Evolutionary hierarchies of conserved blocks in 5'-noncoding sequences of dicot rbcS genes
© Weeks et al; licensee BioMed Central Ltd. 2007
Received: 02 October 2006
Accepted: 02 April 2007
Published: 02 April 2007
Evolutionary processes in gene regulatory regions are major determinants of organismal evolution, but exceptionally challenging to study. We explored the possibilities of evolutionary analysis of phylogenetic footprints in 5'-noncoding sequences (NCS) from 27 ribulose-1,5-bisphosphate carboxylase small subunit (rbcS) genes, from three dicot families (Brassicaceae, Fabaceae and Solanaceae).
Sequences of up to 400 bp encompassing proximal promoter and 5'-untranslated regions were analyzed. We conducted phylogenetic footprinting by several alternative methods: generalized Lempel-Ziv complexity (C LZ ), multiple alignments with DIALIGN and ALIGN-M, and the MOTIF SAMPLER Gibbs sampling algorithm. These tools collectively defined 36 conserved blocks of mean length 12.8 bp. On average, 12.5 blocks were found in each 5'-NCS. The blocks occurred in arrays whose relative order was absolutely conserved, confirming the existence of 'conserved modular arrays' in promoters. Identities of half of the blocks confirmed past rbcS research, including versions of the I-box, G-box, and GT-1 sites such as Box II. Over 90% of blocks overlapped DNase-protected regions in tomato 5'-NCS. Regions characterized by low C LZ in sliding-window analyses were also frequently associated with DNase-protection. Blocks could be assigned to evolutionary hierarchies based on taxonomic distribution and estimated age. Lineage divergence dates implied that 13 blocks found in all three plant families were of Cretaceous antiquity, while other family-specific blocks were much younger. Blocks were also dated by formation of multigene families, using genome and coding sequence information. Dendrograms of evolutionary relations of the 5'-NCS were produced by several methods, including: cluster analysis using pairwise C LZ values; evolutionary trees of DIALIGN sequence alignments; and cladistic analysis of conserved blocks.
Dicot 5'-NCS contain conserved modular arrays of recurrent sequence blocks, which are coincident with functional elements. These blocks are amenable to evolutionary interpretation as hierarchies in which ancient, taxonomically widespread blocks can be distinguished from more recent, taxon-specific ones.
Promoter sequences have been described as a vast and largely uncharted territory for evolutionary biologists . One impediment to exploration is the difficulty of motif prediction in noncoding sequences (NCS): motif-discovery tools achieved detection rates of only 22–35% for transcription factor (TF) binding sites in recent benchmark studies [2, 3]. Although it has long been recognized in principle  that evidence for motifs can be enhanced by comparing sequences of common ancestry, 'phylogenetic footprinting' of higher eukaryotes is still in a development and evaluation phase [5–8]. There are also perceived challenges in the use of sequence alignment for phylogenetic analysis of NCS , as complex mutational processes (slipped-strand mispairing, stem-loop secondary structure excision/repair, minute inversions, intramolecular recombination) are prevalent. In practice, however, Bremer et al.  found chloroplast NCS to be of similar utility to coding sequences in phylogenetic tree construction for asterids. This result confirmed that plant NCS contain evolutionary signal, which might be hypothesized to reside in the conserved motifs sought in phylogenetic footprinting. The present study sought to explore the extent to which phylogenetic footprints in plant 5'-NCS could be subjected to evolutionary analysis and interpretation. For this objective, we needed to conduct sufficiently comprehensive phylogenetic footprinting for meaningful evolutionary analysis of conserved sequence blocks.
We employed a greater taxonomic range than other phylogenetic footprinting studies of plant NCS, which have been confined to single families [6, 8, 11, 12] or to a couple of species . Much of the interest in promoter evolution lies in comparisons of paralogous genes (i.e. genes that diverged after a duplication event). In consequence, it must be noted, our dataset included several multigene families, and therefore was not optimized to investigate taxonomic phylogenies in the manner of Bremer et al. .
Recognizing limitations in individual motif discovery tools [2, 3, 7], we sought to maximize detection of conservation by combining distinct methodologies. Analysis of generalized Lempel-Ziv complexity (C LZ ), played several roles in our study. C LZ measures the complexity of a text as the minimal number of steps in a defined procedure of its synthesis with the parsing rule: the next phrase is the longest seen previously. Many text compression algorithms are based on Lempel-Ziv parsing . Computation of C LZ thus involves a decomposition of the text into repeated blocks, and an application to the discovery of structural regularities in genetic 'texts' was realized by Gusev et al. . This method has identified arrays of conserved sequence blocks in NCS of vertebrates from fish to humans [16, 17]. C LZ analysis has also been used to study human mutagenic mechanisms [18, 19] and genomic architecture .
Our second tool was MOTIF SAMPLER, in which the probability of finding a particular motif is estimated using Gibbs sampling and modelling of the background sequence with a Markov model .
We complemented these tools with sequence alignment, including the DIALIGN and ALIGN-M algorithms designed for highly divergent sequences with only localized similarities, as seen in 5'-NCS. DIALIGN is based on a segment-to-segment comparison [22, 23], while ALIGN-M uses a non-progressive local approach to guide alignments .
We focused on 5'-NCS of ribulose-1,5-bisphosphate carboxylase small subunit (rbcS) genes because of the exceptional corpus of knowledge against which analytical outcomes could be benchmarked. As the earliest nuclear protein-coding sequences in plants to be cloned, rbcS genes became paradigms for functional studies of plant promoters , and several classes of cis-elements were originally defined in rbcS promoters. Thus, the prototype of trihelix TFs was the nuclear protein GT-1, which binds to the 14-bp Box II and related motifs in light-responsive regions of the pea rbcS-3A promoter . Box II versions featured in the earliest rbcS promoter alignments [25, 27], and occur in other light-responsive genes , where they may be targets of calcium/calmodulin phototransduction .
Two further cis-elements discovered in rbcS promoters, the G-box and I-box, are common features in light-responsive promoters [28, 30], and have been functionally characterized as dual components of a minimal light-responsive unit . G-box binding factors (GBFs), identified using tomato rbcS-3A upstream sequences [32, 33], are basic leucine zipper TFs interacting with the G-box core, CACGTG . Dicot rbcS G-boxes interact with the HY5 GBF, which mediates phytochrome and cryptochrome signals in concert with COP and DET regulators [31, 35].
The I-box, core motif GATAAGR, was also defined in rbcS promoters [27, 33, 36]. Its reverse, YCTTATC, was highlighted in rbcS and other light-regulated promoters by early motif searches [37, 38]. Binding factors for the I-box are still being clarified. Functional interactions occurred in yeast between I-box sequences and recombinant zinc-finger GATA TFs from Arabidopsis . I-box binding nuclear proteins reported in several species [40, 41] may therefore include GATA TFs, though the first cloned I-box binding protein was a tomato Myb-like TF . While the above rbcS cis- elements are the most studied, there is evidence for numerous further elements and DNA-protein interactions in rbcS promoters [30, 32, 43–48].
There is a particularly extensive history of characterization of rbcS promoters from pea, Petunia and tomato [25, 49]. We analyzed these along with other studied rbcS 5'-NCS such as those of Arabidopsis  to provide a gradation of taxonomic relations and evolutionary distances. Conserved features shared by the plant families analyzed would have persisted since the Cretaceous, to which can be dated the divergence of eurosids I (represented by the Fabaceae) from eurosids II (Brassicaeae), and both from asterids (Solanaceae) .
