The protein kinase Pstol1 from traditional rice confers tole

LETTER

doi:10.1038/nature11346

The protein kinase Pstol1from traditional rice confers tolerance of phosphorus deficiency

Rico Gamuyao 1,Joong Hyoun Chin 1,Juan Pariasca-Tanaka 2,Paolo Pesaresi 3,Sheryl Catausan 1,Cheryl Dalid 1,Inez Slamet-Loedin 1,Evelyn Mae Tecson-Mendoza 4,Matthias Wissuwa 2&Sigrid Heuer 1

As an essential macroelement for all living cells,phosphorus is indispensable in agricultural production systems.Natural phos-phorus reserves are limited 1,and it is therefore important to develop phosphorus-efficient crops.A major quantitative trait locus for phosphorus-deficiency tolerance,Pup1,was identified in the traditional aus -type rice variety Kasalath about a decade ago 2,3.However,its functional mechanism remained elusive 4,5until the locus was sequenced,showing the presence of a Pup1-specific protein kinase gene 6,which we have named phosphorus-starvation tolerance 1(PSTOL1).This gene is absent from the rice reference genome and other phosphorus-starvation-intolerant modern varieties 7,8.Here we show that overexpression of PSTOL1in such varieties significantly enhances grain yield in phosphorus-deficient soil.Further analyses show that PSTOL1acts as an enhancer of early root growth,thereby enabling plants to acquire more phosphorus and other nutrients.The absence of PSTOL1and other genes—for example,the submergence-tolerance gene SUB1A —from modern rice varieties underlines the importance

of conserving and exploring traditional germplasm.Intro-gression of this quantitative trait locus into locally adapted rice varieties in Asia and Africa is expected to considerably enhance productivity under low phosphorus conditions.

Phosphorus (P)is of unequivocal importance for the production of food crops,and the demand for P fertilizer is increasing worldwide.In Asia,where rice is the main and sometimes the only source of calories,40%of the rice is produced in rain-fed systems,with little or no water control and frequent occurrence of floods,droughts and other calamities.In addition,60%(29Mha)of the rain-fed lowland rice is produced on poor and problem soils 9,10(Fig.1a)that are constrained by a multitude of abiotic stresses and are naturally low in phosphorus or P fixing.Rice yields are therefore low 11and,not surprisingly,poverty in these regions is among the highest in the world (a5b02cc052d380eb63946da9/web/guest/region).A lack of resources or limited access to P fertilizer are some of the constraints for poor farmers.There is a high risk that the situation will be further aggravated given that phosphate rock,the source of P fertilizer,is a finite and non-renewable resource that is

1

International Rice Research Institute (IRRI),DAPO Box 7777Metro,Manila 1301,Philippines.2Japan International Research Center for Agricultural Sciences (JIRCAS),1-1Ohwashi,Tsukuba 305-8686,

Japan.3Dipartimento di Bioscienze,Universita

`degli studi di Milano,20133Milano,Italy.4University of the Philippines,Los Ban ?os,Laguna 4031,Philippines.India Nepal

Bangladesh

Vietnam Thailand

Cambodia

Burma Indonesia

Malaysia

Laos

Origin of aus -type

rice in India a

+Pup1

–Pup1

W i l d t y p e

s t n 7 s t n 8

–P s t o l 1r e c

+P s t o l 1r e c

+P s t o l 1r e c / kDa 46 25 46 25

LHCII

LHCII e c

OsPupK04-1

OsPupK05-1OsPupK20-2OsPupK29-1

PSTOL1GAPDH

OsPupK20-2

PSTOL1GAPDH

N u l l

35S :: P S T O L 1

I R 74

I R 74-P u p 1

f

OsPupK04-1/OsPupK05-1

(fatty acid DOX/hypoth. protein)OsPupK20-2 (dirigent-like) OsPupK29-1 (hypoth. protein) PSTOL1(protein kinase)

Intolerant (Nipponbare; 145 kb)

Indel (~90 kb)

b

Tolerant (Kasalath; 278 kb)

d

1

2

2

4 6 8 0 1

2

1

2

F o l d c h a n g e u n d e r –P

Absent

–Pup1

+Pup1

OsPupK04-1 OsPupK05-1OsPupK20-2OsPupK29-1PSTOL1F o l d c h a n g e u n d e r –P

Pup1+–+–

+–

+–

Root

Shoot Root Shoot –P

+P –L H C I I

Very poor soil Poor soil Problem soil Good soil

Each dot = 5,000 ha

Figure 1|Origin of the donor variety Kasalath

and Pup1candidate genes.a ,Problem soils in Asia and the origin of stress-tolerant aus -type rice varieties 9,10,13.Inlay,breeding lines with and

without the tolerant Pup1locus 8under P-deficient field conditions.b ,Relative position of Pup1candidate genes in Kasalath and the Nipponbare reference genome.OsPupK05-1is part of OsPupK04-1(refs 6,8).DOX,dioxygenase.Hypoth.,hypothetical.c ,Semiquantitative RT–PCR analysis of Pup1candidate genes in

contrasting Nipponbare NILs 1Pup1and –Pup1grown in P-deficient soil 1/–P fertilizer.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)was used as a control.d ,qRT–PCR analysis of Pup1genes in roots of NILs (gene expression 1P 51).Error bars denote s.e.m.e ,Top,gel stained with Coomassie blue.Bottom,phosphothreonine-specific immunoblot showing that recombinant Pstol1protein restores

phosphorylation of the light-harvesting complex II (LHCII)in the Arabidopsis stn7stn8double mutant (lane 4).f ,Semiquantitative RT–PCR analysis of OsPupK20-2in IR6435S::PSTOL1plants and IR74-Pup1NILs grown in 1P hydroponics.

23A U G U S T 2012|V O L 488|N A T U R E |535

Macmillan Publishers Limited. All rights reserved

?2012

concentrated in only a few countries (Morocco,China and the USA),and mining costs are rising 1,12.Apart from the need for long-term strategies to address this problem,the development of rice varieties with high productivity under low P and other stress conditions is a valid and necessary approach to improve yield and enhance food security in rice-dependent countries.

