C 3 –C 4 intermediacy in grasses: organelle enrichment and distribution, glycine decarboxylase expression, and the rise of C 2 photosynthesis

Abstract

Photorespiratory glycine shuttling and decarboxylation in bundle sheath (BS) cells exhibited by C2 species is proposed to be the evolutionary bridge to C4 photosynthesis in eudicots. To evaluate this in grasses, we compare anatomy, cellular localization of glycine decarboxylase
(GDC), and photosynthetic physiology of a suspected C2 grass, Homolepis aturensis, with these traits in known C2 grasses, Neurachne minor and Steinchisma hians, and C3 S. laxum that is sister to S. hians. We also use publicly available genome and RNA-sequencing data to examine the evolution of GDC subunits and enhance our understanding
of the evolution of BS-specific GDC expression in C2 and C4 grasses. Our results confirm the identity of H. aturensis as a C2 species; GDC is confined predominantly to the organelle-enriched BS cells in H. aturensis and S. hians and to mestome sheath cells of N. minor. Phylogenetic analyses and data obtained from immunodetection of the P-subunit of GDC are consistent with the hypothesis
that the BS dominant levels of GDC in C2 and C4 species are due to changes in expression of a single GLDP gene in M and BS cells. All BS mitochondria and peroxisomes and
most chloroplasts in H. aturensis and S. hians are situated centripetally in a pattern identical to C2 eudicots. In S. laxum, which has C3-like gas exchange patterns, mitochondria and peroxisomes are positioned centripetally as they are in S. hians. This subcellular phenotype, also present in eudicots, is posited to initiate a facilitation cascade leading to C2 and C4 photosynthesis.

Full text

Journal of Experimental Botany, Vol. 67, No. 10 pp. 3065–3078, 2016
doi:10.1093/jxb/erw150 Advance Access publication 12 April 2016
RESEARCH PAPER
C
3
–C
4
intermediacy in grasses: organelle enrichment and
distribution, glycine decarboxylase expression, and the rise
of C
2
photosynthesis
RoxanaKhoshravesh
1
, Corey R.Stinson
1
, MattStata
1
, Florian A.Busch
2
, Rowan F.Sage
1
,
MarthaLudwig
3
and Tammy L.Sage
1,
*
1
Department of Ecology and Evolutionary Biology, University of Toronto, 25 Willcocks St., Ontario, ON M5S 3B2, Canada
2
Research School of Biology, Australian National University, Canberra, ACT 2601, Australia
3
School of Chemistry and Biochemistry, University of Western Australia, Crawley, WA 6009, Australia
* Correspondence: tammy.sage@utoronto.ca
Received 21 February 2016; Accepted 21 March 2016
Editor: Christine Raines, University of Essex
Abstract
Photorespiratory glycine shuttling and decarboxylation in bundle sheath (BS) cells exhibited by C
2
species is proposed
to be the evolutionary bridge to C
4
photosynthesis in eudicots. To evaluate this in grasses, we compare anatomy, cellular
localization of glycine decarboxylase (GDC), and photosynthetic physiology of a suspected C
2
grass, Homolepis aturensis,
with these traits in known C
2
grasses, Neurachne minor and Steinchisma hians, and C
3
S. laxum that is sister to S. hians.
We also use publicly available genome and RNA-sequencing data to examine the evolution of GDC subunits and enhance
our understanding of the evolution of BS-specific GDC expression in C
2
and C
4
grasses. Our results confirm the identity of
H. aturensis as a C
2
species; GDC is confined predominantly to the organelle-enriched BS cells in H. aturensis and S. hians
and to mestome sheath cells of N. minor. Phylogenetic analyses and data obtained from immunodetection of the P-subunit
of GDC are consistent with the hypothesis that the BS dominant levels of GDC in C
2
and C
4
species are due to changes
in expression of a single GLDP gene in M and BS cells. All BS mitochondria and peroxisomes and most chloroplasts in H.
aturensis and S. hians are situated centripetally in a pattern identical to C
2
eudicots. In S. laxum, which has C
3
-like gas
exchange patterns, mitochondria and peroxisomes are positioned centripetally as they are in S. hians. This subcellular
phenotype, also present in eudicots, is posited to initiate a facilitation cascade leading to C
2
and C
4
photosynthesis.
Key words: Arthropogoninae, bundle sheath, C
2
Kranz anatomy, C
2
photosynthesis, glycine decarboxylase, grasses,
mitochondria, Homolepis.
Introduction
C
4
photosynthesis has independently evolved >60 times in
angiosperms (R.F. Sage etal., 2011; Grass Phylogeny Working
Group II, 2012). Within the angiosperms, the Poaceae rep-
resents the most prolic family of C
4
origins, with approxi-
mately twice as many origins as any other family (Sage, 2016).
C
4
grasses also make up the greatest number of C
4
species,
comprising ~60% of the 8000 estimated number of C
4
species
(Still etal., 2003; Sage, 2016). Roughly a quarter of global
net primary productivity on land is due to C
4
photosynthesis
(Still etal., 2003), of which the vast majority is contributed by
grasses (Sage etal., 1999). C
4
grasses also have great signi-
cance for humanity as they dominate the fraction of biomass
entering the human food chain as grain (maize, sorghum,
and millets), sugar (sugarcane), and fodder for animals, and
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
© The Author 2016. Published by Oxford University Press on behalf of the Society for Experimental Biology.
This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

3066
|
Khoshravesh etal.
efforts are underway to engineer the C
4
pathway into C
3
grass
crops such as rice and wheat to exploit the superior productiv-
ity of C
4
photosynthesis (Peterhansel, 2011; von Caemmerer
etal., 2012). Given these considerations, there is now a great
interest in understanding how C
4
photosynthesis evolved in
grasses, to understand both how this complex trait repeat-
edly arose, and how we might learn from the evolutionary
examples to direct C
4
engineering in major crops (Hibberd
etal., 2008).
Studies of C
4
evolution are informed by the presence of
species exhibiting intermediate stages between fully expressed
C
3
and C
4
life forms within a single evolutionary clade (Sage
etal., 2014). Ideally, there should be C
3
–C
4
relatives from mul-
tiple independent lineages of C
4
photosynthesis to facilitate
evaluation of evolutionary hypotheses using comparative
approaches. Multiple independent clades provide the pos-
sibility to assess whether evolutionary trends are replicated,
as they should be if C
4
photosynthesis evolved along com-
mon trajectories (Heckmann etal., 2013). To date, the major-
ity (85%) of C
3
–C
4
species occur in eudicots, with the genus
Flaveria standing as the major group used in studies of C
4
evo-
lution (R.F. Sage etal., 2011, 2014). Flaveria has over twice
the number of intermediates (10) as any other evolutionary
clade, and these form two distinct clades that each evolved C
3
-
and C
4
-like species (McKown etal., 2005; Lyu et al., 2015).
In addition to Flaveria, at least 11 other C
4
evolutionary line-
ages have been identied with C
3
–C
4
intermediates branching
in sister positions to the C
4
line (Sage etal., 2014). Most of
these have only one or two species, although in recent years
there is evidence of three clades (Heliotropium, Anticharis,
and Blepharis) potentially having more than ve intermediates
(Muhaidat etal., 2011; Khoshravesh etal., 2012; Fisher etal.,
2015). All but three of the C
4
lineages with C
3
–C
4
intermedi-
ates are eudicots. Among monocots, the genus Neurachne con-
tains a C
3
–C
4
intermediate that branches in a sister position to
a C
4
(Christin etal., 2012). Arecent study has reported C
3
, C
3
–
C
4
, and C
4
photosynthetic genotypes in Alloteropsis semialata
(Lundgren etal., 2015). The genus Steinchisma also contains
C
3
–C
4
intermediates (Brown etal., 1983; Hylton etal., 1988),
but they lack close C
4
relatives in their subtribe, Otachyrinae
(Grass Phylogeny Working Group II, 2012). As a consequence
of the discrepancy between C
3
–C
4
numbers in eudicots and
monocots, our understanding of C
4
evolution is dominated
by information from eudicot clades. If there is important vari-
ation in eudicot versus monocot patterns of C
4
evolution, as
suggested by recent theoretical treatments (Williams et al.,
2013), it could be missed because of low monocot representa-
tion in the C
3
–C
4
intermediate population.
The South American subtribe Arthropogoninae is a hot-
spot for C
4
evolution within the grasses, with four putative
distinct C
4
origins, once in Mesosetum/Arthropogon, a second
time in Oncorachis, and twice in Coleataenia (Grass Phylogeny
Working Group II, 2012). As such, the Arthropogoninae is
a strong candidate to contain numerous C
3
–C
4
species. This
possibility is bolstered by an image from Homolepis aturensis
in Supplementary g. S1 of Christin etal. (2013) that illus-
trates enlarged bundle sheath (BS) cells with chloroplasts
arranged around the periphery in addition to chloroplast
clusters adjacent to the vascular tissue. Although this centrip-
etal chloroplast arrangement led to tentative identication
of H.aturensis as C
4
(Christin etal., 2013), the presence of
centrifugal chloroplasts in the BS cells is a common feature in
C
3
–C
4
intermediate species (Monson etal., 1984; Sage etal.,
2014). Signicantly, because the genus Homolepis is sister to
a clade that contains only C
4
species, it has been identied as
a genus that might exhibit precursor traits that could have
enabled the evolution of the C
4
phenotype (Grass Phylogeny
Working Group II, 2012). To evaluate this possibility, we have
collected H. aturensis in Costa Rica forstudy.
A prominent physiological feature of C
3
–C
4
intermediates
is the transport of photorespiratory glycine from mesophyll
(M) to BS cells for decarboxylation by glycine decarboxy-
lase (GDC), with the released CO
2
then being rexed by BS
Rubisco (Monson and Rawsthorne, 2000). Photorespiratory
glycine shuttling exhibited by C
3
–C
4
intermediates has also
been termed C
2
photosynthesis in reference to the number
of carbons shuttled from M to BS cells (Vogan etal., 2007;
Bauwe, 2011). The BS mitochondria of most C
2
species exam-
ined to date contain the majority of the GDC within the leaf,
with small amounts in M cells potentially for C1 metabolism
(Hylton etal., 1988; Morgan etal., 1993; Rawsthorne etal.,
1998; Voznesenskaya etal., 2001; Ueno etal., 2003; Marshall
etal., 2007; Voznesenskaya etal., 2010; Muhaidat etal., 2011;
T.L. Sage etal., 2011). Decarboxylation of glycine in the BS
cells establishes a glycine gradient between M and BS cells,
and rapid movement to BS cells is facilitated by enhanced vein
density in C
2
relative to C
3
species (Monson and Rawsthorne,
2000). Subsequent glycine decarboxylation within BS cells
increases CO
2
around BS Rubisco ~3-fold (Keerberg et al.,
2014), and the resulting increase in Rubisco efciency reduces
the CO
2
compensation point in C
2
species relative to C
3
by
10–40μmol mol
−1
(Holaday etal., 1984; Monson etal., 1984;
Vogan etal., 2007; T.L. Sage etal., 2011, 2013).
The functional GDC holoenzyme consists of four subunits
encoded by individual genes, GLDH, GLDL, GLDP, and
GLDT (Bauwe 2011). Decarboxylase activity of the com-
plex is located in the P-subunit encoded by the GDLP gene.
In Flaveria, the C
4
photosynthetic mechanism was estab-
lished through gradual pseudogenization of a ubiquitously
expressed GLDP gene, and full activation of a second GLDP
gene that shows BS-specic expression in C
3
Flaveria species
(Schulze etal., 2013). The ancestral duplication of the GLDP
gene in Flaveria is considered a genetic enabler of C
4
evolution
in the genus (Schulze etal., 2013). BS cell-dominant expres-
sion of GDC has been reported in the C
2
grass Steinchisma
hians (=Panicum milioides; Hylton etal., 1988). To date, the
molecular evolution of this trait in S. hians or other C
2
and C
4
grasses has not been assessed. Schulze etal. (2013), highlight-
ing the presence of two GLDP genes in rice [a member of the
C
3
BEP (Bambusoideae, Ehrhartoideae, Pooideae) clade] and
a single copy in maize, sorghum, and Setaria italica [of the
C
4
PACMAD (Panicoideae, Arundinoideae, Chloridoideae,
Micrairoideae, Aristoideae, Danthonioideae) clade], posited
that the ubiquitously expressed GLDP gene(s) in C
4
grasses
were pseudogenized as in Flaveria and subsequently lost from
the genomes.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

