Benefits of co‐translational complex assembly for cellular fitness

Complexes of two or more proteins form many, if not most, of the intracellular “machines” that execute physical and chemical work, and transmit information. Complexes can form from stochastic post‐translational interactions of fully formed proteins, but recent attention has shifted to co‐translational interactions in which the most common mechanism involves binding of a mature constituent to an incomplete polypeptide emerging from a translating ribosome. Studies in yeast have revealed co‐translational interactions during formation of multiple major complexes, and together with recent mammalian cell studies, suggest widespread utilization of the mechanism. These translation‐dependent interactions can involve a single or multiple mRNA templates, can be uni‐ or bi‐directional, and can use multi‐protein sub‐complexes as a binding component. Here, we discuss benefits of co‐translational complex assembly including accuracy and efficiency, overcoming hidden interfaces, localized and hierarchical assembly, and reduction of orphan protein degradation, toxicity, and dominant‐negative pathogenesis, all serving to improve cell fitness.


INTRODUCTION
Proteins are "machines" of the cell, and complexes of proteins facilitate the elaborate but precise activities required for nearly all physiological processes including signal transduction, catalysis, and maintenance of structural integrity, among others. Accurate and timely assembly of constituents is essential for formation of functional protein complexes, that can be homomeric or heteromeric in nature, and with well-defined three-dimensional architectures. A major challenge of complex formation is the identification of partners within the crowded cell interior.
In prokaryotes, polycistronic operons encoding constituents of a complex reduces their diffusion, thereby directing localized assembly of newly translated proteins in a spatially-restricted manner. However, in F I G U R E 1 Cell fitness benefits of co-translational assembly of complexes. Schematic of a uni-directional co-translational interaction to form a heterodimer (top). Three classes of benefits of co-translational interactions during complex formation (bottom).
sites, among others. An alternate mRNA translation-dependent, that is, co-translational, assembly mechanism has recently gained substantial recognition and experimental support. In the canonical co-translational mechanism, a mature, fully formed protein interacts with its partner, generally near the nascent N-terminal polypeptide, as it emerges from the exit tunnel of the ribosome. [1,2] In exceptions to this mechanism, both interacting polypeptides can be mRNA-bound, that is, in-cis, in which polypeptides emerging from the same mRNA interact, or in-trans where interacting polypeptides emerge from distinct translating mRNAs. Global investigations of complex assembly suggest a widespread process where 12 of 31 and 9 of 12 complexes in fission and budding yeast, respectively, appear to be assembled cotranslationally. [3,4] As illustrated here, recognition of an important role of co-translational complex assembly in the mammalian proteome is emerging. Moreover, the co-translational interactions provide novel mechanisms that overcome challenges not adequately addressed by post-translational interactions ( Figure 1).

CO-TRANSLATIONAL COMPLEX ASSEMBLY FACILITATES PROTEIN-PROTEIN INTERACTION SPECIFICITY
Accurate and timely identification of a cognate partner among the dense sea of incorrect partners is a major obstacle during proteinprotein interaction for formation of complexes. The challenge is even greater during assembly of multi-protein complexes where a given constituent interacts with more than one partner, and thus exhibits multiple binding surfaces. A principal advantage of the co-translational mechanism is the reduction of inappropriate or non-specific interactions between proteins with promiscuous binding surfaces. Such interactions reduce the likelihood of formation of the correct complex, and thus reduce cell fitness; also, in cases where a constituent provides protection for its partner, partner absence can cause rapid degradation.
The cell has evolved multiple strategies to overcome the challenges of accurate complex formation, and many involve co-translational interactions.
In view of the likely direct relationship between interface size and tendency for low-specificity interaction, a recent report considered the role of interface size in co-translational interactions. [5] Importantly, in most proteins, interfaces near the N-terminal are larger than C-terminal interfaces, consistent with the concept that N-terminal co-translational interactions has permitted evolution of larger interfaces. A particularly difficult challenge is the correct selection of interacting partners in the situation in which there are multiple similar structural homologs and isoforms. In one case, the selection of homodimers versus heterodimers, there is a relatively simple and elegant solution. As an example, the homodimerization of lamin C, encoded by the human LMNA gene was investigated. [6] Lamin C and lamin A are alternatively spliced isoform products of the same gene that share the same N-terminal dimerization domain, but are present exclusively as homodimers in vivo. The solution, at least for lamin C, is dimer formation from two interacting Nterminal peptides emerging from adjacent ribosomes transiting a single mRNA. The mechanism prevents the diffusion of any constituent away from the biosynthetic machinery, thereby reducing heteromeric interactions. The mechanism features ribosome-ribosome interactions to generate disomes, that are detectable by sucrose gradient analysis.
Importantly, a global analysis of disomes suggests the in-cis, co-translational mechanism is in widespread use for formation of homo-oligomers. [6] The analysis revealed the coiled-coil is the most common interaction domain for formation of homodimers, but other domains utilized include BTB, BAR, SCAN, and RHD domains.
Another structural advantage of the co-translational mechanism is the accurate mediation of interactions between one or more inaccessible interfaces not amenable to post-translational binding. In some cases, proteins known to bind in cells fail to interact during in vitro reconstitution of the purified proteins. [7] The ineffective interaction is often attributed to the absence of unknown assembly factors or posttranslational modifications. Co-translational interaction provides an alternative explanation for these observations. For example, ezrin and ezrin binding protein 50 (EBP50) form a stable heterodimeric complex in cells, but the purified proteins do not interact in vitro. [2] Following maturation, an interaction between ezrin N-and C-termini direct a closed conformation that masks the EBP50 binding site near the Nterminus. A co-translational engagement between fully formed EBP50 and the N-terminus of ribosome-bound ezrin overcomes this steric obstacle and facilitates formation of the EBP50-ezrin complex. [2] In addition to overcoming spatial obstacles, co-translational interactions can overcome topological obstacles, for example, interlocking ring structures where the interacting domains might be surface-accessible, but nonetheless not permissive for interaction of the fully-formed proteins. To our knowledge this class of co-translational interaction has not yet been reported.

