A Conserved Motif Is Prerequisite for the Interaction of NAC with Ribosomal Protein L23 and Nascent Chains*

In eukaryotes, newly synthesized proteins interact co-translationally with a multitude of different ribosome-bound factors and chaperones including the conserved heterodimeric nascent polypeptide-associated complex (NAC) and a Hsp40/70-based chaperone system. These factors are thought to play an important role in protein folding and targeting, yet their specific ribosomal localizations, which are prerequisite for their functions, remain elusive. This study describes the ribosomal localization of NAC and the molecular details by which NAC is able to contact the ribosome and gain access to nascent polypeptides. We identified a conserved RRK(X)nKK ribosome binding motif within the β-subunit of NAC that is essential for the entire NAC complex to attach to ribosomes and allow for its interaction with nascent polypeptide chains. The motif localizes within a potential loop region between two predicted α-helices in the N terminus of βNAC. This N-terminal βNAC ribosome-binding domain was completely portable and sufficient to target an otherwise cytosolic protein to the ribosome. NAC modified with a UV-activatable cross-linker within its ribosome binding motif specifically cross-linked to L23 ribosomal protein family members at the exit site of the ribosome, providing the first evidence of NAC-L23 interaction in the context of the ribosome. Mutations of L23 reduced NAC ribosome binding in vivo and in vitro, whereas other eukaryotic ribosome-associated factors such as the Hsp70/40 chaperones Ssb or Zuotin were unaffected. We conclude that NAC employs a conserved ribosome binding domain to position itself on the L23 ribosomal protein adjacent to the nascent polypeptide exit site.

In eukaryotes, newly synthesized proteins interact co-translationally with a multitude of different ribosome-bound factors and chaperones including the conserved heterodimeric nascent polypeptide-associated complex (NAC) and a Hsp40/70-based chaperone system. These factors are thought to play an important role in protein folding and targeting, yet their specific ribosomal localizations, which are prerequisite for their functions, remain elusive. This study describes the ribosomal localization of NAC and the molecular details by which NAC is able to contact the ribosome and gain access to nascent polypeptides. We identified a conserved RRK(X) n KK ribosome binding motif within the ␤-subunit of NAC that is essential for the entire NAC complex to attach to ribosomes and allow for its interaction with nascent polypeptide chains. The motif localizes within a potential loop region between two predicted ␣-helices in the N terminus of ␤NAC. This N-terminal ␤NAC ribosome-binding domain was completely portable and sufficient to target an otherwise cytosolic protein to the ribosome. NAC modified with a UV-activatable cross-linker within its ribosome binding motif specifically cross-linked to L23 ribosomal protein family members at the exit site of the ribosome, providing the first evidence of NAC-L23 interaction in the context of the ribosome. Mutations of L23 reduced NAC ribosome binding in vivo and in vitro, whereas other eukaryotic ribosome-associated factors such as the Hsp70/40 chaperones Ssb or Zuotin were unaffected. We conclude that NAC employs a conserved ribosome binding domain to position itself on the L23 ribosomal protein adjacent to the nascent polypeptide exit site.
In bacteria, the ribosome-associated chaperone trigger factor (TF) 2 interacts with nascent polypeptides that emerge from the ribosomal exit tunnel. TF uses an exposed loop region within its N terminus to bind to the ribosomal exit site protein L23 (1)(2)(3)(4)(5). This binding to L23 is crucial for the function of TF as a molecular chaperone for nascent polypeptides. The early access of TF to nascent polypeptide chains allows for the controlled entry into the protein folding pathway during ongoing biosynthesis. Whereas TF is only present in eubacteria and chloroplasts, eukaryotes possess other ribosome-associated factors belonging to protein families unrelated to TF (6 -8). In yeast and metazoans, the Hsp70type chaperone Ssz and its DnaJ co-chaperone Zuotin (Zuo) form a ribosome-associated complex (RAC) that is tethered to the ribosome via Zuo (9 -11). In yeast, RAC together with another ribosome-bound Hsp70 homolog, Ssb, forms a functional ribosomal chaperone triad (9,12). In this triad, only Ssb is in direct association with nascent chains. Deletion of either one or all triad members leads to similar phenotypes including poor growth during osmotic stress, cold-sensitivity, and hypersensitivity to aminoglycosides. So far, the molecular basis of these phenotypes remains enigmatic.
In addition to the Hsp70-RAC machinery, a protein complex termed nascent polypeptide-associated complex (NAC) associates with ribosomes and nascent chains in an apparent 1:1 stoichiometry (13)(14)(15). NAC is highly conserved among eukaryotes and consists of two subunits (␣-and ␤NAC) that are both in direct contact with nascent polypeptide chains (13); yet, ␤NAC alone is responsible for binding to the ribosome (16). The yeast genome encodes three known NAC homologs consisting of a single ␣-subunit (encoded by EGD2) and two ␤ subunits (encoded by EGD1 and BTT1). Both ␤ subunits can form heterodimeric complexes with ␣NAC, although Btt1 is significantly less abundant than Egd1 (17,18). The general significance of NAC is emphasized by the embryonic lethality of NAC mutants in mice, nematodes, and fruit flies (19 -21). In contrast, deletion of NAC in yeast is not lethal and growth defects at high temperature are strain dependent (17). 3 It was proposed that NAC has a role in controlling the correct translocation of proteins to the endoplasmic reticulum by regulating the accessibility of the signal recognition particle (SRP) and the translocation pore to ribosome-nascent chain complexes (22,23), yet in vivo data to support this finding is still lacking. The observations that NAC associates with ribosomes and cross-links to short nascent polypeptides (13) triggered the speculation that NAC might play a role in the folding of newly synthesized proteins, however, direct evidence to support this hypothesis has not yet been presented.