5'-NCS of up to 400 bp including proximal promoter and 5'-untranslated regions (5'-UTRs) were analyzed for 27 dicot rbcS genes. The rosid complement comprised all four Arabidopsis genes (three from a tandem locus), plus genes from Brassica and the legumes Phaseolus, Medicago and Pisum (pea). The Lycopersicon (tomato), Solanum (potato), Petunia and Nicotiana genes included representatives of all three solanaceous rbcS loci, which are distinguished by features including an extra (third) intron in 'locus 2' genes, and tandem duplicates at 'locus 3' . Phylogenetic footprinting analyses were performed on the entire dataset, and separately on various subgroups, e.g. rosid, brassica, legume, or solanaceous genes, or genes of each solanaceous locus. Three methodologies were employed:
(1) C LZ analysis was used as proposed by Gusev et al.  to search for recurrent sequence blocks in the rbcS 5'-NCS. The C LZ measure is based on representation of a sequence by fragments that have been encountered before (in the same or other sequences). Let S = s1 ... s L be a nucleotide sequence of length L. Denote by S [i:j] the substring of S that starts at position i and ends at position j. A Lempel-Ziv decomposition of S is a partition of S into m consecutive fragments, S = S [1:i1] S [i1+1:i2]...S [im-1:L], such that the k-th component S [ik-1+1:i k ] is the longest fragment downstream of position ik-1for which an exact repeat has been encountered somewhere upstream of position ik-1+1. The number of fragments in the decomposition, C LZ (S) = m, is called the complexity of S with respect to direct repeats. For example, if S = TCGATCGAGAT, then the decomposition of S with respect to direct repeats is T-C-G-A-TCGA-GAT. Fragments 1, 2, 3 and 4 in this decomposition are of length one since respective nucleotides T, C, G and A occur for the first time. Exact copies of fragments 5 and 6 occur in positions 1 and 3 respectively. The C LZ of the sequence with respect to direct repeats equals 6. To find fragments repeated in different rbcS 5'-NCS, we concatenated multiple sequences for C LZ analysis.
Conserved Blocks in rbcS 5'-NCS
rbcS Box III* 
(?)PHR1 site; (?)MYB site
rbcS I-box ; rbcS-CMA5 
rbcS I-box ; Manzara 5 
rbcS G-box ; rbcS-CMA4/5 
rbcS-CMA4 ; Manzara 4/8 
rbcS I-box ; rbcS-CMA4 
(?)RAV1-A site; (?)Dof site
Manzara 6 ; (?)GT1-core 
rbcS Box II ; rbcS-CMA3 ; Manzara 8 
(?)ACGT element; (-)I-box 
rbcS 3AF-5 site ; rbcS-CMA3 
Manzara 9 ; (?)(-)W-box
rbcS Box III ; rbcS-CMA3 
rbcS-CMA2 ; Manzara 10 
rbcS-CMA2 ; Manzara 12–13 ; CAAT-box
rbcS-CMA2 ; Manzara 15 
rbcS-CMA1 ; (?)(-)CCAAT-box
Manzara 16 ; (?)MYB site
rbcS-CMA1 ; LRE 
rbcS-CMA1 ; TATA-box 
(?)W-box; (?)MYB site
rbcS Box V ; (?)(-)Dof site
(3) Sites of local congruence were sought in multiple sequence alignments produced by CLUSTALW, ALIGN-M and DIALIGN, with various gap penalty options for the first two. Collation of methodologies by mapping output from C LZ and Gibbs sampling analyses onto alignments yielded useful synergies. In particular, the alignments revealed arrays of blocks that occurred in several sequences in the same order, which increased confidence in less conserved block versions that occurred in the appropriate position relative to other blocks.
Our initial C LZ procedure specified blocks ≥ 8 bp with up to two mismatches, which identified 218 instances of 34 conserved blocks (on average 90% identical with their definitions in Table 1). Relaxation of the mismatch criterion for DIALIGN-aligned versions of these 34 C LZ -defined blocks exposed an additional 109 instances (of average 76% identity with definitions).
Conversely, mapping blocks from other tools clarified often complex alignments. When the full dataset was aligned by DIALIGN, 67% of aligned blocks split into an average 3.5 fragments, and 86% of blocks were co-aligned on average with 1.7 others. Nonetheless, with support from C LZ and MOTIF SAMPLER, 323 instances of 35 blocks were identified within alignments. MOTIF SAMPLER used independently found 291 instances of 35 blocks.
Sensitivity = nTP/(nTP+nFN)
Positive Predictive Value (PPV) = nTP/(nTP+nFP)
where nTP = number of 'true' positives (identified blocks found also by other tools), nFN = 'false' negatives (blocks not found though supported by other tools), and nFP = 'false' positives (blocks found but not supported by other tools). (Since every block instance had not been verified as a cis-element, the 'true' and 'false' concepts in these equations reflected sequence analysis performance rather than functionality prediction.)
C LZ analysis and MOTIF SAMPLER showed greater PPV in block prediction, but weaker sensitivities, than the best alignment methods (Figure 1). MOTIF SAMPLER's sensitivity for individual blocks correlated (r = 0.85, P < 0.001) with its log-likelihood statistic  that is optimized during Gibbs sampling. Among the alignment tools, DIALIGN and ALIGN-M, designed for highly divergent sequences with localized alignments, outperformed the CLUSTALW global alignment algorithm (Figure 1). The performance of CLUSTALW was significantly improved by reducing the gap penalties, though the PPV of DIALIGN and ALIGN-M remained superior (Figure 1).
MOTIF SAMPLER outputs statistical data, which helped estimate the significance of our phylogenetic footprinting results. Ten dummy datasets with different randomizations of every sequence were analyzed by MOTIF SAMPLER using background model orders 0–3. Randomization caused MOTIF SAMPLER to find on average 6.5-fold fewer pseudo-motif instances. Log-likelihood scores  for pseudo-motifs in the 10 dummy datasets were much lower (mean = 49.6, standard deviation = 17.3) than those of the original sequence motifs (mean = 188.7, standard deviation = 60.2), which differed from random at significance levels of P < 0.0001 (Kruskal-Wallis tests).
In summation, phylogenetic footprinting defined 36 conserved blocks, representing contiguous nucleotide sequences occurring in two or more rbcS 5'-NCS, and being of sufficient length, sequence fidelity and positional similarity to make their common evolutionary origin probable. A total of 338 instances of these blocks were identified in the dataset. A large majority (275 instances of 33 blocks) were supported by all three methodologies. Of these 33 blocks, another 37 instances were supported only by C LZ and alignments, and 5 more only by C LZ and MOTIF SAMPLER. Two other blocks (11 instances) were defined using only MOTIF SAMPLER and alignments, and a single block (10 instances) only by C LZ analysis.
Over a third of blocks were conserved in two or more plant families, but the remainder were distinctive to single families, or, in the case of the solanaceous genes, to particular orthologous loci identified by Dean et al. .
Blocks are listed in Table 1 with 'definitions' as typical instances, since for variable blocks a consensus would be dominated by ambiguous IUPAC code. The degree of conservation of each instance relative to the 'definition' is indicated by the vertical block dimensions in Figures 2 and 3; the 'definitions' were chosen to maximize these dimensions and do not necessarily represent importance in functional terms. Full sequences and locations of all block instances are in Additional File 1.
The 18 blocks asterisked in Table 1 have been recognized in past rbcS research. Of these, the motif most represented was the I-box (blocks 08, 09, 12, 18, 29). The reverse-strand I-box (block 29) immediately upstream of the TATA-box (block 30) was found by Grob and Stüber , who termed it the light-responsive element (LRE).
The I-box block 08 functions in a light-responsive dual unit with the G-box block 10 . The I-G boxes unit represented by blocks [08–10] was found to be common in light-responsive promoters, and termed rbcS-CMA5 by Argüello-Astorga and Herrera-Estrella . In rosid NCS, a second I-box downstream (block 12) occurred in an I-G-I boxes array postulated as ancestral by these authors. The TG-rich block 11, between the G-box (10) and second I-box (12), formed part of rbcS-CMA4 of Argüello-Astorga and Herrera-Estrella . Block 11 usually corresponded to Motif 4 of Manzara and Gruissem  (but see later on Box II).
