Initiation of protein biosynthesis in eukaryotes


  • William C. Merrick

    Corresponding author
    1. Department of Biochemistry, School of Medicine, Case Western Reserve University, Cleveland, Ohio 44106-4935
    • Department of Biochemistry, School of Medicine, Case Western Reserve University, Cleveland, Ohio 44106-4935. Tel.: 216-368-3578; Fax: 216-368-3419
    Search for more papers by this author


Initiation of protein synthesis in eukaryotes is important both mechanically, because it selects the translation reading frame, and biologically, because it is the primary site for regulation of translation. For approximately 95–98% of the cellular mRNAs, formation of an initiation complex follows a rather specific pathway that is broken down into seven discrete steps. Although almost 35 peptides in 12–14 translation initiation factors participate in this process, three appear to have dominant roles: eukaryotic initiation factor (eIF)3, in building a pool of 40S subunits; eIF2, in binding the initiator tRNA (tRNAi) to the 40S subunit; and eIF4F in activating the mRNA and binding it to the 40S subunit. This process requires both ATP and GTP. The resulting 80S initiation complex contains both the tRNAi and the mRNA, with the anticodon of the tRNAi correctly base paired with the initiating AUG code word. Regulation of translation focuses mostly on controlling the activity of either eIF2 or eIF4F and this regulation has different consequences. Reduction in eIF2 activity influences all mRNA approximately the same, whereas reduction in eIF4F activity drives competition between mRNAs.

The intent of this article is to present the basics of eukaryotic protein biosynthesis as it relates to the translation of most of the cellular mRNA. As such, this review will only explore in cursory detail the more rare but very interesting translation that occurs by alternate mechanisms (re-initiation or internal initiation). This level of presentation can readily be taught to undergraduates and, thus, could also be modified to accommodate either first year graduate or medical students.


The overall process of initiation can be described as follows:

  • Preparation of a pool of small ribosomal subunits on which to build the initiation complex;

  • Binding of the initiator tRNA (Met-tRNAi)11 to the 40S subunit;

  • Activation of the mRNA;

  • Binding of the mRNA to the 40S subunit;

  • Scanning of the mRNA to locate the initiating AUG code word;

  • Subunit joining to form the 80S initiation complex;

  • Recycling of eIF2·GDP to eIF2·GTP.

Each of these steps is described in more detail in the following paragraphs.

Under normal physiological conditions, the two subunits of the ribosome (generally referred to as the large and small subunits, or by their physical size as measured by sedimentation, 60S and 40S, respectively) tend to remain associated as an inactive ribosome (80S), although the equilibrium allows a small portion of the subunits to exist free (Fig. 1). The building of a pool of small ribosomal subunits is achieved by the binding of eukaryotic initiation factor (eIF)3, assisted by the binding of eIF1A. These binding events shift the equilibrium position to the right, in large measure because once eIF3 has bound to the 40S subunit, the 60S subunit can not bind.

The second step is the binding of the tRNAi (Fig. 2). In this step, the tRNAi is bound to the 40S subunit as a ternary complex (eIF2·GTP·Met-tRNAi). It is worth mentioning that there are two Met-tRNA species: Met-tRNAi and Met-tRNAm. The initiation process is specific for Met-tRNAi and the elongation process is specific for the Met-tRNAm. Because both Met-tRNAs respond to the code word AUG, this ensures that there will be independent pools of the tRNAs for both processes. Second, for both initiation and elongation, the aminoacyl-tRNAs are brought to the ribosome as ternary complexes (factor·GTP·aminoacyl-tRNA). The binding of the ternary complex is accomplished due to specific binding sites in eIF2 for both the 40S subunit and for eIF3, which is already on the ribosome.

