A point mutation in the rice alpha‐tubulin gene OsTUBA3 causes grain notching

Abstract

Grain notching is a common deformation that decreases rice (Oryza sativa) quality; however, the underlying molecular basis causing grain notching remains unclear.
We report mechanisms underlying grain notching in Small and notched grain (Sng) mutants, which contained an arginine to histidine substitution at amino acid position 422 (R422H) of the α‐tubulin protein OsTUBA3. The R422H mutation decreased cell length and increased cell width/height of glumes and caryopses, but led to elongated caryopses compressed within shortened glumes, thus giving rise to notched and small grains. Glume and caryopsis cells had different dimensional orientations relative to the directions of organ elongation. Thus, the abnormal cell expansion induced in glumes and caryopses by the R422H mutation had different effects on elongation of these organs.
The R422H mutation in OsTUBA3 compromised β‐tubulin binding and led to formation of defective heterodimers. This in turn affected tubulin incorporation and microtubule (MT) nucleation and regrowth, consequently leading to MT instability and reducing the transverse orientation. The defective MT dynamics affected cell expansion and shape, causing different alterations in glume and caryopsis dimensions and resulting in grain notching.
These data indicate that Arg422 in OsTUBA3 is crucial for MT dynamics and that substitution with His causes grain notching, reducing grain quality and yield. These findings offer valuable insights into the molecular regulation underlying grain development in rice.

Full text

A point mutation in the rice alpha-tubulin gene OsTUBA3 causes
grain notching
Chenshan Xu
1,2
, Bingtang Chen
1
, Shanjin Huang
3
, Zhuyun Deng
1
and Tai Wang
1,4
1
Key Laboratory of Plant Molecular Physiology, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China;
2
Dezhou University, Dezhou, Shandong 253023, China;
3
Center for Plant Biology, School of Life Sciences, Tsinghua University, Beijing 100084, China;
4
College of Life Science, University of Chinese Academy of Sciences, Beijing 100093, China
Authors for correspondence:
Zhuyun Deng
Email:dengzhuyun@ibcas.ac.cn
Tai Wang
Email:twang@ibcas.ac.cn
Received: 20 April 2023
Accepted: 28 July 2023
New Phytologist (2023)
doi: 10.1111/nph.19226
Key words: cell expansion, grain size,
microtubule, notched grains, rice yield and
quality, a-tubulin.
Summary
Grain notching is a common deformation that decreases rice (Oryza sativa) quality; how-
ever, the underlying molecular basis causing grain notching remains unclear.
We report mechanisms underlying grain notching in Small and notched grain (Sng)
mutants, which contained an arginine to histidine substitution at amino acid position 422
(R422H) of the a-tubulin protein OsTUBA3. The R422H mutation decreased cell length and
increased cell width/height of glumes and caryopses, but led to elongated caryopses com-
pressed within shortened glumes, thus giving rise to notched and small grains. Glume and car-
yopsis cells had different dimensional orientations relative to the directions of organ
elongation. Thus, the abnormal cell expansion induced in glumes and caryopses by the
R422H mutation had different effects on elongation of these organs.
The R422H mutation in OsTUBA3 compromised b-tubulin binding and led to formation of
defective heterodimers. This in turn affected tubulin incorporation and microtubule (MT)
nucleation and regrowth, consequently leading to MT instability and reducing the transverse
orientation. The defective MT dynamics affected cell expansion and shape, causing different
alterations in glume and caryopsis dimensions and resulting in grain notching.
These data indicate that Arg422 in OsTUBA3 is crucial for MT dynamics and that substitu-
tion with His causes grain notching, reducing grain quality and yield. These findings offer valu-
able insights into the molecular regulation underlying grain development in rice.
Introduction
A mature rice grain consists primarily of an inner caryopsis and
the outer pair of interlocked lemma and palea collectively called
glumes, which develop asynchronously. Glumes differentiate
during inflorescence development and expand to their largest
dimensions before flowering, whereas caryopses initiate growth
within the fully expanded glumes after fertilization, achieving
their maximum size at c. 15 days after flowering (DAF). Typi-
cally, the mature caryopsis fits perfectly into the space enclosed
by the glumes. Glume shape and size are widely considered to
decide the final dimensions of the caryopsis and therefore have
great effects on the yield and appearance quality of rice. Numer-
ous genes involved in various molecular pathways have been
shown to play important roles in mediating glume cell size and
cell number, thus determining grain shape and size (Li et al.,
2018; Fan & Li, 2019).
Mismatches between glume and caryopsis shape and size have
frequently been observed in rice, but this phenomenon has been
understudied due to the complexity of these traits. Such imbal-
ances in growth often lead to notched grains, which are character-
ized by a V-shaped interstice of variable length in the ventral side
of a mature caryopsis (Takeda, 1982). In a previous study, 5.6%
of the 1366 rice varieties studied had a notched grain
percentage >90% (Xiong et al., 1986). Heavier grains are more
likely to be notched; in the same study, 15.6% of varieties with a
thousand-grain weight higher than 35 g had >90% notched
grains. Grain notching causes substantial reductions in grain
quality; it is associated with increased chalkiness and decreased
head rice yield and milled rice recovery (Xiong et al., 1986; Lin
et al., 2014). For example, Xiong et al.(1986) found that 60% of
Yunan germplasm with a high notched grain percentage (>90%)
had chalky grains, whereas only 15% of Yunan germplasm with-
out notched grains did. Grain notching usually leads to white
belly between the notch and the embryo, because notching may
cause the lower half of the endosperm to be more affected by the
embryo. This inhibits accumulation of starch and other storage
components and results in loose packing of the starch granules
(Tao et al., 2022).
Grain notching is a complex trait that is influenced by both
genetic and environmental factors. Conditions that cause exces-
sive caryopsis expansion or inhibition of glume elongation pro-
mote notched grain formation. Such conditions include excessive
application of nitrogen, phosphorus, or potassium fertilizer; low-
temperature stress during grain filling; and light shading before
heading (Goto & Kumagai, 2009; Zhan et al., 2020). Multiple
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
New Phytologist (2023) 1
www.newphytologist.com
Research

