J Plant Res (2017) 130:625634
DOI 10.1007/s10265-017-0922-8
REGULAR PAPER
Phylogenetic analysis ofproteins involved inthestringent
response inplant cells
DoshunIto1 YutaIhara1 HidenoriNishihara1 ShinjiMasuda2,3
Received: 9 June 2016 / Accepted: 7 February 2017 / Published online: 16 March 2017
The Botanical Society of Japan and Springer Japan 2017
Abstract The nucleotide (p)ppGpp is a second messen-
ger that controls the stringent response in bacteria. The
stringent response modifies expression of a large num-
ber of genes and metabolic processes and allows bacteria
to survive under fluctuating environmental conditions.
Recent genome sequencing analyses have revealed that
genes responsible for the stringent response are also found
in plants. These include (p)ppGpp synthases and hydro-
lases, RelA/SpoT homologs (RSHs), and the pppGpp-spe-
cific phosphatase GppA/Ppx. However, phylogenetic rela-
tionship between enzymes involved in bacterial and plant
stringent responses is as yet generally unclear. Here, we
investigated the origin and evolution of genes involved in
the stringent response in plants. Phylogenetic analysis and
primary structures of RSH homologs from different plant
phyla (including Embryophyta, Charophyta, Chlorophyta,
Rhodophyta and Glaucophyta) indicate that RSH gene
families were introduced into plant cells by at least two
independent lateral gene transfers from the bacterial Deino-
coccus-Thermus phylum and an unidentified bacterial phy-
lum; alternatively, they were introduced into a proto-plant
cell by a lateral gene transfer from the endosymbiotic
Electronic supplementary material The online version of this
article (doi:10.1007/s10265-017-0922-8) contains supplementary
material, which is available to authorized users.
* Shinji Masuda
shmasuda@bio.titech.ac.jp
1
Graduate School ofBioscience andBiotechnology, Tokyo
Institute ofTechnology, Yokohama2268501, Japan
2
Center forBiological Resources andInformatics, Tokyo
Institute ofTechnology, Yokohama2268501, Japan
3
EarthLife Science Institute, Tokyo Institute ofTechnology,
Tokyo1528551, Japan
cyanobacterium followed by gene loss of an ancestral RSH
gene in the cyanobacterial linage. Phylogenetic analysis of
gppA/ppx families indicated that plant gppA/ppx homologs
form an individual cluster in the phylogenetic tree, and
show a sister relationship with some bacterial gppA/ppx
homologs. Although RSHs contain a plastidial transit pep-
tide at the N terminus, GppA/Ppx homologs do not, sug-
gesting that plant GppA/Ppx homologs function in the
cytosol. These results reveal that a proto-plant cell obtained
genes for the stringent response by lateral gene transfer
events from different bacterial phyla and have utilized them
to control metabolism in plastids and the cytosol.
Keywords Chloroplast GppA ppGpp pppGpp Ppx
RelA/SpoT homolog
Introduction
The stringent response is a fundamental regulatory sys-
tem that enables bacterial cells to survive in fluctuating
environments. This response is mediated by the synthe-
sis and hydrolysis of the second messengers guanosine
5-triphosphate 3-diphosphate (pppGpp) and guanosine
5-diphosphate 3-diphosphate (ppGpp), the levels of which
are maintained in Escherichia coli by two enzymes, RelA
and SpoT (Cashel et al. 1996; Dalebroux and Swanson
2012; Hauryliuk et al. 2015; Potrykus and Cashel 2008).
Both enzymes synthesize (p)ppGpp by phosphorylating
GTP or GDP using ATP in response to different environ-
mental changes such as starvation for amino acids, iron,
nitrogen, phosphate or carbon. When GTP is the sub-
strate, pppGpp is initially produced but can be converted
to ppGpp by a specific phosphatase called guanosine
5-triphosphate 3-diphosphate pyrophosphatase (GppA)
13
Vol.:(0123456789)