In recent years,a specific group of rice (aus -type varieties)that originates from a region in India with poor and problem soils 9,10,13(Fig.1a)has been recognized as a valuable source of tolerance genes.For instance,the donor of the submergence-tolerance gene SUB1A is an aus -type variety and rice breeding lines with this gene (Sub1or ‘scuba’rice)survive up to 2weeks in flooded fields 14,15.Likewise,tolerance of drought and heat 16,in addition to other stresses,is present in such varieties.The aus -type variety Kasalath,which is tolerant of P deficiency,was identified about a decade ago,subsequently leading to the identification of a major quantitative trait locus (QTL)associated with P-deficiency tolerance 17.At present,phosphorus uptake 1(Pup1)is the only P-related QTL available for marker-assisted breeding pro-grams,and tolerant Pup1breeding lines have proven effective in field trials 7,8(Fig.1a).Previous efforts to link Pup1with known P-uptake-related mechanisms showed that Pup1near-isogenic lines (NILs)had improved root growth under stress,but the underlying mechanisms remained enigmatic 4,indicating that Pup1might act through a new mechanism or that the underlying gene may be missing in the ref-erence genome.

Indeed,sequencing of the Pup1locus in Kasalath showed the pres-ence of an ,90kilobase transposon-rich insertion–deletion (indel)that is absent from the Nipponbare reference genome and other P-starvation-intolerant rice varieties 6(Fig.1b).A rice germplasm screen conducted with Pup1-specific molecular markers additionally showed that a gene located in the indel,the putative protein kinase gene OsPupK46-2,was most closely associated with tolerance of P deficiency and was highly conserved in stress-adapted rice accessions 7,8.

To gain insight into the function of Pup1and to identify the major genetic determinant of P-deficiency tolerance,the protein kinase OsPupK46-2and four additional Pup1candidate genes 8were short-listed from the initially predicted 68Pup1gene models 6.Gene expression was analysed by semiquantitative reverse transcriptase PCR (RT–PCR)and quantitative (q)RT–PCR analyses in contrasting NILs with (1Pup1)and without (2Pup1)the Kasalath Pup1locus.The data confirmed that OsPupK46-2was absent from the Nipponbare genome and showed that it was upregulated under P-deficient conditions (Fig.1c,d).Expression of the other genes did not change under 2P conditions (OsPupK04-1,OsPupK05-1,OsPupK29-1)or increased in both 1Pup1and 2Pup1NILs (OsPupK20-2)(Fig.1c,d).Additional analyses subsequently showed that OsPupK20-2,which codes for a dirigent protein,is down-stream of the protein kinase (see below).On the basis of these data and the probable role of protein kinases in the sensing and signalling of P homeostasis 18,19,we considered OsPupK46-2the most obvious candidate gene and named it phosphorus-starvation tolerance 1(PSTOL1).

c

a

G r a i n w e i g h t (g p e r p l a n t )

N i p p o n b a r e ?

?

?

= 19 11 23 8 8 6

OX high OX low Null

PSTOL1

GAPDH PSTOL1 GAPDH PSTOL1 GAPDH

T o t a l P (m g p e r p l a n t )

b

G

r a i n w e i g h t (g p e r p l a n t )

I R 64 ?

?

?

?

= 5 5 5 6 9 13 9 12

OX high Null OX low Null

R o o t D W (g p e r p l a n t )

+P

?

??

??

*

T o t a l P (m g p e r p l a n t )

?

?

?

OX high = 5 5 5 6 9 13 9 12 OX high Null OX low Null

= 5 5 5 6 9 13 9 12 OX high Null OX low Null

+P

–P OX high OX low Null = 19 11 23 8 8 6

Figure 2|PSTOL1overexpression enhances tolerance of P deficiency.a ,Representative IR6435S::PSTOL1plants with high (OX high)and low (OX low)transgene expression of independent events (event numbers 20,19and 21)and corresponding null segregants at 8weeks in P-deficient soil (root photos were taken after harvest).b ,Grain weight,P content and root dry weight (DW)of IR64transformants and nulls.n 5number of plants.c ,Grain weight and

total P content of Nipponbare transformants and nulls.Error bars indicate s.e.m.Significance (P ,0.05)is indicated by {(analysis of variance and Tukey’s honestly significant difference test)and asterisks (paired t -test).The mean with symbol {indicates significant difference from the mean with symbol {.The means with the same symbol ({{versus {{)or with a common symbol ({versus {{;{versus {{)are not significantly different from each other.Scale bars,10cm.

RESEARCH LETTER

536|N A T U R E |V O L 488|23A U G U S T 2012

Macmillan Publishers Limited. All rights reserved

?2012

Pstol1shows the highest amino acid sequence similarity with serine/threonine receptor-like kinases of the LRK10L-2subfamily,but lacks the amino-terminal extension typically present in this family 20.This classifies Pstol1as a receptor-like cytoplasmic kinase 21.Interestingly,the conserved kinase domain is most similar to the Arabidopsis defence-related receptor-like kinases PR5K 22(58%)and SNC4(ref.23;57%)(Supplementary Figs 1–3).To assess whether Pstol1is a func-tional protein kinase,an in vitro phosphorylation assay was performed using thylakoid membranes isolated from the Arabidopsis thaliana double mutant stn7stn8,which is defective in STN7and STN8(also known as AT1G68830and AT5G01920,respectively)serine/threonine protein kinases and therefore devoid of phosphorylation of the light-harvesting complex II 24.The data show that recombinant Pstol1protein restored phosphorylation of stn7stn8thylakoids to almost wild-type levels (Fig.1e),confirming that Pstol1is a functional ser-ine/threonine protein kinase.

To quantify the effect of Pstol1on plant performance under low-P stress,we generated transgenic plants with constitutive overexpression of the full-length PSTOL1coding region (35S::PSTOL1).Two rice varieties (IR64and Nipponbare)were used for this experiment,repre-senting two distinct types of modern irrigated varieties (indica and japonica ,respectively)that naturally lack the PSTOL1gene 8(Sup-plementary Fig.4).Phenotypic analyses conducted in two different

locations and P-deficient soil types (from fields that had not received P fertilizer for up to 40years;2P )showed that high expression of the PSTOL1transgene enhanced grain yield by more than 60%under –P conditions in both varieties (Fig.2a–c and Supplementary Fig.5).Transgenic lines with low transgene expression were comparable to segregants without the transgene (null)that were used as controls.These data indicate that expression of PSTOL1above a certain threshold is required to confer tolerance of P deficiency.In both varieties,a significantly higher P content was observed in high PSTOL1-overexpressing lines (Fig.2b,c).For the IR64plants,we further confirmed that the superior performance of PSTOL1lines with high transgene expression was due to a higher root dry weight (Fig.2a,b).The larger root system also enhanced the uptake of other nutrients,as nitrogen and potassium content were also higher in these lines (Supplementary Fig.6).Subsequent phenotypic analyses of IR64PSTOL1-overexpressing lines conducted in nutrient solution with high (100m M)and reduced (10m M)P concentrations showed that,under both P treatments,total root length and root surface area were significantly higher in transgenic seedlings (Fig.3a,c).The same experiment was then repeated with two different contrasting Pup1NILs (IR64and IR74,with (1)or without (–)Pup1)that were developed by marker-assisted introgression of the Kasalath Pup1locus 8.In agreement with the above data,seedlings of 1Pup1