In the present study, the largest CMA found in all three plant families comprised blocks [06-08-10-11], in dotted-line boxes in Figures 2 and 3. Block 06 is a previously overlooked motif, but we found identical versions in similar relative locations in the caryophyllid genes Mesembryanthemum crystallinum rbcS-1 [EMBL L10212, -241 bp] and Spinacia oleracea rbcS-1 [EMBL X73236, -363 bp]. In pea rbcS-3A, block 06 overlapped the 5' flank of the box III* inverted GT-1 site , and so might be a site for a factor like 3AF5, a light-regulated phosphoprotein that binds the 5' flank of the similar downstream Box III . The pea rbcS-3A 3AF5 and Box III sites themselves corresponded to legume-specific blocks 19 and 21, which with block 17 are equivalent to rbcS-CMA3 of Argüello-Astorga and Herrera-Estrella .
Block 17 coincided with the pea rbcS-3A Box II element, which is the prototype of GT-1 trihelix TF binding sites and a target of the calcium phototransduction pathway [26, 29]. The variability of Box II-like motifs  was reflected in the low MOTIF SAMPLER consensus score  for block 17 (1.04), but this block was recognized by MOTIF SAMPLER with 85% sensitivity, and aligned in all dicot NCS by DIALIGN and ALIGN-M on its relatively conserved TGTGG sub-fragment. The Box II motifs of earlier alignments [25, 27] corresponded to block 17 for most sequences, but to block 11 for tomato rbcS-2 and rbcS-3A. (Local alignments of sequences not available to the earlier authors confirm our assignments.)
The solanaceous 5'-NCS (Figure 2) yielded further previously identified motifs, whose functions generally remain uncertain. Blocks 22, 23 and 25 were components of rbcS-CMA2  and identified by Manzara et al.  (Table 1). Likewise, the blocks 20 and 28 flanking rbcS-CMA2 were found by Manzara et al.  (Table 1).
On average, 10.2% of the length of those sequences with known transcription start sites was occupied by 5'-UTRs, though these were highly variable in extent (Figures 2, 3). Blocks 32–34 occurred in the proximities of transcription start sites. Only two blocks, 35 and 36, were located fully within 5'-UTRs, but each featured in multiple sequences in several species (Figures 2, 3).
Precisely half the blocks in Table 1 were newly identified in this study. These novel blocks were confined to single plant families, apart from the brassica blocks 02, 24 and 35 also found in a legume species. In most of the novel blocks, potential cis-elements could be speculatively identified using promoter databases (Table 1).
On the other hand, one-third of DNase-protected regions did not overlap with well defined blocks (Figure 4). These additional DNase-protected sequences tended to be very variable between genes and dominated by particular nucleotides (e.g. AT-rich regions). The latter feature can be formally translated as low complexity, as shown by the sliding-window profiles of C LZ  with respect to the [AT] [GC] alphabet in Figure 4. The association of DNase-protection with C LZ troughs implied functional roles for low-complexity regions.
One characterized mechanism for the introduction of AT-richness (and thus low C LZ ) into an rbcS 5'-NCS sequence is the Stowaway-Le2 inverted repeat element in the tomato rbcS-1 sequence (Figure 3) [56, 57]. Sliding-window C LZ profiles confirmed the Stowaway-Le2 sequence as one of the main low-complexity regions of the tomato rbcS-1 5'-NCS (not shown). DNase-protected regions do occur within the Stowaway-Le2 sequence [53, 56, 57].
Minimum Age Estimates for rbcS 5'-NCS Blocks
Minimum age (106 years)
04, 07, 09, 14, 15, 16, 22, 25, 28, 31, 34, 36
Duplication of ancestral rbcS loci ; divergence of Petunia clade 
19, 26, 27
Pisum-Medicago divergence 
Arabidopsis-Brassica divergence 
02, 03, 05, 13, 18, 32
Duplication of ancestral ats loci 
Divergence of Phaseolus clade 
02, 12, 24, 35
Divergence of eurosids I & II 
06, 08, 10, 11, 17, 20, 23, 29, 30
Brassicaceae and/or Fabaceae, Solanaceae
Divergence of rosids & asterids 
Within the duplicate segments containing the ats1A and B genes, the latter were among several examples of tandem arrays, others including LRK10L receptor-like Ser/Thr kinases . Such tandem arrays, presumed due to unequal crossing over, account for up to 17% of all Arabidopsis genes , but their age range is currently uncertain .
Estimated dates for major lineage divergences implied that 13 taxonomically widespread blocks were of Cretaceous antiquity at least (Table 2). These included the [06-08-10-11] CMA, block 12 (the second I-box in rosids), and blocks 17 (Box II), 23 (CAAT-box), 29 (LRE) and 30 (TATA-box). Another, block 20, remains poorly characterized in functional terms, but does bind protein (Figure 4), and was further noted in genes from the Amaranthaceae (Spinacia oleracea rbcS-1 [EMBL X73236, -191 bp]), and Malvaceae (Gossypium hirsutum rbcS [EMBL X54091, -186 bp]). Other Cretaceous blocks were three rosid blocks (02, 24 and 35) discovered in the present study. The remaining blocks were found only in single families but could be dated by clade divergence or gene duplication events to 18–54 mya (Table 2).
The occurrence of particular phylogenetic footprints at different levels in the taxonomic hierarchy (Table 2) indicated that the 5'-NCS might be amenable to phylogenetic analysis. Opinions differ, however, about phylogenetic analysis of NCS, particularly at higher taxonomic levels. NCS are seen as problematic for alignment and phylogenetic analysis because of their structural constraints, non-randomness of evolution, and mutational changes such as slipped-strand mispairing, stem-loop secondary structure excision/repair, minute inversions, and intramolecular recombination . In practice, however, Bremer et al.  found chloroplast NCS to be of similar utility to coding sequences for asterid phylogenetics.
Secondly, more conventional analyses based on DNA parsimony or distance were applied to 5'-NCS aligned using DIALIGN, or CLUSTALW with the gap penalties found to be most effective in phylogenetic footprinting (Figure 1). (ALIGN-M was not usable as it does not produce complete alignments where sequence tracts are too divergent.) Figure 7B shows a consensus of most-parsimonious trees of DIALIGN-aligned 5'-NCS. (The short Petunia SSU11A sequence was also omitted from this tree for reasons discussed for Figure 7A.)
Our third method (Figure 7C) was a cladistic analysis of character-states defined as presence or absence of conserved blocks. All blocks in Figures 2, 3 were included: of these 96.9% had ≥ 50% identity with the definitions in Table 1. The remainder averaged 45.6% identity, and all but one had been found by three phylogenetic footprinting methods. Close inspections of aligned locations scored as absences were often suggestive of degenerate residues of blocks.
Several points of congruence between the dendrograms produced by these diverse analyses were identifiable, though bootstrap support for nodes was often moderate or weak (Figure 7). Themes included the clustering of the 5'-NCS by gene loci rather than by species. Thus, 5'-NCS of the Arabidopsis atsB tandem locus showed more affinity with the Brassica sequence than with Arabidopsis ats1A. This accorded with the conclusion of Bowers et al.  that the ancestral α duplication event occurred prior to the Brassica-Arabidopsis split, because 49–64% of relevant Brassica genes were more similar to one Arabidopsis gene than was the Arabidopsis duplicate.
The locus 3 5'-NCS presented a consistent picture in that the tomato and potato A genes were resolved as basal members of a monophyletic group (Figure 7). In fact, tomato rbcS-3A was the only gene retaining the ancestral [06-08-10-11] CMA in the analyzed region (Figure 3). The most likely counterpart among the Petunia 5'-NCS analyzed was SSU11A (Figure 7C). Petunia SSU112, SSU491 and SSU911 grouped with the more derived 5'-NCS of the tomato and potato B and C genes.
The remaining solanaceous 5'-NCS grouped into loci 1 and 2 (Figure 7). In evolutionary trees based on CLUSTALW rather than DIALIGN alignments, the Petunia locus 1 gene SSU611 formed the outgroup to locus 2, while the tomato and potato locus 1 genes grouped with the A genes of locus 3 (not shown). The DIALIGN trees were preferred as they were supported by the alternative dendrograms, and because we rated the alignments from this algorithm most highly (Figure 1). Moreover, CLUSTALW alignments of transit peptides supported affinity of SSU611 with the other solanaceous locus 1 genes (not shown).
