Translation Initiation in Eukaryotes

Translation initiation is the process by which the ribosome is assembled at the correct start codon on an mRNA, in the correct reading frame, ready to begin polypeptide synthesis. In eukaryotes this happens through the scanning mechanism: the small ribosomal subunit (40S), pre-loaded with initiator methionyl-tRNA (Met-tRNAi) and initiation factors, binds the 5' 7-methylguanosine cap of the mRNA and then migrates in the 3' direction, testing successive triplets against the anticodon of Met-tRNAi until it finds an AUG codon — usually the first one it encounters — embedded in a favorable sequence context. This contrasts with bacterial initiation (see Translation Initiation in Bacteria), where the 30S subunit is recruited internally, near the start codon, by direct RNA–RNA base-pairing between the 16S rRNA and the Shine-Dalgarno sequence. The eukaryotic pathway is built from a defined series of macromolecular assemblies, each with a name and a measurable sedimentation coefficient (historically defined by sucrose gradient centrifugation, hence "43S," "48S," "80S"). The 43S preinitiation complex consists of the 40S ribosomal subunit, the eIF2 ternary complex (eIF2–GTP–Met-tRNAi), and several accessory factors (eIF1, eIF1A, eIF3, and eIF5). This complex is recruited to the mRNA's 5' cap via the eIF4F complex — a three-subunit assembly comprising cap-binding eIF4E, the scaffold protein eIF4G, and the DEAD-box RNA helicase eIF4A — to form the 48S initiation complex, which then scans. Upon start-codon recognition, GTP hydrolysis and factor release permit joining of the large (60S) ribosomal subunit, producing the elongation-competent 80S initiation complex. Two features distinguish this pathway conceptually from bacterial initiation. First, cap-dependence and mRNA circularization: eIF4G bridges the cap-binding complex at the 5' end to poly(A)-binding protein (PABP) at the 3' poly(A) tail, functionally circularizing the mRNA and coupling translational efficiency to both mRNA ends simultaneously. Second, context-dependent start-site selection: because scanning does not rely on sequence complementarity to a fixed rRNA sequence, the ribosome must use local nucleotide context — the Kozak consensus — together with intrinsic proofreading by eIF1 and eIF5 to discriminate a true start codon from spurious upstream AUGs.

Step 1 — Formation of the ternary complex and the 43S preinitiation complex. Eukaryotic initiation factor 2 (eIF2), a heterotrimeric GTPase, binds GTP and initiator methionyl-tRNA (Met-tRNAi, charged by methionyl-tRNA synthetase and distinct from elongator tRNA-Met) to form the ternary complex. This ternary complex docks onto the 40S ribosomal subunit together with eIF1, eIF1A, eIF3, and eIF5, assembling the 43S preinitiation complex. eIF1 and eIF1A bind near the ribosomal P site and A site respectively, holding the 40S subunit in an "open," scanning-competent conformation. eIF3, a large multi-subunit factor (~13 subunits in mammals), binds the 40S solvent-exposed surface and serves as a hub linking the 43S complex to the cap-binding apparatus on the mRNA.

Step 2 — Cap recognition and mRNA activation via eIF4F. Independently, the mRNA is prepared for ribosome loading. eIF4E binds the m7G cap directly, with high specificity for the methylated guanosine. eIF4G acts as a scaffold, binding eIF4E, eIF4A, eIF3, and PABP simultaneously; this last interaction loops the mRNA into a closed, circular conformation, since PABP sits at the 3' poly(A) tail. eIF4A, a DEAD-box ATPase, unwinds secondary structure in the 5' UTR — often assisted by eIF4B — using cycles of ATP-dependent conformational change rather than processive unwinding. The assembled eIF4F–PABP–mRNA complex presents an accessible, unstructured 5' end to which the 43S complex can attach.

Step 3 — Assembly of the 48S complex and attachment. The 43S preinitiation complex is recruited to the activated mRNA through eIF3–eIF4G contacts, forming the 48S initiation complex positioned at or near the 5' cap.

Step 4 — Scanning. Powered by eIF4A/eIF4B helicase activity and using ATP hydrolysis (not GTP), the 48S complex translocates 5'→3' along the 5' UTR, unwinding secondary structure as it advances. eIF1 plays a critical fidelity role here: it prevents premature, stable engagement of Met-tRNAi with non-AUG or poorly contexted triplets by keeping the 40S P site in an open conformation until a bona fide start codon is found. The scanning rate has been estimated at roughly several nucleotides per second in reconstituted systems, and structured 5' UTRs (long, GC-rich, or containing stable hairpins) slow or block scanning, which is one reason such UTRs are associated with translational repression of the downstream ORF.

