An mRNA blueprint for C4 photosynthesis derived from comparative transcriptomics of closely related C3 and C4 species.

Abstract

C(4) photosynthesis involves alterations to the biochemistry, cell biology, and development of leaves. Together, these modifications increase the efficiency of photosynthesis, and despite the apparent complexity of the pathway, it has evolved at least 45 times independently within the angiosperms. To provide insight into the extent to which gene expression is altered between C(3) and C(4) leaves, and to identify candidates associated with the C(4) pathway, we used massively parallel mRNA sequencing of closely related C(3) (Cleome spinosa) and C(4) (Cleome gynandra) species. Gene annotation was facilitated by the phylogenetic proximity of Cleome and Arabidopsis (Arabidopsis thaliana). Up to 603 transcripts differ in abundance between these C(3) and C(4) leaves. These include 17 transcription factors, putative transport proteins, as well as genes that in Arabidopsis are implicated in chloroplast movement and expansion, plasmodesmatal connectivity, and cell wall modification. These are all characteristics known to alter in a C(4) leaf but that previously had remained undefined at the molecular level. We also document large shifts in overall transcription profiles for selected functional classes. Our approach defines the extent to which transcript abundance in these C(3) and C(4) leaves differs, provides a blueprint for the NAD-malic enzyme C(4) pathway operating in a dicotyledon, and furthermore identifies potential regulators. We anticipate that comparative transcriptomics of closely related species will provide deep insight into the evolution of other complex traits.

Full text

Page 1

An mRNA Blueprint for C4Photosynthesis Derived from
Comparative Transcriptomics of Closely Related C3and
C4Species1[W][OA]
Andrea Bra ¨utigam2, Kaisa Kajala2, Julia Wullenweber, Manuel Sommer, David Gagneul,
Katrin L. Weber, Kevin M. Carr, Udo Gowik, Janina Maß, Martin J. Lercher, Peter Westhoff,
Julian M. Hibberd2, and Andreas P.M. Weber2*
Institute of Plant Biochemistry (A.B., J.W., M.S., D.G., K.L.W., A.P.M.W.), Institute of Plant Molecular and
Developmental Biology (U.G., P.W.), and Institute of Informatics (J.M., M.J.L.), Heinrich-Heine University,
40225 Duesseldorf, Germany; Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA,
United Kingdom (K.K., J.M.H.); and Bioinformatics Core, Research Technology Support Facility, Michigan
State University, East Lansing, Michigan 48824 (K.M.C.)
C4photosynthesis involves alterations to the biochemistry, cell biology, and development of leaves. Together, these
modifications increase the efficiency of photosynthesis, and despite the apparent complexity of the pathway, it has evolved
at least 45 times independently within the angiosperms. To provide insight into the extent to which gene expression is altered
between C3and C4leaves, and to identify candidates associated with the C4pathway, we used massively parallel mRNA
sequencing of closely related C3(Cleome spinosa) and C4(Cleome gynandra) species. Gene annotation was facilitated by the
phylogenetic proximity of Cleome and Arabidopsis (Arabidopsis thaliana). Up to 603 transcripts differ in abundance between
these C3and C4leaves. These include 17 transcription factors, putative transport proteins, as well as genes that in Arabidopsis
are implicated in chloroplast movement and expansion, plasmodesmatal connectivity, and cell wall modification. These are all
characteristics known to alter in a C4leaf but that previously had remained undefined at the molecular level. We also document
large shifts in overall transcription profiles for selected functional classes. Our approach defines the extent to which transcript
abundance in these C3and C4leaves differs, provides a blueprint for the NAD-malic enzyme C4pathway operating in a
dicotyledon, and furthermore identifies potential regulators. We anticipate that comparative transcriptomics of closely related
species will provide deep insight into the evolution of other complex traits.
C4photosynthesis is a complex biological trait that
enables plants to either accumulate biomass at a much
faster rate or live in adverse environments compared
with “ordinary” plants (Hatch, 1987; Osborne and
Freckleton, 2009). These C4plants have added a CO2
concentration mechanism on top of their regular pho-
tosynthetic carbon fixation that makes them not only
more efficient at assimilating inorganic carbon; they
frequently also have higher water and nitrogen use
efficiencies (Black, 1973; Oaks, 1994; Osborne and
Freckleton, 2009). Beyond the basic biochemistry, our
understanding of C4photosynthesis is limited.
The principle of C4photosynthesis is deceivingly
simple: instead of using Rubisco as the primary
carbon-fixing enzyme, C4plants use phosphoenolpy-
ruvate carboxylase (PEPC). Unlike Rubisco, PEPC is
more specific for inorganic carbon (Hatch, 1987). Since
the C4cycle is an add-on rather than a replacement for
Rubisco and the Calvin-Benson cycle, the prefixed CO2
is transported in a bound form, a C4acid (hence the
name), to the site of Rubisco. The C4cycle generates
high concentrations of CO2around Rubisco (Hatch,
1987), and this increases the rate of photosynthesis
because competition between CO2and oxygen at the
active site of Rubisco is reduced (Jordan and Ogren,
1984). In most C4plants, concentrating CO2around
Rubisco involves the reactions of photosynthesis being
partitioned between bundle sheath (BS) and meso-
phyll (M) cells as well as changes to cell biology and
leaf development (Hatch, 1987; Sage, 2004), although
in some lineages, C4photosynthesis operates within
individual cells (Reiskind et al., 1989; Keeley, 1998;
Voznesenskaya et al., 2001, 2002, 2003).
In all known C4plants, CO2enters M cells and is
converted into bicarbonate by carbonic anhydrase.
1This work was supported by the German Research Council
(grant nos. WE 2231/4–1to A.P.M.W., SFB TR1 to P.W. and A.P.M.W.,
and IRTG 1525/1 to P.W. and A.P.M.W.) and the Leverhulme Trust
and Isaac Newton Trust (to J.M.H.).
2These authors contributed equally to the article.
* Corresponding author;e-mail andreas.weber@uni-duesseldorf.de.
The authors responsible for distribution of materials integral to
the findings presented in this article in accordance with the policy
described in the Instructions for Authors (www.plantphysiol.org)
are: Julian M. Hibberd (julian.hibberd@plantsci.cam.ac.uk) and
Andreas P.M. Weber (andreas.weber@uni-duesseldorf.de).
[W]The online version of this article contains Web-only data.
[OA]Open Access articles can be viewed online without a sub-
scription.
www.plantphysiol.org/cgi/doi/10.1104/pp.110.159442
142Plant Physiology?, January 2011, Vol. 155, pp. 142–156, www.plantphysiol.org ? 2010 American Society of Plant Biologists

Page 2

PEPC then combines HCO3
C4oxaloacetic acid, which is rapidly converted into
either Asp or malate. These C4acids then diffuse to the
site of Rubisco through abundant plasmodesmata,
where C4acid decarboxylases release CO2(Hatch,
1987). Three distinct C4acid decarboxylases, known as
NADP-dependent malic enzyme (NADP-ME), NAD-
dependent malic enzyme (NAD-ME), and PEP car-
boxykinase, have been coopted into the C4pathway,
and this has been used to define three biochemical
subtypes of C4photosynthesis. The three-carbon com-
pound released after decarboxylation diffuses back to
the M cells and is converted to PEP catalyzed by
pyruvate,orthophosphate dikinase (PPDK; Hatch and
Slack, 1968). Because the enzymes involved in the C4
cycle are found in the cytosol, chloroplasts, and mito-
chondria, a significant amount of transport across
organellar membranes is required for the C4cycle to
operate. However, few genes encoding transporters
that allow the increased intracellular flux of metabo-
lites required for C4photosynthesis have been identi-
fied (Bra ¨utigam et al., 2008a; Majeran and van Wijk,
2009). In addition, we have a very limited understand-
ing of the mechanisms controlling the altered cell
biology and morphology associated with C4leaves.
The C4cycle likely affects not only the relatively small
number of enzymes and transport proteins needed to
perform the core reactions but, given the consequences
on the ecological performance of the plants, also a
range of other processes.
The gaps in our understanding of the mechanisms
underlying C4photosynthesis limit insight into a met-
abolic pathway that has evolved repeatedly at least 45
times in plants (Sage, 2004) and so is of interest in terms
of understanding a remarkable example of convergent
evolution. In addition, because C4plants are among the
most productive on the planet and the pathway is
associated with increased water and nitrogen use effi-
ciencies (Brown, 1999), it has been suggested that
characteristics of C4photosynthesis should be placed
into C3crops (Matsuoka et al., 2001; Mitchell and
Sheehy, 2006; Hibberd et al., 2008). A more complete
understanding of genes involved in C4photosynthesis
is fundamental to attempts at placing components of
the C4pathway into C3crops to increase yield.
Recently, a new set of tools has become available to
analyze species without sequenced genomes on a ge-
nomic scale: next generation sequencing (NGS) technol-
ogy (summarized in Metzker, 2010). With NGS, the
transcriptome of a tissue can be sequenced and quan-
tified at the same time (RNA-Seq; Wang et al., 2009). The
454 FLX genome sequencer provides a quarter million
sequence reads of 230 bases in each run from a cDNA
template generated from mRNA (http://www.454.
com/; Metzker, 2010). The resulting reads can be
mapped onto a closely related reference to quantify
the number of reads matching a gene locus, thus pro-
viding a measure of transcript abundance (Flicek and
Birney, 2009; Bra ¨utigam and Gowik, 2010). We chose to
compare the C4plant Cleome gynandra with the C3plant
2with PEP to generate the
Cleome spinosa, since they are members of the same
genusandarecloselyrelatedtoArabidopsis(Arabidopsis
thaliana; Brown et al., 2005; Marshall et al., 2007). Given
the close phylogenetic relationship, we can take ad-
vantage of the well-annotated Arabidopsis genome
(Swarbreck et al., 2008) and its known genome history
(Bowers et al., 2003; Haberer et al., 2004; Thomas et al.,
2006) to identify and quantify the biological functions
regulated at the level of transcript abundance in the C4
species compared with the C3species. Although the
experiment will also capture variation in the abundance
of transcripts associated with differences between the
species that do not relate to C4photosynthesis, the close
proximity of the Cleome species should reduce this
effect. We chose to use mature fully differentiated leaves
for the analysis, since we wanted to minimize the
influence of species-specific effects during leaf differen-
tiation but rather focus on transcript profiles when C4
photosynthesis is fully operational. Once this profile is
defined, analysis of developmental stages may reveal
how the profile is achieved during differentiation.
By comparing the transcriptomes of closely related
C3and C4species, we will test (1) whether cross-
species transcriptomic comparisons are feasible, (2)
the degree to which the core C4cycle enzymes and
transport proteins are regulated at the level of tran-
script abundance, and (3) whether the changes in
metabolism associated with C4photosynthesis are
associated with additional unexpected shifts in tran-
script profiles in leaves of C4compared with C3plants,
and (4) define candidates for additional functions
critical to C4photosynthesis based on unbiased obser-
vation of the data. By analyzing the complete tran-
scriptome, we define the maximal extent to which the
C4pathway alters leaf transcript profiles.
RESULTS
Physiological Analysis of C3and C4Leaves Confirms C4
Metabolism in C. gynandra
To confirm that the C. spinosa and C. gynandra leaves
we used for transcriptomic analysis were using C3and
C4 photosynthesis, respectively, we analyzed the
steady-state levels of metabolites associated with the
C4cycle. For example, large quantities of Asp, Ala, and
pyruvate are produced in M and BS cells of NAD-ME
C4leaves, and they were 19, 3.9, and 3.6 times more
abundant, respectively, in C. gynandra compared with
C. spinosa (Supplemental Table S1). In contrast, and in
agreement with the lower demand for the photorespi-
ration in C4leaves, glycerate and glycolate, intermedi-
ates of the photorespiratory cycle, were 4.5 and 1.9
times more abundant in C. spinosa (Supplemental Ta-
ble S1). We also determined the extractable activities
of PEPC, aspartate aminotransferase (AspAT), NAD-
dependent malate dehydrogenase (NAD-MDH), NAD-
ME, and alanine aminotransferase (AlaAT). Except
for NAD-MDH, significantly higher activities of the
enzymes required for the C4cycle were measured in
An mRNA Blueprint for C4Photosynthesis
Plant Physiol. Vol. 155, 2011143

