The α and β Subunit of the Nascent Polypeptide-associated Complex Have Distinct Functions

Nascent polypeptide-associated complex (NAC) is probably the first cytosolic protein to contact nascent polypeptide chains emerging from ribosomes. In this way NAC prevents inappropriate interactions with other factors. Eventually other factors involved in targeting and folding, like the Signal Recognition Particle or cytosolic chaperones, must gain access to the nascent chain. All NAC preparations to date consist of two copurifying polypeptides. Here we rigorously show that these two polypeptides, termed α- and βNAC, form a very stable complex in vivo and in vitroand that a functional complex can be reconstituted from the individual subunits. A dissection of the contributions of the individual subunits to NACs function revealed that both subunits are in direct contact with nascent polypeptide chains on the ribosome and that both contribute to the prevention of inappropriate interactions. However, βNAC alone directly binds to the ribosome and is sufficient to prevent ribosome binding to the endoplasmic reticulum membrane.

Nascent polypeptide-associated complex (NAC) is probably the first cytosolic protein to contact nascent polypeptide chains emerging from ribosomes. In this way NAC prevents inappropriate interactions with other factors. Eventually other factors involved in targeting and folding, like the Signal Recognition Particle or cytosolic chaperones, must gain access to the nascent chain. All NAC preparations to date consist of two copurifying polypeptides. Here we rigorously show that these two polypeptides, termed ␣and ␤NAC, form a very stable complex in vivo and in vitro and that a functional complex can be reconstituted from the individual subunits. A dissection of the contributions of the individual subunits to NACs function revealed that both subunits are in direct contact with nascent polypeptide chains on the ribosome and that both contribute to the prevention of inappropriate interactions. However, ␤NAC alone directly binds to the ribosome and is sufficient to prevent ribosome binding to the endoplasmic reticulum membrane.
Nascent polypeptide-associated complex (NAC) 1 is a very abundant cytosolic protein, which is involved in cotranslational targeting of polypeptides to the endoplasmic reticulum (ER) membrane. The intracellular NAC concentration varies only slightly in different tissues, ranging from 3 to 10 M (1). NAC is highly conserved among eukaryotes, but up to now no functional prokaryotic homolog has been described. The importance of NACs in vivo function is emphasized by early embryonically lethal phenotypes of NAC mutants in mice and fruit flies (2,3).
Recently it was also shown that yeast NAC, although not essential for growth (4), is involved in the import of proteins into mitochondria (5,6).
NAC was originally identified as a ribosome-associated protein when we set out to probe the molecular environment of growing polypeptides on the ribosome using a photo-cross-linking approach (7). With this technique we were able to show that NAC can interact with regions of nascent chains as close as 17 amino acids to the peptidyl transferase center. Additionally, it was shown by protease protection that NAC functions as a dissociable wall of a ribosomal tunnel through which the growing polypeptide emerges (8). Cycles of binding and releasing NAC expose the polypeptide to the cytosol in "quantal units" rather than amino acid by amino acid. We have proposed that exposing the polypeptide chain in functional units contributes to fidelity in cotranslational processes such as targeting and folding (9). Depleting NAC from translating ribosomes led to at least two inappropriate interactions. First, the signal recognition particle (SRP), which interacts with signal peptides of nascent chains on ribosomes during targeting to the ER membrane, could be cross-linked to nascent chains lacking signal peptides (7). Second, in the absence of NAC ribosomes translating non-secretory polypeptides inappropriately interacted with translocation sites on ER membranes and the signal-less chains were even translocated, although with low efficiency. Adding back purified NAC to the system restored the fidelity of cotranslational ribosome binding and translocation of only secretory polypeptides (10). Recently the function of NAC in ribosome targeting to the ER was questioned (11,12) in studies that used subphysiological concentrations of NAC. Although we were able to reproduce these results, we found that, when NAC was restored to physiological levels, our original observations were confirmed (1).
We have hypothesized that NAC normally prevents inappropriate interactions with nascent polypeptides and is thereby involved in transferring growing nascent chains to the appropriate cotranslationally acting factors.
As a first step toward understanding how NAC might transfer control of the nascent chain to the proper factor, we asked whether specific functions could be assigned to each of the two polypeptides, which were always found in NAC preparations purified from bovine brain, and whether these two proteins are always found together as a complex. Here we report by using various techniques that highly purified NAC, as well as NAC in crude extracts, is a very stable complex composed of ␣ and ␤ subunits.
Human ␣and ␤NAC were separately expressed in Escherichia coli and used to characterize the functions of the individual subunits and the reconstituted complex. Each subunit alone is in contact with the nascent chain on the ribosome. ␤NAC, but not ␣NAC, binds to ribosomes and prevents inappropriate ribosome binding to ER membranes. In contrast, prevention of cross-linking of SRP to signal-less nascent chains requires the complete NAC complex. Although ␣NAC alone is incapable of binding to ribosomes, it may contribute to NAC interaction with the ribosome or other cytosolic factors by virtue of its ability to bind tightly to nucleic acids, including those present in ribosomes and SRP.