An important step toward understanding the function of NAC at the ribosome is to define the specific strategy that is used to bind to the ribosome and subsequently interact with nascent chains. Therefore, we set out to gain insights into the molecular details and mechanism of the ribosomal association of NAC. The goals therein included (i) the mapping of a specific binding site of the NAC on the ribosomal surface; (ii) the identification of amino acid residues and structural components that are critical for NAC ribosome binding; and (iii) to determine whether nature has developed similar strategies to mediate the binding of factors that act co-translationally to the ribosome.
Surprisingly, we found that despite the lack of any sequence homology, an intriguing resemblance exists between the bacterial TF and eukaryotic NAC with regard to their ribosomal interplay. Our findings characterize the ribosomal protein L23 as a universally conserved docking platform in different kingdoms of life and show that unrelated ribosome-associated factors share a common strategy for their prime access to nascent polypeptide chains.
The plasmid pL23 (1) was used as a template for site-directed mutagenesis to generate the pL23-SEKAS/AAKAA mutant plasmid, where the TF binding site residues have been mutated (see Fig. 3C and text for details). The Rosetta plasmid was purchased from Novagen.
For expression and purification of NAC (WT or mutant variants) we constructed the plasmid pQE-NAC expressing N-terminal His 6 -tagged EGD2 and untagged EGD1 as an operon. To this end, the EGD1 open reading frame was PCR-amplified from yeast genomic DNA and subcloned into the pTrc-His 6 (1) plasmid to make pTrc-His 6 -EGD1. Using primers that remove the His 6 tag and introduce an upstream ribosome binding sequence, EGD1 was PCR-amplified and inserted into the plasmid pQE31-EGD2 (18) immediately 3Ј to the end of the His 6 -EGD2 open reading frame, resulting in plasmid pQE-NAC, which expresses His 6 -EGD2 and EGD1 as an operon. To create the pQE-NAC-FUS plasmid, an insert encoding the His-tagged Egd2 protein was PCR amplified from plasmid pQE31-EGD2 using primers that omit the terminal EGD2 codon and add a flexible Ser-Gly-rich linker (-AGSSGENLYFQS-GAGASGS-) such that it is in-frame with the EGD1 open reading frame when inserted into the pQE30-EGD1 (18) plasmid. Plasmids pQE-NAC and pQE-NAC-FUS were used as templates for site-directed mutagenesis to generate the Cys mutants (N18C, G22C, L28C, A32C, and A36C) used for cross-linking experiments (the number of the corresponding Egd1 residue is indicated). EGD1 (including endogenous promoter and terminator) was PCR-amplified from yeast genomic DNA and cloned into pRS316 (26) to generate pRS316-EGD1. pRS316-EGD1 was used as a template for site-directed mutagenesis to generate plasmids expressing mutant forms of EGD1 including the EKL/AAA (E7A/K8A/L9A), KL/AA (K14A/L15A), RRK/AAA (R25A/R26A/K27A), and KK/AA (K30A/K31A) mutations. To construct the pRS413-RPL25 plasmid, RPL25 (including endogenous promoter and terminator) was PCR amplified from yeast genomic DNA and was first cloned into the pBluescript vector and then inserted into the yeast pRS413 (26) vector. The plasmid expressing RPL25-GFP (27) was a kind gift of E. Hurt. Plasmids containing His 5 -CHIP (pGM10-His 5 -CHIP) and untagged hCHIP (pUHE21-hCHIP) were described (28). The plasmid pHLH-CHIP was created by annealing primers that coded for the N-terminal 55 residues of Egd1 and ligating them in-frame into a pTrc99B vector (29) that contained the CHIP coding sequence. The plasmid pET-ICDH (isocitrate dehydrogenase) was used to generate an arrested nascent chain of 173 aa in the in vitro transcription/translation system as described (30,31). To generate arrested nascent chains containing the lepB leader peptidase, the plasmid pET-STREP-lepB1-30-Arrest was constructed that fused an N-terminal STREP tag to aa 1-30 of lepB, followed by the SecM arrest sequence (32) to stall translating ribosomes. All plasmids constructed in this study were verified by sequencing and production of target proteins was confirmed.
