Overview of tRNA Modifications in Chloroplasts

Abstract

The chloroplast is a promising platform for biotechnological innovation due to its compact translation machinery. Nucleotide modifications within a minimal set of tRNAs modulate codon–anticodon interactions that are crucial for translation efficiency. However, a comprehensive assessment of these modifications does not presently exist in chloroplasts. Here, we synthesize all available information concerning tRNA modifications in the chloroplast and assign translation efficiency for each modified anticodon–codon pair. In addition, we perform a bioinformatics analysis that links enzymes to tRNA modifications and aminoacylation in the chloroplast of Chlamydomonas reinhardtii. This work provides the first comprehensive analysis of codon and anticodon interactions of chloroplasts and its implication for translation efficiency.

1. Introduction

Chloroplasts are a fascinating platform for gaining insights into fundamental aspects of biology as well as for biotechnological applications [1,2,3,4]. The attraction of chloroplasts is based in part on the relative simplicity of the chloroplast genome and protein synthesis machinery.

The genetic information contained in DNA and its corresponding mRNA is encoded by sequences of nucleotides (A, T/U, C and G). These nucleotide coding sequences are translated into proteins to fulfill specific molecular functions. The genetic code connects nucleotide bases with a biological signification, i.e., amino acids. The genetic code is organized into nucleotide triplets called “codons” that are used as encryption units for each amino acid. Mathematically, triplets of four nucleotides offer 4³ = 64 possible combinations for encryption. However, each codon is not associated with a unique amino acid. In most organisms, 61 codons encode for a total of 20 amino acids and the last three are used as translation termination signals (UAA, UAG and UGA) (Figure 1). As one amino acid is encoded by several codons, a redundancy emerges in the genetic code.

There are two main reasons for this redundancy. The first reason is that 20 amino acids apparently provide sufficient chemical diversity to fulfill protein functional and structural features [5,6]. The second reason is that a triplet-based code not only provides enough encryption capacity, but the excess of combinations is necessary to maintain the fidelity of the genetic code. Indeed, the genetic code is segregated into codon families or “boxes”, due to the absence of strong discrimination by tRNA anticodons. Therefore, most amino acids are associated with several near-cognate codons in duet, triplet or quartet boxes; and some are even decoded by two boxes (Figure 1).

The fidelity of the genetic code is ensured by two mechanisms, codon–anticodon recognition and correct tRNA aminoacylation. For the first mechanism, codons are decrypted through the complementary pairing with a tRNA anticodon. Codon nucleotides in positions 1, 2 and 3 pair with the anticodon positions 36, 35 and 34 respectively (Figure 2). The first and second nucleotide positions of the codon are associated with base 36 and 35 of the anticodon strictly according to the Watson–Crick complementarity rules (A:U, U:A, G:C, C:G) [7]. In contrast, the interaction between the third codon position and the first anticodon base (N₃₄) allows a more flexible set of combinations expressed in the wobble rules [7,8,9]. In addition, a wide variety of post-transcriptional nucleotide modifications affects the anticodon loop especially in the wobble position (N₃₄) and in the dangling base (N₃₇) adjacent to the anticodon [10,11,12,13]. Modifications in the wobble position (N₃₄) modulate the pairing properties of the anticodon [11,14,15,16], while those in the dangling position N₃₇ provide anticodon stacking and avoid cross talk with symmetric boxes [12,15,17,18]. In 1973, Jukes hypothesized that the genetic code evolved from a primitive form coding for 10 amino acids [17]. Crosstalk between some symmetric boxes was possible due to nucleotide pairing rules, namely G:U/C and U:A/G/U pairing between the first codon base and the third position of the anticodon (N₃₆) (Figure 2). It was proposed that some boxes remained unassigned or decoded by lack of specificity or ribosome frameshifts [19,20]. The appearance of new tRNA isoforms along with tRNA modifications permitted to segregate the genetic code into more defined codons boxes (Figure 1). Importantly, tRNA modifications play a fundamental role in the recognition of the appropriate tRNA species during aminoacylation [21,22,23,24] and during translation through structural features of ribosomes that go beyond codon–anticodon recognition [25,26,27,28].

