
Δ double mutant grew better than htz1
Δ, indicating a positive genetic interaction between NAP1
Δ null strains are sensitive to growth on media lacking inositol and media containing galactose as the carbon source, suggesting defects in transcription (11
). We tested nap1
Δ and chz1
Δ double mutants for growth defects on CSM lacking inositol (-INO) and CSM galactose (GAL). Yeast containing htz1
Δ or chz1
Δ grew more slowly on CSM GAL and CSM -INO than wild type, nap1
Δ, or chz1
Δ single mutants (). The nap1
Δ double mutant strain grew slightly less well than wild type but better than htz1
Δ (). These results suggest that deletion of NAP1
relieves the apparent transcriptional defects observed in an htz1
Δ strain, a function that is not shared with CHZ1
. This gives further evidence that the histone chaperones Nap1 and Chz1 have separate Htz1-dependent and -independent functions.
Figure 6 NAP1, SWR1, and SWC2, but not CHZ1, exhibit genetic interactions with HTZ1. Strains of the indicated genotype were grown to logarithmic phase, equalized, and spotted as 10-fold serial dilutions onto CSM, CSM lacking inositol (CSM –INO), or CSM (more ...)