C
2
photosynthesis in grasses
|
3067
The purpose of this study was to determine whether H.
aturensis exhibits C
2
photosynthesis or is a C
4
species as pre-
viously suggested (Christin etal., 2013). We compared anat-
omy, localization of GLDP, and photosynthetic physiology
of H.aturensis with patterns previously identied in the C
2
grasses Steinchisma hians and Neurachne minor (Morgan and
Brown, 1979; Hylton etal., 1988). Current phylogenies place
the subtribe Otachyrinae, containing Steinchisma, as sister
to the Arthropogoninae (Grass Phylogeny Working Group
II, 2012). In addition, we examined S.laxum which has also
been identied as a species that might provide information
on the early stages of C
2
and C
4
evolution (Grass Phylogeny
Working Group II, 2012; Sage etal., 2013). Finally, we use
genome sequence data from publicly available databases
(the Phytozome and NCBI), as well as assembled RNA-
sequencing (RNA-seq) from 16 additional grass species, to
examine evolution of GLDP and genes encoding the other
GDC subunits and provide a broader understanding of the
evolution of BS-specic GDC expression in C
2
and C
4
grasses.
Materials and methods
Plant material
Plants of Homolepis aturensis Chase., Steinchisma hians Raf.,
and S. laxum (Sw.) Zuloaga obtained from sources described in
Supplementary Table S1 at JXB online were grown at the University
of Toronto in a greenhouse in 10–20 liter pots of a sandy-loam soil
and were watered daily to avoid water stress. Fertilizer was supplied
weekly as a 50:50 mixture of Miracle-Grow 24-10-10 All Purpose
Plant Food and Miracle Grow Evergreen Food (30-10-20) at the
recommended dosage (22 ml of fertilizer salt per 6 liters; Scotts
Miracle-Gro; www.scotts.ca). Plants of Neurachne minor from
localities previously described (Christin et al., 2012) were grown
in a naturally illuminated glasshouse with mean temperatures of
25°C/13°C (day/night) at the Plant Growth Facility (PGF) of the
University of Western Australia, Perth, Western Australia (latitude
33°89'S). To provide C
3
and C
4
grass species for comparison, we also
examined leaves of PACMAD species Dichanthelium oligosanthes
(Schult.) Gould (C
3
), Panicum bisulcatum Thunb. (C
3
), and two C
4
species (Panicum virgatum L., NAD-ME subtype and Setaria viridis
P.Beauv., NADP-ME subtype). Seed of these plants, obtained from
sources described in Supplementary Table S1, were also grown at the
University of Toronto.
Leaf anatomy, ultrastructure, and immunolocalizations
The internal anatomy of leaves was assessed on sections sampled
from the middle of the most recent, fully expanded leaves (one leaf
per plant; three plants per species). Plants were sampled from 09:00 h
to 11:00 h between April and August when day length was >11.5 h
and light intensity in the greenhouse regularly exceeded 1400µmol
photons m
−2
s
−1
. The youngest cohort of fully expanded leaves was
sampled in full sun for all procedures. Samples were prepared for
light and transmission electron microscopy (TEM) to assess anat-
omy as previously described (T.L. Sage et al., 2011, 2013; Stata
etal., 2014). For immunolocalization, tissue from the same region
of the leaf was xed overnight in 1% (v/v) paraformaldehyde and 1%
(v/v) glutaraldehyde in 0.05 M sodium cacodylate buffer. Tissue was
then dehydrated and embedded in LR White (Voznesenskaya etal.,
2013). Immunolocalization of GLDP was conducted as outlined by
Khosravesh et al. (2012). Primary and secondary antibody (18 nm
anti-rabbit IgG gold conjugate; Jackson Immunoresearch) dilutions
were 1:50 and 1:20, respectively. Immunodetection of the Rubisco
large subunit was modied from Ueno (1992). Sections were blocked
in 0.5% BSA prior to incubation in primary antibody (1:100) for 3 h.
Incubation in secondary antibody (1:40; 18 nm anti-rabbit IgG gold
conjugate; Jackson Immunoresearch) was for 1 h. To quantify all BS
and M cellular features, TEM images from BS and M cells of the
same grids used for immunogold labeling were analyzed using Image
J software (Schneider et al., 2012) as previously described (Sage
etal., 2013; Stata etal., 2014). The anti-GLDP antiserum was com-
mercially produced (GL Biochem) against a 17 amino acid peptide
showing high conservation in both monocots and dicots. Antisera
recognizing the Rubisco large subunit (RBCL) were obtained from
AgriSera. Three replicate immunolocalizations were conducted on
different days with each replicate including sectioned tissue from all
species.
Leaf gas exchange analysis
Gas exchange of intact, attached leaves was determined using a
LiCor 6400 gas exchange system as previously described (Sage etal.,
2013). Measurement conditions were 31 ± 1°C and a vapor pressure
difference between leaf and air of 2 ± 0.2 kPa. For measurement of
the response of net CO
2
assimilation rate (A) to intercellular CO
2
concentration (C
i
) at light saturation, leaves were rst equilibrated to
1200–1500µmol photons m
−1
s
−1
at an ambient CO
2
concentration
of 400µmol m
−2
s
−1
. Ambient CO
2
levels were then reduced in steps
to 30–50µmol mol
−1
(lower end of this range for C
2
and C
4
species,
upper end of this range for C
3
species), with measurements at each
step after rate equilibration. The ambient CO
2
was then returned
to 400µmol photons m
−1
s
−1
, and A re-measured. If A was within
10% of the original rate, the CO
2
concentration around the leaf
was increased in steps to near 1600µmol mol
−1
, with measurements
made at each step. The linear initial slope of the A/C
i
response was
used as an estimate of carboxylation efciency(CE).
For estimation of the apparent CO
2
compensation point in the
absence of day respiration (C
*
), the Laisk method was used as modi-
ed by Sage et al. (2013). Values of C
*
were not estimated in C
4
plants. Leaves were rst equilibrated at 400µmol mol
−1
and a light
intensity near saturation (900–1500µmol photons m
−2
s
−1
). The CO
2
was then reduced to provide a C
i
near 100µmol mol
−1
, and, after
stabilization, the rate was reduced in a series of steps to near the CO
2
compensation point (Γ), with measurements at each step after signal
stabilization. The C
i
was then returned to near 100µmol mol
−1
, and
the procedure was repeated at a lower light intensity. This cycle was
repeated such that when measurements were completed, there were
4–5 A versus C
i
response curves with over ve measurements at C
i
values <100µmol mol
−1
. Each A versus C
i
response at a given light
intensity was then tted with a linear regression using the lowest 4–6
measurement points that fell on the regression. Points that fell below
the regression line above ~80µmol mol
−1
were not included in the
regression, as A/C
i
responses below light saturation may be prone
to non-linearity above 60–100 µmol CO
2
mol
−1
air. The estimate
of C
*
was taken as the C
i
value where the 4–5 curves at different
light intensities converged. Rarely, however, did all curves converge
at the exact same C
i
; instead, they intersected with each other over
a 5–10 ppm range. In such cases, the middle of the intersects was
taken as the C
*
estimate. However, if there was evidence of a shift to
lower photosynthetic photon ux density (PPFD) of the high light
response, which often occurs in C
2
species, the high light response
was not included in the analysis (Sage etal., 2013).
Phylogenetic analysis of GDC subunitgenes
Genome sequences for Zea mays, Sorghum bicolor, Panicum hallii,
Se. viridis, S. italica, Oryza sativa, Brachypodium distachyon, and
B.stacei were downloaded from the Phytozome v10.3 (https://phyto-
zome.jgi.doe.gov/pz/portal.html). Polyploid species Triticum aesti-
vum and P.virgatum
were omitted. Amborella trichopoda was used as
the outgroup, and we included the Brassicaceae species Arabidopsis
thaliana, Capsella grandiora, and Boechera stricta to show which
duplications in the model plant A. thaliana are conserved across
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

