Nectar secretion requires sucrose phosphate ...

11 downloads 0 Views 11MB Size Report
Mar 16, 2014 - eudicot species: Arabidopsis thaliana, Brassica rapa (extrastaminal ... with the evolution of core eudicots and contributed to the evolution.
LETTER

doi:10.1038/nature13082

Nectar secretion requires sucrose phosphate synthases and the sugar transporter SWEET9 I Winnie Lin1,2, Davide Sosso1,2, Li-Qing Chen2, Klaus Gase3, Sang-Gyu Kim3, Danny Kessler3, Peter M. Klinkenberg4{, Molly K. Gorder4{, Bi-Huei Hou2, Xiao-Qing Qu2,5, Clay J. Carter4{, Ian T. Baldwin3 & Wolf B. Frommer1,2

Angiosperms developed floral nectaries that reward pollinating insects1. Although nectar function and composition have been characterized, the mechanism of nectar secretion has remained unclear2. Here we identify SWEET9 as a nectary-specific sugar transporter in three eudicot species: Arabidopsis thaliana, Brassica rapa (extrastaminal nectaries) and Nicotiana attenuata (gynoecial nectaries). We show that SWEET9 is essential for nectar production and can function as an efflux transporter. We also show that sucrose phosphate synthase genes, encoding key enzymes for sucrose biosynthesis, are highly expressed in nectaries and that their expression is also essential for nectar secretion. Together these data are consistent with a model in which sucrose is synthesized in the nectary parenchyma and subsequently secreted into the extracellular space via SWEET9, where sucrose is hydrolysed by an apoplasmic invertase to produce a mixture of sucrose, glucose and fructose. The recruitment of SWEET9 for sucrose export may have been a key innovation, and could have coincided with the evolution of core eudicots and contributed to the evolution of nectar secretion to reward pollinators. Plants have evolved anatomical and physiological features for attracting animals to promote pollination. Reproductive isolation and thus speciation is thought to be enhanced in animal-pollinated species relative to those reliant on wind pollination3. Floral traits, including animal pollination, can affect subsequent species abundance within clades1. When de Saporta, Hooker, Heer and Darwin discussed the ‘‘abominable mystery’’—the apparent rapid radiation of angiosperms and insects in the mid-Cretaceous period, de Saporta suggested that the development and refinement of insect-assisted pollination through the coevolution of pollinators and flowering plants may have been key to pollinator and angiosperm diversification4. Yet the molecular mechanism of nectar secretion has remained elusive2. Flowering plants evolved intricate methods to secure efficient interaction with pollinators and, thereby, successful reproduction and genetic diversity through cross-pollination. Nectar, which contains sugars and volatile compounds that attract and reward pollinators5 and toxins that repel unwanted visitors6, compels pollinators to optimize outcrossing rates7. Nectar composition varies between species of plants, possibly to reward different animal species2. Depending on the plant species, sucrose, glucose and fructose make up 8–80% (w/w) of nectar. Angiosperm nectar is synthesized and secreted by different types of floral nectaries (FNs) and extrafloral nectaries (EFNs). N. attenuata (coyote tobacco), a self-compatible, hawkmoth- and hummingbird-pollinated asterid, produces nectar containing sucrose, hexoses and numerous secondary metabolites6,8. B. rapa (turnip), comprising self-compatible and incompatible varieties, produces hexose-dominant nectar9,10. A. thaliana, a self-compatible self-fertilizer, develops functional nectaries that produce volatiles and hexose-rich nectar9,11,12. Whether nectar production in selffertilizing plants represents an evolutionary remnant or functions to secure the observed low outcrossing rates has yet to be established13.

To identify transporters involved in nectar secretion, we searched for candidate sugar transporters specifically expressed in nectaries. SWEET sugar transporters appeared as prime candidates for a role in nectar secretion14,15: for example, A. thaliana (At)SWEET11 and 12 are responsible for sucrose efflux, a key step in sucrose translocation from photosynthetic tissues to heterotrophic tissues. A petunia SWEET9 homologue (PhNEC1; Extended Data Figs 1, 2) had been found in nectaries, and expression correlated inversely with nectary starch content16; however, its role remained elusive until the Arabidopsis and rice homologues were shown to function as sugar transporters15. Here we show the critical role of SWEET9 in nectar secretion by confirming its expression in nectaries, sucrose transport activity, and localization at the plasma membrane. Mutation of AtSWEET9 or nectary-expressed sucrose phosphate synthase (SPS) genes led to the loss of nectar secretion. Sugars delivered to defective nectaries accumulated in the stems at the floral base, indicating the lack of negative feedback on phloem delivery and the inability to relocate incoming sucrose. We demonstrate that SWEET9 is either conserved or has been independently co-opted in rosids (Arabidopsis, turnip) and an asterid (coyote tobacco) for nectar secretion. AtSWEET9, which shares ,50% identity with AtSWEET11 or 12 (clade 3 of the SWEET family; Extended Data Figs 1, 2), is highly expressed in Arabidopsis nectaries17 (Extended Data Fig. 3). Transport studies in Xenopus oocytes showed that AtSWEET9 mediates sucrose uptake and efflux (Fig. 1a, b). The sucrose transport activity of AtSWEET9 was confirmed by coexpression of AtSWEET9 with fluorescence resonance energy transfer (FRET) sucrose sensors in human embryonic kidney cells14, indicating that AtSWEET9 functions as a facilitated diffusion transporter for sucrose15. AtSWEET9 weakly transports glucose; possibly contributing to hexose efflux (Extended Data Fig. 4). To determine directly whether AtSWEET9 is involved in nectar secretion, we examined nectar in transfer DNA (T-DNA) insertion mutants. Depending on whether AtSWEET9 has a role in sugar uptake into or efflux from nectaries, nectar from mutants would either contain reduced sugar content or lose fluid secretion. Atsweet9 contained no detectable nectar droplets (Fig. 1c, d, Extended Data Fig. 5 and Supplementary Table 1). To test whether AtSWEET9 activity limits nectar secretion, extra AtSWEET9 copies expressed from the native promoter in wildtype plants increased nectar volume and glucose amount (Fig. 1e and Supplementary Table 2). Restoration of nectar secretion by AtSWEET9 or AtSWEET9 tagged with green fluorescent protein (GFP) in Atsweet9 mutants further supports the role of SWEET9 in nectar secretion (Extended Data Fig. 5). SWEET9 could function in (1) sucrose efflux from phloem, (2) sugar uptake into, or (3) sugar efflux from nectary parenchyma. Translational b-glucuronidase (GUS) and enhanced (e)GFP fusions driven by the AtSWEET9 promoter were specifically expressed in nectaries (Fig. 2a–c), with the highest levels seen in the lower half of nectary parenchyma, but not in guard cells and phloem. The fluorescence intensity of

1

Department of Biology, Stanford University, Stanford, California 94305, USA. 2Carnegie Institution for Science, 260 Panama Street, Stanford, California 94305, USA. 3Max Planck Institute for Chemical Ecology, Jena D-07745, Germany. 4Department of Biology, University of Minnesota Duluth, Duluth, Minnesota 55812, USA. 5Key Laboratory of Plant and Soil Interactions, College of Resources and Environmental Sciences, China Agricultural University, 100193 Beijing, China. {Present address: Department of Plant Biology, University of Minnesota, St Paul, Minnesota 55108, USA. 0 0 M O N T H 2 0 1 4 | VO L 0 0 0 | N AT U R E | 1

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER * 8

**

a

b 1.4

Sucrose efflux (pmol oocyte–1)