genes are known to control grain notching. Using multiple map-
ping populations, quantitative trait locus (QTL) analyses have
identified 12 QTLs for grain notching, distributed across all rice
chromosomes except chromosomes 4 and 9 (Pavithran, 1977;
Sobrizal, 2007;Liet al., 2012). A mutant analysis revealed that
NBG4, which is involved in brassinosteroid biosynthesis, modu-
lates formation of notched grains through differential regulation
of glume and caryopsis elongation (Tong et al., 2018). A muta-
tion in the microtubule (MT) depolymerase OsKinesin-13A
causes grain shrinkage and notching as a result of glume length
reduction (Deng et al., 2015). Few other genes related to grain
notching have been identified to date, and the cellular and mole-
cular mechanisms underlying this unfavorable trait remain largely
unknown. Furthermore, it is unclear how a single gene would dif-
ferentially affect elongation of the inner caryopsis and the outer
glumes of a grain.
In the present study, we analyzed grain notching in the Small
and notched grain (Sng) mutant. We discovered that grain notch-
ing in this mutant resulted from compression of overgrown car-
yopses within shortened glumes. These effects were caused by
SNG
R422H
-mediated abnormalities in the cortical MTs, which
led to alterations in cell dimensions. Specifically, glumes and car-
yopses had different orientation of cell elongation, with glume
and caryopsis cells expanding in orthogonal directions. Growth
imbalances between the glumes and the caryopses in Sng occurred
because cells elongated parallel to the glumes but perpendicular
to the caryopses.
Materials and Methods
Plant materials
The Sng mutant was derived from sodium azide-treated wild-type
(WT) cv Zhonghua 11 (Oryza sativa L. subsp. japonica cv Zhon-
ghua 11). Sng was backcrossed to cv Zhonghua 11 for three gen-
erations, yielding the Sng mutant used in further analyses.
Positional cloning was conducted in the F
2
population of a cross
between Sng and Oryza sativa subsp. indica cv Nanjing 6. All rice
plants were grown in paddy fields in Beijing or Hainan under
natural conditions.
SNG/SNG
R422H
overexpression and knockout
A seamless cloning and assembly kit (NEB) implementing the
Gibson assembly method was used to generate all of the constructs
described here. To generate the 35S::SNG and 35S::SNG
R422H
constructs, the full-length open reading frames (ORFs) SNG and
SNG
R422H
were amplified from cDNA. The amplified ORFs were
tagged with 8 9His, then fused with the NOS terminator, and
inserted into the pCAMBIA1301 plasmid at the BglII site. To
generate the SNG::SNG and SNG::SNG
R422H
constructs, the 2.4-
kb SNG promoter and the N-terminal 8 9His-tagged genomic
sequences of SNG and SNG
R422H
were linked with their 1.6-kb 30
untranslated region (UTR). These DNA fragments were fused
together and inserted into the pCAMBIA1391 plasmid at the
BstEII site. SNG and SNG
R422H
knockouts (CRISPR-SNG and
CRISPR-SNG
R422H
) were generated in the WT and Sng back-
grounds, respectively, with the CRISPR-Cas9 system as described
previously (Ma et al., 2015). All primers used for vector construc-
tion are shown in Supporting Information Table S1. All recombi-
nant plasmids were verified via sequencing before plant
transformation. Each plasmid was inserted into Agrobacterium
tumefaciens strain EHA105. Rice calli were transformed using an
Agrobacterium-mediated method to obtain transgenic plants,
which were genotyped with PCR and sequencing.
Gene and protein expression analyses
Target gene expression levels were analyzed with quantitative
reverse transcription PCR (RT-qPCR). Total RNA was isolated
from multiple tissues using the RNeasy Plant Mini Kit (Qiagen).
cDNA was synthesized using SuperScript III reverse transcriptase
(Invitrogen) and 0.5 lg of isolated RNA. RT-qPCR reactions were
conducted in triplicate with SYBR Green PCR Premix (Invitro-
gen) and primers specific for each target gene on a StepOne Real
Time PCR instrument (Applied Biosystems, Foster City, CA,
USA). Gene expression levels were normalized to the internal con-
trol gene OsTCTP (LOC_Os11g43900; Narsai et al., 2010).
To compare the expression levels of His-tagged proteins in
transgenic plants, total protein was isolated from leaf sheaths with
a lysis buffer (0.1 M PIPES, 10 mM MgSO
4
, 2 mM EGTA
(pH 6.8), 0.5 M 3-(1-pyridinio)-1-propane sulfonate (Solarbio,
Beijing, China), 1% CHAPS (Sigma), 5 mM DTT, 1 mM ATP,
1 mM GTP, 2 mM PMSF (Sigma), and 1 9complete protease
inhibitor cocktail (Roche)). The isolated proteins were separated
via sodium dodecyl sulfate (SDS)–polyacrylamide gel electro-
phoresis (PAGE). The separated proteins were electrophoretically
transferred onto a polyvinylidene difluoride membrane, blocked
with 5% nonfat dried milk, and then probed with primary
monoclonal antibodies against the His tag (MBL) or actin
(HUABIO). Secondary antibody conjugated with horseradish
peroxidase was then applied, and chemiluminescent signals were
visualized on a Tanon 5200 Imaging System (Tanon Science &
Technology, Shanghai, China).
Histological analysis
To observe the inner surfaces of lemmas, mature glumes were col-
lected before anthesis and fixed in formaldehyde-acetic acid-
ethanol (FAA) solution (5% formaldehyde, 5% glacial acetic
acid, and 70% ethanol) for >24 h at 4°C. Samples were then
dehydrated in an ethanol gradient (70%, 80%, 90%, 95%, and
100%), incubated in an ethanol : isoamyl acetate series (3 : 1,
1 : 1, and 1 : 3, followed by 100% isoamyl acetate), and critical-
point dried with CO
2
. The dried glumes were mounted on stubs,
coated with gold, and observed under a scanning electron micro-
scope (Hitachi S-4800, Hitachi, Tokyo, Japan).
To observe pericarp cells, segments (2–3 mm in length or
width) in the middle of young caryopses grown in glumes with
the upper half cut for 12 DAF were sectioned transversely or
longitudinally and then immediately fixed in a modified FAA
solution (5% formaldehyde, 6% acetic acid, 50% ethanol, and
New Phytologist (2023)
www.newphytologist.com
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
Research
New
Phytologist
2
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