EXPERIMENTAL PROCEDURES
Chemical Cross-linking-Purified NAC (6 g in 50 l) was diluted two times with 10 mM Hepes, pH 7.5, 100 mM KOAc, 5 mM Mg(OAc) 2 , and dithiobis(succinimidylpropionate) (DSP, stock solution 10 mg/ml in dimethyl sulfoxide) was added to 0.2 mg/ml and incubated at 4°C for 15 min. The cross-linker was quenched with 200 mM glycine. Proteins were precipitated with 10% trichloroacetic acid and subjected to SDS-PAGE in the first dimension without DTT. The whole lane was cut out and incubated in 100 mM DTT for 10 min at room temperature and then processed for the second dimension SDS-PAGE with DTT (13). Proteins were visualized by Coomassie Brilliant Blue staining.
Cell Culture and Immunohistochemistry-MC3T3-E1 cells (clonal osteogenic cell line, (14)) were grown on glass coverslips to semi-confluency using minimal essential ␣ medium supplemented with 10% fetal calf serum. The medium was exchanged against fresh medium with or without fetal calf serum, and cells were incubated for another 6 or 24 h. Immunohistochemistry was done following standard protocols. Cells were fixed in 4% paraformaldehyde for 30 min at 4°C followed by an incubation with 50 mM NH 4 Cl in phosphate-buffered saline (8 mM Na 2 HPO 4 , 2 mM NaH 2 PO 4 , 70 mM NaCl) for 10 min at room temperature to quench residual paraformaldehyde. Cells were permeabilized and blocked using 0.3% Triton X-100, 2% goat normal serum in phosphate-buffered saline for 1 h at room temperature. Affinity-purified antibodies raised against carboxyl-terminal peptides of ␣and ␤NAC (7) were used. The secondary antibody (goat anti-rabbit) conjugated to FITC (fluorescein isothiocyanate, Jackson Immuno Research Laboratories) was used as recommended by the manufacturer.
Reconstitution of NAC from Recombinant Subunits-The genes coding for human ␣and ␤NAC were cloned into pGEX-4T-1 (Amersham Pharmacia Biotech) resulting in a fusion of the nac genes to the glutathione S-transferase (gst) gene. The fusion genes were separately expressed in E. coli BL21(DE3), and expression products were purified using glutathione-agarose according to the manufacturer's manual (Amersham Pharmacia Biotech). GST-␣NAC was cleaved with thrombin, and the GST moiety was removed on a DEAE-Sepharose column (Amersham Pharmacia Biotech) run with a gradient of 0.15 to 1.0 M KCl in 20 mM Hepes, pH 7.5. ␣NAC eluted at 490 mM salt. GST-␤NAC was eluted from the glutathione-agarose with 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM CaCl 2 . Prior to thrombin cleavage Triton X-100 was added to a final concentration of 0.5% to prevent precipitation of ␤NAC. After thrombin cleavage, the GST moiety was removed on a DEAE-Sepharose column run with a gradient of 0.15 to 1.0 M KCl in 20 mM Hepes, pH 8.0, 0.5% Triton X-100. ␤NAC eluted at Ն700 mM salt. To concentrate ␤NAC and remove the detergent, the eluted protein was pooled, diluted to 150 mM salt, and loaded immediately on a second DEAE column (200-l resin volume). The column was washed with 13 ml of 20 mM Hepes, pH 8.0, 150 mM KCl, and the protein was eluted in one step using 20 mM Hepes, pH 8.0, 2 M KCl. To confirm removal of the detergent (which would interfere with the integrity of microsomal membranes), full-length preprolactin was translated in reticulocyte lysate in the presence of microsomal membranes, SRP, and the purified recombinant proteins. Successful translocation of preprolactin was confirmed by the protease protection assay (Ref. 15, data not shown).
To reconstitute complex from the recombinant subunits, thrombincleaved GST fusion proteins were mixed and denatured by addition of 8 M urea and incubation at room temperature for 1 h. Protease inhibitor (phenylmethylsulfonyl fluoride) was added to 1 mM, and refolding was induced by stepwise dialysis against buffers (20 mM Hepes, pH 7.5, 150 mM KOAc) containing successively lower concentrations of urea (4, 1, then 0 M). Any precipitated protein was removed by centrifugation (14,000 rpm, 15 min, SS-34 rotor, Sorval, ϩ4°C). To remove GST, thrombin, and other copurifying proteins, the supernatant was loaded on a Mono Q column (Amersham Pharmacia Biotech) and eluted with a gradient over 0.15 to 1.0 M KCl (in 20 mM Hepes, pH 7.5). On the following two columns excess, uncomplexed ␣NAC was removed. Mono Q fractions over 400 -470 mM KCl were combined, diluted to 150 mM salt, and applied to a Green A Matrex column (Amicon), which was run with a gradient of 0.15 to 2.0 M KCl (in 20 mM Hepes, pH 7.5). Fractions over 930 -1300 mM salt were combined and dialyzed against 20 mM Hepes, pH 7.5, 100 mM KCl and then applied to a (native) DNAcellulose column (Amersham Pharmacia Biotech; same buffers as for Mono Q) from which the complex eluted at 490 mM salt.