NILs

a

T o t a l r o o t l e n g t h (c m )

T o t a l r o o t s u r f a c e a r e a (c m 2)

OX high Null IR64 transgenics

b

+Pup1 –Pup1

IR64 NILs

High P

T o t a l r o o t l e n g t h (c m )

c

OX high Null

Low P

High P

d

RC

Node

Seminal root

Root tip

roots

1 cm

100 μm 1 cm

e

500 μm

500 μm

**

*

**

PV

ECR

T o t a l r o o t s u r f a c e a r e a (c m 2)

+Pup1 –Pup1 IR64 NILs

Figure 3|PSTOL1is an enhancer of root growth.a ,Total root length and surface area of IR6435S::PSTOL1plants (OX high;T2no.20)and corresponding nulls grown in high-P (100m M)and low-P (10m M)

hydroponics solution for 15days.b ,Root data of sister NILs with and without Pup1grown under the same conditions for 21days.Error bars indicate standard error.Significance was analysed by paired t test (95%).*0.05.P $0.01;**0.01.P $0.001;***P .0.001).c ,Representative root scans.Scale bars,1cm.d ,GUS expression driven by the native PSTOL1promoter in young

IR64seedlings is observed in parenchyma (PC)and outer parenchyma (OP)cells adjacent to the peripheral vascular (PV)cylinder of the coleoptilar node and in crown root primordia (CRP;indicated by asterisks),but not in emerging crown roots (ECR;arrows).RC,root cap.e ,GUS staining in older plants (28days after germination)is likewise seen in crown root primordia (asterisks)and additionally in cells surrounding vascular bundles (VBs),which are interconnected by nodal vascular anastomoses (arrowheads).

LETTER RESEARCH

23A U G U S T 2012|V O L 488|N A T U R E |537

Macmillan Publishers Limited. All rights reserved

?2012

showed significantly enhanced root growth under high and low P conditions (Fig.3b,c and Supplementary Fig.7).The finding that root growth was enhanced in PSTOL1-overexpressing lines as well as in Pup1introgression lines provides strong evidence that PSTOL1is indeed the major tolerance gene within the Pup1QTL and that this gene acts at least partially independently of P.In support of this,down-regulation of PSTOL1by RNA interference in Kasalath caused a sig-nificant reduction in root number and root surface area,which negatively affected overall plant growth (Supplementary Fig.8a–g).To analyse the expression of PSTOL1during root development in more detail,we expressed the b -glucuronidase (GUS)reporter gene under the control of the native PSTOL1promoter in transgenic IR64plants.Specific GUS staining was detected in stem nodes where,in rice,crown roots are formed that constitute the main root system (Fig.3d,e).Within the nodes,GUS staining was restricted to crown root primordia and parenchymatic cells located outside of the peripheral vascular cylinder.This expression pattern is similar to that described for other root-development genes,namely CROWN ROOTLESS 1(also known as AB200234)and RR2,a cytokinin type-A-responsive regulator 25,26.In older plants,GUS staining was additionally detected in the cells surrounding the nodal vascular anastomoses,which interconnect vascular bundles (Fig.3e).No GUS staining was observed in older,emerging crown roots or in the initial (seminal)seedling root.Taken together,our data indicate that PSTOL1is a regulator of early crown root development and root growth in rice.

Because Pstol1is a protein kinase,it cannot directly regulate the expression of genes.However,regulation of transcription factors through phosphorylation has been well studied in the yeast P-starvation response system,which involves the transcription factor PHO4(ref.18),as well as in two-component signalling systems that depend on membrane-bound histidine kinases 27.To gain insight into the downstream responses of Pstol1,we conducted an Affymetrix gene-array analysis using root samples from soil-grown IR64trans-genic plants (high PSTOL1overexpression)and control plants.The data showed that known P-starvation genes were not differentially regulated in the transgenic plants (Supplementary Table 1).Similar results were obtained in a previous Agilent microarray analysis using

Pup1NILs 5.Instead,we identified 23genes with constitutively (that is,independent of the P supply and developmental stage)higher or lower expression in the transgenic plants that are related to root growth and stress response (Supplementary Table 2).Interestingly,21of these differentially expressed genes co-localize with QTLs related to drought tolerance and root growth (Fig.4a),providing further support for an important role of Pup1/PSTOL1during root development and stress tolerance.These findings are also supported by a Pup1-marker analysis that had shown high conservation of PSTOL1in drought-tolerant rice accessions 7,8.In this context,we also determined that the Pup1dirigent gene (OsPupK20-2)is downstream of PSTOL1,as this gene was specifically induced in 35S::PSTOL1plants and in 1Pup1NILs (Fig.1f).

To assess whether the expression of the genes identified is indeed independent of P and/or soil-related factors,a qRT–PCR analysis of selected genes was conducted using root RNA samples of 35S::PSTOL1plants grown under high P conditions in hydroponics.Whereas the data were inconsistent for many of the downregulated genes (data not shown),higher expression was confirmed for six out of the seven genes that were specifically induced in 35S::PSTOL1roots (Fig.4b).Among these are two genes coding for transcription factors;namely HOX1,a positive regulator of root cell differentiation 28,and DOS ,which was shown to delay leaf senescence in rice 29.Altered expression of HOX1is well in agreement with a role for PSTOL1in root development.An association study further showed that a region on chromosome 1,where DOS and a gene coding for a WRKY-type transcription factor are located,was significantly associated with the presence of PSTOL1in a wider range of tolerant rice accessions (Supplementary Fig.9).Interestingly,a putative peptide transporter was among the constitu-tively upregulated genes that might,in addition to P,improve the nitrogen status of the plants 30.