Sucrose uptake (pmol oocyte–1 h–1)

a

7.6 7.2 6.8 6.4

c

1.2 1.0

100 μm

0.6

W at er At SW EE T9 Br SW EE At SW T9 EE T9 Br m SW EE T9 m Col-0

e

Atsweet9-1

10 μm

f

*

Col-0

g

Atsweet9-1

b

0.4 0.2 10

0

20

30

40

50

100 μm

60

m 100 μm

100 μm m

Time (min)

e

d

c

Col-0

0.8

0

0

d

AtSWEET9 BrSWEET9 AtSWEET9m BrSWEET9m

Atsweet9-1

pAtSWEET9-AtSWEET9–GFP/Col0

Figure 1 | AtSWEET9, a sucrose transporter, is necessary for nectar secretion. a, b, Sucrose uptake (a) and efflux (b) activity of AtSWEET9 and BrSWEET9 were determined in Xenopus oocytes. Truncated AtSWEET9(F201*) (AtSWEET9m) and BrSWEET9(L201*) (BrSWEET9m) served as negative controls. a, Oocyte uptake assay: AtSWEET9 and BrSWEET9 mediate 14C-sucrose uptake (error bars show 6 standard error of the mean (s.e.m.), n 5 14, n 5 16). *P , 0.05, **P , 0.01, Student’s t-test. b, Oocyte efflux assay: 14C-sucrose efflux by AtSWEET9 and BrSWEET9 in Xenopus oocytes injected with 14C-sucrose (error bars show 6 s.e.m., n 5 9, n 5 8). c, Nectar droplet clinging to inside of sepal (wild type: Col-0). d, Lack of nectar in nectaries of Atsweet9-1 mutants. e, Increased nectar secreted from nectaries of flowers containing extra copies of AtSWEET9–eGFP, driven by the AtSWEET9 promoter (pAtSWEET9). c–e, Original magnification, 310.

AtSWEET9–eGFP in floral nectaries increased during maturation, becoming highest during maximal nectar secretion (Extended Data Fig. 6). These data do not support a role for AtSWEET9 in phloem unloading, but do not enable us to distinguish between roles in uptake versus efflux from nectarial parenchyma. We conjectured that starch accumulation in Atsweet9 nectaries would be different if SWEET9 were involved in uptake (no starch accumulation in nectary) versus cellular efflux from nectarial cells (accumulation in nectary due to export deficiency). In wildtype plants, starch accumulates in plastids of nectary parenchyma before anthesis and is degraded to produce sugars for secretion16,18. Mutants accumulated starch in all cells of the nectary parenchyma, indicating that AtSWEET9 is responsible for cellular sugar efflux (Fig. 2d–g). Interestingly, at anthesis, nectarial guard cells of wild-type, but not Atsweet9 plants, contained starch granules, possibly due to reabsorption of sugars from nectar19. The functional characterization of AtSWEET9 as a sucrose efflux transporter, the expression of SWEET9 in nectary parenchyma and the starch accumulation pattern in mutants intimate AtSWEET9 as the key transporter for nectarial sugar secretion. The involvement of a putative facilitated-diffusion carrier indicates that high cytosolic levels of sugars in the nectarial parenchyma together with extracellular hydrolysis of sucrose by apoplasmic invertase create the driving force for nectar secretion9. Whether sugar secretion is mediated by plasma membrane carriers or exocytosis has been a matter of debate20. AtSWEET9–eGFP fusions localized at the plasma membrane, the trans-Golgi apparatus and multivesicular bodies (Fig. 2c, Extended Data Fig. 7 and Supplementary Videos 1–3). Given the strength of the SWEET9 promoter, accumulation in the trans-Golgi/multivesicular bodies is probably caused by high

Figure 2 | Cellular and subcellular localization of AtSWEET9 and starch accumulation in Atsweet9 mutants. a, b, Histochemical GUS analysis in Arabidopsis flowers expressing translational GUS fusion of AtSWEET9 (native promoter). Vertical (a) and transverse (b) sections of Arabidopsis flowers show tissue-specific localization of AtSWEET9. Cell walls stained with safranin-O (orange). c, Confocal images of eGFP fluorescence of an AtSWEET9 promoter:AtSWEET9–eGFP fusion showing subcellular localization at the plasma membrane and trans-Golgi network. d, e, Close-up of nectaries from wild-type and Atsweet9-1 flowers stained with Lugol’s iodine solution. Starch accumulated only in the guard cells of wild-type nectaries and in nectary parenchyma in Atsweet9-1 (sampled at the end of the dark period). f, g, LR White resin sections of Arabidopsis nectaries in wild type and Atsweet9-1 mutants stained with Lugol’s iodine solution. Starch grains (dark red) accumulate in nectaries of Atsweet9-1 mutants (g) and in stomata of wild-type nectary (f, indicated by an asterisk). Starch grains in floral stalks and nectaries in wild-type and Atsweet9 mutant lines at anthesis. Cell walls stained with safranin-O (orange).

expression levels21. Thus, although it is conceivable that AtSWEET9 loads vesicles with sugars for exocytosis, plasma-membrane-mediated efflux is more likely. Given that multiple clade 3 SWEETs transport sucrose, we explored whether the well-characterized plasma-membranelocalized paralogues AtSWEET11 and 12 (ref. 15) can complement the loss of nectar secretion in Atsweet9. When expressed from the AtSWEET9 promoter, both transporters restored nectar secretion (Extended Data Figs 5g, h, 6j–k and Supplementary Video 4), indicating that a sugar efflux transporter driven by a promoter that expresses it in nectaries suffices for sugar efflux. Gene expression analysis in Arabidopsis has shown that sucrose biosynthesis genes are upregulated in nectaries17, intimating that resynthesis of sucrose derived from starch fuels sugar efflux. We found that two SPS genes, SPS1F and SPS2F, which encode key enzymes for sucrose biosynthesis, were highly expressed in maturing nectaries (Extended Data Fig. 8). Artificial microRNA (amiRNA) inhibition of SPS gene expression led to loss of nectar secretion and increased starch accumulation (Fig. 3a–d). Thus, the phenotypes of Atsweet9 mutants, SPS amiRNA lines and the nectarial apoplasmic invertase mutant Atcwinv4-1 (ref. 22) are similar. The most parsimonious hypothesis is thus that starch-derived sucrose synthesized in nectaries is exported by SWEET9, and that sucrose hydrolysis by CWINV4 is necessary to create a large enough osmotic gradient to sustain water secretion (Fig. 3e). Species secreting nectar with a very high sucrose content may use additional mechanisms for creating a sufficient osmotic gradient. To explore whether SWEET9 of other Brassicaceae is essential for sugar efflux from nectaries, we identified the orthologous B. rapa (Br)SWEET9 in turnip flowers and showed that it transports sucrose (Fig. 1a, b) and is essential for sugar efflux and nectar secretion (Fig. 4a, b and Extended Data Fig. 9a). Nectar from the rosids Arabidopsis and turnip contains predominantly hexoses9, consistent with the role of apoplasmic invertase in post-secretory sucrose hydrolysis. Other rosids may also use SWEET9 for nectar secretion; for example, microarray data indicate that SWEET9 has a role in nectar secretion from extrastaminal nectaries as it is the highest expressed SWEET in barrel clover (Medicago truncatula) flowers (Extended Data Fig. 10).