Step 3 is the activation of the mRNA (Fig. 3). In most cellular systems, mRNAs exist as complexes of RNA and protein (mRNP) (ribonuclear protein particles). The process of activation of the mRNA appears to require the removal of proteins from the 5′ end of the mRNA and the removal of any secondary structure. This is accomplished by the binding of eIF4F to the 5′ m7G-cap of the mRNA (Fig. 3). eIF4F is composed of three subunits (eIF4E, eIF4A, and eIF4G). The small subunit, eIF4E, specifically recognizes the m7G cap with very high affinity (Kd of approximately 10−8 to 10−9M). This initial binding positions eIF4F at the 5′ terminus of the mRNA and, thus, orients the activity of its resident helicase, eIF4A, to use the energy in ATP to drive the unwinding of RNA secondary structure and to force the release of proteins from the 5′ end of the mRNA (Fig. 4) [1].

In the next step, the association of eIF4F with the mRNA directs mRNA binding to the 40S subunit through the interaction of eIF4G with 40S-bound eIF3 (Fig. 5). This results in the binding of the 5′ end of the mRNA to the 40S subunit, but at this point, there has not been the correct match of the anticodon of the Met-tRNAi with the initiating AUG codon. The correct matching is achieved by an ATP-dependent process termed scanning [2]. Although the details are vague here, it appears that the 40S subunit moves in a 3′ direction, testing each possible codon for a match to the anticodon of the tRNAi. Genetic studies in yeast have confirmed that it is the tRNAi anticodon that recognizes the initiating AUG, not one of the translation factors (although translation factors can influence the accuracy of this process) [3–6]. This recognition of the AUG codon explains why the tRNAi must bind before the mRNA. And, perhaps not as obviously, this also explains why the location of the initiating AUG is simply described as the AUG codon nearest the 5′ end of the mRNA (or the first AUG encountered).

Having established a correct match of the tRNAi with the initiating AUG codon, the major remaining event is to remove the translation factors from the surface of the 40S subunit. This is achieved through the actions of eIF5 and eIF5B, which are associated with two GTP hydrolysis events: the hydrolysis of the GTP in the ternary complex, and a second hydrolysis event by eIF5B (Fig. 6) [7–10]. These two events are sufficient to release the other translation initiation factors from the 40S subunit and allow the joining of the 60S subunit. This yields an 80S ribosome capable of beginning the many elongation steps in protein biosynthesis.

Although this would appear to complete the initiation pathway, there is still one final step. The hydrolysis of the GTP in the ternary complex leads to the release of eIF2·GDP from the 40S subunit. eIF2 is much like the “classical” G proteins in that it binds the product GDP approximately 100 times more tightly than the substrate GTP (relative Kd values of 10−8 and 10−6M, respectively). To achieve the release of GDP and the rebinding of GTP, the nucleotide recycling protein eIF2B catalyzes the exchange of eIF2 bound GDP for GTP.


There are two major points of regulation in eukaryotic translation: regulation through the modulation of either eIF2 and/or eIF4F activity. The simplest is the regulation of eIF2 activity. There are four known protein kinases that phosphorylate eIF2: a heme-regulated inhibitor (HIR), which responds to heme deficiency; GCN2 (from the name of the yeast gene first identified in this regulation), which responds to amino acid starvation; PKR, which responds to the generic “calling card” of viral infection, dsRNA; and pancreatic endoplasmic reticulum kinase (PERK), which responds to stress in the endoplasmic reticulum (although this kinase is essentially in all tissues). With appropriate stimulation, these kinases phosphorylate eIF2. When this phosphorylated eIF2 exits the initiation pathway, it binds more tightly than eIF2·GDP to eIF2B, but nucleotide exchange does not occur. Thus, the limiting amounts of eIF2B end up as an inactive pool of eIF2-PO4·GDP·eIF2B. In the absence of the recycling activity, eIF2·GDP builds up, and the lack of ternary complexes leads to a decrease in 40S·ternary complexes, which slows the overall process of initiation.

The reduction in the level of 40S·ternary complexes reduces translation of essentially all mRNAs equally. That is, if the level of 40S·ternary complexes is reduced by 50%, the level of translation of each mRNA is reduced by 50%. Thus, the level of total amino acid incorporation is also reduced 50%. This direct correlation also points out that, overall, initiation is the rate-limiting step in protein biosynthesis (in keeping with the general biological paradigm that it is the first committed step in a pathway in which most of the regulation occurs).

