Molecular architecture, structure-function relationship, and importance of the Elp3 subunit for the RNA binding of holo-elongator.

The molecular architecture of six-subunit yeast holo-Elongator complex was investigated by the use of immunoprecipitation, two-hybrid interaction mapping, and in vitro studies of binary interactions between individual subunits. Surprisingly, Elp2 is dispensable for the integrity of the holo-Elongator complex, and a purified five-subunit elp2 Delta Elongator complex retains histone acetyltransferase activity in vitro. These results indicate that the WD40 repeats in Elp2 are required neither for subunit-subunit interactions within Elongator nor for Elongator interaction with histones during catalysis. Elp2 and Elp4 were largely dispensable for the association of Elongator with nascent RNA transcript in vivo. In contrast, Elongator-RNA interaction requires the Elp3 protein. Together, these data shed light on the structure-function relationship of the Elongator complex.

The molecular architecture of six-subunit yeast holo-Elongator complex was investigated by the use of immunoprecipitation, two-hybrid interaction mapping, and in vitro studies of binary interactions between individual subunits. Surprisingly, Elp2 is dispensable for the integrity of the holo-Elongator complex, and a purified five-subunit elp2⌬ Elongator complex retains histone acetyltransferase activity in vitro. These results indicate that the WD40 repeats in Elp2 are required neither for subunit-subunit interactions within Elongator nor for Elongator interaction with histones during catalysis. Elp2 and Elp4 were largely dispensable for the association of Elongator with nascent RNA transcript in vivo. In contrast, Elongator-RNA interaction requires the Elp3 protein. Together, these data shed light on the structure-function relationship of the Elongator complex.
The Elongator complex was first identified as a component of a hyperphosphorylated RNA polymerase II (RNAPII) 1 holoenzyme isolated from yeast chromatin (1). One of the subunits of Elongator, Elp3, contains domains conserved among histone acetyltransferases (HATs) (2), and Elongator has HAT activity directed against histone H3 and to a lesser extent H4 in vitro (3), raising the possibility that Elongator acetylates histones during RNAPII transcript elongation.
Elp3 HAT activity is essential for Elongator function in vivo as point mutations that abolish the catalytic function of Elp3 in vitro also confer the full range of elp phenotypes (4). elp phenotypes include temperature sensitivity at 39°C, salt sensitivity, and slow adaptation to growth on carbon sources such as galactose (1,5). elp cells are also sensitive to caffeine, Calcofluor White, and 6-azauracil (6,7). Significantly, in all cases tested, elp phenotypes correlate with a failure of the mutant cells to activate, in a timely manner, the genes required for growth under the new conditions (1).
An elp deletion in combination with the deletion of different genes encoding transcription factors confers synthetic phenotypes. Thus, elp3 in combination with rpb9 (encoding an RNA-PII subunit) is lethal (8), as is the combination of an elp3 and ctk1 mutation (CTK1 encodes a subunit of the CTD kinase, CTDK1) (9). Cells expressing a conditional allele of SPT16 (encoding the largest subunit of yeast FACT-CP complex) also display a synthetic phenotype in combination with elp3 mutation (10). Recent data have shown that Elongator also genetically interacts with Mediator, Rad6 ubiquitin ligase, Paf1 complex, and Rpd3-Sin3 complex. 2 Finally, combining the mutation of elp3 with gcn5 (encoding the catalytic subunit of the SAGA-ADA HAT complexes) confers severe growth defects, which are not seen in either of the single mutants (4). Significantly, the gcn5 elp3 double mutant has reduced levels of histone H3 acetylation in several genes compared with the single mutants, and low levels of acetylation in the coding region of these genes correlates with reductions in gene transcription. By contrast, low levels of acetylation in the promoter of genes does not correlate with reduced transcription nor with reduced promoter occupancy by the TATA-binding protein (11). Recent experiments using RNA-immunoprecipitation have shown that Elongator is indeed present in the coding region of active genes in vivo (12).
The Elongator complex was first thought to consist of three subunits, named Elp1, Elp2, and Elp3 (now called core Elongator) (1), but later purification of the complex from a yeast strain expressing epitope-tagged Elp1 has shown that the active complex, holo-Elongator, consists of six subunits (Elp1-Elp6) (13)(14)(15), which are organized in two three-subunit subcomplexes (13,15). Significantly, deletion of any one of the (ELP) genes encoding these six subunits confers more or less identical phenotypes, and new phenotypes are not detected upon concomitant deletion of two or three ELP genes (5,13). These data suggest a tight functional connection between the proteins comprising the Elongator complex. Consistent with this idea, holo-Elongator, but not core Elongator, has HAT activity even though both complexes contain the catalytic Elp3 subunit (13).