2 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 4

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH a

b Atsps1f/2f amiRNA

Col-0

Figure 3 | SPS1 and SPS2 are necessary for nectar secretion in Arabidopsis. a, b, amiRNA inhibition of AtSPS1 and AtSPS2 gene expression led to a loss of nectar secretion. The circles indicate nectar droplets in wild-type flowers (a), and the absence of nectar in mutant flowers (b). c, d, Starch accumulation in nectaries of Atsps miRNA lines (floral stalk of sps1f/2f mutant lines, red arrow; only in guard cells of wild-type nectaries). Original magnification, 310. e, Model for nectar secretion: starch-derived sucrose is synthesized in nectaries by SPS proteins and exported by SWEET9. Exported sucrose is subsequently hydrolysed by CWINV4, which creates a high osmotic potential to sustain water flow down the osmotic gradient. Frc6P, fructose 6-phosphate; UDPglc, uracil diphosphate glucose.

e Water

Water Evaporation

Starch

c

Nectary parenchyma

100 μm

Hexose reuptake

100 μm Col-0 d Atsps1f/2f amiRNA

Glucose

UDPglc

Fructose

Frc6P SPS1F/2F

CWINV4

SPP

SWEET9 Sucrose Sucrose

a

(Fig. 4e, f). NaSWEET9 was essential for nectar secretion as shown in two independent RNA interference (RNAi) lines (Fig. 4d and Extended Data Fig. 9). Together, our results indicate that SWEET9 either arose early during core eudicot evolution or has been co-opted independently in asterids and rosids for gynoecial and both intra- and extrastaminal nectary function. A phylogenetic analysis tentatively traces the appearance of SWEET9 to a time window nearly at the origin of core eudicots

b

F nectaries EF nectaries SWEET9

Nectaries probably evolved multiple times23. Solanales (asterids) have gynoecial nectaries, whereas Brassicales typically have extra- and intrastaminal nectaries. To identify potential functional orthologues from coyote tobacco, SWEETs were cloned using degenerate primers (Supplementary Table 3). N. attenuata (Na)SWEET9 was most highly expressed in gynoecial nectaries; levels increased during nectary maturation (Fig. 4c). NaSWEET9 mediated sucrose uptake and efflux when expressed in oocytes

g

Asterids

d

3

Legumes

2

Rosids

ty pe ee t9 -1 N as w ee t9 -2

ild

**

1.0 0.5 0

0

10 20 30 40 50 60 Time (min)

Figure 4 | SWEET9 orthologues in B. rapa and N. attenuata are essential for nectar secretion. a, Nectar droplets in lateral nectary of wild-type (R-o-18) B. rapa flowers. b, Lack of nectar in Brsweet9-1 mutants. c, NaSWEET9 transcript accumulation in N. attenuata (error bars show 6 s.e.m., n 5 3). d, Mean (6 s.e.m., n 5 8) nectar volume of flowers measured at 05:00 in wild-type, Nasweet9-1 and Nasweet9-2 plants. e, f, Sucrose uptake (e) and efflux (f) activity of NaSWEET9 in oocytes. Truncated version of NaSWEET9(L201*) (NaSWEET9m) served as control. e, Oocyte uptake: NaSWEET9 mediates 14 C-sucrose uptake (errors bars show 6 s.e.m., n 5 14). f,14C-sucrose efflux by

Vascular plants Flowering plants

1.5

Ranunculales

Land plants

2.0

16 14 12 10 8 6 4 2 0

Green plants

Sucrose uptake (pmol oocyte–1 h–1)

Sucrose efflux (pmol oocyte–1)

2.5

as w

N

W

f 3.0 NaSWEET9 NaSWEET9m

Malpighiales

Eudicots

1

Monocots

Green algae

1 1 1 1 2 1 2 1 1 2 1 1 2 1 2 1 1 1 1 1 1 1

P. trichocarpa T. cacao C. papaya A. lyrata A. thaliana B. rapa C. sinensis A. caerulea

F - -

P. dactylifera M. acuminata Z. mays S. bicolor B. distachyon H. vulgare T. aestivum O. sativa A. trichopoda S. moellendorffii P. patens C. reinhardtii

S -

M. truncatula L. japonicus

4

0

Rosette leaf Stem Stigma

Anthers green

e

Brsweet9-1

5

Nectar volume (μl flower–1)

70 60 50 40 30 20 10 0 Anthers yellow Corolla Nectary green Nectary yellow Nectary orange Ovary yellow Ovary orange

NaSWEET9 signal intensity

c

500 μm

R-o-18

W at er N aS W EE T9 N aS W EE T9 m

500 μm

EF EF EF EF -

F F F F F F F F F F F F F F F F F F F

P. hybrida N. attenuata S. lycopersicum V. vinifera C. sativus C. melo P. persica F. vesca M. domestica

P. vulgaris G. max R. communis M. esculenta

-

-

NaSWEET9 in oocytes (error bars show 6 s.e.m., n 5 10). **P , 0.01, Student’s t-test. g, Data were collected from genome databases (http://phytozome.org; http://genomevolution.org; http://bioinformatics. psb.ugent.be/plaza/) using AtSWEET9 as bait. Tree branches are a schematic representation (Angiosperm Phylogeny Group III system). Species belonging to the core eudicot clades of rosids or asterids are underlined in red and green, respectively; full names are listed in Supplementary Table 5. Species bearing floral (F), extrafloral (EF) and septal (S) nectaries are indicated.

0 0 M O N T H 2 0 1 4 | VO L 0 0 0 | N AT U R E | 3

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER ,120 million years ago24 (Extended Data Figs 1, 2). In all genomes analysed, including grasses, Selaginella and Physcomitrella, we found multiple SWEET paralogues. Thus SWEET9 may have evolved from other clade 3 members when core eudicots merged. SWEET9 seems to be absent from wind-pollinated rice and maize lacking nectaries, banana, which has a typical monocot septal nectary, and Amborella, a member of an early angiosperm lineage. Aquilegia is a highly derived genus within the Ranunculales at the base of the eudicots25 and relies on entomophilous pollination mediated by elaborate nectar-filled nectar spurs. However, its genome seems to lack SWEET9 orthologues (Phytozome search, October 2013). The hypothesis of a single origin of SWEET9 during the evolution of core eudicots is consistent with conclusions from analyses of the master regulator of nectary development CRABS CLAW26. Castor bean and cassava, which produce nectar from EFNs, also contain SWEET9 homologues27 (Fig. 4g and Extended Data Figs 1, 2). We found that the wind-pollinated poplar, which lacks functional FNs but has foliar EFNs, contains SWEET9 and SWEET10 homologues. The sucrose transporter Populus trichocarpa (Pt)SWEET10a (Extended Data Fig. 4) is highly expressed in EFNs28, indicating that the foliar EFN uses SWEET9 paralogues for nectar secretion. Poplar nectar is hexose dominant, potentially due to cell wall invertases, which are highly expressed in the EFN28. These EFNs probably have a role in attracting ants as a defence against other insects29. Starch accumulation in floral stalks of the mutants indicates that phloem-derived sucrose is imported into nectaries symplasmically. Sucrose is then hydrolysed and stored as hexoses in the vacuole, or as starch18. During nectary maturation, sucrose is resynthesized via SPS, and SWEET9 begins to export sucrose down a concentration gradient, leading to sucrose accumulation in the apoplasm (Fig. 3e). As SWEET9 seems to function as a uniporter, and because the cytosol contains other solutes that contribute to the osmotic potential, uniporter-driven efflux is insufficient for osmotically driven water secretion. Thus, apoplasmic sucrose has to be hydrolysed by secreted invertases to produce glucose and fructose, increasing the osmotic driving force and allowing water to be secreted. Ultimately, sugary nectar is secreted through open stomata. Together, our results show that SWEET9 serves as a sugar efflux transporter of the nectary parenchyma and is necessary for the secretion of nectar in two types of eudicot nectaries. Microarray data show that hexose-transporting sugar transporters (STPs) are expressed in nectaries12, indicating that they have a role in reuptake. The relative activities of apoplasmic invertase combined with selective reuptake possibly determine the final ratio of sucrose, fructose and glucose.

5.

METHODS SUMMARY

Supplementary Information is available in the online version of the paper.

Heterologous expression of SWEET9 from in HEK293T cells and Xenopus oocytes was performed as described15. Mutant Arabidopsis was obtained from the Arabidopsis Biological Research Center (ABRC), insertion sites and transcript levels were verified by PCR and qRT–PCR, respectively. BrSWEET9 TILLING mutants were obtained from RevGen UK. NaSWEET9 was silenced in N. attenuata by RNAi (inverted repeat constructs). SPS1F and SPS2F were co-silenced through a single amiRNA targeting the messenger RNAs for both genes. Nectar secretion was evaluated microscopically. Starch was stained using potassium iodide22.

