Sequence analysis of an 81 kb contig from
Arabidopsis thaliana
chromosome III
Sequence analysis of an 81 kb contig from Arabidopsis thaliana chromosome III
Françoise
Quigley*
,
Patrick
Dao
,
Annick
Cottet
and
Régis
Mache
Laboratoire de Génétique Moléculaire des Plantes, Université Joseph Fourier et Centre National de la
Recherche Scientifique, BP 53, 38041
Grenoble
cedex 9,
France
Received June 18, 1996
;
Revised and Accepted September 12, 1996
DDBJ/EMBL/GenBank accession nos
+
ABSTRACT
The nucleotide sequence of a 81 493 bp contig from
Arabidopsis thaliana
chromosome III has been determined together with 11 corresponding cognate cDNAs.
Analysis of the finished sequence and comparison with public databases
indicates a gene density of one gene per 4527 bp and identifies 17 novel genes,
10 of which are totally unknown or have no well-defined function. In addition, the contig contains part of a non-LTR retrotransposon and large direct and inverted repeats. Contig
analysis also provides information on the structure and genomic organization of
plant genes.
INTRODUCTION
The biological importance of genome sequencing is now well recognized.
Information concerning new gene identity and organization, gene density and the
structure of intergenic regions and specific genetic elements is being
obtained. Comparison of sequence information across species brings new insights
into the evolution of organisms. At the present time, total genome sequencing of the three model organisms
Haemophilus influenzae
(
1
)
Mycoplasma genitalium
(
2
) and yeast has been achieved. The next logical step is sequencing of higher
eukaryote complex genomes and in this respect the flowering plant
Arabidopsis thaliana
, a member of the Cruciferae (Brassica) family, is an excellent candidate as a
model organism.
Arabidopsis thaliana
is particularly well suited to large-scale systematic sequencing, with one of the smallest genomes observed in
higher plants and a low content of interspersed repetitive DNA (
3
). The haploid genome of this plant
is distributed into five chromosomes.
A European project, ESSA (European Scientists Sequencing
Arabidopsis
), has been established to sequence large contigs of chromosome IV and other
regions dispersed throughout the genome. As part of this project we sequenced a
81 493 bp contig from chromosome III.
A major problem in sequence analysis of a large contig is gene identification
and the determination of exons and introns and especially of their exact
position. To solve this problem we have used sequence information from cognate
cDNAs and expressed sequence tags (ESTs) together with informatics programs
trained for the
Arabidopsis
genome. We thus identified 18 genes and the remnants of a non-LTR retrotransposon on the 81 kb contig. Results of this analysis are
presented here.
MATERIALS AND METHODS
Chromosome walking
The [lambda] clones covering the contig were generated from two
A.thaliana
ecotype Columbia genomic libraries in EMBL3 and [lambda] GEM11 vectors (the latest from J.T.Mulligan). Selected [lambda] clone DNAs were purified and subcloned into pUC or pBluescript
using standard methods.
Cognate cDNAs
Cognate cDNAs were obtained by screening [lambda] ZAPII cDNA libraries prepared from young shoots (a gift from M.Kreis, Orsay), cell culture (a gift from B.Lescure, Toulouse) or by RT-PCR using 20mer synthetic primers covering the putative gene length followed by cloning into pUC.
Computer assisted analysis
Sequence assembly was performed using the assembly program from the Genetics Computer Group (GCG, University of Wisconsin), version 7.2.
Completed contigs were analyzed at the Martinsried Institute for Protein
Sequences (MIPS) with GENFINDER trained for
Arabidopsis
(C.Wilson and S.Klosterman), FINDORF from Dr S.Liebl and other gene identifying algorithms.
Specific analysis and databank searches were performed with a program package
available at INFOBIOGEN (French National Centre for Bioinformatics). The G+C
content was calculated with FRAQCES from the SQX package. FASTA (
4
) and BLAST (GCG, version 8.1) were used in combination with public, daily
updated databases. Exon-intron boundaries were examined when necessary with the program
NetPlantGen from the Center for Biological Sequence Analysis (Denmark).