Here we investigate the molecular architecture of Elongator complex and show that the only ELP gene that can be deleted without significant loss of Elongator integrity is ELP2. Surprisingly, an Elongator complex lacking Elp2 even retains the ability to acetylate histones in vitro. We also show that the association of Elongator with nascent RNA in vivo requires Elp3 but not Elp2 or Elp4.

EXPERIMENTAL PROCEDURES
Yeast Strains and Phenotypic Analysis-All Saccharomyces cerevisiae stains used for genetic analysis (Table I) were congenic with strain W303 and were grown and manipulated as described previously (1,4).

Expression of Tagged Proteins in
Vivo-For the construction of the Elp4-HisHA strain, part of the ELP4 open reading frame was amplified using primers 5Ј-GCCGGGGTACCGGGGCTGCCGCTGGGAAACTCT-G-3Ј and 5Ј-CCGGCCGGATCCATAGTCTAAAGATATCTTGGTCT-3Ј and cloned into pSE.HISHA-304 (16) using the KpnI and BamHI sites to produce plasmid pELP4-HISHA-304. After yeast transformation, a TRP ϩ clone was isolated in which the 3Ј-end of the ELP4 gene was replaced, resulting in expression of an Elp4-(His) 10 -HA fusion protein.
To ensure that the (His) 10 -HA epitope tag did not interfere with Elp4 function, phenotypic analysis was performed.
Protein Purification-The procedure for purification of Elongator from the elp2⌬ Elp1-HISHA strain has been described elsewhere (12).
Yeast Whole Cell Extract Preparation and Western Blot Analysis-Whole cell extracts were prepared as was described elsewhere (17) from the indicated strains. The antibody against ␣-tubulin was a mouse monoclonal from Oncogene and was used in a final dilution of 1/100 for immunoblotting. The antibodies against Elp1, 2, 3, 4, 5, and 6 were used as described previously (1,2,5,13).
RNA Preparation-For the detection of ELP3 mRNA, total RNA was isolated from an equal number of S. cerevisiae cells from the indicated strains using the RNAeasy mini kit (Qiagen) according to the manufacturer's recommendations. The probe was a PCR product from the coding region of the ELP3 gene. For labeling, Ready-to-Go DNA labeling beads from Amersham Biosciences were used according to the manufacturer's recommendations. Hybridization of the membrane with the probe was performed using the ExpressHyb kit from Clontech. For the detection of 28 and 18 S RNA, the gel was stained with EtBr in a final dilution of 1/10,000 in water.
Co-immunoprecipitation Experiments-For Holo-Elongator co-immunoprecipitation experiments, proteins from the first Bio-Rex chromatography step were used (13). 500 g of protein adjusted to 500 mM salt in buffer A (12) was incubated with Sepharose A beads, which had been previously conjugated with 12CA5 antibody. The beads were washed three times with the same buffer and resuspended in 1ϫ SDS loading buffer, and the bound proteins were subjected to SDS-PAGE.
Two-hybrid Interactions-The coding sequence of all of the genes encoding the Elongator proteins were cloned in both "GAL4-activation domain"-and "GAL4-binding domain"-based vectors provided from the GAL4 two-hybrid phagemid vector kit (Stratagene). To detect interactions, the plasmids were introduced in genetically modified yeast cells provided from the Matchmaker Library Construction & Screening Kit (Clontech). The interaction studies were done according to the manufacturer's recommendations.
Expression of Recombinant GST and GST-Elp5 in Bacteria-To generate a GST-Elp5 fusion protein, the open reading frame of the ELP5 gene was cloned in the pGEX-3X vector (Amersham Biosciences), and the plasmid was introduced in BL21 DE3 competent bacteria. Cells were grown at 37°C and induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h. After cell resuspension in phosphate-buffered saline, the cell suspension was incubated for 30 min on ice with lysozyme (50 mg/ml) and then subjected to sonication. The extract was incubated for 1 h at 4°C with Triton X-100 in a final concentration of 1%, and the soluble supernatant was collected after a 10-min 12,000 rpm spin at 4°C. This soluble supernatant contained GST-Elp5, which was purified and immobilized on glutathione beads (Amersham Biosciences) by incubation for 30 min at room temperature. The soluble GST control protein was produced in the same way.