Acknowledgements We are grateful to D. Ehrhardt, H. Cartwright, J. Lindeboom, K. Barton and T. Liu for help with confocal and scanning electron microscopy. We thank D. Ehrhardt for providing specific endomembrane markers and M. Jia for conducting nectar sugar assays. We thank J. Danielson for help with phylogenetic analyses. This work was made possible by support from the Division of Chemical Sciences, Geosciences and Biosciences, Office of Basic Energy Sciences at the US Department of Energy under grant number DE-FG02-04ER15542 to W.B.F. I.W.L. was supported by the fellowship of Department of Biology, Stanford University and Carnegie. X.-Q.Q. was supported by the Carnegie Institution and Scholarship Program of the Chinese Scholarship Council (2009635108). C.J.C.’s work was supported by a grant from the US National Science Foundation (#0820730). I.T.B. was supported by European Research Council advanced grant ClockworkGreen (293926) and the Max Planck Gesellschaft, and thanks C. Diezel for technical assistance.

Online Content Any additional Methods, Extended Data display items and Source Data are available in the online version of the paper; references unique to these sections appear only in the online paper. Received 12 June 2013; accepted 27 January 2014. Published online 16 March 2014. 1. 2. 3. 4.

Sargent, R. D. Floral symmetry affects speciation rates in angiosperms. Proc. R. Soc. Lond. B 271, 603–608 (2004). Nicolson, S. W., Nepi, M. & Pacini, E. Nectaries and Nectar (Springer, 2007). Kay, K. M. et al. Floral Characters and Species Diversification (Oxford Univ. Press, 2006). Friedman, W. E. The meaning of Darwin’s ‘abominable mystery’. Am. J. Bot. 96, 5–21 (2009).

6. 7. 8. 9.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

De la Barrera, E. & Nobel, P. S. Nectar: properties, floral aspects, and speculations on origin. Trends Plant Sci. 9, 65–69 (2004). Kessler, D., Gase, K. & Baldwin, I. T. Field experiments with transformed plants reveal the sense of floral scents. Science 321, 1200–1202 (2008). Kessler, D. et al. Unpredictability of nectar nicotine promotes outcrossing by hummingbirds in Nicotiana attenuata. Plant J. 71, 529–538 (2012). Sime, K. R. & Baldwin, I. T. Opportunistic out-crossing in Nicotiana attenuata (Solanaceae), a predominantly self-fertilizing native tobacco. BMC Ecol. 3, 6 (2003). Davis, A. R., Pylatuik, J. D., Paradis, J. C. & Low, N. H. Nectar-carbohydrate production and composition vary in relation to nectary anatomy and location within individual flowers of several species of Brassicaceae. Planta 205, 305–318 (1998). Isokawa, S. et al. Novel self-compatible lines of Brassica rapa L. isolated from the Japanese bulk-populations. Genes Genet. Syst. 85, 87–96 (2010). Huang, M. et al. The major volatile organic compound emitted from Arabidopsis thaliana flowers, the sesquiterpene (E)-b-caryophyllene, is a defense against a bacterial pathogen. New Phytol. 193, 997–1008 (2012). Kram, B. W. & Carter, C. J. Arabidopsis thaliana as a model for functional nectary analysis. Sex. Plant Reprod. 22, 235–246 (2009). Hoffmann, M. H. et al. Flower visitors in a natural population of Arabidopsis thaliana. Plant Biol. 5, 491–494 (2003). Chen, L. Q. et al. Sugar transporters for intercellular exchange and nutrition of pathogens. Nature 468, 527–532 (2010). Chen, L. Q. et al. Sucrose efflux mediated by SWEET proteins as a key step for phloem transport. Science 335, 207–211 (2012). Ge, Y. X. et al. NEC1, a novel gene, highly expressed in nectary tissue of Petunia hybrida. Plant J. 24, 725–734 (2000). Kram, B. W., Xu, W. W. & Carter, C. J. Uncovering the Arabidopsis thaliana nectary transcriptome: investigation of differential gene expression in floral nectariferous tissues. BMC Plant Biol. 9, 92 (2009). Ren, G. et al. Transient starch metabolism in ornamental tobacco floral nectaries regulates nectar composition and release. Plant Sci. 173, 277–290 (2007). Langenberger, M. W. & Davis, A. R. Temporal changes in floral nectar production, reabsorption, and composition associated with dichogamy in annual caraway (Carum carvi; Apiaceae). Am. J. Bot. 89, 1588–1598 (2002). Fahn, A. Structure and function of secretory cells. Adv. Bot. Res. 31, 37–75 (2000). Gomord, V. et al. The C-terminal HDEL sequence is sufficient for retention of secretory proteins in the endoplasmic reticulum (ER) but promotes vacuolar targeting of proteins that escape the ER. Plant J. 11, 313–325 (1997). Ruhlmann, J. M., Kram, B. W. & Carter, C. J. CELL WALL INVERTASE 4 is required for nectar production in Arabidopsis. J. Exp. Bot. 61, 395–404 (2010). Johnson, S. D., Hobbhahn, N. & Bytebier, B. Ancestral deceit and labile evolution of nectar production in the African orchid genus Disa. Biol. Lett. 9, 20130500 (2013). Bell, C. D., Soltis, D. E. & Soltis, P. S. The age and diversification of the angiosperms re-revisited. Am. J. Bot. 97, 1296–1303 (2010). Fior, S. et al. Spatiotemporal reconstruction of the Aquilegia rapid radiation through next-generation sequencing of rapidly evolving cpDNA regions. New Phytol. 198, 579–592 (2013). Lee, J. Y. et al. Recruitment of CRABS CLAW to promote nectary development within the eudicot clade. Development 132, 5021–5032 (2005). Rizzardo, R. A., Milfont, M. O., Silva, E. M. & Freitas, B. M. Apis mellifera pollination improves agronomic productivity of anemophilous castor bean (Ricinus communis). An. Acad. Bras. Cienc. 84, 1137–1145 (2012). Escalante-Pe´rez, M. et al. Poplar extrafloral nectaries: two types, two strategies of indirect defenses against herbivores. Plant Physiol. 159, 1176–1191 (2012). Heil, M., Rattke, J. & Boland, W. Postsecretory hydrolysis of nectar sucrose and specialization in ant/plant mutualism. Science 308, 560–563 (2005).

Author Contributions I.W.L., L.-Q.C., C.J.C., I.T.B. and W.B.F. conceived and designed experiments. I.W.L., L.-Q.C., X.-Q.Q., D.S., B.-H.H., K.G., S.-G.K., D.K., P.M.K. and M.K.G. performed experiments. I.W.L., W.B.F., L.-Q.C., X.-Q.Q., D.S., B.-H.H., S.O., P.K., C.J.C. and A.R.F. analysed the data. I.W.L. and W.B.F. wrote the manuscript and I.T.B. and C.J.C revised it. Author Information The sequence for a full-length cDNA for BrSWEET9 is available in the NCBI Reference Sequence database under accession number KC790460. Reprints and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests. Readers are welcome to comment on the online version of the paper. Correspondence and requests for materials should be addressed to W.B.F. ([email protected]).