Because the cleavage of GST-␤NAC resulted only in limited yields of ␤NAC, the gene for ␤NAC was also cloned into pET-28a (Novagen) resulting in an amino-terminal histidine-tagged version of ␤NAC (6Hisx␤NAC). 6Hisx␤NAC was expressed in E. coli BL21(DE3) and purified under denaturing conditions (presence of 6 M urea in all buffers) following the manufacturer's manual (Novagen). 6Hisx␤NAC was refolded, and detergent was removed as described above. Following the same procedure as described for the GST fusion proteins, 6Hisx␤NAC could be reconstituted into a complex with ␣NAC, which was purified according to the same protocol as described above.
In Vitro Transcription and Translation, and Isolation of Ribosome Nascent Chain Complexes-In vitro transcription and translation of truncated mRNAs in rabbit reticulocyte lysates was done as described previously (16,17). Truncated mRNAs were translated for 20 min at 26°C. After translation, 9 volumes of dilution buffer (40 mM Hepes, 0.5 M KOAc, 5 mM Mg(OAc) 2 , 2 mM DTT, pH 7.5) was added, and RNCs (ribosome nascent chain complexes) were recovered by centrifugation (100,000 rpm, 40 min, 4°C, TLA-100.4 rotor, Beckman) through a 1.5-ml high salt-containing sucrose cushion (0.5 M sucrose in dilution buffer supplemented with protease inhibitors (18)) and 0.8 unit/l RNasin (Promega)). RNCs were resuspended in ribosome binding buffer (RBB; 20 mM Hepes, pH 7.5, 100 mM KOAc, 5 mM Mg(OAc) 2 , 1 mM DTT, supplemented with protease inhibitors and 0.8 unit/l RNasin) using 0.5 volume of the buffer unit volume of the original translation mixture. Insoluble material was then removed by centrifugation at 14,000 rpm for 3 min at 4°C. Recovery of the nascent chain complexes was typically 50 -75%. These complexes were free of NAC as assessed by Western blotting (not shown) or by a site-specific photo-cross-linking approach.
Gel Retardation Assays ("Band Shift Assays")-Incubation of DNA (0.4 g of pBluescript SKϩ (Stratagene) digested with BanI) or RNA samples (1 g of complete yeast tRNA (Boehringer), 2 g of bovine ribosomal RNA (Sigma)) in band shift buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 5 mM DTT, 5% glycerin) with increasing amounts of ␣NAC (0.4 -3.0 M) was performed in a total volume of 20 l and at a final salt concentration of 80 mM K ϩ . Samples were incubated for 20 min at 23°C and separated on agarose gels containing ethidium bromide. Gels were prepared with and run in 0.5 ϫ TAE buffer (20).
Preparation of Non-translating Ribosomes and Ribosomal Subunits-Dog pancreas rough microsomes were prepared as described (21). Rough microsomes (in 250 mM sucrose, 50 mM Hepes-KOH, pH 7.6, 50 mM KOAc, 2 mM Mg(OAc) 2 , 1 mM DTT, 10 g/ml phenylmethylsulfonyl fluoride) were adjusted to high salt by addition of 0.5 volume of 1. To separate ribosomal subunits, ribosomes were incubated with 2 mM puromycin for 10 min at 37°C followed by addition of KOAc to a final concentration of 0.5 M. Ribosomes (200 A 260 units in 2 ml) were loaded on top of 7-20% high salt sucrose gradients (50 mM Hepes, pH 7.5, 0.5 M KOAc, 0.1 mM EDTA, 7-20% sucrose, 38 ml total) and spun at 27,000 rpm for 7 h at 20°C in a SW-28 rotor (Beckman). Gradients were unloaded in 1-ml fractions, and fractions containing ribosomal subunits were pooled as judged by the A 260 profile of each gradient (22). To remove salt and concentrate the ribosomal subunits, samples were diluted with 1 volume of RBB and sedimented (100,000 rpm, 1 h, 4°C, TLA-100.4 rotor, Beckman). Ribosomal pellets were resuspended in RBB and homogenized, and insoluble material was removed by centrifugation at 14,000 rpm for 10 min.
Ribosome Binding-80 S ribosomes or 60 S ribosomal subunits were incubated with recombinant NAC or its individual subunits (in 20 mM Hepes, pH 7.5, 150 mM KOAc, 5 mM Mg(OAc) 2 , 2 mM DTT, 0.8 unit/l RNasin, and protease inhibitors, total volume of 20 l) for 2 min at 26°C and 5 min on ice. Ribosomes were then sedimented through a low salt sucrose cushion (750 l of 0.5 M sucrose in RBB, 3 h, 45,000 rpm, 4°C, SW 55 Ti rotor, Beckman). Tubes were frozen in liquid nitrogen and cut into three fractions (Top, Middle, Bottom). A pellet was resuspended in 100 l of 1% SDS. The NAC content of each fraction was analyzed by SDS-PAGE and Western blotting. In case of ␤NAC, all buffers and gradient solutions had to be supplemented with detergent (0.2% Triton X-100) and an inert protein (aprotinin, 20 g/ml) to prevent aggregation of unbound ␤NAC. Centrifugation tubes were precoated with aprotinin to prevent unspecific binding of ␤NAC to the Ultraclear centrifugation tubes (Beckman).