Protein Purification-The yeast NAC complex, NAC-FUS, CHIP, HLH-CHIP proteins, and corresponding mutants were expressed under an isopropyl 1-thio-␤-D-galactopyranoside-inducible promoter and purified under native conditions from E. coli strain MH1. High level expression of the NAC complex was achieved by co-expressing rare tRNAs encoded by the Rosetta plasmid, resulting in the predominantly soluble NAC complex, which could be purified in high quantities via a His 6 tag present at the N terminus of the ␣-subunit. Initial purification of all proteins was achieved using Ni 2ϩ -nitrilotriacetic acid-agarose (Qiagen). The purification of all NAC complex and fusion protein variants included an additional purification step using POROS 20 SP material (PE Biosystems/GE Healthcare-Pharmacia) and dialysis of the highly pure protein-containing fractions against a buffer containing 20 mM Tris/HCl, pH 7.0, 50 mM NaCl, 1 mM EDTA, and 5 mM MgCl 2 . Protein concentrations were determined using the Bradford assay (Bio-Rad). The structural integrity of purified NAC was assessed by circular dichroism spectroscopy according to the described protocol (33). 4 In Vitro and ex Vivo Ribosome Binding Assays-Isolation of ribosomes under high salt conditions from E. coli (33) and yeast (34) and subsequent in vitro rebinding experiments were performed as described. Previously published procedures were used for the ex vivo isolation of ribosomes and associated complexes from E. coli cells (33). Yeast lysates were prepared by growing 25 ml of cells to logarithmic growth phase, cells were pelleted and resuspended in 300 l of lysis buffer (50 mM Tris-HCl, pH 7.5, 0.1 M EDTA, 2 mM phenylmethylsulfonyl fluoride, Complete tablet (Roche), 5 mM MgCl 2 , 0.1 mM dithiothreitol, 100 mM KCl). Acid-washed glass beads were added to the samples and lysis was achieved via standard protocols (24). Lysates were cleared for 10 min at 4°C at 12,000 rpm, and the resultant supernatant was divided into 3 parts: total lysate, a "low salt" sample that was loaded onto a 100 mM KCl 20% sucrose cushion, and a "high salt" sample that was adjusted to a 600 mM final salt concentration and was loaded onto a 600 mM KCl, 20% sucrose cushion. Ribosomes were separated from the rest of the lysate by centrifuging the samples in a TLA100.2 rotor for 90 min, 200,000 ϫ g, and the samples were then analyzed by SDS-PAGE and Western blotting. The precise identity of the faster migrating ␤NAC-derived band that appeared after prolonged incubation with ribosomes ( Fig. 1C) was achieved by mass determination using microflow electrospray ionization mass spectrometry (35).
Generation and Identification of NAC-Ribosome and NAC-Nascent Chain (NC) Cross-links-The UV-activatable cross-linker BPIA (benzophenone-4-iodoacetamide, Molecular Probes) was coupled to the engineered Cys residues of the purified NAC complex or fusion protein (NAC-FUS) according to the described protocol (1,36). The labeled proteins were incubated with either E. coli or yeast ribosomes for 30 min, then were exposed on ice to UV light (365 nm, 100 W; model B-100AP, Ultraviolet Products) at 5 cm for 20 (E. coli ribosome-containing samples) or 60 min (yeast ribosome-containing samples) before loading the reactions onto 20% sucrose cushions to separate soluble material (supernatant, S) from ribosome-bound material (pellet, P) by centrifugation at 200,000 ϫ g in a TLA 100 rotor. Identification of crosslinked NAC-ribosome products by mass spectrometry was performed according to described protocols (1,35). We followed previously described protocols (31,37) to cross-link NAC and TF to radiolabeled [ 35 S]Met-arrested nascent chains generated in an in vitro transcription/ translation system derived from E. coli lysate.
Antibodies and Miscellaneous-Antibodies against Ssb and Zuotin were kindly provided by E. A. Craig, and T. Lithgow generously provided antibodies against the Egd1, Egd2, and the NAC complex. Production of antibodies against Rpl25 (L25) (34) and L23 (1) were described previously. Protein structure prediction was carried out using Jpred (38), and sequence alignments were prepared using ClustalW (39) and Jalview (40).

NAC Binds to Yeast and E. coli Ribosomes in Vitro-
To characterize the binding of NAC to ribosomes in vitro, we incubated recombinantly purified yeast NAC with high salt-washed ribosomes from Saccharomyces cerevisiae in a 3:1 molar ratio, the mixture was loaded onto a sucrose cushion whereby unbound NAC (supernatant) was separated from ribosome-associated NAC protein in the pellet by centrifugation. We monitored NAC association with 80 S ribosomes by Western blotting against the ␤NAC subunit throughout all experiments. Approximately 30% of NAC was recovered in the ribosomal pellet confirming a 1:1 binding stoichiometry to yeast ribosomes in vitro. NAC incubated without ribosomes was exclusively found in the supernatant (Fig. 1A, lane 1). The addition of other purified yeast ribosome-associated factors to the reactions, namely Zuotin and Ssb, did not interfere with the stoichiometric binding of NAC to these ribosomes (data not shown), indicating that NAC and the components of the yeast chaperone triad occupy distinct non-overlapping positions on the eukaryotic ribosome. To determine whether NAC binds to a conserved ribosomal element, we additionally tested NAC binding to purified E. coli 70 S ribosomes. Interestingly, NAC also efficiently associated with E. coli ribosomes in an apparent 1:1 stoichiometry (Fig. 1B, lanes 3 and 4), indicating that NAC binds to a region on the ribosomes that is conserved among bacteria and eukaryotes.
We observed that occasionally during the incubation of NAC with ribosomes, the ␤-subunit, which normally migrates at ϳ20 kDa, was partially degraded resulting in a faster migrating ␤-fragment with an apparent molecular mass of 15 kDa (Fig. 1C). This degradation was most likely caused by a co-purifying proteolytic activity present in our ribosome preparation, which could be eliminated by adding protease inhibitors during the reaction or by purifying E. coli ribosomes from a strain that lacks multiple proteases. Importantly, the degraded ␤NAC fragment was not detected in the ribosomal pellet fraction indicating that the proteolytic cleavage affects a region of ␤NAC that is involved in ribosome binding. By mass spectrometry we identified that the degraded fragment represents a ␤NAC fragment lacking the N-terminal 30 aa. A similar ␤NAC fragment (⌬1-31) deficient in ribosome binding had been identified by tryptic digest of ␤NAC (41).