4 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 4

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH METHODS Constructs for expression in Xenopus oocytes. The open reading frames (ORFs) of AtSWEET9, BrSWEET9 and NaSWEET9 (with stop codon) in vector pDONR221f1 were transferred to the oocyte expression vector pOO2-GW as described previously for other SWEETs14. Non-functional, truncated versions of AtSWEET9(F201*), BrSWEET9(L201*) and NaSWEET9(L201*) were generated by introducing stop codons in transmembrane helix 7 by site-directed mutagenesis (QuickChange II Site-Directed Mutagenesis Kits, Agilent Technology). Primers are listed in Supplementary Table 4. We predicted that proteins lacking parts of the 7th transmembrane spanning domain would not be able to mediate transport. We had previously shown that mutations that lead to truncation in the 7th transmembrane spanning domain lead to loss of function in plant and human SWEET homologues14,15. As predicted, also the mutants shown here are non-functional, and can be used as controls for transport assays. FRET analysis in HEK293T cells. HEK293T cells were co-transfected with FRET glucose sensor FLII12Pglu700mh6 or FRET sucrose sensor FLIPsuc90md3A and different SWEETs including AtSWEET9, BrSWEET9 and PtSWEET9 in 6-well plates, and perfusion experiments were performed as described before30. 25 mM glucose, 25 mM sucrose and 40 mM sucrose were applied to cells after 3–6 min of buffer perfusion. Plasmids for complementation of mutants. For complementation of the Atsweet9 mutant, a 3,327 bp genomic sequence consisting of a 1,815 bp promoter and 1,512 bp of the entire coding region without stop codon from AtSWEET9 was amplified from Col-0 genomic DNA using primers AtSWT9attB1 and AtSWT9attB2. The genomic AtSWEET9 fragment was cloned into the Gateway donor vector pDONR221-f1 (ref. 14) and the pAtSWEET9:AtSWEET9 cassette were re-amplified with the forward primer AtSWT9TKan2F containing a BamHI restriction site and the reverse primer AtSWT9TKan2R containing a PstI restriction site and subcloned into vector pTKan2 (ref. 31) via BamHI and PstI restriction sites. GUS and eGFP fusion constructs under native promoters. For analysing the expression of SWEETs via GUS fusions, the same fragments as used for generating the complementation constructs (promoter and gene including introns for AtSWEET9) were transferred by LR recombination reactions into the plant Gateway vector pMDC163 pGWB3 carrying the GUS gene32. The GUS gene was translationally fused to the carboxy terminus of AtSWEET9. To generate translational GFP fusion constructs, the pAtSWEET9:AtSWEET9 cassette was re-amplified with the forward primer AtSWT9GTKan3F containing a BamHI restriction site and the reverse primer AtSWT9GTKan3R containing a PstI restriction site and subcloned into the eGFP fusion vector pGTKan3 (ref. 31) via BamHI and PstI restriction sites. Construct for NaSWEET9 silencing. For the construction of the inverted repeat plasmid pSOL8SWEET9 used for NaSWEET9 silencing, a partial 311 bp sequence of NaSWEET9 cDNA was amplified by PCR using the primer pairs NEC1-34 and NEC2-34, and NEC3-33 and NEC4-34 (Supplementary Table 4). The 0.6 kb HindIIIPstI and SacI-XhoI inverted repeat fragments of vector pSOL8DC3 (ref. 33) (GenBank HQ698853) were then subsequently replaced with these two PCR fragments cut with the same enzymes. Tracer uptake and tracer efflux in Xenopus oocytes. The method has been described previously15. Linearization of the plasmids in pOO2 vector, capped cRNA synthesis, Xenopus oocytes isolation and cRNA injection, [14C]-labelled sugar uptake and efflux were carried out as described for SWEET1 analysis previously14. As controls we used either oocytes injected with water (here 50 nl RNase-free water was injected instead of cRNA, a standard control for oocyte experiments) or oocytes injected with cRNA for truncated versions of the respective SWEET9 transporters. Both controls yielded comparable data regarding background uptake or efflux. For efflux assays, oocytes were injected on day 3 after injection of the cRNA with 50 nl solution containing 10 mM sucrose (0.18 mCi ml21 [14C(U)] sucrose). The oocyte water space was estimated to be 450 nl (500 nl acutely after 50 nl injection of the radiolabelled sugar solution into the oocyte) as described previously34. Others have estimated the volume to be 1 ml, thus our estimate of the final sugar concentration is coarse and will vary with the batch of oocytes. Plant material and growth conditions. Arabidopsis plants were grown under low light (90–110 mE m22 s21 with 10 h photoperiod) conditions. For the transgenic plants of complemented, GUS and eGFP A. thaliana plants (wild-type Col-0 and Atsweet9 mutants) were transformed by the floral dip method35. Transgenic seedlings were selected on Murashige and Skoog (MS) media with kanamycin (70 mg ml21) (pAtSWEET9:AtSWEET9 and pAtSWEET9:AtSWEET9–eGFP) or hygromycin (40 mg ml21) (pAtSWEET9:AtSWEET9–GUS). BrSWEET9 (Bra000116) is a single copy gene (as identified by BLAST search against the full B. rapa genome (http://brassicadb.org/brad/blastPage.php)) consisting of six exons over 1,513 bp, with an ORF of 813 nucleotides (cDNA accession number KC790460). B. rapa growth was performed in a 16 h light/8 h dark cycle, with a photosynthetic photon flux of 150 mmol m22 s21, at a temperature of 21 uC.

BrSWEET9 was previously described as a gene with nectary-enriched expression in mature flowers36. Wild-type N. attenuata plants were grown from seeds that were field collected from a native population (DI Ranch) and subsequently inbred for 30 generations. These same wild-type lines were transformed by Agrobacterium tumefaciens (strain LBA 4404) to silence Nasweet9. For the cloning of NaSWEET, equal masses of flower organs (corolla, ovary, anther, sepal, leaf and nectary) were harvested and total RNAs were isolated. Original primers to clone NaSWEET9 transcripts were designed based on the NEC1 sequence from Petunia hybrid16 and later full-length sequence of NaSWEET9 was validated using 454-transcriptome sequencing of N. attenuata. A partial sequence of NaSWEET cDNA was used to construct plant silencing vector pSOL8SWEET9 (see above). Transformation procedure has been described previously37. We selected two independent homozygous lines each harbouring single insertions with high silencing efficiencies in the nectaries (Supplementary Figs 10, 12). Genotyping and transcript analysis of mutants. Genomic DNA was extracted from A. thaliana Col-0 and the T-DNA insertion lines (Atsweet9-1, sk225 (T-DNA insertion in position 2308 before the start codon; no detectable transcript); Atsweet92, SALK_060256 (position 2940 before the start codon; reduced transcript level); Atsweet9-3, SALK_202913C (position 779 after the start codon in exon 4), and used as a template for PCR-based genotyping. Primers specific to AtSWEET9 sequences flanking the T-DNA insertion sites were designed from http://signal.salk.edu/ tdnaprimers.2.html (AtSWT9-1LP and AtSWT9-1RP for ssk225; AtSWT9-2LP and AtSWT9-2RP for SALK_060256; and AtSWT9-3LP and AtSWT9-3RP for SALK_202913C). The sequence for the left border primers of pSKtialL3 (for sk225) and LBb1.3 (for SALK_060256 and SALK_202913C) were also obtained from the SALK website (http://signal.salk.edu/). These primers were used to detect the presence of the T-DNA insert. PCR was performed as described on the SALK website. For RT–PCR analyses, total RNA was extracted from leaves and flowers of Arabidopsis from Col-0 and insertion lines using a Spectrum plant total RNA kit (Sigma #120M6117). For sk225 CS1000101 and SALK_060256, first-strand cDNA was synthesized using oligonucleotide dT and M-MuLV Reverse Transcriptase following the instructions of the supplier (Fermentas). Primers for the 751 bp ORF of AtSWEET9 (AtSWT9_2145F and AtSWT9_qPCR R) were used for RT–PCR to determine the expression levels. AtACTIN2 (primers AtACT2F and AtACT2R) served as reference gene. For SALK_202913C (Atsweet9-3), PCR reactions from genomic DNA with the LBb1.3 and AtSWT9-3RP primer pair were sequenced to identify the precise T-DNA insertion site. Homozygous mutants were subsequently evaluated for expression via RT–PCR with the AT2G39060 RT Fwd and AT2G39060 RT Rev primers (Supplementary Table 4) (with AtGAPDH as an internal control) as previously described for wild-type A. thaliana (AtSWEET9 was formerly known as nodulin MtN3 family protein17). Original raw data for the analysis of SWEET family expression in Arabidopsis nectaries via microarray and RNA-seq counts (Supplementary Figs 2, 3 and Supplementary Table 1) were previously described17,38. BrSWEET9 TILLING mutants were obtained from RevGen UK (http://revgenuk. jic.ac.uk/) via screening of previously described mutant populations39. Brsweet9-1 (RevGen UK #JI33127-B) resulted from a transition of nucleotide 407 of the ORF from G to A, causing a premature stop codon (TGGRTAG; W136stop). Brsweet92 (JI32885-B) and Brsweet9-3 (JI32947-B) both resulted from splice site mutations, with G to A transitions at positions 1524 (second intron) and 1827 (third intron) from the start codon of the pre-mRNA (including introns). Each mutant line was backcrossed to the parental line (B. rapa cv. R-o-18), with the subsequent generation being allowed to self-pollinate. DNA from individual lines was then amplified with the primers BrSWEET9 TILLING SEQ F and BrSWEET9 TILLING SEQ R and the product sequenced to identify homozygous mutants. Mutant phenotyping consisted of removal of the short stamen to reveal the lateral and median nectaries to observe nectar droplets. To determine the T-DNA insertion number (Supplementary Fig. 11), genomic DNA from the stably silenced lines and wild-type N. attenuata plants was isolated, digested with Xbal (New England Biolabs) overnight at 37 uC according to the manufacturer’s instructions, and separated on a 0.8% (wt/vol) agarose gel using standard conditions. DNA was blotted onto Gene Screen Plus Hybridization-Transfer membranes (Perkin-Elmer Life and Analytical Sciences) using the capillary transfer method. A specific probe for the hygromycin resistance gene HPTII was labelled with [a-32P] dCTP (Perkin-Elmer) using the RediprimeII kit (GE Healthcare) according to the manufacturer’s instructions. We collected 20–100 mg of tissues: anther, corolla, nectary, ovary, stigma, rosette leaf and stem, to examine tissue-specific expression of NaSWEET. Total RNA was extracted with TRIzol (Invitrogen) and was treated with RNase-free DNaseI (Fermantas) to eliminate contaminating genomic DNA. The first-strand cDNA was synthesized with 1–2 mg of total RNA using SuperScript-II Reverse Transcriptase kit (Invitrogen). Quantitative real-time PCR was performed using the SYBR green reagent (qPCR Core Kit for SYBR Green I; Eurogentec) in a Stratagene 500 Mx3005P (Stratagene) instrument. Gene-specific

