Two rice receptor-like kinases maintain male fertilityunder changing temperaturesJunping Yua, Jiaojiao Hana, Yu-Jin Kima, Ming Songa, Zhen Yanga, Yi Hea, Ruifeng Fua, Zhijing Luoa, Jianping Hub,c,Wanqi Lianga, and Dabing Zhanga,d,e,1

aJoint International Research Laboratory of Metabolic and Developmental Sciences, State Key Laboratory of Hybrid Rice, School of Life Sciences andBiotechnology, Shanghai Jiao Tong University, Shanghai 200240, China; bDepartment of Energy Plant Research Laboratory, Michigan State University, EastLansing, MI 48824; cPlant Biology Department, Michigan State University, East Lansing, MI 48824; dCrop Biotech Institute, Department of Plant MolecularSystems Biotechnology, Kyung Hee University, Yongin 446-701, South Korea; and eSchool of Agriculture, Food and Wine, University of Adelaide,SA 5064, Australia

Edited by George Coupland, Max Planck Institute for Plant Breeding Research, Cologne, Germany, and approved October 6, 2017 (received for review March30, 2017)

Plants employ dynamic molecular networks to control developmentin response to environmental changes, yet the underlying mecha-nisms are largely unknown. Here we report the identification of tworice leucine-rich repeat receptor-like kinases, Thermo-Sensitive GenicMale Sterile 10 (TMS10) and its close homolog TMS10-Like (TMS10L),which redundantly function in the maintenance of the tapetal celllayer and microspore/pollen viability under normal temperatureconditions with TMS10 playing an essential role in higher temper-atures (namely, 28 °C). tms10 displays male sterility under high tem-peratures but male fertility under low temperatures, and thetms10 tms10l double mutant shows complete male sterility underboth high and low temperatures. Biochemical and genetic assaysindicate that the kinase activity conferred by the intracellular do-main of TMS10 is essential for tapetal degeneration and male fer-tility under high temperatures. Furthermore, indica or japonica ricevarieties that contain mutations in TMS10, created by geneticcrosses or genome editing, also exhibit thermo-sensitive genic malesterility. These findings demonstrate that TMS10 and TMS10L act asa key switch in postmeiotic tapetal development and pollen devel-opment by buffering environmental temperature changes, provid-ing insights into the molecular mechanisms by which plants developphenotypic plasticity via genotype–environment temperature inter-action. TMS10 may be used as a genetic resource for the develop-ment of hybrid seed production systems in crops.

temperature | receptor-like kinases | rice | male fertility | hybrid breeding

As sessile organisms, higher plants must adapt to changingenvironmental conditions via physiological and morphologi-

cal responses. Uncovering how dynamic molecular networkscontrol whole-plant morphology and physiology when encounter-ing environmental changes is a key challenge in biology. Recently,H2A.Z nucleosomes were shown to act as conserved thermo-sensors from yeast to plants, which wrap DNA tightly and suppressgene transcription under low temperatures while decreasing theiroccupancy of the promoter regions of target genes to activate genetranscription as the temperature increases (1). In addition, cyclicnucleotide-gated calcium channels were shown to be conservedland-plant thermosensors, which trigger heat-shock response viaCa2+ influx through the plasma membrane in response to mildtemperature increments and lead to transient thermotoleranceagainst heat stress (2, 3). The development of reproductive organsis crucial to crop yield, yet how plants sense and respond totemperature changes to maintain proper development of re-productive organs is poorly understood.The anther is the male organ of higher plants and contains

centrally localized meiotic cells (also called microsporocytes) sur-rounded by four somatic cell layers named the tapetum, middlelayer, endothecium, and epidermis. During organogenesis, andanther formation in particular, higher plants need to mount com-plex and coordinated developmental and physiological eventsto overcome fluctuating environmental conditions (4). Notably,

temperature changes such as heat stress frequently cause prematureprogrammed cell death of the tapetum, resulting in partial orcomplete male sterility (5). The mechanism underlying successfulmale reproductive development during these changes remainslargely unknown. Investigations of several environmentally sensitivegenic male sterile (EGMS) lines in rice (Oryza sativa) revealed thatmutations within a long noncoding RNA (6, 7), a 21-nt phasedsmall-interfering RNA (8), a 21-nt small RNA osa-smR5864w (9),an R2R3MYB transcription regulator (10), and an RNase ZS1 (11)can cause photoperiod-sensitive genic male sterility (PGMS) orthermosensitive genic male sterility (TGMS), traits that have beenused in crop hybrid seed production (12). For practical two-linehybrid rice breeding, a male sterility–fertility transition tempera-ture range of 22–24 °C for current TGMS lines is critical, and ahigher sterility–fertility transition temperature than this tempera-ture range may increase the risk of self-pollination of TGMS linesduring hybrid seed production under fluctuating temperature con-ditions (9, 13). Revealing molecular mechanisms underlying malereproductive development under changing environmental condi-tions will be useful for breeding programs.Multicellular organisms such as plants frequently employ cell-

surface–localized protein receptors to sense and transduce signalsfrom cell to cell. One important class of such receptors is theleucine-rich repeat–receptor-like kinase (LRR–RLK), which typ-ically consists of an extracellular domain for signal perception, atransmembrane domain, and a cytoplasmic kinase domain thattransduces the signal through auto- and transphosphorylation (14,15). In plants, several LRR–RLKs have been shown to be crucialto plant growth, development, pathogen resistance, or cell death(16–20). During early anther development, several LRR–RLKsfromArabidopsis thaliana, rice, andmaize (Zeamays L.) are important

Significance

By affecting male fertility in crops, climate temperature changehas a major impact on global food security. Here we show therole of two rice leucine-rich repeat–receptor-like kinases,TMS10 and TMS10L, which redundantly control male fertilityunder fluctuating temperatures. This finding provides insightsinto how plants overcome adversary temperature changes toachieve normal male fertility and a new genetic resource forcrop hybrid seed production.