Step 5 — Start codon recognition and Kozak context. When the scanning complex encounters an AUG, base-pairing forms between the anticodon of Met-tRNAi and the codon in the ribosomal P site. Recognition triggers a conformational change: eIF1 is displaced from its inhibitory position, the 40S subunit closes around the codon-anticodon duplex, and this closure is markedly favored when the AUG sits within the Kozak consensus sequence: ``` 5'-gccRccATGG-3' ``` where R is a purine (A or G) at position −3 (counting the A of AUG as +1), and a G is strongly preferred at position +4. The −3 purine and +4 G are the two most influential positions; a strong context with both is called "adequate" or "optimal," while a weak context (pyrimidine at −3, non-G at +4) increases the chance that the ribosome will bypass that AUG and continue scanning — a phenomenon called leaky scanning, which allows a single mRNA to produce two protein isoforms from alternative start codons.

Step 6 — GTP hydrolysis by eIF2 and Pi release. Correct codon-anticodon pairing stimulates eIF5, a GTPase-activating protein (GAP), to trigger GTP hydrolysis on eIF2. This step is essentially irreversible under normal conditions and commits the ribosome to that AUG. Inorganic phosphate release causes eIF2–GDP to lose affinity for Met-tRNAi and dissociate from the complex, along with eIF1, eIF3, and eIF5.

Step 7 — 60S joining and 80S formation. eIF5B, a ribosome-dependent GTPase structurally related to bacterial IF2, catalyzes joining of the 60S subunit to the 40S–mRNA–Met-tRNAi complex. GTP hydrolysis by eIF5B, coupled to subunit joining, triggers release of eIF5B and eIF1A. The result is the 80S initiation complex: Met-tRNAi base-paired to the AUG start codon in the ribosomal P site, with the A site empty and ready to accept the first aminoacyl-tRNA of elongation.

Step 8 — Recycling of eIF2. eIF2–GDP must be converted back to eIF2–GTP before it can form a new ternary complex. This exchange is catalyzed by eIF2B, a guanine nucleotide exchange factor (GEF) that is itself the target of regulatory phosphorylation (see below), making eIF2B activity a rate-limiting node for global translation rates.

The special case of leaky scanning and reinitiation. Because start-codon selection is probabilistic rather than absolute, upstream open reading frames (uORFs) in the 5' UTR can dramatically affect downstream translation. A ribosome that initiates at a uORF, translates it, and terminates can sometimes resume scanning and reinitiate at a downstream AUG — but only if enough distance (and time to reacquire a new ternary complex) separates the two ORFs. This mechanism underlies the ATF4 translational control circuit central to the integrated stress response, discussed below.

The special case of cap-independent initiation. Certain viral and cellular mRNAs bypass cap-dependent scanning entirely by folding into an internal ribosome entry site (IRES), a structured RNA element that recruits the 40S subunit (with varying subsets of eIFs, sometimes none at all in the case of the intergenic IRES of dicistroviruses) directly to an internal position near the start codon. This was discovered through poliovirus and encephalomyocarditis virus (EMCV) studies (see Evidence).

Toeprinting locates the ribosome on the mRNA. A key technique for mapping where ribosomal complexes sit on mRNA is primer extension inhibition ("toeprinting"): a radiolabeled or fluorescent DNA primer is annealed downstream of the region of interest, reverse transcriptase extends it, and the ribosome (or any bound complex) physically blocks the enzyme, producing a stop ("toeprint") at a characteristic position roughly 15–17 nucleotides downstream of the P-site codon. Using purified components — 40S subunits, eIF2, eIF3, eIF1, eIF1A, eIF4F, Met-tRNAi, and mRNA — researchers (notably work from the Hellen, Pestova, and Sonenberg laboratories through the 1990s–2000s) reconstituted 48S complex formation in vitro and showed by toeprinting that the complex initially appears at the cap and then moves downstream, coming to rest precisely at the AUG codon, exactly as scanning predicts. Omitting eIF4A or ATP stalled complexes near the 5' end, directly showing that helicase-driven, ATP-dependent movement is required for scanning through structured UTRs.

Marilyn Kozak's mutagenesis defined the consensus. Through the late 1970s and 1980s, Marilyn Kozak systematically mutated the nucleotides flanking AUG codons in cloned genes (notably preproinsulin and other model constructs) expressed in cultured cells and cell-free translation systems, then measured translation efficiency by in vitro translation and by ribosome "toeprinting"-style mapping of initiation sites. She found that changing the base at position −3 from a purine to a pyrimidine sharply reduced initiation at that AUG and increased use of downstream AUGs (leaky scanning); she found a parallel, if smaller, effect for the +4 position. This established the Kozak consensus (gccRccATGG) as a quantitative predictor of translation start-site strength — not an absolute requirement, but a graded determinant of efficiency, later confirmed across thousands of genes by ribosome profiling.

