Introduction

Complete chloroplast genome sequences are available for various vascular and non-vascular land plants, green and red algae, and secondary algal lineages (Odintsova and Yurina 2003; Ohta et al. 2003; Sugiura et al. 2003; Wolf et al. 2003, 2005; Asano et al. 2004; Goremykin et al. 2004, 2005; Hagopian et al. 2004; Kim and Lee 2004; Shahid Masood et al. 2004; Kim et al. 2006; Pombert et al. 2005; Saski et al. 2005; Steane 2005; Turmel et al. 2005) (http://www.ncbi.nlm.nih.gov/genomes/ORGANELLES/plastids_tax.html). Among the dicots, complete chloroplast genome sequences are available for at least one representative from each subclass except Hamamelidae. Mulberry is a dicot (Magnoliopsida) belonging to the subclass Hamamelidae (Family: Moraceae; Order: Urticales) (http://plants.usda.gov/), which is the smallest subclass (in terms of taxa) among the dicots. The plastome sequence of mulberry, therefore, represents the first case study from this subclass. Mulberry (Morus indica) is a perennial tree or shrub cultivated extensively in East, Central, and South Asia for silk production; its foliage is the sole food for the domesticated silkworm (Bombyx mori). Sixty-eight species of mulberry are found mostly in Asia, mainly China and Japan, and continental America. They are poorly represented in Africa and Europe and virtually absent in Australia. Besides silk production, mulberry is used for its medicinal properties (foliage), as fodder (foliage), as a source of firewood (pruned shoots), for preparation of jams (fruit) and also as a shade plant for intercropping between rows of other plants like tea and coffee. Although there are a few reports on phylogenetic studies involving mulberry, these are restricted to only a few genes. A complete repertoire of genes would thus help in establishing the position of mulberry in the tree of life. We present in this paper the complete organization of the mulberry chloroplast genome along with some interesting comparative analyses with other known plastid genomes.

Materials and methods

Chloroplast DNA was isolated from mature green leaves kept in darkness for 48–72 h by the sucrose gradient method (Palmer 1986) followed by purification by cesium chloride–ethidium bromide density gradient centrifugation. As chloroplast DNA could not be separated from the nuclear DNA on the gradient, a long PCR-based strategy was employed to amplify the DNA fragments of interest. Primers were designed from conserved regions of genes from complete chloroplast genomes using Gene Runner (version 3.05) (http://www.generunner.com). Multiple alignments were performed using the ClustalW (Thompson et al. 1994) component of BioEdit sequence alignment editor (Hall 1999). PCR products were generated using the Eppendorf Triple Master PCR System. Products were gel-purified using the QIAquick gel extraction kit (QIAGEN). Products above 3 kb were sheared using HydroShear™ (GeneMachines) and the end-repaired fragments cloned in linearized and dephosphorylated (SmaI and bacterial alkaline phosphatase-treated) pUC19 vector (MBI Fermentas).

Smaller products were sequenced by primer walking. Sequencing reactions were performed with Big Dye Terminators™ (ver. 3.1, Applied Biosystems) and products analyzed on ABI Prism 3700 automated sequencers. Vector trimming, assembly, and editing of sequences was done using Phred-Phrap (Ewing and Green 1998) and Consed (Gordon et al. 1998). Annotation of protein-coding, rRNA and tRNA genes was done by Dual Organellar GenoMe Annotator (Wyman et al. 2004) (http://bugmaster.jgi-psf.org/dogma/).