626
or exopolyphosphatase (Ppx) (Hara and Sy 1983; Keasling
et al. 1993). In Escherichia coli, (p)ppGpp controls tran-
scription of a large number of genes through direct inter-
action with RNA polymerase and also controls translation,
DNA replication, GTP-binding proteins, proteases and
enzymes for purine biosynthesis, which is necessary for
adaptation to environmental stimuli (Potrykus and Cashel
2008). pppGpp and ppGpp have differing affinities for the
target proteins (Mechold et al. 2013), suggesting that the
ratio of the two nucleotides is physiologically important.
SpoT, but not RelA, can also hydrolyze (p)ppGpp under
certain conditions (Cashel etal. 1996; Potrykus and Cashel
2008). Thus, SpoT is a bifunctional protein that catalyzes
both (p)ppGpp synthesis and hydrolysis, whereas RelA has
only the (p)ppGpp synthase activity. SpoT and RelA pro-
teins are only found in - and -proteobacteria (Mitten-
huber 2001), although SpoT homologs, having (p)ppGpp
synthase and hydrolase domains, are widely distributed
among the bacterial kingdom, which are called Rel or Rsh
(Atkinson etal. 2011); we refer the bacterial homologs as
Rel hereafter.
The enzymatic activities of E. coli RelA and SpoT are
also differentially regulated. RelA associates with ribo-
somes and is activated when the ribosome A site is occu-
pied by uncharged tRNA (Cashel etal. 1996; Potrykus and
Cashel 2008). This allows bacteria to rapidly accumulate
(p)ppGpp upon amino acid starvation. RelA, SpoT and Rel
have a threonyl-tRNA synthetase, GTPase, SpoT (TGS)
domain and an aspartate kinase chorismate mutase TyrA
(ACT) domain in its C terminus, where acyl-carrier pro-
tein, ribosomal tRNA, and other regulatory proteins are
associated to control their (p)ppGpp synthesis and hydroly-
sis activities (Brown etal. 2016; Loveland etal. 2016; Pot-
rykus and Cashel 2008). This allows bacteria to control (p)
ppGpp levels in response to environmental changes.
GppA/Ppx protein families are widely distributed
among prokaryotes (bacteria and archaea), fungi, protists
and metazoa; they are classified into six groups: GPPases,
GPPase-like polyPases, high molecular-weight polyPases,
low molecular-weight polyPases, HD-domain polyPases,
and RTG2 proteins (Albi and Serrano 2014). E. coli GppA
and Ppx are classified into GPPases and GPPase-like poly-
Pases, respectively. Previous studies showed that enzy-
matic specificities for each GppA/Ppx family seem to dif-
fer. Specifically, E. coli GppA and Ppx liberate Pi not only
by hydrolysis of pppGpp to generate ppGpp, but also by
progressive hydrolysis of the terminal phosphoanhydride
bonds of the inorganic polyphosphate (polyP) (Hara and
Sy 1983; Keasling etal. 1993). Two GppA/Ppx homologs
belonging to the low molecular-weight polyPases from the
pathogenic bacterium Mycobacterium tuberculosis show
polyP phosphatase (polyPase) and ATPase activity, but do
not catalyze hydrolysis of pppGpp to ppGpp (Choi et al.
13
J Plant Res (2017) 130:625634
2012). Albi and Serrano (2014) characterized two GppA/
Ppx homologs belonging to the HD-domain polyPases
from the green-sulfur bacterium Chlorobium tepidum,
and showed that they are able to hydrolyze polyP, and one
of the two also has the nucleotide triphosphatase activ-
ity (Albi and Serrano 2014). These results suggested that
GppA/Ppx-like proteins have remarkable functional plastic-
ity and some of them do not mediate the (p)ppGpp-depend-
ent stringent response; they mediate only inorganic polyP
metabolism.
Recent genome sequence data have indicated that RelA/
SpoT-like proteins are conserved in algae and land plants
(Masuda 2012; Tozawa and Nomura 2011; van der Biezen
et al. 2000). In Arabidopsis, four genes encoding RelA/
SpoT homologs (RSHs) have been identified: RSH1,
RSH2, RSH3 and Ca2+-activated RSH (CRSH) (Tozawa
and Nomura 2011). All Arabidopsis RSHs are localized in
plastids (Maekawa etal. 2015; Masuda etal. 2008; Mizu-
sawa etal. 2008), indicating that they control plastidial (p)
ppGpp level. Arabidopsis RSH2 and RSH3 are paralogs
with high (~80%) amino acid sequence identity (Atkinson
et al. 2011; Givens et al. 2004; Masuda 2012), and thus
Arabidopsis RSHs are classified into three groups, namely
RSH1, RSH2/RSH3 and CRSH, that differ in primary
structure. Specifically, RSH1 does not have a conserved
Gly residue that is critical for (p)ppGpp synthase activ-
ity in Bacillus subtilis Rel (Wendrich and Marahiel 1997),
indicating that RSH1 does not have (p)ppGpp synthase
activity and has only (p)ppGpp hydrolase activity (Masuda
et al. 2008). CRSH does not have the conserved His-Asp
(HD) domain that is responsible for catalyzing (p)ppGpp
hydrolysis (Masuda etal. 2008; Tozawa etal. 2007). RSH2/
RSH3 contains both the conserved Gly residue and HD
domain (Givens et al. 2004). This suggests that plastidial
(p)ppGpp levels in Arabidopsis are balanced through dif-
ferent activities of the three RSH family enzymes. We pre-
viously showed that over-accumulation of ppGpp in Arabi-
dopsis chloroplasts induces pleiotropic phenotypes, such as
reduced chloroplast size and photosynthetic activities, indi-
cating that (p)ppGpp levels must be maintained within spe-
cific concentration to ensure robust control of plastid activi-
ties in plant cells (Maekawa etal. 2015). We also showed
that artificial over-accumulation of ppGpp in the cytosol
results in reduced plant growth (Ihara and Masuda 2016),
suggesting that (p)ppGpp produced in chloroplasts is trans-
ferred to the cytosol to optimize metabolic processes.
Although the mechanisms of the plastidial stringent
response are beginning to appear, it is unclear how RSH
genes were introduced into plant cells during evolution.
Previously, Givens et al. (2004) performed a phylogenetic
analysis of plant RSHs and reported that they originated