In light of the need to increase rice production for a growing popu-lation despite potentially negative impacts of climate change and increasing scarcity of natural resources,it will be critically important to systematically explore traditional rice varieties in which high-value genes such as PSTOL1are preserved,and to enable breeders to effi-ciently use these genes in breeding programs.

****

Root meta QTLs ref. 42Centromere

b

Hypoth. prot. (Os05g48790)Fold change Peptide SIP23

Null OX

Figure 4|PSTOL1putative downstream genes co-localize with root and drought QTLs.a ,Approximate chromosomal location of genes with

constitutively higher (boxed)or lower (all other genes)expression in roots of 35S::PSTOL1transgenics.QTLs for P-deficiency tolerance are indicated in red and green on the chromosomes.Root-related meta-QTLs and QTLs for grain yield under drought are shown as colour-coded bars (see key).Centromeres are indicated in black.Expr.,expressed.b ,qRT–PCR analysis of upregulated genes in root samples of IR64transgenics (OX)and null controls grown in 1P hydroponics.References for the drought QTLs are *ref.37,**ref.38,***ref.40and ****ref.41.

RESEARCH LETTER

538|N A T U R E |V O L 488|23A U G U S T 2012

Macmillan Publishers Limited. All rights reserved

?2012

METHODS SUMMARY

Pup1candidate gene-expression analysis.Expression of Pup1candidate genes was analysed by semiquantitative RT–PCR as described8and by qRT–PCR of root samples derived from49-day-old plants of a set of Nipponbare NILs with and without the Kasalath Pup1introgression.Plants were grown in P-deficient and P-sufficient soil in Japan.Primer sequences for this and other experiments are provided in Supplementary Table3.

Protein kinase activity of Pstol1.Recombinant Pstol1protein was synthesized in Escherichia coli(strain BL21;pBAD-DEST49expression vector,Invitrogen). Purified protein was incubated with isolated thylakoid membranes from the Arabidopsis double mutant stn7stn8defective in phosphorylating the light-harvesting complex II24.

Transgenic35S::PSTOL1plants and Affymetrix analysis.The PSTOL1coding sequence was cloned into the binary vector pMDC32with the35S promoter and used for Agrobacterium-mediated transformation of immature IR64and Nipponbare embryos.Transgene copy number and expression level were deter-mined by Southern blot analysis and RT–PCR,respectively.Independent trans-genic lines(T1and T2)were phenotyped at the International Rice Research Institute(the Philippines)and Japan International Research Center for Agricultural Sciences(Japan)in P-deficient soil1/2P fertilizer application(equi-valent of60kg P2O5ha21).For the Affymetrix gene-array analysis(line20)root samples from IR6435S::PSTOL1plants and nulls grown in P-deficient soil1/2P fertilizer were used.Plants were at the reproductive/heading stage(2P treatment) and at mid-tillering(1P control treatment).Complementary RNA samples were analysed at ATLAS Biolabs GmbH(Germany)using GeneChip Operating System1.4.

Promoter::GUS analysis.A1,755base pair genomic region upstream of the PSTOL1start ATG codon was amplified from the genomic DNA of1Pup1 NILs using the primer pair oRG107/oRG109(Supplementary Table3)and cloned into the pMDC164vector containing the gene coding for GUS.For GUS staining, IR64transgenic T1plants were germinated on Petri dishes in the dark at room temperature(26u C)and monitored for GUS after1week and4weeks.

Full Methods and any associated references are available in the online version of the paper.

Received23November2011;accepted25June2012.

1.Van Kauwenbergh,S.J.World Phosphate Rock Reserves and Resources http://

a5b02cc052d380eb63946da9/getdoc/56358fb1-fc9b-49ba-92fe-187dc08e9586/T-

75_World_Phosphate_Rock_Reserves_and_Resources(International Fertilizer Development Center,2010).

2.Wissuwa,M.,Yano,M.&Ae,N.Mapping of QTLs for phosphorus-deficiency

tolerance in rice(Oryza sativa L.).Theor.Appl.Genet.97,777–783(1998).

3.Wissuwa,M.&Ae,N.Genotypic variation for tolerance to phosphorus deficiency in

rice and the potential for its exploitation in rice improvement.Plant Breed.120, 43–48(2001).

4.Wissuwa,a5b02cc052d380eb63946da9bining a modeling with a genetic approach in establishing

associations between genetic and physiological effects in relation to phosphorus uptake.Plant Soil269,57–68(2005).

5.Pariasca-Tanaka,J.,Satoh,K.,Rose,T.,Mauleon,R.&Wissuwa,M.Stress response

versus stress tolerance:a transcriptome analysis of two rice lines contrasting in tolerance to phosphorus deficiency.Rice2,167–185(2009).

6.Heuer,S.et a5b02cc052d380eb63946da9parative sequence analyses of the major quantitative trait locus

Phosphorus uptake1(Pup1)reveal a complex genetic structure.Plant Biotechnol.J.

7,456–471(2009).

7.Chin,J.H.et al.Development and application of gene-based markers for the major

rice QTL Phosphorus uptake1.Theor.Appl.Genet.120,1073–1086(2010).

8.Chin,J.H.et al.Developing rice with high yield under phosphorus deficiency:Pup1

sequence to application.Plant Physiol.156,1202–1216(2011).

9.Haefele,S.M.&Hijmans,R.J.In Science,Technology,and Trade for Peace and

Prosperity:Proceedings of the26th International Rice Research Conference,New Delhi,India.(eds Aggarwal P.K.,Ladha J.K.,Singh R.K.,Devakumar C.&Hardy B.) International Rice Research Institute,Indian Council of Agricultural Research,and National Academy of Agricultural Sciences,Los Ban?os,Philippines,and New Delhi, India297–308(2007).

10.Haefele,S.M.&Hijmans,R.J.Soil quality in rainfed lowland rice.Rice Today8,

30–31(2009).

11.Dawe,D.,Pandey,S.&Nelson,A.In Rice in the Global Economy.Strategic Research

and Policy Issues for Food Security(eds Pandey,S.et al.)5–35(International Rice Research Institute,2010).12.Cordell,D.,Drangert,J.O.&White,S.The story of phosphorus:global food security

and food for thought.Glob.Environ.Change19,292–305(2009).