Author contributions: W.L. and D.Z. designed research; J.Y., J. Han, and Z.Y. performedresearch; M.S., Y.H., R.F., and Z.L. contributed new reagents/analytic tools; J.Y. analyzeddata; and J.Y., Y.-J.K., J. Hu, and D.Z. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1To whom correspondence should be addressed. Email: zhangdb@sjtu.edu.cn.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1705189114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1705189114 PNAS | November 14, 2017 | vol. 114 | no. 46 | 12327–12332

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

1

in determining the identity and number of tapetal cells andmeiocytes (17, 20, 21).The tapetum is a somatic helper-cell layer next to the micro-

sporocytes that secretes enzymes to help with the release of mi-crospores from tetrads and provides nutrients for pollendevelopment (22). After meiosis, tapetal cells initiate programmedcell death (PCD)-driven degeneration, which is required for pollengrain maturation, as abnormal tapetal enlargement and prematureor delayed degradation of tapetal cells frequently result in malesterility (23). In this study, we report the discovery of two LRR–RLKs, TMS10 and TMS10L, which can redundantly operate intapetal development, with TMS10 specifically preserving thisfunctionality under moderately higher temperatures. tms10 dis-plays male sterility under high temperatures (>24 °C) due to de-fective tapetal development and male fertility under lowtemperatures (22–24 °C), and the tms10 tms10l double mutantshows complete male sterility under low temperatures as well.These findings provide insights into how plants overcome adversetemperature changes to ensure normal male reproductive devel-opment as well as a genetic resource for the potential establish-ment of a hybrid seed production system in crops.

Resultstms10 Is a Temperature-Sensitive Male Sterile Mutant. To identifyregulators of rice male fertility during changing temperatures, weisolated a 60Co γ-ray radiation-mutagenized TGMS mutant in ssp.japonica cv. 9522 for its complete male sterility at restrictive tem-peratures (>24 °C) and named it thermo-sensitive genic male sterility10 (tms10) because it was the 10th TGMS locus identified in rice(24). Under high temperatures (25–32 °C) in the paddy field duringsummer in Shanghai, tms10 and wild type (WT) exhibited normalvegetative growth and flower development (SI Appendix, Fig. S1 A–E). However, tms10 had white and smaller anthers that do notproduce viable pollen grains, as revealed by I2-KI staining (Fig. 1

A–E and SI Appendix, Fig. S1 F and G), and exhibited completemale sterility. Interestingly, under lower temperatures (22–24 °C),male fertility in tms10 was partially recovered (Fig. 1 A, C, D, andF). All F1 progenies from reciprocal crosses (tms10 under hightemperature as the pollen acceptor, and WT as the pollen donor)were fertile, which indicated that female development was com-pletely normal in the mutant (SI Appendix, Fig. S1H and I). The F2plants segregated as 95 male fertile and 35 male sterile (χ2 =0.256 for 3:1, P > 0.05) under high temperature conditions, sug-gesting that a single recessive mutation is responsible for the malesterile phenotype in tms10.To examine whether both temperature and photoperiod impact

male fertility in tms10, we treated mutants grown in the growthchamber with 12 different temperature–photoperiod combinations(SI Appendix, Table S1). Consistent with the field observations, atan average temperature of 23–24 °C, tms10 was partially fertilewhile WT was fully fertile (SI Appendix, Fig. S1 K and L). Whenthe average temperature was raised to 28 °C, tms10 lacked viablepollen grains and was completely male sterile (SI Appendix, Fig.S1K). At high temperatures, tms10 was male sterile with no ma-ture pollen formation under either long-day or short-day condi-tions, and treatments with various photoperiodic lengths had littleeffect on tms10’s male fertility (SI Appendix, Fig. S1K), suggestingthat tms10 is sensitive to temperature changes but does not havean obvious response to photoperiod.

