Overview of the variation in prokaryotic and plastid translation initiation

Initiation is the rate-limiting step in eubacterial translation and an important point of regulation of gene expression (Laursen et al. 2005). In essence, it is a multi-step process involving the assembly of three translation initiation factors, IF1, IF2, and IF3, GTP, the 30S and 50S subunits of the ribosome, an mRNA, and an initiator N-formylmethionyl–tRNA. Most of our understanding of this process has come from studies of Escherichia coli. Yet, the analysis of the first sequenced cyanobacteria genome, Synechocystis sp. PCC6803 (Kaneko et al. 1996), made it clear that the protein subunit composition of cyanobacterial ribosomes differs from E. coli. For some cyanobacteria with sequenced genomes, the variation in ribosomal protein composition and stoichiometry compared to E. coli has been substantiated at the protein level (Davydov et al. 2013; Mutsuda and Sugiura 2006), providing an initial indication of the degree to which cyanobacterial ribosomes have diverged in form and operation from those of E. coli. The possible functional consequences of heterogeneity between ribosomal populations, between and even within the same organism, are being increasingly recognized and demonstrated (Filipovska and Rackham, 2013; Gilbert 2011; Komili et al. 2007; Schippers and Mueller-Roeber 2010; Sharma et al. 2007).

The mechanics of translation initiation are more varied than initially believed, particularly in the correct initial positioning of the 30S subunit on the mRNA, which can be facilitated by a Shine–Dalgarno (SD) sequence that is usually located 7 ± 2 nucleotides upstream of the initiation codon (Shine and Dalgarno 1974, 1975; Sugiura et al. 1998). Typically identified as the sequence 5′-GGAGG-3′, the SD sequence base pairs with the complementary, antiparallel sequence present at the 3′ end of the 16S rRNA within the 30S subunit. Although this interaction has been widely adopted as an invariant feature of the prokaryotic translation initiation process (Schmeing and Ramakrishnan 2009), prokaryotes actually produce several classes of mRNAs that are distinguished by their 5′ leaders: SD-containing mRNAs, non-SD-containing mRNAs, and mRNAs that lack a 5′ leader (Malys and McCarthy 2011). The percentage of SD-containing mRNAs varies widely between prokaryotic groups. Remarkably, only 39 % of cyanobacterial mRNAs examined contained a recognizable SD element (Nakagawa et al. 2010), which is lower than many other eubacteria and similar to the percentage of chloroplast mRNAs from land plants with correctly positioned SD sequences (Peled-Zehavi and Danon 2007; Sugiura et al. 1998). These findings support the view that although classical SD–16S rRNA interaction can be used during cyanobacterial and plastid translation initiation, the absence or incorrect positioning of SD sequences in the majority of their mRNAs means that these two groups must also use additional mechanisms to insure the proper location and frequency of translation initiation (Barkan 2011; Nakagawa et al. 2010; Sugiura et al. 1998). One additional mechanism for the proper initial localization of the 30S preinitiation complex involves ribosomal protein S1, which has been shown to play a role in translation initiation in cyanobacteria and other eubacteria by providing an RNA chaperone function that helps dock and unfold the mRNA, and to correctly position the initiation codon (Duval et al. 2013; Hajnsdorf and Boni 2012; Mutsuda and Sugiura 2006). Further evidence for the dispensability of SD sequences has come from biochemical and in vitro translation experiments, which have demonstrated that translation initiation can require sequences other than the SD for many plastid and cyanobacterial mRNAs, such as AU-rich regions near the translation start sites that may be the sites for RNA-binding proteins (reviewed in Sugiura et al. 1998; also see Hirose and Sugiura 1996, 2004; Mutsuda and Sugiura 2006; Plader and Sugiura 2003). The AU-rich regions can also be sites at which nucleases act to regulate mRNA stability (Horie et al. 2007).

Both chloroplasts and mitochondria use translation components that are closely related to those of eubacteria. For example, Chlamydomonas and land plant plastids contain some eubacterial ribosomal components like E. coli, but some components are absent or very diverged and some components are unique to plastid ribosomes (Beligni et al. 2004; Manuell et al. 2007; Subramanian 1993; Tiller et al. 2012; Yamaguchi et al. 2002, 2003; Yamaguchi and Subramanian 2003). Because of the similarities between plastid and eubacteria ribosomes, it is plausible that similar mechanisms of translation regulation exist in plastids, including chloroplasts, as in eubacteria.