Cap-binding assays identified eIF4E and proved cap-dependence. Using chemically synthesized cap analogs (m7GTP) as competitive inhibitors, and cap-affinity chromatography with immobilized m7GTP, Sonenberg and Shatkin (1978) and later work identified a 24-kDa polypeptide — eIF4E — as the specific cap-binding protein in reticulocyte lysate. Adding free m7GTP to an in vitro translation reaction selectively inhibited translation of capped mRNAs while leaving uncapped or IRES-driven mRNAs (once these were characterized) relatively unaffected, directly demonstrating that eIF4E-cap interaction is necessary for standard initiation. Cross-linking and mutagenesis subsequently mapped the eIF4E–eIF4G–eIF4A interactions that constitute the eIF4F complex.

Poliovirus cleavage of eIF4G revealed cap-independent (IRES) initiation. Poliovirus infection shuts off host cap-dependent translation while continuing to translate its own uncapped genomic RNA efficiently — a paradox. Etchison, Milburn, and Ehrenfeld (1982) showed that a viral protease cleaves the host factor eIF4G (then called p220), separating the eIF4E-binding N-terminal fragment from the eIF4A/eIF3-binding C-terminal fragment. This disables cap-dependent scanning on host mRNAs but leaves the viral RNA translatable via a distinct route. Pelletier and Sonenberg (1988), followed independently by Jang, Kräusslich, Nicklin, Duke, Palmenberg, and Wimmer for EMCV (1988), demonstrated using bicistronic reporter constructs that a specific structured RNA element in the poliovirus/EMCV 5' UTR could drive translation of a downstream cistron without a cap and without scanning through the intervening sequence — the discovery of the internal ribosome entry site (IRES). This proved that cap recognition and 5'-end scanning, while the dominant mechanism, are not the only route to ribosome recruitment, and it reframed the entire initiation pathway as a modular system that viruses (and, later, some cellular stress-response mRNAs) can bypass.

Reconstitution with purified factors established minimal sufficiency. Pestova, Hellen and colleagues (1998–2000s) reconstituted the entire pathway — 48S and 80S complex formation — from individually purified, recombinant human/rabbit initiation factors, ribosomal subunits, Met-tRNAi, and defined mRNAs. By omitting individual factors one at a time and monitoring 48S/80S formation (via toeprinting, native gel shift, and sucrose gradient sedimentation), they defined the minimal set of factors required at each step: eIF1A and eIF3 for 40S binding to mRNA; eIF4F, eIF4B, and ATP for scanning through structure; eIF1 for discriminating correct start codons and rejecting near-cognate or poorly contexted AUGs; and eIF5/eIF5B GTPase activity for the final commitment and 60S joining steps. This factor-by-factor reconstitution is the eukaryotic analog of the classical bacterial IF1/IF2/IF3 reconstitution experiments and is considered the definitive biochemical proof of the stepwise scanning model.

Example 1 — Predicting the consequence of a uORF mutation on ATF4 translation. The transcription factor ATF4 is translationally controlled by two uORFs in its 5' UTR: uORF1 (short, always efficiently translated) and uORF2 (which overlaps the ATF4 main ORF out of frame, so translating it precludes translating ATF4). Under normal conditions, ternary complex (eIF2–GTP–Met-tRNAi) is abundant. A ribosome translates uORF1, terminates, resumes scanning quickly, reacquires a ternary complex before reaching uORF2, and translates uORF2 instead of ATF4 — repressing ATF4 protein output. Under stress (e.g., amino acid starvation activating the kinase GCN2, or ER stress activating PERK), eIF2α is phosphorylated, eIF2B-mediated GDP/GTP exchange is inhibited, and ternary complex becomes scarce. Now, after translating uORF1, the scanning ribosome takes longer to reacquire a ternary complex — long enough to scan past uORF2's start codon without reinitiating there — and instead reinitiates at the downstream ATF4 start codon. Net result: lower global ternary-complex availability paradoxically increases translation of a specific mRNA (ATF4), because the regulatory logic depends on the spacing and number of uORFs, not simply on total initiation activity. This is the molecular basis of the integrated stress response (ISR), and it explains why global translational shutdown (via eIF2α phosphorylation) is compatible with, indeed causally linked to, upregulation of specific stress-response transcription factors.