Page 3

C. gynandra leaf extracts (Supplemental Fig. S1).
The metabolite profiling of leaf extracts using gas
chromatography-electron impact-time of flight (GC-
EI-TOF) and the enzyme activity assays showed that
the plants we used for digital gene expression analysis
had clear differences in their metabolite profiles and
enzyme activities, and these were consistent with func-
tional C3and C4photosynthesis operating in leaves of
C. spinosa and C. gynandra, respectively.
The Leaf Transcriptomes for Closely Related C3and C4
Species Are Qualitatively Similar
To obtain sequence tags for digital gene expression
(DGE) analysis from C. spinosa (C3) and C. gynandra
(C4), RNA was isolated from mature leaves of each
species and prepared for 454 sequencing. One se-
quencing run on a Genome Sequencer FLX (GS FLX;
Roche) sequencing system was conducted on leaf
cDNA isolated from either C. gynandra or C. spinosa.
From C. spinosa, we obtained 70,564,592 nucleotides,
and from C. gynandra, 91,851,136 nucleotides of raw
sequence were obtained; after quality control, these
corresponded to 65,525,139 and 85,681,233 nucleo-
tides, respectively (Table I). The mean read length of
the cleaned sequence reads was 232 nucleotides for C.
gynandra and 230 nucleotides for C. spinosa (Table I).
To exclude program-specific mapping artifacts and
to test whether the C. gynandra and C. spinosa libraries
behave robustly during mapping, two different pro-
grams, BLAST and BLAT (BLAST-Like Alignment
Tool), were used to align the reads to Arabidopsis as
the reference genome. To define the most suitable
mapping parameters, an array of parameters for map-
pings in both the DNA and protein space were tested
(Table II). Neither the C. gynandra nor the C. spinosa
library mapped well to Arabidopsis cDNAs in the
DNA space using BLAT or BLAST, although the dif-
ferences are more dramatic for BLAT (Table II). In the
protein space, however, the proportion of mapped
reads increased dramatically. When 75% amino acid
sequence identity was required, three-quarters of the
reads could be mapped with BLAT, resulting in 1.48
and 1.57 average mappings per read, respectively.
Even with the most lenient mapping parameters, the
proportion of mapped reads did not exceed 83% with
BLATand 78.8% with BLAST (Table II). In all mapping
attempts, the C. gynandra and C. spinosa read libraries
yielded qualitatively similar mapping results, irre-
spective of mapping program or parameters.
To obtain a stringent yet inclusive mapping, the
mapping conducted in protein space at 75% or greater
identity with BLATwas chosen, and this mapping file
was parsed by in-house scripts to keep only the read
match with the highest number of matching bases. For
a more lenient mapping, a BLAST mapping at a cutoff
of 1e25was chosen and parsed to keep only the best
BLAST hit for each read. For each Arabidopsis Ge-
nome Initiative (AGI) code, the number of matching
reads was counted and the hit count was then trans-
formed to reads per million (RPM) to normalize for the
number of reads available for each species. After
parsing, the sequenced libraries matched between
50.5% and 55.3% of the genes in the Arabidopsis
reference (Supplemental Table S2).
To assess whether the data sets for the two different
species and the two different mappings were qualita-
tively similar, we tested the coverage of the functional
classes. Overall, about 50% of all genes were repre-
sented in both species with the BLAT (Fig. 1A) and the
BLAST mapping (Fig. 1B). Although the majority of
gene classes were represented by more than 50% of
genes in each class for both mappings, the classes
function unknown, putative lipid transfer protein,
storage protein, and defense were underrepresented
compared with all genes (Fig. 1). Genes present in the
organellar genomes were not well represented (Sup-
plemental Table S3). Genes classified into primary
metabolism including photosynthesis, central carbon,
nitrogen metabolism, amino acid, and nucleotide me-
tabolism as well as many cellular processes were well-
represented categories, and about four-fifths of genes
predicted to be involved in the C4pathway were
detected in both species. Overall, the pattern of detec-
tion in the different gene classes was similar for both
species and independent of the program used for the
mapping (Fig. 1).
Transcripts of Known C4Genes Are More Abundant with
One Exception
Detailed analysis of known C4genes showed that all
but one gene necessary for the core C4cycle of NAD-
ME-type plants were massively up-regulated in C.
gynandra compared with C. spinosa. Transcripts encod-
ing PEPC were up-regulated 78-fold, those encoding
AspAT were up-regulated 343-fold, the transcripts
for the two isoforms of NAD-ME were up-regulated
27- and 21-fold, respectively, and AlaAT were up-
regulated 29-fold (Table III). The results for the BLAT
Table I. Massively parallel signature sequencing allows large-scale
assembly of transcripts in both C. spinosa and C. gynandra after
comparison with the TAIR 8 Arabidopsis database
One GS FLX sequencing run allowed significant generation of
sequence for both species, and the vast majority of these could be used
to assemble contigs and then matched to Arabidopsis genes.
DataC. spinosaC. gynandra
Raw reads
Raw nucleotides
Raw mean length
Clean reads
Clean nucleotides
Clean mean length
Contigs
Total length (nucleotides)
Total reads
Percent assembled
313,807
70,564,592
402,674
91,851,136
225228
284,318
65,525,139
368,333
85,681,233
230 232
17,655
7,746,894
245,324
18,992
9,062,043
319,732
86.386.8
Bra ¨utigam et al.
144Plant Physiol. Vol. 155, 2011

