Golgi-localized LOT regulates trans-Golgi network ... · The trans-Golgi network (TGN) is an essential tubular-vesicular or-ganelle derived from the Golgi and functions as an independent

Golgi-localized LOT regulates trans-Golgi networkbiogenesis and pollen tube growthPeng-Fei Jiaa,b,1, Yong Xuea,b,1, Hong-Ju Lia,b,2, and Wei-Cai Yanga,b,2

aState Key Laboratory of Molecular Developmental Biology, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing100101, China; and bUniversity of Chinese Academy of Sciences, Beijing 100049, China

Edited by Zhenbiao Yang, University of California, Riverside, CA, and accepted by Editorial Board Member Joseph R. Ecker October 11, 2018 (received forreview May 29, 2018)

The trans-Golgi network (TGN) is an essential tubular-vesicular or-ganelle derived from the Golgi and functions as an independentsorting and trafficking hub within the cell. However, the molecularregulation of TGN biogenesis remains enigmatic. Here we identifiedan Arabidopsis mutant loss of TGN (lot) that is defective in TGNformation and sterile due to impaired pollen tube growth in thestyle. The mutation leads to overstacking of the Golgi cisternaeand significant reduction in the number of TGNs and vesicles sur-rounding the Golgi in pollen, which is corroborated by the dispersedcytosolic distribution of TGN-localized proteins. Consistently, depo-sition of extracellular pectin and plasma membrane localization ofkinases and phosphoinositide species are also impaired. Subcellularlocalization analysis suggests that LOT is localized on the peripheryof the Golgi cisternae, but the mutation does not affect the locali-zation of Golgi-resident proteins. Furthermore, the yeast comple-mentation result suggests that LOT could functionally act as acomponent of the guanine nucleotide exchange factor (GEF)complex of small Rab GTPase Ypt6. Taken together, these findingssuggest that LOT is a critical player for TGN biogenesis in theplant lineage.

TGN biogenesis | Golgi | vesicle trafficking | pollen tube growth

The Golgi–trans-Golgi network (TGN) system serves as anessential crossroad of the biosynthetic and endocytic recy-

cling pathways. The Golgi plays pivotal roles in sorting proteinsand is also a major biosynthetic factory for cell wall polysac-charides and protein glycosylation in plants (1–3). It consists ofstacks of cisternae that receive cargos at the cis-side from theendoplasmic reticulum (ER) and processes them before sortingat the TGNs into packaged vesicles destined for secretion orvacuoles (4, 5). Distinct from animal cells in which the TGNslocate on the trans-side of the Golgi, the plant TGN is a separatecompartment from the Golgi and functions as an early endosome(EE) as well, thus called TGN/EE (6–8).In plants, the TGN is verified to be an independent organelle

that matures from the trans-most Golgi cisterna, as demonstratedby superresolution live imaging, cell fractionation, and electronmicroscopy-based studies (9, 10). Based on tomography study,the biogenesis of the TGN can be divided into two continuousstages in a coat-independent manner: cisternae peeling from thetrans-side Golgi cisterna, at which stage the TGN is termed theGolgi-associated TGN (GA-TGN), and free TGN, which givesrise to proliferated smaller secretory vesicles (10, 11). De-tachment of GA-TGNs from the Golgi is proposed to give rise tothe free TGNs (10). Genetic studies and drug treatment havealso evidenced the dynamic and physical association of the TGNwith the Golgi (8, 12, 13). Several TGN-localized proteins havebeen implicated in regulating TGN detachment or function (14).However, it is still unclear how TGN biogenesis is regulated, andno Golgi-localized proteins have been discovered to take a rolein this process.Vesicular transport in yeasts, mammalian cells, and plants is

known to be mainly determined by a large family of Ras-relatedGTPases known as Ypt in yeast and Rab proteins in the othertwo kingdoms. These small GTPases are activated by the guanine

nucleotide exchange factor (GEF) that catalyzes the GDP-to-GTP switch at the donor or acceptor membrane (15). The GTP-bound Rabs are required for vesicle generation from the donormembrane and delivery to or fusion with the acceptor mem-brane. A considerable number of Rabs have been identified inflowering plants, with 57 members in Arabidopsis distributed indifferent compartments (16). GEFs for all Ypts have beenidentified in yeast, while compared with the largely extendedRab families, only a few GEFs have been identified in plants.In flowering plants, pollen tubes are deployed to deliver the

sperm cells to the embryo sac for double fertilization (17, 18).Membrane trafficking is critical for driving the tip growth of thepollen tube (19, 20). Distinct from other cell types, pollen tubegrowth requires faster and sustained deposition of the cell wall,proteins, and membrane materials for a shortened fertilizationinterval in angiosperms (21). The cell wall provides tensile andplastic mechanics as well as turgor pressure and membraneproteins that play roles in cell signaling (22). Most of these sig-naling and mechanical materials are packaged and sorted in theform of secretory vesicles from the Golgi apparatus to the TGNand then the plasma membrane (PM) and apoplast. Mutationsaffecting this trafficking pathway may impair pollen tube growth(23–25). Thus, understanding the molecular mechanism of ves-icle secretion en route from the Golgi is important to plant re-production and basic cell biology.In this study, we identified a Golgi-localized RabGEF com-

ponent loss of TGN (LOT) that regulates the TGN biogenesisfrom the Golgi and demonstrated its effect on the secretory andendocytic pathways in the pollen tube.