13.Londo,J.P.,Chiang,Y.C.,Hung,K.H.,Chiang,T.Y.&Schaal,B.A.Phylogeography

of Asian wild rice,Oryza rufipogon,reveals multiple independent domestications of cultivated rice,Oryza sativa.Proc.Natl a5b02cc052d380eb63946da9A103,9578–9583(2006). 14.Xu,K.et al.Sub1A is an ethylene-responsive-factor-like gene that confers

submergence tolerance to rice.Nature442,705–708(2006).

15.Manzanilla,D.O.et al.Submergence risks and farmers’preferences:implications

for breeding Sub1rice in Southeast Asia.Agric.Syst.104,335–347(2011). 16.Gowda,V.R.P.et al.Water uptake dynamics under progressive drought stress in

perse accessions of the OryzaSNP panel of rice(Oryza sativa).Func.Plant Bio.39, 402–411(2012).

17.Wissuwa,M.,Wegner,J.,Ae,N.&Yano,M.Substitution mapping of Pup1:a major

QTL increasing phosphorus uptake of rice from a phosphorus-deficient soil.Theor.

Appl.Genet.105,890–897(2002).

18.Gregory,P.D.,Barbari,S.&Ho¨rz,W.Transcriptional control of phosphate-

regulated genes.Food Technol.Biotechnol.38,295–303(2000).

19.Yang,X.J.&Finnegan,P.M.Regulation of phosphate starvation responses in

higher plants.Ann.Bot.(Lond.)105,513–526(2010).

20.Dardick,C.,Chen,J.,Richter,T.,Ouyang,S.&Ronald,P.The rice kinase database.

A phylogenomic database for the rice kinome.Plant Physiol.143,579–586

(2007).

21.Vij,S.,Giri,J.,Dansana,P.K.,Kapoor,S.&Tyagi,A.K.The receptor-like

cytoplasmic kinase(OsRLCK)gene family in rice:organization,phylogenetic relationship,and expression during development and stress.Mol.Plant1,

732–750(2008).

22.Wang,X.,Zafian,P.,Choudhary,M.&Lawton,M.The PR5K receptor protein kinase

from Arabidopsis thaliana is structurally related to a family of plant defense

proteins.Proc.Natl a5b02cc052d380eb63946da9A93,2598–2602(1996).

23.Bi,D.,Cheng,Y.T.,Li,X.&Zhang,Y.Activation of plant immune responses by a

gain-of-function mutation in an atypical receptor-like kinase.Plant Physiol.153, 1771–1779(2010).

24.Bonardi,V.et al.Photosystem II core phosphorylation and photosynthetic

acclimation require two different protein kinases.Nature437,1179–1182(2005).

25.Inukai,Y.et al.Crown rootless1,which is essential for crown root formation in rice,is

a target of an AUXIN RESPONSE FACTOR in auxin signaling.Plant Cell17,

1387–1396(2005).

26.Zhao,Y.,Hu,Y.,Dai,M.,Huang,L.&Zhou,D.-X.The WUSCHEL-related homeobox

gene WOX11is required to activate shoot-borne crown root development in rice.

Plant Cell21,736–748(2009).

27.Lohrmann,J.&Harter,K.Plant two-component signaling systems and the role of

response regulators.Plant Physiol.128,363–369(2002).

28.Scarpella,E.,Simons,E.J.&Meijer,A.J.Multiple regulatory elements contribute to

the vascular-specific expression of the rice HD-Zip gene Oshox1in Arabidopsis.

Plant Cell Physiol.46,1400–1410(2005).

29.Kong,Z.,Li,M.,Yang,W.,Xu,W.&Xue,Y.A novel nuclear-localized CCCH-type zinc

finger protein,OsDOS,is involved in delaying leaf senescence in rice.Plant Physiol.

141,1376–1388(2006).

30.Ouyang,J.et al.Identification and analysis of eight peptide transporter homologs

in rice.Plant Sci.179,374–382(2010).

Supplementary Information is linked to the online version of the paper at

a5b02cc052d380eb63946da9/nature.

Acknowledgements We would like to thank A.Cruz,E.Ramos and L.Olivo for technical and secretarial support,the staff at the transformation laboratory,F.Rossi for help with the Pstol1kinase assay,M.Akutsu for help with the analysis of transgenic plants,and S.Haefele and his team for their support.We thank J.Prasetiyono,M.Bustamam and S.Moeljopawiro for their long-term collaboration.This project has been primarily funded by the Generation Challenge Program(GCP)since2005.

Author Contributions R.G.cloned and transformed the PSTOL1gene into IR64and Nipponbare.R.G.,J.P.T.and M.W.performed the phenotyping of transgenic plants. J.H.C.conducted the root meta-QTL analysis and J.H.C.and C.D.developed the

IR64-Pup1and IR74-Pup1NILs.P.P.carried out the Pstol1kinase assay.S.C. conducted the expression analysis of putative PSTOL1downstream genes.E.M.T.M. provided advice about the experiments and I.S.-L.provided technical support and infrastructure for rice transformation.R.G.,M.W.and S.H.designed the experiments and wrote the manuscript.

Author Information GenBank protein accession numbers for OsPupK04-1,

OsPupK05-1,OsPupK20-2,OsPupK29-1and PSTOL1/OsPupK46-2are BAH79993, BAH79994,BAK26565,BAH80018and BAK26566,respectively.Reprints and permissions information is available at a5b02cc052d380eb63946da9/reprints.The authors declare no competing financial interests.Readers are welcome to comment on the online version of this article at a5b02cc052d380eb63946da9/nature.Correspondence and requests for materials should be addressed to S.H.(s.heuer@a5b02cc052d380eb63946da9).

LETTER RESEARCH

23A U G U S T2012|V O L488|N A T U R E|539

Macmillan Publishers Limited. All rights reserved ?2012

METHODS

qRT–PCR of Pup1candidate genes.Seeds of NILs segregating for the Pup1locus (1Pup1,NILs6-4,Y-4and14-4;–Pup1,NILs Y6,Y10and Nipponbare)7,17were sown directly in pots filled with P-deficient and P-fixing andosol from a field located at Tsukuba,Japan,that had not received P fertilizer throughout its40-year cropping history(–P).An equivalent of60kg P ha21was applied for the control treatment(1P).Pots were initially watered every2–3days and afterwards the soil was kept near field capacity.The experiment was conducted in a completely randomized design with three replications and four plants per replicate pot. Root tissue samples were taken at49days after sowing.Total RNA was extracted using the RNeasy mini kit according to the instructions of the manufacturer (Qiagen)and treated with RNase-free DNase I(Qiagen).qRT–PCR was per-formed as described5with some a5b02cc052d380eb63946da9plementary DNA synthesis was conducted at37u C for15min followed by5s of85u C using500ng DNase-treated total RNA with PrimeScript RT reagent kit(Takara).qRT–PCR was per-formed with10ng reverse transcriptase template and SYBR Premix Ex Taq (Perfect Real Time;Takara).PCR cycle conditions were94u C for10s as the first denaturing step,followed by40cycles at94u C for5s,55–60u C for10s and72u C for15s,and a gradual increase in temperature from55u C to96u C during the dissociation stage to monitor the specificity of each primer pair.Rice18S(also known as RRN18)was used as an internal control.For primer sequences,see Supplementary Table3.Expression levels were calculated using the delta-delta comparison and expressed as fold changes under–P relative to expression under 1P conditions(expression51).