3068
|
Khoshravesh etal.
the land plants and which are lineage specic. RNA-seq data for 14
BEP clade species as well as the C
3
PACMAD Dicanthelium clan-
destinum and its close C
4
relative Megathyrsus maximus were down-
loaded from the NCBI short read archive (Supplementary Table
S2). Reads mapping to each GDC subunit gene were identied
using BLASTN with orthologs in Z.mays and O.sativa as queries
(S.bicolor was used in lieu of Z.mays for one of the two GLDL
paralogs as Z.mays appears to have lost this copy). For each gene,
alignments were generated based on the Phytozome genomes using
Muscle (Edgar, 2004), a highly conserved region was selected, and a
consensus sequence was generated. These consensus sequences were
used as reference for assembling sequences based on the retrieved
reads using Geneious 8 (http://www.geneious.com).
The assemblies and sequences from Phytozome genomes were
aligned using Muscle, trimmed using Trimal (Capella-Gutierrez
etal., 2009) with the ‘strict’ heuristic option, and used to generate
Bayesian phylogenies using MrBayes (Ronquist and Huelsenbeck,
2003) as follows: four runs, four chains, GTR substitution model, 2
million generations for all trees except GLDT, which was run for 10
million generations to allow better convergence; ≥10 000 trees were
sampled from portions of the end of each run where the average SD
of split frequencies remained below 1%. Tree gures were generated
using Fig Tree (http://tree.bio.ed.ac.uk/software/gtree).
Data analysis
Results were analyzed with Sigmaplot version 12.5 (Systat
Software., San Jose, CA, USA) using one-way ANOVA followed
by a Tukey’s means comparison test. For characterization of leaf
anatomy and ultrastructure, leaf samples were collected from three
plants. For all traits measured, the values per plant were averaged
to give one value for a plant. These individual plant values were
the unit of replication for statistical analysis. For characteriza-
tion of anatomical features, data from 3–5 sections per plant were
averaged. For quantitative assessment of organelles in M and
BS cells and GLDP immunodetection, the data from 10 imaged
cells per cell type per plant were averaged. For leaf gas exchange,
4–15 measurements were conducted on 4–7 plants per species. In
Neurachne species, vascular tissue is surrounded by two layers of
cells, an outermost BS and innermost mestome sheath that func-
tions in the C
4
species as the site of CO
2
rexation (Hattersley
etal., 1986). Comparing data collected from the mestome sheath
of N.minor with data from the BS of the other grasses makes sta-
tistical comparisons invalid except for comparisons between either
the presence or absence of GDLP in M versus the BS or mestome
sheath cells. Hence, N. minor was not included in the statistical
tests involving these other grasses.
Results
Homolepis aturensis possesses structural features
common to C
2
species
Leaves of H. aturensis are anatomically similar to those of
the C
2
species S. hians and S. laxum with respect to M cell
structure (Fig.1). One layer of M cells extends from BS cells
to the adaxial and abaxial epidermis (Fig.1). Approximately
six M cells separate the BS of adjacent veins in S. laxum
whereas four cells separate the BS of adjacent veins in H.atu-
rensis and S. hians (Fig. 1; Supplementary Table S3). The
M:BS tissue ratio for S. laxum, H. aturensis, and S. hians is
~2 (Fig.1; Supplementary Table S3), which is similar to that
of the C
4
species P. virgatum but >2.5 times less than that of
the C
3
species D. oligosanthes and P. bisulcatum. Alarge M
and small BS volume contributes to an M:BS of almost 5 in
Se. viridis even though there are only two cells between each
vein (Supplementary Fig. S1C; Supplementary Table S3).
M cells in D. oligosanthes, P. bisulcatum, and Se. viridis are
more loosely packed and elongate than those of P.virgatum
(Supplementary Fig. S1). The cellular features of N. minor
are similar to those previously reported (Hattersley et al.,
1986), with 2–3 M cells between veins (Supplementary Table
S3).
The BS cells of H.aturensis contain chloroplasts arranged
around the periphery, with a signicantly greater number
clustered centripetally (Figs 1A, B, 2A; Supplementary Table
S4). Mitochondria and peroxisomes localize almost exclu-
sively to the centripetal BS pole (Fig. 2A; Supplementary
Table S4). These spatial arrays of chloroplasts, mitochon-
dria, and peroxisomes are similar in BS cells of S. hians
(Figs 1E, F, 2C; Supplementary Table S4). As observed for
H. aturensis and S. hians, a signicantly greater number of
mitochondria and peroxisomes are positioned at the centrip-
etal BS pole in S. laxum, although chloroplasts are arranged
equally around BS cells (Fig 1C, D; Supplementary Table
S4). BS mitochondria are commonly surrounded by chloro-
plasts in S. hians (Fig 3E), H. aturensis, and S. laxum in a
pattern similar to previous reports for Steinchisma species
(Brown etal., 1983). In contrast to centripetal chloroplasts
displayed with their long axis parallel to the BS wall in S.
laxum, the long axis of these chloroplasts in H. aturensis and
S. hians are perpendicular to the BS wall (Figs 1, 2). BS chlo-
roplasts are situated primarily in a centrifugal position in C
3
grasses D. oligosanthes and P. bisulcatum (Supplementary
Table S4; Supplementary Figs S1, S2), as noted in C
3
eudicots
(Muhaidat et al., 2011; Sage et al., 2013). Approximately
65–69% of mitochondria and 19% (D. oligosanthes) and 41%
(P. bisulctum) of peroxisomes, respectively, are located in the
centripetal position of the C
3
grasses (Supplementary Table
S4).
Quantitative parameters of organelle traits of BS and M
cells are summarized in Supplementary Table S5. BS orga-
nelle parameters of H. aturensis and S. hians are not statis-
tically different from each other except for a signicantly
greater number of mitochondria per planar cell area in S.
hians. Mitochondria planar area per planar BS cell area of
H. aturensis and S. hians is signicantly greater than for S.
laxum. In addition, mitochondria number per planar BS cell
area, chloroplast and peroxisome number, and planar area
per planar BS cell area are signicantly greater in S. hians
than in S. laxum. Results from a one-way ANOVA on ranks
comparing peroxisome planar area per planar BS cell area
between S. hians, S. laxum, and H. aturensis only indicated
that this trait was signicantly higher in S. hians and H. atu-
rensis than in S. laxum (P≤0.001). The C
3
grasses have the
lowest mitochondria planar area per planar BS cell area;
however, BS mitochondria parameters of C
4
grasses are
not different from those of the C
2
grasses H. aturensis and
S. hians. Mitochondria, peroxisomes, and chloroplasts are
abundant in mestome sheath cells of N. minor (Hattersley
etal., 1986; Sage etal., 2014). Unlike H. aturensis, S. hians,
and S. laxum, organelles are not polarized in their distribu-
tion (Supplementary Fig. S3), but are instead positioned
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

