Alternative mechanisms of translation initiation: an emerging dynamic regulator of the proteome in health and disease.

Abstract

Eukaryotic mRNAs were historically thought to rely exclusively on recognition and binding of their 5’ cap by initiation factors to effect protein translation. While internal ribosome entry sites (IRESs) are well accepted as necessary for the cap-independent translation of many viral genomes, there is now recognition that eukaryotic mRNAs also undergo non-canonical modes of translation initiation. Recently, high-throughput assays have identified thousands of mammalian transcripts with translation initiation occurring at non-canonical start codons, upstream of and within protein coding regions. In addition to IRES-mediated events, regulatory mechanisms of translation initiation have been described involving alternate 5’ cap recognition, mRNA sequence elements, and ribosome selection. These mechanisms ensure translation of specific mRNAs under conditions where cap-dependent translation is shut down and contribute to pathological states including cardiac hypertrophy and cancer. Such global and gene-specific dynamic regulation of translation presents us with an increasing number of novel therapeutic targets. While these newly discovered modes of translation initiation have been largely studied in isolation, it is likely that several act on the same mRNA and exquisite coordination is necessary to maintain ‘normal’ translation. In this short review, we summarize the current state of knowledge of these alternative mechanisms of eukaryotic protein translation, their contribution to normal and pathological cell biology, and the potential of targeting translation initiation therapeutically in human disease.

Introduction

Despite the fundamental necessity for protein translation to organismal life, historically research has focused on transcription as the central governor of the proteome. An appreciation for co- and post-transcriptional gene regulation has since developed with the discovery of alternative splicing, RNA modifications, and RNA localization as critical regulators of cellular physiology. One of many posttranscriptional mechanisms capable of altering the cellular proteome is translational regulation of mRNAs, fine-tuning protein expression and rapidly altering the proteome during stress [1, 2]. Alterations in translation initiation are a key feature of cancer, viral infection, and cardiac hypertrophy [3-5].

Modifying the translational program allows a rapid response to extracellular signals, effecting changes in protein expression from translation of existing mRNAs. Translation initiation is a point where many mechanisms of regulation converge to determine if an mRNA will be translated, and a canonical model of translation initiation central to our understanding of this process has arisen [6, 7]. In this model, recruitment of the ribosome is dependent upon recognition of the mRNA’s 5’ cap by eukaryotic initiation factor (eIF) 4E. Alternative mechanisms of translation initiation are characterized by distinct modes of initiation complex recruitment to the mRNA to facilitate translation initiation. These include 5’ cap-dependent and independent models with diverse molecular mechanisms such as alternate cap recognition and mRNA methylation capable of allowing the formation elongation competent ribosomes. There is also growing evidence to support a critical role of the ribosome itself in regulating translation. This brief review will discuss several models of alternative translation initiation and the evidence for their occurrence in eukaryotic cells, and provide examples of mRNAs for which alternative translation is necessary to maintain normal cellular function. Yeast have served as a powerful tool to gain mechanistic insight in studies of translation initiation, however for this review we focus on studies conducted in mammalian model systems. Translation of upstream open reading frames (uORFs), repeat associated non-AUG (RAN), and other models of alternative translation activated during pathological states have been the subjects of recent reviews, to which we refer readers [8-11].

Mechanisms of eukaryotic translation initiation

eIF4E cap-dependent translation

The canonical mode of eukaryotic translation initiation begins with formation of a ternary complex composed of eukaryotic initiation factor (eIF) 2, a methionine charged transfer RNA, and GTP (reviewed in [12]). Binding of the ternary complex with the 40S ribosomal subunit is promoted by eIF1, eIF1A, and eIF3, forming the pre-initiation complex (PIC). Eukaryotic mRNAs bear a 5’ cap of 7-methylguanosine (m⁷G), necessary for this mode of translation initiation and maintaining mRNA stability. The cap recognition protein eIF4E binds the 5’ cap, and together with eIFs 4A, 4G, and 4B recruits the PIC to the 5’ mRNA end. Interactions of the eIF4G scaffolding protein and poly(A) binding protein (PABP) circularize the mRNA and the complex scans in a 5’ to 3’ manner until arriving at a start codon in a favorable context. Upon start codon recognition, GTP is hydrolyzed, binding of the 60S ribosomal subunit forms the 80S complex and translation is initiated (Figure 1A).

Figure 1. Mechanisms of translation initiation currently known.

