The plant family Asteraceae (Compositae), commonly known as the daisy or sunflower family, is among the three megadiverse families that comprise up to 25% of angiosperm species1. The Asteraceae family has between 25,000 and 35,000 species which is ~ 10% of flowering plants and comparable only to the Fabaceae and Orchidaceae families1. These species are diverse in distributions and habitat, exist on every continent, including Antarctica, and occupy every type of habitat1,2. This family is divided into 13 subfamilies, including Barnadesioideae, Famatinanthoideae, Stifftioideae, Mutisioideae, Gochnatioideae, Wunderlichioideae, Hecastocleidoideae, Pertyoideae, Carduoideae, Gymnarrhenoideae, Cichorioideae, Corymbioideae, and Asteroideae1,3,4. Among these families, the four subfamilies Gymnarrhenoideae, Cichorioideae, Corymbioideae, and Asteroideae are considered core Asteraceae5. The subfamily Asteroideae is the youngest and largest subfamily of Asteraceae, comprising more than 17,000 species1,4.

The chloroplast is a vital organelle in plants due to its role in photosynthesis6. It is prokaryotic in origin and shows uniparental inheritance—paternal in some gymnosperms and maternal in most angiosperms7,8,9. The uniparental inheritance and variable mutation rate of different regions of the chloroplast genome make it suitable for studies ranging from population genetics to phylogenetics10,11. Many mutational events occur in chloroplast genomes, including InDels (Insertions-deletions), substitutions, inversions, and copy-number variations12,13,14,15. Some of these mutational events also lead to complete deletion or pseudogenization of genes within the chloroplast genome, including protein-coding genes and transfer RNA genes16,17,18. Pseudogenization is a process by which a functional gene becomes non-functional. Pseudogenes have significant homology to functional genes but disruptive mutations led to lost of  function often via formation of truncated proteins19,20. Pseudogenes reflect the evolutionary past19, thus being important elements shaping genome content21. Pseudogenization has been linked to gene size and shown to occur more frequently in larger genes encoding a protein product of up to 1000 amino acids22. It can also be linked to ecological niches such as utilization of resources related to energy, metabolism, interaction among organisms, host-specific responses, and lifestyle of the organism19,21.

Two transfer RNA genes exist for threonine, located in the large single-copy region of the chloroplast. One copy of threonine (trnT-GGU) lies between protein-coding genes atpA and psbD along with trnG-UCC and trnR-UCU. Another copy of threonine (trnT-UGU) is located between rps4 and trnL-UAA near the trnT-F region, which is widely used in phylogenetic analyses and barcoding studies4,23. Based on the wobbling rule24, threonine is encoded by four codons during translation in which the two codons ACC and ACU are decoded by trnT-GGU, whereas the two remaining codons ACA and ACG are decoded by trnT-UGU. However, a study based on functional analysis of the plastid genome of tabacum, after developing mutant lines, indicates that the trnT-UGU is able to degenerate all four codons of threonine, making trnT-GGU unessential25 for translation of threonine. The pseudogenization of trnF-GAA has also been reported in some plant lineages17,26,27. Previously, the pseudogenization of trnT-GGU has been reported in Cryptomeria japonica D. Don. (family Cupressaceae)28, Pelargonium × hortorum (family Geraniaceae)29, and Anaphalis sinica and Leontopodium leiolepis of the tribe Gnaphalieae (Asteroideae (Asteraceae)30. Here, we are interested in determining the range of trnT-GGU pseudogenization in the family Asteraceae, its possible mechanism of pseudogenization, and the process of codon degeneration in its absence. To the best of our knowledge, we for the first time analyzed the trnT-GGT genes in 134 representative species of Asteraceae belonging to 13 subfamilies, which were diverse in habit and habitat and included 97 species of Asteroideae (Table S1). We report that trnT-GGT is either absent or a pseudogene in core Asteraceae due to an insertion event in the 5′ acceptor stem. Moreover, codon usage analysis indicates that superwobbling may be a possible mechanism by which species decode all four codons using trnT-UGU in the absence/pseudogenization of trnT-GGU.