A Conserved Motif in the N Terminus of ␤NAC Is Involved in Ribosome Binding-To further investigate the contribution of the N terminus of ␤NAC in ribosome binding, we aligned the N termini of ␤NAC from different species and searched for conserved aa residues ( Fig. 2A).
A predicted unstructured loop region consisting of residues 15-40 of the yeast ␤NAC contained an entirely conserved RRK(X) n KK motif of positively charged aa side chains. This prediction, in combination with our observation that the loop region is proteolytically sensitive (Fig. 1C and data not shown), supported our hypothesis that this N-terminal region of ␤NAC might be involved in ribosome binding using an exposed and positively charged motif to contact the ribosomal surface.
To test our assumption, we replaced the RRK as well as the KK residues within the ␤NAC motif with alanine residues, resulting in the NAC RRK/AAA and NAC KK/AA mutants of NAC, respectively ( Fig. 2A). Because it was previously reported that deletion of the first 11 N-terminal aa of ␤NAC abolished ribosome binding (42), we also mutated conserved residues within the 11 N-terminal aa and exchanged the conserved residues EKL and KL to alanines resulting in NAC EKL/AAA and NAC KL/AA ( Fig. 2A). Importantly, none of these purified ␤NAC variants were impaired in their ability to form a stable complex with ␣NAC, and their structural integrity was similar to wild-type NAC as judged by solubility, circular dichroism measurements, and partial proteinase K digest (data not shown). 4 To test whether the NAC mutants could associate with ribosomes in vitro, we performed a ribosomal rebinding experiment as described above to separate ribosome-NAC complexes from unbound NAC. The purified NAC complexes alone could only be found in the supernatant (S, unbound) fraction ( Fig. 1A and data not shown). When the NAC RRK/AAA and NAC KK/AA loop mutants were incubated in the presence of yeast 80 S ribosomes, a severe reduction in the ability to bind to ribosomes in vitro was observed as shown by the presence of the majority of the ␤NAC in the soluble fraction (Fig. 2B, lanes 3 and  FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5 FIGURE 2. A conserved motif in the N terminus of ␤NAC mediates ribosome binding. A, secondary structure prediction of ␤NAC (␣-helices are shown in red, ␤-sheets are indicated by arrows in yellow) and alignment of N-terminal regions of ␤NAC subunits from different species. Identical aa are boxed in yellow and conserved residues that were mutated in this study are shown in blue. B, NAC variants (2 M) were incubated with or without purified ribosomes (1 M) from different species and NAC-ribosome complexes (P) were separated from unbound NAC (S) as described in the legend to Fig. 1. Association of NAC with ribosomes was detected by Western blotting using antibodies against ␤NAC. L23 and L25 Western blots 5). In the same experiment, the NAC EKL/AAA and NAC KL/AA helix mutants were not severely reduced in ribosome binding (Fig. 2B,  lanes 7-10). Very similar results were obtained when NAC binding to purified E. coli ribosomes was investigated. The two loop mutants RRK/AAA and KK/AA revealed a deficiency in binding to E. coli ribosomes supporting the finding that NAC specifically associates with E. coli ribosomes via the conserved RRK(X) n KK "NAC signature" motif (Fig. 2B, lanes 13-16). Because the RRK/AAA mutation in ␤NAC causes this variant to migrate faster in SDS-PAGE compared with WT ␤NAC (Fig. 2B), we were able to perform a competition experiment in which we could monitor both the WT and mutant protein simultaneously. We found that the RRK/AAA mutant could not compete with WT NAC for binding to yeast ribosomes even when added in a 10-fold excess over ribosomes emphasizing the severe ribosome binding defect of the loop mutant in vitro (Fig. 2C, lanes 11 and 12).

Insights into NAC Interaction with the Ribosome
Next, we investigated the ribosome association of the two NAC loop mutants in vivo. To do so, we ectopically expressed either WT or mutant ␤NAC/EGD1 under the control of its own promoter to allow for physiological levels of production in a yeast mutant strain lacking the ␤NAC subunit (egd1⌬). Cells were harvested in the logarithmic growth phase and after cell lysis ribosomes were isolated by sedimentation through a sucrose cushion using physiological salt concentration (100 mM KCl). Wild-type ␤NAC co-sedimented together with ribosomes in the pellet fraction (Fig. 2D, lane 3). Increasing the ionic strength up to 600 mM KCl during the isolation procedure caused the quantitative release of WT NAC from ribosomes (data not shown) in agreement with published data reporting the salt-dependent NAC binding to ribosomes (43). Neither ␤NAC RRK/AAA nor ␤NAC KK/AA co-sedimented with ribosomes and were exclusively found in the soluble fraction (Fig. 2D, lanes 6 and  9). Importantly, the loop mutations in ␤NAC also abolished the association of ␣NAC with ribosomes, clearly indicating that the NAC complex requires the RRK(X) n KK motif of the ␤NAC subunit to be targeted to ribosomes. To investigate whether ribosome association of the yeast triad (Ssb/Ssz/Zuo), a second ribosome-associated chaperone complex that contacts the nascent polypeptide chains on the yeast ribosome, is influenced by the loss of NAC binding, we checked the co-purification of Zuotin and Ssb with ribosomes in parallel. Co-sedimentation of these ribosome-bound chaperones was similar in all yeast strains tested and thus their binding to ribosomes is independent of the association of NAC with ribosomes (data not shown).