A) eIF4E recognition of the 5’ mRNA cap facilitates binding of the 43S preinitiation complex in the canonical pathway of translation initiation. Upon recognition of the initiation codon the 60S subunit joins and translation is initiated. This mechanism of eIF4E and cap-dependent translation initiation is responsible for the majority of mRNA translation in the eukaryotic cell. B) N(6)-methyladenosine (M⁶A) in the mRNA 5’ UTR can directly bind the eIF3 multiprotein complex and allow recruitment of the 43S complex to initiate translation independent of the 5’ cap. C) A subset of mRNAs are bound by the d subunit of the eIF3 complex. eIF3d cap recognition directs translation initiation allowing cap-dependent translation independent of eIF4E, the canonical cap recognition protein. D) Highly structured mRNA elements, in many cases with the assistance of IRES trans acting factors (ITAF) recruit ribosomes to specific start codons in viral mRNAs, allowing cap-independent translation initiation. The requirement for mRNA structure, ITAFs, and other factors in eukaryotic IRES remains a subject of research. E) Ribosomes vary in their ribosomal protein composition, and selectively translate mRNAs. Ribosomal proteins, in tandem with cis elements of the message, may direct ribosomes to translate specific mRNAs in changing conditions.

The necessity for eIF4E in canonical cap-dependent translation make it a critical regulatory factor; eIF4E expression is tightly regulated and eIF4E activity is governed by a family of eIF4E binding proteins (4E-BPs) which sequester and inactivate eIF4E [3, 13, 14]. 4E-BPs are primarily regulated by the mechanistic target of rapamycin (mTOR) pathway, where phosphorylation of 4E-BPs by mTOR releases eIF4E to promote 5’ cap and eIF4E-dependent translation initiation [15, 16]. Degradation or sequestration of components of the eIF4F complex inhibits this mechanism and is targeted by many viruses to induce a ‘global’ shutdown of cellular mRNA translation [17-19]. The finding that a large number of cellular mRNAs are still translated under conditions in which eIF4E cap-dependent translation is inhibited was an early clue that alternative mechanisms of translation initiation exist [20].

Alternative 5’ cap-dependent models of translation initiation

The eIF3 complex promotes efficient canonical, cap-dependent translation initiation and is central to several alternative models of translation initiation [21-23]. Lee et al. (2015) performed photoactivatable ribonucleoside-enhanced crosslinking and immunoprecipitation (PAR-CLIP) to identify eIF3-interacting transcripts [24]. eIF3 selectively bound mRNAs involved in cell growth and proliferation, exerting translational control by binding the 5’ untranslated regions (UTRs) of these transcripts. Four subunits of eIF3 cross-linked directly to mRNAs, and deep sequencing of these eIF3 bound mRNA sequences revealed specific binding to just under 500 cellular transcripts [24]. Notably, the authors find that eIF3 can act as a positive and negative regulator of translation, dependent upon transcript cis elements [24]. eIF3 complex composition ranges from as few as five subunits in yeast, considered the eIF3 “core complex” due to its species conservation, up to thirteen subunits in mammalian cells, with requirement for distinct subunits varying between initiation mechanisms [25, 26]. The eIF3d subunit has been shown to bind the 5’ cap directly and allow cap-dependent translation initiation in the absence of eIF4E [27, 28]. The c-Jun mRNA on which eIF3d cap binding activity has been described, harbors an RNA sequence element that appears inhibitory to eIF4E cap binding. In this study, alternate 5’ cap recognition promotes translation initiation, however an inhibitory role of novel cap binding proteins has also been described. In an effort to identify the mechanism by which La-related protein 1 (LARP1) represses translation of terminal oligopyrimidine (TOP) mRNAs specifically, Lahr et al. reveal 5’ cap binding by LARP1 blocks eIF4F complex assembly to negatively regulate translation [29, 30]. Together, these data highlight the importance of alternate 5’ cap recognition in the regulation of eukaryotic translation initiation.