5% glycerol) for 24 h at 4°C. Samples were dehydrated in an
ethanol gradient and then embedded in LR White resin (The
London Resin Co., Berkshire, UK). Semi-thin sections were pre-
pared on a Leica Ultracut R microtome, stained with 1% tolui-
dine blue, and observed under a bright-field light microscope
(Zeiss Axio Imager A1).
MT fluorescence microscopy analysis
To observe cortical MTs, we first constructed a Ubi::eGFP-b-
tubulin plasmid by inserting a fragment encoding eGFP and the
full-length ORF encoding rice b-tubulin (LOC_Os01g59150)
into the pUN1392 plasmid at the BamHI and BstEII sites. We
also constructed SNG::eGFP-SNG and SNG::eGFP-SNG
R422H
plasmids by fusing the 2.4-kb promoter of SNG,theeGFP frag-
ment, and the genomic sequences of SNG and SNG
R422H
with
the 1.6-kb 30UTR and inserting the assembled fragments into
pCAMBIA1301. WT and Sng transgenic lines expressing
eGFP-b-tubulin, a WT transgenic line expressing eGFP-SNG,
and a Sng transgenic line expressing eGFP-SNG
R422H
were gen-
erated with an Agrobacterium-mediated transformation method.
Lemmas were selected from the elongating glumes (1/3–1/2 of
full length) for observation as described below.
To observe the cortical MT arrays in lemma inner epidermal
cells, elongating glumes were cut from transgenic rice lines
stably expressing eGFP-b-tubulin. The cut glumes were imme-
diately immersed in microtubule-stabilizing buffer (MTSB;
50 mM PIPES, 5 mM MgSO
4
, and 5 mM EGTA at pH 7.0) at
room temperature. For observation of MT nucleation and
regrowth in lemma inner epidermal cells, developing glumes
were cut from transgenic rice lines stably expressing eGFP-SNG
(in the WT background) or eGFP-SNG
R422H
(in the Sng
background). The glumes were incubated in an ice bath for
20–30 min to depolymerize MTs completely. Then, some
glumes were transferred into MTSB buffer immediately,
whereas others were first incubated in MTSB buffer +10 lM
taxol (Sigma) at 37°C for 5 or 10 min for MT polymerization.
After immersion in plain MTSB buffer, all glumes were
photographed under a laser confocal microscope (Zeiss LSM
980 or Olympus FV1000 MPE, Olympus, Tokyo, Japan) at an
excitation wavelength of 488 nm.
To assess the capacity for MT incorporation, lemmas were col-
lected from WT transgenic lines expressing eGFP-SNG and from
Sng transgenic lines expressing eGFP-SNG
R422H
. Lemmas were
immersed in MTSB buffer either immediately or after MT disas-
sembly in an ice bath followed by MT regrowth in MTSB +taxol
for 15 min. Fluorescent images of cell cortical MTs were acquired
using the same parameters of confocal microscopy. Images were
analyzed with ZEISS software (blue edition) to measure the fluor-
escence intensities of MT segments.
To analyze individual MT dynamics, lemmas were cut from
WT transgenic lines expressing eGFP-SNG and from Sng trans-
genic lines expressing eGFP-SNG
R422H
. The lemmas were
immersed in MTSB buffer. A time series was captured over 240 s
at 4-s intervals under a laser confocal microscope (Zeiss LSM 980).
MTs with a clear plus end were selected for measurements, which
were performed using IMAGE-PRO PLUS software. The measurement
data were analyzed in Excel. Individual MT dynamic videos and
kymographs were created with IMAGEJsoftware.
Affinity purification of SNG/SNG
R422H
with b-tubulin
To co-precipitate SNG/SNG
R422H
with b-tubulin, we purified
tubulin heterodimers from the SNG::SNG and SNG::SNG
R422H
transgenic plants with a modified version of a previously
described purification method (Minoura et al., 2013). Briefly,
elongating leaf sheaths were ground into fine powder in liquid
nitrogen, then mixed with lysis buffer at a ratio of 1 : 2 (w/v),
and incubated for 30 min on ice. Lysates were centrifuged at
19 000 gfor 25 min at 4°C, and the supernatants were collected
and centrifuged again with the same parameters. The resulting
supernatant was supplemented with 10% (v/v) glycerol, incu-
bated with DEAE Sepharose Fast Flow resin (Sigma) for 60 min
on ice, and washed with DEAE wash buffer (0.1 M PIPES,
10 mM MgSO
4
, 10% glycerol, 60 mM NaCl, 1 mM ATP,
1 mM GTP, and 1 9complete protease inhibitor cocktail
(Roche), pH 6.8). Samples were then eluted with DEAE elution
buffer (0.1 M PIPES, 10 mM MgSO
4
, 10% glycerol, 0.4 M
NaCl, 1 mM ATP, 1 mM GTP, and 1 9cocktail, pH 7.0). The
eluted solution was incubated with Ni-NTA-His Bind Resin
(Merck, Darmstadt, Germany) for 45 min on ice. Finally, the
His resin was washed with His wash buffer (0.1 M PIPES, 5 mM
MgSO
4
, 10% glycerol, 0.3 M NaCl, and 20 mM imidazole,
pH 7.0) and eluted with His elution buffer (0.1 M PIPES, 5 mM
MgSO
4
, 10% glycerol, 0.3 M NaCl, and 0.25 M imidazole,
pH 7.0). Samples were analyzed with SDS-PAGE after each puri-
fication step. The final eluant (containing recombinant tubulin
heterodimers) was analyzed via western blot with anti-His tag
(MBL) and anti-b-tubulin (HUABIO) antibodies. The protein
band intensities were calculated in IMAGEJ.
Transient expression assays
eGFP-SNG and eGFP-SNG
R422H
were separately transiently
expressed in tobacco (Nicotiana benthamiana) leaves to assess
MT incorporation. The 35S::eGFP-SNG and 35S::eGFP-SNG-
R422H
constructs were generated by fusing eGFP, the full-length
SNG or SNG
R422H
ORF, and the NOS terminator into pCAM-
BIA1301 at the BglII site. Each construct was verified with
sequencing. tobacco leaves were transformed via infiltration with
Agrobacterium containing one of the two plasmids (Drevensek
et al., 2012).
Molecular modeling
The a-tubulin protein SNG and a rice b-tubulin (LOC_Os01g59150)
were used to establish the tertiary structure of the heterodimer.
Protein sequences were downloaded from the Protein Data Bank
(PDB), and homology modeling was conducted with the
SWISS-MODEL online server (Swiss Institute of Bioinformatics,
Basel, Switzerland). The structure of a highly homologous
tubulin heterodimer (1JFF) was used for modeling. The mutate
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
New Phytologist (2023)
www.newphytologist.com
New
Phytologist Research 3
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

tool in SWISS-PDB VIEWER v.4.1.0 software was used to simulate
the effect of the R422H substitution (GlaxoSmithKline R&D,
Geneva, Switzerland; Arnold et al., 2006).
Results
SNG affects grain size and quality
To identify genes associated with grain notching, we performed a
forward genetic screen with chemical mutagenesis, from which
we identified the Sng mutant. Compared with WT rice, Sng had
shorter but wider and thicker grains and a lower thousand-grain
weight (Fig. 1a–e). In the F
1
population of a cross between WT
and Sng, the florets, grains, and caryopses were larger than those
from Sng individuals, but smaller than those from WT plants
(Fig. 1a–e). Each Sng grain had notches on the ventral or/and
dorsal sides, with 18% and 82% of grains having one and two
notches, respectively. By contrast, just 32% of the grains from F
1
plants had one notch on the ventral side, and none of the WT
grains had any notches (Fig. 1f,g). The Sng mutants showed a sig-
nificantly higher percentage of chalky grains and a lower propor-
tion of head rice compared with the WT; grains from F
1
plants
had moderate chalkiness and breakage phenotypes (Fig. 1f,h–j).
Thus, the gene mutated in Sng plants significantly affected grain
yield and quality.
Further observations of caryopsis development showed that
the morphology and size of Sng caryopses grown in intact glumes
were similar to those of WT caryopses from 1 to 4 DAF (Fig. 2).
From 6 DAF, Sng caryopses buckled and wrinkled due to the
reduced glume space caused by the shortened glumes, which led
to production of small, notched grains (Fig. 2). When the Sng
caryopses developed in florets with the upper half cut, they
became longer and narrower than WT caryopses starting from
4 DAF (Fig. 2). These observations indicated that unbalanced
growth between the glumes and caryopses in the Sng mutants led
to limitations on caryopsis elongation, which in turn caused com-
pression and notching of the caryopses. In addition, the Sng
mutants showed a semi-dwarf phenotype with a slightly
decreased number of primary branches and grains per panicle
compared with WT plants (Fig. S1).
Heterozygotes in the F
1
generation showed phenotypes that
were intermediate between the WT and Sng plants. The phenoty-
pic segregation ratio in the F
2
population was the same as the gen-
otypic ratio (1 : 2 : 1 WT, heterozygous, and Sng mutant plants;
Table S2). This indicated that Sng phenotypes were caused by a
mutation in a single gene with incomplete dominant effects. The
mutant locus was fine-mapped to a 24.5-kb region between the
molecular markers RM21859 and D5-1 on the long arm of chro-
mosome 7 (Fig. S2a). This region contained 10 genes, only one of
which had a missense mutation in Sng compared with WT plants.
Fig. 1 Small and notched grain (Sng) mutants showed defects in grain shape, yield, and quality. (a) Florets and mature grains of wild-type (WT), Sng, and
F
1
plants (from a WT 9Sng cross). Bars, 1 mm. (b–e) Length, width, thickness, and weight of grains collected from each genotype. (f) Representative selec-
tion of WT, Sng, and F
1
grains with notched and chalky caryopses indicated. For chalky caryopses, the opaque areas were always on the ventral side of the
brown rice (Oryza sativa). Bar, 1 mm. (g, h) Proportions of notched and chalky grains from WT, Sng, and F
1
plants. (i) Representative images of 100
dehulled and milled rice grains per genotype separated into head and broken grains. Bar, 1 cm. (j) Proportions of head rice from WT, Sng, and F
1
plants.
In (b–e, g, h, j), data are presented as the mean SD from three separate plants each for WT and Sng. For grain size measurements, n≥50 grains per geno-
type. Grain weight was measured three times (with ≥32 grains per measurement) and the thousand-grain weight was calculated from those measure-
ments. For grain percentages, n=3 assessments with 100 grains per replicate. Letters indicate statistically significant differences at P<0.01 (one-way
analysis of variance, ANOVA).
New Phytologist (2023)
www.newphytologist.com
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
Research
New
Phytologist
4
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