Page 4

and the BLAST mappings were similar with one
exception. In the BLAST mapping, the reads mapping
to PEPC were split onto two genes in the Arabidopsis
reference genome, whereas they mapped to only one
gene in the BLAT mapping (Table III). Transcripts
encoding mitochondrial malate dehydrogenases were
increased only 1.3-fold (Supplemental Table S3). Not
only were genes associated with the C4pathway up-
regulated compared with C3, but they also had high
absolute read counts between 1,800 and 4,806 RPM.
The Leaf Transcriptomes for Closely Related C3and C4
Species Are Quantitatively Different
Before undertaking detailed analysis of differences
in transcript abundance between C. gynandra and C.
spinosa, we used quantitative (q)PCR to confirm esti-
mates of transcript abundance identified by RNA-Seq.
We chose genes whose transcript abundance differed
over 4 orders of magnitude and used qPCR to assess
their abundance. qPCR was performed on both the
cDNA used for RNA-Seq and cDNA generated from
RNA isolated from leaves in a separate experiment.
This approach provided strong support for the differ-
ences in abundance of transcripts between the two
species that we determined from RNA-Seq (Fig. 2).
Overall, this showed that the ratios of transcript abun-
dance obtained by RNA-Seq-based DGE are suitable
for calling differentially expressed genes between two
related species.
Of the 13,662 transcripts for which we captured
quantitative data (Supplemental Table S3), we identi-
fied 583 (BLAT) or 603 (BLAST) transcripts whose
abundance differed significantly (P # 0.01) between C.
spinosa and C. gynandra, with 256/258 (1.2%/1.2%)
transcripts being more abundant in C. gynandra and
Table II. Mapping the sequence reads with different BLAT and BLAST parameters to empirically
determine suitable mapping conditions
The percentage of AGI codes with at least one mapped read and the average mappings per read were
determined prior to parsing the tables to retain only the best match. Suitable mapping conditions are
printed in bold; for BLAT, the cutoff value is the minimal number of matching bases; for BLAST, it is the
minimal accepted e-value.
Mapping
Program
Library
Search
Space
Cutoff
Value
Percentage
Reads
with at Least
One Hit in the
Reference
Percentage
AGI
Codes with at
Least One
Mapped Read
Average
Mappings
per Read
BLATC. gynandra DNA60
75
85
90
25
50
75
80
60
75
85
90
25
50
75
80
40.9
40.7
30.2
7.7
82.6
82.6
75.4
56.4
40.8
40.6
29.7
8.5
83.0
83.0
76.0
57.9
68.9
58.8
29.6
9.9
78.0
67.8
29.0
0.1
69.7
59.6
29.4
9.8
78.8
68.3
29.3
0.1
42.0
41.7
35.8
19.5
70.4
70.4
62.6
52.2
38.9
38.5
32.3
17.1
67.7
67.7
58.9
48.4
56.5
49.1
30.5
15.9
76.9
71.0
39.5
0.3
53.0
46.3
28.3
14.4
75.3
68.7
36.0
0.3
1.19
1.19
1.15
1.09
2.35
2.35
1.48
1.27
1.29
1.28
1.21
1.15
2.49
2.46
1.57
1.32
30.7
27.7
18.9
11.5
106.6
64.6
22.9
7.5
28.2
25.1
16.2
9.8
93.7
56.4
21.2
4.6
Protein
C. spinosaDNA
Protein
BLASTC. gynandra DNA1e-05:
1e-10:
1e-30:
1e-50:
1e-05:
1e-10:
1e-30:
1e-50:
1e-05:
1e-10:
1e-30:
1e-50:
1e-05:
1e-10:
1e-30:
1e-50:
Protein
C. spinosaDNA
Protein
An mRNA Blueprint for C4Photosynthesis
Plant Physiol. Vol. 155, 2011 145

Page 5

327/345 (1.5%/1.6%) transcripts being more abundant
in C. spinosa (Fig. 3, “all”). We tested whether signif-
icantly changed transcripts are enriched in functional
categories and whether they were more highly ex-
pressed in the C4or the C3species. While the qualita-
tive classification of detected genes showed a very
similar pattern between C. spinosa and C. gynandra
(Fig. 1), the quantitative analysis revealed massive
differences in representation between gene classes in
the C3and the C4species (Fig. 3). The transcript profile
generated by the BLAT mapping (Fig. 3A) is similar to
the one generated by the BLAST mapping (Fig. 3B),
although not all genes called as significantly regulated
were identical (Supplemental Table S3). The classes
containing the highest percentage of changed genes
are the photosynthetic classes as well as the C4cycle,
Calvin-Benson cycle, and photorespiration (Fig. 3).
The latter two have lower steady-state mRNA levels in
C4leaf tissue (Fig. 3, bottom), while the photosynthetic
classes of PSI, cyclic electron flow, and cytochrome b6/f
complex as well as the C4cycle have higher levels
in C4leaf tissue (Fig. 3, top). A number of classes
involved in primary metabolism also have lower
steady-state transcript levels in C4tissues: one-carbon
compound metabolism, other central carbon metabo-
lism, shikimate pathway, and amino acid metabolism.
Protein synthesis also has lower steady-state tran-
script levels, which are limited to cytosolic and
plastidic protein synthesis genes (Supplemental Fig.
S3). Among the classes with higher steady-state
transcript levels are starch metabolism, cofactor syn-
thesis, putative lipid transfer proteins, nitrogen
metabolism, and b-1,3 glucan metabolism. The quan-
titative pattern (Fig. 3) is similar to the qualitative
pattern (Fig. 1) with regard to the influence of the
mapping program; the BLAT and BLAST mappings
look remarkably similar with the exception of shiki-
mate metabolism.
Figure 1. The qualitative patterns of transcript abundance between C. gynandra and C. spinosa are very similar, with the same
classes underrepresentedand overrepresented in both libraries. A, Analysisbasedon BLAT mapping.B, Analysisbasedon BLAST
mapping. Black bars refer to the C4plant C. gynandra, and white bars refer to the C3plant C. spinosa.
Bra ¨utigam et al.
146Plant Physiol. Vol. 155, 2011

Page 6

Transcripts with Similar Patterns of Abundance
Compared with Bona Fide C4Genes and Rubisco
The list of 13,662 transcripts detected in either C.
spinosa or C. gynandra tissues and the list of 603
transcripts that are differentially regulated between
both species (Supplemental Table S3, BLAST map-
ping) prompted us to determine which transcripts
showed changes in abundance similar to the core C4
genes or Rubisco subunit-encoding genes. Such tran-
scripts display both a large fold change between the C3
and the C4plants and largeabsolute read numbers. For
example, among the transcripts encoding putative
transport proteins, three plastidic transport proteins,
the PEP phosphate translocator PPT, a putative bile
acid:sodium symporter, and a putative proton:sodium
antiporter, two mitochondrial dicarboxylate carriers,
and one plasma membrane intrinsic protein were
massively up-regulated in C4C. gynandra (Table IV).
No transcripts encoding transport proteins were
found to be down-regulated to a comparable degree.
Among metabolic genes, two cytosolic carbonic anhy-
drases, one of which (CA4; Table IV) is likely tethered
to the plasma membrane, an adenylate kinase, and a
pyrophosphatase were up-regulated at levels compa-
rable to those of C4cycle genes. Many proteins of
unknown function showed differential expression, the
most striking case being a putative lipid transfer
protein, also annotated as an extensin-like protein.
Based on annotation and differential expression pat-
tern, several transcripts predicted to encode known C4
functions that have not yet been assigned to genes,
such as CHLOROPLAST UNUSUAL POSITIONING1
(CHUP1) and actin for chloroplast positioning or
callose-degrading enzymes for regulating plasmodes-
matal opening, were identified (Table IV).
Regulatory Genes That Are Significantly Changed
The transcript profiles of these C3and C4species
identify a number of regulatory proteins that are
candidates for maintaining C4status. Among tran-
scripts encoding proteins with regulatory functions, 43
were significantly up-regulated in either C. gynandra
or C. spinosa (Fig. 3). These include bona fide tran-
scription factors, protein phosphatases and kinases,
and the regulatory proteins of the pyruvate dehydro-
genase complex (up-regulated in C4), of PPDK (up-
regulated in C4), and of Rubisco (down-regulated in
C4). Only 17 transcription factors are significantly
changed; seven of those have higher steady-state
mRNA levels compared with the C3leaf tissue, while
10 have lower steady-state mRNA levels (Table V).
In addition to the detailed quantitative and qualita-
tive analysis of read mappings to generate ESTs for
both species, contigs were assembled from cleaned
reads for each species as described previously (Weber
et al., 2007; Bra ¨utigam et al., 2008b) and then annotated
by BLASTX versus The Arabidopsis Information Re-
source (TAIR) 9 protein models. A total of 18,992 and
17,655 contigs representing total sequence lengths of
9,062,043 and 7,746,894 nucleotides were obtained for
C. gynandra and C. spinosa, respectively (Table I).
Table III. Transcript abundance of C4cycle genes that have significantly higher transcript abundance in C4leaf tissue
Asterisks denote changes significant only in BLAST mapping.
EnzymeLocus
BLAT MappingBLAST Mapping
C. gynandra RPMC. spinosa RPMFold ChangeC. gynandra RPM C. spinosa RPMFold Change
AspAT
PPDK
PEPC
AlaAT
NAD-ME1
NAD-ME2
PEPC kinase
NADP-ME*
PEPC*
PPDK regulatory protein*
AT2G30970
AT4G15530
AT2G42600
AT1G17290
AT4G00570
AT2G13560
AT1G08650
AT1G79750
AT1G53310
AT4g21210
4,806
3,262
9,702
7,610
1,357
1,800
230
227
14
14
343.3
233.0
78.2
28.5
26.6
20.7
6.2
3.8
0.4
4.6
4,601
3,216
8,321
7,242
1,326
1,723
226
216
950
198
18
13
257.9
240.3
49.1
28.0
27.0
20.3
6.3
4.8
5.0
7.3
124
267
51
87
37
60
248
32
169
259
49
85
36
45
192
27
94
148
Figure 2. Massively parallel sequencing of mRNAs (RNA-Seq) and
qPCR generate similar profiles of transcript abundance in C. gynandra
and C. spinosa. Ratios of transcript abundance in C. gynandra and C.
spinosa were calculated, and transcripts selected for this analysis
spanned 4 orders of magnitude. CA, Carbonic anhydrase; PPCk, PEPC
kinase; LHCA, light-harvesting complex subunit A; RbcS1a, ribulose
bisphosphate carboxylase oxygenase 1a; RCA, Rubisco activase. Black
bars represent data from RNA-Seq, and white bars represent data from
qPCR. The horizontal dashed line represents a ratio of 1 and indicates
no difference in transcript abundance between the two species.
An mRNA Blueprint for C4Photosynthesis
Plant Physiol. Vol. 155, 2011147