C
2
photosynthesis in grasses
|
3069
around the mestome sheath cell periphery (Hattersley etal.,
1986; Sage etal., 2014).
When considering M cell organelle features, signicantly
fewer mitochondria and peroxisomes are observed in C
4
spe-
cies relative to the C
3
species and the C
2
species H. aturensis,
S. hians, and S. laxum (Supplementary Table S5). The C
4
spe-
cies as well as H. aturensis have signicantly lower chloro-
plast planar area per M planar cell area relative to S. hians, S.
laxum, and the C
3
grass species, and these changes result from
either smaller (H. aturensis, P. virgatum) or fewer (Se. viridis)
M cell chloroplasts (Supplementary Table S5). M cells of C
4
-
like and C
4
Flaveria and other C
4
species have recently been
reported to have signicant reductions in chloroplast volume
(Stata etal., 2014, 2016).
Homolepis aturensis exhibits C
2
levels of GLDP in
bundle sheath and mesophyllcells
Results from quantication of gold particles conjugated
to secondary antibodies that bind to anti-GLDP are sum-
marized in Supplementary Table S5. GLDP is almost
exclusively located in BS cells of H. aturensis and S. hians,
and mestome sheath cells of N. minor (Fig. 3A, B, E, F;
Supplementary Fig. S3). Both BS and M cells contained high
levels of GLDP labeling in S. laxum (Fig.3C, D). In com-
parison with all other C
3
species examined, S. laxum had the
highest GLDP labeling in BS mitochondria (Supplementary
Table S5). These patterns in GLDP distribution in M and
BS mitochondria of S. laxum and S. hians are similar to
earlier reports on these species (Hylton et al., 1988). The
gold density in C
3
species D. oligosanthes and P. bisulcatum
is higher in mitochondria of M than BS cells and is signi-
cantly greater on a planar M cell area basis than in those of
H. aturensis, S. hians, and the C
4
species (Supplementary
Table S5; Supplementary Fig. S4).
Homolepis aturensis exhibits C
2
levels of Rubisco in
bundle sheath and mesophyllcells
Rubisco occurs in both M and BS cell chloroplasts of
H. aturensis, and this pattern was similar to that observed
in S. hians, S. laxum, and M and mestome sheath cells of
N. minor (Supplementary Fig. S5). As expected, Rubisco
labeling is present only in the BS chloroplasts of the C
4
spe-
cies (Supplementary Fig. S6). Although Rubisco is also pre-
sent in the M and BS cells of the C
3
species D. oligosanthes
and P. bisulcatum, there is qualitatively less labeling in the BS
cells (Supplementary Fig. S6).
Fig.1. Leaf cross-sections from (A, B) Homolepis aturensis (Hoat), (C, D) Steinchisma laxum (Stla), and (E, F) S.hians (Sthi). M, mesophyll; asterisk,
bundle sheath. Scale bars=50µm.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

3070
|
Khoshravesh etal.
Homolepis aturensis exhibits photosynthetic
characteristics of a C
2
species
Values of A at 400µmol CO
2
mol
−1
air are statistically simi-
lar, being 21 ± 3 (mean±range) µmol m
−2
s
−1
for all species in
the study except for the C
4
plant Se.viridis, which has higher
A (Table1). Thus, any differences in carboxylation efciency
(CE), Γ, or intrinsic water use efciency (estimated as A/g
s
at 400 µmol CO
2
mol
−1
air) should reect photosynthetic
pathway effects and not variation in photosynthetic capac-
ity. Comparison of the A versus C
i
responses at saturating
light intensities show three equivalent sets of curves that cor-
responded to the photosynthetic pathway (Fig.4). C
3
and C
2
species have similar responses, with the exception that Γ was
reduced ≥30µmol mol
−1
in the C
2
species; consequently, their
A/C
i
responses were shifted to lower C
i
values. Homolepis atu-
rensis has A/C
i
responses identical to those of S.hians and
N.minor, leading us to classify H.aturensis as a C
2
species.
Carboxylation efciencies and A/g
s
of all the C
2
and C
3
spe-
cies are statistically identical and less than those of the C
4
species (Table1).
In the C
3
species D. oligosanthes and P. bisulcatum, we
observed C
*
values near 50µmol mol
−1
(Table1) which is typi-
cal for C
3
species at 31 °C (Busch etal., 2013). Steinchisma
laxum had a similar C
*
value (53µmol mol
−1
; Table1; Fig.5
A) indicating it is functionally C
3
despite the potential func-
tion of the BS organelles. The C
*
values of the C
2
species were
10.8−20 µmol mol
−1
(Table1; Fig.5B–D); the 10.8 value is on
the lower end of C
*
for species with this physiology (Edwards
and Ku, 1987). In the grasses studied here, lower C
*
values cor-
respond to higher values of BS mitochondria per planar cell
area (Fig.6A) and higher BS GLDP density per planar cell
area, except for Se. viridis (Fig.6C). C
*
was lower in species
where BS chloroplast area per planar cell area was greater and
M chloroplast area per planar cell area was lower (Fig.6E,F).
There is little observed shift to lower C
i
in the high light
response of A versus C
i
in H.aturensis, such that all A versus
C
i
curves converge near a common intersection point (Fig.5).
In S.hians and N. minor there is a slight reduction by a C
i
of
~5-7µmol mol
−1
in the high light response. There is no change
in Γ in the C
3
species with variation in light intensity (Fig.7),
while in each of the C
2
species, Γ increases at the lower light
intensity (Fig. 7). Neurachne minor exhibits the greatest
increase in Γ as light declines, while the light response of Γ is
negligible in H.aturensis above 300µmol m
−2
s
−1
. Notably, Γ
of N.minor is twice that of H.aturensis across the range of
light intensities.
GDC BS specificity in C
2
and C
4
grasses probably
results from changes in expression of a single
GLDPgene
Phylogenetic analyses reveal that within the Poaceae, O.sativa
is the only species examined with two gene copies encoding
GLDP, and these sequences are most closely related to each
other, even with inclusion of 14 additional BEP clade spe-
cies (Fig. 8). We nd no evidence of broader GLDP gene
duplication or loss of gene copies in C
4
grass species. Two
paralogs encoding GLDH are present in all angiosperms,
owing to an ancient duplication event (Supplementary Figs
S7, S8). The two paralogs were treated independently as
GLDH1 and GLDH2. Preliminary analysis indicates that
there is a Poaceae-wide duplication only of GLDL, and tar-
geted assemblies were made independently for each paralog,
which we label GLDL1 and GLDL2 (Supplementary Fig.
S9). During each assembly, mapped reads were scrutinized
Fig.2. Bundle sheath ultrastructure of (A) Homolepis aturensis (Hato), (B)
Steinchisma laxum (Stla), and (C) S.hians (Sthi). BS, bundle sheath; C,
chloroplast; M, mesophyll; VB, vascular tissue; arrowheads, mitochondria.
Scale bars=2µm.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

C
2
photosynthesis in grasses
|
3071
manually for evidence of additional paralogs, and none was
detected. No reads were found for GLDL2 in the BEP grass
Dendrocalamus sinicus, suggesting that it may have been
lost in this species. Most other grass species possess both
GLDL gene copies, with the exception of Z. mays, which
also lacks GLDL2; S. bicolor, which shares a common C
4
origin with Z.mays, has both paralogs. For all grass species,
full GLDL1 and GLDL2 sequences from genomes and all
assemblies which were completed to the start codon were
strongly predicted to be mitochondrial-localized by TargetP
(Emanuelsson etal., 2000; data not shown). This is consistent
with targeting of the two copies in Arabidopsis (AT3G17240,
AT1G48030; Fig. S9), which are both mitochondrial
(Lutziger and Oliver, 2001; Rajinikanth etal., 2007). Aplas-
tidial dihydrolipoyl dehydrogenase exists in land plants as
well, but is more distantly related and likely dates to a much
earlier duplication (Lutziger and Oliver, 2000; Rajinikanth
etal., 2007). Finally, GLDT is present as a single-copy gene
in all species examined (Supplementary Fig. S10). While we
nd evidence of local duplication (GLDP, GLDH2) and con-
served paralogs (GLDL), we nd no evidence of C
4
lineage
loss of any GDC subunit genes.
Discussion
Photorespiratory glycine shuttling exhibited by C
2
species
is considered to be the evolutionary bridge from C
3
photo-
synthesis to C
4
photosynthesis, based largely on studies from
eudicot species, particularly Flaveria (Bauwe, 2011; Sage
Fig.3. Immunolocalization of GLDP in bundle sheath and mesophyll cells of (A, B) Homolepis aturensis, (C, D) Steinchisma laxum, and (E, F) S.hians. C,
chloroplasts; p, peroxisomes; black asterisk, mitochondria; white asterisk, mitochondria surrounded by chloroplast. Scale bars=500 nm.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