ResultsLOT Regulates Pollen Tube Growth. In a genetic screen of Dstransposon insertion mutants of Arabidopsis, we identified a male

Significance

The plant trans-Golgi network (TGN) is spatially separated fromthe Golgi and functions as an independent cargo sorting sta-tion. The molecular mechanism and significance of its genera-tion and separation from the Golgi are largely unclear. Thisstudy identified a Golgi-localized component that is essentialfor TGN biogenesis and sheds light on the regulation of post-Golgi trafficking.

Author contributions: Y.X., H.-J.L., and W.-C.Y. designed research; P.-F.J. and Y.X. per-formed research; P.-F.J., Y.X., and H.-J.L. analyzed data; H.-J.L. and W.-C.Y. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. Z.Y. is a guest editor invited by theEditorial Board.

Published under the PNAS license.1P.-F.J. and Y.X. contributed equally to this work.2To whom correspondence may be addressed. Email: [email protected] or [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1809206115/-/DCSupplemental.

Published online November 9, 2018.

www.pnas.org/cgi/doi/10.1073/pnas.1809206115 PNAS | November 27, 2018 | vol. 115 | no. 48 | 12307–12312

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 25

, 202

0

gametophyte defective mutant, kd329, with a distorted Ds seg-regation ratio. The Ds element contains an NPTII expressingcassette, which confers theDs-mutated plants’ kanamycin resistance(KanR). Therefore, the gametophytic transmission can be trackedby the distorted segregation ratio of KanR to Kan sensitivity(KanR/KanS) of the seedlings on Murashige and Skoog mediasupplemented with kanamycin (26). In the self-pollination exper-iment of the heterozygous kd329+/−, the progeny displays a KanR/KanS

ratio of 0.9:1 (1,244:1,392, n = 2,636), suggesting a gametophyticdefect. In reciprocal test crosses with the wild-type (WT) plants,KanR/KanS is 0.9:1 (789:896, n = 1,685) when kd329+/− was used asthe pollen recipient plant. However, the KanR/KanS ratio is 0.005:1(10:2,031, n = 2041) when kd329+/− was used as the pollen donorplant (SI Appendix, Table S1). These results indicate that the kd329mutation severely impairs the male gametophytic function.Through thermal asymmetrical interlaced PCR and sequenc-

ing, the flanking sequence of Ds was obtained. Sequence analysisshows that the Ds element is inserted in the first exon ofAt1G50120, 4 bp downstream of the start codon (Fig. 1A and SIAppendix, Fig. S1A). We designated At1G50120 as LOT (seebelow) and mutant kd329 as lot. Additionally, a homozygoustransfer DNA (T-DNA) insertion allele, Salk_130228 fromArabidopsis Biological Resource Center (ABRC) with T-DNAinserted in the sixth intron, still transcribes a substantial amountof the corresponding mRNA (SI Appendix, Fig. S1B). This sug-gests that Salk_130228 is a knockdown mutant, which may ex-plain the lack of visible phenotype of the plants. Therefore, onlythe Ds mutation lot was further studied. In a screening for ho-mozygotes, a homozygous plant for Ds that is dwarf and almosttotally male sterile, bearing only scarce seeds (1 ± 0.5 seed perpod), was obtained among 300 plants (Fig. 1B).To confirm that LOT is the causal gene for the phenotype, the

fusion construct of the coding sequence of LOT and GFP drivenby its native promoter (1,701 bp upstream of the start codon) was

made and transformed into lot+/−. In T2 transgenic lot+/− plantsexpressing GFP-LOT, the KanR/KanS segregation ratio of fiveindependent transgenic lines recovers from 0.9 to 1.7–2.8 (SIAppendix, Table S2). In the T3 plants, fertile homozygous lotplants carrying the transgene were isolated, which further con-firms that LOT is the causal gene of the male defect of lot.To investigate the biological basis of the decreased male ga-

metophyte transmission efficiency, the pollen morphology andviability of lot homozygotes were examined by DAPI and Alex-ander staining, respectively. As the results have shown, lot pollengrains are morphologically normal with two generative nucleiand one vegetative nucleus at maturity (n = 1,000), indicatingthat pollen development of lot is normal (SI Appendix, Fig. S2).Further examination of pollen germination and tube growthin vitro showed that the germination ratio of lot is lower than thatof theWT (Fig. 1C) and lot pollen tube length is shortened (Fig. 1D).The live imaging result suggests that even though the morphologyof lot pollen tubes is normal, the growth rate is 1.2 μm/s, muchslower than that of the WT (Fig. 1 E and F). However, this defectin pollen tube growth in vitro is not proportionally consistent withthe severely impaired male gametophytic transmission (0.5%).Thus, we speculate that this may result from the low competenceof lot pollen for fertilization in vivo.To test this possibility, the pollen tube growth in the pistil was

examined. Pollen grains were pollinated manually onto the WTpistil, and then at 12 h after pollination the pistils were stainedwith aniline blue, which labels the callose on the pollen tube. Asshown, the WT pollen tubes grew to the base of the pistils, andevery ovule in the pistil was targeted (Fig. 2A), while most of thepollen tubes from lot homozygotes were blocked in the style (Fig. 2B and D). Only a few mutant pollen tubes extended to the trans-mitting tissue, but they failed to target the ovules (Fig. 2C). Whenthe stigma was dissected, most of the blocked pollen tubes wereobserved to exhibit swollen or branching tips in the style (Fig. 2 Eand F). In semi-in vitro conditions, lot pollen tubes never grewout of the cut style even after 8-h culturing on the medium (Fig. 2G and H).Together, these results suggest an indispensable role of LOT

in pollen tube growth in vivo, and different tip morphologies ofthe lot pollen tube in vivo and in vitro imply a pollen–stigmainteraction defect.