Page 7

DISCUSSION
Transcriptomic Comparisons of Different Species with
NGS Technology Are Feasible
Read mapping by alignment is a well-established
tool to quantify transcript abundance and thus deter-
mine mRNA steady-state levels (Wall et al., 2009;
Metzker, 2010). The concept of mapping to a cross-
species reference has also been established theoreti-
cally (Palmieri and Schlotterer, 2009), although the
potential has not been experimentally explored to date
(Bra ¨utigam and Gowik, 2010).
To explore cross-species mapping,the transcriptome
sequencing was carried out using 454 FLX, a long-read
technology, since theoretical work had established that
at least BLATis capable of mapping reads that contain
alterations in comparison with the reference if the reads
are at least 100 bases long (Palmieri and Schlotterer,
2009). We also established a reference database, which
removes the genome history of Arabidopsis as far as
it is known (Bowers et al., 2003; Haberer et al., 2004;
Thomas et al., 2006). Tandem duplicated genes and
segmentally duplicated genes (remnants of the last
whole genome duplications) were removed to pre-
vent genome history from interfering with compara-
tive quantitative mapping (Bra ¨utigam and Gowik,
2010).
Both BLATand BLAST mappings indicate that using
a minimal reference does not diminish read mappings
(Supplemental Table S4) while avoiding mapping
problems based on genome history (Bra ¨utigam and
Gowik, 2010). The mappings in protein space allowed
more successful read mappings, because protein se-
quences diverge more slowly than nucleotide se-
quences. Although the proportion of reads mapped
varied with changing mapping parameters (Table II;
Figure 3. The quantitative patterns of transcript accumulation in C. gynandra and C. spinosa are distinct. A, Analysis based on
BLAT mapping.B, AnalysisbasedonBLASTmapping. Shown arethepercentagesofgeneswithsignificantly higherabundanceof
transcripts in C4(red bars), unchanged (white bars, including genes not detected), and significantly lower abundance of
transcripts in C4(blue bars) based on the total number of genes in each annotation class (in parentheses on the y axis).
Bra ¨utigam et al.
148Plant Physiol. Vol. 155, 2011

Page 8

Supplemental Table S4), the C. spinosa and C. gynandra
libraries yielded similar results, indicating that, evo-
lutionarily, both species are approximately equally
distant from Arabidopsis, with mapping incurring
similar penalties depending on parameters.
Since no read alignment program has emerged as the
consensus program for NGS data analysis, two differ-
ent programs were used for mapping and the output
was compared in all cases. The output proved robust
against changing the mapping program both qualita-
tively and quantitatively. When we mapped the quarter
million reads obtained from each species of Cleome to a
minimized TAIR 9 release of the Arabidopsis genome,
they corresponded to approximately 11,000 loci. As
theminimizedTAIR9datasetcontains21,972geneloci,
the reads we collected in C. gynandra and C. spinosa
represent approximately 50% of the transcriptome.
In Arabidopsis seedlings, approximately 60% of the
loci represented in the TAIR 8 release were detectable
(Weber et al., 2007); hence, we have likely captured a
large proportion of the transcripts associated with
leaves of C. spinosa and C. gynandra.
The qualitative representation of gene classes de-
tected reflects that leaf tissues were analyzed. While
photosynthetic genes as well as primary metabolism
are well represented in all data sets, genes implicated
in cell walls, secondary metabolism, and defense re-
sponses are underrepresented (Fig. 1). These classes
contain genes that are likely specific to certain tissues,
developmentalstages,orenvironmentalchallenges.For
example, cell wall genes may be better represented if
our sampling had included expanding leaf or stem
material(Schmidetal.,2005),and stress-responsegenes
may be better represented if plants were sampled after
exposure to extreme conditions (Kilian et al., 2007).
Likewise, certain pathways of secondary metabo-
lism are likely restricted to defined tissues or devel-
opmental stages, making it unlikely that we would pick
up many of these genes when profiling leaf libraries.
Based on the gene detection pattern, the two plant
species did not encounter different biotic or abiotic
stresses or were not in different stages of growth, as
very similar genes were detected in both species (Figs.
1 and 3).
Finally, only a very small proportion of transcripts
showed significant differences in abundance between
the two different species (Supplemental Tables S2 and
S3), and these changes were enriched in a limited
number of functional classes (Fig. 3). We conclude that
cross-species mapping in protein space is a feasible
strategy to compare different species as long as an
equidistant reference is available.
Transcripts Derived from Core C4Cycle Genes Are More
Abundant in the C4Species
C4photosynthesis has evolved convergently in
many different lineages of plants (Sage, 2004), and in
many cases the alterations to expression of specific
genes has been related to transcriptional regulation
(summarized in Sheen, 1999). Our genome-scale anal-
ysis allowed us to compare the steady-state transcript
levels for all candidate C4genes at the same time. For
all of the enzymes where a change in total extractable
activity could be shown (Supplemental Fig. S1), a
higher mRNA level of at least one isoform as judged
from the read count was also present (Table III). The
only enzyme showing no changes in transcript level is
the mitochondrial NAD-MDH. Possibly, the activity of
the mitochondrial NAD-MDH is high enough already
Table IV. Transcript abundance of selected genes with an expression similar to that of C4cycle genes and Rubisco
All changes are significant at P # 0.01. n/a, Not available.
FunctionLocus Annotation (TAIR 9)C. gynandra RPMC. spinosa RPMRatio
Transport proteins
AT2G26900
AT2G22500
AT4G24570
AT2G45960
AT5G33320
AT1G49810
Bile acid:sodium symporter family protein
Mitochondrial dicarboxylate carrier
Mitochondrial dicarboxylate carrier
Plasma membrane intrinsic protein subfamily protein
Phosphoenolpyruvate/phosphate translocator
Member of Na+/H+antiporter family
4,774
324
148
2,686
1,955
1,321
5586.8
n/a
n/a
20.2
20.2
15.9
0
0
133
97
83
Metabolism
AT3G52720
AT1G23730
AT5G35170
AT5G09650
a-Carbonic anhydrase 1
b-Carbonic anhydrase 4
Adenylate kinase family protein
Inorganic pyrophosphatase
227
497
152
87
235
833
1.5
5.7
8.5
3.2
1,994
2,664
Proteins of unknown function
AT1G12090
Callose-degrading enzymes
AT3G57240
AT1G32860
AT5G42100
Cell biology
AT3G25690
AT3G12110
Extensin-like protein (ELP)6,27814742.7
Member of glycosyl hydrolase family 17, likely b-1,3 glucanase
Member of glycosyl hydrolase family 17, likely b-1,3 glucanase
Plasmodesmal-associated b-1,3-glucanase
436
50
173
0
0
n/a
n/a
5.432
CHUP1
ACTIN
22
122
170
727
0.13
0.2
An mRNA Blueprint for C4Photosynthesis
Plant Physiol. Vol. 155, 2011 149

Page 9

in C3plants to support a C4-type metabolic flux. The
only transport protein known to date that is involved in
the C4cycle, the PEP phosphate translocator (Fischer
et al., 1997; Bra ¨utigam et al., 2008a), is also up-regulated
20-fold, indicating that this transport protein is regu-
lated at the level of mRNA abundance. Based on
similarities in transcript abundance to known C4
genes, our comparative RNA-Seq also identified likely
additional components needed for C4photosynthesis.
When PPDK was characterized, it was proposed that
adenylate kinase aswell as inorganic pyrophosphatase
need to be abundant in C4chloroplasts (Hatch and
Slack, 1968). RNA-Seq confirmed this prediction and
showed that the up-regulation also occurs at the level
of transcript abundance. Taken together, we found that
almost all transcripts encoding the proteins required
for the core C4cycle have higher steady-state mRNA
levels, and we propose that, at least in C. gynandra, the
activity of C4cycle enzymes and transport proteins is
controlled at least partially at the level of transcript
abundance.
Alterations to the Abundance of Transcripts Associated
with Other Metabolic Processes
Changes in the abundance of transcripts that are not
associated with the core C4cycle are also detectable in
leaves of C. gynandra and C. spinosa. The high-flux C4
cycle poses additional demands in terms of ATP and
reduction equivalents on the light reaction (Hatch,
1987). Specifically, the recycling of the initial CO2
acceptor PEP requires additional ATP molecules
(Hatch, 1987). In C4leaf tissue, one-third to one-half
of the genes in the photosynthetic gene classes that
contribute to ATP production by cyclic electron flow
are up-regulated compared with C3leaf tissue: PSI, the
cytochrome b6/f complex, and the genes mediating
cyclic electron flow themselves (Fig. 3). It remains an
open question whether these higher steady-state levels
are caused by higher ATP demand or whether C4
photosynthesis requires up-regulation of these genes
to meet the ATP demand prior to establishing C4
photosynthesis.
On the other hand, the classes of Calvin-Benson
cycle genes and photorespiratory genes are those with
the highest number of genes with significantly lower
steady-state mRNA levels. It is a well-established fact
that most C4plants have less Rubisco protein com-
pared with C3plants (Ku et al., 1979) and that flux
through the photorespiratory pathway is reduced
compared with C3species (Chollet and Ogren, 1975;
Leegood, 2002). Transcripts encoding the large and
small subunits of Rubisco were reduced from 22,968
and 15,442 RPM to 6,984 and 4,900 RPM in C. spinosa
and C. gynandra, respectively. Overall, the trend for
Calvin-Benson cycle genes was for them to be down-
regulated in C. gynandra compared with C. spinosa (Fig.
3). Likewise, a large number of genes encoding photo-
respiratory proteins, proteins involved in one-carbon
compound metabolism, and the genes involved in
ammonia reassimilation, Gln synthetase, and Glu syn-
thase have lower steady-state transcriptional levels
(Fig. 3; Supplemental Table S3). The reduced flow
through the photorespiratory pathway obviously de-
creases the demand on the expression system to main-
tain high steady-state levels of mRNA for many
Calvin-Benson cycle and photorespiratory genes. The
photosynthetic genes, the Calvin-Benson cycle and
photorespiratory genes (in C3), and the C4cycle genes
(in C4) are those with the highest read counts of the
genes with known function (Supplemental Table
S3). Although it is currently not possible to quantify
Table V. Transcription factors that are significantly changed between the leaf tissue samples
Asterisks denote changes significant only in BLAST mapping. n/a, Not available.
Locus
Transcription Factor
Type
BLAT MappingBLAST Mapping
Segmentally
Duplicated?
C. gynandra RPMC. spinosa RPMRatioC. gynandra RPMC. spinosa RPM Ratio
AT1G25560
AT5G07580
AT1G53910
AT5G10570
AT3G21330
AT3G62420
AT2G20570
AT1G72030
AT2G22430
AT1G10200
AT4G30410
AT1G32700
AT5G02810
AT2G36990
AT1G48500
AT1G17380
AT3G02790
AP2-EREBP
AP2-EREBP
AP2-EREBP*
bHLH
bHLH*
bZIP
G2-like
GNAT
HB
LIM
Not specified*
PLATZ
Pseudo ARR-B
Sigma70-like
Tify
Tify*
Zinc finger
176
223
32
919.6
4.4
0.2
n/a
n/a
0.1
n/a
0.1
4.9
0.1
n/a
19.6
n/a
n/a
0.1
0.2
4.3
219
292
84
924.3
8.1
0.3
n/a
n/a
0.1
n/a
0.0
4.4
0.1
n/a
28.8
0.1
n/a
0.1
0.1
3.6
Yes
Yes
Yes
Yes
51
138
83
74
138
36
268
112
107
138
0
0
0
0
1110
220
11
515
22
0292
10
505
21
0
179
106
230
32
330
116
205
76
Yes
00
1769115
10
143
10
24
407
4
0106112
130
11
18
374
00
147
110
87
174
161
112
Yes
Yes
Yes
Bra ¨utigam et al.
150Plant Physiol. Vol. 155, 2011