3072
|
Khoshravesh etal.
etal., 2013; Schulze et al., 2013; Mallmann et al., 2014). It
has been posited that C
2
photosynthesis may serve a bridging
role to C
4
photosynthesis in grasses as well (Schulze et al.,
2013), although the timing of trait acquisition, to include
GDC BS cell specicity and abundance, has been proposed
to differ between eudicots and monocots (Williams et al.,
2013). Here, we evaluated photosynthetic pathway character-
istics and cellular features of BS and M cells in H.aturensis,
a candidate C
2
species which branches in a position sister to a
C
4
clade in the subtribe Arthropogoninae. We used compara-
tive studies with conrmed C
2
grasses, S. hians and N. minor,
and the C
3
grass S. laxum to facilitate our classication of
the photosynthetic type of H. aturensis. In addition, we con-
ducted a phylogenetic study of genes for the GDC subunits
to test the Schulze etal. (2013) inference that the evolution
of BS-specic GDC expression in C
4
grasses was similar
to that in Flaveria. From our physiological and structural
results, we conclude that H. aturensis is indeed a C
2
species,
supporting a hypothesis that the photorespiratory glycine
shuttle is a bridge to C
4
photosynthesis in grasses in the sub-
tribe Arthropogoninae. The characterization of S. laxum and
S. hians also allows us to conclude that activation of the BS
cells during transition from C
3
to C
2
in grasses in the sub-
tribe sister to the Arthropogoninae is similar to what has been
reported for eudicots with respect to BS organelle positioning
and organelle and GDC enrichment (Muhaidat etal., 2011;
Sage etal., 2013). Finally, our phylogenetic and immunohis-
tochemical data are consistent with the notion that BS GDC
in C
2
and C
4
grasses results from changes in expression levels
of a single GLDP gene in M and BScells.
Homolepis aturensis exhibits characteristic features com-
mon to species that concentrate photorespired CO
2
in BS cells
using the C
2
metabolic cycle. One of the key proteins essen-
tial for GDC activity, GLDP, localizes almost exclusively in
BS mitochondria in H. aturensis and S. hians. This pattern is
ubiquitous in C
2
eudicots (Hylton etal., 1988; Voznesenskaya
etal., 2001; Ueno and Sentoku, 2006; Voznesenskaya etal.,
2010; Muhaidat etal., 2011; T.L. Sage etal., 2011). Asecond
characteristic of C
2
species is an abundance of Rubisco in
M and BS cells (Monson etal., 1984). Rubisco is abundant
in M and BS cells in H. aturensis and S. hians, in contrast to
the typical C
3
pattern (high M Rubisco; low BS Rubisco) and
C
4
pattern (Rubisco only in BS cells). Lastly, the number of
Fig.4. Response of net CO
2
assimilation rate to intercellular CO
2
concentration in Homolepis aturensis, two C
2
species (Neurachne minor
and Steinchisma hians), a C
4
species (Setaria viridis), a C
3
species
(Dicanthelium oligosanthes), and a C
3
species with proto-Kranz anatomy
(Steinchisma laxum). Measurement conditions were 31 ± 1°C and
saturating light intensities (1200–1500µmol m
−2
s
−1
). The curves shown
are representative of 2–3 individual A/C
i
responses per species. The linear
regression for points below 100µmol mol
−1
is shown for H.aturensis
(dashed line).
Table1. Summaries of gas exchange values for species included in this study. Values are means ±SE.
Species n C
*
µmol mol
−1
CE
mol m
−2
s
−1
A
at 400
µmol m
−2
s
−1
A/g
s at 400
µmol mol
−1
C
3
species
Dicanthelium oligosanthes 3, 6 48 ± 1 a 0.13 ± 0.01 b 24.5 ± 1.5 b 56 ± 2
Panicum bisulcatum 6, 6 50 ± 2 a 0.11 ± 0.01 b 18.7 ± 1.7 b 62 ± 8
C
3
-Protokranz
Steinchisma laxum 5, 7 53 ± 1 a 0.12 ± 0.0 b 21.7 ± 2.4 b 62 ± 11
C
2
species
Homolepis aturensis 4, 7 10.8 ± 1.4 c 0.09 ± 0.01 b 19.8 ± 1.4 b 57 ± 10
Neurachne minor 3, 3 20.0 ± 2.5 b 0.11 ± 0.02 b 17.7 ± 1.9 b 43 ± 4
Steinchisma hians 5, 8 11.8 ± 1.4 c 0.10 ± 0.01 b 21.6 ± 1.1 b 58 ± 7
C
4
species
Panicum virgatum 0, 2 NA 0.36 ± 0.01 a 24.5 ± 2.5 b 152 ± 5
Setaria viridis 0, 3 NA 0.54 ± 0.05 a 31.9 ± 0.8 a 76 ± 12
Measurement temperature was 31 ± 1°C.
Sample sizes given are n=3-5 for C
*
estimates, and n=3–8 for all other data, with the exception of P. virgatum where n= 2).
Letters indicate statistical groupings at P<0.5 via one-way ANOVA followed by a Student’s–Neumann–Kuehls test.
NA, not applicable.
400 refers to ambient CO
2
concentration in µmol mol
−1
.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

C
2
photosynthesis in grasses
|
3073
M cells between BS cells in H. aturensis and S. hians results
in a reduced M:BS ratio and increased vein density common
to C
2
species (reviewed in Sage etal., 2012). These features
promote rapid ux of photorespiratory metabolites between
M and BS compartments (Monson and Rawsthorne, 2000),
improve water relations under high photorespiratory condi-
tions (Osborne and Sack, 2012), and facilitate an increase in
the volume of leaf tissue where photorespired CO
2
is concen-
trated around Rubisco in C
2
species.
As the rexed fraction of photorespiratory CO
2
increases,
C
*
(and Γ) declines (von Caemmerer, 2000). The effectiveness
of the C
2
process in H. aturensis as well as S. hians is reected
in the Γ and C
*
values that are at the lower range of values
reported for C
2
species (Holaday etal., 1984; Monson etal.,
1984; Ueno et al., 2003; Vogan et al., 2007; Voznesenskaya
et al., 2010; T.L. Sage et al., 2011, 2013). Values of Γ and
C
*
below 15 ppm indicate either that a C
4
cycle is active to
complement the C
2
cycle, or that the CO
2
trap in the inner BS
is particularly effective at recapturing the photorespired CO
2
(von Caemmerer, 2000). Steinchisma hians has weak to negli-
gible C
4
cycle activity (Edwards etal., 1982), and cellular char-
acteristics of organelle orientation in BS cells may explain the
low C
*
in H. aturensis and S. hians. The organelle-enriched
BS cells of these two species exhibit polarity in organelle
positioning such that almost all peroxisomes and GDC-
containing mitochondria, and over half of the chloroplasts
are situated adjacent to the vascular tissue. This arrange-
ment of BS organelles is posited to be particularly effective in
enhancing recapture efciency of photorespired CO
2
before it
can escape the BS cell in C
2
species (Rawsthorne, 1992). One
notable feature we observed is that the parallel orientation of
the centripetal BS chloroplasts to the inner wall in S. laxum
shifts to a perpendicular orientation in S. hians. A similar
perpendicular chloroplast orientation is present in H. aturen-
sis. This pattern allows for packaging of the more numerous,
larger chloroplasts in the centripetal position, which could
be important for increasing the surface area for rexing pho-
torespired CO
2
from adjacent mitochondria. Moreover, BS
mitochondria are physically surrounded by chloroplasts in S.
hians and H. aturensis, and these close physical associations
have been proposed to enhance rexation of photorespired
CO
2
(Brown et al., 1983).
The ne structure of C
2
BS cells is dened as C
2
Kranz,
reecting a view that this photosynthetic carbon-concentrat-
ing mechanism is associated with its own enabling Kranz-
like structure (Sage etal., 2014). Multiple convergence of C
2
Kranz in eudicots and grasses is strong evidence that this par-
ticular BS anatomy is specically adapted for the C
2
pathway.
The earliest recognizable subcellular events that ‘increase the
accessibility’ (Grass Phylogeny Working Group II, 2012) of
C
2
Kranz from C
3
in eudicots are an enhancement in num-
bers and size of mitochondria per BS cell, and positioning of
Fig.5. Representative A/C
i
responses below 80µmol mol
−1
determined on single leaves at four distinct light intensities of (A) Panicum virgatum (C
4
) and
Steinchisma laxum (C
3
proto-Kranz), (B) the C
2
species Steinchisma hians, (C) the C
2
species Homolepis aturensis, and (D) the C
2
species Neurachne
minor. Measurement light intensities are indicated beside each curve in µmol photons m
−1
s
−1
. The C
*
estimate is indicated by arrows. Curves shown are
representative of 3–6 measurement sets per species, except for P.virgatum where two sets of measurements were obtained. Measurement temperature
was 31 ± 1°C.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