For the regulation of translation in plastids, genetic screens have uncovered many nuclear-encoded RNA-binding proteins that, often together with elements primarily located within the 5′ leader of mRNAs, influence the extent of translation and stabilization of specific chloroplast transcripts (reviewed in Peled-Zehavi and Danon 2007). Unlike eubacteria, transcription and translation are typically uncoupled in chloroplasts and mRNAs are generally very stable (Marin-Navarro et al. 2007; Rochaix 1996). Consequently, posttranscriptional control has emerged as the primary form of regulation of plastid gene expression. Stem-loops at various locations in the 5′ leaders of chloroplast mRNAs have been identified as central structures in this control mechanism (Mayfield et al. 1994; Zerges and Rochaix, 1994). These stem-loops can be bound by one or several nuclear-encoded proteins, which typically activate translation (reviewed in Barkan 2011). The binding activities of such proteins are generally specific to one or a few chloroplast mRNA species in both the green alga Chlamydomonas and land plants (see Marin-Navarro et al. 2007; Peled-Zehavi and Danon 2007; Rochaix 1996). A major family of such proteins are the pentatricopeptide repeat (PPR) proteins, which bind 5′ stem-loops in mRNAs and influence translation as well as many other aspects of chloroplast mRNA metabolism by blocking exoribonuclease activity and remodeling the structure of the ribosome binding site, possibly functionally mimicking the action of small bacterial RNAs (Barkan 2011; Stern et al. 2010). This family has not been found in cyanobacteria or other eubacteria thus far.

Posttranscriptional regulation in cyanobacteria

The majority of posttranscriptional regulation described thus far in cyanobacteria has been at the level of mRNA stability and/or translation elongation, with the 5′ leader regions of the regulated genes often implicated as having a role in regulation. Regulation by antisense RNAs (e.g., Duhring et al. 2006), which appears to be very prevalent (Georg and Hess 2011), will not be addressed here. In Microcystis aeruginosa K-81, the 5′ leader of psbA2 confers posttranscriptional regulation, apparently by changing the stability of the transcript (Agrawal et al. 2001). Light regulation of hliA mRNA stability also occurs, perhaps due to ribosome stalling (Salem and van Waasbergen 2004), and iron-dependent stabilization of ferredoxin I mRNA was determined to act through the 5′ end of the transcript in Synechococcus PCC 7942 (Bovy et al. 1993). Cyanobacterial responsiveness to cold shock also involves RNA stability changes of the des genes (Sakamoto and Bryant 1997) and the cold-regulated expression of rbpA1 and rbp2, which encode RNA-binding proteins in many cyanobacteria (Hamano et al. 2004; Mutsuda et al. 1999). For rbpA1, this is controlled in part at the level of mRNA stability and involves the formation of a stem-loop structure in the 5′ leader region (Ehira et al. 2003, 2005; Sato and Nakamura 1998). In Synechococcus PCC 7942, posttranscriptional regulation of psbA involves both the 5′ leader and an element in the coding sequence (Kulkarni and Golden 1997; Kulkarni et al. 1992; Li and Golden 1993); this was found to be due to changes in mRNA stability and translation elongation rather than differential translation initiation (Mulo et al. 2012; Tyystjarvi et al. 2004). In Synechocystis PCC 6803, psbA is posttranscriptionally regulated at the level of translation elongation (Tyystjarvi et al. 2001) and stability (Horie et al. 2007), and the structure of the 5′ leader accounts for the differential use of two different translation start sites of the petH gene (Omairi-Nasser et al. 2011; Thomas et al. 2006). The regulation of Photosystem I by light in cyanobacteria appears to act primarily during translation or assembly of the complex, and at least part of psaA regulation also occurs at the level of translation (Aizawa and Fujita 1997; Fujita 1997; Herranen et al. 2005).