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER primers were used to detect NaSWEET9 transcripts and eukaryotic elongation factor 1A-a (as an internal standard) were used to calculate expression levels. SPS1F (At5g20280) and SPS2F (At5g11110) were co-silenced via a single amiRNA targeting the mRNAs for both genes. Primers for generating DNA encoding the amiRNA (Supplementary Table 4) were designed and used to modify MIR319a by PCR as previously described40. The resultant product was directionally cloned into the EcoRI and SpeI sites of the plant transformation vector pPMK1, which contains 1.44 kb of the SWEET9 promoter upstream of the multiple cloning site and also confers resistance to kanamycin to transformed plants41. Starch staining. Flowers (stage 14–15, at anthesis) were examined for starch accumulation by iodine–potassium iodide (IKI) staining as previously described22. Freshly collected flowers were placed in IKI solution and subjected to a 1 min vacuum infiltration, followed by 5 min incubation at room temperature. Flowers were then rinsed twice in distilled water and immediately imaged after removing sepals. Plastic-embedded sections of wild-type and sweet9 plants were first stained with 0.1% (w/v) safranin-O for 3 min, washed three times with distilled water, stained with IKI for 20 min, and washed with distilled water. Sections were then mounted with CytoSeal 60 (Electron Microscopy Sciences). Plastic embedding and sectioning. Arabidopsis was grown until flower stage 14– 15 (at anthesis) as described previously15. Plastic embedding followed the protocol provided with the LR White embedding kit (Sigma). Semi-thin cross sections (1 mm) were cut by Ultracut (Reichert) and stained with 0.1% (w/v) safranin-O for 3 min, then washed three times with distilled water. Sections for starch staining were stained with Lugol’s solution IKI for 20 min and washed with distilled water. All sections were mounted with CytoSeal 60 (Electron Microscopy Science). GUS histochemistry. GUS staining was performed following standard procedures with minor changes42,43. Potassium ferrocyanide/ferricyanide was used at a final concentration of 5 mM. Staining intensity and diffusion were checked under a microscope and controlled by modulating incubation time at 37 uC. For cross-sections, flowers were stained for 6–8 h to reduce diffusion depending on the expression levels in the individual lines. Transient gene expression in N. benthamiana leaves. The method a previously described method44 with modifications. The Agrobacterium tumefaciens strain GV3101 was transformed with the binary expression clones. Bacterial cultures were grown, precipitated and dissolved in infiltration buffer (10 mM MES, pH 5.6, 10 mM MgCl2 and 200 mm acetosyringone) to the cell density of OD600 nm approximately 0.5–1 for 2 h. Aliquots of Agrobacterium cells containing each co-expression clone were mixed together and then a syringe was used to infiltrate the mixture into the lower surface of N. benthamiana leaves. Plants were incubated for 36–48 h before image. Nectar collection and glucose content analysis. Nectar collection was previously described41. Briefly, a single biological replicate for nectar glucose assays consisted of nectar collected from the lateral nectaries of ten stage 14–15 flowers using wicks cut from Whatman no. 1 filter paper. The wicks were placed in 250 ml 13 Reaction buffer, diluted ten times and the glucose contents were analysed by using Amplex Red Glucose/Glucose Oxidase Assay Kit (Life Technology). Fluorescence microscopy. Fluorescence imaging of plants was performed on a Leica TCS SP5 microscope. eGFP was visualized by standard procedures as described before14. Sepals were removed before imaging and flowers were mounted onto the cover glass in distilled water. Fluorescence was detected with a water-corrected 320 or glycerol-corrected 363 objective at 488 nm excitation and 492–600 nm emission using 30% transmittance at minimum laser power. Image analysis was performed using Fiji software (http://fiji.sc). Scanning electron microscopy. Sepals were removed from flowers before imaging. Samples were then mounted and fixed on stage by CRYO-GEL (EMS). Imaging was performed under low-vacuum conditions (60 Pa) and an accelerating voltage of 30 kV with an FEI Quanta 200 scanning electron microscope. Phylogenetic analysis. Protein sequences were retrieved from NCBI Blast (http:// blast.ncbi.nlm.nih.gov/Blast.cgi) and Phytozome (http://www.phytozome.net/search.