Nascent Chain Targeting Assay (Flotation Assay)-Truncated RNCs in 0.5 volume of RBB (see above) were incubated with NAC or its individual subunits for 2 min at 26°C and 5 min on ice prior to addition of 0.2 eq/l EDTA/KOAc stripped rough microsomes (EKRM) (21). After 2 min at 26°C and 5 min on ice, membranes and bound ribosomes were floated as described (23).

RESULTS
NAC Is a Complex of ␣ and ␤ Subunits-Purified NAC from bovine brain obtained after several chromatography steps consists of two polypeptides with apparent molecular masses of 33 and 22 kDa (7). To determine whether both polypeptides are required for NAC function we first asked whether they form a stable complex. Purified NAC migrated as a single moiety under non-denaturing conditions on PAGE (Fig. 1A, lane 1). When the band was excised and subjected to SDS-PAGE (under denaturing conditions) ␣and ␤NAC were resolved as two bands (Fig. 1a, lane 2). Scanning and integrating the bands stained with Coomassie Brilliant Blue in lane 2 suggested a 1:1 stoichiometry. Other independent methods also support the hypothesis that NAC forms a heterodimeric complex of a 1:1 stoichiometry. Using the bi-functional, thiol-cleavable crosslinking reagent DSP (dithiobis(succinimidylpropionate), ␣and ␤NAC were shown to form a complex, which was resolved by treatment with DTT (Fig. 1B). No homo-oligomers were detected with this method. Additionally, we subjected purified NAC to analysis by mass spectrometry (Fig. 1C). Three peaks were obtained that correspond to the predicted molecular masses of ␣NAC (23,492 Da), ␤NAC (17,352 Da), and the heterodimeric complex (40,864 Da), demonstrating that the differences in the calculated and apparent molecular masses deduced by SDS-PAGE are not the result of modifications but of an unusual running behavior in SDS-PAGE. The detection of a peak for the intact complex is an indication for a very stable complex.
Sizing chromatography reveals that in crude bovine brain extracts the NAC subunits are also in a complex (Fig. 1D). Note that no single subunits are detected in the appropriate fractions where the single subunits would be expected. NAC was also subjected to immunoprecipitation followed by Western blotting either without or after treating NAC samples with bifunctional cross-linkers. These data (not shown) also indicate that ␣and ␤NAC form a complex in crude cytosol.
NAC Is Localized in the Cytosol-NAC was originally purified from bovine brain cytosol and was associated with cytosolic ribosomes, suggesting that NAC is a cytosolic protein. Using a polyclonal anti-GST-␣NAC serum (rather than affinity-purified antibodies) for immunofluorescence, it was recently claimed that ␣NAC is localized in the nucleus after serum starvation of MC3T3-E1 cells, a clonal osteogenic cell line (24). Therefore, we performed immunofluorescence microscopy using affinity-purified anti-␣-and anti-␤NAC peptide-specific antibodies to determine the localization of NAC in intact cells. Different cell lines (including COS-, HeLa-, Hep-G2, and HL60 cells) were tested for NAC localization by immunostaining. In all of these cases the vast majority of NAC was found in the cytoplasm (data not shown). Staining of MC3T3-E1 cells also suggested a cytosolic localization for both subunits regardless of whether the cells had been starved or not (Fig. 2). Cells were grown to semi-confluency. In parallel samples, cells were either arrested at the G 0 /G 1 border by serum deprivation for 24 h ( 2, b and d) or allowed to continue to grow after exchange of the medium (Fig. 2, a and c). The vast majority of ␣NAC (Fig. 2, a  and b) as well as ␤NAC (Fig. 2, c and d) was clearly localized in the cytoplasm. Control staining without a first antibody showed a complete absence of signal (Fig. 2e). The same result was obtained after starving the cells for only 6 h (data not shown). The very high intracellular concentration of NAC (3-10 M (1)) naturally results in such an intense staining of the cells that we cannot exclude that a minor fraction of NAC might enter the nucleus to perform a different function. However, our results clearly demonstrate that NAC is not depleted from the cytosol in a cell cycle-dependent manner but is one of the most abundant cytosolic proteins.