The role of eIF3 as a governor of the translational program during stress is now coming into focus, with subunits including eIF3d central to translation initiation during chronic endoplasmic reticulum (ER) stress [31], Guan et al. describe an eIF3-dependent translation initiation program that is largely independent of eIF4E and necessary for translation of stress-induced proteins and “translational recovery” during stress [31]. This study highlights eIF3 complex modularity by demonstrating unique signatures of protein synthesis upon knockdown of individual eIF3 subunits during ER stress. Depletion of eIF3d, the subunit proposed to act as an alternate cap binding protein by Lee et al. (2016), had the most profound effect on translational recovery and stress-induced protein synthesis, supporting a model where eIF3 complex components regulate mRNA translation [27, 31].

mRNA sequence elements in translation initiation

Alternative mechanisms of translation initiation can involve mRNA cis-elements that facilitate translation initiation. The most intensely studied of these is the internal ribosome entry site (IRES), first described in viral mRNAs [32-35]. Viruses utilize IRESs to hijack cellular translation machinery, concomitantly shutting down cap-dependent translation while maintaining translation of viral proteins [17, 19, 32, 36]. Viral IRESs are categorized into four classes based upon size, structure, and requirement for eIFs [37]. The viral IRES is typified by its ability to facilitate ribosome assembly and translation initiation independent of a 5’ mRNA cap [32, 38]. Two of the simplest viral IRESs, often serving as models in research, are derived from the hepatitis C (HCV) and cricket paralysis (CrPV) viruses. These IRESs are highly structured regions of the viral mRNA, with secondary and tertiary folding essential for function [39-41]. The HCV and CrPV IRESs are examples of type 3 and 4 IRESs, respectively; the 40S ribosomal subunit binds these IRESs with high affinity, followed by binding of 60S subunit and ternary complex. They differ in their requirement for eIFs, with HCV IRES binding to the 40S subunit dependent upon eIF3 [39]. Type 4 IRESs such as that of CrPV are capable of direct recruitment of the 40S subunit [41, 42]. The mechanism by which an IRES facilitates open reading frame (ORF) selection by the ribosome remains a topic of study, however it is clear that components of the IRES are capable of directing the ribosome to a specific ORF [37]. More complex viral IRESs require additional host cell initiation factors and have variations in mechanism of action, however all catalyze formation of an elongation competent ribosome complex independent of 5’ cap binding and/or a free 5’ mRNA end [37].

Though Pelletier and Sonenberg (1988) speculated that IRES-like translation initiation could occur on cellular mRNAs, the phenomenon was thought to occur exclusively on viral mRNAs until the discovery that the cellular protein GRP78/BiP was translated during viral infection [32, 43]. Since, evidence for alternative translation initiation driven by mRNA sequence has arisen in several cellular mRNAs [44-47]. Cap-independent translation has been attributed to IRES activity within many of these, and high-throughput screening has identified thousands of putative cellular IRESs often occurring within the 5’ UTR [48-52]. Molecular virology has been critical to our understanding of IRES translation initiation but it is increasingly appreciated that eukaryotic ‘IRESs’ differ significantly from viral IRESs in structure and function. Recent studies have reported IRES activity in eukaryotic circular mRNAs, which have to date only been described in a small number of viruses and viroids [53-55]. Advancements in technology have revealed alternate promoter usage and transcriptional changes leading to generation of distinct mRNA transcripts responsible for the expression of some proteins previously attributed to IRES activity [56, 57]. This work has highlighted the need for rigorous controls when evaluating IRESs, and for a more nuanced view of translation initiation including categories beyond the ‘cap-dependent’ or ‘IRES’ designation [58]. Importantly, IRESs alone cannot explain the multitude of mRNAs that continue to be translated despite inactivation of cap-dependent pathways and increasing examples of translation initiation within protein coding sequences [44, 45]. Cap-independent translation elements (CITEs) are mRNA sequence elements capable of recruiting the translation initiation machinery, initially described in plant viruses [59, 60]. Shatsky and colleagues have proposed the use of this term to describe mRNA sequence features driving cap-independent translation initiation but require a free 5’ end, thus likely requiring ribosome scanning [61, 62]. The nature of these elements and the mechanism by which they drive translation remains elusive, but the recent discovery of an RNA modification capable of driving cap-independent translation alludes to the probability that many non-canonical mechanisms of translation initiation have yet to be discovered.