TMS10 Is Required for Tapetum Development and Microspore Formation.Transverse section analysis demonstrated that, at an average tem-perature of 28 °C, tms10 had enlarged tapetal cells and displayedaborted pollen development at stage 9 of anther development, afterthe microspores were released from the tetrads (Fig. 1 G–I). Thissuggested that the deficiency caused by the tms10 mutation occursat the postmeiotic stage, a finding that differs from previouslyreported TGMS mutants such as tms5 and PA64S that showeddefects during the formation of microspore mother cells (MMCs)(9, 11). In support of this, 4′,6-diamidino-2-phenylindole stainingfor the observation of chromosome behaviors confirmed that tms10had a normal meiotic process under 28 °C, as evidenced by thenormal formation of tetrads that each contained four young mi-crospores (SI Appendix, Fig. S1J).Transmission electron microscopy (TEM) showed that, at

stages 8b and 9, WT anthers formed tetrads, released the freemicrospores, and had a condensed and degenerated tapetal layerwith abundant lipidic granules (Ubisch bodies) attached to theirinner surface (Fig. 1J and SI Appendix, Fig. S2 A, C, E, G, I, K, M,and O). In contrast, tapetal cells in tms10 exhibited no obviousdegeneration and became highly vacuolated, containing no visibleUbisch bodies and showing a multinuclei cell cluster pattern atstages 8b and 9 under high temperature (Fig. 1 K and L and SIAppendix, Fig. S2 D, H, L, and P). At these stages, microspores oftms10 seemed to develop normally, but lacked primexine, astructure of the template for pollen wall formation (SI Appendix,Fig. S2 B, F, J, and N). Further analysis using scanning electronmicroscope revealed that, under the average temperature of 28 °C,tms10 had less wax and cutin deposition on anther epidermis,fewer Ubisch bodies on the inner surface of tapetal cells, and lessaccumulation of sporopollenin on the surface of microsporescompared with the WT (SI Appendix, Fig. S2Q). These observa-tions suggested that the TMS10 gene product plays an essentialrole in both tapetal degeneration and pollen wall formation.

TMS10 Encodes an LRR–RLK Highly Expressed in the Anther. To clonethe TMS10 gene, we collected 281 male sterile individuals fromthe F2 mapping population by crossing the mutant with O. sativassp. indica cv. 9311. Map-based cloning revealed that the tms10locus was located within three bacterial artificial chromosomeclones on chromosome 2 (Fig. 2A). Sequence analysis revealed a7-bp deletion in the fourth exon (Fig. 2B) of a putative LRR–RLK

A G

D

J

B

E

F

H K

C I L

Fig. 1. tms10 displays temperature-dependent male fertility. (A–C) Spikelets atstage 13 after the removal of half of lemma and palea. (Scale bars, 2 mm.) (D–F)Anthers that contain pollen grains stained by I2-KI. (Scale bars, 200 μm.) (G–I)Semithin section analysis at stage 9, when microspores were released from tet-rads. (Scale bars, 20 μm.) (J–L) TEM at stage 9. Shown are the anther wall layerswith a focus on tapetal cells. (Scale bars, 5 μm.) DMs, degenerated microspores;E, epidermis; En, endothecium; Ml, middle layer; Ms, microspore; T, tapetum.

12328 | www.pnas.org/cgi/doi/10.1073/pnas.1705189114 Yu et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

1

(LOC_Os02g18320, annotated in www.gramene.org/), which causeda frame shift that resulted in the pretermination of proteintranslation (Fig. 2C). To confirm the identity of this gene, wetransformed the full-length genomic DNA of the putative TMS10gene—which included its 3,031-bp native promoter, 5,804-bpcoding region, and 400-bp 3′-untranslated region (UTR)(pTMS10::TMS10gDNA)—into the tms10 mutant for complemen-tation. All 50 transgenic plants that expressed the putative TMS10gene showed rescue of the mutant phenotype at 28 °C (Fig. 3 Dand G), confirming that the mutation in LOC_Os02g18320 is re-sponsible for the male sterile phenotype in tms10.To understand the evolution of TMS10 in plants, we performed a

phylogenetic analysis using the full-length TMS10 protein sequenceand 43 homologous sequences retrieved from 10 plant species frompublic databases. TMS10 was found in a clade that contained bothdicot and monocot proteins with no reported functions (26) (Fig.2E), suggesting that TMS10 may play a conserved role in dicots andmonocots during male reproduction. In addition to the TMS10clade, members of the clade containing SERKs are known to beinvolved in brassinosteroid (BR) signaling, pathogen-related de-fense, or male organ development (27–30), and members in theNSP-INTERACTING KINASE (NIK) clade were reported toconfer resistance to geminivirus infection (31). Arabidopsis SERK1/2(32, 33) and rice MULTIPLE SPOROCYTE 1 (MSP1) (25) wereshown to specify tapetal cell identity, which differs from TMS10’sobvious function in tapetal degeneration, suggesting that TMS10evolved a new function in anther development after the function ofMSP1 was established. These data implied that TMS10 functions ina pathway that controls the TGMS trait in rice and that this pathwaymay be conserved in monocots and dicots.Reverse transcription–quantitative PCR (RT-qPCR) was