RESULTS
Assembly of the sequence
The 81 kb sequence was determined from a set of eight partially overlapping [lambda] clones obtained by chromosome walking. The starting point of the walk was a [lambda] clone containing the gene coding for eRF1-3, a eukaryotic peptide chain early release factor (
5
). At each walking step, several independent overlapping [lambda] clones were obtained and their relative position mapped by restriction
analysis. The mapping data was confirmed by end sequencing of both end and
common restriction fragment subclones of the [lambda] inserts, in order to make sure that each region of the contig was
covered by at least two independent [lambda] clones. The collinearity of the contig was finally confirmed by hybridization to independent YAC clones (CIC5H4, CIC6B11 and CIC11D3) covering this
region of chromosome III (data not shown).
The sequencing strategy evolved from directed sub-cloning and manual sequencing at the start of the project to a random
approach (shotgun cloning into M13) followed by automated fluorescent
sequencing after DNA nebulization of overlappping [lambda] inserts subcloned into pUC. At this stage sequence assembly was
performed using the assembly program of the GCG package, making sure that both
DNA strands were covered. Eventual ambiguities were solved by resequencing or
directed sequencing with specific primers.
List of ORFs/genes and genetic element identified in the 81 kb contig and short
description of their homologies
*Highest scoring homologue from SwissProt. Only scores >200 were considered.
The cognate cDNAs identification and sequencing which was undertaken during this
work allowed us to verify the accuracy of the genomic sequencing. A 23.8%
coverage of the 81 kb contig was realized by sequencing of the cognate cDNAs on
one strand at least. The overall sequence accuracy achieved was 99.98%, a value
of the same order as that reported in similar studies.
At this stage, the finished sequence was sent to MIPS, where gene identification analyses were performed. More specific sequence analyses and similarity searches of the public sequence databases as described in Materials and Methods were done locally.
The annotated sequence was submitted to the EMBL databank (accession no.
X98130). Cognate cDNA accession numbers are listed in Table
1 .
Chromosomal location of the contig
At an early stage in the project similarity searches of the public databases
showed a region located ~5000 bp downstream of the sequencing starting point presenting a 100% match with
GAP
A, the gene for subunit A of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The latest map using RI lines released for chromosome III
shows
GAP
A at 56 cM down the chromosome (
6
). The orientation of the 81 kb contig relative to
GAP
A on chromosome III is not known.
Definition of ORFs and candidate genes
A total of 19 ORFs and putative genes were identified on the 81 kb contig. The identified genes were named ORFs 01-19 as a working nomenclature and will be mentioned indifferently as ORF or gene in the text.
The ORFs and putative genes were identified according to a combination of the
following principles: the use of the program GENEFINDER trained for plant sequences, comparison with public databases, localization of expressed sequence tags (ESTs) on the sequence and,
finally, examination of exon-intron boundaries with the program NetPlantGen (see Materials and
Methods).
An essential source of additional information was the identification and
sequencing of the cDNAs corresponding to 11 of the putative genes (Fig.
1
). Three of the putative genes were characterized by sequencing the
corresponding EST clones: ORF 15 (accession no. T22276); ORF 16 (accession nos
R65277 and T46660); ORF 18 (accession no. T43095). The sequence data generated
by sequencing the cognate cDNAs and ESTs was used to define or confirm exons
and gene boundaries and define splice sites exactly.
Sequence organization and composition
Genes account for 44% of the sequence (30% for exons alone), with a gene every
4500 bp, which is within the range of 2-7 kb predicted prior to large-scale sequencing data being available (
7
). The average gene length including introns and the 3' non-coding region is 2300 bp.
The different genes are dispersed evenly along the region, with some appearance
of clustering (ORFs 01-04, 08-11 and 14-19). Intergenic space varies from 327 to 5293 bp, with a
mean value of 2300 bp. Some differences in intergenic mean values are observed
according to gene orientation: terminator followed by promoter (1960 bp),
convergent promoters (1500 bp) and terminators (2600 bp). The most striking
feature of the gene organization in the 81 kb region is the unequal gene
distribution between the Watson and Crick strands, with 15 (35% occupancy) and
three genes (11% occupancy) respectively.