from an endosymbiotic cyanobacterium (Givens et al.
2004). However, several other groups reported different
J Plant Res (2017) 130:625634 627
models indicating that plant RSHs were introduced into
plant cells by lateral gene transfer from pathogenic bacteria
and/or the Deinococcus-Thermus bacterial phylum (Atkin-
son et al. 2011; Masuda 2012; Suzuki and Miyagishima
2010; van der Biezen etal. 2000). It is also unknown how
the three plant RSH families diverged and were established
during plant evolution.
To clarify the origin and evolution of RSHs involved in
the land plant stringent response, we compared the primary
structures of RSHs conserved in five plant phyla. We then
performed a phylogenetic analysis of the plant RSHs iden-
tified. Our results suggest that plant RSHs originated from
the Deinococcus-Thermus phylum and an unidentified bac-
terial phylum by independent lateral gene transfer events.
We also found that genes encoding GppA/Ppx homologs
are conserved in plant cells, which are in the same branch
with those of several bacterial species. These results indi-
cate that the plant stringent response was established with
enzymes from different origins.
Materials andmethods
To identify plant orthologs of RSHs, BLAST searches
were performed with amino acid sequences of Arabidop-
sis RSH1, RSH2 and/or CRSH against nucleotide and
amino acid sequence databases of the National Center for
Biotechnology Information (NCBI), Phytozome database
(https://phytozome.jgi.doe.gov/pz/portal.html), Cyan-
ophora Genome Project (Price et al. 2012), and Klebsor-
midium flaccidum genome project (Hori etal. 2014). Iden-
tified plant RSHs were then separately applied for BLAST
analysis against the Arabidopsis database in NCBI, and if
the top hit successfully identified the original Arabidop-
sis sequence (RSH1, RSH2 or CRSH), it was taken as an
ortholog. Bacterial Rel proteins were searched via BLAST
analysis using the amino acid sequence of E. coli SpoT. For
phylogenetic analysis of plant RSHs, amino-acid sequences
of bacterial Rel were selected based on datasets used in
previous phylogenetic studies (Givens et al. 2004; Mas-
uda 2012; Suzuki and Miyagishima 2010; van der Biezen
et al. 2000), which include those from Cyanobacteria,
Deinococcus-Thermus, Proteobacteria, Firmicutes, Ther-
momicrobium, and Chlamydiae. Accession numbers for the
sequences and identifiers were as described in TableS1.
To identify plant orthologs of GppA/Ppx, BLAST
searches were performed with the amino acid sequence of
E. coli GppA against the same nucleotide and amino acid
sequence databases of the National Center for Biotechnol-
ogy Information (NCBI), Phytozome database (https://phy-
tozome.jgi.doe.gov/pz/portal.html), Cyanophora Genome
Project (Price et al. 2012), and Klebsormidium flaccidum
genome project (Hori et al. 2014). Identified plant GppA/
Ppx were separately applied for BLAST analysis against
the E. coli database in NCBI, and if the top hit success-
fully identified the original E. coli GppA or Ppx sequence,
it was taken as an ortholog. For constructing phylogenetic
trees, bacterial GppA/Ppx were searched via BLAST analy-
sis on the NCBI website using the amino acid sequence of
the identified Arabidopsis GppA/Ppx homolog as a query
sequence. Obtained sequences were cutoff by an E value
of 11024, and then GppA/Ppx homologs were selected
from all bacterial phyla found in the list. Accession num-
bers for the sequences and identifiers were as described in
TableS2.
Domain structures of the identified proteins were ana-
lyzed using InterProScan (Mitchell et al. 2014). Based on
multiple protein sequence alignments, constructed below,
critical amino acid residues for enzymatic activities (e.g.,
ppGpp synthesis and hydrolysis) were analyzed for each
protein sequence based on previous studies (Aravind and
Koonin 1998; Hogg et al. 2004; Wendrich and Marahiel
1997).
Multiple amino-acid sequence alignments of RSH and
GppA/Ppx homologs were constructed using MAFFT ver.
7 (Katoh and Standley 2013), followed by manual adjust-
ment based on structural considerations. The region used
to construct the RSH phylogenetic trees corresponds to
residues 42342 of E. coli SpoT. The region used to con-
struct the GppA/Ppx phylogenetic tree corresponds to resi-
dues 18323 of Arabidopsis GppA/Ppx homolog. After
constructing the original alignments, we excluded gap-rich
regions from each alignment by a home-made Perl script,
which removes sites where gaps were found in over 95% of
the sequences. All alignments used for phylogenetic analy-
sis are shown in Supplementary Dataset. Substitution mod-
els were tested for each of the alignments with ProtTest ver.
3.4.2 (Darriba etal. 2011) under consideration of empirical
amino acid frequencies (+F) and a discrete gamma distri-
bution (+G). The LG substitution model (Le and Gascuel
2008) with +G was chosen as the best-fit model based
on Bayesian information criterion (BIC) for all the dataset.
Bayesian inference was performed using MrBayes 3.2.3
(Ronquist et al. 2012) under the LG+G model. Two par-
allel runs of 1107 generations with twelve chains were
carried out with tree sampling every 100 generations. The
first 2.5106 generations were discarded as burn-in. The
run parameter convergence was assessed by average stand-
ard deviation of split frequencies (<0.01) and by monitor-
ing the trace plots and effective sample size (>200) using
Tracer v.1.6 (Rambaut etal. 2013). The trees were visual-
ized by FigTree v1.4.2 (http://tree.bio.ed.ac.uk/software/
figtree/). Phylogenetic trees were also constructed by the
maximum likelihood (ML) method using RAxML ver. 8
(Stamatakis 2014) under the LG+G model. Bootstrap (BP)
values were obtained by 1,000 replications.
13