3074
|
Khoshravesh etal.
these organelles from the centrifugal C
3
position to the cen-
tripetal BS pole. The BS chloroplast numbers also increase
in tandem with alterations in mitochondria placement, along
with a rearrangement of many, but not all, BS chloroplasts
from the centrifugal to centripetal pole (Muhaidat et al.,
2011; Sage et al., 2013; Voznesenskaya et al., 2013). The
anatomy associated with these earliest subcellular events has
been termed proto-Kranz (Muhaidat etal., 2011; Sage etal.,
2013). The transition to full C
2
BS patterns from proto-Kranz
in eudicots results from further amplication in centripetal
mitochondria volume (size and numbers) and relocation of
a greater fraction of enlarged chloroplasts to the centripetal
pole (Muhaidat etal., 2011; Sage etal., 2013). Proto-Kranz
and the shift to C
2
Kranz occurs with increasing vein den-
sity in Flaveria and Heliotropium (Muhaidat etal., 2011; Sage
etal., 2013).
The subcellular framework of C
3
S. laxum BS cells and
subsequent changes to that conguration from S. laxum to C
2
S. hians are similar to those observed in eudicots, supporting
a hypothesis that proto-Kranz facilitates the C
3
to C
2
transi-
tion (Sage et al., 2014). In comparison with the C
3
species
D. oligosanthes and P. bisulcatum, mitochondria and peroxi-
somes are situated along the centripetal poles of BS cells in S.
laxum, classifying this species as proto-Kranz. Previous char-
acterizations of proto-Kranz species have not presented data
on peroxisomes, and this additional focus in the present study
provides critical information on the positioning of the other
organelle involved in C
2
photosynthesis in the BS. A3-fold
increase in mitochondrial volume and corresponding increase
in GDC, as well as a 2.5-fold increase in peroxisome volume,
and 2-fold increase in chloroplast volume accompany the
transition to the C
2
BS pattern in S. hians. Also, as observed
in eudicots, the increase in BS chloroplast volume in S. hians
from proto-Kranz is associated with more of these organelles
in the centripetal location. Notably, although the number and
size of mitochondria per BS cell area are lower in S. laxum
than in S. hians, the GLDP label intensity is similar per mito-
chondrion and signicantly higher than that observed in the
Fig.6. C
*
(Γ, for C
4
species), organelle traits, and GLDP labeling in bundle sheath (A, C, E) and mesophyll (B, D, F) cells of Dichanthelium oligosanthes
(triangle, C
3
), Panicum bisulcatum (inverted triangle, C
3
), Steinchisma laxum (diamond, C
3
), S.hians (circle, C
2
), Homolepis aturensis (hexagon, C
2
),
Se.viridis (square, C
4
), and P.virgatum (star, C
4
). Filled and open symbols represent mitochondria and peroxisomes, respectively. Mean ±SE.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

C
2
photosynthesis in grasses
|
3075
C
3
grasses. These results indicate that increased BS GDC per
mitochondrion is also a functionally important development
early in C
2
evolution in grasses. The enhanced BS GDC den-
sity per mitochondrion in S. laxum is present in tandem with
C
3
-like values of vein density. The C
2
levels of BS GDC in
S. hians are present in high vein density leaves. The high lev-
els of BS GDC in S. laxum and S hians contrasts with the
theoretical predictions of Williams etal. (2013) who modeled
C
4
evolution in eudicots and monocots based on observed
patterns of trait acquisition in C
3
–C
4
intermediates. For
monocots, they predicted that an increase in vein density pre-
ceded enhanced GDC specicity and abundance in BS cells.
However, their model relied on a relatively small data set with
signicant gaps. The results here indicate that C
2
evolution in
grasses follows a pattern more typical of eudicots, which the
model of Williams etal. (2013) may support when reparam-
eterized with a richer data set.
In Neurachne, as in many other grasses, the mestome
sheath cell is the site of the Calvin–Benson cycle in C
4
spe-
cies (Hattersley and Browning, 1981; Hattersley etal., 1986;
Dengler and Nelson, 1999; Edwards and Voznesenskaya,
2011). We demonstrate that the mestome sheath cells in the
C
2
species N. minor are functionally similar to the C
2
BS cells
of H. aturensis and S. hians because GDC is almost exclu-
sively located in organelle-enriched mestome sheath cells in
the high vein density leaves. However, unlike H. aturensis and
S. hians, there is no polarized orientation of organelles in the
GDC-enriched mestome sheath cells of N. minor (Hattersley
etal., 1986; this study), indicating that N.minor utilizes a dif-
ferent strategy from H. aturensis and S. hians to trap photore-
spired CO
2
. In Neurachne, as in many other grasses, the thick
mestome sheath cell wall with a suberized lamella becomes the
Fig.7. The response of the CO
2
compensation point of A (Γ) as a function
of measurement light intensity in the C
2
species Homolepis aturensis,
Neurachne minor, Steinchisma hians, the proto-Kranz species Steinchisma
laxum, and the C
3
species Dicanthelium oligosanthes and Panicum
bisulcatum. Values of Γ were determined from the A/C
i
curves used in the
sequence of measurements to determine C
*
. Symbols are means ±SE
(n=2–6), except where no error bar is shown, in which case n=1.
Fig.8. A Bayesian phylogenetic tree of GLDP nucleotide sequences from 24 grasses, three Brassicaceae species, and Amborella. Long branches
between distantly related groups are condensed for visibility, denoted with a gap. Dashed red lines denote inferred gene duplication events. Numbers at
nodes indicate posterior probability.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

3076
|
Khoshravesh etal.
trap (Hattersley and Browning, 1981; Hattersley etal., 1986;
Dengler and Nelson, 1999; Edwards and Voznesenskaya,
2011). The C
2
species Alloteropsis semialata ssp semialata
has a similar C
2
Kranz anatomy to N. minor; GDC levels
are highest in mestome sheath cells with a suberized lamella
and the abundant organelles are equally partitioned within
those cells (Hattersley and Browning, 1981; Ueno and
Sentoku, 2006). Intriguingly, although organelle orientation
in mestome sheath cells is not important in the evolution of
C
2
photosynthesis in N. minor, chloroplasts do have a cen-
trifugal orientation in C
4
Neurachne species (Hattersley etal.,
1986). Qualitative observations on C
3
Neurachne species
indicate that some of the C
3
species have enhanced numbers
of chloroplasts and mitochondria in mestome sheath cells
(Hattersley etal., 1986), leading us to posit that organelle and
GDC enrichment may have been important during the early
stages of C
2
evolution in thegenus.
The evolutionary transition from C
3
to C
2
has been pro-
posed rst to involve a change in cell type-specic expression
of GDC from M to BS in tandem with a loss of M GDC
(Bauwe, 2011). A comparison of the cellular site of GDC
expression in proto-Kranz S. laxum with that of the sister
species S. hians indicates that the severe reduction in M GDC
in the C
2
species is preceded by increased expression in BS
cells. This is consistent with patterns observed in the eudicots
Flaveria and Heliotropium, and supports a model of gradual
GDC loss in M cells following a physiological activation of
the BS (Muhaidat etal., 2011; Sage etal., 2013; Schulze etal.,
2013). C
3
species of Flaveria contain two copies of the gene
encoding GLDP resulting from gene duplication (Schulze
etal., 2013). One of these is BS dominant in expression and
the second is expressed ubiquitously throughout the leaf in
C
3
Flaveria (Schulze et al., 2013). During the evolution of
C
4
photosynthesis, the loss of M GDC function in Flaveria
resulted from pseudogenization of the gene coding for the
ubiquitously expressed GLDP (Schulze etal., 2013). Schulze
etal. (2013) speculated that BS-dominant GLDP expression
arose in a similar manner in C
4
grasses, because O.sativa (C
3
BEP clade) has two GLDP genes, but Z.mays and other C
4
species in the C
4
PACMAD clade have only one. To provide an
understanding of the evolution of BS-specic GDC expres-
sion in C
2
and C
4
grasses, we conducted phylogenetic analyses
of genes encoding GLDP using 17 BEP and seven PACMAD
grass species, three Brassicaceae species, and Amborella as
outgroup. Our analyses demonstrate that, with the exception
of O. sativa, all grasses have one copy of GLDP. The two
copies of the genes encoding GLDP in rice are more closely
related to each other than to any other GLDP gene included
in the analysis and therefore represent a local duplication.
Combined, the phylogenetic and immunohistochemical
observations on C
3
and C
4
PACMAD species are consistent
with a hypothesis that BS-dominant quantities of GDC in
C
2
and C
4
grasses resulted from modications in regulatory
mechanisms controlling the levels of expression of a single
GLDP gene present in M and BS cells; C
3
species have high
levels of M GDC and low levels of BS GDC, and the oppo-
site pattern is present in the C
2
and C
4
species. Mesophyll
tissue specicity of phosphoenolpyruvate carboxylase has
evolved through modication of cis-regulatory elements in
C
4
Flaveria (Gowik etal., 2004). Studies examining promoter
regions of the GLDP subunits in closely related C
3
, C
2
, and
C
4
species should provide insights into the evolution of the
regulatory mechanisms that confer the requisite expression
patterns for C
2
and subsequently C
4
photosynthesis in grasses.
Since it is conceivable that BS specicity of the GDC
complex arose via duplication and pseudogenization of one
of the other subunits, we also included analyses for GLDL,
GLDH1 and GLDH2, and GLDT. In land plants, GLDH1
functions in photorespiration and GLDH2 is associated
with C1 metabolism (Rajinikanth et al., 2007). We found
no evidence of broad duplication for either of the ancient
GLDH copies or GLDT in the grasses; however, GLDL is
encoded by two conserved paralogs in most grass species,
and may have arisen from the whole-genome duplication in
the ancestor of the grass family. GLDL is the only GDC
subunit gene for which we nd evidence of subfunctional-
ized and conserved paralogs analogous to the two GLDP
copies in Flaveria. There is, however, no evidence that either
of these copies has been lost or pseudogenized (e.g. via
nonsense or frameshift mutation) in any C
4
species except
Z.mays, which has a local duplication of GLDL1 and lacks
a gene encoding GLDL2. Yet it is unlikely that this loss was
involved in the evolution of C
4
photosynthesis, as GLDL2 is
present in S.bicolor, which shares a common C
4
origin with
maize (Grass Phylogeny Working Group II, 2012). The two
GLDL paralogs in grasses may be partially reduntant with
one another (Rajinikanth et al., 2007). The nature of the
subfunctionalization of GLDL which resulted in the evolu-
tionary retention of two paralogs in Poaceae is not known,
but our results indicate that these paralogs did not play a
role in the evolution of C
4
photosynthesis analogous to the
GLDP paralogs in Flaveria. It is similarly unlikely that the
two ancestral copies of GLDH played a role in C
4
evolution
analogous to that of GLDP in Flaveria as both are present
in all grass species examined here, and only GLDH1 plays a
role in photorespiration (Rajinikanth etal., 2007).
Conclusion
Evolution of C
2
photosynthesis in the grasses and eudicots is
conferred by organelle enrichment, BS- or mestome sheath-
dominant GDC accumulation, and centripetal positioning of
organelles when the BS is the carbon-concentrating tissue. In
Steinchisma, the earliest recognized subcellular event that facil-
itates C
2
photosynthesis is the placement of BS mitochondria
and peroxisomes exclusively to the centripetal pole. This fea-
ture, also present in eudicots, is posited to set in motion a feed-
forward facilitation cascade that leads to C
2
and subsequently
C
4
photosynthesis (Sage etal., 2013). How and why changes in
BS or mestome sheath GDC levels and organelle volume, and
BS organelle positioning were initiated during the early stages
of C
2
evolution in grasses remains a mystery. Identication of
mechanisms controlling these processes should be a primary
focus of research. The Arthropogoninae, the subtribe contain-
ing Steinchisma (Otachyrinae), and Neurachninae will be key
to these future studies.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

