Report

Operon structure and cotranslational subunit association direct protein assembly in bacteria

See allHide authors and affiliations

Science  06 Nov 2015:
Vol. 350, Issue 6261, pp. 678-680
DOI: 10.1126/science.aac8171

Proximity best for building protein complexes

The synthesis of protein subunits and their assembly into a fully functional complex are generally thought to be two distinct processes. Shieh et al. studied the synthesis and assembly of the luciferase complex in Escherichia coli. Organization of the luciferase subunits LuxA and LuxB side by side into an operon promotes their colocalized synthesis and assembly into an active enzyme complex. Indeed, the association between the subunits occurs as they are being synthesized on ribosomes, which helps order the sequence of subunit interactions.

Science, this issue p. 678

Abstract

Assembly of protein complexes is considered a posttranslational process involving random collision of subunits. We show that within the Escherichia coli cytosol, bacterial luciferase subunits LuxA and LuxB assemble into complexes close to the site of subunit synthesis. Assembly efficiency decreases markedly if subunits are synthesized on separate messenger RNAs from genes integrated at distant chromosomal sites. Subunit assembly initiates cotranslationally on nascent LuxB in vivo. The ribosome-associated chaperone trigger factor delays the onset of cotranslational interactions until the LuxB dimer interface is fully exposed. Protein assembly is thus directly coupled to the translation process and involves spatially confined, actively chaperoned cotranslational subunit interactions. Bacterial gene organization into operons therefore reflects a fundamental cotranslational mechanism for spatial and temporal regulation that is vital to effective assembly of protein complexes.

Oligomeric protein complexes are thought to assemble by diffusion and random collision of subunits within the cytosol (1) [“trans-assembly” (fig. S1)]. However, this mechanism does not explain how unassembled subunits avoid nonspecific interactions, aggregation, and quality control sequestration to proteases and chaperones or how subunits navigate crowded and occluded cellular environments. We postulated that for complex subunits translated from polycistronic mRNAs, local confinement of assembly around the translation sites [“cis-assembly” (fig. S1)] would promote efficient, nonstochastic assembly.

To test our hypothesis, we used the bacterial Vibrio harveyi heterodimeric luciferase complex, which contains the subunits LuxA (α) and LuxB (β) (2), encoded by the luxCDABE operon (3). To express luxA and luxB from the same or distinct operons, we integrated plasmids harboring artificial lux operons (fig. S2B) into the Escherichia coli chromosome at distinct sites (fig. S2A). We fused genes encoding monomeric variant forms of enhanced yellow fluorescent protein (YFP) or cyan fluorescent protein (CFP) to the 5′ end of each lux gene (fig. S2B). We created four different strains (1 through 4 in Fig. 1A), each with two artificial operons carrying different tag configurations of the yfp/cfp-luxA and yfp/cfp-luxB fusion genes integrated into the chromosome (see supplementary materials and methods).

Fig. 1 Assembly of bacterial luciferase.

(A) E. coli strains 1 to 4 with two bicistronic operons integrated into the chromosome. Operons encode fluorescently tagged luxA or luxB genes under control of isopropyl-β-d-thiogalactopyranoside–inducible promoters and strong Shine-Dalgarno sequences. (B) Relative specific luciferase activities (mean ± SEM, n = 3 replicates, unpaired t test, P = 0.9207) (fig. S3A) and relative FRET efficiencies (mean ± SEM, n = 3, unpaired t test, P < 0.0001) of strains 1 and 2 determined in vivo. (C) Relative specific luciferase activities of strains 3 and 4 determined in vivo (mean ± SEM, n = 19, unpaired t test, P < 0.0001) (fig. S3B, right). We used GraphPad Prism 6 software for the significance tests.

To assess luciferase assembly in the vicinity of synthesis, we assayed specific luciferase activities and fluorescence resonance energy transfer (FRET) efficiencies in strains 1 and 2 (Fig. 1A). Both strains encode luxA followed by luxB in each operon. The FRET pair (YFP and CFP) in strain 1 tags LuxA and LuxB within each operon, whereas in strain 2, YFP tags both LuxA and LuxB subunits in one operon and CFP tags both subunits in the other. The strains exhibit similar specific luciferase activities (Fig. 1B, left, and fig. S3A), consistent with the identical genetic order of luxA and luxB within the operons. However, strain 2, which encodes each partner of the FRET pair in separate operons, displays only 40 ± 3% of the FRET efficiency of strain 1, which encodes the FRET pair within each operon (Fig. 1B, right). This observation demonstrates that bacterial luciferase heterodimers are predominantly assembled in a cis configuration from subunits synthesized from the same bicistronic mRNA molecule.