conducted to analyze the spatiotemporal expression pattern of

TMS10 and showed that TMS10 is expressed mainly in the earlystages of anther and leaf development (SI Appendix, Fig. S3A).In situ hybridization analysis revealed TMS10 expression in al-most all layers of the anther wall and MMCs at stage 5. At stages7 and 9, the transcripts of TMS10 were detectable in antherwall layers and MMCs/microspores and conjunction tissues,with particularly strong signals in tapetal cells (SI Appendix, Fig.S4F). Consistently, in tms10 lines complemented by pTMS10::TMS10gDNA-GUS, β-glucuronidase (GUS) signals were de-tected in anthers from early anther primordium formation tostage 9 (SI Appendix, Fig. S4C). Confocal microscopic analysis oftms10 lines complemented by pTMS10::TMS10gDNA-GFP alsorevealed TMS10 expression in both anther sac and conjunctiontissues, but not in meiotic cells or microspores (Fig. 2D and SIAppendix, Fig. S4G). Inconsistent results in TMS10 transcriptswere detected in MMCs/microspores by in situ assay, but theTMS10-GFP fused protein was undetectable in MMCs/micro-spores, which indicated that there might be some posttranscrip-tional regulation to inhibit TMS10 protein formation in MMCs/microspores. Immunoblot analysis of the complemented tms10lines that expressed pTMS10::TMS10gDNA-GFP or pTMS10::TMS10gDNA-9×myc showed that the TMS10 protein level inthese lines under 22 °C had no obvious changes from that inplants under 28 °C (SI Appendix, Fig. S4 D and E). The obser-vation that no obvious difference in TMS10 transcript (SI Ap-pendix, Fig. S4 A and F) or protein levels (SI Appendix, Fig. S4C–E and G) in anthers between plants grown at 22 °C and 28 °Csuggested that TMS10 expression and protein accumulationwere not significantly affected by the tested temperatures. Thedata indicating that TMS10 is expressed in anther tissue, par-ticularly in tapetal cells, support the function of TMS10 in antherdevelopment.

A E

B

C

D

Fig. 2. Molecular identification, localization, and phylogenetic analysis of TMS10. (A) Map-based cloning of TMS10. (B) Gene structure of TMS10. Black boxesare exons, gray boxes represent the 5′ and 3′ UTRs, and lines between boxes are introns. Red arrow indicates the mutation site in tms10. (C) Genomic andprotein sequences around the mutation sites in tms10. Nucleotides in red are deleted in tms10. Amino acids in blue are the changed residues in tms10.Asterisk, stop codon. (D) Localization of the TMS10-GFP protein in anther sacs at stage 8. (Scale bars, 25 μm.) (E) Phylogenetic analysis of TMS10 and itshomologs in monocots and dicots. MSP1, a rice LRR–RLK involved in male reproductive development (25), was included in this phylogeny as an outgroup. TheSERK and NIK subclades include only the members from Arabidopsis (At) and rice (Os) species. The TMS10 clade contains homologs from nine plant speciesincluding Triticum aestivum (Ta), Hordeum vulgar (Hv), Brachypodium distachyon (Bd), Z. mays (Zm), Sorghum bicolor (Sb), Medicago truncatula (Mt), Populustrichocarpa (Pt), Arabidopsis lyrata (Al), and Picea sitchensis (Ps).

Yu et al. PNAS | November 14, 2017 | vol. 114 | no. 46 | 12329

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

1

TMS10 Is Localized to the Plasma Membrane and Has Kinase Activity.TMS10 encodes a 607-aa LRR–RLK (34), which has four LRRsin its extracellular region, a single transmembrane domain, and acytoplasmic region that contains the juxtamembrane, kinase, andC-terminal domains (Fig. 3A). TMS10 is localized mainly to theplasma membrane, as demonstrated by confocal microscopicanalysis of stably or transiently expressed TMS10-GFP fusionprotein in rice anther, rice protoplasts, tobacco leaves, and onionepidermis cells (Fig. 2D and SI Appendix, Fig. S3 B and C).Furthermore, we compared the localization of transientlyexpressed TMS10-GFP in tobacco leaves under 22 °C and 28 °Cand confirmed that the protein’s plasma membrane localizationwas not affected by temperature changes (SI Appendix, Fig. S4B).To determine whether TMS10 has kinase activity, we used

recombinant proteins that contained the TMS10 kinase domain (K),kinase with C-terminal domain (KC), and cytoplasmic region (JKC)to conduct in vitro kinase assays. JKC can phosphorylate a universalsubstrate, myelin basic protein (MBP), and this kinase activity isdependent on TMS10’s juxtamembrane domain (Fig. 3 A–C). Se-quence alignment showed that the kinase domain is highly conservedin TMS10 and its homologs in rice (TMS10L) and Arabidopsis (SIAppendix, Fig. S5A). Mutation of K312 (JKCK312E, mJKC), a con-served residue at the ATP-binding site, abolished the phosphoryla-tion activity of TMS10 in vitro (Fig. 3 A and B), demonstrating thecritical role of this conserved residue in TMS10’s kinase activity. Totest the biological significance of K312 in TMS10, mTMS10-9×myc(pTMS10::TMS10gDNAK312E-9×myc) was introduced into tms10.TMS10-9×myc (pTMS10::TMS10gDNA-9×myc) can rescue antherfertility in tms10 under high temperature (Fig. 3 E andH). However,mTMS10-9×myc failed to rescue tms10’s male sterility under hightemperature in all 35 transgenic lines (Fig. 3 F and I), suggesting that

the kinase activity of TMS10 is required for controlling tapetalfunction and pollen formation under high temperatures. Massspectrometry (MS) analysis of bacterially expressed recombinantJKC and mJKC revealed that, of the 13 putative phosphorylationsites, T392, T576, Y458, and S464 are auto-phosphorylated inrecombinant JKC but not in mJKC (SI Appendix, Table S2), sup-porting the auto-phosphorylation of TMS10. These results suggestthat K312 and the juxtamembrane domain of TMS10 are essential forits kinase activity and that T392, T576, Y458, and S464 are the potentialphosphorylation sites for TMS10.