LOT Is Widely Expressed and Encodes a Golgi-Localized Protein. Todetermine the subcellular localization and expression pattern ofLOT, a genomic fragment including the native promoter andcoding region fused C-terminally with β-glucuronidase (GUS) orGFP was introduced to the lot heterozygotes. The KanR/KanSsegregation ratio was highly elevated in the T2 transgenic prog-eny (SI Appendix, Table S3), suggesting that the C-terminaltagged LOT also functions properly in vivo. GUS staining showedthat GUS activity is detected in the root tip, lateral root primor-dium, leaf, stamen filament, pollen, and pollen tube (Fig. 3 A–E). Inthe genetically rescued GFP-LOT and LOT-GFP plants, the fusionproteins display punctate structures in the pollen grain, pollen tube,and root cells (Fig. 3 F–H). To further dissect the subcellularlocalization, the plant expressing Golgi marker 35S:ST-mCherry(sialyltransferase-mCherry) was crossed with GFP-LOT trans-genic plants. ST-mCherry and GFP-LOT totally colocalize in thesame punctate structure in root cells, which indicates that LOT islocalized on the Golgi apparatus (Fig. 3 I–K). Immunoblot assaywith GFP-LOT seedlings showed that GFP-LOT is ∼100 kDa, aspredicted (Fig. 3L). Furthermore, superresolution microscopywas deployed to obtain clearer observation of LOT on the Golgi(Fig. 3M). The series of optical sections (Fig. 3 N and P) andprojection images (Fig. 3Q and Movie S1) showed that GFP-LOT is localized on the rim of the Golgi apparatus, similar tothe pattern of ST-mCherry (Fig. 3O).

LOT Is Required for TGN Formation and Golgi Structure Maintenance.To further investigate the function of LOT on the Golgi, weperformed high-pressure freezing and freeze-substitution-based

Fig. 1. Characterization of lot and its phenotype. (A) Schematic diagram of Dsand T-DNA insertion. Black box, exon; black line, intron; gray box, untranslatedregion; arrow, primers used. (B) The lot plant is sterile and dwarf. (C and D) Invitro pollen germination and tube growth of lot at 4, 6, and 8 h after plating.Data are shown as means ± SD of three repeats (n > 1,000). Student’s t test,**P < 0.01; ***P < 0.001. (E) Time series of pollen tube growth. (Scale bar,20 μm). (F) In vitro pollen tube growth rate of lot and WT. Data are shown asmeans ± SD of three repeats (n = 10 pollen tubes). Student’s t test, *P < 0.05.

12308 | www.pnas.org/cgi/doi/10.1073/pnas.1809206115 Jia et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 25

, 202

0

transmission electron microscopy (TEM) analysis with pollengrains to examine the ultrastructure of the WT and lot. TEManalysis revealed that the WT pollen grains display a regularGolgi morphology, with vesicles of different sizes around theGolgi (Fig. 4 A and D and SI Appendix, Fig. S3). By contrast,whole-cell survey suggested that most of the Golgi in lot pollenexhibits overstacking and elongated Golgi cisternae with narrowerspace in between (Fig. 4 B, C, and E and SI Appendix, Fig. S4). Inaddition, a small proportion (<5%) of the cisternae, withoutvesicles around the Golgi, is dilated (Fig. 4F), which was not ob-served in the WT. Nevertheless, it could not be ruled out that thedilated structure is caused by the sectioning angle. A 3D recon-struction would be able to clarify this caveat. Statistical resultssuggest that the length of the Golgi cisternae is longer, and thenumbers of cisternae in the Golgi stack are increased in lot (Fig. 4G–I). Importantly, vesicles surrounding the Golgi are drasticallyreduced (Fig. 4J), implying defective vesicle budding from theGolgi. These results suggest the possibility of defective formation ofTGN and vesicles.To confirm this speculation, the TGN-specific maker YFP-

RabA4b (27) and endosome marker mCherry-RabA5D (28)driven by the pollen-specific LAT52 promoter and UBQ10 pro-moter, respectively, were introduced to track the number anddistribution of TGN/EE in the pollen grain and pollen tube.RabA4b labels budding vesicles that peel off from the GA-TGN,and the knockout mutant of RabA4b displays morphologicaldefects of the GA-TGN and fewer budding vesicles from theseGA-TGNs (10). Our results show that these two markers aredispersive in the cytosol of lot instead of on the punctate TGN/EE as in the WT (Fig. 5 A–C). As Rab proteins dynamicallyrecycle between the membrane and cytosol by the switch of GTPto GDP binding, we further introduced transmembrane TGNmarker protein VTI12, which is a component of Q-SNARE (28,29). Similarly, VTI12 is also dispersed in the cytosol of lot pollenand the pollen tube (Fig. 5 D–G). These results corroborate theobservation of TGN deficiency in lot pollen by TEM.The lack of TGNs may also disrupt the endocytic process due

to the reduced pool of TGN/EE compartments. To test thishypothesis, we tracked the endocytosis of lot pollen tubes withFM4-64, a lipophilic probe that labels the outer leaflet of thePM. Under in vitro germination conditions, the WT pollen tubesinternalized the dye within 5 min (Fig. 5H), while endocytosis