The second mechanism that safeguards the fidelity of the genetic code is the correct association of a tRNA with its corresponding amino acid. Aminoacylation of tRNAs is a highly specific reaction that is accomplished by aminoacyl tRNA synthetases (aaRSs). Each tRNA aminoacylation is performed by a corresponding aaRS that recognizes specifically one tRNA species and ligates the corresponding amino acid [29,30] (Figure 3). There are two classes of aaRS (class I and II) that vary widely in sequence and structural features, pointing toward distinct phylogenic origins [31,32,33]. Each class of aaRSs recognizes opposite sides of the tRNA acceptor stem (minor or major groove) [34]. The two classes are subdivided into three subclasses (a, b and c) based on mechanistic properties, anticodon-binding domain features and conserved structural motifs [35]. Each class of aaRS is responsible for the aminoacylation of half of the 20 proteogenic amino acids. Aminoacylation is a two-step mechanism in which an aaRS catalyzes the ATP-dependent activation of a specific amino acid, forming an aminoacyl-adenylate intermediate, that is subsequently ligated onto the 3’ end of the corresponding tRNA [36,37]. The structural features leading to the activation of amino acids are conserved within each aaRS class [38]. However, the recognition mechanisms of each amino acid differ substantially among aaRSs [29]. The binding site of an aaRS is usually composed of two sub-domains, each involved in one respective step of tRNA aminoacylation [29]. The enzymatic specificity of the aminoacylation process is a major determinant of the emergence of the genetic code and is an essential step within protein biosynthesis [39], because aminoacyl-tRNAs must consistently carry the same amino acid. Non-covalent interactions play a crucial role in the specific recognition of the correct amino acid and the tRNA species [40,41,42]. It is important to note that the recognition of tRNAs by aaRS enzymes is sensitive to nucleotide modifications occurring on the tRNA, particularly on the anticodon [21,22,23,24]. Interestingly, aaRS enzymes are also involved in non-canonical activities such as cell signaling, RNA splicing, translational and transcriptional regulation.

Chloroplasts originate from the endocytosis of an ancestral prokaryote, that is affiliated with modern cyanobacteria, by a eukaryotic cell [43,44,45]. The eukaryotic host enslaves chloroplasts as power plants by transferring essential prokaryotic genes to the nucleus and rerouting gene products to the chloroplast [1,46]. Therefore, the genome of the chloroplast (plastome) is considerably reduced [47,48], limiting its genetic information to proteins that require co-translational assembly into a multi-protein complex or the association with cofactors such as chlorophyll [49]. Despite the genome reduction, chloroplasts sustain their own translation machinery, which is able to decipher the genetic code with a minimal set of 24 tRNA species by superwobbling [50,51]. The plastidial tRNA set follows the evolutionary strategy that consists in a total depletion of tRNA harboring A₃₄ and C₃₄ in the anticodon [16]. This minimalist strategy conserved a single tRNA species containing G₃₄ and U₃₄ in anticodons to decrypt duet and quartet codon boxes [16].

Microorganisms with minimal genomes (<1 Mb) often utilize a reduced tRNA set for protein translation. The same evolutionary pressure appears to be experienced by organelles, such as the chloroplasts or mitochondria, which maintain a small genome (200 kb and 20 kb resp.) and a simple translation machinery that allows for complete decoding of the genetic code (with some exceptions for mitochondria). Mycoplasma capricolum possesses a minimal genome (1 Mb) that shows striking resemblance to the plastome based on their similar tRNA set characteristics and AT-content equilibria [13,15,52]. E. coli, on the other hand, possesses a more complex genome with 43 different tRNA species, of which many have several genomic copies. Decryption the E. coli genome relies primarily on tRNA isoacceptor concentrations and to a lower extend on codon–anticodon affinity [53]. By contrast, minimal genomes often possess single tRNA copies and mainly rely on codon–anticodon affinity to decipher the genetic code. In the latter case, the characteristics of the anticodon loop, especially nucleotide modifications at base N₃₄ and N₃₇, ensure accurate mRNA decoding and influence codon usage [12,13,14,15,16,52]. The study of the translation machinery of minimal organisms provides fundamental insights into codon–anticodon recognition and protein synthesis dynamics. Furthermore, this knowledge is necessary for synthetic biology approaches, such as heterologous protein expression and genome recoding schemes. We currently do not have a comprehensive overview of tRNA modifications and the enzymes involved in tRNA maturation (aminoacylation and modifications) for chloroplasts. Here we focus on establishing the tRNA modifications of the green algae Chlamydomonas reinhardtii. This choice is based on the wide breath of available data that is present for this organism and its promise as a platform for biotechnological applications.

As only few tRNA modifications have been conclusively demonstrated in Chlamydomonas reinhardtii, we assembled a detailed picture of tRNA structural modification using molecular data from published studies of the chloroplast of algae and plants. We correlated this information with bioinformatics analyses concerning the availability of genes that mediate required biochemical reactions and their prospective targeting to the chloroplast of Chlamydomonas reinhardtii. We cross-validated the information gathered on tRNA modifications with codon–anticodon translation efficiency to obtain a coherent picture of the deciphering of the genetic code in algal chloroplasts.