TMS10L and TMS10 Play Redundant Roles in Anther DevelopmentUnder Low Temperature. The rice genome has a close homologof TMS10 within the TMS10 clade (TMS10L, LOC_Os03g49620,Fig. 2E), which shares a similar domain structure and 88% aminoacid sequence identity with TMS10. TMS10L has a predictedextracellular domain with four LRRs, one transmembrane do-main, and an intracellular kinase domain that also contains theconserved ATP activation loop and especially the conserved K310

(Figs. 3A and 4A and SI Appendix, Fig. S5A). To determine thefunction of TMS10L, we used the CRISPR-Cas9 system to createtwo tms10l alleles that contain mutations within the second exon:tms10l-1 has a 17-bp deletion that leads to a pretranslationaltermination in the LRR domain, and tms10l-2 has an 18-bp de-letion that causes a 6-aa deletion in the LRR domain (Fig. 4A).The double mutants tms10 tms10l-1 and tms10 tms10l-2 werelater generated by genetic crosses.Both tms10l-1 and tms10l-2 showed normal fertility under high

and low temperatures (Fig. 4 B and C). However, tms10 tms10l-1and tms10 tms10l-2 double mutants were male sterile under bothhigh and low temperatures (Fig. 4 B and C). Semithin section

Fig. 3. TMS10 has in vitro kinase activity required for anther fertility. (A) TMS10 protein structure. Green, blue, and purple boxes represent LRR, trans-membrane (TM), and kinase (K) domains, respectively. KC, kinase domain plus C terminus (amino acids 284–607). JKC, intracellular domain (amino acids 241–607) including the juxtamembrane (J), kinase (K), and C-terminal domains (C). mJKC, kinase-dead intracellular domain (amino acids 241–607 with K312E).(B and C) In vitro kinase assays of TMS10. Derivatives (Left) and Coomassie Brilliant Blue staining (Right) in B; derivatives (Top) and Coomassie Brilliant Bluestaining (Bottom) in C. MBP, myelin basic protein. BRI1-phosphorylated MBP used as the positive control. (D–I) Complementary of tms10 at 28 °C. Comparedwith rescued anther fertility by TMS10 (pTMS10::TMS10gDNA) or TMS10-9×myc (pTMS10::TMS10gDNA-9×myc) in tms10 (D, E, G, and H), mTMS10-9×myc(pTMS10::TMS10gDNAK312E-9×myc) in tms10 failed to rescue male fertility (F and I). (Scale bars, 1 mm in D–F; 200 μm in G–I.) (J) Immunoblot analysisto qualitatively show the expression level of TMS10-9×myc and mTMS10-9×myc in tms10 in plants containing the pTMS10::TMS10gDNA-9×myc andpTMS10::TMS10gDNAK312E-9×myc transgenes growing under 28 °C.

12330 | www.pnas.org/cgi/doi/10.1073/pnas.1705189114 Yu et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

1

analysis confirmed that double mutants grown at 22 °C had de-fective tapetal degeneration and collapsed microspores in an-ther development (SI Appendix, Fig. S6), suggesting that bothTMS10 and TMS10L are involved in anther development underlow temperatures. RT-qPCR analysis showed that, in WT antherat stages 8b and 9, TMS10L transcript levels were three- to-fivefold higher at 22 °C than at 28 °C (SI Appendix, Fig. S5C),suggesting that TMS10L may play an important role in antherdevelopment under lower temperature. Additionally, in tms10anthers, the mRNA levels of TMS10L were reduced at these twotemperatures compared with the WT (SI Appendix, Fig. S5C),implying that TMS10 affects the expression of TMS10L via a yet-unknown pathway. Furthermore, transient expression analysisshowed that TMS10L-YFP was located on plasma membrane intobacco leaves under 28 °C or 22 °C (SI Appendix, Fig. S5D),similar to that of TMS10-GFP, suggesting that the temperaturechange does not obviously affect the protein localization of bothTMS10L and TMS10. Given that LRR–RLKs normally form di-mers to perform their biological function (35, 36), we testedwhether TMS10 and TMS10L are able to form hetero- orhomodimers. Our DUALmembrane yeast two-hybrid assays didnot reveal interactions between TMS10 and TMS10L (SI Appen-dix, Fig. S5B). Taken together, our data suggest that TMS10 andTMS10L may play redundant roles in anther development underlow temperatures.

tms10 Is Applicable to Hybrid Seed Production in Both Indica andJaponica Rice Varieties. Previous studies showed that the samemutation in a noncoding RNA, named long-day–specific male-fertility–associated RNA, caused different EGMS phenotypes in adifferent rice background, i.e., photoperiod-dependent male ste-rility in japonica cv. NK58S but thermal-dependent male steril-ity in indica cv. PA64S (7, 9). To test whether the tms10 mutationcan cause the TGMS trait in indica subspecies, the dominant