In Vitro Transcribed/Translated Proteins-The recombinant Elp4 and Elp6 proteins used for the in vitro pull-down experiments were produced using the TNT T7-coupled wheat germ extract systems (Promega) according to the manufacturer's recommendations. The DNA template used was pBluescript II KS (Ϯ), into which the coding sequence of the two genes was cloned. Details are available on request.
In Vitro Pull-down Experiments-Immobilized baculovirus-expressed histidine-tagged Elp1 or Elp2 and bacterially expressed GST or GST-Elp5 proteins were mixed with the product of in vitro transcription/translation (Elp4 or Elp6) reactions in a buffer containing 250 mM potassium acetate, 100 mM Hepes-KOH, pH 7.6, 20% glycerol, 0.1% Nonidet P-40, 1ϫ proteinase inhibitors, and 4 mM ␤-mercaptoethanol and incubated overnight. The beads were washed three times before the bound proteins were separated by PAGE on a 10% SDS gel. The Elp4 and Elp6 proteins showed weak nonspecific binding to nickel-agarose. Therefore, these beads were washed three times with the above buffer but containing 30 mM imidazole.
Other Assays-Histone acetyltransferase reactions (15 l) were carried out as described (3). RNA immunoprecipitation experiments were carried out as described by Gilbert et al. (12).

RESULTS
In a multisubunit protein complex, the stability of an individual subunit can be dependent on the presence or integrity of the other subunits. We examined the amount of individual Elongator subunits present in extracts derived from different elp deletion strains. Fig. 1A (top panels) shows the results of an analysis where whole cell extracts were immunoblotted and probed with anti-Elp3 antibodies. As a loading control, reactivity with an anti-tubulin antibody was used. Interestingly, the Elp3 protein either could not be detected or was present at very low levels in cells lacking the ELP1 gene. In cells lacking any of the other ELP genes, the Elp3 levels were also somewhat reduced, but the protein was still detectable (Fig. 1A, top panel, compare the tubulin/Elp3 ratio in wild type (WT) cells with that in the mutants). By contrast, Elp4, Elp5, and Elp6 were only absent from extracts derived from strains in which their encoding gene was deleted (Fig. 1A, middle and bottom panels). Importantly, Northern blot analysis of RNA from the same cells showed that the mRNA level of the ELP3 gene was comparable in wild type and all of the elp⌬ cells (Fig. 1B) suggesting that reduced protein stability rather than gene expression was causing the effect. We conclude that deletion of the ELP1 results in substantial destabilization of the Elp3 protein, suggesting that the Elp1 protein is required for the integrity of Elp3, in all likelihood because of a direct interaction between these proteins. Protein-Protein Interactions among Elongator Subunits-To investigate protein-protein interactions in the Elongator complex, we next performed co-immunoprecipitation experiments in different elp deletion strains. For these experiments whole cell extracts were prepared from cells expressing a tagged version of an Elongator subunit while lacking another one, for example from elp2 cells expressing Elp1-HA. To enable high efficiency immunoprecipitation and to increase specificity, whole cell extracts were first subjected to chromatography on Bio-Rex before being loaded onto 12CA5-conjugated Sepharose A beads. As shown in Fig. 2A, Elongator could, under those conditions, be specifically immunoprecipitated from wild type cell extracts carrying either Elp1 or Elp4 tagged with an HA epitope but not from control cells that did not express a tagged Elp protein. Remarkably, when the ELP2 gene was deleted, Elp1 still interacted with the five remaining subunits, although the three smaller proteins Elp4, Elp5, and Elp6 appeared to be somewhat substoichiometric (Fig. 2B, left panel). This experiment indicates that the Elp2 protein is not required for the integrity of the Elongator complex.
When extracts from cells lacking the ELP3 gene were used, only the Elp1 protein was detected in the precipitates, which indicates that none of the remaining four subunits interact strongly with Elp1 (Fig. 2B, middle panel). We conclude that the Elp3 protein is not only essential for catalytic function but also for the integrity of the Elongator complex.