Reconstitution of Functional NAC from Individual Recombinant Subunits-To define the role of NAC subunits in the described biological functions, the two were separately expressed in E. coli and purified to homogeneity (Fig. 3, lanes 3  and 4). A 70-kDa protein often copurified during the initial steps with GST-␣NAC, which was identified as DnaK (the Hsp70 homolog of E. coli) by Western blotting. Additionally, ␤NAC was expressed and purified as a his-tagged protein (Fig.  3, lane 4), because much higher yields could be achieved using this construct. Both ␤NAC variants (with or without his-tag) could be reconstituted into a functional ␣/␤ complex (see below). Recombinant reconstituted NAC (recNAC) could be purified to concentrations of 10 -15 M (Fig. 3, lane 2), which is 3to 4-fold more concentrated than the highest achievable concentration of NAC purified from bovine brain (Fig. 3, lane 1).
Both Subunits of NAC Interact with Nascent Chains on the Ribosome-NACs contact to a nascent chain on the ribosome can be shown in a cross-linking approach where a photoactivatable cross-linker is incorporated instead of lysine into the nascent chain. For the assay documented in Fig. 4 RNCs (ribosome nascent chain complexes) containing the amino-terminal 77 amino acids of peroxisomal firefly luciferase (77aaffLuc), which lacks an ER signal sequence, were prepared in a reticulocyte lysate translation system, supplemented with TDBAmodified lys-tRNA and [ 35 S]methionine. RNCs were high saltextracted and incubated with recombinant ␣NAC, his-tagged ␤NAC, recNAC, or NAC purified from bovine brain. After irradiation, cross-linking products were analyzed by SDS-PAGE followed by fluorography. The mass of the cross-linked protein can be estimated by deducting the mass of the nascent chain from that of the cross-linked product. In the absence of any added factors, no prominent cross-link appeared after irradiation (Fig. 4, lane 2). Both subunits, if added separately (Fig. 4,  lanes 3 and 4), as well as in the reconstituted complex (Fig. 4,  lane 5), are in close contact with the nascent chain, because they give rise to strong cross-links to the nascent chain. The cross-links are comparable in strength to those obtained with NAC purified from bovine brain (Fig. 4, lane 6), indicating that with regard to this function recNAC is as active as bovine NAC.
No cross-links to the NAC subunits were detected if the nascent chains were released from the ribosome by puromycin treatment prior to irradiation (data not shown). Although ␣NAC is clearly in close proximity to the nascent chain (a prerequisite for obtaining a cross-linking product with this method), its interaction with the RNC seems to be weaker than FIG. 2. NAC proteins are localized in the cytoplasm. MC3T3-E1 cells at semi-confluency were either incubated with medium without the addition of fetal calf serum ("starvation") for an additional 24 h or, in a parallel sample, the medium was exchanged against fresh medium containing 10% fetal calf serum. Cells were fixed and analyzed for NAC localization using peptide-specific affinity-purified antibodies. In the presence of fetal calf serum the vast majority of ␣NAC (a) and ␤NAC (c) are localized in the cytoplasm. Upon starvation no obvious relocation of either of the subunits to the nucleus was detectable (b and d). The same distribution of NAC subunits was found after 6 h of starvation (data not shown). In the absence of a first antibody no unspecific signal was obtained with the secondary FITC-labeled anti rabbit antibody (e).

FIG. 3.
Reconstitution of NAC complex from recombinant proteins. SDS-PAGE of purified NAC proteins. Lane 1, NAC purified from bovine brain cytosol. Lane 2, NAC reconstituted from purified recombinant subunits, which were initially expressed as fusions to glutathione S-transferase and then cleaved with thrombin prior to reconstitution and further purification. The subunits show a slightly slower mobility in the gel due to additional six amino-terminal amino acids remaining after thrombin cleavage of the GST fusion proteins (GSIEGR, lanes 2 and 3). Lane 3, purified ␣NAC. Lane 4, purified histidine-tagged ␤NAC, which has a 20-amino acid amino-terminal extension (MGSSHHHHHHSSGLVPRGSH).

FIG. 4. Recombinant NAC interacts with nascent chains on ribosomes.
High salt-stripped 77aaffLuc RNCs prepared in rabbit reticulocyte lysate with TDBA-lys-tRNA and [ 35 S]methionine were incubated with NAC or its individual subunits as indicated. Samples were irradiated and analyzed by SDS-PAGE and fluorography. Both subunits as well as the recombinant NAC could be cross-linked to nascent chains on the ribosome. Presence of the ribosome was essential, because puromycin treatment prior to irradiation abolished the cross-links (data not shown). that of ␤NAC. If the RNCs are sedimented through a low salt-containing (150 mM KOAc) sucrose cushion prior to irradiation, the cross-link to ␣NAC is hardly detectable, whereas the cross-link to ␤NAC remains unchanged (data not shown). This result implies that ␤NAC can bind to RNCs (see below) and ␣NAC, in the absence of the ␤ subunit, is in rather loose contact with the nascent chain (see below).