was not obvious even after 30 min of staining in lot pollen tubes(Fig. 5I). This result indicates that the endocytic pathway is alsoimpaired in lot pollen tubes. To further analyze the endocyticpathway, we introduced the late endosome/multivesicular body(MVB) markers ARA6 and ARA7, which label the homotypi-cally fused endosomes (30–32). In WT pollen tubes, YFP-ARA6and YFP-ARA7 label the late endosomes (Fig. 5 J and L), whilein lot pollen tubes, they display a dispersed pattern with no vis-ible late endosomal compartments (Fig. 5 K and M). This resultis consistent with the lack of TGN/EE in lot pollen tubes. AsMVBs progress and integrate with the vacuole, one intriguingquestion is whether the vacuole formation in lot is also disrupted.To investigate this question, we introduced the vacuole markermCherry-VAMP711 (28). The results suggest that no obviousabnormal vacuole pattern was observed in lot pollen tubes (SIAppendix, Fig. S5). One possible explanation is that the ER, notthe Golgi and TGN, is the major membrane source of vacuolebiogenesis (33). Together, these results suggest that the TGNbiogenesis is impaired in lot.Membrane-associated Rab GTPases regulate membrane traf-

ficking events in Rab-GTP bound active state by specificallyrecruiting cytosolic effectors. TGN-localized RabA4b recruitsPI4Kβ, which catalyzes the formation of PI(4)P, the precursor ofPI(4,5)P2 (34). PI(4,5)P2 is required for PM targeting of diverseproteins, which in turn affects exocytosis and endocytosis and,subsequently, pollen tube growth (35). Thus, we raised the pos-sibility that failed generation or targeting of phosphoinositidespecies might be the cause of defective endocytosis. To test thishypothesis, we examined their distribution in lot pollen tubes usingphosphoinositide-specific biosensors mCITRINE-1×TUBBY and

Fig. 3. The expression pattern and subcellular localization of LOT. (A–E) GUSstaining of pLOT: gLOT-GUS-3UTR plants. (A) Seedling. (B) Leaf. (C) Inflores-cence. (Scale bar in A–C, 1 mm). (D) Anther. (Scale bar, 100 μm.) (E) Pollen tube.(Scale bar, 10 μm.) (F–H) Image of GFP from pLOT: gLOT-GFP-3UTR plants. (F)Pollen. (G) Pollen tube. (H) Root. (Scale bar, 5 μm.) (I–K) Colocalization of GFP-LOT and ST-mCherry in roots. (Scale bar, 1 μm.) (L) Immunoblot of GFP-LOTfusion protein. 1, GFP-LOT plants; 2, WT plants. kD, kilodalton. (M–Q) GFP-LOTand ST-mCherry in the root by superresolution microscopy. (M) GFP-LOT imagewith lower magnification. (Scale bar, 2 μm.) (N) Enlarged view ofM. (Scale bar,0.7 μm.) (O) ST-mCherry. (Scale bar, 0.7 μm.) (P) Stacked image series of GFP-LOT. (Scale bar, 0.5 μm.) (Q) Three-dimensional projection of stacked images ofGFP-LOT.

Fig. 2. The lot pollen tubes grow slower in vivo. (A) WT pollen tubes in theWT pistil. Asterisks, targeted ovules. (B) The lot pollen tubes in the WT pistil.Arrow, where most pollen tubes reached. (C) The lot pollen tubes fail totarget the ovule. Arrow, failed pollen tube. (D) Pollen tube length in vivo.Data are shown as means ± SD of three repeats (n = 20 pistils). Student’s t test,***P < 0.001. (E) Swelled lot pollen tubes in the style. (F) Statistics of E. Dataare shown as means ± SD of three repeats (n = 120 pollen tubes). Student’st test, ***P < 0.001. (G) WT pollen tubes grow out of the WT cut style. (H) Thelot pollen tubes fail to penetrate out of theWT style. (Scale bars inA, B, G, andH, 400 μm; C and E, 100 μm.)

Jia et al. PNAS | November 27, 2018 | vol. 115 | no. 48 | 12309

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 25

, 202

0

mCITRINE-2×PH (FAPP1), which label PI(4,5)P2 and PI(4)P,respectively. In the WT pollen tube, both phosphoinositides aretargeted to the PM of pollen grains and tubes (Fig. 5 N and P),while in lot, they are distributed evenly in the cytosol (Fig. 5 Oand Q). This result supports the speculation that the failed en-docytosis may be caused by, or at least correlated with, the im-paired PI(4,5)P2 targeting on the PM due to the lack of TGNs.To investigate whether the localization of Golgi-resident

proteins is affected in lot pollen, three Golgi-resident proteinswere examined: COG3, which plays key roles in Golgi structure(23) in pollen; Got1 (28); and EMP12 (36). In the pollen pop-ulations from WT and lot+/− plants, consistent localization patternsfor all three markers were observed, which is similar to what isfound in the literature (23, 36) (SI Appendix, Fig. S6). Theseresults suggest that lotmutation most likely affects the biogenesisof the TGN from the Golgi, although the localization of theGolgi resident proteins is normal.