Page 10

absolute transcript levels, since the genome of neither
Cleome species has been sequenced, the high read
counts obtained for the genes of central carbon me-
tabolism and photosynthesis indicate that the steady-
state levels of transcripts are high. Since the most
altered gene classes are also those that contain the
genes with the highest absolute read counts, it is not
clear whether C4photosynthesis lowers or raises the
demand on protein synthesis and accessory pathways
such as amino acid synthesis. However, both the
protein synthesis and the amino acid metabolism
classes contain more genes that have lower steady-
state levels in C4leaf tissue (Fig. 3). Within the protein
synthesis gene class, many transcripts encoding struc-
tural components of plastidic and cytosolic ribosomes
were reduced (Supplemental Fig. S3). This was not the
case for components of mitochondrial ribosomes (Sup-
plemental Fig. S3), indicating that there is not a general
effect on translation but that the effect is likely specific
to ribosomes involved in translation for the Calvin-
Benson cycle and photorespiration. The protein-to-
fresh weight ratio is also lower in C4leaf tissue
compared with C3leaf tissue (Supplemental Fig. S2).
We propose that plastidic ribosomes are relieved of the
high translation load associated with the large subunit
of Rubisco and that the cytosolic ribosomes need to
translate fewer transcripts associated with central
carbon metabolism as well as the small subunit of
Rubisco. The reduced production of proteins in the
leaves of C4plants is considered important in increas-
ing nitrogen use efficiency, because the rate of photo-
synthesis per unit of nitrogen in the leaf is increased
(Oaks, 1994). Our data indicate that there is also likely
a significant saving in the nitrogen provision in the
leaf, because fewer ribosomes as well as fewer proteins
for central carbon metabolism are required.
The data set contains two additional gene classes,
b-1,3 glucan metabolism and putative lipid transfer
proteins, that showed differences in transcript abun-
dance between C. gynandra and C. spinosa that could be
explained within the current framework of knowledge
of C4photosynthesis. The C4pathway requires effi-
cient exchange of metabolites between M and BS cells
via large numbers of plasmodesmata connecting both
cell types, while the BS cell wall of many C4plants is
suberized to reduce diffusion of CO2away from
Rubisco (Hatch, 1987). Transcripts encoding three dis-
tinct glucan 1,3-b-glucosidases (Table IV) involved in
governing plasmodesmatal conductivity by regulating
the turnover of the b-1,3-glucan callose (Levy et al.,
2007) were up-regulated in leaves of C. gynandra
compared with C. spinosa. Therefore, it is possible
that these genes are involved in increasing the open
probability of plasmodesmata (Roberts and Oparka,
2003), which allows the efficient flux of organic acids
between M and BS cells required during C4photosyn-
thesis (Evert et al., 1977; Botha, 1992; Roberts and
Oparka, 2003). A transcript annotated as a putative
lipid transfer protein is among those that are most
highly up-regulated in C. gynandra compared with C.
spinosa. Lipid transfer proteins are required for the
export of lipids to the cell wall during cutin biosyn-
thesis (DeBono et al., 2009). Interestingly, in Arabi-
dopsis, some lipid transfer proteins are exclusively
and abundantly expressed in the root endodermis,
where suberin biosynthesis is required to establish the
Casparian strip.
There are additional changes in the transcript profile
that are less easily explained. Among the gene classes
containing more genes with significantly higher tran-
script levels in C4leaf tissue are starch metabolism,
cofactor synthesis and nitrogen metabolism, and heat
shock/protein folding (in order of decreasing number
of significantly different genes). On the other hand, it
is difficult to conceive why genes involved in metal
handling are frequently lower in transcript level in C4
leaf tissues (Fig. 3). These changes may be connected
to currently unknown phenomena relating to the C4
pathway or may be part of differences not relating
to C4photosynthesis between the two species. Overall,
the global analysis of transcription on the level of
functional classes reveals unexpected shifts in tran-
script profiles that can be explained based on the
current knowledge about the C4pathway, while a
range of smaller changes remain enigmatic.
Finally, our global transcriptional analysis of C4and
C3leaf tissues not only allows testing hypotheses
about the C4pathway on a global scale but also allows
genes with expression patterns similar to those of
known C4genes to be identified. The phylogenetic
proximity of the Cleomaceae to Arabidopsis allows the
identification of the orthologs in Arabidopsis, which
will facilitate translational research into the model
species (Brown et al., 2005).
Candidates for Additional C4-Related Genes
The identification of transport proteins involved in
the C4cycle lags behind that of enzymes, considering
that the C4cycle requires the intracellular transport of
pyruvate, PEP, Asp, and Ala across different organ-
ellar membranes (Bra ¨utigam and Weber, 2011). Awide
range of C4plants take up pyruvate into chloroplasts
from the M in cotransport with sodium (Aoki et al.,
1994; Aoki and Kanai, 1997), which might explain the
requirement for sodium as a micronutrient in many C4
species (Brownell and Crossland, 1972). Since the rate
of pyruvate transport into C4M cell chloroplasts
occurs at or exceeds the apparent rate of CO2assim-
ilation, sodium-coupled pyruvate import implies a
large influx of sodium into these chloroplasts, but the
transporter has not yet been identified at the molecular
level (Aoki and Kanai, 1997). Our finding that a
putative plastidic proton:sodium symporter (NHD1)
is 16-fold up-regulated in C. gynandra prompts us to
hypothesize that it functions in exporting sodium from
the chloroplast in order to maintain the sodium gra-
dient required for import of pyruvate. In addition, we
found strong up-regulation of a putative bile acid:
sodium cotransporter in C. gynandra. Interestingly, up-
An mRNA Blueprint for C4Photosynthesis
Plant Physiol. Vol. 155, 2011151