2. Materials and Methods

The tRNA modifications present in chloroplasts were investigated by performing a meticulous literature review of demonstrated tRNA modifications from biochemical analytical data in various chloroplasts. In order to cross-validate these data or find other potential tRNA modifications, we searched for enzymes implicated in tRNA modifications and tRNA charging. C. reinhardtii nuclear genes coding for aaRS, release factors and tRNA modification enzymes that may be directed to the chloroplast were identified from JGI online resource Phytozome (phytozome.jgi.doe.gov/ (accessed on 5 December 2021)). The analytical data found in the literature, describing plastidial tRNA modifications, and enzymes identified as responsible for these transformations were used to suggest biochemical modifications. In order to discriminate the proteins targeted to the chloroplast from the proteins that remain in the eukaryotic context, we performed chloroplast-targeting analysis (https://services.healthtech.dtu.dk/service.php?TargetP-2.0 (accessed on 10 December 2021)). Based on the scores from TargetP, we estimated the potential presence of these enzymes in the chloroplast (Supplemental Datasheet). Protein blast (blast.ncbi.nlm.nih.gov/Blast.cgi (accessed on 5 December 2021)) was used to obtain C. reinhardtii enzymes that are homologous to identified chloroplastic enzymes in other species (such as Arabidopsis thaliana, cyanobacteria and other algal species). Chloroplastic aaRS genes were previously identified in A. thaliana [54], hence, these enzymes were compared to C. reinhardtii aaRS enzymes by protein blast. The highest percentage of identity permitted to establish with confidence which enzymes were cytosolic or targeted to the chloroplast C. reinhardtii (Supplemental Datasheet). The sequence of tRNA genes found in the chloroplast are presented in Supplemental Table S1. Secondary tRNA structures and potential uridine modifications were predicted with tRNAmod (see Supplemental Figure S1) (webs.iiitd.edu.in/raghava/trnamod/index.html (accessed on 10 December 2021)).

Relative translation efficiency of each codon within each codon box was inferred from literature describing codon–anticodon energy stability or codon decoding rates from ribosome profiling studies (referred in the main text), and/or derived from codon usage evolution (only for quartet boxes possessing modified or unmodified U₃₄) from chloroplast and Mycoplasma [16,52].

3. Results and Discussion

A significant amount of tRNA bases are post-transcriptionally modified, allowing complex interactions that go beyond the standard base-pairing rules. Modifications occur within the anticodon loop as well as on the rest of the tRNA. All these modifications play crucial roles in the recognition of tRNA by aaRS enzymes and by the ribosome.

3.1. Common Nucleotide Modifications of tRNA Backbones in Chloroplasts

Common modifications of chloroplast tRNA that have been observed outside the anticodon arm include 2-o-oxymethyl-Guanine Gm₁₈, pseudouridine ψ₅₅, ψ₂₆ and ψ₂₇, methyl guanidine m⁷G₄₆, m²G₁₀, dihydrouridine D_{(16-20-21-47)} [55,56,57,58,59,60,61]. The stem preceding the anticodon loop is very often composed of a Watson–Crick base pair at position referred as B_29-41, followed by a G/C Watson–Crick pair at B_30-40 to provide a stable helical conformation. One base pair further, either a C=G or a A–ψ pair is preferred at B_31-B39 [15]. At the start of the anticodon loop, B₃₂ and B₃₈ form a non-Watson–Crick pair. The former being a pyrimidine (U, C) occasionally modified to ψ, Cm or Um; and the latter being most frequently an A or less often a C or ψ [62]. Notably, the invariant residue U₃₃ constitute the U-turn between the 5′ and 3′ helical stacks of the anticodon loop [12]. These characteristics are widely found among bacteria and thought to maintain the biophysical properties of the stem preceding the anticodon loop [15]. A list of all plastidial tRNAs and their predicted structures is presented in Supplemental Table S1 and Figure S1. We investigated the nuclear genome of C. reinhardtii to identify genes involved in the various tRNA modifications. Each enzyme was subject to a subcellular targeting analysis to evaluate their potential presence in the chloroplast (Supplemental Datasheet).

3.2. Modifications of the Anticodon-Adjacent Nucleotide (N₃₇) Maintains the Fidelity of the Genetic Code

Modifications of the “dangling” base in position 37 affects the stability and specificity of the anticodon loop. Its purpose is to increase stacking of the anticodon first nucleotide in order to ensure proper pairing of N₁:N₃₆ [10,63]. These modifications are usually dependent on the first base of the codon (N₁). In the case of C₁NN codons, with the exception of arginine codons, the purine 37 of each tRNA is a guanine that is invariably changed to N1-methyl-guanosine (m¹G₃₇) in all three kingdoms of life [15]. Hence, in the chloroplast of C. reinhardtii, tRNAs of proline (UGG), leucine (UAG), histidine (GUG) and glutamine (UUG) all possess m¹G₃₇ [57,59,61,64,65], while arginine (ICG) contains m⁶A₃₇ [65,66]. The 1-methylguanosine in position 37 (m¹G₃₇) is determinant for the quality of codon–anticodon pairing as well as the avoidance of frameshifts for C-starting codons (C₁NN) [12,67].