Further confirmation of gene predictions, gene locations, intron–exon boundaries, locations of single-copy regions, inverted repeats, and IR expansion/contraction was done using different programs of NCBI BLAST (Altschul et al. 1990) (http://www.ncbi.nlm.nih.gov/BLAST/). All plastid sequences were obtained from NCBI except Populus (available at http://genome.ornl.gov/poplar_chloroplast/). Global alignments were performed using VISTA (Mayor et al. 2000), which uses AVID (Bray et al. 2003) as the alignment program. Global alignment-level phylogenetic tree was generated using MAVID/AMAP (Bray and Pachter 2004). The tRNAs were confirmed using tRNAscan-SE ver. 1.21 (Lowe and Eddy 1997). Phylogenetic analysis was done using TREECON version 1.3b (Van de Peer and De Wachter 1994), Tree-Puzzle 5.1 (Strimmer and von Haeseler 1996) (available online at http://bioweb.pasteur.fr/intro-uk.html) and PHYLIP (the PHYLogeny Inference Package, v3.6; Felsenstein 1989). Neighbor joining (NJ) tree was created using distance calculation based on Kimura (1983) and Tajima and Nei (1984) as implemented in the TREECON package. Dayhoff et al. (1978) and Jones–Taylor–Thornton (JTT; Jones et al. 1992) substitution models were employed for phylogenetic reconstruction in case of Tree-puzzle. Protein parsimony method was also tested using PROTPARS from the Phylip (Felsenstein 1989) package.

Results

Plastome organization

The chloroplast genome of Morus indica (GenBank Accession Number DQ226511) is a circular double-stranded DNA of 158,484 bp with an overall A+T content of 63.63%. The plastome harbors a pair of identical inverted repeat regions (IRA and IRB), which are 25,678 bp each. The inverted repeats are separated by a large single-copy (LSC) region of 87,386 bp and a small single-copy (SSC) region of 19,742 bp. The positions of all the genes identified in the mulberry plastome and category-wise distribution of these genes are presented in Fig. 1 and Table 1, respectively. More than half of the plastome is composed of coding regions (90,532 bp; 57.12%) with the peptide-coding regions accounting for the major portion (78,681 bp; 49.65%) followed by ribosomal RNA genes (9,050 bp; 5.71%) and transfer RNA genes (2,801 bp; 1.77%). The remaining 42.88% (67,952 bp) is covered by intergenic regions (46,923 bp; 29.61%) and a total of 20 introns (21,029 bp; 13.27%) present within 18 genes (counting the IR genes only once) (see Supplementary Table S2).

Fig. 1
figure 1

Organization of the mulberry chloroplast genome. The IRA/LSC junction (JLA) represents the start of the genome (base position 1). Genes drawn outside and inside the circle represent those on the positive and negative strands, respectively. Genes joined by dotted lines represent intron-containing genes. Different categories of genes are color-coded. *intron containing gene, **genes showing trans splicing

Full size image
Table 1 Genes present in the mulberry plastome
Full size table

Codon usage

The codon usage (Supplementary Table S3) of the mulberry plastome strongly reflects the AT bias. Majority of the codons (70.75%) end in A or T. Even the stop codons are biased with 73.49% ending in A or T. The A+T content of the IR region is the least and amounts to 57.08% reflecting the low A+T content of the ribosomal RNA genes (44.62%), whereas the A+T content in the LSC and SSC regions is 65.88 and 70.65%, respectively. There are in total 26,179 codons, which represent the total coding capacity of the mulberry plastome; 2,803 (10.71%) of these are for amino acid leucine, 2,281 (8.71%) for isoleucine, 1,990 (7.6%) for serine, and 1,750 (6.68%) are for glycine. One-third of the total codons are represented by these four amino acids. Again, most of the codons end in A or T for these four amino acids (66.82% for leucine, 80.62% for isoleucine, 68.54% for serine, and 74.51% for glycine). The least number of codons (excluding stop codons) are for cysteine (307; 1.17%) and these too are with an AT bias (76.87% ending in T).