In the absence of Elp4, the Elp1 protein could still interact with Elp3 (Fig. 2B, right panel). However, in this case the Elp2, Elp5, and Elp6 proteins could not be detected even in the inputs. This can be explained in two ways: either those proteins no longer co-elute with Elp1 and Elp3 from Bio-Rex if Elp4 is absent, or deletion of the ELP4 gene results in reduced Elp2, Elp5, and Elp6 protein levels. Our previous experiments showed that Elp5 and Elp6 protein levels were not decreased significantly in any of the elp mutants (Fig. 1A), pointing to a lack of interaction with core Elongator as the likely explanation. Unfortunately, the Elp2 protein could not be detected in this case in the Bio-Rex-70 eluate (and could generally not be detected in crude whole cell extracts with our polyclonal anti-Elp2 antibody), precluding any conclusions on the fate of this protein in elp4 extracts. We conclude that Elp4 is required for the association of Elp2 and the small Elp4/5/6 complex with Elp1 and Elp3.
When extracts from elp5 cells expressing Elp4-HA protein were used, Elp4 only co-immunoprecipitated Elp6 and low amounts of Elp3 (Fig. 2C, left panel). This suggests that Elp4 interacts directly with Elp6 and that neither of these two proteins interacts strongly with any of the three larger proteins Elp1, Elp2, and Elp3, although a weak Elp4/Elp6-Elp3 interaction was evident. The absence of a signal for Elp1, even in the input, again precluded firm conclusions on the fate of this protein in elp5 cells. However, it was obvious from the previous experiment with elp3⌬ cells that Elp1 does not interact directly with Elp4 or Elp6 (Fig. 2B, middle panel). In elp6 cells expressing Elp4-HA protein, Elp4 also only co-precipitated a small amount of Elp3 (Fig. 2C, right panel), supporting the notion that these proteins interact directly, albeit weakly. Elp4 did not interact with Elp5 in the absence of Elp6. As Elp5 is stable in elp6 cells (Fig. 1A, middle panels), this indicates that it is incorporated into the small subcomplex in an Elp6-dependent manner.
We conclude from the above experiments that there is a direct interaction between Elp1 and Elp3, as well as between Elp4 and Elp6. Because a core-Elongator complex consisting of

FIG. 2. Elongator subunit interactions investigated by co-immunoprecipitation.
Immunoprecipitation of a tagged subunit using 12CA5 monoclonal antibodies and Western blotting with antibodies directed against the six Elongator subunits were used to investigate the integrity of the Elongator complex in different mutant strains. A, precipitations from wild type strains carrying either no tag, Elp1-HA, or Elp4-HA as indicated. B, Elp1-precipitations from the indicated mutant strains. C, Elp4 precipitations from the indicated mutant strains. Elp1, Elp2, and Elp3 has previously been isolated (1), the absence Elp2 in Elp1-precipitates from elp3 cells also indicate that interactions with Elp3 are required for the incorporation of Elp2 into core-Elongator. Surprisingly, the co-immunoprecipitation experiments do not indicate any strong direct binary interactions between any individual protein in the Elp4/5/6 module and any one of the proteins in core Elongator. This suggests that it is primarily novel interaction surfaces created by the association of Elp1, Elp2, and Elp3 in core-Elongator, and Elp4, Elp5, and Elp6 in the small subcomplex, respectively, that are important for the later association of these subassemblies into holo-Elongator.
Pairwise Elongator Protein Interactions-To compliment the co-immunoprecipitation studies, pairwise interactions between individual Elongator proteins were examined by use of the yeast two-hybrid system. The genes encoding all the Elongator subunits were cloned in-frame with the activation domain and the DNA-binding domains of the GAL4 activator, respectively, and tester yeast strains were transformed with combinations of these 12 plasmids. All possible combinations of pairwise Elongator proteins interactions were tested in both directions, but, surprisingly, the only interaction detected was between Elp4 and Elp6 (Fig. 3A, negative results not shown).
Finally, we sought to investigate the in vitro pairwise interactions of several Elongator components (Fig. 3B). Purified recombinant His-tagged Elp1 and His-tagged Elp2 proteins were obtained via expression in insect cells and recombinant GST-Elp5 via expression in bacteria, and Elp4 and Elp6 were transcribed/translated in vitro. Unfortunately, despite attempts in several expression systems, we were unable to obtain soluble recombinant or in vitro transcribed/translated Elp3 protein. Examples of the results obtained are shown in Fig. 3B. Of all of the many combinations that were tested, only a direct interaction between Elp5 and Elp6 was detected.