␣NAC Has a High Affinity for Nucleic Acids-NAC acts at the ribosome where it is in close vicinity not only to various cytosolic and ribosomal proteins but also to several different RNA molecules, e.g. ribosomal RNA, mRNA, tRNA, and 7SL RNA of the ribonucleoprotein complex SRP. Therefore, we were interested in the question whether ␣NAC, which does not bind to non-translating ribosomes (see below), can bind to nucleic acids. Interestingly, it had been recently reported that ␣NAC binds to DNA in a sequence-specific manner. A consensus sequence for binding of ␣NAC to DNA was published (5Ј-(G/C)(G/ C)A(G/C)A(G/C)ANNNG-3Ј) (25). We confirmed binding of ␣NAC to oligonucleotides containing this motif; however, we also observed binding to oligonucleotides where this motif was mutated (data not shown). To further characterize the binding of ␣NAC to nucleic acids, we tested binding to DNA, rRNA, tRNA, and 7SL RNA. DNA binding was tested using the plasmid pBluescript SKϩ, which was cleaved with BanI, resulting in four easily distinguishable fragments (244, 386, 1097, and 1231 bp). Only one contains the so-called "NAC motif" (1097 bp). The DNA fragments were incubated with increasing amounts of ␣NAC and separated on agarose gels containing ethidium bromide. All four DNA fragments exhibit a slower mobility in the gel, indicating that ␣NAC bound to them regardless of whether they contained the NAC motif or not (Fig.  5a). Furthermore, the more ␣NAC is added, the stronger the retardation effect becomes, indicating that more than one ␣NAC molecule binds to each of the DNA fragments.
␣NAC binding is not restricted to DNA molecules. Incubation of ribosomal RNA from bovine liver (Fig. 5b) or complete tRNA from yeast ( Fig. 5c) with rising amounts of ␣NAC also resulted in a complete shift of the RNA band(s). ␣NAC also bound to the 7SL RNA in an almost 1:1 stoichiometry (data not shown). In summary, ␣NAC binds to all nucleic acids we have tested. Clearly, this affinity is not restricted to DNA molecules and is not dependent on a specific sequence motif. The ability of ␣NAC to bind to RNA and DNA molecules might contribute to the overall binding of the NAC complex to ribosomes or may even underlie its ability to compete with SRP for access to nascent chains (see below). If this capacity contributes to the overall binding of the complex to translating ribosomes, RNA molecules should compete for the interaction of ␣NAC with the nascent chain. A 10-fold molar excess of mRNA (coding either for the first 77 amino acids of firefly luciferase or the first 86 amino acids of preprolactin), tRNA, or rRNA led to a 50% reduction of the cross-link of ␣NAC to 77aaffLuc RNCs, whereas cross-links to the NAC complex remained unchanged in the presence of RNAs (data not shown).
␤NAC, but Not ␣NAC, Can Bind to the Ribosome in the Absence of the Corresponding Subunit-NAC can be sedimented with RNCs and prevents RNCs as well as non-translating ribosomes from binding to translocation sites at the ER membrane by blocking a putative ribosomal membrane attachment site (26). We investigated whether the affinity for ribosomes could be ascribed to one subunit or whether this binding requires NAC complex. recNAC or the individual subunits were incubated with non-translating cytosolic ribosomes or with the large 60 S ribosomal subunit. Samples were sedimented through a low salt sucrose cushion. Each tube was divided into four fractions (top (T), middle (M), bottom (B), and pellet (P)). Fig. 6, shows the distribution of NAC subunits in these fractions after SDS-PAGE and Western blotting. Bottom and pellet fractions contained sedimented ribosomes as judged by Ponceau and Coomassie staining (data not shown). In the absence of ribosomes, none of the NAC proteins sedimented (Fig. 6,  -ribos). The reconstituted, recombinant complex bound to nontranslating ribosomes and, although less efficiently, to 60 S ribosomal subunits (Fig. 6, recNAC, B and P). ␣NAC alone is not capable of binding to non-translating ribosomes or 60 S ribosomal subunits. It was always found in the top fraction. ␤NAC showed the same binding pattern as recNAC. It bound to 80 S ribosomes and less efficiently to 60 S subunits.
␤NAC Alone Prevents Default Targeting of Ribosomes Containing Signal-less Nascent Chains to the ER-In addition to its role in shielding the nascent chain on the ribosome toward the cytosol, NAC prevents inappropriate targeting of nontranslating ribosomes and RNCs bearing signal-less nascent chains to the ER membrane by occupying the proposed membrane-attachment site (M-site) on the ribosome (23,26). We next investigated whether the binding of ␤NAC and recNAC to ribosomes observed in Fig. 6 reflects functional binding to the M-site on the ribosome. This function is tested by using a flotation assay where high salt-stripped RNCs are incubated with microsomal membranes (EKRM), which contain the com-  1 and 2). Addition of ␣NAC (lanes 3-5) led to a band shift indicating the binding of this subunit to tRNA. plete repertoire of ER membrane proteins but were stripped of their bound ribosomes by an EDTA/high salt treatment (21). After incubation of RNCs with these membranes, in the presence or absence of additional factors, samples were separated on discontinuous sucrose gradients. Two fractions were collected; the top fraction (T in Fig. 7) contains membranes and all RNCs bound to them, and the bottom fraction (B in Fig. 7) contains unbound, sedimented RNCs. Fig. 7 shows the result of an experiment where signal-less 77aaffLuc RNCs were incubated with EKRMs in the absence or presence of NAC or its individual subunits. In the absence of any added factors approximately 50% of signal-less RNCs bound to the membranes (Fig. 7, lanes 1 and 2), a result which is in agreement with previously published data (10). In the presence of bovine NAC RNCs were recovered in the bottom fraction (Fig. 7, lanes 3 and  4). The same result was achieved by adding recNAC to the incubation mix (Fig. 7, lanes 9 and 10), indicating that NAC and recNAC both prevent binding of RNCs that do not carry a signal sequence to the ER membrane. Lanes 7 and 8 show that ␤NAC is sufficient for preventing signal-less RNCs from binding to membranes. In contrast, ␣NAC had no influence on the binding of RNCs to EKRMs, because in this case again approximately 50% of the RNCs were targeted to the membranes (Fig.  7, lanes 5 and 6).