subspecies used for current hybrid seed technology, we introducedthe tms10 mutation to ssp. indica cv. HuangHuaZhan (HHZ).After twice backcrosses followed by three times self-pollinations,BC2F4 plants containing the tms10 mutation and HHZ charac-teristics (i.e., longer palea and lemma) exhibited the TGMS trait(SI Appendix, Fig. S7 D–F). Finally, we used the CRISPR-Cas9 genome-editing system to create mutations in the secondexon of TMS10 in japonica cv. KY131 (37) and obtained fourtms10 alleles (tms10-2/-3/-4/-5), all of which displayed the TGMStrait (SI Appendix, Fig. S7 A–C). We thus concluded that tms10has a consistent TGMS trait in both indica and japonica cultivars.In past decades, PGMS and TGMS lines have been widely used

for two-line hybrid breeding in rice (38). To test the applicabilityof tms10-derived cultivars to two-line hybrid seed production, themale-sterile tms10 lines in cv. 9522 and HHZ backgrounds underhigh temperature conditions were, respectively, pollinated withpollen grains from a widely compatible restorer line JP69. The F1hybrid generation had an about 80% seed setting and displayedhybrid vigor, including increases in the number of tillers, grains perinflorescence, and 1,000-grain weight (SI Appendix, Fig. S7 G–P).These results together suggested that tms10 is a TGMS traitsuitable for the development of hybrid rice seed production sys-tems in both indica and japonica subspecies.

DiscussionRice is cultivated in many countries. However, temperature hasbeen a limiting factor in achieving high yield. Global climatewarming generally has negative effects on the length of both veg-etative and reproductive growth periods and on rice yield. Pollenformation is highly sensitive to temperature stress during flowering,which therefore poses a serious threat to current and long-termcrop yields (3, 39, 40). Our work identified two homologous LRR–RLKs, TMS10 and TMS10L, which appear to redundantly bufferthe adverse effect of fluctuating temperatures on rice male devel-opment. TMS10 (but not TMS10L) is essential to the maintenanceof normal male fertility under higher temperatures (i.e., 28 °C),while TMS10L together with TMS10 may play an important role inmale fertility control under lower temperatures. These findingsprovide an understanding of how plants develop male organs viagenotype–environment temperature interactions.The use of hybrid plants has significantly transformed agricul-

tural practices and seed production in the past decades (12, 38). Inthe efforts to increase crop yield, male sterility has played a centralrole in cytoplasmic male sterility (CMS) and photoperiod-sensitive/thermosensitive genic male sterile (P/TGMS) in hybrid breedingprograms (12, 38). Compared with traditional CMS-based three-line hybrid breeding, two-line hybrid breeding is based on nuclearencoded genes, thus widening the choice of restorer lines in pro-ducing F1 hybrids. Discovering new resources and mechanisms willprovide additional tools to implement P/TGMS traits in hybridseed production. To date, nine types of TGMS lines have beenisolated, several of which have been widely used in two-line hybridbreeding (24). For example, PA64S was shown to be caused by anucleotide mutation in a 21-nt small RNA osa-smR5864w (9). TheTGMS line tms5 is controlled by a mutation in RNase ZS1 that isinvolved in UbL40 mRNA degradation (11). In this study, weidentified tms10 mutation(s) as TGMS alleles and discovered thatTMS10 encodes a LRR–RLK with its kinase activity essential foranther development and fertility under high environmental tem-peratures. Anther phenotypes start to show in tms10 after meiosis,and therefore at a stage later than that in tms5 (11), PA64S (9), orUGPase-suppressed lines (41). TMS10 regulates tapetal de-generation and pollen development, which is different from pre-viously reported LRR–RLKs that function in male reproduction,all of which are involved in tapetal cell identity specification duringanther development (25, 31, 32).Our results suggested that, through its kinase activity, TMS10

plays a major role in tapetal cell degeneration and pollen

A

B

C

Fig. 4. TMS10L and TMS10 play redundant roles in anther development andpollen fertility under low temperature. (A) The second exon of TMS10L wastargeted by CRISPR-Cas9 to generate tms10l-1 and tms10l-2. Nucleotides inblue indicate the target site and in red the changed sites, and dots indicatedeletions. K, kinase domain; LRR, leucine-rich repeat domain; TM, trans-membrane domain. (B and C) Anther phenotype and fertility of tms10l-1,tms10l-2, tms10 tms10l-1, and tms10 tms10l-2 mutants under high temper-ature (28 °C) (B) and low temperature (22 °C) (C). (Scale bars for flowers,1 mm; for pollen grains, 200 μm.)

Yu et al. PNAS | November 14, 2017 | vol. 114 | no. 46 | 12331

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

1

formation under high temperatures. Due to functional redundancy,LRR–RLKs coordinately participate in many developmental pro-cesses. For example, AtSERK1 and AtSERK2 are functionallyredundant in regulating early tapetal-layer formation in Arabidopsis(32, 33), and their extracellular LRR domain has specific functionsthat cannot be replaced by that of AtSERK3 although the kinasedomain of these three SERKs are interchangeable (42). The ob-servation that under low temperatures tms10 is male-fertile whiletms10 tms10l is male-sterile prompts us to speculate that TMS10Lfunctions redundantly with TMS10 under low temperature duringanther development and pollen formation. LRR–RLKs normallydimerize to transphosphorylate (18, 35). However, TMS10 andTMS10L do not interact directly in our assays, so whether and howthey function together to regulate anther development under lowtemperature remains to be elucidated. In the BR signal trans-duction cascade, BR signal is perceived by the heterodimer of twoLRR–RLKs, BR-INSENSITIVE1 (BRI1) that has 25 LRRs andits coreceptor BRI1-ASSOCIATED RECEPTOR KINASE1(BAK1) that contains 5 LRRs (36, 43). It will be exciting to revealthe partner(s) of TMS10 and/or TMS10L, which could be un-known LRR–RLKs with longer LRRs and capable of forming aheterodimer with TMS10 and/or TMS10L to transphosphorylateand magnify the signals in promoting tapetal PCD and normalpollen formation. Identifying components of the TMS10- andTMS10L-mediated signaling pathway in the thermosensitivecontrol of rice male fertility will uncover how LRR–RLKs buffer