The average base composition for the 81 kb sequence is 36.1% G+C. As expected,
coding sequences have a higher than average G+C content relative to non-coding regions, with average values of 43.9% (40.1-47%) and 32.6% (29.6-38.1%) respectively. Intron G+C base composition varies from
24.6% in ORF 06 to 36.6% in ORF 03.
Gene structure
A total of 72 introns were identified in 12 genes (Fig.
1
), the number of introns per gene ranging from 0 to 26. A surprisingly high
number of genes had no introns: ORFs 05, 08, 09, 10 and probably 17. Genes with
a high number of small sized introns of average length 149 bp have already been identified in
Arabidopsis
(
8
). The intron mean AU content of 68% of these introns corresponds to the AU
composition found for dicotyledon plant introns (
9
). Confirming the observations of Goodall and Filipowicz (
10
), none of the introns contains the consensus
Saccharomyces cerevisiae
branching sequence (UACUAAC). Also, there is no 3' polypyrimidine region necessary for the branching lariat in the course
of vertebrate intron splicing.
Cognate cDNA and EST sequence data allowed precise definition of 45 donor and 47 acceptor intron sites, thus giving the following consensus sequence: CAG/GTAAGT...TKCAG/GTT, with 100% conservation at the GT...AG borders. This data agrees with the consensus
sequences established from nearly 900 databank entries (
11
), with a slight variation in the acceptor site sequence found to be TGYAG/GT.
Exons are relatively small in size, the majority being <67 codons (Fig.
2
). The smallest one with 12 codons is found in ORF 06. The larger exon size
corresponds to ORF 01, with 991 codons.
Overview of identical and similar
Arabidopsis
ESTs and cross-phylum matches for
Arabidopsis
novel genes
dbEST search was performed using BLASTN. Only scores >400 were considered. The highest scoring database entry is shown for each
organism. FASTA scores are indicated beneath each entry. Scores <200 were not considered significant and are indicated by a dash. Database
entries are from SwissProt, PIR* and GenBank
+
. ORF 11 (
GAP
A), a gene of eubacterial origin (16) is omitted from the cross-phylum matches list.
Comparison of the contig nucleotide sequence and of the predicted protein products with public databases gives indications on the putative functions of the genes (Table
1
). Twelve of the 18 genes find significant matches with proteins from
Arabidopsis
or other organisms. The function of seven of these putative proteins has been determined experimentally in
Arabidopsis
or other organisms. The remaining five present a protein signature or are similar to genes of
unknown function found in other organisms. The last six genes do not exhibit
any significative similarities with databank entries. A variety of different
functions are represented, with, however, a greater representation of
fundamental functions, but it may just be that such genes are well represented
in public databases and thus more likely to be identified (Table
1
). There is apparently no clustering of genes with related functions or
belonging to the same gene family. The latter could have been expected, as gene
family members have been found to be organized in tandem arrays in
Arabidopsis
(
12
).
Closely related genes can be identified by searching dbEST for sequences showing
a high score with the query sequence. Visual inspection of the resulting data
allows distinction between ESTs identical to the genomic sequences and ESTs
which differ slightly and are indicative of closely related genes. The summary
of this search for the contig coding sequences is shown in Table
2
and it can be concluded that nearly half of the genes found on the contig show
similarity to other genes. As an example, the gene encoding the eukaryotic
peptide chain release factor eRF1-3 (ORF 08) is closely related to two other genes, one of them, coding for
eRF1 2, being highly similar, with a 87% nucleotide identity where comparing
partial (915 bp) cDNA sequences (
5
). The gene corresponding to eRF1-2 has been linked to marker m219 on the top arm of chromosome I (P.Carol,
personal communication) and is an example of gene redundancy between two
chromosomes.
Database searches at the protein level also give interesting information on the
conservation of gene functions between organisms at different stages of evolution (Table
2
; see Discussion).