Comparison with other plastomes

The gene content, order, and organization of the Morus indica chloroplast genome is similar to that of higher plants except for the inverted repeat/single copy (IR/SC) junction regions (see IR expansion/contraction section). Genes absent in the mulberry plastome are infA (translation initiation factor 1), sprA (small plastid RNA), and rpl21 (ribosomal protein large subunit 21). Two non-functional genes ycf15 and ycf68 (Supplementary Table S4), are also present in the mulberry plastome. The ycf15 gene is 5′-truncated with respect to Nicotiana, Atropa, Panax, and Nymphaea genes (determined by TBLASTN). When compared to the Panax and Nymphaea, the reading frame of mulberry reveals stop codons and an insertion corresponding to 20 amino acids (Supplementary Fig. S1c). The ycf68 gene on the other hand is a non-truncated pseudogene in mulberry having accumulated stop codons in its reading frame. Upon close scrutiny, it was observed that in maize and rice, there are two ‘AAAC’ units one after another in a region before the first stop codon of mulberry. One of the ‘AAAC’ units was missing from the mulberry ycf68 gene, which was causing the frameshift and the resulting stop codons. When the four bases were added, the frame was restored (Supplementary Fig. S1a,b). In Pinus, although four bases ‘CAAA’ are missing in the same region, an additional stretch of four bases ‘TGTG’ restores the frame and, hence, does not have stop codons.

Eighteen genes in the mulberry plastome contain one or two introns. This is identical to the Panax, Cucumis, and Calycanthus plastome. In comparison, there are 17 intron-containing genes in the Spinacia plastome and 15 in the tobacco plastome. Supplementary Table S2 summarizes the sizes of exons and introns for each gene. Five of these introns, rpl2, ndhB, rps12, trnI-GAU, and trnA-UGC, are located within the IR regions. Genes clpP and ycf3 have two introns each. A total of 30 tRNA gene species coding for all 20 amino acids were identified from the M. indica chloroplast genome. The numbers and kinds of tRNA genes from M. indica are identical to well characterized vascular plant chloroplasts. The codon usage of the M. indica chloroplast genome and the anticodons present in the 30 tRNA species are summarized in Supplementary Table S3.

IR expansion/contraction

The borders between the two inverted repeats (IRA and IRB) and the two single-copy regions (LSC and SSC) usually differ in plastomes of various species. The length variations in chloroplast genomes of different plant groups are often due to expansions and contractions in the IR regions. These expansions and contractions result in the truncation of genes at or near the boundaries (e.g., rps19 and ycf1). Comparisons of IR boundaries reveal that these changes are not uncommon in higher plants. Figure 2 shows the detailed view of the IR/SC border positions with respect to adjacent genes of Morus, Eucalyptus, Calycanthus, Panax, Nicotiana, Atropa, Spinacia, and Arabidopsis. A close-up of the border regions reveals differences even between Nicotiana and Atropa—members of the same family.

Fig. 2
figure 2

Detailed view of the inverted repeat-single copy (IR/SC) border regions from various dicot species with respect to the genes located at or near the boundaries. Gene names suffixed by a single quotation mark represent pseudogenes. The figure is not to scale and only shows relative changes at or near the IR-SC boundaries. Mi: Morus indica, Nt: Nicotiana tabacum, Ab: Atropa belladonna, At: Arabidopsis thaliana, Cf: Calycanthus fertilis, Eg: Eucalyptus globulus, Pg: Panax ginseng, So: Spinacia oleracea

Full size image

The IRA/SSC borders are located in the 3′ region of the ycf1 gene and create the ycf1 pseudogenes at the IRB/SSC border with lengths of 996 bp and 1,438 bp in Nicotiana and Atropa, respectively. The same is the case with Morus where a ycf1 pseudogene of 1,001 bp is produced at the IRB/SSC border. The situation is, however, different in the sense that this ycf1 pseudogene has an overlap of 25 bp with the ndhF gene. The IRB/SSC border passes through the ndhF gene in Morus just as it does in Arabidopsis. The IR/LSC borders, unlike their SSC-counterparts, show much less variation between different genera. The IRA/LSC borders are usually located downstream of the non-coding region of trnH-GUG. The IRB/LSC borders are either located upstream of the non-coding region or within the coding region of rps19.

In Morus, a single base pair separates the rps19 gene from the IRB/LSC border and thus the gene marks the end of the LSC. Thus, no rps19 pseudogene is created at the other border. A similar situation is seen in Nicotiana, where 4 bp separate the rps19 from the IRB/LSC border, and in Eucalyptus, where there are 6 bp in between. For other genera the case is different and the IRB/LSC borders pass through the rps19 gene to create rps19 pseudogenes of lengths 59 bp, 113 bp, 51 bp, and 143 bp in Atropa, Arabidopsis, Panax, and Spinacia, respectively. On the other extreme is Calycanthus, where the rps19 gene is separated from the IRB/LSC border by 1,552 bp. The expansions/contractions of IR, as observed in the IR/SSC borders, are probably mediated by gene conversion and recombinational repair of double-strand breaks (Goulding et al. 1996).