In contrast, the regulation of eIF4F activity has a different consequence. In general, the major pathway for eIF4F regulation is the availability of the eIF4E subunit, which has the potential to bind to either eIF4G (and with eIF4A to form active eIF4F) or 4E-BP, a protein that binds to eIF4E and thereby blocks its association with eIF4G (Fig. 7). The ability of 4E-BP to bind eIF4E is regulated by its state of phosphorylation, and in general, increased phosphorylation of 4E-BP inhibits its ability to bind to eIF4E. There appear to be several pathways that lead to the inactivation of 4E-BP via phosphorylation, which include the target of rapamycin (TOR), which is a drug that specifically blocks the action of certain kinases, and S6 (named after the ribosomal protein S6) kinase pathways. In general, conditions that favor growth (elevated insulin, specific growth factors, etc.) favor the activation of these two kinase cascades, whereas conditions that are stressful (heat shock, apoptosis, nutritional deprivation, etc.) down-regulate these pathways and, thus, down-regulate protein synthesis [11–13].

An immediate assumption might be that such regulation would lead to the same reduction in protein synthesis as seen with eIF2 phosphorylation, an across-the-board reduction in the translation of all mRNAs. This is not the case. To the translational machinery (mostly eIF4F), not all mRNAs are created equal. Although there are many elements that will influence how efficiently an mRNA will be translated, the most dominant is accessibility of the m7G cap structure in an mRNA [14]. Although this has been tested experimentally, it is difficult to predict given the sequence of the mRNA because current RNA folding programs are not very good at predicting weak, tertiary interactions (i.e. pseudoknots) and are poor predictors of three-dimensional structure. There is also the added complication, noted above, that most mRNAs exist as mRNPs that are 50% protein and 50% RNA, and the protein composition of mRNPs is not known.

As a consequence of RNA structure and protein binding, mRNAs have different affinities for the translational machinery, probably varying by at least 20-fold. Under conditions optimal for growth and maximal eIF4F activity, mRNAs are generally translated in accordance with their abundance. However, when eIF4F becomes limiting, mRNAs must compete for the few eIF4F molecules available, and as might be anticipated, the most efficiently translated mRNAs out-compete the poorly translated mRNAs [15–17]. A simple example of this is given below for the efficient mRNA A, which has an apparent Kd for the translational machinery of 10−8M and a poor mRNA B, whose apparent Kd is 10−7M.

As can be seen from the numbers in Table II, even though the number of A and B mRNA molecules remains constant, the ratio of synthesis of the resulting proteins shifts from 1:1 (100 eIF4F molecules) to 10:1 (50 or fewer eIF4F molecules). Thus, although a 50% reduction in eIF4F activity leads to the predicted 50% reduction in the incorporation of amino acids, the reduced eIF4F activity drives competition for the most efficient mRNAs, and, as noted in the above example, the most immediate loss of translation is observed for the poor mRNA.


The two presentations above on the pathway of 80S initiation complex formation and regulation of this pathway are complete enough to obtain a general idea of the interplay of initiation factors with mRNA, Met-tRNAi and the 40S subunit, and the overall regulation of translation via regulation of either ternary complex levels or eIF4F activity. In real, quantitative terms, there are inadequacies in this pathway that require a more sophisticated presentation (i.e. the ratio of translation factors associated with ribosomes varies almost 100-fold, with eIF4A being the most abundant initiation factor and eIF5B being among the least abundant). However, this type of pathway is useful in orienting the reader to the temporal order of utilization of the mRNA, Met-tRNAi and initiation factors.

The above description has focused on the major players: eIF3 for providing the pool of 40S subunits, eIF2 for binding the tRNAi; and eIF4F as activating and binding the mRNA to 40S subunits. For the terminally curious, a complete list of the translation factors is shown in Table I, and the initiation pathway using these translation factors is shown in Figure 8. The key feature of the initiation factors not mentioned in the first two parts of this article is that they enhance the kinetic rate of initiation complex formation, and they enhance the accuracy of this process (correct identification of the initiating AUG). When tested in the yeast system, the absence of almost any one of the factors (or a protein subunit in an initiation factor composed of multiple subunits) was incompatible with growth. Thus, kinetic rate and accuracy are essential for living organisms.