All other codons are decrypted with tRNAs containing an adenine in position 37. In the chloroplast of C. reinhardtii, A₃₇ can be modified to 2-methylthio-N6-isopentenyladenosine (ms²i⁶A₃₇), N6-threonylcarbamoyladenosine (t⁶A₃₇) or N6-methyladenosine (m⁶A₃₇) [16,55,56,57,58,59,60,64,65,67,68,69,70]. In many organisms, it is generally modified when the adjacent base B₃₆ is either an A or a U; this pattern of modification correlates with the necessity to stabilize the weak neighboring codon–anticodon pair A₁:U₃₆ and U₁:A₃₆ [10,12,15,63].

For the first modification, MiaA and MiaB are the enzymes known to modify A₃₇ to i⁶A₃₇ and ms²i⁶A₃₇, respectively, for all U-starting codons (U₁NN) in bacteria [71,72,73], namely Tyr, Phe, Cys, Ser (UCN), Trp and Leu (UUG/A). The role of this modification is to maintain correct U₁:A₃₆ pairing and prevent any symmetric crosstalk [12,17]. We found one orthologue of MiaA and two of MiaB that have moderate to high likeliness to be translocated to the chloroplast (Supplemental Datasheet). The i⁶ or ms²i⁶A₃₇ modification were confirmed experimentally for tRNA-Phe [59], tRNA-Trp [58], tRNA-Tyr [68] and tRNA-Cys [69] in plant and algal chloroplasts.

Two additional adenine modifications t⁶A₃₇ and m⁶A₃₇ have been identified on chloroplastic tRNAs but their respective enzymes, which modify base 37, remain unknown is C. reinhardtii. A potential candidate for this function is the gene Cre10.g455400, which is annotated as coding for an enzyme adding m⁶A to tRNA-Val (however, valine was not shown to contain m⁶A₃₇, but the enzyme may act on other tRNAs) and displays features that may allow it to be translocated to the chloroplast (Supplemental Datasheet). The residue t⁶A₃₇ plays a role for preventing G₁:U₃₆ mispairing for all A₁NN codons therefore ensuring fidelity of the genetic code. m⁶A₃₇ has a similar role but applies to various tRNAs [12,17]. In chloroplasts, the t⁶A₃₇ was identified in tRNA-Ile (GAU) [60] and (k²C₃₄AU) [56], tRNA-Lys [55], tRNA-Thr₃ is modified in spinach to m⁶t⁶A₃₇ [70] while the tRNAs for Asn, Ser (AGU/C) and Arg (AGA/G) have not been investigated. Finally, m⁶A₃₇ occurs in tRNA-Met_e (in line with experimental data for tRNA-Arg-ICG) [55,66] while A₃₇ of tRNA-Met_i remains unmodified [74].

In contrast to the aforementioned modifications, plastidial tRNAs reading G₁NN codons do not possess modifications of A₃₇ [55,75,76], with the exception of tRNA-Asp [77] which has not been identified. Indeed, according to nucleotide pairing rules [7], cytosine strictly pairs with guanosine, which is sufficient to prevent frameshifts and mispairing of N₁:C₃₆.

The modifications affecting the dangling base are crucial to maintain genetic code accuracy. However, modifications occurring in the anticodon, especially in base 34, are of particular importance because they dictate tRNA decoding properties. These modifications result either in a higher codon–anticodon specificity or a less stringent recognition for superwobbling [10].

3.3. Nucleotide Modifications Affecting Anticodons Shape Their Deciphering Properties

In the next section we analyze all plastidial tRNA species with their anticodon modifications regarding their effect on decoding properties.

3.3.1. The Anticodon Base 34 of Single Codon Box

Amino acids encoded by a single codon require a high codon–anticodon specificity to avoid misreading of near-cognate codons. Only methionine (AUG) and tryptophan (UGG) possess a single codon. For both amino acids, the anticodon of the plastidial tRNA ends with C₃₄, which, according to the wobble theory [7], reads only G-ending codons.

In the case of methionine, there are actually two distinct tRNAs, one optimized for translation initiation (tRNA-Met_i) [78] and one for polypeptide elongation (tRNA-Met_e). The former contains a ψ in position 39 and an unmodified anticodon loop [74], while in the latter, ψ is found in position 32 and the aforementioned m⁶A₃₇ residue [55,66]. Interestingly, in some prokaryotes C₃₄ can be modified to ac⁴C₃₄ to maintain accurate translation of only the AUG codon [79,80], although near-cognate codon reading is rare. While an orthologue of an RNA cytidine acetyltransferase is present in the nuclear genome (Cre03.g192850) the modification was not shown to appear on tRNA-Met [66] but was detected in rice chloroplast lysate [81]. Hence, C₃₄ certainly provides enough discrimination to read only the AUG codon.