Comparison by means of global and local alignments

Comparison of the mulberry plastome sequence with other plastomes at the global level using VISTA brought out Eucalyptus (a tree species) as the closest relative, followed by Panax, Populus (another tree species), Cucumis, Nicotiana, and Atropa. Interestingly, Acorus, which is thought to be one of the ancient lineages of monocots, appears to be closer to the dicots. Morphological similarities with dicots have already been reported for Acorus (Grayum 1987), and the VISTA alignments further reinforce the fact. A similar kind of result was obtained when concatenated local alignments obtained by BLAST were compared (Fig. 3).

Fig. 3
figure 3figure 3

Plot of concatenated local alignment lengths and identities of several dicot, monocot, and lower plant groups compared to mulberry plastome (a). A cut-off of 90% identity was used. VISTA plot (global alignment) comparison of mulberry chloroplast genome with 25 chloroplast genomes (b). Y-scale represents the percent identity ranging from 50 to 100%. Genomes are arranged according to the number of conserved bases with respect to mulberry. Double asterisk: Acorus is a monocot but shows more similarity to dicots in terms of global alignment

Full size image

In this case also, Eucalyptus came out as the closest relative of Morus, followed by Cucumis, Panax, Populus, Nicotiana, and Atropa. Acorus, again, was found to be close to the dicot cluster. Both VISTA and local alignment comparisons clearly bring out three distinct groups—dicots, monocots, and lower plants (including algae). The vista-plot patterns produced are remarkably group-specific and each group shows nearly identical patterns among themselves. The gap regions in Oenothera and Lotus represent the ∼54 and ∼51 kb inversions, respectively (Hupfer et al. 2000; Kato et al. 2000) in the large single-copy regions of both the plastomes. A similar gap (∼28 kb) representing a large single-copy inversion is seen in the monocots (Doyle et al. 1992).

Phylogenetic positioning of mulberry

Amino acid sequences from 43 protein-coding genes common to 26 genera were used to create a concatenated data set. The total alignment length was 12,299 positions long. Another alignment of 9,727 positions was created after removal of gaps. Similar topologies were obtained with distance-based, maximum parsimony and maximum likelihood approaches using the models specified in the methods section (Fig. 4a–c). In all the cases, mulberry paired up with Cucumis, which was highly similar at the global and local alignment levels, too. Bootstrap values of 100 and 53 were obtained in case of NJ and ML trees, respectively. With PROTPARS, the 9,727-position-long alignment gave 44 and 46 trees out of a total of 100 having Morus/Cucumis and Morus/Lotus pairs, respectively (almost 1:1). The remaining ten trees had Lotus/Cucumis or Lotus/Arabidopsis pairs. The 12,299-position-long alignment, however, gave a very strong support for Morus/Cucumis (89 out of 100 trees).

