Cotranslational Ubiquitination of Cystic Fibrosis Transmembrane Conductance Regulator in Vitro *

Ubiquitination is a covalent protein modification that can target proteins in eukaryotic cells for degradation by the 26 S proteasome. Substrates for this degradation pathway include abnormal proteins that arise from misfolding and/or mutation. How and when the ubiquitination machinery recognizes misfolded proteins and targets them for degradation remains largely unknown. We have previously shown that cystic fibrosis transmembrane conductance regulator (CFTR), is rapidly degraded in a ubiquitin-dependent fashion, without any detectable lag following its synthesis (Ward, C. L., and Kopito, R. R. (1994) J. Biol. Chem. 269, 25710–25718), suggesting that ubiquitination and protein synthesis may be temporally linked. In the present study, we have investigated the timing of CFTR ubiquitination relative to its translation in reticulocyte lysates containing 125 I-ubiquitin. In synchronized, proteasome-inhibited lysates, translation of full-length CFTR chains was com-pleted in approximately 30 min, whereas modification of CFTR with [ 125 I]ubiquitin was evident by 20 min, indicating that ubiquitination precedes the completion of full-length polypeptide chains. Moreover, ubiquitin was also found to be transferred to nascent CFTR chains while attached to ribosomes. Together, these data establish that ubiquitination, which is widely assumed to be a post-translational event, can occur cotranslationally and suggest a role for ubiquitination early in protein biosynthesis.

Ubiquitin is a highly conserved 76-amino acid polypeptide that plays a central role in determining protein turnover (reviewed in Refs. [1][2][3]. Covalent attachment of ubiquitin serves as signal that targets proteins for degradation by the 26 S proteasome. Substrates for this degradation pathway include potentially toxic, aggregation-prone misfolded proteins that can be synthesized de novo as a result of mutation or of errors in translation or transcription. Although the enzymatic mechanism by which ubiquitin is conjugated to degradation sub-strates has been well characterized (reviewed in Ref. 4), the processes by which misfolded protein substrates are initially recognized by the Ub 1 conjugation machinery remain largely unknown.
Genetic and biochemical studies suggest that some substrate specificity resides in the enzymatic machinery of the ubiquitin coupling system, particularly the ubiquitin-conjugating enzymes (E2s) and ubiquitin-protein ligases (E3s), which function in transferring thiol-activated ubiquitin intermediates to lysine ⑀-amino groups on the target polypeptide (2,3). However, attachment of a single ubiquitin is not sufficient to mark a protein for degradation (5). Efficient degradation by the 26 S proteasome appears to require the presence on the target protein of "multiubiquitin" polymers often composed of more than 20 ubiquitins (5,6). Proteasomal degradation of a protein depends therefore not only on the presence of covalently attached ubiquitin but also on the length of the multiubiquitin chain and possibly the type of internal ubiquitin-ubiquitin linkage. An additional level of specificity is suggested by the presence in cells of multiple ubiquitin isopeptidases that appear to "edit" multiubiquitin chains and suggest that the ultimate fate of a protein may be determined by kinetic partitioning between ubiquitination and de-ubiquitination (7,8).
Recent data implicate the ubiquitin-proteasome pathway in the degradation of misfolded integral membrane proteins (9) including CFTR (10,11). It is well established that CFTR folding is an inefficient process in vivo (12,13). The majority (50 -75%) of newly synthesized wild type CFTR molecules are rapidly degraded by a pathway that involves both ubiqutination and proteaseome activity (10,11). The common ⌬F508 mutation increases the ineffienciency of CFTR folding (to greater than 99%) but does not change the kinetics or ubiquitin dependence of its degradation (12,13). CFTR and ⌬F508 degradation in vivo occur without a detectable lag following biosynthesis (12). Moreover, the apparent rate of CFTR synthesis in vivo (assessed by the incorporation of [ 35 S]Met) is increased 2-3-fold in the presence of proteasome inhibitors. 2 Together these data suggest that recognition of CFTR misfolding by the ubiquitin-conjugating machinery may be closely linked to translation and may not be easily reconciled with the generally held assumption that ubiquitination is a "post-translational modification." To clarify this point, we address in this paper when nascent CFTR chains are tagged with ubiquitin by employing a cell-free system. Our data indicate that nascent CFTR polypeptides can become ubiquitinated prior to release from the translation apparatus. These data establish that ubiquitin-conjugating machinery can recognize nascent polypeptides during their translation and suggest that the destiny of nascent polypeptides may be established by ubiquitination, whereas the polypeptide chains are still elongating on ribosomes.