In vitro phosphorylation assay.Seeds of A.thaliana ecotype Col-0and of the stn7stn8double mutant were sown in plastic trays containing one portion of Techinc and one portion of Flox6soils and incubated for3days at5u C in the dark to break the dormancy.Plants were grown in a greenhouse under long-day conditions(16h light and8h dark)for4weeks.Thylakoids were isolated from 4-week-old plants as described24in the presence of the phosphatase inhibitor sodium fluoride(10mM).The coding sequence(CDS)of PSTOL1was cloned into the pBAD-DEST49vector(Invitrogen),and recombinant Pstol1(Pstol1rec) was expressed in the E.coli strain BL21with a carboxy-terminal63His-tag. Pstol1rec was purified under denaturing conditions following a nickel-nitrilotriace-tic acid batch purification procedure according to the instructions of the manufac-turer(Qiagen).After protein precipitation in10%trichloroacetic acid followed by three washing steps with absolute ethanol,around500m g of Pstol1rec protein was re-suspended in500m l1%(w/v)lithium dodecyl sulphate,12.5%(w/v)sucrose, 5mM e-aminocaproic acid,1mM benzamidine and50mM HEPES potassium hydroxide buffer,pH7.8,as previously described31.Subsequently,Pstol1rec protein was boiled for2min at100u C and incubated for15min at25u C.Dithiothreitol (DTT;75mM final concentration)was then added and the solution was subjected to three freeze–thaw cycles(20min at–20u C,20min at–80u C and20min at 220u C,thawing in an ice-water bath,and5min at25u C).After completion of the three freeze–thawing cycles,octylglucopyranoside(1%(w/v)final concentra-tion)was added and the solution was kept on ice for15min before potassium chloride(75mM,final concentration)was added to precipitate the lithium dodecyl sulphate detergent.After centrifugation at16,000g at4u C for10min,the supernatant containing the re-folded Pstol1rec in the presence of1%(w/v) octylglucopyranoside was collected.Subsequently,1m l of kinase was incubated together with thylakoids corresponding to5m g of total chlorophyll.The phosphor-ylation reaction was performed in50m l total volume containing0.06%(w/v) dodecyl-B-D-maltoside,5mM magnesium acetate,5mM DTT,100mM HEPES potassium hydroxide,pH7.8,200mM ATP and10mM sodium fluoride at37u C for2h.The reaction mixture was loaded on an SDS–PAGE,and immunoblot analyses with phosphothreonine-specific antibodies(Cell Signaling)were per-formed as described32.A replicative SDS–PAGE was stained with Coomassie blue. Generation of35S::PSTOL1transgenic plants.The CDS of PSTOL1was amp-lified from Kasalath genomic DNA using the primer pair oKas4603and oKas4604 (all primer sequences are provided in Supplementary Table3),cloned into pCR8/ GW/TOPO TA cloning vector(Invitrogen)and sent for sequencing(Macrogen). Through LR clonase recombination reaction(Invitrogen),the CDS was sub-cloned into the pMDC32binary destination vector33containing the35S promoter and NOS terminator(35S::PSTOL1).The construct was sequenced using primer pairs amplifying the35S promoter(oRG89)and the NOS-terminator(oSH07) with adjacent CDS,respectively.The correct sequence of PSTOL1was re-confirmed by sequencing with the primer pair oKas4603and oKas4604. Transformation of the construct into the indica-type IR64and japonica-type Nipponbare rice varieties,which naturally lack the PSTOL1gene,was mediated by the Agrobacterium tumefaciens strain LBA4404according to a published pro-tocol34with modifications(I.S.-L.et al.,in preparation).Transgenic plants were tested by genomic PCR in the T1generation for the presence of the hygromycin phosphotransferase gene(hpt;primer pair oRG127and oRG128)and the35S promoter with part of the CDS(primer pair oRG89and oRG88).PCR was carried out in a total volume of20m l with the following conditions:100ng genomic DNA, primers(0.2m M each of forward and reverse),13PCR buffer,0.5mM dNTP mix and1.5U iTaq DNA polymerase(Intron Biotechnology).The PCR cycle settings were94u C for5min,followed by30cycles of94u C for30s,Ta(55u C for primer pair oRG127and oRG128,60u C for primer pairs oRG88and oRG89and GAPDH forward(-F)and GAPDH reverse(-R)for30s,72u C for extension time(30s for oRG127and oRG128and45s for oRG88and oRG89and GAPDH-F and GAPDH-R)and a final extension at72u C for10min.As a control,the GAPDH gene was amplified using the primer pair GAPDH-F and GAPDH-R.PCR pro-ducts were separated by agarose gel electrophoresis and stained with SYBR Safe (Invitrogen).The copy number of the transgene in selected plants was determined by Southern blot analysis using genomic DNA digested with XbaI and SacI, respectively,and hybridized with a digoxigenin-labelled hpt probe.Plants with independent transformation events were selected for phenotypic analysis in the T1 generation(Supplementary Fig.4).

Phenotyping of35S::PSTOL1plants.T1seeds from selected independent IR64 transgenic lines(Supplementary Fig.4b)were pre-germinated in Petri dishes for 3days in the dark at room temperature(26u C)before seedlings were transferred into seedling trays.At21days after germination(DAG),transgenic plants and the corresponding null segregant were transferred into pots filled with P-deficient soil (P-Bray, 1.2360.30mg kg21;P-Olsen,0.7760.46mg kg21)from Siniloan (Luzon,the Philippines).To control for pot-to-pot variation,one transgenic plant and one null segregant were always grown together in each pot.Each pot received the equivalent of90kg nitrogen ha21,40kg potassium ha21and20kg zinc ha21. The equivalent of60kg P ha21was applied only to the1P control treatment that was done in parallel.To mimic upland field conditions,plants grown under–P conditions were exposed to a dry-down treatment until leaf rolling at about60days after germination.Control pots were kept well watered but aerobic.