Ribosome-binding Domain of ␤NAC Is Portable-The loop in the N-terminal region of ␤NAC containing the NAC signature binding motif is flanked on both sides by ␣-helices, thus forming a helix-loophelix (HLH) structure ( Fig. 2A). We therefore tested whether this HLH, containing the NAC signature, was the equivalent of a portable domain that could be used to target an otherwise completely unrelated soluble cytosolic protein to the ribosome. For this purpose, we selected the human E3 ubiquitin ligase CHIP, a protein implicated in the discrimination between folding and degradation pathways in mammalian cells with no known homologs in bacteria or yeast (44,45). We fused the N-terminal 55 aa of ␤NAC, corresponding to the HLH, to the N terminus of CHIP, thus generating the HLH-CHIP chimera. Like NAC, HLH-CHIP bound quantitatively to purified yeast and E. coli ribosomes in vitro when HLH-CHIP and NAC were incubated with the ribosomes at an equimolar ratio (Fig. 2E). Furthermore, when E. coli lysates prepared from logarithmically growing cells expressing either WT CHIP or HLH-CHIP were separated into ribosome-containing (P) and ribosome-free (S) fractions at physiological salt conditions (100 mM KCl), only HLH-CHIP was observed in the ribosome-containing fraction (Fig. 2F, compare lanes 3 and 8). This in vivo association of HLH-CHIP with ribosomes could be disrupted when ionic concentrations were increased to 600 mM KCl (Fig. 2F, compare lanes 8 and 10). Thus, the HLH of ␤NAC is completely portable and can generally function as a unit that imparts the ability of an otherwise soluble protein to bind to ribosomes from different kingdoms both in vivo and in vitro.
NAC Cross-links to L23 and L29-To identify the ribosomal docking partner of NAC we used a covalent cross-linking approach using NAC variants that had been modified with a UV-activatable cross-linker. Cross-linked adducts of NAC with ribosomal proteins visible on Coomassie-stained gels could then be excised and identified by mass spectrometry. For this purpose, we created a NAC fusion protein for which we covalently linked ␣NAC and ␤NAC subunits by introducing a large flexible linker between the subunits. The NAC fusion (NAC-FUS) has a molecular mass of 47 kDa and thus fulfills the criteria that it does not migrate within the ensemble of ribosomal proteins in SDS-PAGE, which would complicate subsequent mass spectrometric analysis (Fig.  3A, lane 1). Importantly, the purified fusion protein efficiently associated with E. coli and yeast ribosomes in vitro comparable with WT NAC, indicating that the covalent linkage did not interfere with its ribosome binding properties (data not shown and Fig. 3A, lane 8). In the next step, we replaced the Leu residue within the NAC signature (-RRKLNKK-) of the NAC fusion with cysteine. The lack of additional cysteines in the NAC fusion allowed us to attach the thiol-specific UV activatable cross-linker BPIA specifically to this engineered cysteine residue. We chose to first use E. coli ribosomes in our cross-linking experiments because (i) they could be isolated at a higher level of purity than yeast ribosomes, and (ii) NAC could specifically associate with E. coli ribosomes in a manner dependent on an NAC signature motif (Fig. 2B). After being coupled with BPIA, the NAC fusion (NAC-FUS-BPIA) was incubated with purified E. coli ribosomes. The samples were UV-irradiated and samples were removed at specific time intervals. After UV irradiation, the samples were separated into ribosome-free and ribosome-containing fractions as described previously. UV irradiation resulted in two dominant cross-linking products with a relative molecular mass of about 55 and 60 kDa in the presence of ribosomes (Fig. 3A, lanes 10 and 12), whereas no cross-linking product was observed after 15 min of UV exposure in the absence of ribosomes (Fig.  3A, lanes 1-6), or in the presence of ribosomes without UV exposure (Fig. 3A, lane 8). We verified by Western blotting that these crosslinking products contain the ␤NAC subunit (Fig. 3A, lower panel). Upon prolonged UV exposure some additional high molecular weight were performed in parallel to control the experimental setup and amount of loaded ribosomes. C, Western blot against ␤NAC showing a competition experiment for ribosome binding. Yeast ribosomes, WT NAC, and NAC RRK/AAA mutants were co-incubated at the concentrations indicated and subsequently subjected to sucrose cushion centrifugation to isolate unbound (S) and ribosome-bound NAC (P). L25 Western blot was performed in parallel as a control (not shown). D, total lysates (T) prepared from exponentially growing egd1⌬ yeast cells expressing either plasmid-encoded WT or mutant ␤NAC subunits were separated into post-ribosomal supernatant (S) and ribosomal pellet (P). Samples were separated by SDS-PAGE and subsequently analyzed by Western blotting to detect ribosome-associated chaperones as indicated. An anti-L25 immunoblot is shown as control. E, NAC and HLH-CHIP (2 M) were incubated in vitro with purified ribosomes (2 M) from yeast (left panel) or E. coli (right panel). Ribosome-bound complexes (P) were separated from unbound protein (S) as described in the legend to Fig. 1. Association of NAC or HLH-CHIP with ribosomes was detected by Western blotting using antibodies against ␤NAC, which efficiently recognizes both proteins. L23, L25, and CHIP Western blots were performed in parallel as a control (not shown). F, the binding of HLH-CHIP to ribosomes in vivo is demonstrated by generating total lysates (T) prepared from exponentially growing E. coli cells expressing either WT CHIP (left panel) or HLH-CHIP (right panel), which were then separated into post-ribosomal supernatant (S) and ribosomal pellet (P) under physiological (100 mM KCl) or non-physiological (600 mM KCl) ionic conditions. Samples were separated by SDS-PAGE and subsequently