The mutation was in LOC_Os07g38730, which encodes the a-
tubulin protein OsTUBA3. A G to A substitution in the fifth exon
at position 1265 of the full-length ORF (G1265A) caused a mis-
sense Arg to His substitution at position 422 (R422H; Fig. S2b,c).
R422 is conserved among a-tubulins. Homology structural mod-
eling showed that R422 was in the C-terminal H12 a-helix
(Fig. S3a;Nogaleset al., 1998; Fourel & Boscheron, 2020). The
guanidinium ion of R422 formed 3 H bonds with the carboxylate
group of D396 and a salt bridge with the carboxylate group of
D392 in the H11 helix (Fig. S3b), whereas the R422H substitu-
tion likely abolished these interactions because the His residue
would have been too far away from the D392 and D396 residues
(Fig. S3c). This suggested important roles of R422 in SNG.
The R422H substitution in SNG (SNG
R422H
) is a dominant-
negative mutation
To confirm that the R422H mutation in SNG/OsTUBA3 was
responsible for Sng phenotypes, we generated four transgenic rice
lines in WT backgrounds that expressed His-tagged WT or
mutated SNG under the control of the 35S or the SNG promoter,
yielding the 35S::SNG,SNG::SNG,SNG::SNG
R422H
,and35S::
SNG
R422H
lines. The 35S::SNG and 35S::SNG
R422H
lines had
higher levels of the target proteins than the SNG::SNG and SNG::
SNG
R422H
lines, respectively (Fig. S4). We also generated SNG
and SNG
R422H
knockout lines in the WT and Sng backgrounds,
respectively, with CRISPR (the CRISPR-SNG and
CRISPR-SNG
R422H
lines, respectively; Fig. S5). The 35S::
SNG
R422H
line produced small, notched grains and had a semi-
dwarf phenotype, comparable to Sng mutants. The SNG::
SNG
R422H
line displayed weak Sng phenotypes, similar to the het-
erozygotes (Fig. 3). By contrast, the 35S::SNG,SNG::SNG,
CRISPR-SNG, and CRISPR-SNG
R422H
lines were phenotypically
similar to WT plants (Fig. 3). These results suggested that
SNG
R422H
expression led to Sng-like phenotypes in a dosage-
dependent manner and that knocking out SNG
R422H
rescued the
Sng phenotype. We therefore inferred that SNG
R422H
was a
dominant-negative mutation with dosage effects. Together, these
observations demonstrated that the R422H substitution in SNG
led to the grain and plant phenotypes observed in Sng mutants.
Thus, we also referred to Sng as SNG
R422H
.
SNG
R422H
affects cell anisotropy and induces imbalanced
glume and caryopsis growth
To determine the role of SNG in grain development, we first ana-
lyzed SNG expression patterns by RT-qPCR. Although SNG was
ubiquitously expressed in various organs including the leaves,
stems, florets, and caryopses, it was most highly expressed in
elongating or developing tissues (Fig. S6). This suggested that
SNG was likely required for organ growth.
We then performed cytological observations to establish the
mechanism by which the R422H mutation in SNG caused
shortened glumes but elongated caryopses. Scanning electron
Fig. 2 Small and notched grain (Sng) mutant
caryopsis phenotypes in intact glumes and
glumes with the upper half cut. (a)
Developing caryopses grown in intact glumes
or glumes with the upper half cut. The
numbers above the images indicate the
number of days after flowering (DAF). Bar,
1 mm. (b) Quantification of caryopsis length,
width, and weight after development in
intact glumes or glumes with the upper half
cut. Data are presented as the mean SD
(n=21–52 caryopses at each timepoint from
≥3 plants per genotype).
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
New Phytologist (2023)
www.newphytologist.com
New
Phytologist Research 5
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

microscopy revealed that the inner epidermal cells of Sng lem-
mas were much shorter and slightly wider than those of WT
plants (Fig. 4a,b). Similarly, pericarp cells of Sng caryopses
were greatly reduced in length and increased in height, but
unchanged in width compared with those of WT plants
(Fig. 4d,e). There were no significant differences in the number
of inner epidermal cells in lemmas or the number of pericarp
cells between the caryopses of WT and Sng plants (Fig. 4c,f).
Importantly, glumes elongated in a direction parallel to the
length of the glume cells, whereas caryopses elongated in a
direction that was parallel to the height and perpendicular to
the length of the pericarp cells (Fig. 4a,d). Therefore, the
reduction in glume cell length and the increase in pericarp cell
height caused glume shortening but caryopsis elongation in
Sng. These results demonstrated that the differential effects of
SNG
R422H
on glume and caryopsis elongation were not due to
differences in expansion or the dimensions of glume and car-
yopsis cells; these differing effects arose because the direction
of cell elongation in glumes and caryopses was orthogonal.
SNG
R422H
leads to defects in MT orientation
The orientation and dynamics of cortical MTs directly affect the
direction of cell expansion, which determines cell shape (Smith
& Oppenheimer, 2005). We therefore first observed cortical MT
arrays in the inner epidermal cells of immature lemmas from
transgenic WT and Sng rice expressing an eGFP-b-tubulin fusion
protein. Elongating lemmas in both WT and Sng plants had three
major types of cortical MT arrays in cells: a transverse array, with
most MTs perpendicular to the cell length; an oblique or ran-
domly aligned MT array; and a longitudinal array, with most
MTs parallel to the direction of cell elongation (Fig. 5a). Sng
lemmas contained a relatively low proportion of cells with the
transverse MT array and a high percentage of cells with the obli-
que and longitudinal MT arrays compared with WT lemmas
(Fig. 5b). We then observed the distribution of cells with trans-
verse MT arrays in WT and Sng plants. WT cells with the trans-
verse MT array were primarily distributed in a length range of
30–75 lm, whereas Sng cells with this array ranged in length
Fig. 3 Small and notched grain (Sng) was a dominant-negative mutation. (a, b) Images of mature florets, grains, and caryopses (a) and grain phenotypes
(b) in wild-type (WT), Sng,SNG/SNG
R422H
-overexpression, and SNG/SNG
R422H
knockout lines. Bar, 1 mm. Data are presented as the mean SD (n≥50
grains from three plants or lines each). Letters indicate statistically significant differences at P<0.01 (one-way ANOVA). (c, d) Representative images of
whole mature plants (c) and quantification of mature plant height (d) for the lines indicated in (a). Bar, 10 cm. Data are presented as the mean SD (n≥30
plants). Letters indicate statistically significant differences at P<0.01 (one-way ANOVA).
New Phytologist (2023)
www.newphytologist.com
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
Research
New
Phytologist
6
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Fig. 4 SNG
R422H
affected the anisotropic elongation of cells in both glumes and caryopses. (a) Inner surfaces of mature lemmas as observed with scanning
electron microscopy. Areas outlined in pink in the images at left are magnified in the middle and right images to show inner epidermal cells of the lemmas.
Bars: (left) 0.5 mm; (right) 100 lm. (b) Length and width of the inner epidermal cells shown in (a). (c) Total number of inner epidermal cells on the
longitudinal and transverse axes of mature lemmas. (d) Longitudinal and transverse sections of pericarps from caryopses grown in glumes with the upper
half cut for 12 DAF (days after flowering). Blue and pink lines on the caryopses indicate the positions of the longitudinal and transverse sections,
respectively. Bars: (left) 1 mm; (right) 50 lm. (e) Length, width, and height of the pericarp cells shown in (d). (f) Total number of pericarp cells on the
longitudinal and transverse axes of caryopses shown in (d). In (a, d), red lines mark the length (L), width (W), and height (H) of representative cells.
In (b, c), cell length was measured in 136 cells from three wild-type (WT) lemmas and 160 cells from five Small and notched grain (Sng) mutant lemmas;
cell width was measured in 195 and 205 cells from six WT and six Sng lemmas, respectively; and cell number was measured in 10 WT and Sng lemmas.
In (e, f), cell length and width were measured in 360 and 392 cells from six WT and six Sng caryopses, respectively; cell height was measured in 415 and
313 cells from three WT and three Sng caryopses, respectively; cell number in the longitudinal direction was measured in three WT and Sng caryopses; and
cell number in the transverse direction was measured in six WT and Sng caryopses. All lemmas and caryopses were selected from three WT and three Sng
plants. Data are presented as the mean SD. **,P<0.01; ns, not significant (Student’s t-test).
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
New Phytologist (2023)
www.newphytologist.com
New
Phytologist Research 7
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