MATERIALS AND METHODS
In Vitro Transcription, Translation, and Immunoprecipitation-HA-CFTR and HA-⌬F508 cDNA were generated by introducing the HA epitope (YPYDVPDYA) between codons 109 and 110 of human CFTR * This work was supported by Grant DK43994 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Separation of Ribosome-bound Polypeptide Chains-Run-off translation of HA-CFTR RNA (truncated at codon 920) was terminated with cycloheximide (2 mM) to stabilize the ribosome nascent chain complex or puromycin (2 mM) to release the polypeptide from ribosomes together with an ATP depletion mixture containing hexokinase (0.4 mg/ml) and D-glucose (5 mM). Samples to which puromycin was added were further incubated at 30°C for 5 (for translation without RM) or 30 min (for translation with RM) as indicated. The RM fraction was sedimented as above and solubilized in 3% digitonin in buffer E (50 mM HEPES-KOH (pH 7.5), 500 mM potassium acetate, 5 mM magnesium acetate, 0.5 mM EDTA) for 30 min on ice. The solubilized fraction was precleared by 10 min of centrifugation at 14,000 ϫ g at 4°C and then applied onto a 1.2-ml sucrose gradient (15-50%) in 3% digitonin/buffer E and centrifuged for 90 min in TLA 100.2 at 100,000 ϫ g. The gradient was fractionated (0.1 ml/fraction), and one-fifth of each fraction was used to monitor truncated HA-CFTR polypeptide by immunoblotting. The remainder was diluted in buffer A and immunoprecipitated with anti-HA antibody. For purification of ribosome-polypeptide in the absence of RM, translation mixture was precleared and subjected to sedimentation on a sucrose density gradient in buffer E without detergent.

RESULTS AND DISCUSSION
To monitor the fate of nascent CFTR polypeptide chains, an HA epitope tag was inserted near the N terminus of CFTR and ⌬F508 (Fig. 1a). This construct, designated HA-CFTR (or HA-⌬F508) was efficiently translated in vitro in a RRL supplemented with canine pancreas microsomes (RM), giving rise to a single predominant polypeptide of molecular mass of 140 kDa (Fig. 1b) that was identified as full-length HA-CFTR because it was immunoprecipitated by antibodies both to the extreme C terminus of CFTR and to the N-terminal HA epitope. Most HA-CFTR translated in the presence of RM bound to concanavalin A-Sepharose, suggesting that it is glycosylated (data not shown). Moreover, the mobility of HA-CFTR translated in the absence of RM was ϳ2-4 kDa lower than its mobility in the presence of RM, consistent with the presence of 1-2 core Nlinked oligosaccharide chains (Fig. 1c). Finally, HA-CFTR was not extracted from RM membranes at pH 11.5 under conditions sufficient to extract the lumenal protein ␤-lactamase (Fig. 1d). Together, these data demonstrate that full-length HA-CFTR is efficiently translocated, glycosylated, and integrated into RM membranes in vitro and establish that this cell-free system is a suitable model for the early stages of CFTR biosynthesis.