Page 11

regulation of the putative bile acid:sodium cotrans-
porter or of NHD1 was not observed in maize (Zea
mays; Bra ¨utigam et al., 2008a), which belongs to a
group of C4plants that show proton-dependent, not
sodium-dependent, transport of pyruvate into M cell
chloroplasts (Aoki et al., 1994; Aoki and Kanai, 1997).
PEP generated from pyruvate in M cell chloroplasts is
exported from these chloroplasts by PPT, thereby
providing the substrate for the cytosolic PEPC reac-
tion. Accordingly, transcripts encoding PPTare 20-fold
up-regulated in C. gynandra, likely reflecting the in-
creased requirement for transport of PEP (Table III). In
contrast to what has been observed for the NADP-ME-
type C4plant maize by quantitative proteomic analysis
(Bra ¨utigam et al., 2008a), we did not detect increased
transcript abundance of the putative M chloroplast
oxaloacetate/malate exchanger DiT1 (Taniguchi et al.,
2002, 2004; Renne et al., 2003; Supplemental Table S3).
This is consistent with the fact that oxaloacetic acid/
malate shuttling across the M cell chloroplast envelope
membrane is not required for NAD-ME-type C4pho-
tosynthesis(WeberandvonCaemmerer,2010;Bra ¨utigam
and Weber, 2011). The mitochondrial dicarboxylate
carriers are prime suspects for the C4acid importer
into the mitochondria, where decarboxylation takes
place (Table IV). The initial uptake of inorganic carbon
and its conversion to bicarbonate may be facilitated by
the concerted action of a membrane intrinsic protein
channeling the gas and a carbonic anhydrase that is
predicted to be membrane bound (Table IV).
Chloroplasts in the BS of C. gynandra are larger than
those in the BS of C3species and, as in many other C4
plants, are positioned in a strictly centripetal pattern
(Marshall et al., 2007; Voznesenskaya et al., 2007).
Transcripts derived from the GIANT CHLOROPLAST1
(GC1) gene were more abundant in C. gynandra than in
C. spinosa (Table IV). Although overexpression of GC1
in Arabidopsis is reported not to effect chloroplast
division (Maple et al., 2004), it is possible that it does
so in C. gynandra. In addition, we also detected re-
duced accumulation of transcripts derived from the
CHUP1 and ACTIN11 genes. In Arabidopsis, the outer
chloroplast envelope membrane protein CHUP1 con-
tains an actin-binding motif and is required for pre-
venting chloroplast aggregation (Oikawa et al., 2003).
Differential positioning of chloroplasts in BS and M
cells of the C4plants finger millet (Eleusine coracana)
and maize requires the actomyosin system (Kobayashi
et al., 2009). Since AtCHUP1 is involved in positioning
chloroplasts at the periclinal plasma membrane dur-
ing the weak-light acclimation response via a coiled-
coil domain and interaction with the cytoskeleton
(Oikawa et al., 2003), it is possible that the centripetal
positioning of chloroplasts in BS cells is linked to lower
expression of the CgCHUP1 and ACTIN11 genes.
Controlling and Maintaining a C4State in Leaf Tissue
Our estimate that around 603 transcripts accumulate
differentially in leaves of C3and C4species provides
insight into the extent to which gene expression pro-
files change in C4leaves. For example, the fact that 258
transcripts were more abundant in the leaves of C4
compared with C3species indicates that about 2.8% of
the leaf transcriptome differentially accumulates in C4
leaves (Supplemental Tables S2 and S6). To compare
the complexity of the C4pathway with other multi-
genic traits, we assessed the number of transcripts that
are known to be regulated by sugars, cold, diurnal and
Table VI. Comparison of alterations in transcript abundance in C4and C3leaves with those induced by
cold, sugar feeding, attack by pests or pathogens, diurnal changes to light, or circadian rhythms
Cause Estimated Change in TranscriptomeChange Reference
%
2.1 Cold treatment
C4leaves and C3leaves
Glc feeding
Pseudomonas syringae
Myzus persicae
Diurnal regulation
Circadian regulation
514 (24,000) ATH1
583/603 (13,443/13,662)
978 (22,500) ATH1
2,034 (23,750) ATH1
2,181(23,750) ATH1
1,115 (11,521) cDNA array
2,282 (18,890) Galbraith
Vogel et al. (2005)
This study
Price et al. (2004)
De Vos et al. (2005)
De Vos et al. (2005)
Schaffer et al. (2001)
Dodd et al. (2007)
2.7/2.8
4.4
8.6
9.1
11
12
Figure 4. Schematic of components associated with the C4cycle in the
NAD-ME subtype based on interpretation of RNA-Seq. Proteins that
have been described previously are in gray, and novel proteins are
marked in red. Metabolites are in black. PIP1B:CA4, PIP1B plasma
membrane aquaporin:membrane-tethered carbonic anhydrase; OAA,
oxaloacetic acid; ACT11-CHUP11, ACTIN11-CHUP1 complex; Pyr,
pyruvate; OEP24, chloroplast outer envelope protein 24.
Bra ¨utigam et al.
152Plant Physiol. Vol. 155, 2011

Page 12

circadian rhythms, as well as attack by pests and
pathogens (Table VI). Interestingly, the alterations in
transcript abundance of leaves of C. gynandra com-
pared with those of C. spinosa were greater than those
observed in response to cold treatment and lower than
those induced by Glc feeding, those occurring during
pathogen attack, and the response to both diurnal and
circadian rhythms. As significant progress has been
made in understanding sugar signaling (Rolland et al.,
2006), pathogen attack (Wise et al., 2007), and the
control of gene expression in response to the diurnal
cycle and circadian rhythms (Imaizumi et al., 2007), it
should be possible to identify the regulators responsi-
ble for these alterations in transcript abundance in a C4
leaf compared with a C3leaf. The changes in transcript
abundance that we document in a C4leaf compared
with a C3leaf likely overrepresent the changes in
transcript abundance actually associated with C4pho-
tosynthesis on a whole leaf basis, as some differences
in gene expression are likely due to the phylogenetic
distance between C. gynandra and C. spinosa. A more
confident estimate of the extent to which the leaf
transcriptome is altered in association with C4photo-
synthesis will be generated when additional conge-
neric pairs of C3and C4species are subjected to deep
transcriptome analysis and shared transcripts are
identified. Between M and BS cells, the alterations in
gene expression may be greater than those that we
have defined for whole leaves. For example, up to 18%
of genes are estimated to be differentially expressed
between M and BS cells of maize (Sawers et al., 2007).
However, it is not clear how different the transcript
profiles of M and BS cells are in a dicot C3leaf, and
until this is defined, it is not possible to infer the extent
to which transcript abundance alters in these cell types
in association with C4photosynthesis.
As we sampled from mature leaves to capture the
differences between C3and C4leaves at the point of
fully differentiated pathways, we likely also captured
regulatory genes needed to maintain C4architecture
and metabolism in mature leaves. Of the 17 transcrip-
tion factors significantly altered (Table V), GOLDEN2-
LIKE1 (GLK1) has previously been implicated in reg-
ulating genes important in C4photosynthesis. In
maize, GOLDEN2 controls functional differentiation
of chloroplasts in BS cells (Langdale and Kidner, 1994),
and GLK1 has been implicated in the expression of
photosynthesis genes in M cells (Rossini et al., 2001).
The fact that GLK1 transcripts are significantly more
abundant in leaves of C. gynandra would not neces-
sarily be predicted, as previous work indicates that it
becomes specialized in BS cells of C4leaves but not
that its abundance is altered significantly. This implies
that the increase in abundance of GLK1 transcripts
may not simply be due to its involvement in C4
photosynthesis. When overexpression of GLK1 was
induced in Arabidopsis, the abundance of 114 tran-
scripts was altered (Waters et al., 2009). We assessed
the extent to which the genes that are controlled by
GLK1 change in abundance in leaves of C. gynandra
compared with C. spinosa and found that only 19 genes
were shared between the two data sets. This may be
due to a number of factors that could include the
following: that there are differences in the targets of
GLK1 in Arabidopsis and C. gynandra; that a number
of other transcriptional regulators are more important
than GLK1 in maintaining patterns of photosynthesis
gene expression in C. gynandra; and that a rapid
induction of GLK1 gene expression has more impact
than increasing the steady-state level of GLK1. This
analysis is also subject to the caveat that in neither case
was the amount of GLK1 protein measured.
In all of our analyses, differences in transcript abun-
dance between the leaves of C. gynandra and C. spinosa
may reflect the operation of the C4and C3photosyn-
thetic pathways; alternatively, they may be due to
differences in metabolism and cell biology associated
with the phylogenetic distance between the two spe-
cies. However, in many cases, it is striking that our
analysis has identified differences in the abundance of
transcripts derived from genes that have been docu-
mented to be involved in processes known to alter in a
C4leaf. Taken together, the analysis allows us to sig-
nificantly extend the number of C4-related genes con-
trolled at the level of transcript abundance and to
extend the current model for C4-related processes in
NAD-ME C4plants (Fig.4).Analysisofadditionalpairs
of C3and C4species will likely facilitate the identifica-
tion of genes specifically involved in the C4pathway
and exclude genes that are modified for other reasons.
MATERIALS AND METHODS
Plant Material and 454 Sequencing
Cleome spinosa and Cleome gynandra plants for transcript profiling by RNA-
Seq were grown in standard potting mix in a glasshouse in August and
September 2007. To obtain sequence tags for DGE analysis from C. spinosa and
C. gynandra, total RNAs were isolated from fully expanded leaves sampled
from 56-d-old plants of each species. mRNAwas reverse transcribed to cDNA
after two consecutive rounds of oligo(dT) purification and prepared for 454
sequencing as described previously (Weber et al., 2007).
Mapping and Quantification of the Sequence Reads
Evolution did not stop in the lineage to the reference genome of Arabi-
dopsis (Arabidopsis thaliana) after the Cleomaceae branch diverged. Hence,
there may be genes that were tandem duplicated or retained after the whole
genome duplication event of the Brassicaceae that are absent in either of the
Cleomaceae species (Bra ¨utigam and Gowik, 2010). To avoid mapping prob-
lems such as splitting of reads or mapping errors due to differential retention
of genes in either Cleomaceae or Arabidopsis, we created a minimal genome
for mapping. The remnants of the last whole genome duplication in the
lineage of the Brassicaceae (Bowers et al., 2003; Thomas et al., 2006) and the
tandem duplicated genes (Haberer et al., 2004) were reduced to one repre-
sentative for each based on the TAIR 9 coding sequence set. In each case, the
gene with the lowest AGI code was retained for mapping. For each gene, the
Supplemental Data store whether there are duplicates and which duplicates
match the gene (Supplemental Tables S3 and S5). We recommend recovery of
the associated duplicated genes followed by a detailed analysis with phylo-
genetic trees to define the true ortholog when translating the results of
Cleomaceae analyses to Arabidopsis research.
The 454 sequence reads were mapped onto coding sequences of the
minimalized TAIR 9 genome by BLAT (Kent, 2002) and BLAST (Altschul et al.,
An mRNA Blueprint for C4Photosynthesis
Plant Physiol. Vol. 155, 2011 153