Analysis of trnT-GGU among species of Asteraceae

We compare the trnT-GGU gene among 13 subfamilies of Asteraceae. The analyses revealed an insertion event (i.e., CTTTT/TTTTC/TTTCC) at the 5′ acceptor stem of the trnT-GGU gene in core Asteraceae, while this was lacking in the species of other subfamilies of Asteraceae (Fig. 1a,b). This insertion event was found to be linked to the pseudogenization of the trnT-GGU in all subfamilies of core Asteraceae based on the result of ARAGORN, as the gene was not annotated in any species. However, the gene was found to be non-functional in three subfamilies of core Asteraceae, as a functional copy was predicted in single species of Corymbioideae with low infernal score based on the result of tRNAscan-SE (Fig. 1b). The high infernal score indicates high matching with other tRNAs of the database and reflects high accuracy of the predicted tRNAs. Therefore, the infernal  program was integrated into tRNAscan-SE to improve performance and prediction accuracy and to achieve a better functional classification of tRNA. The infernal score of the trnT-GGU gene ranged from 49.4 to 65 in the species of those subfamilies that lacked the insertion event (Table 1), indicating that the gene is completely functional in these species. In contrast, tRNAscan-SE detected mismatch isotypes of the trnT-GGU gene in Gymnarrhenoideae, with a low infernal score and pseudogene in Cichorioideae, functional copy with low infernal score in Corymbioideae, and diverse types of results in the subfamily of Asteroideae (Table S2). The structure of the trnT-GGU gene of the species of each subfamily showed that the mismatch/mismatches is/are present in the species of Gymnarrhenoideae, Cichorioideae, Corymbioideae, and Asteroideae mostly at the 5′ acceptor stem and anticodon loop, whereas the species that lacked the aforementioned insertion have a complete cloverleaf structure (Figs. 2, S1). These data suggest that the pseudogenization event might be widespread in the Asteraceae family and may be limited to core Asteraceae.

Figure 1
figure 1

Multiple sequence alignment of the plastid threonine (trnT-GGU) gene and the position of pseudogenes within the phylogenetic tree. (a) All functional parts of the gene have been noted above the alignment. The insertion occurring in the acceptor stem is highlighted. Co-occurrence of mutational events in some species are shown above and below the alignment. (b) The black block indicates the starting node of the insertion event in the 5′ acceptor stem and pseudogene detected by ARAGORN among species of the ‘Core Asteraceae’ clade, whereas the purple block indicates the presence of pseudogenes based on tRNAscan-SE. Leaves of the phylogenetic tree from Barnadesioideae to Corymbioideae represent subfamilies of Asteraceae, while all other leaves of phylogenetic trees represent 13 tribes of the Asteroideae subfamily (indicated in blue background), followed by the icon size photo of a representative species used in our analysis. The species from 13 tribes of Asteroideae were included in the analysis and their names are noted at each node in the highlighted background. The representative photos of each subfamily and tribe are included in the figure and the species names are provided in front of each photo.

Table 1 Prediction of trnT-GGU genes in the representative species of 13 subfamilies.
Figure 2
figure 2

taken from each subfamily of core Asteraceae. The trnT-GGU gene of Barnadesia lehmannii is labeled to show the functional parts as representative of all species. The perfect clover leaf structure of trnT-GGU exists in the species of nine subfamilies, including Barnadesioideae, Famatinanthoideae, Stifftioideae, Mutisioideae, Gochnatioideae, Wunderlichioideae, Hecastocleidoideae, Pertyoideae, and Carduoideae. The B. lehmannii represent the structure of trnT-GGU of the species of all nine subfamilies. The insertion occurred in the species of four subfamilies of core Asteraceae (Gymnarrhenoideae, Cichorioideae, Corymbioideae, and Asteroideae), which also correspond to mismatches above the anticodon loop. The insertion is highlighted with a box.

Structure of trnT-GGU gene of species of 13 subfamilies. One species was

Analyses of trnT-GGU genes among the species of Carduoideae

The analyses of 11 species from 11 different genera of Carduoideae showed that the functional trnT-GGU gene with a high infernal score ranged from 55.7 to 57.1 (Table S3). Except for Atractylodes chinensis (DC.) Koidz, the analyses of the other ten species revealed the presence of anticodon CGU (Fig. S2). We also found an insertion (CTCAG) in the D-loop of Saussurea inversa Raab-Straube, which slightly decreases the infernal score to 55.7. The structure analyses support the presence of all functional parts of the gene in the species of Carduoideae (Fig. S1).