Secretion Is Defective in lot Pollen Tubes. The Golgi is the mainsynthetic station of cell wall polysaccharides and glycolipids forthe PM as well as glycoprocessing of proteins. Deposition of newcell wall materials and the PM targeting of receptor-like kinasesderived from the Golgi are indispensable for pollen tube growth(22, 37). To elucidate the direct cause of defective growth of lotpollen tubes, we analyzed the secretion of a cell surface receptorkinase and pectin deposition. First, we examined the expressionand localization of PRK2, a member of the receptor-like kinasesrequired for pollen tube growth (38). PRK2-GFP is localizedcontinuously on the PM in the WT pollen tube (Fig. 6 A, C, andE); however, in the lot pollen tube, it appears to be localized

on punctuate structures at and/or beneath the PM (Fig. 6 B,D, and F).Next, we observed the cell wall deposition using the carbo-

hydrate antibody JIM7 and ruthenium red to visualize pectinwith a high level of methylesterification (39). In the WT, theJIM7 antibody and ruthenium red predominantly label the apexof pollen tubes with a dome shape (Fig. 6 G and I), while in lot,no positive staining was observed in the pollen tube tip (Fig. 6 Hand J). Receptor-like kinases, like PRK2, are synthesized in theER and processed in the Golgi and then transported to the PMthrough TGNs. Pectin is synthesized and processed in the Golgiand then secreted to the extracellular matrix by the bulk secretionpathway, instead of the Golgi-TGN-PM/cell wall pathway (40, 41).Other reports also suggest that a defect in the TGN also impairspectin secretion (42). Collectively, these results imply that lotdisrupts the Golgi and post-Golgi secretion pathways.

LOT Functions as a Component of GTP Exchange Factor of Rab GTPase.LOT contains 532 amino acids and encodes an unidentified pro-tein with two overlapping domains that could be identified fromthe known domain repertoires: a predicted Rgp1 domain (Inter-Pro: IPR014848) that is found only among its orthologs in otherorganisms and an Ig E-set (InterPro: IPR014756) domain (Fig.7A). The Rgp1 protein was suggested to function as a functionalcomponent of GEF of Ypt6 and Rab6 in yeast and mammaliancells, respectively (43, 44). Single-copy orthologs of Rgp1 arefound in all eukaryotes (Fig. 7B), although Rgp1 in Saccharomycescerevisiae and LOT share only 14.6% identity and 27% similarityin the amino acid sequence. In contrast to the biological functionof LOT in TGN biogenesis, Rgp1 in yeast and mammalian cellsfunctions in endosome-to-Golgi retrograde trafficking (43–45). Toelucidate whether LOT is a functional component of RabGEF, weperformed a yeast complementation assay with the yeast rgp1 mu-tant which shows a temperature-sensitive growth defect at 37 °C(44). The yeast rgp1mutant cells transformed with LOT indeed cansurvive at 37 °C (Fig. 7C). This indicates that LOT can functionallyreplace the yeast Rgp1 as a component of RabGEF.

DiscussionIn summary, we demonstrated that LOT regulates Golgi-derivedTGN formation and is essential for the entire post-Golgi

Fig. 5. The TGN formation in lot pollen tubes is disrupted. (A and B) Con-focal images of YFP-RabA4b in WT (A) and lot (B) pollen tubes. (C) Pollenfrom mCherry-RabA5D; lot+/−. (D–G) mCherry-VTI12 in WT (D) and lot (E)pollen and WT (F) and lot (G) pollen tubes. (Scale bar, 5 μm.) (H–M) Confocalimages of FM4-64 staining (H and I) and the YFP signal (J–M) in pollen tubes.YFP-ARA6 in WT (J) and lot (K). YFP-ARA7 in WT (L) and lot (M). mCITRINE-1×TUBBY in WT (N) and lot (O). mCITRINE-2×PH(FAPP1) in WT (P) and lot (Q).N–Q, insets, pollen grain. (Scale bar, 5 μm.)

Fig. 4. The lot causes reduced TGNs, vesicles, and deformed Golgi. (A–F) TEMimages of the Golgi inWT (A and D) and lot (B, C, E, and F) pollen. (D) Zoomed-in image of A. (E) Zoomed-in image of B. Arrows, vesicular structures aroundthe Golgi (G). (Scale bar, 200 nm.) (G) Diagram of the Golgi structure andsurrounding vesicles. (H) Ratio of Golgi length/width in WT (n = 21) and lot(n = 26) pollen. Data are shown as means ± SD. Student’s t test, ***P < 0.01. (I)Number of Golgi of different cisternae numbers in theWT (n = 19) and lot (n =25). (J) Quantification of vesicles around the Golgi. Areas of (800 nm × 400 nm)square with a Golgi centered were scored for the vesicle numbers in WT andlot. (n = 21 Golgi for WT; n = 26 Golgi for lot). Data are shown as means ± SD.Student’s t test, ***P < 0.01.

12310 | www.pnas.org/cgi/doi/10.1073/pnas.1809206115 Jia et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 25

, 202

0

trafficking during pollen tube growth (Fig. 7D). GFP-LOT islocalized on the rim of Golgi cisternae and might regulate vesicleformation therein, which is indispensable for TGN formation.The overstacking of Golgi cisternae and the proper Golgi lo-calization of EMP12 exported from the ER network in lot sug-gest that the cargo input from the ER to cis-cisternae may beunimpaired. Furthermore, the dispersed distribution of the TGNmarkers implies that the earlier stage of TGN formation beforeTGN detachment is likely defective in lot. An early report sug-gested the role of PI4Kβ1/2 in TGN detachment since loss ofPI4Kβ1 and PI4Kβ2, which catalyze PI(4)P generation, causesTGN aggregation at the trans-side of the Golgi and fewer TGNbudding profiles (34). Such Golgi-associated TGN aggregationwas not observed in lot, which suggests a mechanism distinct fromPI4Kβ1/2. Thus, LOT provides a novel clue to study how the cister-nae rim contributes to the TGN formation, which may also bevaluable to the understanding of intra-Golgi trafficking.Secretion and endocytosis underline the cellular basis for