CO-TRANSLATIONAL COMPLEX ASSEMBLY PROMOTES CELL HEALTH BY REDUCING ADVERSE CONSEQUENCES FROM INCORRECT OR FAILED INTERACTIONS
Protein steady-state levels are determined by rates of synthesis and degradation. In many cases, constituents of complexes are rapidly degraded when not associated with its cognate binding partner. Multiple in vivo studies support this mechanism as deletion of a complex constituent can reduce the level of their partner protein without altering mRNA expression. [8] Likewise, some successful protein purifications require the presence of protective interaction partners. A substantial influence of co-translational interactions on protein stability has been reported by several laboratories. Saccharomyces cerevisiae exhibits a heterotrimeric multi-tRNA synthetase complex (MSC) consisting of two aminoacyl tRNA synthetases, GluRS and MetRS, and the Arc1p scaffolding protein that binds the complex. [9] All contain GSTlike structures near their N-termini. GluRS and MetRS interact with each other by a bi-directional, co-translational mechanism, and both co-translationally bind Arc1p. [3] Glucose depletion disassembles the complex, with MetRS and GluRS relocalization to the nucleus and mitochondria, respectively, accompanied by rapid degradation of Arc1P.
Thus, the co-translational interactions of the synthetases result in stabilization of Arc1p.
A global analysis of yeast complexes revealed that complex constituents that are nascently engaged, that is, ribosome bound, are prone to aggregation or degradation in the absence of their partner subunits. [3] Beyond influencing degradation and aggregation, complex constituents can have cell-deleterious effects when present in free, uncomplexed form. Caspase-activated DNAse (CAD) causes chromosomal DNA fragmentation and apoptotic cell death. CAD binding to inhibitor of CAD (ICAD) in proliferating cells prevents the deleterious CAD activity. ICAD binds nascent CAD polypeptide while ribosomebound, preventing even a transient appearance of free protein and thereby minimizing susceptibility to degradation. [10] Intriguingly, the chaperones Hsc70 and Hsp40 also bind nascent CAD illustrating the collaboration of co-translational interaction and co-translational folding.
A surprising advantage of co-translational assembly is the ability to buffer the deleterious effect of dominant-negative (DN) mutations that occur when expression of a mutant allele disrupts the function of the wild-type (WT) allele. [1] DN mutations with a dominant mode of inheritance are more frequently observed in homomeric complexes as the interaction between the mutant and WT protein disrupts activity of the holo-complex, despite the presence of one functional subunit. Random post-translational interaction between the WT and DN proteins would generate dimers in which about three-quarters contain one or more of the "poisonous" DN form and thus has reduced or no activity. The cis mode of co-translational assembly of homomeric complexes theoretically reduces this challenge by generating equal amounts of WT or mutant homomers, thereby reducing the injurious effect of the DN mutation. Consistent with the theory, homodimeric subunits assembled co-translationally were shown to be less likely associated with autosomal dominant relative to recessive disorders. [11] Moreover, interaction interfaces in DN mutant proteins are generally exposed late during translation consistent with post-translational assembly of associated complexes. Certain proteins when disengaged from their host macromolecular complex can induce activities that can be considered to be inappropriate rather than injurious. For example, the tRNA synthetases in the yeast heterotrimeric MSC described above perform specific functions when released from the MSC in response to glucose deprivation; the inappropriate performance of these functions under basal conditions could possibly disturb metabolic homeostasis. The corresponding mammalian MSC is much larger than the yeast version, containing eight tRNA synthetases and three auxiliary proteins. Several of these proteins are released from the MSC by specific condition-dependent stimuli to perform moonlighting functions unrelated to their primary role in protein synthesis. Like the yeast version many of the sub-complexes of the human MSC are formed cotranslationally, thus reducing the likelihood of their presence in free form performing an inappropriate activity. [12]