N⁶-methyladenosine (m⁶A) residues are found throughout the coding sequence and 5’ and 3’ UTRs of cellular mRNAs [63]. The presence of m⁶A in mRNA 5’ UTRs can effect translation initiation independent of the eIF4F complex and 5’ cap binding [64, 65]. Intriguingly, 5’ UTR m⁶A can promote translation initiation from specific start codons, a function previously attributed to the cap binding complex and scanning mechanism. Ribosome toeprinting and in vitro translation assays indicate that 5’ UTR m⁶A preferentially facilitates translation initiation at the most 5’ AUG start codon in some transcripts [64]. Cap-independent translation of a subset of mRNAs is regulated by the methyltransferase METTL3, and together with ABCF1 mediates cap-independent protein synthesis in response to heat shock stress [65]. Combining identification of m⁶A modifications with deep sequencing to describe m⁶A initiated translation revealed the necessity for this mechanism of translation during the integrated stress response, a signaling pathway activated in response to a range of pathological changes [66, 67]. Mutation of the site of m⁶A methylation in ATF4 suppressed translation of an upstream ORF, thus promoting expression of the downstream main ATF4 ORF. This finding provided insight into the mechanism by which m⁶A mRNA methylation allows start codon selection, as ribosomes reinitiate at a downstream codon followed by m⁶A-mediated translation of uORFs [66]. While the role of m⁶A within the 5’ UTR has primarily been investigated in the context of translation initiation, the presence of m⁶A within the coding sequence of human mRNAs and the ability of m⁶A to direct ribosomes to specific sites raises the possibility that m⁶A could recruit ribosomes to sites within the coding sequence, representing a potential mechanism by which truncated protein isoforms could be translated [63].

Trans-acting mechanisms of translation initiation

Proteins acting in trans have been implicated in alternative translation initiation and can be separated into two groups; RNA binding proteins (RBPs), and the ribosome itself. RBPs play a pivotal role in almost every step of RNA transcription and processing, while also acting on viral and cellular mRNA translation initiation [68, 69]. As an mRNA is transcribed, it is almost immediately bound by trans-acting proteins and RNAs that form the ribonucleoprotein (RNP) complex. This complex is critical in maintaining RNA stability and assisting in remaining steps of mRNA biogenesis, as well as in governing mRNA splicing, translation, and degradation [70]. A group of RBPs described as IRES trans-acting factors (ITAFs) are thought to recruit ribosomes to cellular IRESs and drive mRNA translation [71-73]. Insulin-like growth factor 2 mRNA binding protein 1 (IMP1) is one of three IMPs (IMPs 1-3) with a proposed role in carcinogenesis where increased IMP1 expression promotes cap-independent translation initiation of oncogenic proteins including cellular inhibitor of apoptosis 1 (cIAP1), inhibiting apoptosis [73, 74]. Action of IMP1 as an ITAF was initially recognized in translation of insulin-like growth factor 2 (IGF2), where binding of IMP1 to the IGF2 transcript’s 5’ UTR is necessary for cap-independent translation [75]. In the absence of IMP1, IGF2 transcripts dissociated from the translational polysome, indicating that IMP1 is necessary for translation initiation and leading the authors to identify several residues necessary for IMP1’s mRNA binding activity [76]. IMP1 also targets and modulates internal translation of cIAP1, where a similar dissociation of the cIAP1 mRNA from polysomes occurs when IMP1 is silenced. IMP1 binds directly to the cIAP1 5’ UTR, and as in the case of IGF2, the IMP1 interaction with the 5’ UTR alone appears sufficient to initiate internal translation [77].

Polypyrimidine tract-binding protein (PTB) is a ubiquitous protein initially recognized for its role in RNA splicing, later found to have a number of roles in the RNA life cycle including mRNA stability and translation initiation [78]. The role of PTB in IRES-mediated translation initiation has been studied extensively, and it has been described as essential or a strong enhancer of viral translation initiation [35, 79, 80]. Early in the study of putative cellular IRESs, PTB was identified as an ITAF with a number of targets and necessary for their translation [79, 80]. PTB recognizes and binds polypyrimidine-rich tracts, and synthetic mRNAs with inserted tract sequences demonstrated IRES activity in a bicistronic reporter assay in the presence of PTB [81, 82]. Complexing of PTB with RNA during splicing events results in structural RNA remodeling partially responsible for PTB’s ability to repress alternative splicing [83]. A similar mechanism of regulation via mRNA structural changes has been proposed for PTB’s role in activating translation initiation. Binding of PTB alters mRNA secondary structure, and therefore nucleotides exposed for ribosome binding. This facilitates binding of the ribosome within the mRNA and activates translation initiation [71, 84]. While PTB is one of the most rigorously studied, other RNA binding proteins necessary for translation initiation of viral and putative cellular IRES have been identified, examples of which are listed in Table 1.

Table 1. Examples of IRES trans acting factors (ITAFs).

Cellular proteins that have been identified as necessary and/or sufficient to initiate translation of viral and eukaryotic mRNAs. Many of these proteins act in concert with cis elements of the mRNA to drive translation initiation.