from 0 to 60 lm and were concentrated chiefly from 15 to
30 lm (Fig. 5c,d). It is generally accepted that transverse MT
arrays promote cell elongation. Thus, these observations partially
explained the reduced cell length in Sng.
SNG
R422H
decreases the MT growth rate and increases MT
instability
We next traced the dynamics of individual MTs in WT and Sng
lemma epidermal cells and measured the associated dynamic
parameters, including the MT growth and shrinkage rates and
the transition frequencies between the MT growth, shrinkage,
and pause phases (Fig. 6; Videos S1–S4). The plus ends of indivi-
dual MTs in Sng cells differed significantly from those in WT
cells (Fig. 6). No significant differences were detected in the MT
shrinkage rate between Sng and WT cells (Fig. 6e). However, the
MT growth rate was significantly lower in Sng than in WT cells
(P<0.01 (Student’s t-test); Fig. 6a,b,e; Videos S1,S2). The res-
cue and catastrophe frequencies of MTs appeared to be
equivalent in Sng cells, whereas WT cells had a higher frequency
of MT rescue than catastrophe (Fig. 6c–e; Videos S3,S4). The
MTs in Sng cells were in the growth phase for shorter periods of
time and in the shrinkage phase for longer than those in WT cells
(Fig. 6e). MTs in Sng cells also displayed lower and higher Kg-g
and Ks-s transition rates, respectively, than those in WT cells.
Furthermore, there were more total MT events in Sng than in
WT cells (Fig. 6e). These data suggested that MT growth was
impaired and MT dynamics were upregulated in Sng cells.
Consistently, after cold-induced disassembly, MT nucleation
sites were visible in WT cells within 10 min. By contrast, these
sites only became visible in Sng cells at 30 min after returning to
room temperature (Fig. 7a). In response to taxol treatment, both
WT and Sng cells displayed three general categories of MT
regrowth: (1) depolymerized arrays without regrown MTs; (2)
partially regrown arrays with radial MTs; and (3) fully regrown
arrays with crosslinked MT networks (Fig. 7b). However, Sng
plants had a significantly lower percentage of cells with fully
regrown MT arrays and a higher percentage of cells without MT
Fig. 5 Small and notched grain (Sng) mutants had fewer cells with cortical microtubules (MTs) in the transverse orientation. (a) Representative microscopic
images showing the three main types of MT arrays in the inner epidermal cells of elongating lemmas collected from wild-type (WT) and Sng plants expres-
sing eGFP-tagged b-tubulin. Bars, 1 lm. (b) Distribution of transverse, oblique/randomly aligned, and longitudinal MT arrays in lemma inner epidermal cells
(n>600 cells from three transgenic WT and four transgenic Sng lines, respectively). P=3.15E-07 (chi-squared test). (c) Elongating lemma inner epidermal
cells with a transverse MT array. Bars, 5 lm. (d) Distribution of cell lengths among lemma inner epidermal cells with a transverse MT array (n>200 cells
from three transgenic WT and four transgenic Sng lines, respectively). P=1.71E-49 (chi-squared test).
New Phytologist (2023)
www.newphytologist.com
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
Research
New
Phytologist
8
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

regrowth than WT plants (Fig. 7c). These results demonstrated
that the R422H substitution in SNG led to defective MT nuclea-
tion and regrowth. Collectively, these results indicated that the
R422H substitution impaired MT growth and increased MT
instability.
SNG
R422H
is defective in interactions with b-tubulin and in
MT incorporation
To understand how the R422H mutation in SNG affected Sng
cell function, we analyzed the localization of SNG
R422H
and
SNG in vivo. Both SNG and SNG
R422H
could be successfully
incorporated into cortical MTs, although fluorescence was
much weaker in MTs incorporating eGFP-SNG
R422H
than
eGFP-SNG (Fig. 8a,c). Transient expression of eGFP-tagged
SNG or SNG
R422H
in tobacco leaf epidermal cells demon-
strated eGFP-SNG assembly into cortical MTs but diffusion of
eGFP-SNG
R422H
into the cytoplasm (Fig. 8d). Interestingly,
after 15 min of treatment with the MT stabilizer taxol, the
MT fluorescence intensities were similar between cells expres-
sing SNG
R422H
and those expressing SNG (Fig. 8b,c). This
suggested that MTs incorporating SNG
R422H
were less stable
compared with those incorporating SNG. To determine
whether the instability of MTs containing SNG
R422H
resulted
from altered interactions between SNG
R422H
and b-tubulin,
we performed pulldown experiments with purified His-tagged
SNG and SNG
R422H
(Fig. 9a). The amount of b-tubulin that
co-precipitated with SNG
R422H
was much lower than the
amount that co-precipitated with SNG (Fig. 9b,c), implying
that interactions between SNG
R422H
and b-tubulin were rela-
tively weak and that the SNG
R422H
-b-tubulin heterodimer may
have been defective. Collectively, these observations suggested
that the R422H substitution in SNG impaired interactions of
the mutated protein with b-tubulins and reduced their incor-
poration into MTs.
Discussion
A proportion of rice varieties produce notched grains under nat-
ural growth conditions, and this phenotype also frequently occurs
in response to environmental stress or excess fertilizer application.
Grain notching is a common deformation that is accompanied
by reductions in rice yield and quality (Xiong et al., 1986), but
the mechanisms underlying this trait remain largely unknown.
Fig. 6 Dynamic instability behavior of individual microtubules (MTs) in vivo. (a, b) Representative images of a growing MT in lemma inner epidermal cells
from wild-type (WT) (a) and Small and notched grain (Sng) mutant (b) lines expressing eGFP-SNG and eGFP-SNG
R422H
, respectively. Corresponding
kymographs are shown at the far right. (c, d) Representative images of catastrophizing and rescuing MTs in lemma inner epidermal cells from WT (c) and
Sng (d) lines expressing eGFP-SNG and eGFP-SNG
R422H
, respectively. Corresponding kymographs are shown at the far right. In (a–d), yellow arrowheads
indicate the plus ends of selected MTs and white numbers indicate the elapsed time in s. In (c, d), the yellow letters C and R highlight MT catastrophe and
rescue events, respectively. Bars, 1 lm. Supporting Information Videos S1–S4 show the entire sequences of events shown in (a–d), respectively. (e) Key
parameter dynamics for individual MTs in vivo. g, growth; K, rate of transitions between dynamic states (events min
1
); p, pause; s, shrinkage. For exam-
ple, Kg-p represents the switching frequency from the growth state to pause state. n=146 MTs from three transgenic WT lines and 142 MTs from three
transgenic Sng lines. Dynamicity was calculated as the sum of the total grown and shortened length divided by the total time a particular MT was observed.
Growth rate, shrinkage rate, and dynamicity are expressed as the mean SD.
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
New Phytologist (2023)
www.newphytologist.com
New
Phytologist Research 9
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