The tRNA of tryptophan contains ψ in position 39 and 38, an unmodified C₃₄CA anticodon and an i⁶A₃₇ or ms²i⁶A₃₇ modification [58]. In eubacteria, C₃₄ is usually modified to Cm₃₄ to avoid misreading of UGA stop codons [16]. The absence of UGA codons in the chloroplast genome may also indicate that tRNA-Trp could interact slightly with this codon without detrimental consequences. For example, in M. capricolum, the two codons UGG and UGA encode tryptophan while RF2 has been lost, showing the lack of requirement for efficient discrimination between the two codons [16,82]. We hypothesize that the enzyme performing the modification to Cm₃₄/Um₃₄ is Cre10.g417650, which is likely shuttled to the chloroplast (Supplemental Datasheet); however, the enzymatic product was found only on tRNA_Leu-Um₃₄AA. Either the modification was not detected by Canaday et al. at the time [58] or it is not present and the low affinity A₁:G₃₄ does not interfere with RF2 termination as demonstrated by Young et al. [83] if RF2 is present at all in chloroplasts.

The last case to discuss in this section is the isoleucine tRNA reading AUA codons. Isoleucine is encoded by three codons, AUU/C and AUA, which are recognized by two distinct tRNA species. The tRNA reading AUA actually originates from a tRNA with methionine-like anticodon C₃₄AU. It is post-transcriptionally modified at position 34 into k²C₃₄ (lysidine) [21,56,84]. Thus, tRNA-Ile-(k²C₃₄AU) accomplishes specific decoding of AUA codons for isoleucine. The enzyme responsible for lysidination was identified as Cre13.g572800 and is predicted to be exported to the chloroplast (Supplemental Datasheet).

3.3.2. The Anticodon Base 34 of NNU/C Duet Codon Boxes Are Mostly Unmodified

Duet codon boxes are divided into two groups, pyrimidine-ending codons (NNU/C) and purine-ending codons (NNA/G). In the chloroplast, within each (NNU/C) box, codons are decoded with a single tRNA containing G₃₄ in the first position of the anticodon. This group is composed of the amino acid boxes coding for Phe, Cys, Tyr, Asp, His, Asn, Ile (AUC/U) and Ser (AGC/U). All these codons possess less than two G or C in codon positions N₃₅ and N₃₆, which infers a low to intermediate codon–anticodon binding energy [15]. The anticodon G₃₄NN recognizes NNC₃ codons by Watson–Crick pairing and less efficiently NNU₃ codons by U₃:G₃₄ wobble pairing [15,85,86,87,88]. The C₃:G₃₄ Watson–Crick pair forms three hydrogen bonds, compared to two for the U₃:G₃₄ wobble pair, which increases the codon–anticodon binding energy and its stability. This increase in codon–anticodon energy and the Watson–Crick geometry improves codon recognition by the ribosome, thus allowing faster translation of the C₃:G₃₄ Watson–Crick by the ribosome [28,88].

Interestingly, there is evidence pointing to the presence of queuine (Q) instead of G₃₄ in certain plant tissues or under particular growth conditions [77,89,90]. To corroborate these observations, two queuine tRNA-ribosyltransferases were identified in the nuclear genome of C. reinhardtii and one of them (Cre12.g558400) showed a high probability of being transferred to the chloroplast. This enzyme acts on all tRNAs harboring a U₃₅ in the second anticodon position (His, Asn, Asp and Tyr). Nonetheless, queuine has not been found in the tRNA of His and Asn and is only present in specific conditions for Asp and Tyr in plants [77,89,90]. One explanation for this is that most eukaryotes, including C. reinhardtii, are not able to synthesize queuine. Instead, eukaryotes rely on an external bacterial source for queuine supply. Interestingly, a salvage pathway was identified in C. reinhardtii [91]. In queuine replete conditions, C. reinhardtii may incorporate Q₃₄ in place of G₃₄ in G₃₄U₃₅N plastidial tRNAs. This may underlay specific gene expression regulation due to the ability of queuine (Q₃₄) for decoding NNC/U codons that differs from G₃₄ [92]. Indeed, G₃₄ pairs preferentially with NNC compared to NNU [15,85], while Q₃₄ exhibits less bias between C₃ and U₃ deciphering depending on the context of the anticodon loop [8,86]. Interestingly, tRNA modifications play a role in differential gene expression in response to a stress factor [93,94].