The basal solanaceous locus could not be confidently identified, as the basal position of locus 2 in two of the dendrograms in Figure 7 had only moderate bootstrap support. Clear guidance was not forthcoming from the coding sequence Ks values (Figure 6), or from parsimony analyses (Figure 8), in which outgroup choice influenced topology with respect to these two loci.
Conserved blocks revealed by phylogenetic footprinting in dicot rbcS 5'-NCS formed an evolutionary hierarchy, from those common to plant families that diverged in the Cretaceous, to family-specific blocks with minimum estimated ages of only about 20 million years. Similar heterogeneity in longevity and clade-specificity of promoter motifs has been found in other organisms of ancient divergence. Among homologous human and rodent TF binding sites, for example, Dermitzakis and Clark  found 33 with shared functions, while 14 were human-specific and 17 rodent-specific.
The most ancient conserved blocks we found included those recognized earliest in rbcS research on the basis of functional importance (I-boxes, G-box, Box II, CAAT-box, TATA-box) [25, 27, 32], though several relatively unknown ones also fell in this category. Furthermore, we were able to extend CMAs postulated in previous studies . Younger blocks were generally of less widely recognized function, and presumably had acquired roles in the more recent clades in which they had evolved. Simulations by Stone and Wray  of the acquisition by point mutation of novel TF binding sites, and their subsequent fixation within populations, indicated the evolution of new sites must be virtually inevitable over millions of years. In a theoretical population of 106 Arabidopsis plants with two generations per year, the fixation time for two 6-bp binding sites in a 200-bp region was only 270,000 years.
Evolutionary information in the 5'-NCS was sufficient for several formal computational methods to produce dendrograms in accordance with the existing classification of solanaceous rbcS loci based descriptively on sequence similarities, intron features and linkage relations [25, 27, 67]. The solanaceous locus 2 genes are distinguished as the only land plant rbcS genes with introns at three positions, while locus 3 is a distinctive tandem array of three 2-intron rbcS genes in tomato and potato, and probably six in Petunia. Gene duplications appear to have provided additional impetus in functional evolution of rbcS genes. For instance, the strongly expressed locus 3 tomato genes rbcS-3B and rbcS-3C  represented the most derived members of tandem arrays according to our dendrograms. It has been suggested that gene duplicates are conserved and subfunctionalized by regulatory mutations, because each duplicate must survive to complement lost expression for essential subfunctions in the other . Duplicate gene preservation by such a process could be < 4 million years for a gene with ≥ 5 regulatory elements and a mutation rate of 10-7 per year . Such a rapid preservation of duplicates may need to be invoked for locus 3, because of coincident estimates (18–23 mya) for the major ancestral genome duplication event  and for divergence of the Petunia clade . In our dendrograms, segregation of Petunia SSU112, SSU491 and SSU911 with the more derived tomato and potato genes of locus 3 indicated that tandem duplications at locus 3 had occurred prior to the Petunia divergence, and had undergone relatively little subsequent sequence evolution.
Point mutations do not appear to have been the only evolutionary processes governing protein interactions in the rbcS 5'-NCS. Mechanisms such as slipped-strand mispairing  probably generated the relatively extensive and variable low-complexity tracts that coincided with known DNase footprints in the locus 3 tomato genes. Another example of the gross mutational processes that can occur in 5'-NCS was the Stowaway-Le2 transposable element in the tomato rbcS-1 sequence (Figure 3). The absence of this transposable element from the potato sequence  implies a recent insertion event in tomato.
A primary factor that facilitated our study was a suite of phylogenetic footprinting tools that complemented and cross-validated each other. The least known member of our toolkit was probably C LZ analysis, whose use deserves to increase with its availability as an internet tool . Its intuitive process of sequence decomposition by repeated fragments proved useful not only for identification of conserved motifs, but also for highlighting low-complexity regions such as AT-rich tracts, and as a similarity measure for global sequence comparisons and hence dendogram construction. Otu and Sayood  formally examined C LZ as a new sequence distance measure for phylogenetic tree construction, and demonstrated that its lack of dependence on alignments or evolutionary models was particularly suited for sequences subject to segment-based modifications, including whole mitochondrial genomes of eutherian mammals. Promising alternative alignment-independent methods of sequence comparison have also been proposed using the general information theoretical concept of Kolmogorov complexity [71, 72], of which C LZ is one explicitly computable implementation.
The dendrograms we produced using C LZ , and those obtained by parsimony analysis of DIALIGN alignments or block characters, were of sufficient consistency to confirm the presence of evolutionary information in plant 5'-NCS. The dataset was not designed to investigate taxonomic phylogenies, as it included several multigene families. Moreover, we would not claim that the dendrograms rival in quality those produced using coding sequences, as bootstrap support for nodes was often moderate or weak, and there were points of variance between the dendrograms. Further investigation is needed to establish the extent to which NCS might contribute to molecular phylogenetics. We do, however, conclude that current computational methods provide the potential for analysis of the evolution of gene expression in terms of promoter structure.
Comprehensive phylogenetic footprinting of dicot 5'-NCS revealed conserved modular arrays of recurrent sequence blocks. Transcriptional functionality was confirmed as an evolutionary basis for this conservation by coincidence of recurrent blocks with cis-elements and protein-binding sites. Evolutionary hierarchies were discernible within the assemblage of blocks, such that taxonomically widespread, and hence ancient, blocks could be distinguished from taxon-specific, more recent, ones.
Noncoding sequences (NCS) up to 400 bp including and immediately 5' to the ATG codon were obtained for the following genes [accession numbers, bp analyzed]: Arabidopsis thaliana ats1A [EMBL:X13611, 400], ats1B [EMBL:X14564, 400], ats2B [EMBL:X14564, 400], ats3B [EMBL:X14564, 400]; Brassica napus rbcS [EMBL:X61097, 400]; Phaseolus vulgaris rbcS-2 [EMBL:AF028707, 400]; Pisum sativum rbcS-E9 [EMBL:X00806, 400], rbcS-3A [EMBL:M21356, 400], rbcS-3C [EMBL:X04334, 331]; Medicago sativa rbcSK-1A [EMBL:X96847, 400]; Lycopersicon esculentum rbcS-1 [EMBL:X05982, 338], rbcS-2 [EMBL:X05983, 400], rbcS-3A [EMBL:X05984, 380], rbcS-3B [EMBL:X05985, 283], rbcS-3C [EMBL:X05986, 300]; Petunia × hybrida SSU112 [EMBL:X12990, 351], SSU11A [EMBL:X03821, 281], SSU301 [EMBL:X12986, 400], SSU491 [EMBL:X12988, 400], SSU611 [EMBL:X12987, 400], SSU911 [EMBL:X12989, 400]; Nicotiana plumbaginifolia rbcS-8B [EMBL:X13711, 400]; Solanum tuberosum rbcS-1 [EMBL:X69759, 400], rbcS-2A [EMBL:X69760, 400], rbcS-2B [EMBL:X69761, 400], rbcS-2C [EMBL:X69762, 400], rbcS-3 [EMBL:X69763, 382]; Zea mays rbcSZm1 [EMBL:S42508, 400]. Coding sequences were from the same accessions, except P. sativum rbcS-3A [EMBL:X04333] and Zea mays ZmrbcS [EMBL:Y00322].
Duplicated ancestral chromosome segments encompassing the Arabidopsis ats genes were identified (as block 0105451100840) in the PARALOGONS IN ARABIDOPSIS THALIANA database [58, 73]. Potential cis-elements in the 5'-NCS were identified using the PLACE [74, 75] database.