628
To assess the possibility for the sister-relationship
between the cyanobacterial Rel and RSH1, RSH2 or CRSH,
the approximately unbiased (AU) test was carried out by
using each sequence data shown in Supplementary Data-
set. For each of the RSH2 and CRSH data, the approximate
likelihoods were calculated using the ProtML program in
MOLPHY 2.3b (Adachi and Hesegawa 1996) or all the
945 possible trees under the constraints of RSH2/CRSH,
Cyanobacteria, Proteobacteria, Firmicutes+Chlamydiae,
Thermomicrobium, Dictyoglomus, and Deinococcus-Ther-
mus according to the results in Figs. S1, S2, respectively.
For RSH1, the best 3,200 trees with the highest approxi-
mate likelihood were selected for this evaluation from the
10,395 combinations because Meiothermus Rel and other
Deinococcus-Thermus clade were not constrained (see Fig.
S3). Per-site log-likelihoods were calculated with RAxML
under the LG+G model and were subsequently used for
the AU test with CONSEL (Shimodaira and Hasegawa
2001).
Chloroplast transit peptide prediction was performed
with the ChloroP 1.1 server (http://www.cbs.dtu.dk/
services/ChloroP/).
Results
We searched for RSH family proteins in genomic sequences
from five plant phyla (Embryophyta, Charophyta, Chloro-
phyta, Rhodophyta, Glaucophyta) using Arabidopsis RSH1,
RSH2 and CRSH as query sequences. To find evolutional
trails for RSHs in land plants, we mainly focused on the
green plant linage, but not other lineages including those of
secondary symbionts. Figure1 shows the phylogenetic tree
based on amino acid sequence alignment of the (p)ppGpp
hydrolase and synthase domains of bacterial Rel and plant
RSHs. Strikingly, plant RSHs are classified into three indi-
vidual clusters, which are separated from each other. The
classified families were termed as RSH1, RSH2 and CRSH
families, since Arabidopsis RSH1, RSH2 and CRSH were
grouped in the clusters, respectively. RSH1 homologs were
found in species from three plant phyla, other than Embry-
ophyta, including K. flaccidum (Charophyta), Ostreococcus
tauri (Chlorophyta) and Cyanidioschyzon merolae (Rhodo-
phyta). No gene encoding an RSH1-like protein was found
in the genome sequence of Cyanophora paradoxa (Glauco-
phyta). RSH2-like proteins were found in four plant-phyla
Embryophyta, Charophyta, Chlorophyta and Glaucophyta,
but not found in Rhodophyta (Fig.1). CRSH-like proteins
were found in all plant phyla except Glaucophyta (C. para-
doxa; Fig. 1). Deinococcus-Thermus Rel is grouped out
from other bacterial Rel, and form an individual cluster
with plant RSH1- and CRSH-family proteins (Fig.1). We
also constructed phylogenetic trees based on each plant
13
J Plant Res (2017) 130:625634
RSH family with bacterial SpoT. We found that RSH1 fam-
ily is in the same clade comprising Deinococcus-Thermus
Rel (Fig. S3), but RSH2 family is not clearly resolved from
the clade comprising bacterial Rel (Fig. S1). CRSH family
is clearly in the same clade comprising Deinococcus-Ther-
mus Rel with a BIPP value of 1.00 (Fig. S2).
We carried out the AU test to assess the possibility of
the cyanobacterial origin for each of the RSH1, RSH2 and
CRSH families by using CONSEL (for detail, see Materi-
als and methods section). The sister relationships of the
cyanobacterial Rel with RSH1 or CRSH were significantly
rejected by the AU test (p=0.00005 and p=0.012, respec-
tively). On the other hand, the sister-relationship between
RSH2 and cyanobacterial Rel was not significantly rejected
by the AU test (p=0.144).
Amino-acid sequences of RSHs from each phylum were
then applied for analysis by InterProScan (Mitchell et al.
2014) to identify domain structures. As shown in Fig. 2a,
InterProScan revealed that all RSH1 sequences had simi-
larity to (p)ppGpp hydrolase and synthase domains. How-
ever, the Gly residue that is critical for (p)ppGpp synthase
activity in B. subtilis Rel was not conserved in all RSH1
homologs (Fig. 3a), suggesting that they do not have (p)
ppGpp synthase activity. All RSH1-like proteins had the
TGS domain found in bacterial Rel (e.g., Synechocystis and
Meiothermus Rel) (Fig. 2a). In addition, the ACT domain
was conserved among RSH1-like proteins from all plant
phyla except Embryophyta (Arabidopsis and Physcom-
itrella) (Fig. 2a) . The lack of ACT domain in Embryo-
phyta RSH1 may not be due to alternative splicing of the
gene since the primary structures were revealed from their
genomic sequences, but not cDNA sequences.
Recently, the complex structure of the ribosome and E.
coli RelA was solved by cryo-electron microscopy (Brown
et al. 2016; Loveland et al. 2016). In the structure, the
TGS domain of RelA binds the CCA tail to orient the free
3 hydroxyl group of the terminal adenosine, which leads
steric preclusion of an aminoacylated tRNA at this posi-
tion (Brown etal. 2016), indicating the importance of the
interaction for sensing ribosome activity. More specifically,
Pro411, Lys412, His432 and Arg438 of E. coli RelA inter-
act with the 3 CCA tail of the A-site tRNA (Fig. S4, red
boxes), and Arg487, Arg489, Lys491, His493, Arg497,
and Lys498 interact with the tRNA phosphate backbone
(Fig. S4, blue boxes). Among the amino-acid residues,
positively-charged amino acids (e.g., Lys, Arg, and His)
were particularly important for the interaction (Brown etal.
2016). The amino-acid sequence alignment of the TGS
domain of bacterial Rel and plant RSH1 reveals that at least
half of the positively-charged amino-acids were conserved
in both bacterial Rel, and plant RSH1 (Fig. S4), suggest-
ing that RSH1 homologs interact with the ribosomes in
chloroplasts.
J Plant Res (2017) 130:625634 629