NAC Complex, but Not the Individual Subunits, Prevent Inappropriate Interaction of SRP with Signal-less Chains on Ribosomes-In vivo SRP usually only interacts with the signal sequence of nascent chains on the ribosome. In contrast, in the absence of NAC, SRP can also be cross-linked to RNCs bearing signal-less nascent chains. Addition of NAC or complete cytosol to RNCs prevents this inappropriate interaction (1,10). Because both single NAC subunits could be cross-linked to signalless RNCs, we asked whether one of the subunits would be sufficient to compete with SRP for cross-linking to a signal-less chain. High salt-stripped 77aaffLuc RNCs were incubated with excess SRP. Prior to a second incubation round, the individual NAC subunits, bovine NAC or recombinant NAC, were added to the samples, followed by irradiation and analysis by SDS-PAGE and fluorography. Bovine NAC (Fig. 8, lane 6) and recNAC (Fig. 8, lane 5), both successfully compete with SRP for interaction with a signal-less chain on the ribosome. But neither ␣NAC (Fig. 8, lane 3) nor ␤NAC (Fig. 8, lane 4) alone could prevent SRP from interacting with this chain. Changing the order in which the factors were added to the mix had no influence on the result (data not shown). The results shown in Fig. 8 demonstrate that the functional complex is required to prevent inappropriate interactions with the nascent chain and that neither subunit alone can substitute for this function.

DISCUSSION
Between synthesis on the ribosome and arrival at a final destination, cellular proteins undergo a journey whose regulation and control is vital to the proper functioning of the eukaryotic cell (27). Thus, depending on their final localization, proteins may interact with the cytosolic folding machinery, i.e. chaperones (28,29), or with factors like SRP and SRP receptor, which direct cotranslational translocation into the ER (30 -33). Although the mechanisms by which proteins are translocated into the ER are very well understood, our knowledge about other cotranslationally acting factors such as cytosolic chaperones and nascent polypeptide-associated complex (NAC) is still rudimentary.
NAC is the first cytosolic factor to contact the nascent polypeptide chain as it grows on the ribosome and thus may be key in regulating early events in protein folding and transport (8). In fact, it has been established that NAC prevents SRP binding to inappropriate nascent chains, i.e. those lacking a signal peptide (7). NAC also prevents binding of inappropriate ribosomes to translocons on the ER by blocking the putative ribosomal M-site (1,10,23,26). We have previously hypothesized that, as the nascent chain lengthens, NAC binding weakens, and for signal sequence containing proteins, allows SRP to compete effectively for binding to signal peptides. Once SRP displaces NAC, the ribosome is targeted to the ER, and after the SRP receptor displaces the SRP, the M-site is free to bind to the translocon and the nascent chain enters the lumen of the ER (23).
The physiological importance of NAC cellular function is underscored by the embryonically lethal phenotype of NAC deletion mutants in both mouse (2) and fruit fly (3). An insertional mutation in the gene coding for ␤NAC led to early postimplantational lethality in mice (2). In Drosophila a mutation affecting ␤NAC expression was shown to cause the long known bicaudal phenotype (3). Bicaudal was the first Drosoph- , and bottom (B) fractions, and pellets (P) were resuspended using 1% SDS. Ribosomes are recovered under these conditions in the bottom and pellet fraction. NAC was detected in the different fractions by Western blotting with peptide-specific antibodies against ␣and ␤NAC. In the absence of ribosomes or 60 S ribosomal subunits, NAC and its subunits did not sediment into the sucrose gradient (lanes 9 -12, -ribos), they were always recovered in the top fraction. ␣NAC did not bind to 80 S ribosomes or 60 S subunits. It was always recovered in the top fraction (lanes 1 and 5). In contrast, recombinant NAC as well as ␤NAC alone bound to complete 80 S ribosomes and 60 S ribosomal subunits (lanes 3, 4, and 8). Note that the binding to 60 S ribosomal subunits appears to be weaker than to 80 S ribosomes. ila mutation identified as producing mirror-image pattern duplications along the anterior-posterior axis of the embryo (35). In contrast, deletion mutants in yeast, lacking both ␣NAC and ␤NAC, show only slight growth defects (4,5), but SRP is also not essential in yeast (35,37).