A recently identified form of posttranscriptional regulation in cyanobacteria that is not the result of differential mRNA stability is the light-color responsiveness of the cpeCDESTR operon (hereafter cpeC), which encodes proteins involved in biosynthesis of the photosynthetic light harvesting-antennae of Fremyella diplosiphon UTEX481 (also called Tolypothrix PCC 7601). Along with a transcriptional control system, cpeC regulation is mediated by the Cgi (control of green light induction) pathway, a posttranscriptional regulatory system that represses cpeC expression during growth in red light but not in green light (Kehoe and Gutu 2006). The Cgi system does not influence the stability of cpeC mRNA and operates through a region of the cpeC mRNA 5′ leader approximately 30 nucleotides 5′ of the start codon for CpeC that is predicted to form a stem-loop with a ΔG of −3.4 kcal mol−1 (Bezy et al. 2011). Nucleotide substitutions within either the stem or loop of this putative stem-loop structure abolish repression by the Cgi system. Collectively, these data suggest that cpeC mRNA levels may change in response to shifts in light color through modulation of transcription attenuation, with more attenuation occurring during growth in red light than green light (Bezy et al. 2011). Although it has not yet been shown in cyanobacteria, transcription attenuation is a relatively common form of gene regulation in prokaryotes and often controlled by the degree of coupling between the transcription and translation machinery (Gollnick and Babitzke 2002; Henkin and Yanofsky 2002). However, because chloroplast gene expression is mainly regulated at the translational level (Deng and Gruissem 1987) and mostly uncoupled from transcription, a plastid gene regulation system that operates in a manner directly parallel to the putative transcription attenuation system controlling cpeC seems unlikely.

Use of multiple IF3s in F. diplosiphon and perhaps other bacteria

A transposon mutagenesis screen for Cgi pathway components in F. diplosiphon that failed to light regulate cpeC mRNA levels uncovered multiple mutants with insertions in infCa, which encodes an IF3 named IF3α (Gutu et al. 2013). IF3s are essential in most bacteria.They are dumbbell-shaped proteins, consisting of C- and N-terminal domains (Cheng and Kaiser 1989; Springer et al. 1977). However, an F. diplosiphon infCa deletion mutant was viable due to the presence of infCb, which encodes a second IF3 called IF3β (Gutu et al. 2013). Each gene was capable of complementing an Escherichia coli infC mutant, demonstrating that both genes encoded bona fide IF3s (Gutu et al. 2013). Interestingly, deletion of infCb did not affect light regulation of cpeC mRNA, demonstrating that IF3α and IF3β have unique cellular roles even with their functional overlap as translation initiation factors (Gutu et al. 2013). Although the selective use of multiple IF3s has not been previously found, there have been reports of regulatory roles for other IF3s in addition to translation initiation and ribosome recycling. Cold shock causes the single E. coli IF3 to increase in abundance relative to ribosome abundance, causing the preferential translation of cold-shock mRNAs (Giuliodori et al. 2004). At low temperature, IF3 increases the binding rate of fMet-tRNA to ribosomes that are associated with cold-shock mRNAs, increases the pool size of 30S subunits, and also lowers the translation fidelity of non-cold-shock mRNAs (Giuliodori et al. 2007). In Myxococcus xanthus, the IF3 family member D signal (Dsg) is involved in both translation initiation and developmental processes, including the control of sporulation and the development of fruiting bodies (Cheng et al. 1994; Kalman et al. 1994). In Rhodobacter sphaeroides, an IF3-like protein called PifC influences the production of bacteriochlorophyll and photosynthetic complexes (Babic et al. 1997). The mechanisms through which Dsg and PifC operate are unknown.

The absence of either IF3α or IF3β affects the expression of many genes in F. diplosiphon, and thus these initiation factors are more widely employed than simply regulating cpeC expression by light color. Recent microarray studies of ΔinfCa and ΔinfCb mutants suggest that the expression levels of 80 % of the 298 light-color-regulated genes in F. diplosiphon were directly or indirectly controlled by IF3α, and that an additional 1,806 genes that were not light-color-regulated were differentially expressed when either IF3α or IF3β was missing (A. Nesbit and D. Kehoe, unpublished results). While it has not been established how many of these gene expression changes are direct, these results indicate that in F. diplosiphon these IF3s are important components of a previously unrecognized gene regulatory system. It is noteworthy that these gene expression differences were measured at the mRNA level because it raises  the possibility that their regulation also could be via transcription attenuation.