Fig.1Phylogenetic tree based on (p)ppGpp synthase and hydrolase tutions per site, as indicated by the scale bar. Gain and loss of the
domains of plant RSHs and bacterial Rel. The trees were constructed critical Gly residue, aspartate kinase chorismate mutase TyrA domain
based on Bayesian inference and ML using the LG+G model in (ACT), Ca2+-binding helix E and F motif (EF hand), and tetratrico-
MrBayes and RAxML, respectively. Support values are Bayesian peptide repeat motif (TPR) are expressed by solid and dotted arrows,
posterior probabilities (left) and bootstrap values of ML (right), and respectively
branch lengths are proportional to the expected number of substi-

Primary structures of RSH2-like proteins found in
four plant-phyla Embryophyta, Charophyta, Chlorophyta
and Glaucophyta were characterized (Fig. 2b). One of
the two RSH2-like proteins in O. tauri (RSH2b) lacks
Gly residue, although the sequences are general similar
to other RSH2-like proteins in these regions. The TGS
domain is only found in RSH2-like proteins of C. para-
doxa (Glaucophyta). The lack of TGS domain in other
phyla is not due to alternative splicing of the gene since
we obtained primary structures of RSH2 homologs form

13

630 J Plant Res (2017) 130:625634

Fig.2Schematic depiction
of the primary structures of
RSH1 (a), RSH2 (b) and CRSH
(c) family proteins. Hyd: (p)
ppGpp hydrolase domain; hyd:
(p)ppGpp hydrolase domain
lacking some critical amino-
acids; Syn: (p)ppGpp synthase
domain; syn: (p)ppGpp synthase
domain lacking the critical
Gly residue; ACT: aspartate
kinase chorismate mutase TyrA
domain; EF hand: C a2+-binding
helix E and F motif; TGS:
threonyl-tRNA synthetase,
GTPase, SpoT domain; TPR:
tetratricopeptide repeat motif;
AA: amino-acids

genomic sequences of K. flaccidum, Arabidopsis and
Physcomitrella.
CRSH-like proteins were found in all plant phyla except
Glaucophyta (C. paradoxa; Fig. 2c). These CRSH-like
proteins lack certain critical residues in the HD domain
(Fig. 3b), suggesting that they lack (p)ppGpp hydro-
lase activity. The CRSH-like proteins from Chlorophyta
(O. tauri and Volvox carteri) lack the conserved Gly
residue in the (p)ppGpp synthase domain (Fig. 3a), sug-
gesting that they have lost (p)ppGpp synthase activity.
The Ca2+-binding EF hand motifs have been conserved
in CRSH-like proteins from most plant phyla, except
Rhodophyta (C. merolae; Fig. 2c). CRSH-like proteins
from O. tauri and K. flaccidum have C-terminal tetratrico-
peptide repeat motifs (Fig.2c), which mediate proteinpro-
tein interactions.
A computer-aided similarity search showed that
GppA/Ppx homologs have been conserved in the plant
phyla Embryophyta (e.g. Arabidopsis and Populus) and
Charophyta Klebsormidium; these contain both a GppA/
Ppx phosphatase domain and an HD domain or an HD-
like domain, which resembles to the bacterial HD-domain
polyPases, including E. coli GppA and Ppx (Fig. 4).

13
J Plant Res (2017) 130:625634 631

Fig.3Alignment of partial a Arabidopsis RSH2 419 GI - - - - - - -

- S Y HV L CGRHK S L Y S I Y S K
amino acid sequences for the Arabidopsis RSH3 423 GI - - - - - - -
- S Y HV V S GRHK S L Y S I Y CK
(p)ppGpp synthase (a) and Physcomitrella RSH2a 256 GV - - - - - - -
- Q F VD L S GRP K N L Y S I Y K K RSH2 family
Klebsormidium RSH2 517 GI - - - - - - -
- GY L D L S GRP K N L Y S I HS K
hydrolase (b) domains. The Gly Ostreococcus RSH2a 278 NA - - - - - - -
- QV VD L CGRQK S V Y S V Y K K
residue that is critical for syn- Cyanidioschyzon RSH 642 PQL - HS Y I -
- WSWR I RGRV K S L Y S T Y RK
thase activity and the His-Asp ** * ** *
Arabidopsis CRSH 296 L V L - A EMV - - NDV Y I K GRY K S RY SMMK K
(HD) critical for hydrolysis are Physcomitrella CRSH 239 P EM - QS L I - - E SV K I S GRCK S K Y S TMK K
indicated in red Klebsormidium CRSH 552 L L I T QP Y VPGMEV T L S GR L K S L Y S AHCK CRSH family
Ostreococcus CRSHa 157 P TM - NA L I GR GGV RV L ARRK S RY S TMK K
Cyanidioschyzon CRSH 305 P E L - RR - - - - YRY E I T GRA K N L F S T F K K
* * * *
Synechocystis Rel 235 G I - - - - - - - - E H F E L QGRP K H L Y G I Y Y K
Bacterial Rel
Meiothermus Rel 245 S V L - GA S I - - QGY E V T GR T K H L Y S I WK K
* * *
Arabidopsis RSH1 363 QF L - D L V T - - V N T DV RS V CK E T Y S I Y K A
Physcomitrella RSH1 327 QF L - GYMT - - A D I K V A T L S K E L Y S I Y K K
Klebsormidium RSH1 356 A C L - - C L T - - HS F RV E T L CK E A Y S I Y K K RSH1 family
Ostreococcus RSH1 304 D F L - RGHA - - V A V S L GA K QK E P Y GV HRS
Cyanidioschyzon RSH1 413 S I L - K P L V - - RE I C T K V A L K N L F S I Y QR
*
b Arabidopsis RSH1 176 P V A V A R I L GE L E L DWE S I V A GL L HD
Physcomitrella RSH1 135 P V E V A R I L GE L EMDWE T V V A GL L HD
Klebsormidium RSH1 169 P V A V A E I L GQMQMDCE T I I A GL L HD RSH1 family
Ostreococcus RSH1 104 P V A V A A I L A D L E L DHE S L I A GL L HD
Cyanidioschyzon RSH1 139 P V A V A L L L A E L RMD VD T L I A GL L HD
** ** * * * * ** *
Synechocystis Rel 56 P V A V A GL L RD L GGDE AM I A A GF L HD
Meiothermus Rel 45 P V A V A E I L A E L GL DA E T L MA GL L HD Bacterial Rel
** ** * * * * ***
Arabidopsis RSH2 237 CV E T AML L A N I GA NS T V V V A GL L HD
Arabidopsis RSH3 241 CV E T AML L A D I GA NS T V V V A G I L HD
Physcomitrella RSH2 a 44 CV E T A L I L A A T GV GN T V V A A GL L HD RSH2 family
Klebsormidium RSH2 334 C L E A A K I L A E S T RDRVMV A A GL L HD
Ostreococcus RSH2a 99 CV E T A V T L A K L GMD T K T V A V GL L HD
* * * **
Arabidopsis CRSH 116 A L S L S I I L AD L QMDA E V I S A S I L S E
Physcomitrella CRSH 59 A L S V A F I L AD L RMDA E V I A A GL L RE CRSH family
Klebsormidium CRSH 373 - - - - - - - VWGV K A E L E D L C F A V L Q -
Cyanidioschyzon CRSH 123 S V RNA A F L AR L GV DA D T I CA A L V - D