We show here that NAC is detected solely as a heterodimer of ␣ and ␤ subunits associated in a very stable complex in vivo (Fig. 1). Consistent with this, peptide-specific antibodies raised against the individual subunits demonstrate that both have a cytoplasmic localization in a number of cell lines, including HepG2, HeLa, and COS and during different cell cycle stages of HL60 cells. 2 This is also true in an osteogenic cell line, MC3T3-E1 after serum starvation (Fig. 2), conditions previously reported to result in movement of ␣NAC to the nucleus (24). Because of NAC's extremely high intracellular concentration of 3 to 10 M in tissues tested (1), it is impossible to exclude that a minor fraction of NAC might be in the nucleus but, clearly, NAC is not depleted from the cytosol in favor of a nuclear localization under these conditions.
We present here a further characterization of NAC based on recombinant subunits purified from E. coli (Fig. 3). Full activity in all of our assays was obtained using holoNAC reconstituted from these recombinant subunits (Figs. 4, 7, and 8). This confirms that NAC is composed solely of ␣ and ␤ subunits as well as showing that our recombinant subunits are intact and functional.
Cross-linking experiments demonstrate that both ␣ and ␤ subunits are in contact with nascent chains on the ribosome (7). However, ␤NAC binds more tightly than ␣NAC, because its binding is resistant to low salt extraction and it retains binding to non-translating ribosomes (Fig. 6). In fact, ␤NAC binding to ribosomes is sufficient to prevent the mistargeting of signalless RNCs to the ER (Fig. 7). This suggests that ␤NAC binds at or near the proposed M-site on the ribosome, preventing binding to the translocon.
The only activity that we can so far attribute to ␣NAC on its own is the ability to bind nucleic acids (Fig. 5). Yotov and St-Arnaud have also published that ␣NAC binds to a highly redundant consensus DNA sequence and thereby is implicated in transcription activation (25). Consistent with this idea, it has been reported that a tissue-specific isoform of ␣NAC (skNAC) is up-regulated in differentiating myoblasts and osteoblasts and in cells involved in wound healing (24,25,39,40). Note, however, that all biochemical activities reported up to now have been for ␣NAC and not skNAC. In contrast to Yotov and St-Arnaud, we find that ␣NAC binding is not restricted to DNA, nor is it sequence specific. Rather, ␣NAC binds to a variety of DNA and RNA molecules, including ribosomal RNA, tRNA (Fig. 5), and the 7SL RNA of SRP. 3 These results suggest that ␣NAC affinity for nucleic acid may contribute to the overall binding and possibly positioning of NAC in complex with the ribosome. This hypothesis is supported by the observation that RNase A treatment of ribosomes results in less holoNAC binding to ribosomes. 3 Interestingly, evidence has recently been provided that 28 S rRNA mediates the interaction of the ribosome with the SEC61 complex (38), which together with NAC and SRP, has been shown to compete for binding to the M-site (23). Also consistent with this idea is our unpublished finding that both tRNA and mRNA compete for the cross-link of ␣NAC to the nascent chain but have no effect on the cross-link of holoNAC to nascent chains. However, ribosomal RNA may only become accessible to ␣NAC during translation or in the presence of ␤NAC, because ␣NAC alone is not capable of binding to 80 S ribosomes. A computer model constructed with Predict-Protein software (41) assigns the sequence of ␣NAC from Arg-71 to Lys-82 to an ␣-helix. Because this putative ␣-helix would carry a substantial positive charge, it may be responsible for a strong but nonspecific interaction with the negatively charged phosphates of nucleic acids.
At least one key function of NAC requires both ␣ and ␤ subunits in complex. HoloNAC is required to prevent inappropriate interactions of SRP with non-signal sequence nascent chain complexes (Fig. 8). Thus, although ␤NAC is able to bind the ribosome on its own, ␣ and ␤ subunits together bind in a functionally different manner.
These data together suggest a model in which NAC is bound to the ribosome via its ␤ subunit binding to the M-site. ␤NAC, by complexing tightly to ␣NAC also serves to strengthen ␣NACs interaction with the ribosome. Together, the subunits shield the nascent chain against early inappropriate interactions with other factors. We suggest that cycles of binding and releasing of NAC gradually allow other cytosolic factors to access the growing polypeptide, in "quantal" units, rather than one amino acid at a time. A question to be addressed in the future is whether NAC interacts with any of the protein factors that replace it on the nascent chain. For example, ␣NAC contains a domain with homology to DnaJ and GST-␣NAC copurifies with DnaK, the Hsp70 homolog in E. coli, 4 suggesting that ␣NAC interacts with Hsp70 in eukaryotic cells. Our purified recombinant ␣and ␤NAC subunits constitute a powerful tool to answer such questions and to further characterize the M-site.