Recurrent sequence blocks were identified in rbcS 5'-NCS by Lempel-Ziv complexity (C LZ ) decomposition. Lempel and Ziv  suggested measurement of sequence complexity by the number of steps required for the iterative generation (recovery) of a given sequence S from scratch, using two possible 'recovery' operations per iteration: either copy a fragment that has already been encountered in the recovered part of the sequence; or add (generate) a new symbol not encountered before. This iterative process, called a decomposition, represents a sequence S as a concatenation of m consecutive fragments, H(S) = S [1:i1]S [i1+1: i2]...S [im-1+1: i m = N], where S [ik-1+1: i k ] is a fragment copied or generated at k-th step, N is the length of the sequence and m = m H (S) is the number of steps in decomposition process. Among all possible decompositions the one with the minimum number of steps defines sequence complexity, i.e. . The minimum is ensured by copying at each step of the decomposition process the longest fragment that has been encountered before. Similarly, one can define a pair-wise complexity of sequences S and Q, C(S|Q), as the number of steps needed to recover Q from S (or S from Q). In this case, each fragment in the decomposition of Q is the longest one whose copy occurs anywhere in sequence S. Gusev et al.  proposed a linear algorithm for sequence decomposition and computation of C LZ with respect to various types of repeat (including direct and inverted repeats or any combination of them). Its implementation is available online at LZCOMPOSER [55, 77].
The full algorithm used in this study follows. Step 1. For N 5'-NCS denoted as S1,..., S N (with corresponding lengths |S1|,..., |S N |), a new, concatenated sequence ŋ = S1#...# S N of length L = Σ |S i |, i = 1,..., N was defined. (The arbitrary symbol # separated the concatenated sequences.) Step 2. A Lempel-Ziv decomposition of ŋ into m consecutive fragments, ŋ [1:i1]ŋ [i1+1:i2] ... ŋ [im-1: L], was computed, such that ŋ [ik-1+1:i k ] was the longest fragment downstream of position i k-1 for which a direct repeat occurred starting from position j(k) somewhere upstream of position ik-1+1, and ŋ [i k-1 +1:i k ] did not contain #. Pointers j(k) were expressed as pairs (sequence number, position within the sequence). Step 3. Fragments ≥ 8 bp that were common for at least two sequences were included in a vocabulary of 'blocks'. Only exact matches were considered in the decomposition process. However, when two or more consecutive fragments of a decomposition were identical to the respective substrings in another sequence, and when these fragments were separated by a similar number of nucleotides (± 1) then they were merged into a single block. All remaining sequences from the given dataset were scanned for the occurrence of these blocks. For each block, the origin in the decomposition and the entire track of occurrences in different sequences were traced, ensuring that the fragments found were independent of the sequence order in ŋ. Step 4. Fragments defined as the same block were aligned, including an extra 10 bp either side to check for possible block extension, and their consensus sequence was defined allowing for a given number of mismatches (initially two). Step 5. All sequences were then scanned for each block defined by its consensus. No steps involved a priori knowledge of cis-elements. A similar algorithm was used to search for inverted repeats , but we found too few of these for detailed analysis. The decomposition process is available online at LZCOMPOSER [55, 77].
Matrices (N × N) of pairwise C LZ values for N sequences were produced on LZCOMPOSER (using the symmetrized matrix output with diagonals adjusted to 0). Sliding-window profiles of local C LZ along single sequences were also generated on LZCOMPOSER.
Overrepresented motifs in the 5'-NCS were also sought using MOTIF SAMPLER v3.1 . The 5'-NCS were analyzed in 19 combinations, with program options in the following ranges: search (s), single stranded; prior probability of motif (p), 0.3–0.8; length of motif (w), 9–25 bp; number of different motifs (n), 3–20; number of instances of each motif per sequence (M), 1, 2 or undefined; allowed overlap (x), 1–9 bp; program repeat runs (r), 0–99. Background models of order 0–3 were used in the analysis.
Multiple alignments of 5'-NCS were performed with three algorithms: CLUSTALW v1.83 ; DIALIGN 2  in the QALIGN v1.10T software of Sammeth et al. ; and ALIGN-M v2.3 . Unless stated, gap penalties in both CLUSTALW and the S2P step of ALIGN-M were: opening 8.0; extension0.5. In the search process for conserved blocks, a total of 26 different sequence combinations were aligned with DIALIGN and/or ALIGN-M, and the blocks from C LZ analysis and MOTIF SAMPLER were mapped in the alignments.
Levels of synonymous substitutions (Ks) were obtained by multiple alignment of all the tomato and potato rbcS coding sequences by CLUSTALW (default gap penalties), followed by estimation of the matrix of pairwise Ks values by the method of Li  implemented in the R package SEQINR .
Three strategies were used to produce dendrograms of the 27 dicot sequences, with the 5'-NCS of Zea rbcSZm1 included as outgroup. (1) Hierarchical cluster analysis was performed on matrices of pairwise C LZ values (see above), using Euclidean distance to measure similarity of the different rows. Dendrograms were produced by the unweighted pair group method with arithmetic mean (UPGMA), using PAST v1.34  and the R package PVCLUST . Statistical support was assessed using PVCLUST to calculate the approximately unbiased (AU) values of Shimodaira  by multiscale bootstrap resampling of 1000 pseudoreplications. (2)Evolutionary trees were produced from multiple sequence alignments created with DIALIGN or CLUSTALW. Trees were obtained, using PAUP* v4.0b10  and PHYLIP v3.64 , by DNA parsimony or, by the neighbour-joining, UPGMA or Fitch-Margoliash methods, from DNA distance matrices produced with the Jukes-Cantor substitution model. (3)Cladistic analysis, using PAUP* v4.0b10 and PAST v1.34, was performed on the conserved blocks identified by sequence analyses. A character-state matrix of absence (0) or presence (1) of each block was created. Characters were assigned equal weight and Dollo status (i.e. a block could evolve only once, but could disappear at several points on the tree). The tree-bisection-reconnection heuristic was used to search for the most parsimonious topologies. For methods (2) and (3), the Zea sequence rbcSZm1 was specified as outgroup, and nodal support was estimated from 100 tenfold-replicated bootstrap pseudoreplicates.
Evolutionary trees of coding sequences were obtained by bootstrapped parsimony analysis in PAUP*v4.0b10 of sequences aligned by CLUSTALW (default gap penalties) or DIALIGN.
List of abbreviations
- C LZ :
conserved modular array
G-box binding factor
- Ks :
level of synonymous substitutions
million years ago
Positive Predictive Value
unweighted pair group method with arithmetic mean
This research was funded by the BBSRC Genes and Developmental Biology Committee. We are also grateful to Prof. N.J. Fiddian of Cardiff University School of Computer Science for his support.