Fig.4Schematic representa-
tion of the primary structures of
GppA/Ppx homologs. Domain
structures were analyzed using
InterProScan. AA amino-acids

When we performed BLAST search, with Arabidopsis Discussion

GppA/Ppx homolog used as a query sequence, to find
genes showing similarity more than a 1104 E value,
only three bacterial GppA/Ppx families were found in
addition to plant GppA/Ppx, which belong to HD-domain
polyPases, GPPases and GPPase-like polyPases. By the
analysis, high molecular-weight polyPases, low molecu-
lar-weight polyPases, and RTG2 proteins including fun-
gal, protistan and/or metazoan GppA homologs were
not found. Phylogenetic analysis revealed that the plant
GppA/Ppx homologs are in the sister clade comprising
HD-domain polyPases from bacterial phyla Spirochetes,
Actinobacteria and Chlorobi (Fig.5).
It has been shown that Arabidopsis RSH1 does not func-
tionally complement E. coli relA or the relA-spoT double
mutation, although Arabidopsis RSH2, RSH3 and CRSH
do complement these mutations (Mizusawa et al. 2008).
These results indicate that Arabidopsis RSH1 does not have
(p)ppGpp synthase activity. Here, we confirmed that Arabi-
dopsis RSH1 does not have the critical Gly residue in the
(p)ppGpp synthase domain (Fig. 3a). Indeed, RSH1-like
proteins were found in Embryophyta, Charophyta, Chloro-
phyta and Rhodophyta (Figs.1, 2a), none of which have the
Gly residue (Fig.3a), suggesting that they had lost the (p)
ppGpp synthase activity. The RSH1 family proteins from