Analyses of trnT-GGU genes among the species of Cichorioideae

The analyses of 13 species from 13 genera of Cichorioideae revealed the pseudogenization of the trnT-GGU gene in all species based on the result of ARAGORN, whereas the gene was found to be pseudo in four species based on the prediction of tRNAscan-SE. The tRNAscan-SE predicted trnT-GGU with mismatch isotypes of lysine along with truncated start and truncated end in Hypochaeris radicata L., pseudogene in Lactuca raddeana Maxim. (Fig. S3), and the gene was not predicted due to deletion events in Stebbinsia umbrella (Franch.) Lipsch. and Ixeris polycephala Cass. The structure of the species H. radicata and L. raddeana showed certain mismatches at the acceptor stem and specific variations in the variable loop (Fig. 3). In other species, the trnT-GGU gene was predicted with a low infernal score of 31 to 34.6 (Table S4).

Figure 3
figure 3

Structure of pseudo or low infernal score trnT-GGU gene in Asteroideae and Cichorioideae. The structure of the gene from ‘a’ to ‘f’ shows the species of Asteroideae, whereas from ‘g’ to ‘i’ represents species of Cichorioideae. (a and b) Pseudogenization of the gene occurred due to loss of the acceptor stem. (c and d) The genes are predicted with low infernal score (22.6) only by tRNAscan-SE and were not predicted by ARAGORN. However, the mismatch at 5′ and 3′ indicates that this gene may also be non-functional. (e and f) The gene of trnT-GGU predicted as mismatch isotypes for isoleucine. The clear insertion is visible in the acceptor stem, which disturbs the cloverleaf structure. (g) Pseudogenization of the gene occurred due to loss of the acceptor stem. (h) Indicates the tRNAscan predicted pseudo gene. (i) Indicates the tRNAscan predicted gene with low infernal score of 34.6. All the species show the mismatch of c–c, which forms an extra loop-like structure above the codon loop. * indicates loss of the acceptor arm, ** indicates mismatch at 5′ and 3′, + indicates the missing of base pair of uridines in the acceptor arm due to insertion.

Analyses of trnT-GGU genes among the species of Asteroideae

We analyzed 97 species belonging to 78 genera and 13 tribes of the Asteroideae subfamily. The analyses revealed that the trnT-GGU gene exists as a pseudogene in the species of all tribes (Table S4). This gene was not detected by ARAGORN in any species, whereas tRNAscan-SE did not predict this gene in 60 species. The trnT-GGU gene was predicted as a pseudogene in 9 species and as mismatch isotypes of isoleucine and lysine in 16 species (Table S2. We also detected this gene with a low infernal score of 22.6 in 12 species. However, the manual analyses of the structure revealed truncation at the 5′ and 3′ ends, indicating that the gene might also be non-functional in these species. The structure of representative species is shown in Fig. 3. The pseudogenization of the trnT-GGU gene occurs throughout the Asteroideae subfamily due to a high mutational rate (substitutions and insertion-deletion) in all functional parts of the genes. However, the highest mutations and degradation were recorded in the 5′ acceptor arm and the 3′ arm (Fig. S4). Pseudogenization occurred in the species of Asteroideae irrespective of the habit, habitat, and native range (Table S1, Table S2). We analyzed the trnT-GGU gene of 22 species of Artemisia L., 21 species of Aldama La Llave, and 25 species of Diplostephium Kunth to determine the extent of similarities and differences existing within this gene among closely related species (species of same genus). The analyses showed high similarities in the pseudogene of the trnT-GGU gene and fewer variations among species of the same genus (Figs. S5, S6, S7).

Codon usage analysis

The codon usage analysis of five representative species, four (Artemisia ordosica, Aster hersileoides, Symphyotrichum subulatum, and Helianthus annuus) of which represent species that lack the trnT-GGU gene or contained a pseudo copy, while Barnadesia lehmannii had a functional copy of trnT-GGU with a high infernal score of up to 65. The aforementioned species revealed high similarities in codon usage for amino acid threonine (Table S5). These findings showed that the pseudogenization of trnT-GGU did not cause any alteration in codons of protein-coding sequences and translates proteins similar to the species that have both functional tRNA.