- Wray GA, Hahn MW, Abouheif E, Balhoff JP, Pizer M, Rockman MV, Romano LA: The evolution of transcriptional regulation in eukaryotes. Mol Biol Evol. 2003, 20: 1377-1419. 10.1093/molbev/msg140.View ArticlePubMedGoogle Scholar
- Hu J, Li B, Kihara D: Limitations and potentials of current motif discovery algorithms. Nucleic Acids Res. 2005, 33: 4889-4913.Google Scholar
- Tompa M, Li N, Bailey TL, Church GM, De Moor B, Eskin E, Favorov AA, Frith MC, Fu Y, Kent WJ, Makeev VJ, Mironov AA, Noble WS, Pavesi G, Pesole G, Régnier M, Simonis N, Sinha S, Thijs G, Van Helden J, Vandenbogaert M, Weng Z, Workman C, Ye C, Zhu Z: Assessing computational tools for the discovery of transcription factor binding sites. Nature Biotech. 2005, 23: 137-144. 10.1038/nbt1053.View ArticleGoogle Scholar
- Karlin S, Ghandour G, Foulser DE: DNA sequence comparisons of the human, mouse, and rabbit immunoglobulin kappa gene. Mol Biol Evol. 1985, 2: 35-52.PubMedGoogle Scholar
- Clark AG: The search for meaning in noncoding DNA. Genome Res. 2001, 11: 1319-1320. 10.1101/gr.201601.View ArticlePubMedGoogle Scholar
- Guo H, Moose SP: Conserved noncoding sequences among cultivated cereal genomes identify candidate regulatory sequence elements and patterns of promoter evolution. Plant Cell. 2003, 15: 1143-1158. 10.1105/tpc.010181.PubMed CentralView ArticlePubMedGoogle Scholar
- Van Hellemont R, Monsieurs P, Thijs G, De Moor B, de Peer YV, Marchal K: A novel approach to identifying regulatory motifs in distantly related genomes. Genome Biol. 2005, 6 (13): R113-10.1186/gb-2005-6-13-r113.PubMed CentralView ArticlePubMedGoogle Scholar
- Colinas J, Birnbaum K, Benfey PN: Using cauliflower to find conserved non-coding regions in Arabidopsis. Plant Physiol. 2002, 129: 451-454. 10.1104/pp.002501.PubMed CentralView ArticlePubMedGoogle Scholar
- Kelchner SA: The evolution of non-coding chloroplast DNA and its application in plant systematics. Ann Mo Bot Gard. 2000, 87: 482-498. 10.2307/2666142.View ArticleGoogle Scholar
- Bremer B, Bremer K, Heidari N, Erixon P, Olmstead RG, Anderberg AA, Källersjö M, Barkhordian E: Phylogenetics of asterids based on 3 coding and 3 non-coding chloroplast DNA markers and the utility of non-coding DNA at higher taxonomic levels. Mol Phylogenet Evol. 2002, 24: 274-301. 10.1016/S1055-7903(02)00240-3.View ArticlePubMedGoogle Scholar
- Koch MA, Weisshaar B, Kroymann J, Haubold B, Mitchell-Olds T: Comparative genomics and regulatory evolution: conservation and function of the Chs and Apetala3 promoters. Mol Biol Evol. 2001, 18: 1882-1891.View ArticlePubMedGoogle Scholar
- Kaplinsky NJ, Braun DM, Penterman J, Goff SA, Freeling M: Utility and distribution of conserved noncoding sequences in the grasses. Proc Natl Acad Sci USA. 2002, 99: 6147-6151. 10.1073/pnas.052139599.PubMed CentralView ArticlePubMedGoogle Scholar
- De Bodt S, Theissen G, Van de Peer Y: Promoter analysis of MADS-box genes in eudicots through phylogenetic footprinting. Mol Biol Evol. 2006, 23: 1293-1303. 10.1093/molbev/msk016.View ArticlePubMedGoogle Scholar
- Verdú S: Fifty years of Shannon theory. IEEE T Inform Theory. 1998, 44: 2057-2078. 10.1109/18.720531.View ArticleGoogle Scholar
- Gusev VD, Nemytikova LA, Chuzhanova NA: On the complexity measures of genetic sequences. Bioinformatics. 1999, 15: 994-999. 10.1093/bioinformatics/15.12.994.View ArticlePubMedGoogle Scholar
- Chuzhanova NA, Krawczak M, Nemytikova LA, Gusev VD, Cooper DN: Promoter shuffling has occurred during the evolution of the vertebrate growth hormone gene. Gene. 2000, 254: 9-18. 10.1016/S0378-1119(00)00308-5.View ArticlePubMedGoogle Scholar
- Chuzhanova NA, Krawczak M, Thomas N, Nemytikova LA, Gusev VD, Cooper DN: The evolution of the vertebrate beta-globin gene promoter. Evolution. 2002, 56: 224-232. 10.1554/0014-3820(2002)056[0224:TEOTVG]2.0.CO;2.PubMedGoogle Scholar
- Bacolla A, Jaworski A, Larson JE, Jakupciak JP, Chuzhanova N, Abeysinghe SS, O'Connell CD, Cooper DN, Wells RD: Breakpoints of gross deletions coincide with non-B DNA conformations. Proc Natl Acad Sci USA. 2004, 101: 14162-14167. 10.1073/pnas.0405974101.PubMed CentralView ArticlePubMedGoogle Scholar
- Chuzhanova NA, Anassis EJ, Ball EV, Krawczak M, Cooper DN: Meta-analysis of indels causing human genetic disease: Mechanisms of mutagenesis and the role of local DNA sequence complexity. Hum Mutat. 2003, 21: 28-44. 10.1002/humu.10146.View ArticlePubMedGoogle Scholar
- Kehrer-Sawatzki H, Sandig C, Chuzhanova N, Goidts V, Szamalek JM, Tanzer S, Muller S, Platzer M, Cooper DN, Hameister H: Breakpoint analysis of the pericentric inversion distinguishing human chromosome 4 from the homologous chromosome in the chimpanzee (Pan troglodytes). Hum Mutat. 2005, 25: 45-55. 10.1002/humu.20116.View ArticlePubMedGoogle Scholar
- Thijs G, Marchal K, Lescot M, Rombauts S, de Moor B, Rouzé P, Moreau Y: A Gibbs sampling method to detect overrepresented motifs in the upstream regions of coexpressed genes. J Comput Biol. 2002, 9: 447-464. 10.1089/10665270252935566.View ArticlePubMedGoogle Scholar
- Morgenstern B: DIALIGN 2: improvement of the segment-to-segment approach to multiple sequence alignment. Bioinformatics. 1999, 15: 211-218. 10.1093/bioinformatics/15.3.211.View ArticlePubMedGoogle Scholar
- Morgenstern B, Frech K, Dress A, Werner T: DIALIGN: Finding local similarities by multiple sequence alignment. Bioinformatics. 1998, 14: 290-294. 10.1093/bioinformatics/14.3.290.View ArticlePubMedGoogle Scholar
- Van Walle I, Lasters I, Wyns L: Align-m - a new algorithm for multiple alignment of highly divergent sequences. Bioinformatics. 2004, 20: 1428-1435. 10.1093/bioinformatics/bth116.View ArticlePubMedGoogle Scholar
- Dean C, Pichersky E, Dunsmuir P: Structure, evolution and regulation of RbcS genes in higher plants. Annu Rev Plant Physiol Plant Mol Biol. 1989, 40: 415-439. 10.1146/annurev.pp.40.060189.002215.View ArticleGoogle Scholar
- Gilmartin PM, Memelink J, Hiratsuka K, Kay SA, Chua NH: Characterization of a gene encoding a DNA binding protein with specificity for a light-responsive element. Plant Cell. 1992, 4: 839-849. 10.1105/tpc.4.7.839.PubMed CentralView ArticlePubMedGoogle Scholar
- Manzara T, Gruissem W: Organization and expression of the genes encoding ribulose-1,5-bisphosphate carboxylase in higher plants. Photosynth Res. 1988, 16: 117-139. 10.1007/BF00039489.View ArticlePubMedGoogle Scholar
- Argüello-Astorga G, Herrera-Estrella L: Ancestral multipartite units in light-responsive plant promoters have structural features correlating with specific phototransduction pathways. Plant Physiol. 1996, 112: 1151-1166. 10.1104/pp.112.3.1151.PubMed CentralView ArticlePubMedGoogle Scholar
- Wu Y, Hiratsuka K, Neuhaus G, Chua NH: Calcium and cGMP target distinct phytochrome-responsive elements. Plant J. 1996, 10: 1149-1154. 10.1046/j.1365-313X.1996.10061149.x.View ArticlePubMedGoogle Scholar