php) using AtSWEET9 and AtSWEET1 as baits. For non-fully annotated genome species CoGePedia (http://genomevolution.org/wiki/index.php) was used to refine the search; C. sinensis sequences were provided by F. F. White (personal communication). Sequences were aligned using ClustalW. The evolutionary history was inferred using the maximum likelihood method using a frequency model45. The tree with the highest log likelihood (234,944) is shown (Extended Data Fig. 1). Initial trees for the heuristic search were obtained automatically by applying NeighbourJoin and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model, followed by selecting the topology with superior log likelihood value. The analysis involved 200 amino acid sequences. All positions with ,95% site coverage were eliminated. The evolutionary history was inferred using the neighbourjoining method. The optimal tree with the sum of branch lengths of 40.89 is shown (Extended Data Fig. 2). The evolutionary distances were computed using the JTT matrix-based method and are in units of the number of amino acid substitutions per site. The analysis involved 200 amino acid sequences. All positions with less than 95% site coverage were eliminated. Analyses were conducted using MEGA5 (http://www.megasoftware.net). 30. Hou, B. H. et al. Optical sensors for monitoring dynamic changes of intracellular metabolite levels in mammalian cells. Nature Protocols 6, 1818–1833 (2011). 31. Kasaras, A. & Kunze, R. Expression, localisation and phylogeny of a novel family of plant-specific membrane proteins. Plant Biol (Stuttg) 12 (suppl. 1), 140–152 (2010). 32. Nakamura, S. et al. Gateway binary vectors with the bialaphos resistance gene, bar, as a selection marker for plant transformation. Biosci. Biotechnol. Biochem. 74, 1315–1319 (2010). 33. Gase, K., Weinhold, A., Bozorov, T., Schuck, S. & Baldwin, I. T. Efficient screening of transgenic plant lines for ecological research. Mol. Ecol. Resour. 11, 890–902 (2011). 34. Chernova, M. N. et al. Electrogenic sulfate/chloride exchange in Xenopus oocytes mediated by murine AE1 E699Q. J. Gen. Physiol. 109, 345–360 (1997). 35. Davis, A. M., Hall, A., Millar, A. J., Darrah, C. & Davis, S. J. Protocol: Streamlined sub-protocols for floral-dip transformation and selection of transformants in Arabidopsis thaliana. Plant Methods 5, 3 (2009). 36. Hampton, M. et al. Identification of differential gene expression in Brassica rapa nectaries through expressed sequence tag analysis. PLoS ONE 5, e8782 (2010). 37. Kru¨gel, T., Lim, M., Gase, K., Halitschke, R. & Baldwin, I. T. Agrobacterium-mediated transformation of Nicotiana attenuata, a model ecological expression system. Chemoecology 12, 177–183 (2002). 38. Bender, R. L. et al. PIN6 is required for nectary auxin response and short stamen development. Plant J. 74, 893–904 (2013). 39. Stephenson, P. et al. A rich TILLING resource for studying gene function in Brassica rapa. BMC Plant Biol. 10, 62 (2010). 40. Ossowski, S., Schwab, R. & Weigel, D. Gene silencing in plants using artificial microRNAs and other small RNAs. Plant J. 53, 674–690 (2008). 41. Bender, R. et al. Functional genomics of nectar production in the Brassicaceae. Flora 207, 491–496 (2012). 42. Bauby, H., Divol, F., Truernit, E., Grandjean, O. & Palauqui, J. C. Protophloem differentiation in early Arabidopsis thaliana development. Plant Cell Physiol. 48, 97–109 (2007). 43. Gallagher, S. R. GUS protocols: using the GUS gene as a reporter of gene expression. (Academic Press, 1992). 44. Bleckmann, A., Weidtkamp-Peters, S., Seidel, C. A. & Simon, R. Stem cell signaling in Arabidopsis requires CRN to localize CLV2 to the plasma membrane. Plant Physiol. 152, 166–176 (2010). 45. Jones, D. T., Taylor, W. R. & Thornton, J. M. The rapid generation of mutation data matrices from protein sequences. Computer applications in the biosciences. CABIOS 8, 275–282 (1992). 46. Smyth, D. R., Bowman, J. L. & Meyerowitz, E. M. Early flower development in Arabidopsis. Plant Cell 2, 755–767 (1990). 47. Benedito, V. A. et al. A gene expression atlas of the model legume Medicago truncatula. Plant J. 55, 504–513 (2008).

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 1 | SWEET9 molecular phylogenetic analysis using the neighbour-joining method. A phylogenetic tree of SWEET proteins collected from different species: A. thaliana (At), Manihot esculenta (Me), Solanum lycopersicum (Sl), P. trichocarpa (Pt), M. truncatula (Mt), Glycine max (Gm), Oryza sativa (Os), Amborella trichopoda (Ambo), Aqulegia caerulea (Ac), Physcomitrella patens (Pp) and Chlamydomonas reinhardtii (Cr).

Accessions are listed in Supplementary Table 6. The SWEET9 orthologue cluster (blue) is part of clade 3 and was completed with SWEET9 collected also from B. rapa (Br), Petunia hybrida (Ph), N. attenuata (Na), Theobroma cacao (Tc), Fragaria vesca (Fv), Ricinus communis (Rc), Malus domestica (Md) and Carica papaya (Cp). Bootstrap values are out of 1,000 replicates.

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

Extended Data Figure 2 | SWEET9 molecular phylogenetic analysis using the maximum-likelihood method. A phylogenetic tree of SWEET proteins collected from different species: A. thaliana (At), M. esculenta (Me), S. lycopersicum (Sl), P. trichocarpa (Pt), M. truncatula (Mt), G. max (Gm), O. sativa (Os), A. trichopoda (Ambo), A. caerulea (Ac), P. patens (Pp) and C. reinhardtii (Cr). Accessions are listed in Supplementary Table 6.

The SWEET9 orthologue cluster (red) is part of clade 3 and was completed with SWEET9 collected also from B. rapa (Br), P. hybrida (Ph), N. attenuata (Na) (denoted with an asterisk), T. cacao (Tc), F. vesca (Fv), R. communis (Rc), M. domestica (Md) and C. papaya (Cp). Bootstrap values are out of 1,000 replicates.

©2014 Macmillan Publishers Limited. All rights reserved

a

La te Probe set signal inetnsity ra ln e c la te tar r y m al n , im ed ec m ia n tary atu r ne , m e c C tary atu ar pe , m re at l, u C imm re ar pe atu Pe l, re ta ma l, t u im r Pe ma e t Se al, tur pa m e a l, im tu Se m re a St p am al, ture en ma St , im ture am m en atu , m re at ur Pe e tio le R R os e oo C tte t au L e Po l i n e af lle Le Pe n, a di ma f ce tu l, m re N a In od ture In tern e S h flo re ode oot sc en Sho ce ot Sh oo t

LETTER RESEARCH

16000 14000 12000

10000 8000 6000 4000 2000

b

Probe set signal inetnsity

0

16000

SWEET1 SWEET4 SWEET5 SWEET8 SWEET9 SWEET16 SWEET17

14000 12000 10000 8000 6000 4000 2000

c

ln

ed

ra

te

la

m

La

te

ra

ln

ec

ta

ry ,i ec mm ia n tary atu ne , m re c C tary atu ar r pe , m e at l, u C imm re ar pe atu re Pe l, ta ma l, t im ure Pe ma Se tal, tur pa m e a l, im ture m S a St ep am al ture , en m at , St im ure m am a en tur ,m e at u Pe re tio le R os R o ot et C te L au e Po line af ll L Pe en, ea di ma f ce l, ture m at N In ode ure t In ern S h flo o re de o o t sc S en ho ce ot Sh oo t

0

Name

Locus

AtSWEET1

AT1G21460

Read counts 27

AtSWEET2

AT3G14770

50

AtSWEET3

AT5G53190

0

AtSWEET4

AT3G28007

41

AtSWEET5

AT5G62850

3

AtSWEET6

AT1G66770

0

AtSWEET7

AT4G10850

3

AtSWEET8

AT5G40260

27

AtSWEET9

AT2G39060

598206

AtSWEET10

AT5G50790

0

AtSWEET11

AT3G48740

0

AtSWEET12

AT5G23660

0

AtSWEET13

AT5G50800

2

AtSWEET14

AT4G25010

6

AtSWEET15

AT5G13170

3

AtSWEET16

AT3G16690

0

AtSWEET17

AT4G15920

0

Extended Data Figure 3 | AtSWEET9 and other AtSWEET genes expression in Arabidopsis nectaries and reference tissues. a, Normalized mean ATH1 GeneChip probe set signal intensity for AtSWEET9 in nectaries and other tissues. Original data for all tissues were described previously17. Mean value derived from an n of 3 (6 s.e.m.) for all tissues except mature median nectaries

(n 5 2). b, Normalized mean ATH1 GeneChip probe set signal intensity for all Arabidopsis SWEET family proteins expressed in nectaries17. c, Normalized RNA-seq counts for AtSWEET family gene expression in mature lateral nectaries (n 5 24; original raw data from ref. 38).