The loss/pseudogenization of the trnT-GGU gene was investigated in 13 subfamilies of Asteraceae. Our findings show the loss/pseudogenization of the trnT-GGU gene in the species of the four subfamilies of Gymnarrhenoideae, Cichorioideae, Corymbioideae, and Asteroideae (collectively known core Asteraceae) based on the result of ARAGORN, whereas pseudogenization of the trnT-GGU gene was predicted in Gymnarrhenoideae, Cichorioideae, and Asteroideae based on the result of tRNAscan-SE. In addition, pseudogenization of the trnT-GGU gene was reported in previous studies, including the Cryptomeria japonica D. Don. of the family Cupressaceae28, Pelargonium × hortorum of the family Geraniaceae29, and Anaphalis sinica and Leontopodium leiolepis of the tribe Gnaphalieae of Asteroideae (Asteraceae)30. The species of the aforementioned subfamilies are diverse in terms of habit, habitat, and geographical distribution (Table S1). This demonstrates that pseudogenization is not linked to convergent evolution or environmental factors and a clade-specific event was found following clear phylogenetic patterns that agreed with the previously established phylogeny of the family Asteraceae1,4,30. The pseudogenization was linked to an insertion event in the 5′ acceptor stem. Earlier studies have demonstrated that insertions and deletions generate substitutions18,31,32,33 due to the recruitment of error-prone DNA polymerase34,35. Hence, this insertion may increase the rate of mutations of the trnT-GGU gene either causing pseudogenization of the gene by affecting the functional parts of the gene or leading to complete deletion of the gene. Previously, loss of the trnT-GGU gene was noted to be linked to an inversion event in Pelargonium × hortorum (Geraniaceae)29. A similar large inversion event (22.8 kb) has also been reported in Asteraceae, except for species of Barnadesioideae (earlier diverged clade)36. One endpoint of this inversion is located between trnS-GCU and trnG-UCC genes, whereas the other endpoint is present between trnE-UUC and trnT-GGU36. However, the absence of pseudogenization of trnT-GGU in species of subfamilies Famatinanthoideae, Stifftioideae, Mutisioideae, Gochnatioideae, Wunderlichioideae, Hecastocleidoideae, Pertyoideae, and Carduoideae reveals that the inversion event is not responsible for the pseudogenization of the trnT-GGU gene. Therefore, the insertion event may be responsible for pseudogenization and provide a plausible explanation for the pseudogenization of trnT-GGU.

Insertion-deletion and pseudogenization events are also considered important to gain insight into the evolutionary past19 and phylogenetic patterns37. Previously, the 9 bp deletion in ndhF gene was shared by three subfamilies of core Asteraceae, including Cichorioideae, Corymbioideae, and Asteroideae38,39, the 9 bp and 18 bp deletion in rpoB gene was shared by the six subfamilies Carduoideae, Pertyoideae, Gymnarrhenoideae, Cichorioideae, Corymbioideae, and Asteroideae40, and the 15 bp deletion in rpoB was shared by the seven subfamilies Hecastocleidoideae, Carduoideae, Pertyoideae, Gymnarrhenoideae, Cichorioideae, Corymbioideae, and Asteroideae40. All of these deletion events in protein-coding genes were used to gain insights into the phylogenetics of Asteraceae. Our result provides new support for the phylogenetic history and evolution of core Asteraceae based on specific insertion events and pseudogenization of the trnT-GGU gene, which is limited to core Asteraceae.

The loss/pseudogenization of the trnT-GGU gene was not reflected in codon usage analysis, which is in agreement with the previous report on Pelargonium × hortorum from the family Geraniaceae29. However, the conventional wobble rules described by Crick (1966)24 suggest the presence of 32 tRNA in the plastid genome and consider both threonine genes essential, trnT-GGU for decoding codons ACC and ACU of mRNA, and trnT-UGU for decoding codons ACA and ACG of mRNA. Therefore, the wobble rules cannot describe the adapted mechanism of the species of core Asteraceae by which they cover the deficiency of the trnT-GGU gene. The functional study of tRNAs indicates that 25 tRNA will be sufficient to decode all 61 codons by using superwobbling phenomena25 in which a single tRNA species containing an unmodified uridine in the wobble position of the anticodon can read an entire fourfold degenerate codon box25,41. The study of Alkatib et al.25 experimentally proved, based on knock-out mutants of tabacum chloroplast, that the trnT-UGU followed the superwobbling rule and degenerated all four codons of threonine, thus making trnT-GGU nonessential for translation of threonine codons. The deletion/pseudogenization of the trnT-GGU gene in the core Asteraceae, specifically subfamily Asteroideae (comprising about 17,000 species1) shows pseudogenization of this gene within the representative species of 13 tribes (from early diverged tribe Senecioneae to recently diverged tribe Eupatorieae), suggests that superwobbling may be responsible for the translation of threonine codons, as the species of core Asteraceae did not show any adverse events, and supports the findings of Alkatib et al.25 in naturally growing species.