C
2
photosynthesis in grasses
|
3077
Supplementarydata
Supplementary data are available at JXB online.
Table S1. List of species studied and source of species.
Table S2. Species for which RNA-seq data were down-
loaded and assembled for phylogenetic analysis of GDC
subunits.
Table S3. Anatomical parameters of C
3
, C
2
, and C
4
leaves.
Table S4. Organelle distribution in bundle sheath cells of
C
3
and C
2
species.
Table S5. Quantication of organelle numbers, size, and
density of gold labeling (GLDP) in mesophyll and bundle
sheath cells of C
3
, C
2
, and C
4
leaves.
Figure S1. Light micrographs of C
3
and C
4
species.
Figure S2. Bundle sheath cell ultrastructure of C
3
and C
4
species.
Figure S3. Leaf structure and anatomy and immunolocali-
zation of GLDP in N. minor.
Figure S4. Immunolocalization of GLDP in M and BS
cells of C
3
and C
4
species.
Figure S5. Immunolocalization of Rubisco large subunit in
M and BS cells of C
3
and C
2
species.
Figure S6. Immunolocalization of Rubisco large subunit in
M and BS cells of C
3
and C
4
species.
Figure S7. ABayesian phylogenetic tree of GLDH1.
Figure S8. ABayesian phylogenetic tree of GLDH2.
Figure S9. ABayesian phylogenetic tree of GLDL.
Figure S10. ABayesian phylogenetic tree of GLDT.
Acknowledgements
We thank E.Kellog, T.Brutnell, and M.Tanaiguchi for providing seed, and
N.Dakin for assistance with preparation of N. minor leaves. This research
was supported by funding from the Natural Science and Engineering
Research Council of Canada (grant nos 2015-04878 to TLS and 154273-2012
to RFS) and an Australian Research Council Discovery Project Grant to
ML, TLS, and RFS (DP130102243).
References
Bauwe H. 2011. Photorespiration: the bridge to C
4
photosynthesis. In:
Raghavendra AS, Sage RF, eds. C
4
photosynthesis and related CO
2
concentrating mechanisms. Advances in Photosynthesis and Respriration,
Vol. 32. Dordrecht, The Netherlands: Springer, 81–108.
Brown H, Bouton JH, Rigsby L, Rigler M. 1983. Photosynthesis
of grass species differing in carbon dioxide fixation pathways. VIII.
Ultrastuctural characteristics of Panicum species in the LAXA group. Plant
Physiology 71, 425–431.
Busch FA, Sage TL, Cousins AB, Sage RF. 2013. C
3
plants enhance
rates of photosynthesis by reassimilating photorespired and respired CO
2
.
Plant, Cell and Environment 36, 200–212.
Capella-Gutierrez S, Silla-Martinez J, Gabaldon T. 2009. trimAl: a tool
for automated alignment trimming in large-scale phylogenetic analyses.
Bioinformatics 25, 1972–1973.
Christin P-A, Osborne CP, Chatelet DS, Columbus JT, Besnard
G, Hodkinson TR, Garrison LM, Vorontsova MS, Edwards EJ.
2013. Anatomical enablers and the evolution of C
4
photosynthesis in
grasses. Proceedings of the National Academy of Sciences, USA 110,
1381–1386.
Christin PA, Wallace MJ, Clayton H, Edwards EJ, Furbank RT,
Hattersley PW, Sage RF, Macfarlane TD, Ludwig M. 2012. Multiple
photosynthetic transitions, polyploidy, and lateral gene transfer in the grass
subtribe Neurachninae. Journal of Experimental Botany 63, 6297–6308.
Dengler NG, Nelson T. 1999. Leaf structure and development in
C
4
plants In: Sage RF, Monson RK, eds. C
4
plant biology . New York:
Academic Press, 133–172.
Edgar RC. 2004. MUSCLE: multiple sequence alignment with high
accuracy and high throughput. Nucleic Acids Research 32, 1792–1797.
Edwards GE, Ku MSB. 1987. Biochemistry of C
3
–C
4
intermediates.
In: Hatch MD, Boardman HK, eds. The biochemistry of plants, Vol. 10.
London: Academic Press, 275–325.
Edwards GE, Ku MSB, Hatch MD. 1982. Photosynthesis in Panicum
milioides, a species with reduced photorespiration. Plant and Cell
Physiology 23, 1185–1195.
Edwards GE, Voznesenskaya EV. 2011. C
4
photosynthesis: Kranz forms
and single-cell C
4
in terrestrial plants. In: Raghavendra AS, Sage RF, eds.
C
4
photosynthesis and related CO
2
concentrating mechanisms. Advances
in Photosynthesis and Respriration, Vol. 32. Dordrecht, The Netherlands:
Springer, 29–61.
Emanuelsson O, Nielsen H, Brunak S, von Heijne G. 2000. Predicting
subcellular localization of proteins based on their N-terminal amino acid
sequence. Journal of Molecular Biology 300, 1005–1016.
Fisher AE, McDade LA, Kiel CA, Khoshravesh R, Johnson MA,
Stata M, Sage TL, Sage RF. 2015. Evolutionary history of Blepharis
(Acanthaceae) and the origin of C
4
photosynthesis in section Acanthodium.
International Journal of Plant Sciences 176, 770–790.
Gowik U, Burscheidt J, Akyildiz M, Schlue U, Koczor M, Streubel
M, Wessthoff P. 2004. cis-Regulatory elements for mesophyll-specific
gene expression in the C
4
plant Flaveria trinervia, the promoter of the C
4
phosphoenolpyruvate carboxylase gene. The Plant Cell 16, 1077–1090.
Grass Phylogeny Working Group II. 2012. New grass phylogeny
resolves deep evolutionary relationships and discovers C
4
origins. New
Phytologist 193, 304–312.
Hattersley PW, Browning AJ. 1981. Occurrence of the suberized lamella
in leaves of grasses of different photosynthetic types. I.In parenchymatous
bundle sheaths and PCR (‘Kranz’) sheaths. Protoplasma 109, 371–401.
Hattersley PW, Wong S-C, Perry S, Roksandic Z. 1986. Comparative
ultrastructure and gas exchange characteristics of the C
3
–C
4
intermediate
Neurachne minor S.T. Blake (Poaceae). Plant, Cell and Environment 9,
217–233.
Heckmann D, Schulze S, Denton A, Gowik U, Westhoff P, Weber
APM, Lercher MJ. 2013. Predicting C4 photosynthesis evolution:
modular, individually adaptive steps on a Mount Fuji fitness landscape. Cell
153, 1579–1588.
Hibberd JM, Sheehy JE, Langdale JA. 2008. Using C
4
photosynthesis
to increase the yield of rice—rationale and feasibility. Current Opinion in
Plant Biology 11, 228–231.
Holaday AS, Lee KW, Chollet R. 1984. C
3
−C
4
intermediate species in
the genus Flaveria: leaf anatomy, ultrastructure, and the effect of O
2
on the
CO
2
compensation concentration. Planta 160, 25–32.
Hylton CM, Rawsthorne S, Smith AM, Jones DA, Woolhouse HW.
1988. Glycine decarboxylase is confined to the bundle-sheath cells of
leaves of C
3
–C
4
intermediate species. Planta 175, 452–459.
Keerberg O, Pärnik T, Ivanova H, Bassüner B, Bauwe H. 2014. C
2
photosynthesis generates about 3-fold elevated leaf CO
2
levels in the
C
3
–C
4
intermediate species Flaveria pubescens. Journal of Experimental
Botany 65, 3649–3656.
Khoshravesh R, Akhani H, Sage TL, Nordenstam B, Sage RF. 2012.
Phylogeny and photosynthetic pathway distribution in Anticharis Endl.
(Scrophulariaceae). Journal of Experimental Botany 63, 5645–5658.
Lundgren MR, Christin PA, Gonzalez Escobar E, Ripley BS, Besnard
G, Long CM, Hattersley PW, Ellis RP, Leegood RC, Osborne CP.
2015. Evolutionary implications of C
3
–C
4
intermediates in the grass
Alloteropsis semialata. Plant, Cell and Environment (in press).
Lutziger I, Oliver DJ. 2000. Molecular evidence of a unique lipoamide
dehydrogenase in plastids: analysis of plastidic lipoamide dehydrogenase
from Arabidopsis thaliana. FEBS Letters 484, 12–16.
Lutziger I, Oliver DJ. 2001. Characterization of two cDNAs encoding
mitochondrial lipoamide dehydrogenase from Arabidopsis. Plant
Physiology 127, 615–623.
Lyu MJ, Gowik U, Kelly S, etal. 2015. RNA-Seq based phylogeny
recapitulates previous phylogeny of the genus Flaveria (Asteraceae) with
some modifications. BMC Evolutionary Biology 15, 116.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from