Fig. 4
figure 4

Phylogenetic position of Morus as inferred from 43 chloroplast proteins. The same topology was obtained with both Neighbor-joining method (a) (Kimura, Tajima-Nei, and Poisson correction distance calculation as implemented in TREECON); Maximum Likelihood method (b) (Dayhoff and Jones–Taylor–Thornton models of substitution as implemented in Tree Puzzle 5.1) and PROTPARS (c) (Protein parsimony method: PHYLIP package, Joseph Felsenstein). A consensus tree is shown in the figure from the 12,299-position-long alignment. Trees with Morus/Cucumis pairs were 89 out of a total of 100 trees. Phylograms are displayed for a and b, while c is shown in the form of a dendrogram with color-coded groups. d Tree topology obtained with the 12,299-position-long alignment (gapped) using ProML with the JTT model as implemented in BioEdit. Bootstrap values for support are indicated at branch-points. The scale represents the number of substitutions per site. Following are the genes used in the analysis, with the lengths of alignment used in parentheses: atpA (512), atpB (502), atpE (143), atpF (187), atpH (82), atpI (251), petB (234), petG (38), psaA (754), psaB (735), psaC (81), psaI (52), psaJ (57), psbC (487), psbD (354), psbE (83), psbF (41), psbH (88), psbI (54), psbJ (42), psbK (65), psbN (46), psbZ (ycf9) (62), rpl2 (279), rpl14 (123), rpl16 (143), rpl20 (142), rpl36 (38), rpoA (533), rpoB (1153), rpoC1 (904), rpoC2 (2041), rps2 (317), rps3 (257), rps4 (211), rps7 (157), rps8 (141), rps11 (147), rps12 (138), rps14 (103), rps18 (171), rps19 (95), ycf4 (256)

Full size image

There was only one tree with Morus/Lotus pairing, and the remaining ten trees had Lotus/Arabidopsis pairs. In all the distance-based approaches, Lotus appeared sister to Morus/Cucumis. Eucalyptus, which was closest to mulberry in terms of global and local alignments, however, grouped with Oenothera in all phylogenetic analyses. This is no surprise as both belong to the same order Myrtales. Acorus, the ancient Liliopsid, grouped with the ‘basal angiosperms’ and paired up with Calycanthus in all cases. The monocots appeared as the sister group to all the angiosperms, which in itself is highly controversial (Goremykin et al. 2003a,b–2005; Soltis et al. 2004; Stefanovic et al. 2004; Martin et al. 2005). Monocots did not appear to be monophyletic in all the above cases. However, when protein maximum likelihood program ProML (Felsenstein 1989) employing the JTT model, as implemented in BioEdit (Hall 1999), was used for phylogeny construction, monophyly for the monocots was observed with Acorus as the sister genus to the rest of the monocots (Fig. 4d). In the tree obtained from the 9,727-position-long alignment (ungapped), Nymphaea appeared as the basal angiosperm, next to Amborella, while in the other tree, obtained with the 12,299-position-long alignment (with gaps), Amborella-basal topology was obtained. However, Morus paired up with Lotus in both these cases, Cucumis being sister to Morus/Lotus. When ProtML with JTT model (Molphy, Adachi, and Hasegawa, http://bioweb.pasteur.fr/intro-uk.html) was used with the 12,299-position-long alignment, 34 out of a total of 47 (72.3%) trees gave Morus/Cucumis pairs (tree not shown).

The remaining trees comprised of Lotus/Morus pairs (10) and Cucumis/Lotus pairs (3). The 9,727 (ungapped)-position-long alignment gave a much stronger support to the Morus/Cucumis pair (91.8% trees; 45 out of a total of 49). Only one tree had a Morus/Lotus pair and the remaining three trees had Lotus/Cucumis pairs. Thus, Morus appears to be closer to Cucumis phylogenetically, although Lotus cannot be ruled out. If we take both the genome level and phylogenetic comparison, then Cucumis appears to be the closest. The dendrogram (Fig. 4c) obtained using PROTPARS clearly shows different groups, namely, the dicots, monocots, basal angiosperms, and lower plants. Even among the dicots, there is a clear distinction between the rosid and asterid groups. Finally, a global alignment-level phylogenetic tree was generated using the MAVID/AMAP (Bray and Pachter 2004) multiple-alignment server (http://baboon.math.berkeley.edu/mavid/) and viewed using the ATV applet. Morus paired up with Cucumis in this case, too (Supplementary Fig. S2).