Repeats
Several types of repeated sequences are present in the 81 kb region. Figure
1
indicates the locations of the larger repeats. Three types of repeats have been
found: inverted, tandem and interdispersed. Only repeats >30 bp were considered. The most striking repeats are two large
inverted repeats located in the
GAP
A vicinity. The two repeated segments show 96% identity in a 792 bp overlap. One repeat is located in the intergenic region between ORF 08 and ORF
09 and includes the ORF 09 putative promoter region. The corresponding inverted segment is located downstream of
GAP
A and starts within the 3' non-coding trailer of that gene. The 5798 bp loop thus contains
GAP
A, ORF 09 and ORF 10 and the total region corresponds to 7645 bp end-to-end.
The second inverted repeat is located in the intergenic region between ORFs 13
and 14. The repeated segments are 76.5% identical over a 503 bp overlap with a
putative loop of 362 bp. Upstream of this inverted repeat, still in the same
intergenic region, is located the only large tandem repeat found in the 81 kb
contig. The two repeated segments are next to each other and present 68.1%
identity in a 503 bp overlap. The intergenic region between ORFs 13 and 14 is
relatively G+C poor (29.6%). The presence of these two features and the
relatively large AT content of this region may be related.
Another feature similar to those described above is also observed for a region located between ORFs 04 and 05. This region shows 69.5% identity in
a 1922 bp fragment with a region corresponding to part of the intergene space
of two genes,
RPL
12A and
RPL
12B, coding for chloroplast ribosomal proteins (
13
). The duplicate region is likewise AT-rich (G+C = 28.3%). It is not known whether
RPL
12A and
RPL
12B are located on chromosome III or another chromosome. Such long range
duplications/inversions have been described for large genomes, in particular
Caenorhabditis
elegans
(
14
).
Short repetitive sequences of the dinucleotide repeat type defined for
microsatellite motifs are found scattered in the intergenic and intronic
regions of the 81 kb contig. The three most significant of these AT
(15-17)
repeats are indicated in Figure
1
. Two of these AT repeats are located in ORF 04 and ORF 06 introns respectively.
The different motifs with a length of over six repeats were AA/TT, AT/TA and
CT/GA. This is in agreement with published data (
15
). These motifs were found in relatively equal abundance.
Expression
The number of
Arabidopsis
ESTs present in the databank is at this time >28 000 (release 052396). The data
represented by the collection of ESTs can be used to give a first insight into
the expression of a gene, by looking at the number of identical EST hits found
for each gene candidate (Table
2
). From this data it can be seen that five of the genes are relatively well
expressed. One of these genes,
GAP
A, coding for a key enzyme in chloroplast metabolism, is highly expressed, as
expected. Four genes are hit by only one EST. The last nine genes are
presumably rarely expressed under the conditions under which the cDNA libraries
were made. A pattern for trancription according to the location of genes on the
contig is not evident and physical closeness of the genes does not apparently
mean the same pattern of expression. The two genes
GAP
A (
17
) and ORF 19 (
18
,
19
), which have been experimentally demonstrated to be regulated by light, are >30
kb apart.
DISCUSSION
The sequencing presented here of a 81 kb contig from chromosome III gives a
first insight into the organization of a relatively large fragment of the
A.thaliana
genome.
The gene density of one gene every 4500 bp confirms estimations derived from other data for
Arabidopsis
(
7
) and is of the same order as one gene every 5 kb found in
C.elegans
, which has a genome size (100 kb) comparable with that of
Arabidopsis
. The gene density could reach a higher value on a small fragment. As an
example, four genes (ORFs 8-11) are concentrated within a 9000 bp fragment in the 81 kb contig. A
similar density has also been found in other parts of the
Arabidopsis
genome (
19
).