Genome database searches of all sequenced non-cyanobacterial, eubacterial genomes revealed that 1 % apparently encode more than one IF3 (Gutu et al. 2013). In cyanobacteria, 22 % of all available sequenced genomes encode multiple putative IF3s. The majority of cyanobacteria with multiple IF3s contain at least two putative IF3s, while the greatest number of putative IF3s in any cyanobacterial species uncovered thus far is five (Gutu et al. 2013). Half of the genomes with multiple IF3s are members of the Nostoc clade, which typically have large genome sizes and extensive phenotypic plasticity in their environmental responsiveness and developmental pathways (Castenholz and Waterbury 1989). Because many of the species with multiple IF3s are not able to regulate cpeC gene expression in response to light color (Tandeau de Marsac 1977), it is likely that IF3 family members control different cellular responses in those cyanobacteria.

Multiple IF3s in Arabidopsis thaliana chloroplasts

Many diatom, green algal, and land plant nuclear genomes also encode multiple putative prokaryotic-like IF3s, which are easily distinguished from their multi-subunit eukaryotic eIF3 counterparts (Gutu et al. 2013). For photosynthetic eukaryotes, it is likely that at least one of the nuclear-encoded IF3s is targeted to the mitochondria and one to the chloroplast. Therefore, only species encoding three or more putative IF3s are considered to possess “multiple” IF3s for an organelle. Genomes encoding multiple putative IF3s were more common in land plants (6 out of 23 genomes, or 26 %) than in cyanobacteria (22 %), while none have been found in algae (Gutu et al. 2013).

Because prokaryotic-like IF3s could localize to either the mitochondria and/or the chloroplast, it is of interest to know where such IF3 family members localize in photosynthetic eukaryotes. In some land plants, the ultimate subcellular destination of any nuclear-encoded putative IF3 can be predicted with reasonable certainty by analysis of the protein’s transit peptide sequence (Small et al. 2004). The nuclear genome of A. thaliana (Arabidopsis, thale cress), a model plant that is a member of the Brassicaceae, is predicted to encode three prokaryotic-like IF3s (Gutu et al. 2013). Encoded on chromosomes 1, 2, and 4, we have tentatively designated these genes AtINFC-1 (At1g34360), AtINFC-2 (At2g24060), and AtINFC-4 (At4g30690) and their corresponding proteins AtIF3-1, AtIF3-2, and AtIF3-4. These are shown, aligned with the F. diplosiphon IF3s, in Fig. 1. AtIF3-1, with a predicted molecular mass of 58.6 kDa, consists of a slightly longer N-terminal region and a significantly longer C-terminal half than IF3α (20.6 kDa), IF3β (20.6 kDa), AtIF3-2 (35.1 kDa), or AtIF3-4 (31.8 kDa). Predotar N-terminal targeting sequence analysis (Small et al. 2004) predicted an 86 % chance that AtIF3-1 was mitochondrial localized and 0 % chance that it was chloroplast targeted. AtIF3-2 was predicted to have a 1 % chance of mitochondrial localization and a 96 % probability of chloroplast localization. AtIF3-4 also had a 1 % probability of mitochondrial localization and an 87 % chance of chloroplast localization. These predictions were tested by stably transforming Arabidopsis with the genes encoding AtIF3-2 or AtIF3-4, fused in frame to DNA sequence encoding yellow fluorescence protein (YFP) (Shaner et al. 2004), and determining the subcellular location of each using a Leica scanning confocal microscope. Based on the co-localization of the YFP and chlorophyll fluorescence signals, both AtIF3-2-YFP and AtIF3-4-YFP localized to the chloroplast (Fig. 2) (A. Nesbit, C. Whippo, R. Hangarter, and D. Kehoe, unpublished). Although the reporter gene encoding AtIF3-1-YFP could not be stably transformed into Arabidopsis, it was transiently expressed in Nicotiana benthamiana, where it did not localize to the chloroplast. Subsequent co-localization studies using AtIF3-1-YFP and Mitotracker-Red demonstrated that AtIF3-1-YFP was targeted to mitochondria in N. benthamiana (Fig. 2) (A. Nesbit, C. Whippo, R. Hangarter, and D. Kehoe, unpublished).