Taken together, these results make it possible to propose a model to describe the molecular architecture of the Elongator complex (Fig. 4). In the large core-Elongator subcomplex, Elp1 protein interacts with Elp3 but does not interact strongly with Elp2 in the absence of Elp3. This suggests a direct Elp2-Elp3 interaction. On the other hand, the small subcomplex is formed based on direct interactions between Elp4 and Elp6 and between Elp6 and Elp5. Although not absolutely required, the Elp2 protein serves to somewhat stabilize the core Elongator-Elp4/5/6 interaction, whereas Elp3 is essential for the integrity of the complex. Furthermore, a weak interaction between Elp3 and Elp4 was detected. Based on these results we suggest that, rather than relying solely on strong binary interactions between individual subunits of the respective subcomplexes, the formation of the small subcomplex creates new interaction surfaces to enable contacts with Elp3 and Elp2, as well as with novel interaction surfaces created by the association of Elp1, Elp2, and Elp3 into core-Elongator (Fig. 4).
The WD40 Repeat Protein, Elp2, Is Dispensable for the in Vitro HAT Activity of Elongator-The data presented above revealed that although an elp2 strain displayed the typical Elongator phenotypes (5), the remaining five subunits still exist as a complex in these cells. To investigate whether the five-subunit elp2⌬ complex also retained Elongator activity, we purified the "mutant" complex and tested its in vitro HAT activity. This was particularly relevant, because several HAT and histone deacetylase complexes harbor subunits, which, like Elp2, contain WD40 repeats. It has thus been suggested that WD40 repeats might be involved in contacting histones to facilitate the acetyltransferase and deacetylation reactions, respectively (18,19). The purification scheme employed was identical to that used previously for the purification of a six-subunit Elongator complex from cells carrying a double affinity-tagged Elp1 protein (13). As indicated by Fig. 5A, a purified elp2⌬ Elongator complex from two independent purifications retained the five other subunits, although the three smaller subunits as well as the Elp3 protein appeared to be somewhat substoichiometric compared with the wild type complex. To be able to compare the HAT activity of the elp2⌬ complex with that purified from wild type cells, we therefore used similar amounts of the catalytic Elp3 subunit in the HAT reactions, with the consequence that other subunits, particularly Elp1, were present in larger amounts in the reactions containing the mutant five-subunit complex. Fig. 5B shows the result of HAT assays comparing wild type and the two independently purified mutant complexes. Surprisingly, Elongator retained in vitro HAT activity directed against H3 even in the absence of Elp2. Adding recombinant, His-tagged Elp2 to the reactions did not dramatically affect the activity (the apparent stimulation of wild type complex (Fig. 5B, lanes 1 and 2) was because of loading differences). These data demonstrate that the WD40 repeat protein Elp2 is not required for Elongator HAT activity in vitro.
Elp3, but Not Elp2 or Elp4, Is Required for RNA Binding in Vivo-We have recently shown that the Elongator complex binds to RNA in vitro and that it is in close vicinity of nascent pre-mRNA in vivo (12). These results indicate that Elongator is associated with active genes in living cells. To investigate the subunit requirements for RNA association in vivo, we compared Elongator RNA-immunoprecipitation in wild type, elp2, and elp4 cells expressing myc-tagged Elp3 (Fig. 6A). As shown previously (12), efficient cross-linking of Elp3 protein to GAL1 mRNA was observed in wild type cells (Fig. 6A, compare Elp3myc with no tag). Interestingly, although the levels were somewhat reduced, RNA interaction was also observed in cells lacking Elp2 (a component of core Elongator) and Elp4 (a core component of the small Elongator subcomplex). This indicates that neither the small subcomplex, nor Elp2 play crucial roles in the association of Elongator (Elp3) with transcripts in vivo. Because Elp3 levels are dramatically reduced in cells lacking Elp1, we could not investigate whether Elp1 is required for Elongator-RNA interaction. However, by tagging Elp1 rather than Elp3, the importance of the Elp3 protein for Elongator-RNA interaction could be investigated (Fig. 6B). Remarkably, Elongator-RNA interaction was ϳ5-fold reduced in the absence of Elp3, and this dramatic decrease was not because of changes in the expression level of the Elp1 protein (Fig. 6B, bottom  panel). Taken together, these data indicate that the Elp3 protein plays a crucial role for the association of Elongator with active genes. DISCUSSION The results presented here represent the first comprehensive investigation of the structure-function relationship of Elongator, a histone acetyltransferase complex with a role in RNAPII transcription. Our results reveal the detailed molecular architecture of the complex and suggest roles, or unexpected lack of roles, for individual subunits.