To establish whether operon organization promoting cis-assembly confers any advantage to the assembly process, we compared specific luciferase activities in cells where luxA and luxB are expressed from the same (strain 3) or separate (strain 4) operons (Fig. 1A). Expression levels for the fusion proteins are comparable between the two strains (fig. S3B). Expression from separate mRNAs (strain 4), however, shows only 60 ± 3% of the specific luciferase activity measured in strain 3 (Fig. 1C). This difference was not caused by subunit aggregation (fig. S3C). Synthesis of complex subunits from one bicistronic mRNA encoded by an operon therefore strongly increases the efficiency of cis-assembly compared with trans-assembly.

We postulated that complexes might be assembled cotranslationally, effectively preempting preassembly diffusion. We used selective ribosome profiling (SeRP) to uncover any nascent subunit interactions during luciferase complex assembly in vivo (4). This method (4, 5) allowed us to compare the distribution profile of nuclease-protected mRNA fragments (ribosome footprints) isolated from all translating ribosomes to the profile of ribosomes selected by immunopurification (IP) of YFP-tagged luciferase subunits (fig. S4A, IP ribosomes). We constructed two strains of E. coli (strains 5 and 6), each containing a single luciferase operon where either luxA (strain 6) or luxB (strain 5), but not both, is fused at the 5′ end with yfp (fig. S4B). Genetic fusion of YFP to LuxA or LuxB N termini allows enrichment of potentially two classes of translating ribosomes: those directly translating the YFP-Lux protein (fig. S4A, IP part) versus those translating the untagged partner subunit, if and when cotranslational engagement of the YFP-Lux protein occurs (fig. S4A, co-IP part). The N-terminal YFP tagging allows IP of the first class of ribosomes engaged with the YFP-tagged subunit. Independent selection of yfp-luxB (strain 5) and yfp-luxA (strain 6) gene products generates a profile for each that shows strong ribosome footprint enrichment once the YFP part (248 residues) of the nascent YFP-LuxB and YFP-LuxA chains has emerged (fig. S5, upper panel), confirming that IP is highly selective.

We then compared the density of ribosome footprints across lux genes of both data sets, the immunopurified ribosomes and the total pool of all ribosomes. We find strong evidence for the second class of translating ribosomes: IP of YFP-LuxA copurifies ribosomes synthesizing LuxB, which directly demonstrates that nascent LuxB cotranslationally interacts with YFP-LuxA [Fig. 2A, right, without trigger factor (TF)]. This interaction is unlikely to occur after cell lysis because the crowded cytoplasm becomes highly diluted, which greatly reduces the possibility of nascent subunit interactions (supplementary text). As translation of luxB starts, footprint enrichment (reflecting interaction with YFP-LuxA) fluctuates around background levels but rises with increasing length of nascent LuxB. We defined a stable twofold enrichment threshold to reliably indicate initiation of cotranslational subunit interactions. According to this threshold, interaction of YFP-LuxA with nascent LuxB starts after ~120 residues of LuxB are synthesized and becomes more robust with increasing length until a 20-fold enrichment, reflecting strongly interacting subunits, is reached near the termination of luxB translation. Assuming the ribosomal exit tunnel protects 30 C-terminal residues, emergence of the 90 N-terminal residues of LuxB is evidently sufficient to initiate the assembly interactions. We also detected interactions of YFP-LuxB with nascent LuxA, at much lower enrichment (maximally sixfold), reflecting less robust interactions (Fig. 2A, left, without TF). We conclude that both nascent luciferase subunits can initiate complex assembly, but cotranslational assembly initiates predominantly on nascent LuxB.

Fig. 2 Assembly of bacterial luciferase occurs cotranslationally.

(A) Engagement of (left) nascent LuxA by YFP-LuxB and (right) nascent LuxB by YFP-LuxA without TF (strains 5 and 6, –TF) or with TF (strains 9 and 10, +TF). (Top) Cartoons illustrate luxA and luxB operons and the principle of SeRP analysis (fig. S4A). (Middle) Mean enrichments were corrected for nonspecific background binding analyzed with strains 11 and 12 (fig. S7A and table S2). Gray zones denote local regression curves (span parameter = 0.1) with a 95% confidence interval. Colored numbers indicate ribosome positions when the mean enrichments stably cross the twofold threshold. (Bottom) Cartoons show exposed nascent chain (n.c.) lengths. Orange, dimer interface; red, helices α2 and α3. aa, amino acids. (B) Crystal structure of the heterodimer of LuxA (blue) and LuxB (green) (Protein Data Bank accession code 1LUC). Orange, dimer interface; red, helices α2 and α3. LuxA and LuxB subunit residues involved in subunit interaction are listed in table S1.