3078
|
Khoshravesh etal.
Mallmann J, Heckmann D, Brautigam A, Lercher MJ, Weber AP,
Westhoff P, Gowik U. 2014. The role of photorespiration during the
evolution of C
4
photosynthesis in the genus Flaveria. Elife 3, e02478.
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
C
3
to C
4
photosynthesis. The Plant Journal 51, 886–896.
McKown AD, Moncalvo JM, Dengler NG. 2005. Phylogeny of Flaveria
(Asteraceae) and inference of C
4
photosynthesis evolution. American
Journal of Botany 92, 1911–1928.
Monson R, Rawsthorne S. 2000. CO
2
assimilation in C
3
–C
4
intermediate
plants. In: Leegood R, Sharkey T, von Caemmerer S, eds. Photosynthesis,
Vol. 9. Dordrecht, The Netherlands: Springer, 533–550.
Monson RK, Edwards GE, Ku MSB. 1984. C
3
–C
4
intermediate
photosynthesis in plants. BioScience 34, 563–574.
Morgan JA, Brown RH. 1979. Photosynthesis in grass species
differing in carbon dioxide fixation pathways: II. Asearch for species
with intermediate gas exchange and anatomical characteristics. Plant
Physiology 64, 257–262.
Morgan CL, Turner SR, Rawsthorne S. 1993. Coordination of the
cell-specific distribution of the 4 subunits of glycine decarboxylase and of
serine hydroxymethyltransferase in leaves of C
3
–C
4
intermediate species
from different genera. Planta 190, 468–473.
Muhaidat R, Sage TL, Frohlich MW, Dengler NG, Sage RF. 2011.
Characterization of C
3
–C
4
intermediate species in the genus Heliotropium
L.(Boraginaceae): anatomy, ultrastructure and enzyme activity. Plant, Cell
and Environment 34, 1723–1736.
Osborne CP, Sack L. 2012. Evolution of C
4
plants: a new hypothesis for
an interaction of CO
2
and water relations mediated by plant hydraulics.
Philosophical Transactions of the Royal Society B: Biological Sciences
367, 583–600.
Peterhansel C. 2011. Best practice procedures for the establishment
of a C
4
cycle in transgenic C
3
plants. Journal of Experimental Botany 62,
3011–3019.
Rawsthorne S. 1992. C
3
–C
4
intermediate photosynthesis—linking
physiology to gene-expression. The Plant Journal 2, 267–274.
Rawsthorne S, Morgan CL, O’Neill CM, Hylton CM, Jones DA, Frean
ML. 1998. Cellular expression pattern of the glycine decarboxylase P
protein in leaves of an intergeneric hybrid between the C
3
–C
4
intermediate
species Moricandia nitens and the C
3
species Brassica napus. Theoretical
and Applied Genetics 96, 922–927
Ronquist F, Huelsenbeck JP. 2003. MrBayes 3: Bayesian phylogenetic
inference under mixed models. Bioinformatics 19, 1572–1574.
Rajinikanth M, Harding SA, Tsai CJ. 2007. The glycine decarboxylase
complex multienzyme family in Populus. Journal of Experimental Botany
58, 1761–1770.
Sage R. 2016. A portrait of the C
4
photosynthetic family on the 50th
anniversary of its discovery: species number, evolutionary lineages, and
hall of fame. Journal of Experimental Botany 67, 2919–2922.
Sage RF, Christin PA, Edwards EJ. 2011. The C
4
plant lineages of
planet Earth. Journal of Experimental Botany 62, 3155–3169.
Sage RF, Khoshravesh R, Sage TL. 2014. From proto-Kranz to C
4
Kranz: building the bridge to C
4
photosynthesis. Journal of Experimental
Botany 65, 3341–3356.
Sage RF, Sage TL, Kocacinar F. 2012. Photorespiration and the
evolution of C
4
photosynthesis. Annual Review of Plant Biology 63, 19–47.
Sage RF, Wedin DA, Li M. 1999. The biogeography of C
4
photosynthesis: patterns and controlling factors. In: Sage RF, Monson RK,
eds. C
4
plant biology . San Diego: Academic Press: 313–374.
Sage TL, Busch FA, Johnson DC, Friesen PC, Stinson CR, Stata M,
Sultmanis S, Rahman BA, Rawsthorne S, Sage RF. 2013. Initial events
during the evolution of C
4
photosynthesis in C
3
species of Flaveria. Plant
Physiology 163, 1266–1276.
Sage TL, Sage RF,Vogan PJ, Rahman B, Johnson DC, Oakley JC,
Heckel MA. 2011. The occurrence of C
2
photosynthesis in Euphorbia
subgenus Chamaesyce (Euphorbiaceae). Journal of Experimental Botany
62, 3183–3195.
Schneider CA, Rasband WS, Eliceiri KW. 2012. NIH Image to ImageJ:
25years of image analysis. Nature Methods 9, 671–675.
Schulze S, Mallmann J, Burscheidt J, Koczor M, Streubel M, Bauwe
H, Gowik U, Westhoff P. 2013. Evolution of C
4
photosynthesis in the
genus Flaveria: establishment of a photorespiratory CO
2
pump. The Plant
Cell 25, 2522–2535.
Stata M, Sage TL, Hoffman N, Covshoff S, Ka-Shu Wong G, Sage
R F. 2016. Mesophyll chloroplast investment in C
3
, C
4
and C
2
species of
the genus Flaveria. Plant and Cell Physiology 57 (in press).
Stata M, Sage TL, Rennie TD, Khoshravesh R, Sultmanis S, Khaikin
Y, Ludwig M, Sage RF. 2014. Mesophyll cells of C
4
plants have fewer
chloroplasts than those of closely related C
3
plants. Plant, Cell and
Environment 37, 2587–2600.
Still CJ, Berry JA, Collatz GJ, DeFries RS. 2003. Global distribution of
C
3
and C
4
vegetation: carbon cycle implications. Global Biogeochemical
Cycles 17.
Ueno O. 1992. Immunogold localization of photosynthetic enzymes
in leaves of Aristida latifolia, a unique C
4
grass with a double
chlorenchymatous bundle sheath. Physiologia Plantarum 85, 189–196.
Ueno O, Bang SW, Wada Y, Kondo A, Ishihara K, Kaneko Y,
Matsuzawa Y. 2003. Structural and biochemical dissection of
photorespiration in hybrids differing in genome constitution between
Diplotaxis tenuifolia (C
3
–C
4
) and radish (C
3
). Plant Physiology 132,
1550–1559.
Ueno O, Sentoku N. 2006. Comparison of leaf structure and
photosynthetic characteristics of C
3
and C
4
Alloteropsis semialata
subspecies. Plant, Cell and Environment 29, 257–268.
Vogan PJ, Frohlich MW, Sage RF. 2007. The functional significance of
C
3
–C
4
intermediate traits in Heliotropium L.(Boraginaceae): gas exchange
perspectives. Plant, Cell and Environment 30, 1337–1345.
von Caemmerer S. 2000. Biochemical models of leaf photosynthesis .
Australia: CSIRO publishing.
von Caemmerer S, Quick WP, Furbank RT. 2012. The development
of C
4
rice: current progress and future challenges. Science 336,
1671–1672.
Voznesenskaya EV, Artyusheva EG, Franceschi VR, Pyankov VI,
Kiirats O, Ku MSB, Edwards GE. 2001. Salsola arbusculiformis, a
C
3
–C
4
intermediate in Salsoleae (Chenopodiaceae). Annals of Botany 88,
337–348.
Voznesenskaya EV, Koteyeva NK, Akhani H, Roalson EH,
Edwards GE. 2013. Structural and physiological analyses in
Salsoleae (Chenopodiaceae) indicate multiple transitions among C
3
,
intermediate, and C
4
photosynthesis. Journal of Experimental Botany
64, 3583–3604.
Voznesenskaya EV, Koteyeva NK, Edwards GE, Ocampo G.
2010. Revealing diversity in structural and biochemical forms of
C
4
photosynthesis and a C
3
–C
4
intermediate in genus Portulaca
L.(Portulacaceae). Journal of Experimental Botany 61, 3647–3662.
Williams BP, Johnston IG, Covshoff S, Hibberd JM. 2013.
Phenotypic landscape inference reveals multiple evolutionary paths to C
4
photosynthesis. Elife 2, e00961.
by guest on May 15, 2016http://jxb.oxfordjournals.org/Downloaded from