Fig. 1
figure 1

Clustal Omega alignments (Goujon et al. 2010; Sievers et al. 2011) of the two F. diplosiphon IF3s (IF3β, IF3α), the mitochondria-localized (AtIF3-1), and chloroplast-localized (AtIF3-2 and AtIF3-4) putative IF3s from Arabidopsis. Numbering is from the start site of each translated protein. An asterisk indicates a position with a single, fully conserved residue; a colon indicates conservation between groups of strongly similar properties; a dot indicates conservation between groups of weakly similar properties. The color code is: red small (+hydrophobic), blue acidic, magenta basic, green hydroxy, sulfhydryl, amine, and glycine

Full size image
Fig. 2
figure 2

a Micrograph of fluorescence from Arabidopsis cells stably transformed with a gene encoding AtIF3-2 (top row), AtIF3-4 (bottom row) fused at its C-terminus to the yellow fluorescent protein (YFP). In the first two panels of AtIF3-2, brackets denote cell diameter; arrows indicate individual chloroplasts. Fluorescence emission was measured from either YFP (left column), plastid-localized chlorophyll (middle column), or both emissions were merged (right column). Regions of intense yellow fluorescence for both AtIF3-2- and AtIF3-4-transformed cells may be aggregates of YFP-tagged proteins. b Micrograph of fluorescence from N. benthamiana transiently transformed with AtIF3-1 fused at its C-terminus to the YFP and injected with Mitotracker-Red. Fluorescence emission was measured from either YFP (left column), Mitotracker-Red (middle column), or merge of both emissions (right column). White bars indicate 5 μm. Excitation at 514 nm was provided by an argon laser; chlorophyll emission was measured  from 650 to 720 nm and YFP emission measured  from 525 to 600 nm, and sample preparation and microscopy was carried out as previously described (Ruppel et al. 2011)

Full size image

The above results do not address two central issues: are these three Arabidopsis proteins bona fide IF3s, and do AtIF3-2 and AtIF3-4 play unique roles in regulating chloroplast gene expression? We do not yet have direct answers for either of these questions. However, we reasoned that if AtIF3-2 and AtIF3-4 have different roles in Arabidopsis chloroplasts, their expression patterns should differ at different developmental stages and in different tissue and organ types. Using the Arabidopsis eFP Browser developed at the University of Toronto (Winter et al. 2007), we examined AtINFC-2 and AtINFC-4 mRNAs ratios in many locations and under many conditions and found that these varied widely, with an AtINFC-2: AtINFC-4 mRNA ratio of 12.2 in the cotyledon tissue of the heart embryo (see Fig. 3) and an AtINFC-4: AtINFC-2 mRNA ratio of 50.0 in four-hour post-germination pollen (Fig. 3; Table 1). These data strongly support the hypothesis that Arabidopsis optimizes the expression and use of its IF3s throughout development and in response to various abiotic cues. The results of this analysis are therefore consistent with the proposal that AtIF3-2 and AtIF3-4 have unique roles in regulating chloroplast gene expression throughout the plant life cycle. It can also be predicted that other photosynthetic eukaryotes containing multiple chloroplast-targeted IF3s may also use them to regulate chloroplast gene expression.

Fig. 3
figure 3

Using the Arabidopsis eFP Browser at the University of Toronto (Winter et al. 2007), AtINFC-2 (At2g24060) and AtINFC-4 (At4g30690) mRNAs ratios were calculated in many locations and conditions, a subset of which is shown here. For heat map, a higher AtINFC-2: AtINFC-4 mRNA ratio (Log2) is represented by red and a higher AtINFC-4: AtINFC-2 mRNA ratio (Log2) is represented by blue (see scale in lower left corner)

Full size image
Table 1 Arabidopsis AtINFC-2 and AtINFC-4 mRNA Ratios (greater than 4-fold marked in italics)
Full size table

How might the use of multiple IF3s control gene expression?