analyzed by anti-CHIP immunoblot. An anti-L23 immunoblot was performed as control (not shown). FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5 cross-links were detectable by Western blotting, albeit much less abundant compared with the 55-and 60-kDa cross-linking products. The two prominent cross-link bands were excised from the Coomassiestained SDS-PAGE gel, digested with trypsin, and the mass and sequence of the resulting peptides were obtained by nanoelectrospray tandem mass spectrometry. Both cross-linking products contained multiple peptides corresponding to NAC. In addition, we found that the cross-linking product migrating at 60 kDa also contained two unambig- uous peptides derived from the ribosomal protein L23 and the 55-kDa product includes two peptides specifically corresponding to the ribosomal protein L29. By size, these cross-linking products corresponded to a 1:1 covalent interaction of NAC-FUS-BPIA/L23 or NAC-FUS-BPIA/ L29. The presence of these ribosomal proteins in the cross-links was confirmed by Western blotting using antibodies specific for E. coli L23 and L29, respectively (Fig. 3B, lane 4). Additional NAC contacts to the RNA moiety are possible, but were not detected here in this study. Similar results were obtained using four additional variants of NAC-FUS-BPIA labeled within the unstructured loop region either N-or C-terminal to the NAC signature (data not shown).

Insights into NAC Interaction with the Ribosome
To exclude that the labeled NAC fusion (NAC-FUS-BPIA) associated non-specifically with ribosomes, we added unlabeled WT NAC or NAC-FUS in a 5-fold excess to compete for ribosome binding. We observed a significant decrease in the amount of cross-linking products to L23 and L29 (Fig. 3B, lanes 6 and 8), whereas the addition of the NAC RRK/AAA mutant did not outcompete cross-linking to L23 and L29 (data not shown). We conclude that NAC specifically cross-links to L23 and L29. Both proteins are universally conserved in ribosomes and are located in close proximity to each other at the exit site of the ribosomal tunnel. Remarkably, the same proteins were identified earlier as binding partners of the bacterial ribosome-associated chaperone TF (1, 3-5). Indeed, cross-linking of the NAC fusion to L23 and L29 could be efficiently competed by the addition of a 5-fold excess of TF (Fig. 3B, lane  10), suggesting their co-occupation of a similar binding site at the ribosome.
NAC Interacts with the L23 Docking Site of TF-To further characterize the interaction of NAC with ribosomes, we investigated the crosslinking of NAC using E. coli mutant ribosomes either lacking the nonessential L29 protein or carrying point mutations in the conserved region of L23 (L23 SEKAS/AAKAA , positions 17-21 in E. coli L23, Fig. 3C). Point mutations in L23 such as SEKAS/AAKAA involving the con-served Glu 18 residue are known to significantly decrease ribosome association with bacterial TF (Ref. 1 and data not shown). Mutant ribosomes were purified from E. coli cells lacking TF and subsequently tested for NAC interaction in vitro. Cross-linking of the NAC fusion protein to ribosomes lacking L29 resulted in the loss of the L29 cross-link with NAC, whereas the L23 cross-link to NAC was maintained or even slightly increased in its intensity, indicating that NAC efficiently associates with ribosomes in the absence of L29 (Fig. 3D, lane 8). Using L23 SEKAS/AAKAA mutant ribosomes for cross-linking, we found that both cross-linking products were severely reduced (Fig. 3D, lane 12). This finding suggests that the NAC fusion is generally impaired in its ribosome association by the mutation in L23. Taken together, these data raise the hypothesis that NAC, like bacterial TF, contacts the ribosome via a similar surface exposed region in L23.

Mutation of NAC Results in a Functional Decrease in Nascent Chain
Interactions-The finding that although NAC does not exist in eubacteria, it can interact with and cross-link to the L23 protein located adjacent to the ribosomal exit tunnel inspired us to test whether NAC could also interact with nascent chains generated in a coupled in vitro transcription translation system derived from E. coli lysates that lack TF. Using previously described approaches (31,37) we tested the ability of purified NAC or the NAC signature mutant NAC RRK/AAA to cross-link to radiolabeled arrested NC derived from the LepB leader peptidase (Fig. 4A, 8 kDa) or isocitrate dehydrogenase (data not shown) proteins using the nonspecific cross-linker disuccinimidyl suberate. TF, which is known to cross-link to these nascent chains, was added to demonstrate the functionality of the assay (Fig. 4, A, lane 4, and B, lane 2). The addition of NAC to the reactions resulted in three NAC:NC cross-links, corresponding to the covalent interaction of the nascent chain with ␤NAC (30-kDa cross-link), ␣NAC (38 kDa), or the ␣␤NAC heterodimer (60 kDa) (Fig. 4, A, lane 2 and B, lane 3, and data not shown). The presence of each NAC component in the cross-linking bands was  FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5  Fig. 3A. After separation of the ribosomal pellet fraction (P) and the post-ribosomal supernatant (S) by centrifugation over a sucrose cushion, components of the NAC complex or L25 were identified by immunoblotting using anti-NAC (left panels) or anti-L25 antibodies (right panels), respectively. NAC/L25 crosslinking adducts are indicated with a star; WB, Western blot. C, ⌬rpl25 yeast strains expressing plasmid-encoded L25 variants were grown to logarithmic phase. Cells were harvested and soluble components (S) were separated from the ribosomal pellets (P) as described in the legend to Fig. 2D. Proteins of interest were detected using antibodies directed against ␤NAC, ␣NAC, Ssb, Zuo, and L25. The asterisk indicates a consistently appearing degradation product of the L25-GFP fusion protein.