3.3.3. The Anticodon Base 34 of NNA/G Duet Codon Boxes Restricts Deciphering to Purine Codons

In the second group of duet boxes, NNA/G codons are decoded by single tRNA containing a modified U₃₄. This group is composed of the 2-codon boxes of Lys, Glu, Gln, Arg (AGA/G) and Leu (UUA/G). In bacteria, the most common modifications of U₃₄ are 5-iminomethyl-U₃₄ derivatives (nm⁵U, mnm⁵U, cmnm⁵U). In addition, the same uridine is sometimes thiolated (s²U). We identified candidate enzymes in C. reinhardtii responsible for the addition of cmnm⁵, mnm⁵s² and cmnm⁵s² to U₃₄ that are likely to be shuttled to the chloroplast (Supplemental Datasheet). These modifications affect the tRNAs of Glu, Gln, Lys and probably Arg (AGA/G) [55,64,75]. Although the thiolation of U₃₄ generally affects the tRNAs of glutamate, lysine and glutamine in bacteria [95], U₃₄ thiolation was previously detected solely for glutamate in chloroplasts [75]; however, the inaccuracy of the chromatographic method used for the lysine and glutamine plastidial tRNAs [55,64] may leave some place for the presence of thiolation on U₃₄ of these two tRNAs. The bacterial tRNA-Leu (UUA/G) also contains the cmnm⁵s² group on U₃₄ [15], although it was not found in plastid versions [61]. However, just as its bacterial counterpart, the plastidial tRNA-Leu (UUA/G) contains a 2′-O-methyluridine (Um₃₄) in the wobble position [61]. These different U₃₄ modifications restrict the reading to purine codons with a strong preference for A-ending codons [8,15,16,96,97,98,99].

3.3.4. The Anticodon Base 34 of Quartet Codon Boxes Expands tRNA Reading Properties by Superwobbling

In the chloroplast of C. reinhardtii, each amino acid box encoded by four codons is read by a single respective tRNA. Only the glycine box possesses an additional tRNA (anticodon GCC) that was shown to be dispensable [51]. Decoding of four-codon boxes requires the presence of at least one G or C in position 35/36 of the anticodon [12,100] and other important residues in the anticodon loop such as C₃₂ [101,102]. These residues increase the pairing strength between codon and anticodon, making the third codon base less significant for decoding [15]. According to the superwobble theory, a single tRNA containing U₃₄ is able to read all codons of quartet family boxes [12,13,50,51,100]. In a majority of bacteria, U₃₄ is modified to xo⁵U derivatives [15,16]. In E. coli, it is modified to cmo⁵U₃₄ that provides the ability to read all four codons of the family boxes [103,104,105]. However, synthesis of these xo⁵U derivatives were not found in C. reinhardtii by mass spectrometry [81]. Remarkably, Mycoplasma and mitochondria use unmodified U₃₄ to read the quartet boxes [15]. The chloroplast shows striking resemblance to the M. capricolum tRNA set and the reading properties within quartet boxes due to its minimalistic features and similar GC content [15]. In M. capricolum, modifications such as s²U₃₄ and Um₃₄ are always absent from this type of tRNA to prevent steric hindrance, thus making these tRNA less stringent at the wobble position [16,105]. In the chloroplast, U₃₄ is modified in the tRNAs of Pro, Ala, Val and potentially Ser (UCN₃) and Thr as well, but the precise nature of the modification remains unidentified [57,65,76]. The tRNA of glycine (U₃₄CC) contains an unmodified U₃₄ able to read all four codons with a reasonable efficiency, although less efficient decoding of some codons may be the reason why the second tRNA (G₃₄CC) is maintained in the plastome [55]. Additionally, it was demonstrated that the presence of C₃₂ is crucial in helping tRNA-Gly to decode all four codons [101,102]. In the case of leucine (CUN₃), U₃₄ is not modified; however, G₃₆ is methylated to m⁷G₃₆ [59,61]. This modification allows tRNA-Leu (UAm⁷G₃₆) to read the four codons of the associated box. Additionally, base 37 is modified to m¹G₃₇ to provide 5′ stacking power to the anticodon [59].

The above-mentioned tRNA species containing an unmodified U₃₄ nucleotide efficiently recognize A-ending codons, whilst the G₃:U₃₄ wobble base is fairly unstable [15], C-ending codons present a poor pairing efficiency [106], and there is a lack of data for U-ending codons. The deciphering properties of the other tRNAs from quartet boxes, harboring the unknown modification of U₃₄, have not been investigated. However, the deciphering properties can be deduced from the codon usage of quartet boxes containing a unique deciphering tRNA. Indeed, according to Ikemura, the evolutionary forces shaping codon usage are tRNA concentrations, in the case of several tRNA isoacceptors for the same amino acid; or codon–anticodon affinity in the case of a single tRNA decoding an entire codon box [53,107]. Here, the tRNA set of quartet boxes of the chloroplast follows the second case: the higher the codon–anticodon affinity, the higher the codon usage. The codon usage of the chloroplast of C. reinhardtii is strongly biased toward A- and U-ending codons for quartet boxes [52], as for M. capricolum [12,15]. Even though evolutionary mutational bias leads to a higher AT composition in the chloroplast, there is an active pressure toward a strong enrichment of A- and U-ending codons for quartet boxes [52]. Therefore, we hypothesize that anticodons containing modified or unmodified U₃₄ lead to higher translation efficiency of A- and U-ending codons within quartet boxes in the chloroplast of C. reinhardtii [52].