CO-TRANSLATIONAL INTERACTIONS FACILITATE EFFICIENCY, HIERARCHY, AND LOCALIZATION OF COMPLEX ASSEMBLY
The speed and efficiency of complex formation can be critical for responses to stimuli and for cell fitness in general. Clearly, in-cis assembly of homodimers offers a highly efficient and rapid mechanism as the nascent chains are adjacent in the polysomes ensuring both speed and accuracy. In the case of proteins interacting in-trans, that is, assembled from peptides emerging from distinct mRNAs other distinct mechanisms are used. As an example, ribosome profiling of components of the yeast proteasome base showed the ribosome pauses on the mRNA near the N-termini of the constituents Rpt1 and Rpt2, thereby increasing the dwell time and facilitating the co-translational interaction of the emerging polypeptides. [13] Intriguingly, the interaction overcomes ribosome pausing, permitting rapid completion of translation of the interacting constituents. Possibly, codon bias, or other sequence or structural features, facilitates pausing by mechanisms yet to be determined.
Large multi-protein complexes are generally formed in an ordered fashion in which the assembly of well-defined sub-complexes precedes the formation of the mature complex, possibly requiring temporal identification and interaction of partners. The co-translational assembly mechanism provides a foundation in which post-translational and co-translational interactions between constituents are intermixed in a temporally ordered manner. For example, constituents of the mammalian nucleopore complex (NPC) interact both co-and posttranslationally. [14] Nsp1 interacts co-translationally with Nup57 allowing the resultant Nsp1:Nup57 heterodimer to post-translationally interact with Nup49 to form the Nsp1:Nup57:Nup49 heterotrimer, that is, the central transport Nups (CTN) sub-complex. Once formed, the CTN complex interacts co-translationally with Nic96. This hierarchical mechanism permits efficient assembly of complex multi-protein complexes in a temporally ordered manner. Many, if not most, constituents of multi-protein complexes interact with more than one partner. In one scenario, the co-translational interaction with the Nterminal of an emerging protein certainly precedes formation and folding of the C-terminal which is subsequently available for co-or post-translational interaction with a third constituent, thereby determining a hierarchical assembly. Sub-complex oligomerization by cotranslation can reduce this obstacle by restoring a first-order reaction mechanism. [14] The human MSC, mentioned above, is also assembled by a combination of co-and post-translational events, in fact, of the 121 possible one-to-one interactions between the 11 constituents, 15 are co-translational. [12] Several co-translational events involve more than a duo of interacting proteins. A "multi-site" mechanism in which two or more mature proteins bind the same nascent peptide at distinct sites was observed, and a "piggy-back" mechanism in which a mature carrier protein bears a second protein and binds a single site on an emerging peptide. The multimodal mechanisms of co-translational interaction offer diverse pathways for ordered, piecewise assembly of sub-complexes into larger heteromultimeric complexes.
Interaction of sub-complexes to form large multimeric complexes is hampered by the relatively low cytoplasmic concentration of the sub-complexes compared to monomeric forms. Localized translation and co-translation of protein partners at their site of function can lower the diffusion barrier to interaction. Karyopherins, the superfamily of nuclear transport receptors, interact co-translationally with Nup1/Nup2 proteins via their N-terminal during localized translation at the NPC. [15] The mechanism facilitates both accurate complex formation and delivery to the nuclear pore. Uncoupling of co-translation induces Nup1 aggregation and limits NPC biogenesis. The generality of co-translation-coupled complex formation in localized translational foci remains to be determined.

CONCLUSION
Burgeoning reports of co-translational interactions in eukaryotes is strongly indicative of a pervasive mechanism of complex formation.
Despite the elucidation of novel mechanisms, much remains to be discovered. Although certain N-terminal structures are prevalent in cotranslational events, general rules for regulation of co-translation are not known. Such rules might entail specific amino acid sequences, preferred codons, or specialized ribosomes. Elucidation of the cell typeand condition-specific homomeric and heteromeric co-translatome of mammalian cells is likely to expand our understanding of regulatory mechanisms. Equally important is the elucidation of pathology induced by disrupted or inappropriate co-translation. Finally, identification of therapeutic modalities to induce or restrict specific co-translational events might provide a new tool in the armamentarium of disease targeting.

AUTHOR CONTRIBUTIONS
Krishnendu Khan and Paul L. Fox have contributed equally to the conceptualization and writing of the manuscript.