We here identified a novel grain quality gene, SNG/OsTUBA3,
and revealed that the Arg residue at position 422 of OsTUBA3
was crucial for the roles of this protein in maintaining the normal
behavior of MTs that affects grain notching and size, thus deter-
mining grain quality and yield. Overall, our study provides novel
insights into the molecular regulation of grain notching in rice, as
detailed below.
The distinctive layout of glume and caryopsis cells that
elongate in orthogonal directions is a crucial cytological
basis of grain notching
Our observations clearly revealed that glume and caryopsis cells
expanded in orthogonal directions. Glume cell lengths were paral-
lel to the longitudinal axes of the glumes, whereas the lengths of
caryopsis cells were perpendicular to the longitudinal axes of car-
yopses; caryopsis cell heights were parallel to the longitudinal axes
(Fig. 10). In Sng mutants, the R422H substitution decreased
glume and caryopsis cell length, but increased caryopsis cell height.
Thus, glume length was reduced as caryopsis length was dramati-
cally increased, causing severe compression and notching of the
caryopses enclosed in the shortened glumes (Fig. 10a,b). The con-
trasting effects of R422H substitution on glume and caryopsis
elongation were not due to differences in the regulation of cell
expansion, since both glume and caryopsis cells were shorter but
wider/higher in Sng than WT cells (Fig. 10c,d). The contrasting
effects resulted from the perpendicular arrangement of caryopsis
cells to glume cells with respect to the direction of elongation. The
lengths of glume cells but the heights of caryopsis cells were
aligned with the direction of organ elongation (Fig. 10). These
observations suggested that genes or/and environmental stresses
regulating cell expansion and size, rather than factors controlling
cell division, can differentially affect glume and caryopsis dimen-
sions, thus being more likely to affect grain notching. These results
provide essential cytological knowledge about glume and caryopsis
cell and organ growth, informing our understanding of the mole-
cular mechanisms underlying grain notching.
Mutation of R422 in SNG/OsTUBA3 affects MT dynamics
SNG is one of the four a-tubulin genes in the rice genome (Jayas-
wal et al., 2019). Although SNG is abundantly expressed in devel-
oping tissues, plants overexpressing SNG or with SNG knocked
out showed no detectable abnormalities. This suggested that SNG
functioned redundantly with other a-tubulin genes in cell and
organ development. Previous studies have shown that missense
mutations in genes encoding a-orb-tubulins lead to a wide spec-
trum of phenotypic defects in many species, including rice (Segami
et al., 2012), Arabidopsis thaliana (Ishida et al., 2007), and mam-
malian species (Kumar et al., 2010;Tischfieldet al., 2011).
Fig. 7 Small and notched grain (Sng) mutants were defective in microtubule (MT) regrowth. (a) Time-lapse images showing MT regrowth after cold-
induced MT disassembly in the inner epidermal cells of developing glumes from wild-type (WT) and Sng lines expressing eGFP-SNG and eGFP-SNG
R422H
,
respectively. Bar, 5 lm. White arrows indicate MT nucleation sites. (b) Three typical types of MT regrowth status after taxol treatment. Bar, 5 lm. (c) Distri-
bution of cells in each MT regrowth state shown in (b) at 5 and 10 min after taxol treatment (n>200 cells from each of three transgenic WT and Sng lines,
respectively). P=7.88E-14 and 2.46E-09 at 5 and 10 min, respectively (Chi-square test).
New Phytologist (2023)
www.newphytologist.com
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
Research
New
Phytologist
10
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Fig. 8 SNG
R422H
was incorporated into
microtubules (MTs) less efficiently than SNG.
(a) Incorporation of eGFP-SNG and eGFP-
SNG
R422H
into MTs in rice (Oryza sativa)
lemma inner epidermal cells from wild-type
(WT) and Small and notched grain
(Sng) mutant lines expressing eGFP-SNG and
eGFP-SNG
R422H
, respectively. Bar, 10 lm.
(b) Incorporation of eGFP-SNG and eGFP-
SNG
R422H
into MTs in rice lemma inner
epidermal cells from WT and Sng lines
expressing eGFP-SNG and eGFP-SNG
R422H
,
respectively, after taxol treatment. Bar,
10 lm. (c) Quantification of fluorescence
intensity from eGFP-SNG and eGFP-
SNG
R422H
incorporated into MTs in rice
lemma inner epidermal cells. Untreated
samples, n=80 MTs in eight cells from four
transgenic WT and Sng lines, respectively;
taxol-treated samples, n=75 MTs in eight
cells from four transgenic WT and Sng lines,
respectively. Data are presented as the
mean SD. Letters indicate statistical
significance groups at P<0.01 (one-way
ANOVA). (d) Incorporation of eGFP-SNG
and eGFP-SNG
R422H
into MTs in tobacco leaf
epidermal cells. Bar, 10 lm.
Fig. 9 SNG
R422H
had decreased binding affinity for b-tubulin compared with SNG. (a) Purification of His-tagged SNG and SNG
R422H
proteins. Sodium
dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) results showing purification of His-tagged SNG and SNG
R422H
from transgenic rice (Oryza
sativa) lines expressing SNG::SNG and SNG::SNG
R422H
, respectively. Wild-type (WT) plants served as the control. DEAE, diethylaminoethyl, a resin used to
adsorb anion; Ni-NTA, Ni
2+
cation held by chelation with nitriloacetic acid (NTA), a resin used to bind His-tagged protein. (b) Co-precipitation of His-
tagged SNG and SNG
R422H
with b-tubulin. L1–L3 and mL1–mL3 are independent transgenic rice lines expressing His-tagged SNG and SNG
R422H
, respec-
tively. (c) Relative intensity ratios of b-tubulin co-precipitated with His-SNG and His-SNG
R422H
based on normalized intensity of bands from the blots
shown in (b). Data are presented as the mean SD from three independent transgenic lines. **,P<0.01 (Student’s t-test).
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
New Phytologist (2023)
www.newphytologist.com
New
Phytologist Research 11
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