At last, we investigate the special case of the arginine box (CGN₃). Here, the first base of the anticodon is modified from uridine to inosine (I₃₄) [55,65,108]. This gives the tRNA the ability to decode preferentially U-ending codons; inosine also pairs well with CGC codons, but poorly recognizes CGA. It has been shown that I₃₄ can decode NNG₃ albeit very inefficiently [66,104]. This might explain the extremely low occurrence of CGG codon in the plastome (only three occurrences). Other chloroplast-containing organisms such a liverwort retained a tRNA-Arg (C₃₄CG) that can read the CGG codon more efficiently [109].

We summarized the most important modifications of the anticodon loop in Figure 4. Additionally, we present the deciphering properties of each tRNA based on the translation efficiency of their respective codons (Figure 5).

3.4. Translation Termination

In prokaryotes and organelles, there are two distinct release factors responsible for terminating translation. The first release factor (prfA/RF1) recognizes UAA and UAG codons while the second (prfB/RF2) recognizes UAA and UGA and halt peptide synthesis [110,111]. Interestingly, the specificity of the release factors RF1 and RF2 toward stop codons is ensured by the tripeptides Pro-Ala-Thr and Ser-Pro-Phe respectively [112]. Release factors possess a universally conserved Gly–Gly–Gln (GGQ) motif that is responsible for the hydrolysis of the ester bond between the last tRNA in the ribosomal P-site and the peptide [113]. Among the 67 coding sequences present in the chloroplast, only three genes (psbL, clpP, ftsH) possess a UAG stop codon while all the other genes use UAA to stop translation. It is interesting to note that the Opal stop codon UGA is not present at all. This led to the hypothesis that only release factor 1 may perform translation termination [83]. Both release factors prfA (Cre16.g673617 or Cre06.g289350) and prfB (Cre01.g010864) are nuclear-encoded. While it remains unclear if the prfA gene product is effectively translocated to the chloroplast, protein alignments show that Cre16.g673617 is more closely related to RF1 from Synechocystis PCC 6803 and to AtcpRF1 from Arabidopsis thaliana, the latter of which has been shown to be translocated into the chloroplast [114] (Supplemental Data S3). Surprisingly, C. reinhardtii retained prfB (Cre01.g010864), which bioinformatics analysis indicates is likely being targeted to the chloroplast (Supplemental Data S3) and may still provide active plastidial UGA termination. It was demonstrated that UGA occurrence increases compared to UAG relatively to the abundance of RF2, suggesting that in this case RF2 concentration is low or inexistent [115]. In addition, organisms with very high AT3 content (i.e., third codon position), like the chloroplast, tend to evolve toward decreasing RF2 concentration close to zero and axiomatically drive the reassignment of UGA codons to UAA that is consistently over-represented in highly expressed genes [115]. Young et al. showed that UGA opal codons can be translated as tryptophan when a genomically integrated tRNA-_Trp-UCA containing a modified UCA anticodon is expressed [116]. A temperature-sensitive tRNA was used to engineer a cold-inducible translational system (CITRIC) responding to the reassigned UGA codons [116]. These special characteristics regarding translation termination constitute an opportunity for stop codon reassignment, where the newly freed codon is used to integrate non-canonical amino acids into proteins.

3.5. Comparison of tRNA Modifications within Minimal Genomes

Here, we investigate the nucleotide modification strategies of the chloroplast and other minimal genomic systems (mitochondria and M. capricolum) as well as the eubacterial model E. coli (

Table 1. Comparison of the tRNA modifications occurring in base N₃₄ and N₃₇ for chloroplast, mitochondria, M. capricolum and E. coli. Compiled information collected in this paper for the chloroplast and sourced from Grosjean et al. for the other organisms [15]. The type of codons affected by the modifications is indicated on the lefthand side of the table. Asterisk next to a nucleotide refers to an unknown modification.

Modified Base		Codons		E. Coli	M. Capricolum	Mammalian Mitochondria	Chloroplast
N37	A37	(U1NN)		ms2i6 A		ms2i6 A	ms2i6 A
					m6 A	i6 A	i6 A
						unmodified A
		(C1NN)		m2 A	m6 A	unmodified A	m6 A
		(G1NN)		m2 A	m6 A		A*
				unmodified A	unmodified A	unmodified A	unmodified A
		(A1NN)		m6t6 A			m6t6 A
				ct6A	ct6A	ct6A	t6A
					m6 A		m6 A
					unmodified A	unmodified A	(unmodified A)
	G37	(C1NN)		m1G	m1G	m1G	m1G
N34		NNU/C (duet boxes)	NNU/C	unmodified G	unmodified G	unmodified G	unmodified G
			NAU/C	Q	unmodified G	Q	unmodified G (maybe Q)
				GluQ
		NNA/G (duet boxes)		mnm5s2U
				mnm5Um	cmnm5U	τm5U	cmnm5U
				cmnm5s2U	cmnm5s2U	τm5s2U	cmnm5s2U
				cmnm5Um	cmnm5Um		Um
				Cm
		NNU/A/C/G (quartet boxes)		cmo5U	unmodified U	unmodified U	unmodified U
				unmodified G			U*
				unmodified C
		quartet Arginine		I	I	unmodified U	I
		NNG	Trp	Cm	Cm	τm5U	unmodified C
					cmnm5Um
			Met	ac4C	unmodified C	f5C	unmodified C
		Isoleucine AUC		k2C	k2C	(NA)	k2C