the adverse effect of changing temperatures on male develop-mental programs in plants. Finally, the finding that tms10 confersthe TGMS trait in both japonica and indica cultivars can help towiden the genetic resources for manipulating male fertility, a keyfactor in hybrid seed production.

Materials and MethodsDetails of materials and methods are provided by SI Appendix, SI Materialsand Methods, including plant materials and growth conditions, plasmidconstruction and transformation, phenotypic analysis of anther and GUSstaining, phylogenetic analysis, RT-qPCR analysis, in situ hybridization, yeasttwo-hybrid (Y2H) assays, subcellular localization, protein expression, purifi-cation and immunoblot analysis, in vitro kinase assays, and MS analysis.

ACKNOWLEDGMENTS. We thank Prof. Malcolm J. Bennett for commentingon and editing the manuscript; Professor Hongwei Xue and Dr. Shutang Tanfor advice on kinase analysis; Dr. Huamin Si and Dr. Guocheng Hu forproviding assistance in testing the relationship between temperature andmale fertility in tms10 in the growth chamber; and Prof. Laigeng Li andDr. Jinshan Gui for assistance with the DUALmembrane yeast system. Thisresearch was supported by the National Natural Science Foundation of China(Grants 31430009, 31322040, and 31271698); the National Key Research and De-velopment Program of China (Grants 2016YFD0100804 and 2016YFE0101000);National Key Basic Research Development Program of the Ministry of Scienceand Technology, China (Grant 2013 CB126902); the Innovative Research Team,Ministry of Education, and 111 Project (Grant B14016); the Science and Technol-ogy Commission of Shanghai Municipality (Grant 13JC1408200); Australia-ChinaScience and Research Fund Joint Research Centre (Grant ACSRF48187); and theNational Transgenic Major Program (Grant 2016ZX08009003-003-007).

1. Kumar SV, Wigge PA (2010) H2A.Z-containing nucleosomes mediate the thermo-sensory response in Arabidopsis. Cell 140:136–147.

2. Finka A, Cuendet AF, Maathuis FJ, Saidi Y, Goloubinoff P (2012) Plasma membranecyclic nucleotide gated calcium channels control land plant thermal sensing and ac-quired thermotolerance. Plant Cell 24:3333–3348.

3. Bita CE, Gerats T (2013) Plant tolerance to high temperature in a changing environ-ment: Scientific fundamentals and production of heat stress-tolerant crops. FrontPlant Sci 4:273.

4. Zhang D, Luo X, Zhu L (2011) Cytological analysis and genetic control of rice antherdevelopment. J Genet Genomics 38:379–390.

5. De Storme N, Geelen D (2014) The impact of environmental stress on male re-productive development in plants: Biological processes and molecular mechanisms.Plant Cell Environ 37:1–18.

6. Ding J, et al. (2012) RNA-directed DNA methylation is involved in regulating photo-period-sensitive male sterility in rice. Mol Plant 5:1210–1216.

7. Ding J, et al. (2012) A long noncoding RNA regulates photoperiod-sensitive malesterility, an essential component of hybrid rice. Proc Natl Acad Sci USA 109:2654–2659.

8. Fan Y, et al. (2016) PMS1T, producing phased small-interfering RNAs, regulatesphotoperiod-sensitive male sterility in rice. Proc Natl Acad Sci USA 113:15144–15149.

9. Zhou H, et al. (2012) Photoperiod- and thermo-sensitive genic male sterility in rice arecaused by a point mutation in a novel noncoding RNA that produces a small RNA. CellRes 22:649–660.

10. Zhang H, et al. (2013) Mutation in CSA creates a new photoperiod-sensitive genic malesterile line applicable for hybrid rice seed production. Proc Natl Acad Sci USA 110:76–81.

11. Zhou H, et al. (2014) RNase Z(S1) processes UbL40 mRNAs and controls thermosensitivegenic male sterility in rice. Nat Commun 5:4884.

12. Zhang DB, Liang WQ (2016) Improving food security: Using male fertility for hybridseed breeding. Science Suppl:45–48.

13. Li SQ, Yang DC, Zhu YG (2007) Characterization and use of male sterility in hybrid ricebreeding. J Integr Plant Biol 49:791–804.

14. Torii KU (2004) Leucine-rich repeat receptor kinases in plants: Structure, function, andsignal transduction pathways. Int Rev Cytol 234:1–46.

15. Wang X, et al. (2008) Sequential transphosphorylation of the BRI1/BAK1 receptorkinase complex impacts early events in brassinosteroid signaling. Dev Cell 15:220–235.

16. Shiu SH, Bleecker AB (2001) Receptor-like kinases from Arabidopsis form a mono-phyletic gene family related to animal receptor kinases. Proc Natl Acad Sci USA 98:10763–10768.