Page 13

1997) with varying parameters, and the output was parsed with in-house PERL
scripts toretainonly the best matchingAGI codes foreach sequence read and the
best BLAST hit, respectively. Differentially expressed transcripts were identified
usingthePoissonstatisticsdevelopedbyAudicandClaverie(1997)followedbya
Bonferroni correction to account for the accumulation of a-type errors when
conducting multiple pair-wise comparisons (Audic and Claverie, 1997).
Plant Material and qPCR Analysis
Both species were grown in a growth chamber in long-day conditions (16 h
of light/8 h of dark) under 350 mmol photons m22s21, at 22?C, and 65%
relative humidity prior to samples being taken for qPCR. qPCR was con-
ducted on the same samples used for RNA-Seq and also on mature leaves
collected at noon grown in the growth cabinet. For qPCR, RNA was isolated
using TriPure reagent (Roche Applied Science). RNAwas treated with DNase
I (Promega) and purified with the RNeasy Mini Kit (Qiagen). First-strand
cDNA was then synthesized with SuperScriptII reverse transcriptase (Invi-
trogen) using 4 mg of RNA and oligo(dT) primers (Roche Applied Science).
Quantitative reverse transcription-PCR was carried out with 96-well plates
using a DNA Engine thermal cycler, Chromo4 real-time detector (Bio-Rad),
SYBR Green JumpStart Taq Ready Mix (Sigma), and 15-fold dilution of the
cDNA as a template. Initial denaturation was carried out at 94?C for 2 min,
followed by 40 cycles of 94?C for 20 s, 60?C for 30 s, 72?C for 30 s, and 75?C for
5 s. Primers were designed to have melting temperatures of 60?C 6 0.5?C and
to produce amplicons of 91 to 189 bp. The specificity of the primers and lack of
primer dimers in the PCR were verified using agarose gel electrophoresis and
melting curve analysis. For each product, the threshold cycle CT, where the
amplification reaction enters the exponential phase, was determined for three
technical replicates and four independent biological replicates per species.
The comparative 22DDCTmethod was used to quantify relative abundance of
transcripts (Livak and Schmittgen, 2001). ACTIN7 was chosen as a reference
because the 454 sequencing data showed equal, intermediate levels of ACTIN7
transcripts in both species. For the qPCR, SE values were calculated from
22DDCTvalues of each combination of biological replicates.
Polar Metabolite, Chlorophyll, Protein, and Enzyme
Activity Analyses
For metabolite analysis, mature leaves from 56-d-old plants were collected
in the middle of the light period and immediately frozen in liquid nitrogen.
Three independent biological replicates were used. The tissues were ground
in a mortar, and a 50-mg fresh weight aliquot was extracted using the pro-
cedure described by Lee and Fiehn (2008). Ribitol was used as an internal stan-
dard for data normalization. For GC-EI-TOF analysis, samples were processed
and analyzed according to Lee and Fiehn (2008). Enzyme activities, chlorophyll,
and protein content were determined according to Hausler et al. (2001).
The Cleome read data have been submitted to the National Center for
Biotechnology Information Short Read Archive: C. spinosa = SRS002743.1 and
C. gynandra = SRS002744.2.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Quantitation of marker enzyme activities in leaf
extracts of C. spinosa and C. gynandra.
Supplemental Figure S2. Protein-to-fresh weight and protein-to-chloro-
phyll ratios in leaves of C. gynandra and C. spinosa.
Supplemental Figure S3. Changes in transcript abundance for ribosomal
proteins.
Supplemental Table S1. Relative abundance of predominant metabolites
detected by GC-EI-TOF in C. gynandra and in C. spinosa.
Supplemental Table S2. Number of gene loci and number of differentially
expressed genes detected with BLAT and BLAST.
Supplemental Table S3. Quantitative information for all reads mapped
onto the reference genome from Arabidopsis.
Supplemental Table S4. Comparison of mapping parameters.
Supplemental Table S5. Segmental and tandem duplicates in the Arabi-
dopsis genome.
ACKNOWLEDGMENTS
We thank Tom Hardcastle for help with R.
Received May 18, 2010; accepted June 9, 2010; published June 11, 2010.
LITERATURE CITED
Altschul SF, Madden TL, Schaffer AA, Zhang JH, Zhang Z, Miller W,
Lipman DJ (1997) Gapped BLAST and PSI-BLAST: a new generation of
protein database search programs. Nucleic Acids Res 25: 3389–3402
Aoki N, Kanai R (1997) Reappraisal of the role of sodium in the light-
dependent active transport of pyruvate into mesophyll chloroplasts of
C4plants. Plant Cell Physiol 38: 1217–1225
Aoki N, Ohnishi JI, Kanai R (1994) Proton/pyruvate cotransport into
mesophyll chloroplasts of C-4 plants. Plant Cell Physiol 35: 801–806
Audic S, Claverie JM (1997) The significance of digital gene expression
profiles. Genome Res 7: 986–995
Black CC (1973) Photosynthetic carbon fixation in relation to net CO2
uptake. Annu Rev Plant Physiol Plant Mol Biol 24: 253–286
Botha CEJ (1992) Plasmodesmatal distribution, structure and frequency in
relation to assimilation in C3and C4grasses in southern Africa. Planta
187: 348–358
Bowers JE, Chapman BA, Rong JK, Paterson AH (2003) Unravelling
angiosperm genome evolution by phylogenetic analysis of chromo-
somal duplication events. Nature 422: 433–438
Bra ¨utigam A, Gowik U (2010) What can next generation sequencing do for
you? Next generation sequencing as a valuable tool in plant research.
Plant Biol 12: 831–841
Bra ¨utigam A, Hofmann-Benning S, Weber APM (2008a) Comparative
proteomics of chloroplast envelopes from C3and C4plants reveals
specific adaptations of the plastid envelope to C4photosynthesis and
candidate proteins required for maintaining C4metabolite fluxes. Plant
Physiol 148: 568–579
Bra ¨utigam A, Shrestha RP, Whitten D, Wilkerson CG, Carr KM, Froehlich
JE, Weber APM (2008b) Comparison of the use of a species-specific
database generated by pyrosequencing with databases from related
species for proteome analysis of pea chloroplast envelopes. J Biotechnol
136: 44–53
Bra ¨utigam A, Weber APM (2011) Transport processes: connecting the
reactions of C4photosynthesis. In AS Raghavendra, RF Sage, eds, C4
Photosynthesis and Related CO2Concentrating Mechanisms. Advances
in Photosynthesis and Respiration, Vol 32. Springer, Berlin, pp 199–219
Brown NJ, Parsley K, Hibberd JM (2005) The future of C4research: maize,
Flaveria or Cleome? Trends Plant Sci 10: 215–221
Brown RH (1999) Agronomic implications of C4photosynthesis. In RF
Sage, RK Monson, eds, C4Plant Biology. Academic Press, San Diego, pp
473–507
Brownell PF, Crossland CJ (1972) The requirement for sodium as a
micronutrient by species having the C4dicarboxylic photosynthetic
pathway. Plant Physiol 49: 794–797
Chollet R, Ogren WL (1975) Regulation of photorespiration in C3and C4
species. Bot Rev 41: 137–179
DeBono A, Yeats TH, Rose JKC, Bird D, Jetter R, Kunst L, Samuelsa L
(2009) Arabidopsis LTPG is a glycosylphosphatidylinositol-anchored
lipid transfer protein required for export of lipids to the plant surface.
Plant Cell 21: 1230–1238
De Vos M, Van Oosten VR, Van Poecke RM, Van Pelt JA, Pozo MJ, Mueller
MJ, Buchala AJ, Metraux JP, Van Loon LC, Dicke M, et al (2005) Sig-
nal signature and transcriptome changes of Arabidopsis during pathogen
and insect attack. Mol Plant Microbe Interact 18: 923–937
Dodd AN, Gardner MJ, Hotta CT, Hubbard KE, Dalchau N, Love J, Assie
JM, Robertson FC, Jakobsen MK, Goncalves J, et al (2007) The Arab-
idopsis circadian clock incorporates a cADPR-based feedback loop.
Science 318: 1789–1792
Evert RF, Eschrich W, Heyser W (1977) Distribution and structure of
plasmodesmata in mesophyll and bundle-sheath cells of Zea mays L.
Planta 136: 77–89
Fischer K, Kammerer B, Gutensohn M, Arbinger B, Weber A, Hausler RE,
Flugge UI (1997) A new class of plastidic phosphate translocators: a
putative link between primary and secondary metabolism by the phos-
phoenolpyruvate/phosphate antiporter. Plant Cell 9: 453–462
Bra ¨utigam et al.
154Plant Physiol. Vol. 155, 2011