Insights into NAC Interaction with the Ribosome
confirmed by immunoblots against the individual NAC components (data not shown). Importantly, the introduction of mutations in the NAC signature, in this case the mutant NAC RRK/AAA , eliminates the ability of all NAC components to interact with these arrested nascent chains (Fig. 4, A, lane 3, and B, lane 4, and data not shown). Furthermore, the ability of NAC to cross-link to arrested nascent chains can be outcompeted when TF is added to the reactions (Fig. 4B, compare lanes  3 and 5), showing that TF displaces NAC from the ribosome by competing for binding to their shared attachment site on the ribosome, L23.
NAC Cross-links to L25 on Yeast Ribosomes-Given the high degree of homology between the surface-exposed region of E. coli L23 and its yeast homolog L25 (Fig. 3C) and the observation that NAC binds to ribosomes of both kingdoms (Fig. 1) via its conserved signature loop (Fig. 2), we tested whether NAC could cross-link to purified yeast ribosomes in vitro. Applying the same approach used to identify L23 and L29 as cross-linking partners of the NAC fusion (NAC-FUS) on E. coli ribosomes, we investigated whether L25 serves as a docking site for NAC by probing for the co-existence of L25 (right panel of Fig. 5A) and NAC (left panel of Fig. 5A) in cross-link bands using an immunoblotting approach. The incubation of the NAC-FUS-BPIA fusion protein with yeast ribosomes results in a major cross-linking product that on SDS-PAGE runs at ϳ70 kDa and corresponds to a cross-link between NAC-FUS-BPIA and L25 (Fig. 5A, lanes 6 and 12). To rule out any nonspecific interactions of the NAC fusion protein with yeast ribosomes, and to confirm our results, we coupled the cross-linker BPIA to the same residues of the ␤NAC subunit and purified this labeled complex for in vitro cross-linking studies. The incubation and UV irradiation of only the NAC complex (Fig. 5B), built by the heterodimer of ␣NAC and ␤NAC-BPIA, resulted in 3 major cross-linking products with approximate sizes of 45, 55, and 60 kDa, corresponding to inter-and intramolecular crosslinks of the components of the NAC complex (Fig. 5B, lane 3). When the NAC complex was incubated with yeast ribosomes, additional crosslinking bands were observed (Fig. 5B, lane 6), one of which migrated at ϳ40 kDa and cross-reacted with both anti-NAC and anti-L25 antibodies (Fig. 5B, lanes 6 and 12) and corresponds in size to a 1:1 ␤NAC-BPIA/L25 cross-linking adduct. These data demonstrate for the first time an interaction of NAC with L23 family members in the ribosomal context and complement a recent study that showed an interaction between isolated L25 protein and NAC (41).
The identities of the other two cross-linked products appearing at ϳ37 and 43 kDa in the presence of ribosomes are unknown (Fig. 5B, lane  6). As was the case for NAC cross-linking to E. coli ribosomes (Fig. 3), these additional NAC cross-links are likely to correspond to ribosomal proteins that are present in the immediate vicinity of L25 on the surface of the ribosome.
Mutation in L25, the Yeast L23 Homolog, Reduce NAC Binding in Vitro and in Vivo-To verify that NAC associates with the yeast homolog of L23, L25, on the eukaryotic ribosome, we employed a yeast strain that expressed an L25-GFP fusion protein instead of WT L25. As reported earlier, L25 C-terminally tagged with GFP is functional in yeast, because it can replace authentic L25 (27). We confirmed that the RPL25-GFP gene fully complemented the loss of the essential RPL25 (codes for L25) gene and mutant cells grew comparably well to WT cells (data not shown), indicating that L25-GFP is efficiently incorporated into ribosomes and does not interfere with basic ribosome functions. We assumed that the presence of the 25-kDa GFP moiety, predicted to be at the surface of the ribosome in these cells, may sterically interfere with endogenous NAC binding to L25. Indeed, ribosomes isolated from L25-GFP expressing yeast cells exhibited decreased NAC binding under physiological salt conditions compared with WT ribosomes both in vivo ( Fig. 5C, lanes 3 and 6) and in vitro (data not shown). Moreover, we tested whether the yeast ribosome-associated chaperones Ssb and Zuo were altered in their ribosome binding because of the L25 mutation. Neither Ssb nor Zuo showed any pronounced difference in ribosome binding when co-purified with L25-GFP or WT ribosomes (Fig. 5C). Thus L25 specifically serves as the NAC docking site but is not involved in binding of Ssb or Zuo. Taken together that NAC cross-links to L25 and is reduced in binding to L25-GFP ribosomes, we confirm that NAC uses one and the same conserved binding surface presented by the L23 protein family to contact bacterial and eukaryotic ribosomes.