To study ubiquitination of nascent CFTR polypeptides in vitro, HA-CFTR or HA-⌬F508 were translated in the presence of [ 125 I]Ub, and HA-CFTR polypeptide chains were purified from the translation mixture by immunoprecipitation with anti-HA antibody. The presence of hemin (20 M), a proteasome inhibitor, in this translation-optimized RRL ensured that proteins could be ubiquitinated by the extract but remain undegraded by the proteasome (15). Ubiquitinated protein was detected as an intense smear on autoradiograms of SDS-PAGE separated proteins from RRL programmed with either HA-CFTR or HA-⌬F508 mRNA but not when translation was conducted in the absence of mRNA (Fig. 2a), suggesting that Ub can be transferred onto newly synthesized HA-CFTR polypeptides. The high molecular weight of 125 I label relative to unmodified CFTR could be due to multiubiquitination at a single lysine or to monoubiquitination at multiple lysines, although the data do not discount the possibility that conjugation of single Ub could promote CFTR aggregation. There was no significant difference in the degree of ubiquitination between HA-CFTR and HA-⌬F508 (Fig. 2a), in agreement with previous data indicating that both ⌬F508 and the ϳ75% of CFTR that Sites of N-linked glycosylation are indicated, as are the two cytoplasmic ATP binding sites (ATP), the location of the inserted epitope tag (HA) and the locations of the N (N) and C (Ct) termini. b, detection of CFTR synthesized in vitro. mRNAs encoding CFTR or ⌬F508 with or without an HA tag as indicated were translated in RRL supplemented with canine microsomes in the presence of [ 35 S]Met at 30°C for 60 min. CFTR was immunoprecipitated (IP) with an antibody that recognizes the C terminus of CFTR (lanes 4 -6) or with anti-HA monoclonal antibody (lanes 7-9) or were used without immunoprecipitation (lanes 1-3). Samples were separated by SDS-PAGE and detected by fluorography. WT, wild type. c, HA-CFTR translated in vitro in the presence of microsome is glycosylated. RRL was programmed either with HA-CFTR RNA or no RNA. Translation was conducted in the presence or the absence of RM as indicated. d, in vitro translated HA-CFTR is resistant to alkali extraction from membranes. After translation programmed with RNAs encoding HA-CFTR or ␤-lactamase, RM were treated with sodium carbonate buffer (pH 11.5) and separated into microsome (P) and supernatant (S) fractions by centrifugation through 0.5 M sucrose at 100,000 ϫ g for 15 min. fails to fold are ubiquitinated in vivo (10). By contrast, significantly more label was incorporated into HA immunoprecipitates when the translation was conducted in the absence of RM, consistent with the expectation that CFTR, a hydrophobic polytopic integral membrane protein, is likely to be severely misfolded when translated in the absence of membranes. To test if the 125 I-labeled smear associated with HA-CFTR is due to the small fraction of CFTR molecules that fail to be translocated and to become membrane integrated, we isolated RM. As expected, the majority of core-glycosylated CFTR (Fig. 2b, lower panel) together with the majority of the 125 I label (Fig. 2b, upper panel) cosedimented with RM. When we tested for glycosylation in detergent extracts of RM from HA-CFTR-programmed lysates, we found that most of the 125 I label was adsorbed with concanavalin A-Sepharose (data not shown). Together, these data strongly suggest that RRL can transfer Ub onto both glycosylated membrane-associated and untranslocated CFTR polypeptides.
To precisely determine the timing of ubiquitination relative to translation, we monitored the kinetics of [ 125 I]Ub incorporation into a cohort of HA-CFTR molecules synthesized in a synchronized RRL extract. 10 min after translation was initiated by the addition of HA-CFTR mRNA, the translation reaction was synchronized with aurintricarboxylic acid to prevent re-initiation (16). Full-length HA-CFTR monitored by immunoblotting (Fig. 3a, lower panel) or by [ 35 S]Met incorporation in parallel translations (data not shown) was first detectable after 30 min, corresponding to a rate of CFTR synthesis of 50 -70 amino acid residues/min, comparable with previously reported rates (17). By contrast, high molecular weight, [ 125 I]ubiquitinated HA-immunoreactive polypeptide was evident in the autoradiograms after 20 min (Fig. 3a, upper panel), suggesting that ubiquitination of the translating polypeptide precedes the completion of full-length HA-CFTR (similar results were obtained with a 3-5-min initiation window; data not shown). To further examine this conclusion, we investigated whether [ 125 I]Ub label was associated with incomplete HA-CFTR chains that were immunoprecipitated with anti-HA antibody following immunodepletion of full-length CFTR chains with antibody to the extreme C terminus (Fig. 3b). These data indicate that incomplete HA-CFTR chains are ubiquitinated in the translation system.