pollen tube growth. In lot plants, pollen tube growth is blocked inthe style due to failed TGN formation from the Golgi that causesthe depletion of secretory vesicles which are the carriers of the post-Golgi traffic, as well as the endocytic pathway. Thus, the entirepost-Golgi trafficking of cargo repertoires, such as phospholipids,

cell wall components, and membrane-bound proteins, is abolished.Of note, the lot pollen tubes swell in the style but not in vitro,implying that LOT regulates the pollen tube–style interaction. Thefailed secretion of PRK2 suggests that more receptor-like kinasesare possibly defective in PM targeting in lot, which could con-tribute to the failed male–female interaction. In Arabidopsis, threeclades of Rab subfamilies have been confirmed to be present onthe Golgi, and each clade contains members highly expressed inpollen. Among the Golgi-localized Rabs, only the RabD2 sub-family has been determined to function in pollen development,germination, and pollen tube growth (46). However, none of theirGEF activators have been identified. The Arabidopsis RabH1b wasable to complement the temperature-sensitive growth defect ofyeast ypt6 (47), and activation of Ypt6/Rab6 requires Rgp1 (44),which can be functionally replaced by LOT in the yeast com-plementation assay, as shown above. In Arabidopsis, the orthologof Ypt6 is the RabH1 subfamily, among which four membershave been indicated to be expressed in pollen and none havebeen functionally studied (48). RABH1b and RABH1c localize onthe Golgi and physically interact with Golgi-tethering proteins (49–51). Rgp1 and Ric1 form a functional GEF complex in mammaliancells and yeast (44, 45), while the two orthologs of Ric1 in Arabi-dopsis are large proteins which are difficult to express and purify inour hands. Further endeavor is needed to clarify the potential re-lationship between LOT and Rab GTPases on the Golgi, as well astheir downstream effectors.Golgi-localized LOT regulates the Golgi-derived TGN bio-

genesis, a role that has not been identified in its animal and yeasthomologs. The yeast Rgp1-activated Ypt6 has been recognized tomediate the binding of endosome-derived vesicles to the late Golgithrough the SNARE protein, that is, the endosome-to-late Golgiretrograde trafficking (43). In human cells, Rgp1 also functions asa component of GEF in the TGN-localized small GTPase Rab6A,which regulates tethering of endosome-derived vesicles to thetrans-Golgi cisternae for retrograde transport of the internalizedproteins by endocytosis (43, 45). In animal cells, TGN is associatedwith the Golgi at the trans-face, and the cargo sorting at the TGNto endosome is accompanied by endosome-to-TGN retrograde traf-ficking of a catalog of factors, such as cargo receptors and SNAREs.However, in plants, the TGN is an independent compartmentseparated from the Golgi and functions as a genuine early endosome,which might explain the lack of an endosome-to-TGN retrogradepathway. In this cellular context, LOT likely maintains the GEFmolecular function but is adapted to function in the TGN bio-genesis from the Golgi. However, whether the plant lineage indeedlacks TGN/EE-to-Golgi retrograde trafficking and the role of LOTorthologs in yeast and animals in TGN biogenesis still requirefurther investigation.

Fig. 6. Biosynthetic secretion in lot pollen tubes is defective. (A–F) Confocalimages of PRK2-GFP in WT (A, C, and E) and lot (B, D, and F) pollen or pollentubes. Arrows, punctate structures. (Scale bar, 5 μm.) (G and H) Pollen tubesof WT (G) and lot (H) stained with the JIM7 antibody. Arrows, pollen tubetip. (Scale bar, 5 μm.) (I and J) Pollen tubes of WT (I) and lot (J) stained withruthenium red. Arrows, pollen tube tip. (Scale bar, 5 μm.)

Fig. 7. LOT shares conserved molecular function with yeast Rgp1. (A) The protein structure of LOT. (B) Phylogenetic tree of LOT orthologs. (C) LOT rescuesthe growth defect of the rpg1 yeast mutant. (D) A proposed model of LOT in TGN and vesicle formation. In WT pollen tubes, LOT is required for the activationof Rab GTPases on the Golgi periphery to promote TGN formation. In lot pollen tubes, Rab GTPases and possibly other proteins fail to be activated or recruitedto the Golgi, thus causing failed vesicle formation or budding and lack of TGN. The lack of TGN and vesicles causes failed secretion and endocytosis, whichleads to the pollen tube growth defect.

Jia et al. PNAS | November 27, 2018 | vol. 115 | no. 48 | 12311

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 25

, 202

0

In summary, this work identified LOT as a potential RabGEFon the Golgi and provided a foundation for dissecting the mo-lecular mechanisms that underpin TGN formation and thesorting regulation therein. Second, this work evidenced the cel-lular significance of Golgi in cargo trafficking and endocytosisand provided valuable material for the study of Golgi-dependentand -independent membrane trafficking pathways.

Materials and MethodsPlant Materials and Growth Conditions. Salk_130228 was obtained from ABRC;kd329 was selected from a Ds insertion mutant pool. See SI Appendix, Ma-terials and Methods for details.

Phenotype Analysis. For in vivo pollen tube growth, pollen grains weremanually pollinated on the WT pistil and fixed 12 h after pollination withethanol and acetic acid (3:1) for 2 h. See SI Appendix,Materials and Methodsfor details.

Plasmid Construction. See SI Appendix, Materials and Methods for details.

Western Blot, Immunofluorescence, and Microscopy. Total proteins were re-solved with SDS/PAGE and immunodetected with anti-GFP. See SI Appendix,Materials and Methods for details.

LOT Protein Sequence and Phylogenetic Analysis. For bioinformatics analysis,see SI Appendix, Materials and Methods for details.