In an initial experiment,seven independent lines and the corresponding nulls were analysed(data not shown)and two lines(19and20)with high transgene expression and three lines(5,14and21)with low transgene expression (Supplementary Figs4and5)were selected for detailed analyses.A similar phe-notyping experiment was conducted at the Japan International Research Center for Agricultural Sciences using independent T2Nipponbare transgenic lines grown in well-watered(aerobic)P-deficient soil from Tsukuba(Japan) (Supplementary Fig.4a).For the1P control,soil from a field that had regularly received P fertilizer was used and60kg P h21was additionally applied. Macronutrients in roots,shoots and grains of IR64transgenic plants and null controls were analysed by the Analytical Services Laboratory(ASL)at the International Rice Research Institute.The Kjeldahl method was used to determine the percentage of nitrogen,whereas a modified ASL nitric/perchloric acid digestion was done for inductively coupled plasma analysis of phosphorus and potassium.

Semiquantitative RT–PCR analysis of transgene expression.RT–PCR analysis of35S::PSTOL1expression was conducted using leaf samples.Total RNA was extracted using Trizol(Invitrogen)or RNeasy mini kit(Qiagen)and DNA con-taminations were removed with RNase-free DNase I(Promega or Qiagen).cDNA synthesis in the IR64experiment was performed at55u C for1h in a20m l reaction with1m g RNA template,2.5m M oligo dT,0.5mM dNTP mix,0.01M DTT,13 first-strand buffer and200U of Superscript III RT(Invitrogen).For the Nipponbare experiment,500ng RNA template was used for cDNA synthesis in a total volume of10m l using PrimeScript RT reagent kit(Takara)at37u C for 15min and then85u C for5s.For standard PCR analyses,0.5–1m l cDNA was used as template for amplification of the transgene with iTaq DNA polymerase(Intron Biotechnology)or Takara Taq(Takara)using gene-specific primers(0.2m M each of oKas4603and oKas4604;Supplementary Table3).GAPDH was used as a positive control.

Root scan of IR6435S::PSTOL1T2plants and Pup1NILs grown in hydro-ponics.Seeds of the IR64T2transgenic line20and seeds of IR64-Pup1and IR74-Pup1NILs8were pre-germinated in Petri dishes in the dark at room temperature. After3days,germinated seeds were transferred to Yoshida culture solution35with 100m M and10m M NaH2PO4,respectively.The solution was replaced every 3days.Total root length and root surface area of seedlings(11–21DAG)were measured using WinRhizo(MAC STD1600;Regent Instruments).Each root system was evenly spread out and scanned at least twice to obtain average values. Each experiment was reproduced at least once.Null controls and NILs without Pup1were always grown and analysed in parallel.

PSTOL1promoter::GUS IR64transgenic plants.The1,755base pair promoter of PSTOL1was amplified from the genomic DNA of1Pup1NILs using the primer pair oRG107and oRG109(Supplementary Table3),cloned into the pCR8/GW/ TOPO TA cloning vector(Invitrogen)and sent for sequencing(Macrogen).The promoter fragment was sub-cloned into a pMDC164binary destination vector31

RESEARCH LETTER

Macmillan Publishers Limited. All rights reserved

?2012

through LR clonase recombination reaction(Invitrogen).The final construct contained the GUS gene driven by the PSTOL1promoter,which was confirmed using the forward primer oRG120,sequencing from the39end of the promoter extending to the CDS of the GUS gene.The construct was transformed into IR64 using the same protocol as described above.Transformed T0plants were identified by genomic PCR using oRG120and oRG134,also verifying the fusion of the PSTOL1promoter with the GUS gene.PCR conditions were the same as described above.For expression analyses,1-week-old T1seedlings grown in Petri dishes at room temperature in the dark were incubated in GUS staining solution according to the protocol36.Samples were stored in70%ethanol before embedding in agarose for sectioning(200m m)and brightfield microscopy(Olympus BX53with attached Olympus DP70camera).

Affymetrix gene-expression analysis.For microarray analyses,root samples of IR6435S::PSTOL1and the corresponding null segregants were collected from T1 plants of line20grown in pots with P-deficient soil under stress(–P;dry-down) and control(1P fertilizer;well-watered aerobic)conditions.Plants grown under control conditions were sampled at the four-tiller stage at33DAG.The stress treatment delayed development,and plants were collected at the heading stage when plants had developed two to four tillers.For all treatments,samples of two biological replicates were analysed.Total root RNA was extracted using Trizol according to the instructions from the manufacturer(Invitrogen),with modifica-tions.The RNA was re-precipitated by adding2.53volume absolute ethanol and one-tenth volume3M NaOAc,pH5.2,washed twice(70%and100%ethanol),air-dried and dissolved in RNase-free water before treatment with RNase-free DNase I (Promega).cRNA synthesis and labelling,hybridization and data analysis with the GeneChip operating system1.4were performed by ATLAS Biolabs GmbH using Affymetrix GeneChip rice genome arrays.Identification of genes with differential expression between transgenics and nulls and between P treatments was restricted to probe-set IDs with consistent data in both replicates.For the identification of genes with lower expression in transgenic plants compared with nulls,all IDs ‘present’(expressed)in the nulls were used.For the identification of genes with higher expression,all IDs present in transgenic plants were used.Genes classified as‘constitutively’changed in the transgenics showed significantly(P,0.05) altered expression in all data sets.

Expression analysis and physical location of putative PSTOL1downstream genes.For qRT–PCR analysis of the genes indentified in the Affymetrix study, roots from IR6435S::PSTOL1T2and null control plants grown hydroponically in Yoshida culture solution with100m M P were collected at49DAG.Total RNA extracted with Trizol(Invitrogen)was treated with RNase-free DNase I(Promega) and cDNA synthesis was performed with Transcriptor First Strand cDNA Synthesis Kit(Roche)using1m g DNase-treated RNA.qRT–PCR was conducted with LightCycler480SYBR Green I Master(Roche)using0.5m l cDNA template with the following PCR conditions:94u C for5min,40cycles at94u C for10s, 55u C for5s and72u C for20s.Primer sequences are provided in Supplementary Table3.GAPDH was used as an internal control.Expression levels were calculated using the delta-delta comparison and expressed as fold change relative to the expression in null controls(expression51).The physical location of genes was derived from the Rice Genome Browser(a5b02cc052d380eb63946da9/cgi-bin/ gbrowse/rice/)and the physical position of drought tolerance and meta-QTLs for roots and drought was derived from published data2,37–42.The data were manually summarized and graphically illustrated.