Page 14

Flicek P, Birney E (2009) Sense from sequence reads: methods for align-
ment and assembly. Nat Methods 6: S6–S12
Haberer G, Hindemitt T, Meyers BC, Mayer KFX (2004) Transcriptional
similarities, dissimilarities, and conservation of cis-elements in dupli-
cated genes of Arabidopsis. Plant Physiol 136: 3009–3022
Hatch MD (1987) C4photosynthesis: a unique blend of modified biochem-
istry, anatomy and ultrastructure. Biochim Biophys Acta 895: 81–106
Hatch MD, Slack CR (1968) A new enzyme for interconversion of pyruvate
and phosphopyruvate and its role in C4dicarboxylic acid pathway of
photosynthesis. Biochem J 106: 141–147
Hausler RE, Rademacher T, Li J, Lipka V, Fischer KL, Schubert
S, Kreuzaler F, Hirsch HJ (2001) Single and double overexpression of
C4-cycle genes had differential effects on the pattern of endogenous
enzymes, attenuation of photorespiration and on contents of UV pro-
tectants in transgenic potato and tobacco plants. J Exp Bot 52: 1785–1803
Hibberd JM, Sheehy JE, Langdale JA (2008) Using C4photosynthesis to
increase the yield of rice: rationale and feasibility. Curr Opin Plant Biol
11: 228–231
Imaizumi T, Kay SA, Schroeder JI (2007) Circadian rhythms: daily watch
on metabolism. Science 318: 1730–1731
Jordan DB, Ogren WL (1984) The CO2/O2specificity of ribulose 1,5-
bisphosphate carboxylase oxygenase: dependence on ribulosebisphos-
phate concentration, pH and temperature. Planta 161: 308–313
Keeley JE (1998) C4photosynthetic modifications in the evolutionary
transition from land to water in aquatic grasses. Oecologia 116: 85–97
Kent WJ (2002) BLAT: the BLAST-Like Alignment Tool. Genome Res 12:
656–664
Kilian J, Whitehead D, Horak J, Wanke D, Weinl S, Batistic O, D’Angelo
C, Bornberg-Bauer E, Kudla J, Harter K (2007) The AtGenExpress
global stress expression data set: protocols, evaluation and model data
analysis of UV-B light, drought and cold stress responses. Plant J 50:
347–363
Kobayashi H, Yamada M, Taniguchi M, Kawasaki M, Sugiyama T,
Miyake H (2009) Differential positioning of C4mesophyll and bundle
sheath chloroplasts: recovery of chloroplast positioning requires the
actomyosin system. Plant Cell Physiol 50: 129–140
Ku MSB, Schmitt MR, Edwards GE (1979) Quantitative determination of
ribulose bisphosphate carboxylase oxygenase protein in leaves of sev-
eral C3and C4plants. J Exp Bot 114: 89–98
Langdale JA, Kidner CA (1994) Bundle-sheath defective, a mutation
that disrupts cellular differentiation in maize leaves. Development
120: 673–681
Lee DY, Fiehn O (2008) High quality metabolomic data for Chlamydomonas
reinhardtii. Plant Methods 4: 7
Leegood RC (2002) C4photosynthesis: principles of CO2concentration and
prospects for its introduction into C3plants. J Exp Bot 53: 581–590
Levy A, Erlanger M, Rosenthal M, Epel BL (2007) A plasmodesmata-
associated beta-1,3-glucanase in Arabidopsis. Plant J 49: 669–682
Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data
using real-time quantitative PCR and the 2(T)(-Delta Delta C) method.
Methods 25: 402–408
Majeran W, van Wijk KJ (2009) Cell-type-specific differentiation of chlo-
roplasts in C4plants. Trends Plant Sci 14: 100–109
Maple J, Fujiwara MT, Kitahata N, Lawson T, Baker NR, Yoshida S,
Moller SG (2004) GIANT CHLOROPLAST 1 is essential for correct
plastid division in Arabidopsis. Curr Biol 14: 776–781
Marshall DM, Muhaidat R, Brown NJ, Liu Z, Stanley S, Griffiths H, Sage
RF, Hibberd JM (2007) Cleome, a genus closely related to Arabidopsis,
contains species spanning a developmental progression from C3to C4
photosynthesis. Plant J 51: 886–896
Matsuoka M, Furbank RT, Fukuyama H, Miyao M (2001) Molecular
engineering of C4photosynthesis. Annu Rev Plant Physiol Plant Mol
Biol 52: 297–314
Metzker ML (2010) Applications of next-generation sequencing technolo-
gies: the next generation. Nat Rev Genet 11: 31–46
Mitchell PL, Sheehy JE (2006) Supercharging rice photosynthesis to
increase yield. New Phytol 171: 688–693
Oaks A (1994) Efficiency of nitrogen utilization in C3and C4cereals. Plant
Physiol 106: 407–414
Oikawa K, Kasahara M, Kiyosue T, Kagawa T, Suetsugu N, Takahashi F,
Kanegae T, Niwa Y, Kadota A, Wada M (2003) CHLOROPLAST UN-
USUAL POSITIONING1 is essential for proper chloroplast positioning.
Plant Cell 15: 2805–2815
Osborne CP, Freckleton RP (2009) Ecological selection pressures for
C4photosynthesis in the grasses. Proc R Soc Lond B Biol Sci 276:
1753–1760
Palmieri N, Schlotterer C (2009) Mapping accuracy of short reads from
massively parallel sequencing and the implications for quantitative
expression profiling. PLoS ONE 4: 10
Price J, Laxmi A, St Martin SK, Jang JC (2004) Global transcription
profiling reveals multiple sugar signal transduction mechanisms in
Arabidopsis. Plant Cell 16: 2128–2150
Reiskind JB, Berg RH, Salvucci ME, Bowes G (1989) Immunogold local-
ization of primary carboxylases in leaves of aquatic and a C3-C4
intermediate species. Plant Sci 61: 43–52
Renne P, Dressen U, Hebbeker U, Hille D, Flugge UI, Westhoff P, Weber
APM (2003) The Arabidopsis mutant dct is deficient in the plastidic
glutamate/malate translocator DiT2. Plant J 35: 316–331
Roberts AG, Oparka KJ (2003) Plasmodesmata and the control of sym-
plastic transport. Plant Cell Environ 26: 103–124
Rolland F, Baena-Gonzalez E, Sheen J (2006) Sugar sensing and signaling
in plants: conserved and novel mechanisms. Annu Rev Plant Biol 57:
675–709
Rossini L, Cribb L, Martin DJ, Langdale JA (2001) The maize Golden2 gene
defines a novel class of transcriptional regulators in plants. Plant Cell 13:
1231–1244
Sage RF (2004) The evolution of C4photosynthesis. New Phytol 161:
341–370
Sawers RJH, Liu P, Anufrikova K, Hwang JTG, Brutnell TP (2007) A
multi-treatment experimental system to examine photosynthetic dif-
ferentiation in the maize leaf. BMC Genomics 8: 12
Schaffer R, Landgraf J, Accerbi M, Simon V, Larson M, Wisman E (2001)
Microarray analysis of diurnal and circadian-regulated genes in Arabi-
dopsis. Plant Cell 13: 113–123
Schmid M, Davison TS, Henz SR, Pape UJ, Demar M, Vingron M,
Scholkopf B, Weigel D, Lohmann JU (2005) A gene expression map of
Arabidopsis thaliana development. Nat Genet 37: 501–506
Sheen J (1999) C4gene expression. Annu Rev Plant Physiol Plant Mol Biol
50: 187–217
Swarbreck D, Wilks C, Lamesch P, Berardini TZ, Garcia-Hernandez M,
Foerster H, Li D, Meyer T, Muller R, Ploetz L, et al (2008) The
Arabidopsis Information Resource (TAIR): gene structure and function
annotation. Nucleic Acids Res 36: D1009–D1014
Taniguchi M, Taniguchi Y, Kawasaki M, Takeda S, Kato T, Sato S, Tahata
S, Miyake H, Sugiyama T (2002) Identifying and characterizing
plastidic 2-oxoglutarate/malate and dicarboxylate transporters in Arab-
idopsis thaliana. Plant Cell Physiol 43: 706–717
Taniguchi Y, Nagasaki J, Kawasaki M, Miyake H, Sugiyama T,
Taniguchi M (2004) Differentiation of dicarboxylate transporters
in mesophyll and bundle sheath chloroplasts of maize. Plant Cell
Physiol 45: 187–200
Thomas BC, Pedersen B, Freeling M (2006) Following tetraploidy in an
Arabidopsis ancestor, genes were removed preferentially from one
homeolog leaving clusters enriched in dose-sensitive genes. Genome
Res 16: 934–946
Vogel JT, Zarka DG, Van Buskirk HA, Fowler SG, Thomashow MF
(2005) Roles of the CBF2 and ZAT12 transcription factors in config-
uring the low temperature transcriptome of Arabidopsis. Plant J 41:
195–211
Voznesenskaya EV, Edwards GE, Kiirats O, Artyusheva EG, Franceschi
VR (2003) Development of biochemical specialization and organelle
partitioning in the single-cell C4system in leaves of Borszczowia
aralocaspica (Chenopodiaceae). Am J Bot 90: 1669–1680
Voznesenskaya EV, Franceschi VR, Kiirats O, Artyusheva EG, Freitag H,
Edwards GE (2002) Proof of C4photosynthesis without Kranz anatomy
in Bienertia cycloptera (Chenopodiaceae). Plant J 31: 649–662
Voznesenskaya EV, Franceschi VR, Kiirats O, Freitag H, Edwards GE
(2001) Kranz anatomy is not essential for terrestrial C4plant photosyn-
thesis. Nature 414: 543–546
Voznesenskaya EV, Koteyeva NK, Chuong SDX, Ivanova AN, Barroca J,
Craven LA, Edwards GE (2007) Physiological, anatomical and bio-
chemical characterisation of photosynthetic types in genus Cleome
(Cleomaceae). Funct Plant Biol 34: 247–267
Wall PK, Leebens-Mack J, Chanderbali AS, Barakat A, Wolcott E, Liang
HY, Landherr L, Tomsho LP, Hu Y, Carlson JE, et al (2009) Comparison
An mRNA Blueprint for C4Photosynthesis
Plant Physiol. Vol. 155, 2011155

Page 15

of next generation sequencing technologies for transcriptome charac-
terization. BMC Genomics 10: 347
Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: a revolutionary tool for
transcriptomics. Nat Rev Genet 10: 57–63
Waters MT, Wang P, Korkaric M, Capper RG, Saunders NJ, Langdale JA
(2009) GLK transcription factors coordinate expression of the photo-
synthetic apparatus in Arabidopsis. Plant Cell 21: 1109–1128
Weber APM, von Caemmerer S (2010) Plastid transport and metabolism of
C3and C4plants: comparative analysis and possible biotechnological
exploitation. Curr Opin Plant Biol 13: 256–264
Weber APM, Weber KL, Carr K, Wilkerson C, Ohlrogge JB (2007) Sam-
pling the Arabidopsis transcriptome with massively parallel pyrose-
quencing. Plant Physiol 144: 32–42
Wise RP, Moscou MJ, Bogdanove AJ, Whitham SA (2007) Transcript
profiling in host-pathogen interactions. Annu Rev Phytopathol 45:
329–369
Bra ¨utigam et al.
156Plant Physiol. Vol. 155, 2011