Many of the features described here for the
Arabidopsi
s 81 kb contig and relative to the contig structure and gene pattern have been
observed on a larger scale in the
Caenorhabditis
genome, of which 2.2 Mb have been sequenced and analysed. The similarities
include the organization of gene families, the presence of remnants of genetic
elements and, more importantly, the presence of complex duplicated and inverted
genomic pieces. The 81 kb fragment sequenced presents, however, its own
characteristics. All the genes identified as members of gene families are of
the dispersed type: no close family member is in direct vicinity on the contig.
Another characteristic is the apparently high number of genes without introns
(five out of 18).
At this point, the EST information generated by the EST sequencing effort and
the EST assembly done at TIGR (The Institute for Genome Research) can be used
to estimate the total number of genes in
Arabidopsis
. At the moment, there are 28 007 ESTs deposited in dbEST (release 052396). The
latest assembly (release 1.1) was performed on 25 282 ESTs, which after assembly gave a total of ~13 600 different sequences. Of the 18 genes identified in the contig, nine
were hit by ESTs, which means that by extrapolation 27 200 genes could be
present in the
Arabidopsis
genome. However, this estimation has to be taken with caution, due to the
probable over-representation in the EST database of highly expressed genes. The number
obtained in this way is close to the most recent estimation of 25 000 genes (
7
).
As mentioned earlier, most of the genes presenting matches with databases entries represent fundamental functions or mechanisms and present cross-phylum similarities. Some are highly conserved in distant phyla. ORF 01, an RNA helicase, is such an example, with 47
and 61.5% amino acid identity to its yeast and human genes respectively.
The availability of the complete genomes of three lower organisms in the
databases (see Introduction) makes it possible to infer which of the candidate genes may be specific to multicellular organisms. Eleven of the genes identified during this work do not show any
significant match with either bacterial or yeast ORFs in BLASTP searches (Table
2
) and might be good candidates for these functions. Two genes, ORFs 06 and 14, a
polygalacturonase and a serine-threonine kinase respectively, which belong to
large gene families widespread in plants (see Table
1
for highest scoring similar genes) do not show significant cross-phylum similarity and thus could possibly be considered as specialized for
plant functions. More interestingly, eight out of the nine remaining genes do
not show any similarities with vertebrate genes and may be putative candidates
for plant-specific genes. As an example, ORF 19, of unknown function, has been shown
to be regulated by light in rice (
20
) and in citrus (
18
) plants and is very likely plant specific. The complete sequence of another
organism of similar size to
Arabidospsis
is however needed before assigning plant specificity to an unknown gene.
To summarize, this work has allowed extension of the catalogue of genes detected
in
Arabidopsis
, with 17 novel genes and characterization of 11 cognate cDNAs. The function of
eight of these genes is totally unknown. Most of the genes described here have no
homologous function described in yeast or prokaryotes and might perform differentiated functions in multicellular
organisms and are of particular interest for more detailed study.
ACKNOWLEDGEMENTS
We thank M.Bevan for coordinating the ESSA Program, P.Slonimski, Director of the
GREG (Groupement de Recherches et d'Etudes sur les Génomes) for his support, H.W.Mewes, S.Klosterman and M.Chalwatzis from
the MIPS for computer data analysis, G.Clabault for computer assistance, Y.-F.Li for excellent sequencing support, J.Dangl and J.T.Mulligan for
supplying the
Arabidopsis
[lambda] GEM11 library, G.Picard and J.Lafleuriel for helping with the YAC
library hybridization, the
Arabidopsis
Biological Resource Center at Ohio State for supplying ESTs clones,
D.Tremoussaygue for the cell culture cDNA library and M.Kreis for the green
shoots cDNA library. This work was supported by the EU under the ESSA program
and by the the French government under the GREG program.
4 Pearson,W.R. and Lipman,D.S.(1988) Proc. Natl. Acad. Sci. USA, 85, 2444-2448.MEDLINE Abstract
5 Brown,C.M., Quigley,F.R. and Miller,W.A. (1995) Plant Physiol., 110, 336.
6 Lister,C. and Dean,C. (1995) Weeds World, 2, 11-18.
7 Meyerowitz,E.M. (1994) In Meyerowitz,E.M. and Somerville,C.R. (eds), Arabidopsis. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 89-120.