). All these systems possess the m¹G₃₇ modification for the C₁NN codons. When adenine is present in position 37, the chloroplast follows similar strategies as the bacteria and mitochondria with slight exceptions. For U₁NN codons, the chloroplast utilizes the modifications ms²i⁶A₃₇ and i⁶A₃₇ that are also present in mitochondria and E.coli. For C₁NN and G₁NN codons, the chloroplast presents similar tRNA characteristics as Mycoplasma with m⁶A₃₇ and unmodified A₃₇. Some differences occur for A₁NN codons, where the chloroplast displays the m⁶A₃₇, unmodified A₃₇ as in Mycoplasma. Furthermore, the chloroplast possesses a t⁶A₃₇ modification instead of ct⁶A₃₇ found in the three other organisms, which have similar functional implications. The chloroplast also contains the m⁶t⁶A₃₇ modification that is present only in E. coli. Mitochondria display the lowest amount of modifications in base 37, while the chloroplast possesses similar characteristics as M. capricolum, at the notable exceptions of m⁶t⁶A₃₇ and ms²i⁶A₃₇ that resemble the more complex organism E. coli.

The plastidial modifications occurring on base N₃₄ are similar to the ones occurring in M. capricolum, with a few interesting differences (Table 1). Within the NNU/C duet boxes, the two bacteria and the two organelles use unmodified G₃₄ to read both codons, except for codons containing A₃₅ (NA₃₅U/C). For these codons, tRNAs from E. coli and mitochondria harbor the queuosine modification that might be present in chloroplast under specific growth conditions. The NNA/G duet boxes are decrypted using the restrictive groups cmnm⁵, cmnm⁵s² added to U₃₄ or Um₃₄, which are similar deciphering strategies as E. coli and Mycoplasma. Interestingly, mitochondria opted for different modifications, τm⁵ and τm⁵s², which have the same restrictive deciphering properties as in plastids. When it comes to quartet boxes, E. coli utilizes the cmo⁵ modification of U₃₄ to read all four codons of the quartet box but also contains additional tRNAs harboring unmodified C₃₄ and G₃₄. In contrast, organelles and Mycoplasma decipher the quartet boxes with tRNAs containing unmodified U₃₄, known to confer superwobbling [50,51], which is a common strategy for decoding minimal genomes. Strikingly, the chloroplast displays an unknown modification of U₃₄, different from cmo⁵, that can decipher all codons of quartet boxes. In the case of methionine, the chloroplast and M. capricolum display an unmodified cytosine to read uniquely the AUG codon, which may offer enough discrimination through Watson–Crick base pairing to avoid reading near-cognate codons; E. coli evolved the ac⁴C₃₄ modification for more accurate codon discrimination, while the mitochondrial modification f⁵C₃₄ reads both AUG and AUA. The last-mentioned AUA codon is allocated to isoleucine using the k₂C₃₄ modification for the three other examples. Last, in the case of tryptophan, the chloroplast adopts a strategy consisting of an unmodified C₃₄, while E. coli possesses the Cm₃₄ modification, both sufficient to read only the UGG codon. M. capricolum and mitochondria possess the cmnm⁵Um₃₄ and τm⁵U₃₄ modifications respectively, which also assign the UGA codon to tryptophan instead of stop signal.

Overall, the chloroplast shares several tRNA modifications with other minimal genomic systems.

4. Conclusions

This study presents a comprehensive view of codon–anticodon interactions in the chloroplast of C. reinhardtii. Many of the experimentally established and postulated tRNA deciphering strategies and their basis in tRNA modifications for C. reinhardtii, are likely also found in the chloroplasts of other members of the Viridiplantae. The nature of these tRNA modifications defines how the genetic code is deciphered in chloroplasts. Each modification presents specific aptitudes in reading a codon family based on codon–anticodon pairing affinity and interactions with ribosomes. The chloroplast displays a complex interplay of evolutionary forces that lead to a minimum set of tRNAs, which is modified to maintain tRNA-aminoacyl synthesis fidelity, as well as tRNA–ribosome recognition and protein synthesis.

The high codon–anticodon affinity is associated with a decreased codon decoding time during translation [15,117], thus improving protein yield. The relationship between codon–anticodon pairing affinity and protein yield has been recognized in the codon usage of highly expressed genes [52]. Appreciating and manipulating this interplay is vital for realizing the potential of heterologous gene expression and biotechnological applications in the chloroplast.