17. Zhao D (2009) Control of anther cell differentiation: A teamwork of receptor-likekinases. Sex Plant Reprod 22:221–228.

18. Clouse SD (2011) Brassinosteroid signal transduction: From receptor kinase activationto transcriptional networks regulating plant development. Plant Cell 23:1219–1230.

19. Osakabe Y, Yamaguchi-Shinozaki K, Shinozaki K, Tran LS (2013) Sensing the envi-ronment: Key roles of membrane-localized kinases in plant perception and responseto abiotic stress. J Exp Bot 64:445–458.

20. Zhang D, Yang L (2014) Specification of tapetum and microsporocyte cells within theanther. Curr Opin Plant Biol 17:49–55.

21. Feng X, Zilberman D, Dickinson H (2013) A conversation across generations: Soma-germ cell crosstalk in plants. Dev Cell 24:215–225.

22. Shi J, Cui M, Yang L, Kim YJ, Zhang D (2015) Genetic and biochemical mechanisms ofpollen wall development. Trends Plant Sci 20:741–753.

23. Li N, et al. (2006) The rice tapetum degeneration retardation gene is required for

tapetum degradation and anther development. Plant Cell 18:2999–3014.24. Qi Y, et al. (2014) Fine mapping and candidate gene analysis of the novel thermo-

sensitive genic male sterility tms9-1 gene in rice. Theor Appl Genet 127:1173–1182.25. Nonomura K, et al. (2003) The MSP1 gene is necessary to restrict the number of cells

entering into male and female sporogenesis and to initiate anther wall formation in

rice. Plant Cell 15:1728–1739.26. Sasaki G, et al. (2007) Multiple receptor-like kinase cDNAs from liverwort Marchantia

polymorpha and two charophycean green algae, Closterium ehrenbergii and Nitella

axillaris: Extensive gene duplications and gene shufflings in the early evolution ofstreptophytes. Gene 401:135–144.

27. Li D, et al. (2009) Engineering OsBAK1 gene as a molecular tool to improve rice ar-

chitecture for high yield. Plant Biotechnol J 7:791–806.28. Gou X, et al. (2012) Genetic evidence for an indispensable role of somatic embryo-

genesis receptor kinases in brassinosteroid signaling. PLoS Genet 8:e1002452.29. Chen X, et al. (2014) An XA21-associated kinase (OsSERK2) regulates immunity me-

diated by the XA21 and XA3 immune receptors. Mol Plant 7:874–892.30. Zuo S, et al. (2014) OsSERK1 regulates rice development but not immunity to Xanthomonas

oryzae pv. oryzae or Magnaporthe oryzae. J Integr Plant Biol 56:1179–1192.31. Santos AA, Lopes KVG, Apfata JA, Fontes EPB (2010) NSP-interacting kinase, NIK: A

transducer of plant defence signalling. J Exp Bot 61:3839–3845.32. Colcombet J, Boisson-Dernier A, Ros-Palau R, Vera CE, Schroeder JI (2005) Arabidopsis

SOMATIC EMBRYOGENESIS RECEPTOR KINASES1 and 2 are essential for tapetum

development and microspore maturation. Plant Cell 17:3350–3361.33. Albrecht C, Russinova E, Hecht V, Baaijens E, de Vries S (2005) The Arabidopsis

thaliana SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASES1 and 2 control male

sporogenesis. Plant Cell 17:3337–3349.34. Shiu SH, et al. (2004) Comparative analysis of the receptor-like kinase family in Ara-

bidopsis and rice. Plant Cell 16:1220–1234.35. He K, Xu S, Li J (2013) BAK1 directly regulates brassinosteroid perception and

BRI1 activation. J Integr Plant Biol 55:1264–1270.36. Zhang Z, Thomma BP (2013) Structure-function aspects of extracellular leucine-rich

repeat-containing cell surface receptors in plants. J Integr Plant Biol 55:1212–1223.37. Zhang H, et al. (2014) The CRISPR/Cas9 system produces specific and homozygous

targeted gene editing in rice in one generation. Plant Biotechnol J 12:797–807.38. Chen L, Liu YG (2014) Male sterility and fertility restoration in crops. Annu Rev Plant

Biol 65:579–606.39. Sakata T, et al. (2014) Reduction of gibberellin by low temperature disrupts pollen

development in rice. Plant Physiol 164:2011–2019.40. Peng S, et al. (2004) Rice yields decline with higher night temperature from global

warming. Proc Natl Acad Sci USA 101:9971–9975.41. Chen R, et al. (2007) Rice UDP-glucose pyrophosphorylase1 is essential for pollen

callose deposition and its cosuppression results in a new type of thermosensitive genic

male sterility. Plant Cell 19:847–861.42. Aan den Toorn M, Albrecht C, de Vries S (2015) On the origin of SERKs: Bioinformatics

analysis of the somatic embryogenesis receptor kinases. Mol Plant 8:762–782.43. Wang ZY, Bai MY, Oh E, Zhu JY (2012) Brassinosteroid signaling network and regu-

lation of photomorphogenesis. Annu Rev Genet 46:701–724.

12332 | www.pnas.org/cgi/doi/10.1073/pnas.1705189114 Yu et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

1