Example 2 — Quantifying Kozak-context effects on isoform ratios. Consider an mRNA with two in-frame AUG codons, 45 nucleotides apart: AUG1 in a weak context (pyrimidine at −3, non-G at +4) and AUG2 (the next in-frame AUG) in an optimal context (purine at −3, G at +4). Suppose empirical data (as in Kozak's constructs) show that a weak-context AUG is recognized by the scanning complex only ~20% of the time, with the remaining ~80% of ribosomes leaky-scanning past it to the next AUG. If 100 ribosomes initiate scanning on this mRNA:

• ~20 initiate at AUG1, producing the long isoform (extra 15 amino acids at the N-terminus, since 45 nt = 15 codons).
• ~80 continue scanning and initiate at AUG2 (in strong context, so essentially all of these ~80 initiate there, since it is now the next candidate and strongly favored).

This yields a 20:80 (1:4) ratio of long-isoform to short-isoform protein from a single mRNA species — exactly the kind of leaky-scanning-generated isoform diversity seen for genes like c-myc (which produces both a full-length and an N-terminally truncated product from alternative AUGs) as well as many mitochondrial-targeting/cytosolic protein pairs where the longer isoform's extra N-terminal residues encode an organelle-targeting signal. Note the asymmetry: the effect isn't symmetric competition between two equally strong sites, but a strongly context-dependent probabilistic bypass — small changes in the −3/+4 nucleotides can shift this ratio dramatically, which is precisely why Kozak's mutagenesis data (changing a single base) produced large, measurable shifts in isoform output.

Marilyn Kozak (1936–2022) was a molecular biologist whose painstaking, largely single-author mutagenesis studies through the 1970s–1990s defined how the eukaryotic ribosome selects its start codon. Working initially at the University of Pittsburgh, she used cell-free translation systems and, later, transfected mammalian cells expressing systematically mutated 5' UTR constructs to measure how nucleotide identity around the AUG codon affected translational efficiency and start-site choice. Her work was notable for its rigor and its willingness to challenge oversimplified models — she was also a persistent critic of loosely supported claims about "leaky scanning" and IRES elements in the literature, insisting on quantitative, carefully controlled evidence, which occasionally put her at odds with parts of the field but sharpened the standards of proof used to establish the modern scanning model.

Cancer biology and therapeutics. eIF4E is a proto-oncogene: its overexpression preferentially boosts translation of mRNAs with structured 5' UTRs encoding growth-promoting proteins (cyclin D1, MYC, VEGF), and eIF4E overexpression alone can transform cultured cells and cause tumors in mouse models. This has motivated development of eIF4E inhibitors and mTOR-pathway drugs (mTOR phosphorylates 4E-BP1, a repressor that normally sequesters eIF4E away from eIF4G) as anticancer strategies, since hyperactive mTOR signaling in many cancers drives excess cap-dependent initiation.

Antiviral and vaccine biotechnology. Because many viruses manipulate eIF4G cleavage, IRES-mediated translation, or eIF2α phosphorylation (via PKR) to redirect the translation machinery toward viral mRNAs, understanding this pathway is central to antiviral drug design. IRES elements themselves are widely used as biotechnology tools, allowing bicistronic vectors to express two proteins from a single mRNA (common in CAR-T cell engineering and some viral vector vaccine designs).

Metabolic and neurological disease. Mutations in eIF2B subunits cause vanishing white matter disease, a fatal leukoencephalopathy, by impairing eIF2B's GEF activity and dysregulating the integrated stress response in glial cells — a direct clinical illustration of how central eIF2/eIF2B function is to normal physiology, particularly in the CNS.

Agricultural and RNA vaccine engineering. Understanding 5' UTR structure, cap chemistry, and Kozak context has been essential to engineering synthetic mRNA constructs — including mRNA vaccine platforms — for maximal translational efficiency, informing choices of cap analogs, UTR sequences, and start-codon context in therapeutic mRNA design.

Stress-response pharmacology. Small-molecule modulators of the integrated stress response (targeting PERK, GCN2, or the eIF2B complex) are in development for neurodegenerative disease, based directly on the ATF4/uORF translational control logic described above. This assembled 80S ribosome, sitting with Met-tRNAi in the P site and an empty A site, is the direct substrate for the next stage of protein synthesis — the topic taken up in Elongation and Codon Recognition, where aminoacyl-tRNAs delivered by eEF1A are tested against each successive codon and peptide bonds are formed processively down the reading frame established during initiation.

Ask the tutor: How does eIF1's role in rejecting incorrect start codons mechanistically differ from eIF5's role in triggering GTP hydrolysis, and why does the cell need both as separate checkpoints?