Yeast Complementation Assay. For constructs and the following analysis, see SIAppendix, Materials and Methods for details.

ACKNOWLEDGMENTS. We thank Prof. Yan Zhang (Shandong AgricultureUniversity) for kindly providing the PRK2-GFP and YFP-ARA6 and -ARA7 seeds,Prof. Yi-Hua Zhou [Institute of Genetics and Developmental Biology (IGDB),Chinese Academy of Sciences] for kindly providing the JIM7 antibody, Prof.Pelham (Medical Research Council Laboratory of Molecular Biology, Cambridge)for kindly providing the rgp1 yeast mutant, Prof. Yiqun Bao (Nanjing AgricultureUniversity) for kindly providing the Golgi markers, Prof. Shilai Bao (IGDB) for thecritical reading of the manuscript, and the staff of the live-cell imaging facilitiesof IGDB for assistance with TEM, superresolution, and confocal microscopy. Thiswork is supported by the National Natural Science Foundation of China (Grant31330053 to W.-C.Y., and Grants 31622010 and 31270351 to H.-J.L.).

1. Faso C, Boulaflous A, Brandizzi F (2009) The plant Golgi apparatus: Last 10 years ofanswered and open questions. FEBS Lett 583:3752–3757.

2. Kim SJ, Brandizzi F (2016) The plant secretory pathway for the trafficking of cell wallpolysaccharides and glycoproteins. Glycobiology 26:940–949.

3. Nakano A, Luini A (2010) Passage through the Golgi. Curr Opin Cell Biol 22:471–478.4. Brandizzi F, Barlowe C (2013) Organization of the ER-Golgi interface for membrane

traffic control. Nat Rev Mol Cell Biol 14:382–392.5. Hwang I (2008) Sorting and anterograde trafficking at the Golgi apparatus. Plant

Physiol 148:673–683.6. Neumann U, Brandizzi F, Hawes C (2003) Protein transport in plant cells: In and out of

the Golgi. Ann Bot 92:167–180.7. Viotti C, et al. (2010) Endocytic and secretory traffic in Arabidopsis merge in the trans-

Golgi network/early endosome, an independent and highly dynamic organelle. PlantCell 22:1344–1357.

8. Dettmer J, Hong-Hermesdorf A, Stierhof Y-D, Schumacher K (2006) Vacuolar H+-ATPaseactivity is required for endocytic and secretory trafficking in Arabidopsis. Plant Cell 18:715–730.

9. Uemura T, Suda Y, Ueda T, Nakano A (2014) Dynamic behavior of the trans-Golginetwork in root tissues of Arabidopsis revealed by super-resolution live imaging.Plant Cell Physiol 55:694–703.

10. Kang BH, Nielsen E, Preuss ML, Mastronarde D, Staehelin LA (2011) Electron to-mography of RabA4b- and PI-4Kβ1-labeled trans Golgi network compartments inArabidopsis. Traffic 12:313–329.

11. Staehelin LA, Kang BH (2008) Nanoscale architecture of endoplasmic reticulum exportsites and of Golgi membranes as determined by electron tomography. Plant Physiol147:1454–1468.

12. Gendre D, et al. (2011) Conserved Arabidopsis ECHIDNA protein mediates trans-Golgi-network trafficking and cell elongation. Proc Natl Acad Sci USA 108:8048–8053.

13. Shirakawa M, et al. (2014) Continuous vascular ring (COV1) is a trans-Golgi network-localized membrane protein required for Golgi morphology and vacuolar proteinsorting. Plant Cell Physiol 55:764–772.

14. Rosquete MR, Davis DJ, Drakakaki G (2018) The plant trans-Golgi network: Not just amatter of distinction. Plant Physiol 176:187–198.

15. Segev N (2001) Ypt and Rab GTPases: Insight into functions through novel interac-tions. Curr Opin Cell Biol 13:500–511.

16. Vernoud V, Horton AC, Yang Z, Nielsen E (2003) Analysis of the small GTPase genesuperfamily of Arabidopsis. Plant Physiol 131:1191–1208.

17. Dresselhaus T, Franklin-Tong N (2013) Male-female crosstalk during pollen germina-tion, tube growth and guidance, and double fertilization. Mol Plant 6:1018–1036.

18. Li H, Yang WC (2016) RLKs orchestrate the signaling in plant male-female interaction.Sci China Life Sci 59:867–877.

19. Lee YJ, Yang Z (2008) Tip growth: Signaling in the apical dome. Curr Opin Plant Biol11:662–671.

20. Cheung AY, Wu HM (2008) Structural and signaling networks for the polar cellgrowth machinery in pollen tubes. Annu Rev Plant Biol 59:547–572.

21. Williams JH (2008) Novelties of the flowering plant pollen tube underlie di-versification of a key life history stage. Proc Natl Acad Sci USA 105:11259–11263.

22. Hepler PK, Rounds CM, Winship LJ (2013) Control of cell wall extensibility duringpollen tube growth. Mol Plant 6:998–1017.

23. Tan X, et al. (2016) Arabidopsis COG complex subunits COG3 and COG8 modulateGolgi morphology, vesicle trafficking homeostasis and are essential for pollen tubegrowth. PLoS Genet 12:e1006140.

24. Chen LY, et al. (2015) The Arabidopsis alkaline ceramidase TOD1 is a key turgorpressure regulator in plant cells. Nat Commun 6:6030.

25. Wang L, et al. (2013) Arabidopsis galacturonosyltransferase (GAUT) 13 and GAUT14have redundant functions in pollen tube growth. Mol Plant 6:1131–1148.