In conclusion, the pseudogenization of the trnT-GGU gene occurred in core Asteraceae and is linked to the insertion event in the 5′ acceptor arm. The insertion event provides new insight into the evolution of core Asteraceae and broadens our knowledge of the evolution of the chloroplast genome in angiosperms. The codon usage analysis of the species indicates that superwobbling may be the universal phenomena in core Asteraceae by which they proceed to translate all four codons using only trnT-UGU in the absence of trnT-GGU.

Materials and methods

The complete chloroplast genome sequences of 124 species belonging to six subfamilies of Asteraceae were retrieved from the National Center for Biotechnology and Information (NCBI) (Table S1). These included the chloroplast genome sequences of 96 species of Asteroideae, 13 species of Cichorioideae, 11 species of Carduoideae, 2 species of Barnadesioideae, and 1 species each of Mutisioideae and Pertyoideae. The raw reads of 10 other species (Table S1) were retrieved from the Sequence Read Archive (SRA) to extract trnT-GGU gene. This enabled us to include the data of seven other subfamilies, including one species each of Gymnarrhenoideae, Corymbioideae, Famatinanthoideae, Hecastocleidoideae, and Stifftioideae and two species each of Gochnatioideae and Wunderlichioideae (Table S1). Moreover, the trnT-GGU gene of Chrysanthemoides incana (Asteroideae) was also extracted from raw reads to cover the tribe Calenduleae. The raw read of these species was retrieved and mapped to Silybum marianum (L.) Gaertn. (KT267161) in Geneious R8.142 using Medium–Low Sensitivity/Fast option, keeping all other parameters as default. The consensus was annotated and extracted after confirmation of mapping quality, specifically focusing on the trnT-GGU region. This approach enabled us to include diverse species in our study regarding geographical distribution, habit, and habitat (Table S1). We also retrieved chloroplast genome sequences of 25 species of Diplostephium, 22 species of Artemisia, and 21 species of Aldama to perform comparative analyses of the trnT-GGU gene at genus level in the Asteroideae subfamily (Table S6). The pseudogenization of the trnT-GGU gene was confirmed by reannotation of the trnT-GGU region by ARAGORN v.1.2.3843 and tRNAscan-SE v.2.0.744 whereas the infernal score was calculated for each tRNA using Infernal v.1.145 integrated in tRNAscan. The prediction of ARAGORN and/or tRNAscan-SE v.2.0.7 was recorded for each species.

The structural variations within trnT-GGU were analyzed by utilizing multiple alignment tool using clustalW46 integrated into Geneious R8.1 and inspected manually at the family, subfamily, and genus levels. To determine the position of pseudogenes on the phylogenetic tree, we drew a representative phylogenetic tree based on the previously reported data set of Panero et al.4 by running IQ-Tree with settings reported in Mehmood et al.11 while a high-quality representative tree was drawn using Integrative Tree Of Life (iTOL v.4.0)47.

We analyzed codon usage of protein-coding genes in five representative species to examine the effect of pseudogenization of trnT-GGU on the sequences of protein-coding genes. We included four species: Artemisia ordosica (contained tRNAs with mismatch isotypes of isoleucine with infernal score 27.8 instead of trnT-GGU gene), Aster hersileoides (copy of trnT-GGU gene not predicted by ARAGORN and tRNAscan-SE), Symphyotrichum subulatum (trnT-GGU gene predicted with infernal score of 22.6 and with loss of 5’ acceptor stem), and Helianthus annuus (trnT-GGU gene predicted as pseudo copy with infernal score 21.8), whereas Barnadesia lehmannii was selected from the subfamily Barnadesioideae, which showed the presence of the functional copy of the trnT-GGU gene with infernal score of 65.8.

Plant collection and deposition of voucher specimens to herbarium

The publicly available genomics sequences were taken from the National Center for Biotechnology Information. None of the plants was collected and sequenced in the current study. Hence, permission for plant collection and submission to herbarium under a voucher specimen are not applicable.