Previous work (13)(14)(15) showed that holo-Elongator is composed of two weakly associated subcomplexes. Subsequent Elp2 subunit is present only in wild type, but this Western blot is not shown. B, HAT activity of wild type and mutant Elongator complex with or without the addition of Elp2 as indicated. Please note that the activity of Elongator in HAT assays differs significantly from preparation to preparation and with freeze-thawing. It is therefore not possible to make a definitive judgment on the possible minor quantitative differences in activity, if any, between wild type and mutant Elongator.
FIG. 6. Elp3, but not Elp2 and Elp4, is required for Elongator-RNA association in vivo. RNA-immunoprecipitation experiments in strains expressing myc-tagged Elongator subunits. GAL1 mRNA was quantitated using reverse transcriptase real time PCR. Bars represent the average of two independent experiments. Error bars indicate variance. A, RNA-immunoprecipitation experiments with the indicated strains expressing Elp3-myc or the untagged control. B, RNA-immunoprecipitation experiments with the indicated strains expressing Elp1myc or the untagged control. Bottom panel shows the result of an Elp1-myc Western blot with loading control (Ponceau S-stained membrane section) to demonstrate that Elp1 is stable in elp3 cells. The resolution of Elp1 into two distinct bands has previously been described (22). work by Schaffrath and co-workers (20, 21) using immunoprecipitation of tagged Elongator subunits and detection of a coprecipitated differently tagged subunit gave some information about subunit requirements for Elongator stability. Although the present work confirms and significantly extends the conclusions derived from these studies, there are also surprising contradictions. For example, Frohloff et al. (21) detected the Elp3 protein in cells lacking ELP1, whereas our results indicate that normal Elp3 stability in vivo requires ELP1 (Fig. 1). Likewise, these authors concluded that the structural integrity of the small Elp4/5/6 subcomplex requires the ELP4, ELP5, and ELP6 genes, whereas our work shows that Elp4 can interact with Elp6 in the absence of ELP5 (Fig. 2C). The reasons for these differences are presently unclear.
One of the theses for the architecture of the Elongator complex before starting our studies was that the WD40 repeatcontaining Elp2 protein would turn out to either be essential for the integrity of the Elongator complex, or be required for its HAT activity. WD40 repeats are protein-protein interaction domains, and WD40-repeat proteins have been identified in other histone modifying/interacting proteins (18,19). Surprisingly, Elp2 turned out to be dispensable for both; a five-subunit elp2⌬ Elongator complex can be isolated, and this complex retains the ability to acetylate histones (Fig. 5). Elongator complex is also capable of binding RNA in the absence of ELP2 (Fig. 6). Data from Schaffrath and co-workers (20) suggests that the role of Elp2 might be to allow interactions with other proteins, such as Kti12/Tot4. The Elongator-interacting protein, Kti12/Tot4 has been proposed to bridge interactions between RNAPII and Elongator. For example, the protein was shown to co-immunoprecipitate promoter DNA (chromatin immunoprecipitation) from the ADH1 gene and to also be able to co-immunoprecipitate hyperphosphorylated RNAPII (20,21). These authors also reported co-immunoprecipitation of hyperphosphorylated RNAPII with Elongator from the DNA-free soluble fraction of a yeast whole cell extract (21). For unknown reasons, we have so far not been able to reproduce these coimmunoprecipitation results, although we can detect Elongator-Kti12 interaction. 3 We therefore presently favor the idea that besides the hyperphosphorylated C-terminal domain, RNA in the ternary complex might play an important stabilizing role for the Elongator-RNAPII interaction (12).
In contrast to the Elp2 protein, the Elp3 subunit appears to play a crucial role for all Elongator functions. It is the catalytic subunit (2), it is crucial for the integrity of the holo-Elongator complex (Fig. 2), and it is essential for RNA binding (Fig. 6). In light of the fact that Elp3, but not the other subunits, is conserved from Archaea to man (2), this is perhaps not surprising. It thus seems reasonable to expect that the fundamental functions of the Elongator complex be supplied by the abilities intrinsic to Elp3 with the other subunits playing primarily function/augmentative or regulatory roles.