Incomplete ubiquitinated chains could arise from cotranslationally ubiquitinated nascent HA-CFTR molecules or from polypeptide chains that were released prematurely from the translational apparatus. To discriminate between these two possibilities, we assessed the incorporation of [ 125 I]Ub into nascent, ribosome-bound HA-CFTR polypeptides translated from a truncated mRNA construct lacking a stop codon and then fractionated on sucrose density gradients (Fig. 4). When the translation was conducted in the absence of RM (Fig. 4a), virtually all of the truncated HA-CFTR chains of the expected size (110 kDa), assessed by immunoblotting, cosedimented in the heavier fractions near the bottom of the gradient together with 125 I-labeled protein. The presence of 125 I label in the lighter fractions (2-5) that lacked significant 110-kDa HA immunoreactivity suggests that some HA-CFTR chains that had been prematurely released from ribosomes were ubiquitinated. Treatment of the lysates with puromycin B (Fig. 4b), which releases nascent chains from ribosomes, shifted most of the 125 I label and 110-kDa HA-CFTR to lighter fractions, indicating that the majority of the ubiquitinated HA-CFTR polypeptides Cotranslational Ubiquitination of CFTR 7191 in fractions 7-10 were ribosome associated. Although the release of nascent chains from membrane-bound ribosomes was inefficient, when truncated HA-CFTR was translated in the presence of RM, a fraction of ubiquitinated truncated HA-CFTR chains were shifted to lighter fractions by puromycin treatment (data not shown), suggesting that nascent membrane-integrated chains that are attached to ribosomes are also modified by Ub. Taken together, these data show that ribosome-associated nascent CFTR polypeptides chains can be modified cotranslationally with ubiquitin. Protein folding is the process by which the linear information in the genetic code is converted into the three-dimensional structures of proteins. Because incorrectly folded proteins expose hydrophobic surfaces and therefore have a strong tendency to form insoluble potentially toxic aggregates, eukaryotic cells possess "quality control" mechanisms that monitor the progress of protein folding and selectively degrade misfolded proteins (reviewed in Ref. 18). Intermediates in the normal pathway of protein folding also transiently expose aggregationprone hydrophobic surfaces; the formation of aggregated, offpathway products in cells is minimized by sequential, transient interaction of nascent, partially folded proteins with molecular chaperones (19).
How the cellular quality control machinery discriminates between bona fide intermediates in the folding pathway and aggregation-prone products that lie off of the folding pathway is not known. Misfolded ⌬F508 CFTR molecules appear to remain associated with the molecular chaperones calnexin (20) and Hsp70 (21) until they are degraded, suggesting that molecular chaperones may contribute to quality control. However, because these chaperones also lie on the folding pathway, it is unlikely that chaperone binding is the sole signal for the degradation machinery. The existence of an editing isopeptidase recently identified as a component of the 26 S proteasome (7) raises the possibility that multiubiquitination is reversible and that Ub-dependent proteasomal targeting is not sufficient for efficient degradation.
The data in this paper are the first evidence demonstrating that protein ubiquitination can occur while the nascent polypeptide chain is still attached to the ribosome. Cotranslational ubiquitination could occur because of early recognition of a misfolding/ubiquitination signal by the cellular quality control apparatus. Alternatively, the reversible conjugation of Ub to nascent chains might itself serve a chaperone-like role where the choice between folding and degradation would be determined by the competition between net elongation or shortening of a Ub chain. Future studies will be needed to discriminate between these two possibilities.