However, the mechanisms underlying the phenotypic defects
caused by missense mutations in tubulins are not clear. The Sng
mutant generated in the present study served as an important tool
to delineate the associated mechanism. SNG
R422H
produced small,
notched grains, and higher SNG
R422H
expression was associated
with more severe phenotypes (Fig. 3). SNG
R422H
knockout lines in
the Sng substitution mutant (CRISPR-SNG
R422H
) had WT pheno-
types (Fig. 3). These results indicated that the R422H substitution
in SNG was a dominant-negative mutation with dosage effects.
This finding is consistent with observations of human tubulin; a
mutation causing neurological disorders is considered to function
via a dominant-negative mechanism rather than tubulin haploin-
sufficiency (Kumar et al., 2010; Tischfield et al., 2011).
Our results showed that SNG
R422H
was defective in b-tubulin
binding (Fig. 9b,c), but still had the capacity to be incorporated
into MTs (Fig. 8a). Because a-tubulin must be associated with
b-tubulin in heterodimers for MT assembly, we speculate that
SNG
R422H
was competent to form heterodimers and could
therefore be incorporated into MT, but that the heterodimer
was defective in the conformation and/or strength of the bond to
b-tubulin. Consistent with this hypothesis, a previous study indi-
cated that the D417H mutation in human TUBB3, which lies
on the MT lattice surface and is proximal to the kinesin binding
site but distal from the intra-subunit contacts, alters heterodimer
conformation (Ti et al., 2016). Thus, mutations in similar sites
appear to cause conformational alterations of the heterodimer.
Furthermore, MTs assembled with SNG
R422H
displayed
abnormalities in nucleation, regrowth (Fig. 7), and dynamic
instability behavior (Fig. 6). Similarly, a previous study showed
that the P263T and R402H mutations in the mammalian a-
tubulin protein TUBA1A cause brain malformation and neuro-
nal differentiation defects in tubulinopathy patients; this is asso-
ciated with delayed MT nucleation and regrowth, and the
mutant a-tubulins are not incorporated into MTs as efficiently
as WT a-tubulins (Yu et al., 2016). Interestingly, the MT-
stabilizing agent taxol improves incorporation of TUBA1A-
P263T and TUBA1A-R402H into MTs (Yu et al., 2016), simi-
lar to the effects of taxol on SNG
R422H
that we observed here
(Fig. 8b,c). We therefore hypothesize that SNG
R422H
partici-
pated in altered interactions with adjacent tubulins when they
were incorporated into MTs. Together, these results suggested
that rice TUBA3-R422H and mammalian TUBA1A-P263T/
TUBA1A-R402H may exhibit similar defects in MT dynamics
and organization in both plants and animals.
Mutation of R422 in SNG/OsTUBA3 may alter lateral
contacts between protofilaments and interactions between
MTs and MT-associated proteins (MAPs)
R422 was a conserved amino acid residue among a-tubulins and
was in the C-terminal H12 a-helix (Fig. S3a; Nogales
et al., 1998; Fourel & Boscheron, 2020). The guanidinium ion
of R422 formed 3 H bonds with the carboxylate group of D396
and a salt bridge with the carboxylate group of D392 in the H11
helix (Fig. S3b). The R422H substitution likely abolished these
interactions because the His residue would have been too far
away from the D392 and D396 residues (Fig. S3c). Therefore,
the R422H substitution likely affected the interactions, orienta-
tions, and conformations of the H11-H12 helices (Kumar
et al., 2010). Lateral contacts between protofilaments entail
extensive interactions between homologous subunits (L€owe
et al., 2001), involving interactions of the M-loop of one subunit
with the H12 and H5 helices in the adjacent subunit (L€owe
et al., 2001). The R422H substitution in helix H12 could there-
fore have altered the interactions between the adjacent a-tubulin
subunits. In addition, the H11 and H12 helices are on the out-
side surface of the MT lattice, which serves the main binding
region for many MAPs and motor proteins (Nogales
et al., 1998). The R422H substitution and resulting alterations
of the fine helix H11–H12 structures could have altered MT
interactions with MAPs. Future experiments should address
whether altered interactions between protofilaments and
Fig. 10 Working model of SNG-mediated grain notching. (a, b) Illustration
of mature grains from wild-type (WT) (a) and Small and notched grain
(Sng) mutant (b) plants showing the outer glumes (yellow) and inner car-
yopses (gray). Compared with WT plants, Sng mutants had shorter and
wider glumes, which led to round, notched caryopses due to compression
of the elongated caryopses within shortened glumes. (c, d) Illustration of
caryopsis (gray) and glume (yellow) cells from WT (c) and Sng (d) plants.
The directions of caryopsis and glume elongation were parallel to the
length of the glume cells and the height of the caryopsis cells. Both glume
and caryopsis cells in Sng plants were shorter but wider/higher than those
in WT plants due to SNG
R422H
mutation-induced reductions in the trans-
verse orientation of cortical microtubules (MTs; green lines).
New Phytologist (2023)
www.newphytologist.com
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
Research
New
Phytologist
12
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

interactions between MTs and MAPs contribute to the observed
changes in MT organization and dynamics in Sng mutants.
Acknowledgements
We thank Fengqin Dong, Jingquan Li, and Xiuping Xu at the
Institute of Botany, Chinese Academy of Sciences (IB, CAS) for
their contributions to histological analysis, confocal microscopy,
and scanning electron microscopy, respectively. We thank
Zhaosheng Kong at Shanxi Agricultural University for his techni-
cal advice for fluorescence microscopy of MTs. This work was
supported by the Chinese Academy of Sciences (grant no.
XDA24010103) and the Ministry of Science and Technology of
the People’s Republic of China (grant no. 2020YFE0202300).
Competing interests
None declared.
Author contributions
CX, ZD and TW designed research. CX performed most experi-
ments. BC conducted Sng mutant screen and mapping-based
cloning of SNG. CX and ZD analyzed the data. ZD and CX pre-
pared the figures and tables and wrote the original manuscript.
SH revised the manuscript. TW contributed grants, reagents, and
materials, and revised the paper.
ORCID
Zhuyun Deng https://orcid.org/0000-0001-8286-1960
Shanjin Huang https://orcid.org/0000-0001-9517-2515
Tai Wang https://orcid.org/0000-0003-2752-697X
Chenshan Xu https://orcid.org/0000-0001-9909-7671
Data availability
All data supporting the findings of this study are available within
this article, and its Supporting Information is published online.
References
Arnold K, Bordoli L, Kopp J, Schwede T. 2006. The SWISS-MODEL workspace: a
web-based environment for protein structure homology modeling.
Bioinformatics 22: 195–201.
Deng ZY, Liu LT, Li T, Yan S, Kuang BJ, Huang SJ, Yan CJ, Wang T.
2015. OsKinesin-13A is an active microtubule depolymerase involved in
glume length regulation via affecting cell elongation. Scientific Reports 5:
9457.
Drevensek S, Goussot M, Duroc Y, Christodoulidou A, Steyaert S, Schaefer E,
Duvernois E, Grandjean O, Vantard M, Bouchez D et al. 2012. The
Arabidopsis TRM1–TON1 interaction reveals a recruitment network common
to plant cortical microtubule arrays and eukaryotic centrosomes. Plant Cell 24:
178–191.
Fan Y, Li Y. 2019. Molecular, cellular and Yin-Yang regulation of grain size and
number in rice. Molecular Breeding 39: 163.
Fourel G, Boscheron C. 2020. Tubulin mutations in neurodevelopmental
disorders as a tool to decipher microtubule function. FEBS Letters 594: 3409–
3438.
Goto E, Kumagai S. 2009. The effect of temperature and shading on notched-
belly rice kernel in Hokkaido rice variety "Nanatsuboshi". Japanese Journal of
Crop Science 78:35–42.
Ishida T, Kaneko Y, Iwano M, Hashimoto T. 2007. Helical microtubule arrays
in a collection of twisting tubulin mutants of Arabidopsis thaliana.Proceedings
of the National Academy of Sciences, USA 104: 8544–8549.
Jayaswal PK, Shanker A, Singh NK. 2019. Phylogeny of actin and tubulin gene
homologs in diverse eukaryotic species. Indian Journal of Genetics and Plant
Breeding 79: 284–291.
Kumar RA, Pilz DT, Babatz TD, Cushion TD, Harvey K, Topf M, Yates L,
Robb S, Uyanik G, Mancini GMS et al. 2010. TUBA1A mutations cause wide
spectrum lissencephaly (smooth brain) and suggest that multiple neuronal
migration pathways converge on alpha tubulins. Human Molecular Genetics 19:
2817–2827.
Li L, Cheng Z, Zhai H. 2012. QTL analysis on notched-belly kernal in rice.
Crops 28:13–17.
Li N, Xu R, Duan P, Li Y. 2018. Control of grain size in rice. Plant Reproduction
31: 237–251.
Lin Z, Zhang X, Yang X, Li G, Tang S, Wang S, Ding Y, Liu Z. 2014.
Proteomic analysis of proteins related to rice grain chalkiness using iTRAQ and
a novel comparison system based on a notched-belly mutant with white-belly.
BMC Plant Biology 14: 163.
L€owe J, Li H, Downing KH, Nogales E. 2001. Refined structure of ab tubulin at
3.5