Although the process(es) by which multiple IF3s are able to influence gene expression remains to be discovered, in this section, we will provide some hypotheses that may be useful starting points for mechanistic investigations. At the outset, it should be noted that the use of multiple IF3s in either prokaryotes or chloroplasts is not an essential capacity, but may provide significantly more variation in the ways that gene expression can be regulated. For cyanobacteria, this additional regulatory layer may be particularly important for species with large genomes and complex responses to environmental cues. A second issue is that because transcription and translation are coupled in cyanobacteria and predominantly uncoupled in plastids, it is hard to imagine that there could be a common regulatory process in cyanobacteria and plastids which involves a translation initiation factor despite their evolutionary relatedness. We propose that the commonly regulated process is the differential efficiency of translation initiation between IF3 family members. We predict that in cyanobacteria, the consequence of this would be variation in the extent of transcriptional/translational coupling, leading to changes in the amount of transcriptional attenuation, while in plastids it would simply alter the degree of translational activity of a specific mRNA or a set of mRNAs (Fig. 4). These hypotheses, and the data supporting them, are explored further below.

Fig. 4
figure 4

Model of the possible similarities and differences in the use of IF3 family members in prokaryotes versus eukaryotic chloroplasts. Both groups use IF3 (brown oval), IF2 (purple oval) and IF1 (red circle) to correctly position the initiator fMet-tRNA (orange oval) within the 30S subunit (light blue oval) along the mRNA (pink 5′ untranslated region, yellow translated region). Some also may use stem-loops 5′ of the ribosome binding site (RBS); these have been shown to be binding sites for RNA-binding proteins in chloroplasts (dark blue and green) (see text) and may or may not also bind protein(s) in prokaryotes (gray question mark). a In prokaryotes such as cyanobacteria, translation initiates during transcription by RNA polymerase (RNAP), providing the potential for translational coupling/uncoupling, possibly through the differential usage of IF3 family members and specific RNA-binding protein(s), and regulation of gene expression by transcriptional attenuation. b In chloroplasts, significant regulation of gene expression occurs at the level of translation initiation, which often involves stem-loops and RNA-binding proteins that may interact with different IF3 family members to further modify the translatability of specific mRNAs

Full size image

Cyanobacteria

In F. diplosiphon, the post-transcription initiation regulation of cpeC is strongly dependent on a region of the 5′ leader approximately 30 nucleotides upstream of the translation start site (Bezy et al. 2011). Mfold predicts that this region folds into a stem-loop with a ΔG of −3.4 kcal mol−1, sufficiently strong to be folded virtually always (de Smit 1998) but not strong enough to cause attenuation of transcription by DNA-dependent RNA polymerase. Yet cpeC post-transcription initiation regulation appears to be at the level of transcriptional attenuation, since cpeC mRNA stability differences were not involved, and differences in extent of translation elongation would have required differences in mRNA stability, which were not detected (Bezy et al. 2011).

Our data also indicate that the amount of IF3α present in the cell, rather than the IF3α:IF3β ratio, was the factor determining cpeC mRNA levels, since infCa overexpression lowered cpeC levels while deletion of infCb, which dramatically shifted the IF3α:IF3β ratio, had no effect (Gutu et al. 2013). Interestingly, cryo-EM structures show that IF3 is positioned on the 30S subunit in a location relatively close to the 5′ leader region of the mRNA (Fig. 4) (Julian et al. 2011), and recent single-molecule FRET studies have demonstrated that there is movement of the C-terminal and N-terminal domains relative to each other while IF3 is bound to the 30S subunit (Elvekrog and Gonzalez 2013). Thus, both domains of IF3 are near the 5′ leader of the mRNA and could directly or indirectly interact with secondary structures in that region, such as the putative stem-loop adjacent to the start codon of cpeC (Fig. 4). IF3s are quite variable in the length and composition of both their N- and C-terminal ends within and between cyanobacterial and plant species (Gutu et al. 2013) (Fig. 5), providing significant potential for variation in how different IF3s might interact with the 5′ leader of any given mRNA.

Fig. 5
figure 5

Clustal Omega alignments (Goujon et al. 2010; Sievers et al. 2011) of a sampling of putative chloroplast and prokaryotic IF3s showing the variability in lengths of the N- and C-terminal regions of these proteins. The protein designations are: Arabidopsis1, the Arabidopsis mitochondria-localized IF3 (AtIF3-1); glycine, Glycine max (soybean) (Glyma02g36710.1); Nostoc71201, Nostoc sp. PCC 7120 (BAB74067.1); Ecoli, E. coli infC; and Nostoc71202, Nostoc sp. PCC 7120 (BAB76322.1). Numbering is from the start site of each translated protein. An asterisk indicates a position with single, fully conserved residues; a colon indicates conservation between groups of strongly similar properties; a period indicates conservation between groups of weakly similar properties. The color code is: red small (+hydrophobic), blue acidic, magenta basic, green hydroxy, sulfhydryl, amine, and glycine