Finally, there are two rare initiation schemes that appear to be crucially important but for which we currently lack the kind of detail that is available for the cap-dependent initiation pathway. The first is re-initiation. In this scheme (shown diagrammatically in Fig. 9), there is a normal initiation process that initiates at an AUG and then after polymerizing 3–50 amino acids, goes through the normal termination pathway. Following termination and the loss of the 60S subunit, the 40S scans in a 3′ direction looking for the next AUG start codon. In the process of this scanning, a new ternary complex is acquired to allow the accurate identification of the second AUG. Following the match of this AUG with the anticodon of the tRNAi, subunit joining occurs, likely in a manner very similar to that seen in the cap-dependent pathway described in the beginning of this article. In general, this pathway is inefficient, with at best only 25% of the 40S subunits that terminate going on to initiate at the second AUG. In most instances, removal of the upstream AUG would lead to at least a 10-fold increase in production of the protein in question. Therefore, the purpose of re-initiation is not to provide an efficient process for protein expression but, rather, a regulated process for protein expression.

The second rare initiation scheme is termed internal initiation (Fig. 10). In this scheme, the 40S subunit binds directly to an internal portion of the mRNA at an internal ribosome entry site (IRES). In most cellular mRNAs, this process is not very efficient. However, the down-regulation of eIF4F activity and the commensurate down-regulation of cap-dependent translation make initiation factors available for internal initiation and, thus, although total protein synthesis is decreased, IRES-mediated translation is increased. Many of the IRES-containing mRNAs encode proteins that facilitate a cell's recovery from stress. As noted above, eIF4F activity is generally down-regulated under conditions of stress. In this manner, by the regulation of eIF4F activity, the cell has the capability to immediately respond to stressing agents without the need for transcription.

At present, the exact details of the mechanism for re-initiation and internal initiation appear to vary, depending on the mRNA in question. In some instances of re-initiation, the position of the small upstream open reading frame (ORF) is important, and in some instances the amino acid sequence of the ORF is important [18, 19]. In the case of internal initiation, for the few examples studied to date, the initiation factors required for optimal in vitro 80S complex formation vary from a requirement for all of the normal factors to the requirement for none of the initiation factors [20]. This variety for both of these rare schemes make it likely that we will only know the mechanistic details for initiation of these mRNAs in the years to come.


From the above presentation, one might infer that 3–5% of the cellular mRNAs are regulated at the level of translation, rather than at the level of their synthesis. In fact, translational regulation is involved (by mass) in the synthesis of approximately 30% of the normal cellular proteins. This additional regulation comes from cis-acting sequences in either the 5′ or 3′ end of the mRNA (for which there is, of course, a trans-acting factor). Thus, the regulated expression of proteins by transcription accounts for approximately two thirds of the cellular proteins, and translation accounts for approximately one third. This increase from the predominant transcriptional regulation in bacterial systems likely reflects the compartmentalization of mRNA synthesis and translation in the nucleus and cytoplasm, respectively, in contrast to both events occurring in a coupled fashion in the cytoplasm of the bacteria. Second, many viral infections lead to an alteration of the translational machinery to favor the translation of the viral mRNAs over the translation of cellular mRNAs. (This appears to be especially true for lytic viruses that in essence are trying to replicate and then lyse the cell in 6–24 h to release progeny.) As a consequence, the “translational status” of the cell may very well be taken as the thermometer for the health of the cell. This often under-appreciated fact has left open many doors to revisit translation as it relates to both general molecular mechanisms and a wide variety of clinical situations, especially viral infection. Hopefully this short review has provided the reader with a general understanding of the initiation and regulation of translation so that he/she may be prepared to incorporate some of these findings into their own efforts.

Figure FIGURE 1..

Building the 40S subunit pool.

Figure FIGURE 2..

Binding the ternary complex (eIF2·GTP·Met-tRNAi) to the 40S subunit. Note in the figure that the formation of the ternary complex is ordered, GTP is bound before the tRNAi. Second, the presence of eIF3 on the 40S subunit stabilizes the binding of the ternary complex.

Figure FIGURE 3..