- Argüello-Astorga G, Herrera-Estrella L: Evolution of light-regulated plant promoters. Annu Rev Plant Physiol Plant Mol Biol. 1998, 49: 525-555. 10.1146/annurev.arplant.49.1.525.View ArticlePubMedGoogle Scholar
- Martínez-Hernández A, López-Ochoa L, Argüello-Astorga G, Herrera-Estrella L: Functional properties and regulatory complexity of a minimal RBCS light-responsive unit activated by phytochrome, cryptochrome, and plastid signals. Plant Physiol. 2002, 128: 1223-1233. 10.1104/pp.010678.PubMed CentralView ArticlePubMedGoogle Scholar
- Gilmartin PM, Sarokin L, Memelink J, Chua NH: Molecular light switches for plant genes. Plant Cell. 1990, 2: 369-378. 10.1105/tpc.2.5.369.PubMed CentralView ArticlePubMedGoogle Scholar
- Guiliano G, Pichersky E, Malik VS, Timko MP, Scolnik PA, Cashmore A: An evolutionarily conserved protein binding sequence upstream of a plant light-regulated gene. Proc Natl Acad Sci USA. 1988, 85: 7089-7093. 10.1073/pnas.85.19.7089.View ArticleGoogle Scholar
- Sibéril Y, Doireau P, Gantet P: Plant bZIP G-box binding factors. Eur J Biochem. 2001, 268: 5655-5666. 10.1046/j.0014-2956.2001.02552.x.View ArticlePubMedGoogle Scholar
- Chattopadhyay S, Ang LH, Puente P, Deng XW, Nei W: Arabidopsis bZIP protein HY5 directly interacts with light-responsive promoters in mediating light control of gene expression. Plant Cell. 1998, 10: 673-683. 10.1105/tpc.10.5.673.PubMed CentralView ArticlePubMedGoogle Scholar
- Donald RGK, Cashmore A: Mutation of either G box or I box sequences profoundly affects expression from the Arabidopsis rbcS1A promoter. EMBO J. 1990, 9: 1717-1726.PubMed CentralPubMedGoogle Scholar
- Grob U, Stüber K: Discrimination of phytochrome dependent light inducible from non-light inducible plant genes. Prediction of a common light-responsive element (LRE) in phytochrome dependent light inducible plant genes. Nucleic Acids Res. 1987, 15: 9957-9973. 10.1093/nar/15.23.9957.PubMed CentralView ArticlePubMedGoogle Scholar
- Dean C, Vandenelzen P, Tamaki S, Dunsmuir P, Bedbrook J: Differential expression of the 8 genes of the petunia ribulose bisphosphate carboxylase small subunit multi-gene family. EMBO J. 1985, 4: 3055-3061.PubMed CentralPubMedGoogle Scholar
- Teakle GR, Manfield IW, Graham JF, Gilmartin PM: Arabidopsis thaliana GATA factors: organisation, expression and DNA-binding characteristics. Plant Mol Biol. 2002, 50: 43-57. 10.1023/A:1016062325584.View ArticlePubMedGoogle Scholar
- Borello U, Ceccarelli E, Guiliano G: Constitutive, light-responsive and circadian clock-responsive factors compete for the different I box elements in plant light-regulated promoters. Plant J. 1993, 4: 611-619. 10.1046/j.1365-313X.1993.04040611.x.View ArticlePubMedGoogle Scholar
- Buzby JS, Yamada T, Tobin EM: A light-regulated DNA-binding activity interacts with a conserved region of a Lemna gibba rbcS promoter. Plant Cell. 1990, 2: 805-814. 10.1105/tpc.2.8.805.PubMed CentralView ArticlePubMedGoogle Scholar
- Rose A, Meier I, Wienand U: The tomato I-box binding factor LeMYB1 is a member of a novel class of Myb-like proteins. Plant J. 1999, 20: 641-652. 10.1046/j.1365-313X.1999.00638.x.View ArticlePubMedGoogle Scholar
- Sarokin L, Chua NH: Binding sites for two novel phosphoproteins, 3AF5 and 3AF3, are required for rbcS-3A expression. Plant Cell. 1992, 4: 473-483. 10.1105/tpc.4.4.473.PubMed CentralView ArticlePubMedGoogle Scholar
- Lam E, Kano-Murakami Y, Gilmartin P, Niner B, Chua NH: A metal-dependent DNA-binding protein interacts with a constitutive element of a light-responsive promoter. Plant Cell. 1990, 2: 857-866. 10.1105/tpc.2.9.857.PubMed CentralView ArticlePubMedGoogle Scholar
- Meier I, Callan KL, Fleming AJ, Gruissem W: Organ-specific differential regulation of a promoter subfamily for the ribulose-1,5-bisphosphate carboxylase/oxygenase small subunit genes in tomato. Plant Physiol. 1995, 107: 1105-1118. 10.1104/pp.107.4.1105.PubMed CentralView ArticlePubMedGoogle Scholar
- Urwin NAR, Jenkins GI: A sucrose repression element in the Phaseolus vulgaris rbcS2 gene promoter resembles elements responsible for sugar stimulation of plant and mammalian genes. Plant Mol Biol. 1997, 35: 929-942. 10.1023/A:1005950915499.View ArticlePubMedGoogle Scholar
- Ueda T, Pichersky E, Malik VS, Cashmore A: Level of expression of the tomato rbcS-3A gene is modulated by a far upstream promoter element in a developmentally regulated manner. Plant Cell. 1989, 1: 217-227. 10.1105/tpc.1.2.217.PubMed CentralView ArticlePubMedGoogle Scholar
- Datta N, Cashmore A: Binding of a pea nuclear protein to promoters of certain photoregulated genes is modulated by phosphorylation. Plant Cell. 1989, 1: 1069-1077. 10.1105/tpc.1.11.1069.PubMed CentralView ArticlePubMedGoogle Scholar
- Manzara T, Carrasco P, Gruissem W: Developmental and organ-specific changes in promoter DNA-protein interactions in the tomato rbcS gene family. Plant Cell. 1991, 3: 1305-1316. 10.1105/tpc.3.12.1305.PubMed CentralView ArticlePubMedGoogle Scholar
- Krebbers E, Seurinck J, Herdies L, Cashmore AR, Timko MP: Four genes in two diverged subfamilies encode the ribulose-1,5-bisphosphate carboxylase small subunit polypeptides of Arabidopsis thaliana. Plant Mol Biol. 1988, 11: 745-759. 10.1007/BF00019515.View ArticlePubMedGoogle Scholar
- Wikström N, Savolainen V, Chase MW: Evolution of the angiosperms: calibrating the family tree. Proc R Soc Lond B. 2001, 268: 2211-2220. 10.1098/rspb.2001.1782.View ArticleGoogle Scholar
- Thijs G, Lescot M, Marchal K, Rombauts S, De Moor B, Rouze P, Moreau Y: A higher-order background model improves the detection of promoter regulatory elements by Gibbs sampling. Bioinformatics. 2001, 17: 1113-1122. 10.1093/bioinformatics/17.12.1113.View ArticlePubMedGoogle Scholar
- Manzara T, Carrasco P, Gruissem W: Developmental and organ-specific changes in DNA-protein interactions in the tomato rbcS1, rbcS2 and rbcS3A promoter regions. Plant Mol Biol. 1993, 21: 69-88. 10.1007/BF00039619.View ArticlePubMedGoogle Scholar
- Carrasco P, Manzara T, Gruissem W: Developmental and organ-specific changes in DNA-protein interactions in the tomato rbcS3B and rbcS3C promoter regions. Plant Mol Biol. 1993, 21: 1-15. 10.1007/BF00039613.View ArticlePubMedGoogle Scholar
- Orlov YL, Potapov VN: Complexity: an internet resource for analysis of DNA sequence complexity. Nucleic Acids Res. 2004, 32: W628-W633. 10.1093/nar/gkh466.PubMed CentralView ArticlePubMedGoogle Scholar
- Blume B, Barry CS, Hamilton AJ, Bouzayen M, Grierson D: Identification of transposon-like elements in non-coding regions of tomato ACC oxidase genes. Mol Gen Genet. 1997, 254: 297-303. 10.1007/s004380050419.View ArticlePubMedGoogle Scholar
- Bureau TE, Wessler SR: Stowaway: a new family of inverted repeat elements associated with the genes of both monocotyledonous and dicotyledonous plants. Plant Cell. 1994, 6: 907-916. 10.1105/tpc.6.6.907.PubMed CentralView ArticlePubMedGoogle Scholar
- Blanc G, Hokamp K, Wolfe K: A recent polyploidy superimposed on older large-scale duplications in the Arabidopsis genome. Genome Res. 2003, 13: 137-144. 10.1101/gr.751803.PubMed CentralView ArticlePubMedGoogle Scholar