Association analysis.A total of79rice varieties with different Pup1haplotypes8 were genotyped with379single-nucleotide polymorphism(SNP)markers using the RiceOPA2.1BeadXpress platform43and analysed using structure44to identify co-ancestry subgroups.The optimum number of populations(K)was selected by testing for K51to K58using ten independent runs of10,000burn-in runs followed by100,000iterations with a model allowing for admixture and correlated allele frequencies45.K56provided the best distinction and two subgroups with the most contrasting Pup1haplotypes(Kasalath type,1Pup1;Nipponbare type,–Pup1)were selected for further analysis.SNP markers located within the putative PSTOL1downstream genes(Fig.4)are not present in the379SNP set and markers located within approximately1megabase distance from the genes were therefore used for analysis of allelic associations with PSTOL1using TASSEL 3.046 (Supplementary Fig.9).Rice accessions included in this study were:Kas group, Kasalath,AUS196,AUS257,Dular,IR84144-11-12,Lemont and Vandana;Nip group,Bala,CT6510-24-1-2,IR42,IR64,IR66424-1-2-1-5,IR73678-6-9-B,IR74, IR74371-46-1-1,K36-5-1-1BB,Nipponbare,PM-36and Vary Lava701.

RNA interference(RNAi)transgenic plants.A322bp fragment specific to the PSTOL1gene was amplified using the primer pair oSH07and oSH08and cloned into pENTR/D-TOPO vector(Invitrogen).The cloned fragment was transferred into pANDA RNAi vector47through LR clonase recombination reaction (Invitrogen).The RNAi construct was transformed into the Pup1donor variety Kasalath using the rice-transformation protocol described above.Six RNAi lines (T2and T3generation)were selected on the basis of semiquantitative RT–PCR showing downregulation of PSTOL1in roots using the oSH07and oSH08primer pair as described above.To verify whether the RNAi cassette is active,the expres-sion of the GUS linker between the sense and antisense sequence47of the cloned PSTOL1fragment was determined.Selected RNAi lines were grown in hydroponics culture solution and in P-deficient soil and phenotyped(see supplementary Fig.8). Wild-type Kasalath and null segregants were analysed in parallel.

31.Haldrup,A.,Naver,H.&Scheller,H.V.The interaction between plastocyanin and

photosystem I is inefficient in transgenic Arabidopsis plants lacking the PSI-N subunit of photosystem I.Plant J.17,689–698(1999).

32.Pribil,M.,Pesaresi,P.,Hertle,A.,Barbato,R.&Leister,D.Role of plastid protein

phosphatase TAP38in LHCII dephosphorylation and thylakoid electron flow.PLoS Biol.8,e1000288(2010).

33.Curtis,M.D.&Grossniklaus,U.A Gateway cloning vector set for high throughput

functional analysis of genes in planta.Plant Physiol.133,462–469(2003).

34.Hiei,Y.,Ishida,Y.,Kasaoka,K.&Komari,T.Improved frequency of transformation in

rice and maize by treatment of immature embryos with centrifugation and heat prior to infection with Agrobacterium tumefaciens.Plant Cell a5b02cc052d380eb63946da9.87,233–243 (2006).

35.Yoshida,S.,Forno,D.A.,Cock,J.H.&Gomez,a5b02cc052d380eb63946da9boratory Manual for

Physiological Studies of Rice2nd edn1–70(International Rice Research Institute, 1972).

36.Jefferson,R.A.,Kavanagh,T.A.&Bevan,M.W.GUS fusions:b-glucuronidase as a

sensitive and versatile gene fusion marker in higher plants.EMBO J.6,3901–3907 (1987).

37.Bernier,J.,Kumar,A.,Ramaiah,V.,Spaner,D.&Atlin,G.A large-effect QTL for grain

yield under reproductive-stage drought stress in upland rice.Crop Sci.47,

507–518(2007).

38.Bimpong,I.K.et al.Determination of genetic variability for physiological traits

related to drought tolerance in African rice(Oryza glaberrima).J.Plant Breed.Crop.

Sci.3,60–67(2011).

39.Courtois,B.et al.Rice root architecture:meta-analysis from a drought QTL

database.Rice2,115–128(2009).

40.Gomez,M.S.et al.Mapping QTLs linked to physio-morphological and plant

production traits under drought stress in rice(Oryza sativa L.)in the target

environment.Am.J.Biochem.Biotechnol.2,161–169(2006).

a5b02cc052d380eb63946da9nceras,J.C.,Pantuwan,G.,Jongdee,B.&Toojinda,T.Quantitative trait loci

associated with drought tolerance at reproductive stage in rice.Plant Physiol.135, 384–399(2004).

42.Ni,J.J.,Wu,P.,Senadhira,D.&Huang,N.Mapping QTLs for phosphorus deficiency

tolerance in rice(Oryza sativa L.).Theor.Appl.Genet.97,1361–1369(1998). 43.Thomson,M.J.et al.High-throughput single nucleotide polymorphism genotyping

for breeding applications in rice using the BeadXpress platform.Mol.Breed.29, 875–886(2011).

44.Pritchard,J.K.,Stephens,M.&Donnelly,P.Inference of population structure using

multilocus genotype data.Genetics155,945–959(2000).

45.Falush,D.,Stephens,M.&Pritchard,J.K.Inference of population structure using

multilocus genotype data:linked loci and correlated allele frequencies.Genetics 164,1567–1587(2003).

46.Bradbury,P.J.et al.TASSEL:software for association mapping of complex traits in

perse samples.Bioinformatics23,2633–2635(2007).

47.Miki,D.&Shimamoto,K.Simple RNAi vectors for stable and transient suppression

of gene function in rice.Plant Cell Physiol.45,490–495(2004).

LETTER RESEARCH

Macmillan Publishers Limited. All rights reserved ?2012

本文来源：https://www.bwwdw.com/article/qogl.html