A resolution. Journal of Molecular Biology 313: 1045–1057.
Ma X, Zhang Q, Zhu Q, Liu W, Chen Y, Qiu R, Wang B, Yang Z, Li H, Lin Y
et al. 2015. A robust CRISPR/Cas9 system for convenient, high-efficiency
multiplex genome editing in monocot and dicot plants. Molecular Plant 8:
1274–1284.
Minoura I, Hachikubo Y, Yamakita Y, Takazaki H, Ayukawa R, Uchimura S,
Muto E. 2013. Overexpression, purification, and functional analysis of
recombinant human tubulin dimer. FEBS Letters 587: 3450–3455.
Narsai R, Ivanova A, Ng S, Whelan J. 2010. Defining reference genes in Oryza
sativa using organ, development, biotic and abiotic transcriptome datasets.
BMC Plant Biology 10: 56.
Nogales E, Wolf SG, Downing KH. 1998. Structure of the ab tubulin dimer by
electron crystallography. Nature 391: 199–203.
Pavithran K. 1977. Inheritance and linkage relationship of notched kernel in rice
(Oryza sativa). Canadian Journal of Genetics and Cytology 19: 483–486.
Segami S, Kono I, Ando T, Yano M, Kitano H, Miura K, Iwasaki Y. 2012.
Small and round seed 5 gene encodes alpha-tubulin regulating seed cell
elongation in rice. Rice 5:4.
Smith LG, Oppenheimer DG. 2005. Spatial control of cell expansion by the plant
cytoskeleton. Annual Review of Cell and Developmental Biology 21: 271–295.
Sobrizal YA. 2007. Identification of a notched kernel gene associated with pre-
harvest sprouting using Oryza glumaepatula introgression lines in rice.
Indonesian Journal of Agronomy 35:63–67.
Takeda K. 1982. Notched grains developed by the minute genes of rice/
unbalanced growth in floral glumes and caryopsis in rice VI. Japanese Journal of
Breeding 32: 353–364.
Tao Y, Din AMU, An L, Chen H, Li G, Ding Y, Liu Z. 2022. Metabolic
disturbance induced by the embryo contributes to the formation of chalky
endosperm of a notched-belly rice mutant. Frontiers in Plant Science 12:
760597.
Ti SC, Pamula MC, Howes SC, Duellberg C, Cade NI, Kleiner RE, Forth S,
Surrey T, Nogales E, Kapoor TM. 2016. Mutations in human tubulin
proximal to the kinesin-binding site alter dynamic instability at microtubule
plus- and minus-ends. Developmental Cell 37:72–84.
Tischfield MA, Cederquist GY Jr, Engle EC. 2011. Phenotypic spectrum of the
tubulin-related disorders and functional implications of disease-causing
mutations. Current Opinion in Genetics & Development 21: 286–294.
Tong X, Wang Y, Sun A, Bello BK, Ni S, Zhang J. 2018. Notched belly grain 4,a
novel allele of Dwarf 11, regulates grain shape and seed germination in rice
(Oryza sativa L.). International Journal of Molecular Sciences 19: 4069.
Xiong Z, Min S, Kong F, Zhu X. 1986. Genetic analysis of notched grain in rice.
Chinese Journal of Rice Science 1:26–34.
Yu N, Signorile L, Basu S, Ottema S, Lebbink JHG, Leslie K, Smal I, Dekkers
D, Demmers J, Galjart N. 2016. Isolation of functional tubulin dimers and of
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
New Phytologist (2023)
www.newphytologist.com
New
Phytologist Research 13
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

tubulin-associated proteins from mammalian cells. Current Biology 26: 1728–
1736.
Zhan C, Yamamori K, Koide Y, Kishima Y. 2020. Genetically-controlled
morphological abnormality during endosperm maturation at low temperatures
in a rice variety in Hokkaido. In: Report of the Hokkaido Branch, the Japanese
Society of Breeding and Hokkaido Branch, the Crop Science Society of Japan,vol.
61,60–61.
Supporting Information
Additional Supporting Information may be found online in the
Supporting Information section at the end of the article.
Fig. S1 SNG is a pleiotropic gene.
Fig. S2 SNG encodes the a-tubulin TUBA3 in Oryza sativa.
Fig. S3 Structural modeling of SNG and SNG
R422H
.
Fig. S4 Protein levels of SNG and SNG
R422H
in transgenic lines.
Fig. S5 CRISPR-mediated knockout of SNG and SNG
R422H
.
Fig. S6 Measurement of SNG expression in the indicated tissues
with quantitative reverse transcription PCR.
Table S1 Primers used in this study.
Table S2 Segregation ratio in an F
2
population derived from a
cross between wild-type and Sng plants.
Video S1 Microtubule growth in a lemma inner epidermal cell
from the wild-type line expressing an eGFP-SNG fusion protein.
Video S2 Microtubule growth in a lemma inner epidermal cell
from the Sng mutant line expressing an eGFP-SNG
R422H
fusion
protein.
Video S3 Microtubule catastrophe and rescue in a lemma inner
epidermal cell from the wild-type line expressing an eGFP-SNG
fusion protein.
Video S4 Microtubule catastrophe and rescue in a lemma inner
epidermal cell from the Sng mutant line expressing an eGFP-
SNG
R422H
fusion protein.
Please note: Wiley is not responsible for the content or function-
ality of any Supporting Information supplied by the authors. Any
queries (other than missing material) should be directed to the
New Phytologist Central Office.
New Phytologist (2023)
www.newphytologist.com
Ó2023 The Authors
New Phytologist Ó2023 New Phytologist Foundation
Research
New
Phytologist
14
14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19226 by Institute Of Botany, Wiley Online Library on [24/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License