DISCUSSION
This study reveals that NAC contacts the surface of the L23 ribosomal protein family members (including L25 in yeast) via a conserved consensus motif rich in positively charged residues RRK(X) n KK (the NAC signature), located in the exposed loop of a predicted HLH motif in the N terminus of ␤NAC.
The highly conserved NAC signature is likely to constitute the major ribosome-binding component of all eukaryotic NACs including the second ␤NAC in yeast, Btt1, and the NACs of worms, flies, and humans. The previously observed loss of ribosome binding upon deletion of the 11 N-terminal aa of ␤NAC (42) might be attributed to local structural disturbances that may either affect the conformation and flexibility of the residual HLH region and/or the positioning of the NAC signature within the exposed loop structure. Interestingly, in the Archaea kingdom the ␤NAC subunit is absent and the NAC complex is instead built by a homodimer of two ␣NAC subunits. This Archaea NAC complex lacks a homologous HLH motif, yet it is still able to bind ribosomes via an unknown mechanism (46).
Our findings emphasize that L23 constitutes a general docking site for various factors involved in protein folding and targeting to gain early access to nascent polypeptides (Fig. 6). The best characterized example of this phenomenon is the binding of bacterial TF to L23 where it partakes in the folding program of nascent chains (1,3). Moreover, it is known that the SRP utilizes L23 as a major ribosomal contact point (47). Recent data indicate a direct interaction between SRP and ␣NAC in yeast (41). Our finding that ␤NAC permits the NAC complex to bind at the L23 site places NAC in close proximity to SRP and thus supports the previous observations that SRP and NAC can interact directly with each other (41) and compete for access to nascent polypeptides (13,48). Furthermore, ribosomes translating nascent chains that are destined to be translocated via the Sec translocase into the endoplasmic reticulum  FEBRUARY 3, 2006 • VOLUME 281 • NUMBER 5 make contact to the translocation channel at ribosomal proteins of the large subunit, including L25 (the L23 homolog) and L35 (L29 family member) (49). Thus, the occupation of L25 by NAC can now also explain the observation that NAC prevents the mistargeting of translating ribosomes to the endoplasmic reticulum membrane (22,23).

Insights into NAC Interaction with the Ribosome
The observation that NAC and these other factors compete with each other for binding to L23 raises the question of whether NAC has developed a novel mechanism to bind to L23, or do NAC and other factors employ a similar strategy to gain access to L23? Based on the following observations, several intriguing overlapping features among non-homologous ribosome binding factors are evident. The binding of NAC to ribosomes via its HLH motif containing the NAC signature is reminiscent of the strategy used by TF to bind to ribosomes, which involves the TF signature motif (GFRXGXXP) (1). The TF signature motif is localized in a flexible loop region that is flanked by two ␣-helical structures (2)(3)(4)(5). Mutations of the positively charged residues within the TF signature severely reduced ribosome binding of TF (1,3). Although rRNA contacts additionally contribute to the ribosome association of TF, a crucial contact is formed between the Arg of the TF signature and a conserved Glu residues of L23 (Glu 18 for E. coli L23, Fig. 3C) (2)(3)(4)(5). A mutation of the Glu residue in L23 causes the loss of the ribosome binding of TF (1,3). Additionally, the L23 binder SRP utilizes a fourhelix bundle to stabilize an exposed positively charged loop that come in close contact to the ribosome (50,51). Thus, the use of positively charged residues in an exposed loop inserted between helical structures seems to be a common theme for binding to the ribosome.
The portability of the N-terminal HLH ribosome-binding domain of ␤NAC that contains the positively charged NAC signature (Fig. 2, D and E) emphasizes that the functionality of this domain is not context-dependent and can target unrelated proteins to the ribosome in a general manner. Finally, the ability of NAC to prevent non-productive interactions of ribosomes with Sec61 at the endoplasmic reticulum membrane (22,23) is likely to critically depend on the association of NAC with the ribosome via its positively charged NAC signature motif. Such a competition between NAC and Sec61 for ribosome binding would be consistent with the recent finding that positively charged residues located in cytoplasmic loops, inserted between membrane-spanning helices, mediate the stable contact of Sec61 to ribosomes (52).
It is intriguing that TF in bacteria and NAC in eukaryotes occupy the same prime location on ribosomes to gain access to nascent chains. This result was surprising because no homology on the amino acid or structural level is evident. Despite their lack of sequence and structural homology, do NAC and TF function in a similar manner? In a recent study it was shown that Archaea ␣NAC, but not TF, partially restores the temperature-sensitive phenotype of a yeast NAC mutant (41). We intended to test the ability of NAC to complement the synthetic lethality of cells depleted of both TF and DnaK (53), but for unknown reasons NAC could not be stably expressed in ⌬dnaK cells. Instead, we employed an in vitro translational lysate to demonstrate that disruption of the NAC signature motif constitutes a disruption of NAC function. Indeed, NAC could efficiently cross-link to nascent chains, but only in a manner dependent on an intact NAC signature. Hence, the docking of NAC to the same ribosomal attachment point as bacterial TF underscores its primary access to nascent chains and supports the idea of a crucial role for NAC in protein biogenesis. Furthermore, our data demonstrates that NAC has a different docking site at the ribosome than ribosome-associated proteins Ssb and RAC (Ssz/Zuo). The action that NAC exerts on nascent chains remains a topic of debate and of further extensive investigations.