13

632 J Plant Res (2017) 130:625634

13
 J Plant Res (2017) 130:625634 633
Fig.5Phylogenetic tree based on GppA/Ppx domains of GppA/Ppx
homologs. The trees were constructed based on Bayesian inference
and ML using the LG+G model in MrBayes and RAxML, respec-
tively. Support values are Bayesian posterior probabilities (left), boot-
strap values of ML (right), and branch lengths are proportional to the
expected number of substitutions per site, as indicated by the scale
bar. Organisms whose GppA/Ppx primary structures are shown in
Fig.4 are highlighted by red squares
the plant phyla have retained the conserved TGS domain,
like bacterial Rel (Fig. 2a), which may associate with the
ribosome in chloroplasts (Fig. S4). Furthermore, the ACT
domain has been conserved in the RSH1 homologs from
Charophyta, Chlorophyta and Rhodophyta, but not in
Embryophyta. These results indicate that RSH1 is structur-
ally similar to bacterial Rel proteins. Phylogenetic analysis
indicates that RSH1 proteins are clustered with Deinococ-
cus-Thermus Rel proteins (Fig.1, S3). It has been reported
that nuclear-encoded plastidial proteins are derived from
the endosymbiotic cyanobacterium as well as other bacte-
rial species (Suzuki and Miyagishima 2010). These results
suggest that the plant RSH1-like protein was introduced
into a proto-plant cell by lateral gene transfer at least before
the separation of Chlorophyta and Rhodophyta, followed
by loss of the functional ACT domain after separation of
Embryophyta and Charophyta (Fig.1).
CRSH-like proteins were found in Embryophyta, Cha-
rophyta, Chlorophyta and Rhodophyta (Figs. 1, 2c). Phy-
logenetic analysis clearly showed that CRSH and RSH1
family proteins are in the same cluster but different line-
ages; CRSH was separated from RSH1 and Deinococcus-
Thermus Rel with high BIPP and BP values of 0.83 and 63,
respectively (Fig. 1). This suggests that CRSH and RSH1
were introduced into a proto-plant cell by lateral gene trans-
fer from Deinococcus-Thermus prior to the separation of
Chlorophyta and Rhodophyta, and the introduced gene(s)
was functionally and genetically divided at an early step
after gene transfer. Arabidopsis CRSH does not have (p)
ppGpp hydrolase activity owing to loss of the HD domain,
and this property is conserved in all CRSHs from different
plant phyla (Fig.2c). Thus, CRSH and RSH1 have adopted
separate (p)ppGpp synthase and hydrolase roles, respec-
tively, during the evolution of plant cells.
Complementation analysis of the E. coli relA mutant
and relA-spoT double mutant suggests that Arabidopsis
RSH2 and RSH3 have both (p)ppGpp synthase and hydro-
lase activities (Mizusawa et al. 2008). RSH2-like proteins
were found in four plant phyla including Embryophyta,
Charophyta, Chlorophyta and Glaucophyta, and most have
the conserved (p)ppGpp synthase and hydrolase domains
(Fig.2b). Notably, two RSH2-like proteins in C. paradoxa
have the TGS domain, like bacterial Rel (Fig. 2b), sug-
gesting that RSH2 was introduced into an ancestral Glau-
cophyta by lateral gene transfer. The phylogenetic analyses
clearly show that the origin of RSH2 differs from that of
RSH1 and CRSH (Fig. 1). Specifically, the sister rela-
tionships of the cyanobacterial Rel with RSH1 or CRSH
were significantly rejected by the AU test (p=0.00005
and p=0.012, respectively); on the other hand, the sister-
relationship between RSH2 and cyanobacterial Rel was
not significantly rejected by the AU test (p=0.144) (see
Results section). From these observations, we propose
that plant RSHs were introduced into proto-plant cells by at
least two different horizontal gene transfer events: one from
Deinococcus-Thermus (for RSH1 and CRSH families) and
the other from an undefined bacterial species (for RSH2
family). Although the RSH2 origin could not be resolved
by this phylogenetic analysis because of low BIPP or BP
values between RSH2 and bacterial Rel, it is still possible
that RSH2 was derived from an endosymbiotic cyanobac-
terium. If this is the case, one could assume that the RSH
homologs (RSH1, CRSH and RSH2 families) were intro-
duced into a proto-plant cell by the endosymbiosis of the
ancestral cyanobacterium and the RSH1 and CRSH gene
families were then selectively lost in the cyanobacterial
linage. A close evolutionary relationship of Cyanobacteria
and Deinococcus-Thermus phyla was reported (Gupta and
Johari 1998), supporting the idea. In any case, plant RSH
families were sub-functionalized during plant evolution;
e.g., RSH1 may have been arranged to associate with ribo-
some through the TGS domain, but others have not (Fig.2).
This is similar to the bacterial (p)ppGpp synthases/hydro-
lases; RelA and SpoT were sub-functionalized as the major
(p)ppGpp synthesis and the hydrolysis, respectively, in the
- and -proteobacterial linage (Potrykus and Cashel 2008).
We found that GppA/Ppx-like proteins, having both
GppA/Ppx and HD (or HD-like) domains, are conserved
in Embryophyta and Charophyta, and the protein family
is in the clade comprising bacterial HD-domain polyPases
and the GPPases/GPPase-like polyPases (Figs. 4, 5). This
suggests that GppA/Ppx-like protein was introduced into a
proto-plant cell by lateral gene transfer, although the donor
bacterial species could not be defined by the phylogenetic
analysis. The HD-domain (or HD-like domain) polyPases
are structurally similar to the GPPases and GPPase-like
polyPases including E. coli GppA and Ppx that convert
pppGpp to ppGpp to mediate the stringent response (Hara
and Sy 1983; Keasling et al. 1993), suggesting that plant
GppA/Ppx homologs may have such an enzymatic activity.
It should be noted that Arabidopsis GppA/Ppx do not have
a predicted chloroplast transit peptide at their N terminus
(Fig. S5); no transit peptide is also predicted for the GppA/
Ppx homolog from K. flaccidum (Charophyta), although it
has a long N-terminal sequence (Fig.4). These results sug-
gest that plant GppA/Ppx homologs localize in the cytosol.
We previously showed that over-accumulation of ppGpp in
the cytosol results in reduced plant growth, indicating that
13