- Bowers JE, Chapman BA, Rong J, Paterson AH: Unravelling angiosperm genome evolution by phylogenetic analysis of chromosomal duplication events. Nature. 2003, 422: 433-438. 10.1038/nature01521.View ArticlePubMedGoogle Scholar
- Shiu SH, Bleecker AB: Expansion of the receptor-like kinase/Pelle gene family and receptor-like proteins in Arabidopsis. Plant Physiol. 2003, 132: 530-543. 10.1104/pp.103.021964.View ArticlePubMedGoogle Scholar
- Zhang L, Gaut BS: Does recombination shape the distribution and evolution of tandemly arrayed genes (TAGs) in the Arabidopsis thaliana genome?. Genome Res. 2003, 13: 2533-2540. 10.1101/gr.1318503.PubMed CentralView ArticlePubMedGoogle Scholar
- Blanc G, Wolfe KH: Widespread paleopolyploidy in model plant species inferred from age distributions of duplicate genes. Plant Cell. 2004, 16: 1667-1678. 10.1105/tpc.021345.PubMed CentralView ArticlePubMedGoogle Scholar
- Angers B, Gharbi K, Estoup A: Evidence of gene conversion events between paralogous sequences produced by tetraploidization in Salmoninae fish. J Mol Evol. 2002, 54: 501-510. 10.1007/s00239-001-0041-x.View ArticlePubMedGoogle Scholar
- Dean C, van den Elzen P, Tamaki S, Black M, Dunsmuir P, Bedbrook J: Molecular characterization of the rbcS multi-gene family of Petunia (Mitchell). Mol Gen Genet. 1987, 206: 465-474. 10.1007/BF00428887.View ArticleGoogle Scholar
- Dermitzakis ET, Clark AG: Evolution of transcription factor binding sites in mammalian gene regulatory regions: conservation and turnover. Mol Biol Evol. 2002, 19: 1114-1121.View ArticlePubMedGoogle Scholar
- Stone JR, Wray GA: Rapid evolution of cis-regulatory sequences via local point mutations. Mol Biol Evol. 2001, 18: 1764-1770.View ArticlePubMedGoogle Scholar
- Fritz CC, Wolter FP, Schenkemeyer V, Herget T, Schreier PH: The gene family encoding the ribulose-(1,5)-bisphosphate carboxylase/oxygenase (Rubisco) small subunit of potato. Gene. 1993, 137: 271-274. 10.1016/0378-1119(93)90019-Y.View ArticlePubMedGoogle Scholar
- Sugita M, Gruissem W: Developmental, organ-specific, and light-dependent expression of the tomato ribulose-1,5-bisphosphate carboxylase small subunit gene family. Proc Natl Acad Sci USA. 1987, 84: 7104-7108. 10.1073/pnas.84.20.7104.PubMed CentralView ArticlePubMedGoogle Scholar
- Force A, Lynch M, Pickett FB, Amores A, Yan YL, Postlethwait J: Preservation of duplicate genes by complementary, degenerative mutations. Genetics. 1999, 151: 1531-1545.PubMed CentralPubMedGoogle Scholar
- Otu HH, Sayood K: A new sequence distance measure for phylogenetic tree construction. Bioinformatics. 2003, 19: 2122-2130. 10.1093/bioinformatics/btg295.View ArticlePubMedGoogle Scholar
- Varré JS, Delahaye JP, Rivals E: Transformation distances: a family of dissimilarity measures based on movements of segments. Bioinformatics. 1999, 15: 194-202. 10.1093/bioinformatics/15.3.194.View ArticlePubMedGoogle Scholar
- Li M, Badger JH, Chen X, Kwong S, Kearney P, Zhang H: An information-based sequence distance and its application to whole mitochondrial genome phylogeny. Bioinformatics. 2001, 17: 149-154. 10.1093/bioinformatics/17.2.149.View ArticlePubMedGoogle Scholar
- Paralogons in Arabidopsis thaliana. [http://wolfe.gen.tcd.ie/athal/dup]
- Higo K, Ugawa Y, Iwamoto M, Korenaga T: Plant cis-acting regulatory DNA elements (PLACE) database: 1999. Nucleic Acids Res. 1999, 27: 297-300. 10.1093/nar/27.1.297.PubMed CentralView ArticlePubMedGoogle Scholar
- Plant cis-acting regulatory DNA elements (PLACE) database. [http://www.dna.affrc.go.jp/PLACE/]
- Lempel A, Ziv J: Complexity of finite sequences. IEEE T Inform Theory. 1976, 22 (1): 75-81. 10.1109/TIT.1976.1055501.View ArticleGoogle Scholar
- LZcomposer. [http://wwwmgs.bionet.nsc.ru/mgs/programs/lzcomposer/complete.html]
- Thompson JD, Higgins DG, Gibson TJ: CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position specific gap penalties and weight matrix choice. Nucleic Acids Res. 1994, 22: 4673-4680. 10.1093/nar/22.22.4673.PubMed CentralView ArticlePubMedGoogle Scholar
- Sammeth M, Rothgänger J, Esser W, Albert J, Stoye J, Harmsen D: QAlign: quality-based multiple alignments with dynamic phylogenetic analysis. Bioinformatics. 2003, 19: 1592–1593-PubMedGoogle Scholar
- Li WH: Unbiased estimation of the rates of synonymous and nonsynonymous substitution. J Mol Evol. 1993, 36: 96-99. 10.1007/BF02407308.View ArticlePubMedGoogle Scholar
- The R Project for Statistical Computing. [http://www.r-project.org/]
- Hammer O, Harper DAT, Ryan PD: PAST: paleontological statistics software package for education and data analysis. Palaeontologia Electronica. 2001, 4(1): http://palaeo-electronica.org.Google Scholar
- Shimodaira H: An approximately unbiased test of phylogenetic tree selection. Syst Biol. 2002, 51: 492-508. 10.1080/10635150290069913.View ArticlePubMedGoogle Scholar
- Swofford DL: PAUP*: Phylogenetic Analysis using Parsimony (* and Other Methods). Version 4. 1998, Sunderland, Massachusetts , Sinauer AssociatesGoogle Scholar
- Felsenstein J: PHYLIP (Phylogeny Inference Package). 2004, Seattle , University of Washington, Version 3.64Google Scholar
- Fiebig C, Link G: 5'-upstream cis-elements and binding factor(s) potentially involved in light-regulated expression of a Brassica napus rbcS gene. Curr Genet. 1992, 21: 161-168. 10.1007/BF00318476.View ArticlePubMedGoogle Scholar
- Fluhr R, Moses P, Morelli G, Coruzzi G, Chua NH: Expression dynamics of the pea rbcS multigene family and organ distribution of the transcripts. EMBO J. 1986, 5: 2063-2071.PubMed CentralPubMedGoogle Scholar
- Dean C, Favreau M, Dunsmuir P, Bedbrook J: Confirmation of the relative expression levels of the Petunia (Mitchell) rbcS genes. Nucleic Acids Res. 1987, 15: 4655-4668. 10.1093/nar/15.11.4655.PubMed CentralView ArticlePubMedGoogle Scholar
- Poulsen C, Chua NH: Dissection of 5' upstream sequences for selective expression of the Nicotiana plumbaginifolia rbcS-8B gene. Mol Gen Genet. 1988, 214: 16-23. 10.1007/BF00340173.View ArticlePubMedGoogle Scholar
- Kuhlemeier C, Cuozzo M, Green PJ, Goyvaerts E, Ward K, Chua NH: Localization and conditional redundancy of regulatory elements in rbcS-3A, a pea gene encoding the small subunit of ribulose-bisphosphate carboxylase. Proc Natl Acad Sci USA. 1988, 85: 4662-4666. 10.1073/pnas.85.13.4662.PubMed CentralView ArticlePubMedGoogle Scholar
- Lavin M, Herendeen PS, Wojciechowski MF: Evolutionary rates analysis of Leguminosae implicates a rapid diversification of lineages during the Tertiary. Syst Biol. 2005, 54: 530–549-10.1080/10635150590947131.View ArticleGoogle Scholar
- Koch MA, Haubold B, Mitchell-Olds T: Comparative evolutionary analysis of chalcone synthase and alcohol dehydrogenase loci in Arabidopsis, Arabis, and related genera (Brassicaceae). Mol Biol Evol. 2000, 17: 1483-1498.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.