Structure of the m7G cap. Shown in the figure is the chemical structure for the m7G cap with an adenosine residue as the normal 5′ terminal nucleotide in the mRNA.

Figure FIGURE 4..

Activation of an mRNA A. Depiction of a normal mRNA coated with protein (small circles) is shown. The eIF4E subunit recognizes specifically the m7G cap structure whereas the eIF4A subunit unwinds RNA secondary structure (or removes proteins) in an ATP-dependent manner.

Figure FIGURE 5..

Binding and “scanning” of the mRNA.

Figure FIGURE 6..

Subunit joining. Note that this step requires the utilization of two GTP molecules, one in the ternary complex and one via the action of eIF5B. It is possible that the hydrolysis of the second GTP by eIF5B is concomitant with the joining of the 60S subunit.

Figure FIGURE 7..

Regulation of eIF4F activity via the sequestration of eIF4E by 4E-BP. The protein kinases that regulate the phosphorylation state of 4E-BP emanate from the mTOR and S6 kinase pathways. In general, these pathways are activated by conditions that favor growth and are inactivated by stress conditions.

Figure FIGURE 8..

The “real” 80S pathway. This initiation pathway demonstrates the full utilization of the many initiation factors (for greater detail on the description of this pathway, please see one of the books or reviews cited at the end of this article).

Figure FIGURE 9..

Re-initiation. In this sequence, the translation initiation of the first ORF (AUG-[XYZ]3–50-UAA) occurs as described in the text and as shown in total in Fig. 8. Following termination, the 40S subunit scans to the next AUG, acquiring a ternary complex in the process. Subsequent subunit joining allows translation of the downstream or second open reading frame.

Figure FIGURE 10..

Internal initiation. In this scheme, the 40S subunit (with the ternary complex, eIF3 and eIF1A) binds directly to an internal portion of the mRNA described as an IRES.

Table Table I. Eukaryotic initiation factors
NameMolecular weight × 10−3Number of SubunitsFunction(s)
  • a

    a 40S pool, facilitates the generation of free 40S subunits for the initiation pathway.

  • b

    b Bridging function, binds to and helps orient other translation factors on the surface of the 40S subunit and, with the poly(A) binding protein, facilitates enhanced translation in polysomes.

  • c

    c functions in parentheses are suggested, but not proven.

  • d

    d GTPase activating proteins (GAPs) trigger the hydrolysis of GTP bound to other proteins, generally referred to as “G proteins,” which display a much greater binding affinity for the product GDP than the substrate GTP.

eIF1121AUG recognition
eIF1A16140S pool,a ternary complex binding to 40S, IF1 homolog
eIF21253GTP-dependent binding of Met-tRNAi
eIF2A651Represses IRES-mediated translation
eIF2B2705Guanine nucleotide exchange factor for eIF2
eIF36501140S pool, orients ternary and eIF4F·mRNA complexes on 40S subunit, general bridging functionb
eIF4A451ATP-dependent RNA helicase (scanning?)c
eIF4B1402Stimulates eIF4A and eIF4F activities
eIF4F2403m7G recognition, RNA helicase, 3′ to 5′ bridging functionb
eIF4H281Stimulates eIF4A and eIF4F activities
eIF5501GTPase activating proteind for the ternary complex (eIF2·GTP·Met-tRNAi)
eIF5A181Stimulates subunit joining, facilitates conformation change of 80S
eIF5B1391Ribosome-dependent GTPase, subunit joining, IF2 homolog
eIF6251(Binds 60S to help 40S pool formation?)c
Table Table II. Ratio of protein synthesis
Number of active eIF4F molecules50 mRNA A relative amount of protein made50 mRNA B relative amount of protein made


The author would like to thank Dr. Anton A. Komar for his preparation of the figures used in this article and Drs. Anton A. Komar and C. Raman Bhasker for their critical reading and helpful suggestions on this article.


  1. 1

    The abbreviations used are: tRNAi, initiator tRNA; eIF, eukaryotic initiation factor; mRNP, messenger RNA·protein particle; 4E-BP, eIF4E binding protein; IRES, internal ribosome entry site; ORF, open reading frame.