634
ppGpp is utilized in the cytosol (Ihara and Masuda 2016).
If plant GppA/Ppx-like proteins could convert pppGpp to
ppGpp, plant-type GppA/Ppx-like proteins may control
the ratio of pppGpp and ppGpp in the cytosol during the
stringent response in plant cells. Genetic and biochemical
analyses of plant GppA/Ppx-like proteins will help clarify
the exact function of (p)ppGpp in the cytosol, and such a
study is currently under way in our laboratory.
Acknowledgements This work was supported in part by JSPS
KAKENHI Grant Nos. 16H03280 and 16K14694 to SM.
References
Adachi J, Hesegawa M (1996) MOLPHY version 2.3: programs for
molecular phylogenetics based on maximum likelihood. Comput
Sci Monogr 28:1150
Albi T, Serrano A (2014) Two exopolyphosphatases with distinct
molecular architectures and substrate specificities from the
thermophilic green-sulfur bacterium Chlorobium tepidum TLS.
Microbiology 160:20672078
Aravind L, Koonin EV (1998) The HD domain defines a new super-
family of metal-dependent phosphohydrolases. Trends Biochem
Sci 23:469472
Atkinson GC, Tenson T, Hauryliuk V (2011) The RelA/SpoT
homolog (RSH) superfamily: distribution and functional evolu-
tion of ppGpp synthetases and hydrolases across the tree of life.
PLoS One 6:e23479
Brown A, Fernandez IS, Gordiyenko Y, Ramakrishnan V (2016)
Ribosome-dependent activation of stringent control. Nature
534:277280
Cashel M, Gentry DR, Hernandez VJ, Vinella D (1996) The stringent
response. In: Neidhardt FC, Curtiss IR, Ingraham JL etal (eds)
Escherichia coli and Salmonella: cellular and molecular biology,
2ndedn. ASM Press, Washington DC, pp14581496
Choi MY, Wang Y, Wong LLY et al (2012) The two PPX-GppA
homologues from Mycobacterium tuberculosis have distinct bio-
chemical activities. PLoS One 7:e42561
Dalebroux ZD, Swanson MS (2012) ppGpp: magic beyond RNA pol-
ymerase. Nat Rev Microbiol 10:203212
Darriba D, Taboada GL, Doallo R, Posada D (2011) ProtTest 3: fast
selection of best-fit models of protein evolution. Bioinformatics
27:11641165
Givens RM, Lin M-H, Taylor DJ et al (2004) Inducible expression,
enzymatic activity, and origin of higher plant homologues of
bacterial RelA/SpoT stress proteins in Nicotiana tabacum. J Biol
Chem 279:74957504
Gupta RS, Johari V (1998) Signature sequences in diverse proteins
provide evidence of a close evolutionary relationship between
the Deinococcus-Thermus group and cyanobacteria. J Mol Evol
46:716720
Hara A, Sy J (1983) Guanosine 5-triphosphate, 3-diphosphate
5-phosphohydrolase. Purification and substrate specificity. J
Biol Chem 258:16781683
Hauryliuk V, Atkinson GC, Murakami KS etal (2015) Recent func-
tional insights into the role of (p)ppGpp in bacterial physiology.
Nat Rev Microbiol 13:298309
Hogg T, Mechold U, Malke H et al (2004) Conformational antago-
nism between opposing active sites in a bifunctional RelA/SpoT
homolog modulates (p)ppGpp metabolism during the stringent
response. Cell 117:5768
Hori K, Maruyama F, Fujisawa T etal (2014) Klebsormidium flaccidum
genome reveals primary factors for plant terrestrial adaptation. Nat
Commun 5:3978
13
J Plant Res (2017) 130:625634
Ihara Y, Masuda S (2016) Cytosolic ppGpp accumulation induces
retarded plant growth and development. Plant Signal Behav
11:e1132966
Katoh K, Standley DM (2013) MAFFT multiple sequence alignment
software version 7: improvements in performance and usability.
Mol Biol Evol 30:772780
Keasling JD, Bertsch L, Kornberg A (1993) Guanosine pentaphosphate
phosphohydrolase of Escherichia coli is a long-chain exopolyphos-
phatase. Proc Natl Acad Sci USA 90:70297033
Le SQ, Gascuel O (2008) An improved general amino acid replacement
matrix. Mol Biol Evol 25:13071320
Loveland AB, Bah E, Madireddy R, etal (2016) Ribosome-RelA struc-
tures reveal the mechanism of stringent response activation. Elife
5:e17029
Maekawa M, Honoki R, Ihara Y, et al (2015) Impact of the plastidial
stringent response in plant growth and stress responses. Nat Plants
1:15167
Masuda S (2012) The stringent response in phototrophs. In: Najafpour
M (ed) Advances in photosynthesis. In Tech, Rijeka, pp487500
Masuda S, Mizusawa K, Narisawa T etal (2008) The bacterial stringent
response, conserved in chloroplasts, controls plant fertilization.
Plant Cell Physiol 49:135141
Mechold U, Potrykus K, Murphy H etal (2013) Differential regulation
by ppGpp versus pppGpp in Escherichia coli. Nucleic Acids Res
41:61756189
Mitchell A, Chang H-Y, Daugherty L etal (2014) The InterPro protein
families database: the classification resource after 15years. Nucleic
Acids Res 43:D213D221
Mittenhuber G (2001) Comparative genomics and evolution of genes
encoding bacterial (p)ppGpp synthetases/hydrolases (the Rel, RelA
and SpoT proteins). J Mol Microbiol Biotechnol 3:585600
Mizusawa K, Masuda S, Ohta H (2008) Expression profiling of four
RelA/SpoT-like proteins, homologues of bacterial stringent factors,
in Arabidopsis thaliana. Planta 228:553562
Potrykus K, Cashel M (2008) (p)ppGpp: still magical? Annu Rev
Microbiol 62:3551
Price DC, Chan CX, Yoon HS et al (2012) Cyanophora paradoxa
genome elucidates origin of photosynthesis in algae and plants.
Science 335:843847
Rambaut A, Suchard M, Drummond A (2013) Tracer v1.6. http://tree.
bio.ed.ac.uk/software/tracer/. Accessed 14 Mar 2017
Ronquist F, Teslenko M, van der Mark P, etal (2012) MrBayes 3.2: effi-
cient Bayesian phylogenetic inference and model choice across a
large model space. Syst Biol 61:539542
Shimodaira H, Hasegawa M (2001) CONSEL: for assessing the confi-
dence of phylogenetic tree selection. Bioinformatics 17:12461247
Stamatakis A (2014) RAxML version 8: A tool for phylogenetic
analysis and post-analysis of large phylogenies. Bioinformatics
30:13121313
Suzuki K, Miyagishima S (2010) Eukaryotic and eubacterial contribu-
tions to the establishment of plastid proteome estimated by large-
scale phylogenetic analyses. Mol Biol Evol 27:581590
Tozawa Y, Nomura Y (2011) Signalling by the global regulatory mol-
ecule ppGpp in bacteria and chloroplasts of land plants. Plant Biol
(Stuttg) 13:699709
Tozawa Y, Nozawa A, Kanno T et al (2007) Calcium-activated (p)
ppGpp synthetase in chloroplasts of land plants. J Biol Chem
282:3553635545
van der Biezen EA, Sun J, Coleman MJ, etal (2000) Arabidopsis RelA/
SpoT homologs implicate (p)ppGpp in plant signaling. Proc Natl
Acad Sci 97:37473752
Wendrich TM, Marahiel MA (1997) Cloning and characterization of
a relA/spoT homologue from Bacillus subtilis. Mol Microbiol
26:6579