26. Sundaresan V, et al. (1995) Patterns of gene action in plant development revealed byenhancer trap and gene trap transposable elements. Genes Dev 9:1797–1810.

27. Thole JM, Vermeer JE, Zhang Y, Gadella TW, Jr, Nielsen E (2008) Root hair defective4encodes a phosphatidylinositol-4-phosphate phosphatase required for proper roothair development in Arabidopsis thaliana. Plant Cell 20:381–395.

28. Geldner N, et al. (2009) Rapid, combinatorial analysis of membrane compartments inintact plants with a multicolor marker set. Plant J 59:169–178.

29. Sanderfoot AA, Kovaleva V, Bassham DC, Raikhel NV (2001) Interactions betweensyntaxins identify at least five SNARE complexes within the Golgi/prevacuolar systemof the Arabidopsis cell. Mol Biol Cell 12:3733–3743.

30. Ueda T, Yamaguchi M, Uchimiya H, Nakano A (2001) Ara6, a plant-unique novel type RabGTPase, functions in the endocytic pathway ofArabidopsis thaliana. EMBO J 20:4730–4741.

31. Zhang Y, He J, Lee D, McCormick S (2010) Interdependence of endomembrane traf-ficking and actin dynamics during polarized growth of Arabidopsis pollen tubes. PlantPhysiol 152:2200–2210.

32. Lee GJ, Sohn EJ, Lee MH, Hwang I (2004) The Arabidopsis rab5 homologs rha1 andara7 localize to the prevacuolar compartment. Plant Cell Physiol 45:1211–1220.

33. Viotti C, et al. (2013) The endoplasmic reticulum is the main membrane source forbiogenesis of the lytic vacuole in Arabidopsis. Plant Cell 25:3434–3449.

34. Preuss ML, et al. (2006) A role for the RabA4b effector protein PI-4Kbeta1 in polarizedexpansion of root hair cells in Arabidopsis thaliana. J Cell Biol 172:991–998.

35. Zhao Y, et al. (2010) Phosphoinositides regulate clathrin-dependent endocytosis atthe tip of pollen tubes in Arabidopsis and tobacco. Plant Cell 22:4031–4044.

36. Gao C, et al. (2012) The Golgi-localized Arabidopsis endomembrane protein12 con-tains both endoplasmic reticulum export and Golgi retention signals at its C terminus.Plant Cell 24:2086–2104.

37. Li HJ, Meng JG, Yang WC (2018) Multilayered signaling pathways for pollen tubegrowth and guidance. Plant Reprod 31:31–41.

38. Kim HU, et al. (2002) New pollen-specific receptor kinases identified in tomato, maizeand Arabidopsis: The tomato kinases show overlapping but distinct localizationpatterns on pollen tubes. Plant Mol Biol 50:1–16.

39. Szumlanski AL, Nielsen E (2009) The Rab GTPase RabA4d regulates pollen tube tipgrowth in Arabidopsis thaliana. Plant Cell 21:526–544.

40. Micheli F (2001) Pectin methylesterases: Cell wall enzymes with important roles inplant physiology. Trends Plant Sci 6:414–419.

41. Wang H, et al. (2016) A distinct pathway for polar exocytosis in plant cell wall for-mation. Plant Physiol 172:1003–1018.

42. Gendre D, et al. (2013) Trans-Golgi network localized ECHIDNA/Ypt interacting pro-tein complex is required for the secretion of cell wall polysaccharides in Arabidopsis.Plant Cell 25:2633–2646.

43. Bonifacino JS, Rojas R (2006) Retrograde transport from endosomes to the trans-Golginetwork. Nat Rev Mol Cell Biol 7:568–579.

44. Siniossoglou S, Peak-Chew SY, Pelham HRB (2000) Ric1p and Rgp1p form a complexthat catalyses nucleotide exchange on Ypt6p. EMBO J 19:4885–4894.

45. Pusapati GV, Luchetti G, Pfeffer SR (2012a) Ric1-Rgp1 complex is a guanine nucleotideexchange factor for the late Golgi Rab6A GTPase and an effector of the medial GolgiRab33B GTPase. J Biol Chem 287:42129–42137.

46. Peng J, Ilarslan H, Wurtele ES, Bassham DC (2011) AtRabD2b and AtRabD2c haveoverlapping functions in pollen development and pollen tube growth. BMC Plant Biol11:25.

47. Bednarek SY, et al. (1994) A small GTP-binding protein from Arabidopsis thalianafunctionally complements the yeast YPT6 null mutant. Plant Physiol 104:591–596.

48. Wang Y, et al. (2008) Transcriptome analyses show changes in gene expression toaccompany pollen germination and tube growth in Arabidopsis. Plant Physiol 148:1201–1211.

49. Johansen JN, Chow CM, Moore I, Hawes C (2009) AtRAB-H1b and AtRAB-H1c GTPases,homologues of the yeast Ypt6, target reporter proteins to the Golgi when expressedin Nicotiana tabacum and Arabidopsis thaliana. J Exp Bot 60:3179–3193.

50. Latijnhouwers M, et al. (2007) Localization and domain characterization of Arabi-dopsis golgin candidates. J Exp Bot 58:4373–4386.

51. Osterrieder A, et al. (2009) Fluorescence lifetime imaging of interactions betweenGolgi tethering factors and small GTPases in plants. Traffic 10:1034–1046.

12312 | www.pnas.org/cgi/doi/10.1073/pnas.1809206115 Jia et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 25

, 202

0