ITAF	References	Organism
PTB/hnRNP1	[112-115]	Viral, Eukaryotic
hnRNPA1	[116-118]	Eukaryotic
ITAF(45)	[112]	Viral
La autoantigen	[119]	Viral, Eukaryotic
UNR	[120, 121]	Viral, Eukaryotic

The eukaryotic ribosome comprises ~80 ribosomal proteins (RPs) assembled with four ribosomal RNA (rRNA) species to generate the small and large ribosomal subunits. The concept of “specialized” ribosomes capable of selectively translating mRNAs was alluded to in 1987, when in vitro mutations introduced within RPs directed the ribosome to specific transcripts [85]. Evidence for eukaryotic ribosome heterogeneity arises at multiple levels: RP expression, post-translation modifications, ribosome associated proteins or factors, and rRNA [86]. Stoichiometric variation in RP levels have been described, and a functional role for specific RPs implicated in vertebrate development [86-88]. Utilizing a viral IRES to understand mechanisms of ribosome recruitment, ribosomal protein s25 (rps25) was identified as necessary for hepatitis C viral RNA cap-independent translation initiation [89]. A quantitative study of ribosome composition in which mass spectrometry was used to determine absolute abundance identified four RPs existing in sub-stoichiometric quantities [90]. This suggests ribosome populations that differ in RP composition, and using immunoprecipitation and ribosome footprint sequencing revealed distinct populations of mRNAs translated by rps25 containing ribosomes. The same study investigated rpl10a, a large subunit protein also sub-stoichiometric in polysomes, finding that it too is necessary for translation of functionally linked mRNAs [90]. These data highlight a mechanism by which RPs confer ribosomal specificity, perhaps rendering it more permissive to initiate through the canonical pathway or proceed via an alternative translation initiation mechanism. This could be particularly important during normal conditions when canonical translation is not compromised, as it would allow selection from various modes of translation initiation [44, 45].

Human RP mutations can perturb ribosome function and result in a number of disorders described as ribosomopathies. A unifying feature of many of these ribosomopathies is tissue specific phenotypes, despite dysfunction of a protein complex essential in every cell. The ribosome concentration model serves to explain this this phenomenon, as an alternate to the ribosome specialization model (reviewed in [91]). Diamond-Blackfan Anemia (DBA) is a ribosomopathy in which the majority of cases involve mutations of RPs resulting in RP haploinsufficiency [92]. DBA causing mutations in the hematopoietic transcription factor GATA1 are directly linked to this mRNA’s translational sensitivity to ribosome availability rather than altered ribosome composition [93]. Khajuria et al. (2018) further defined the role of ribosome levels not only in DBA but in normal hematopoietic differentiation [94]. The authors reconstitute DBA causing mutations in human hematopoietic cells and identify a pool of transcripts, including GATA1, sensitive to RP haploinsufficiency and the resulting reduced ribosome availability. The 5’ UTRs of these ‘ribosome concentration’ sensitive transcripts were shorter and unstructured; features determinate of translation initiation rates [94, 95]. The authors exclude a contribution of variations in ribosome composition in this particular context. Each of these models highlights the ability of the ribosome to regulate translation globally and at the gene-specific level. The ribosome concentration model directly links translation initiation rate with ribosome availability to dictate the rate of protein synthesis, and the purported role of RP composition in alternative modes of translation initiation specifically alludes to the possibility of these models coexisting.

New roles for alternative translation initiation

Several eukaryotic mRNAs that undergo alternative translation initiation to generate novel protein isoforms with distinct functions have now been described [44, 45, 96-99]. GJA1 and MAVS are discussed here as exemplars of this growing family of mRNAs undergoing alternative translation initiation in basal conditions.