Full size image

Proposed model

Our model proposes a mechanism for post-transcription initiation regulation of gene expression by alternative IF3s in cyanobacteria in which the efficiency of translation initiation of each IF3, either by direct interactions with the 5′ mRNA leader or with one or more gene-specific RNA-binding proteins as in plants, alters the extent of transcriptional-translational coupling and thus transcriptional attenuation. In addition to controlling such a system by changing IF3 levels, IF3 could be posttranslationally modified and there could be changes in ribosomal protein structure or abundance, or changes in the structure or abundance of RNA 5′ leader-binding proteins. This last alternative would provide an explanation for why, in F. diplosiphon, our unpublished microarray data show that some genes are both light color and IF3 regulated, while many others genes are only IF3 regulated, since there could be differences in the extent to which the abundance or activity of different RNA-binding proteins is regulated by light color. Also, because many cyanobacterial genes do not contain a canonically positioned SD sequence, 5′ stem–loop interactions with IF3 could also help correctly position the ribosome on the mRNA, augmenting SD usage or replacing it entirely.

Chloroplasts of land plants

The general absence of transcriptional and translational coupling in plastids strongly indicates that the overall mechanism(s) through which multiple IF3s operate in chloroplasts is likely to be different from those in cyanobacteria (Fig. 4). One way in which IF3s might function to regulate translation of mRNAs in chloroplasts is by increasing the frequency of translation initiation of specific mRNAs. The variety of stem-loops found in the 5′ leader regions of plastid mRNAs (Marin-Navarro et al. 2007) and the many proteins that bind to these regions (Barkan 2011) have the potential to provide tremendous complexity to the regulation of plastid gene expression.

Plastids with multiple IF3s may use one particular IF3 to boost the translation initiation of a specific mRNA as a result of its tighter interactions with the 5′ RNA-binding protein(s) and/or stem-loops of that mRNA than exist for the other IF3s. This would result in the creation of more efficient translation initiation complexes and greater protein production levels overall. These IF3–RNA-binding protein interactions may also partially or completely replace the function of the SD sequence in properly positioning the 30S subunit on the mRNA.

Summary

The discovery that cyanobacteria and photosynthetic eukaryotes have multiple IF3s adds an interesting, new layer of complexity to translation initiation in these organisms by expanding the possible roles of IF3s (Gualerzi et al. 2010) as regulators of gene expression (Gutu et al. 2013). How such regulation is achieved mechanistically is an important question, and the models presented in this review hopefully will help to guide such studies. Another central issue is why the capacity to use multiple IF3s exists predominantly in photosynthetic organisms. One possibility is that multiple IF3s and stem-loops in the 5′ leader regions of mRNAs existed in the progenitor cyanobacteria that were engulfed during the creation of photosynthetic eukaryotes.

Our proposals for how IF3 families posttranscriptionally regulate gene expression in prokaryotes and eukaryotes also may provide an answer to the interesting question of why “translational modulators in chloroplasts are consistently activators of chloroplast translation, (while) translational regulation in bacteria is generally mediated by negative regulators” (Barkan 2011). In our models, plastid translation is weak unless it is positively amplified by increasing the recruitment of the translation initiation machinery to a completed mRNA, which in many cases may involve specific RNA-binding proteins and, in some species, the “correct” IF3 family member. Conversely, in prokaryotes such as cyanobacteria, strong transcriptional activity from genes such as cpeC (Gutu et al. 2013) is acted upon by the translation initiation machinery shortly after transcription initiation and exerts its effect negatively, by attenuating the recently initiated transcription.

Finally, unraveling the roles of IF3 family members in chloroplasts may be of significant agricultural importance, since several economically important crop plants, including soybean and many members of the Brassicaceae family, appear to have multiple chloroplast-localized IF3s. In Arabidopsis, the genes encoding these proteins have very different patterns of expression during the plant life cycle and in response to many different environmental cues. Some of the strongest differences are during seed development, a critical stage of the life cycle for many commercial crops. The modulation of the expression of different IF3s may allow powerful modifications of plant development in key production stages that lead to improved crop quality and yield.