For these experiments, we used a Δtig mutant background lacking the ribosome-associated TF chaperone. TF interacts transiently with most nascent cytosolic proteins a nascent chain with minimal average length of ~110 residues is synthesized (4). To test the effect of TF, we expressed wild-type levels of TF in strains 3 and 4 (making strains 7 and 8, respectively) and strains 5 and 6 (making strains 9 and 10). In vivo luciferase activity measurements of strains 7 and 8 indicate no effect of TF on the enhanced efficiency of cis-assembly (fig. S6). However, SeRP experiments of strains 9 (yfp-luxB in bicistronic organization with unlabeled luxA) and 10 (yfp-luxA in bicistronic organization with unlabeled luxB) reveal that TF does affect cotranslational interactions of the nascent luciferase subunits (Fig. 2A). Comparison of the cotranslational interaction of YFP-LuxB with nascent LuxA in the absence (strain 5) and presence (strain 9) of TF shows that TF efficiently suppresses interactions of YFP-LuxB with nascent LuxA (Fig. 2A, left). In contrast, TF delays but does not block the interaction of YFP-LuxA with nascent LuxB, shifting the minimal length of nascent LuxB that is required for cotranslational interaction from 90 to 152 residues (Fig. 2A, right), without affecting the ~20-fold enrichment of ribosome footprints near the end of luxB translation. We infer that TF chaperone activity shields nascent LuxB from premature interactions with LuxA but allows the exposure of the dimerization fold for timely interactions with LuxA molecules in the vicinity of the ribosome.

The luciferase heterodimer interface is large (residues 17 to 161 in LuxA and 11 to 163 in LuxB). Extensive contacts occur via a parallel four-helix bundle formed by helices α2 and α3 from either subunit (residues 55 to 65 and 83 to 97 in LuxB) (Fig. 2B and table S1) (2). Given the length of the ribosomal tunnel, α2 and α3 of nascent LuxB will be fully exposed after translation of the first 127 codons; the complete subunit interface of LuxB will be exposed after translation of codon 193. In the absence of TF, initial contacts of nascent LuxB with LuxA are established upon emergence of helix α3 (translation of the first 120 codons). With TF present, we detected initial interactions of nascent LuxB with LuxA only upon exposure of ~152 N-terminal LuxB residues (translation of the first 182 codons), constituting nearly the complete subunit interface (Fig. 2A, right).

We conclude that association of fully synthesized LuxA with nascent LuxB requires ribosome-associated exposure of critical structural elements of the dimerization interface on LuxB. TF further increases the specificity of subunit interactions by preventing association of LuxB with nascent LuxA and premature association of LuxA with short nascent chains of LuxB, but it selectively allows LuxA association with long nascent chains of LuxB, exposing the complete dimer interface. We propose that TF engagement with nascent LuxA prevents premature interactions of the latter with LuxB and effectively promotes timely delivery of full-length LuxA to nascent LuxB. Likewise, during the early stages of translation, interaction of TF with nascent LuxB may protect the nascent subunit from incorrect and premature folding and interactions, until the assembly-competent part of LuxB is sufficiently exposed at the ribosomal surface for productive interaction with LuxA. Our results provide proof of principle for organized and regulated protein assembly in bacteria and validate previous observations that suggested translational coupling affects protein complex assembly (6). The SeRP data provided for bacterial luciferase are consistent with the concept that a fully synthesized subunit encoded by an upstream gene interacts with the nascent subunit encoded by the downstream gene of an operon. Assembly, which is considered a final, posttranslational step in the generation of oligomeric proteins, emerges as a process that is physically and kinetically coupled to the fundamental processes of protein biogenesis, folding, and translation. Our findings add a further dimension to the concept of the operon, originally defined by Jacob and Monod as a genetic unit for coordinated regulation of transcription (7). The genetic organization into operons of genes for products destined for assembly into protein complexes determines local concentration of subunits and crucial timing of assembly.

Supplementary Materials

www.sciencemag.org/content/350/6261/678/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S8

Tables S1 to S3

References (820)

Notes S1 and S2

References and Notes

  1. ACKNOWLEDGMENTS: We thank A. Becker, V. van Noort, and C. Andréasson for valuable contributions; V. Sourjik and co-workers for help with in vivo FRET analysis; M. Langlotz and the fluorescence-activated cell sorting facility of the ZMBH for support; D. Ibberson for optimizing DNA sample preparation protocols; Y. Vainshtein for support with data analysis; and B. Zachmann-Brand for technical support. Sequencing was carried out at Genomics and Proteomics Core (DKFZ) and Genomics Core (EMBL) facilities. This work was supported by research grants from the German Science Foundation (DFG SFB638 and FOR1805) to G.K. and B.B.
View Abstract

Stay Connected to Science

Navigate This Article