Discussion

The complete chloroplast genome of mulberry has been determined using a combination of long PCR and shotgun-based approaches using purified chloroplast DNA as a template. The long PCR approach has been used to determine the entire plastome sequence in other organisms like sugarcane (Asano et al. 2004), Calycanthus (Goremykin et al. 2003a), Amborella (Goremykin et al. 2003b), Nymphaea (Goremykin et al. 2004), and Acorus (Goremykin et al. 2005) using total DNA as the template. Our approach uses DNA isolated from a chloroplast-enriched preparation, thus making the amplifications even more reliable and decreasing the chances of nuclear-localized plastid DNA (nuptDNA) getting amplified. The chloroplast genome of mulberry is highly similar in organization, gene content, order, and A+T content to other known land plant plastomes (Shinozaki et al. 1986; Sato et al. 1999; Kato et al. 2000; Schmitz-Linneweber et al. 2001, 2002; Odintsova and Yurina 2003; Steane 2005).

The organization of the rps12 gene exhibits signatures of trans-splicing. To produce mature rps12 transcript, the 5′-exon present in the large single-copy region and the 3′-exons present in the inverted repeats as duplicates should splice. This discontinuous arrangement of the first exon with respect to the 3′ exons is typical of plant chloroplasts. Unlike tobacco, the mulberry plastome does not contain a small plastid RNA-encoding gene sprA. This gene, believed to be involved in 16S rRNA maturation (Vera and Sugiura 1994), has only been found in plastomes of Solanaceae members. Other genes absent from the mulberry plastome are infA and rpl21.

The infA gene is known to have been lost from almost all known rosid plastomes (Millen et al. 2001) having been transferred to the nucleus. The rpl21 gene is present only in plastomes of ferns and bryophytes. The mulberry plastome also contains two pseudogenes ycf15 and ycf68. The presence of portions of the ycf15 gene indicates that it is probably a remnant of a functional gene in one of its predecessors. It is believed that ycf15 is not a protein-coding gene (Schmitz-Linneweber et al. 2001; Steane 2005). The deletion observed in the ycf68 gene, which causes the frameshift, does not appear to be a sequencing problem, as the coverage and read quality in the concerned region are high.

Moreover, the reads in this region are of two types—one from a whole chloroplast genome shotgun and the other from a PCR-based approach, indicating that the deletion is part of the plastome and not an artifact. If these highly conserved genes are essential for the organism, then it is possible that they have a counterpart in the nuclear genome and are in different stages of degeneration. If not, the conservation of these sequences might be signatures of regulatory regions. C to U transitions in chloroplast genomes are known, but the same is not true for reverse editing (U to C). This phenomenon has been observed in mitochondria and only in the plastome of Anthoceros (Kugita et al. 2003a,b). Thus, it does not seem likely that the stop codons get converted to sense codons.

IR expansion/contraction studies in Atropa and Nicotiana (Kim and Lee 2004) reveal that there is considerable difference even between members of the same family. However, it would be of great interest to compare IR/SC junctions in different varieties of Morus along with other members of Moraceae. Similarity at this level could indicate common ancestry and will be a useful evolutionary tool besides the conventional ones.

Plastome comparison at the global and local level brought out a tree species, Eucalyptus, as the closest out of the 25 genera taken for comparison. Cucumis and Panax were the next in line. However, phylogenetic analysis using amino acid sequences from common protein coding genes brought out Cucumis as the closest followed by Lotus. This was confirmed by several phylogenetic methods, and similar results were obtained with global-alignment level phylogenetic analysis also. Thus, Cucumis seems to be the closest to mulberry if results of both methods are combined, i.e., global/local comparison and phylogeny.

In summary, the complete plastome sequence of an economically important plant—mulberry—has been determined, thus filling in the gap representing the sixth subclass of the Magnoliopsida. Global- and local-alignment-based comparison and phylogenetic positioning of mulberry have been presented. It appears to be closest to Cucumis and Lotus phylogenetically, but at the genome level, Eucalyptus appears to be closest. This may be attributed to the non-coding portions, which were not part of phylogenetic analysis. The difference in the results obtained from the gene-level and genome-level studies are probably due to the differences in the selection pressures operating on the coding and non-coding regions. This is the third angiosperm tree species after Populus and Eucalyptus, whose plastome sequence has been completely deciphered. Studies based on IR/SC junction regions and other variable regions from different Morus species would be of great help in systematics. The information thus generated will be also useful for taxonomic analyses of other species of Morus, other genera within Moraceae, and other families within the same subclass.