Gap junctions, composed of connexin proteins, allow direct intercellular communication by connecting the cytoplasms of many cell types. Gap junction loss is a hallmark of cardiac disease in addition to a number of other pathologies [100-102]. Smyth and Shaw (2013) demonstrated that GJA1 mRNA which encodes Cx43, the most ubiquitously expressed connexin, undergoes an alternative translation initiation event to generate a truncated protein isoform termed GJA1-20k. Levels of GJA1-20k regulate formation of Cx43 gap junctions and are dynamically regulated by growth factor signaling [44]. The presence of an internal ribosome entry site (IRES) in the GJA1 mRNA 5’ UTR was first proposed by Schiavi et al. (1999), and is thought to allow cap-independent translation of full length Cx43 protein encoded for by GJA1 [50]. Ul-Hussain et al. (2014) later proposed the existence of IRES activity within the coding region of Cx43, and that cap-independent translation initiation from this site could be responsible for synthesis of GJA1-20k [103]. Experiments performed by Salat-Canela et al. (2014) however demonstrate the requirement for a 5’ cap and upstream ribosome scanning for translation of GJA1-20k [104]. While it is unlikely these internal coding sequence elements of GJA1 act as a bona fide IRES, it is clear their activity is important for alternative translation initiation and could be working in concert with other components to guide the translation machinery to this internal start codon. In the cardiomyocyte, where Cx43 gap junctions are necessary for the electrical conduction that facilitates each heartbeat, ectopic expression of GJA1-20k regulates actin dynamics. This promotes gap junction formation and protects against gap junction loss during acute ischemia, providing proof of principle that targeting translation initiation could protect against loss of electrical coupling in the heart [105]. We have recently found that GJA1-20k positively regulates trafficking of full length Cx43 through promoting Cx43 hemichannel oligomerization, and suppression of alternative translation initiation is a mechanism by which Cx43 gap junctions are downregulated during epithelial-mesenchymal transition (EMT) [106]. Ectopic expression of GJA1-20k again rescues gap junction formation in this model of gap junction loss, providing further evidence that targeting of translation initiation could be utilized to rescue gap junction formation in disease. GJA1-20k also appears to have a role in the cellular response to oxidative stress, preserving mitochondrial localization and function [107]. The diversity of GJA1-20k function highlights the importance of alternative translation initiation in regulation of normal cellular functions across tissue types, and alterations in translation initiation in response to stress demonstrate this is a dynamically regulated process.

MAVS is an intermediate adaptor protein that induces an antiviral response upon recognition of viral RNA by RIG-1. Alternative translation initiation at a methionine start codon within the MAVS protein coding sequence generates a truncated isoform termed ‘miniMAVS’ that regulates interferon (IFN) production [108]. The ratio of full length MAVS to miniMAVS determines IFN production and modulation of miniMAVS expression directly affects viral replication. Protective effects of IFN production are numerous and include suppression of viral replication, however during chronic infections sustained IFN production can result in immunosuppression and an inability to clear viruses [109, 110]. Negative regulation of IFN by miniMAVS and the importance of miniMAVS/MAVS ratio to IFN production suggests alternative translation initiation of the MAVS transcript as a mechanism by which the cell response to infection is fine tuned. Regulation of translation initiation of the MAVS transcript therefore has significant implications for the innate cellular immune response. Brubaker et al. also shed light on the mechanism of miniMAVS translation, revealing a role for ‘translational context,’ of the mRNA sequence surrounding a start codon in regulation of MAVS and miniMAVS expression. Insertion of artificial start codons within the MAVS mRNA indicate ‘leaky’ ribosomal scanning as ribosomes read through 5’ end AUGs and proceed to a start codon within the coding sequence, generating miniMAVS. This could also be a scenario in which ribosomes could act synergistically to regulate alternative translation initiation, with specific RPs or ribosome availability resulting in leaky scanning and thus translation at the internal start site. Ribosome profiling, in which harringtonine can be used to stall ribosomes at the point of translation initiation, provided one of the first lines of evidence that alternative translation initiation events could occur on significantly more transcripts than previously thought [111]. Brubaker et al. (2014) then performed ribosome profiling in human monocytes, and describe a subset of proteins of the innate immune response that also undergo alternative translation initiation events [45]. Identification of a group of mRNAs with similar function that undergo alternative translation initiation leads one to speculate that their regulation may allow unified alterations in expression upon infection.

Conclusion and Future Directions

Given the high cellular energy demand of protein translation, it is unsurprising that distinct mechanisms that exquisitely regulate this process occur. Canonical, cap-dependent translation initiation proceeds with greater efficiency than many alternative mechanisms of translation initiation and remains responsible for a large portion of protein output during cellular homeostasis. Previously believed to contribute primarily to the cellular stress response, it is now clear that alternative translation initiation also makes significant contributions to protein expression under basal conditions. Regulation of translation initiation and alternative mechanisms of translation initiation are key components of cellular physiology and disruption of this intricate regulation plays a role in disease. While many of these novel modes of translation initiation have been discovered and investigated in isolation, future studies will need to consider how multiple mechanisms of translation initiation function together to regulate cellular physiology.