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

D 2.1

FLII12Pglu700μδ6+BrSWEET9

1.02

25mM Sucrose

1

FLII12Pglu700μδ6+PtSWEET10 FLII12Pglu700μδ6+AtSWEET1

1.9

Norm. intensity ratio [535/470 nm]

Norm. intensity ratio [535/470 nm]

E

FLII12Pglu700μδ6 FLII12Pglu700μδ6+AtSWEET9

1.7

1.5

1.3

1.1

0.98 0.96 0.94 0.92 0.9 0.88 FLIPsuc90μδ3A FLIPsuc90μδ3A+AtSWEET9 FLIPsuc90μδ3A+BrSWEET9 FLIPsuc90μδ3A+PtSWEET10 FLIPsuc90μδ3A+AtSWEET12

0.86

25mM glucose

0.9 0

2

4

6

8

10

12

0.84

14

0

2

Time [min]

4

6

Time [min]

8

10

12

c 1D6:((7

1D6:((7

$W6:((7

$W6:((7

$W6:((7

$W6:((7

$W6:((7

$W6:((7

6F+;7

6F+;7

FRQWURO

FRQWURO

Extended Data Figure 4 | Sucrose, glucose and fructose transport activity of SWEETs in HEK293T cells and yeast cells. a, b, Detection of glucose and sucrose uptake activity in HEK293T cells using the FRET glucose sensor FLII12Pglu700md6 and FRET sucrose sensor FLIPsuc90md3A. a, Glucose transport activity as detected for by co-expression with the cytosolic FRET glucose sensor FLII12Pglu700md6 in HEK293T cells. Individual cells were analysed by quantitative ratio imaging of enhanced cyan fluorescent protein (eCFP) and Citrine emission (acquisition interval, 5 s). HEK293T cells were perfused with culture medium, followed by square pulses of increasing glucose concentrations. Grey circle indicates cells expressing sensor alone; green triangle, red diamond and blue square indicate cells co-expressing sensor and AtSWEET9, BrSWEET9 or PtSWEET10, respectively, and orange circle indicates the positive control AtSWEET1; accumulation of glucose is indicated by a positive FRET ratio change (mean 1 s.e.m.; n . 10). Experiments were repeated with comparable results at least four times. b, Sucrose transport activity as detected for by co-expression with the cytosolic FRET sucrose sensor

FLIPsuc90md3A in HEK293T cells. Individual cells were analysed by quantitative ratio imaging of eCFP and Aphrodite emission (acquisition interval, 10 s). HEK293T cells were perfused with culture medium, followed by square pulses of increasing sucrose concentrations. Grey circle indicates cells expressing sensor alone; green triangle, red diamond and blue square indicate cells co-expressing sensor, AtSWEET9 and BrSWEET9, respectively, circle indicates the positive control AtSWEET12 or PtSWEET10; accumulation of sucrose is indicated by a negative FRET ratio change (mean 1 s.e.m.; n . 10). Experiments were repeated with comparable results at least four times. c, Complementation of yeast EBY4000 (ref. 45) lacking 18 hexose transporter genes with AtSWEET1, NaSWEET9, AtSWEET9, AtSWEET11 or yeast HXT5 (positive control) transporters and empty vector (negative control). Left panel showed the medium containing glucose and right panel showed the medium containing fructose. AtSWEET1 showed glucose/fructose transport activity in yeast cells but NaSWEET9, AtSWEET9 and AtSWEET11 did not.

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

/)/)/)

Atsweet9-3

&RO

Atsweet9-2

Atsweet9-1

b &RO

a

AtSWEET9

AtSWEET9 DFWLQ

*$3'+

c

d

e

Atsweet9-2 pAtSWEET9-AtSWEET9/sweet9-2

pAtSWEET9-AtSWEET9-GFP/sweet9-1

f Col-0

gg

Atsweet9-3

hh

 pAtSWEET9-AtSWEET11/ Atsweet9-2 pAtSWEET9-AtSWEET12/ Atsweet9-1

i Col0

˩P

Extended Data Figure 5 | AtSWEET9 is necessary for nectar secretion. a, Identification of three Atsweet9 T-DNA insertion lines. RT–PCR (40 cycles) performed on RNA isolated from leaves (L) and whole flowers (F) of wild-type, Atsweet9-1 and Atsweet9-2 plants. AtSWEET9 expression can only be detected in wild-type flowers. Actin was used as a constitutively expressed control. b, RT–PCR performed on RNA isolated from wild-type and Atsweet9-3 (SALK_202913C) plants. GAPDH (At3g04120) was used as a constitutively expressed control. c, Lack of nectar in nectaries of Atsweet9-2 mutants. d, e, Nectar secreted from nectaries of complemented Atsweet9 mutants under its native promoter: AtSWEET9 (d) or AtSWEET9–eGFP (e). f, Nectar production in wild-type and Atsweet9-3 flowers. Atsweet9-3 mutant lines do

Atsweet9-1

 ˩P

not secrete nectar, similar to Atsweet9-1 and Atsweet9-2. g, h, Nectar production in Atsweet9 mutants complemented with AtSWEET9, 11 (g) or 12 (h) expressed under the control of the AtSWEET9 promoter. Nectar production was restored by expression of AtSWEET11 and AtSWEET12 under the AtSWEET9 promoter in the Atsweet9 mutant plants. Arrows indicate nectar droplet on the peeled-down sepals. i, Ultrastructure of lateral nectaries of wild-type and Atsweet9-1 mutant flowers. The morphology of wild-type and Atsweet9-1 mutant flowers was observed by scanning electron microscopy. Sepals were removed before imaging. As judged by scanning electron microcopy, mutant nectaries appeared normal, indicating that loss of nectar secretion was not caused by physical defects of nectaries. c–h, Original magnification, 310.

©2014 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER a

b

c

e

d

f

HSLGHUPLVFHOOV JXDUGFHOOV

˩P

)ORZHUVWDJHa /DWHUDOQHFWDULHV

g

˩P

˩P

l

)ORZHUVWDJHa PHGLXPQHFWDULHV

)ORZHUVWDJHa /DWHUDOQHFWDULHV

h

HSLGHUPLVFHOOV

˩P

i

˩P

˩P

)ORZHUVWDJHa S$W6:((7$W6:((7H*)3

j

k

˩P

m

Col-0

Atsweet9-1

Extended Data Figure 6 | Cellular and subcellular localization of AtSWEET9 and starch accumulation in Atsweet9 mutants. a–f, Histochemical GUS analysis in Arabidopsis flowers expressing translational GUS fusion of AtSWEET9 (native promoter). a, b, GUS staining in lateral (a) and median (b) nectaries. c–f, Transverse (c, d) and vertical (e, f) sections of Arabidopsis flowers showing tissue-specific localization of AtSWEET9. a, b, Original magnification, 310. Cell walls stained with safranin-O (orange). GUS activity was highest in the lower (basal) half of the nectary parenchyma, less in the top part of the parenchyma and weak or absent from epidermis and guard cells. g–i, Arabidopsis plants expressing translational AtSWEET9–eGFP fusions under control of its native promoter: lateral nectaries in an immature flower at floral stage 12–13 (unopened flower) (g), lateral nectaries in a mature flower at floral stage 14–15 (open flower)

(h) and median nectaries in a mature flower at floral stage 14–15 (i). Auto-fluorescence from chloroplasts in magenta. Fluorescence imaging was performed on a Leica SP5 confocal microscope. Shown are maximum projections of Z-stacks. The definition of flower stages is as described previously46. j–k, Confocal images of eGFP fluorescence of proAtSWEET9:AtSWEET11–eGFP fusion showing subcellular localization at plasma membrane and Golgi of lateral nectaries in a mature flower at floral stage 14–15 (open flower); and a close-up image of the subcellular localization of AtSWEET11–eGFP in the nectary cells (k). j, k, Scale bars, 5 mm. l, m, Flowers of wild-type (l) and Atsweet9-1 mutant (m) plants stained with Lugol’s iodine solution 4 h after dawn: starch in the floral stalk of Atsweet9-1. Arrow indicates starch accumulation in the petiole. l, m, Original magnification, 310.

©2014 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH $W6:((7*)3 AtSWEET9-GFP D

E

H