Cytoskeletal and membrane dynamics during higher plant cytokinesis




Following mitosis, cytoplasm, organelles and genetic material are partitioned into daughter cells through the process of cytokinesis. In somatic cells of higher plants, two cytoskeletal arrays, the preprophase band and the phragmoplast, facilitate the positioning and de novo assembly of the plant-specific cytokinetic organelle, the cell plate, which develops across the division plane and fuses with the parental plasma membrane to yield distinct new cells. The coordination of cytoskeletal and membrane dynamics required to initiate, assemble and shape the cell plate as it grows toward the mother cell cortex is dependent upon a large array of proteins, including molecular motors, membrane tethering, fusion and restructuring factors and biosynthetic, structural and regulatory elements. This review focuses on the temporal and molecular requirements of cytokinesis in somatic cells of higher plants gleaned from recent studies using cell biology, genetics, pharmacology and biochemistry.

I. Introduction

Cytokinesis, the final step of cell division, results in the partitioning of cellular components to form distinct daughter cells. In animals, cytokinesis begins at the cell periphery where an actinomyosin contractile ring drives invagination of the plasma membrane (PM) as new membrane is added to the growing furrow to facilitate abscission (Neto et al., 2011). By contrast, higher plant cytokinesis is an inside-out process mediated by the cell plate (CP), whose orientation and assembly is orchestrated by the plant-specific cytoskeletal preprophase band (PPB) and phragmoplast (Samuels et al., 1995; Jürgens, 2005; Backues et al., 2007). In this review we highlight and focus on recent advances in our understanding of the cytoskeletal and membrane dynamics that mediate cytokinesis in somatic plant cells. For additional background on this subject, the reader is referred to several previous reviews concerning plant cytokinesis: Samuels et al. (1995); Otegui & Staehelin (2000); Verma (2001); Seguí-Simarro et al. (2004); Jürgens (2005); and Backues et al. (2007). For convenience, the proteins mentioned in this review are summarized in Tables 1 and 2.

Table 1. Cytoskeleton-associated proteins involved in plant cytokinesis
ProteinsAGI Accession NumbersOrthologsaLocalizationb,cFunctionb,cReference(s)
  1. a

    Homologous localization and/or function; Zm, Zea mays (corn); Nt, Nicotiana tabacum (tobacco); Os, Oryza sativa (rice); Dc, Daucus carota (carrot); Tv, Tradescantia virginiana (spiderwort); Gs, Gossypium hirsutum (cotton); Hk, Haemanthus katherinae (blood lily).

  2. b

    Known or predicted.

  3. c

    PPB, preprophase band; CDZ, cortical division zone; CP, cell plate; CW, cell wall; MTs, microtubules; MF, actin microfilament.

TON1a, TON1bAt3g55000, At3g55005 PPB MTsOrganize/stabilize PPB MTsAzimzadeh et al. (2008); Malcos & Cyr (2011)
MOR1/GEM1At2g35630 PPB MTsOrganize/stabilize PPB MTsWhittington et al. (2001); Kawamura et al. (2006)
AtKINUa/ARK3/PAKAt1g12430 PPB MTsOrganize/stabilize PPB MTsSakai et al. (2008)
TON2/FASS/EMB40At5g18580(Zm) DCD1, ADD1PPB MTsRegulate TON1a/b, MOR1, TAN1Camilleri et al. (2002); Wright et al. (2009)
TAN1At3g05330(Zm) TAN1PPB; CDZPPB memoryWalker et al. (2007)
RanGAP1At3g63130 PPB; CDZPPB memoryXu et al. (2008)
POK1, POK2At3g17360, At3g19050 PPB; CDZRegulate TAN1, RanGAP1Müller et al. (2006)
AIR9At2g34680 PPB; CP leading edge; CWCP recognition/maturationBuschmann et al. (2006)
KCA1/KAC1At5g10470 Non-CDZ cortical MTs; phragmoplast midzone(–)-end-directed kinesinVanstraelen et al. (2004, 2006); Malcos & Cyr (2011)
EB1a/EB1H2, EB1b, EB1c/EB1H1At3g47690, At5g62500, At5g67270 Phragmoplast MTsMT (+)-end tracking proteinChan et al. (2003); Van Damme et al. (2004); Bisgrove et al. (2008)
NEDD1At5g05970 Spindle and phragmoplast MTsγ-TuRC: nucleate/organize MTsZeng et al. (2009)
γ-tubulin/TUBG1At3g61650 Spindle and phragmoplast MTsγ-TuRC: nucleate/organize MTsVan Damme et al. (2004)
GCP2/TUBG2At5g05620 Spindle and phragmoplast MTsγ-TuRC: nucleate/organize MTsNakamura et al. (2010)
AUG1/EMB2819, AUG3At2g41350, At5g48520 Spindle and phragmoplast MTsAugmin complex: recruit γ-tubulin and γ-TuRCHo et al. (2011)
AtFH5, AtFH8, AtFH14At5g54650, At1g70140, At1g31810 PPB, spindle and phragmoplast MFs and MTsAlign phragmoplast MFs and MTsLi et al. (2010b); Wang et al. (2012)
PAKRP2At4g14330 PhragmoplastVesicle transport on MTsLee et al. (2001)
AtNACK1/HIK, AtNACK2/TES/STDAt1g18370, At3g43210(Nt) NACK1, NACK2; (Os) DBS1/OsNACKPhragmoplast MTs; phragmoplast midzonePhragmoplast organization/dynamics; kinesin; activate MAPK cascade (NACK-PQR)Nishihama et al. (2002); Strompen et al. (2002); Tanaka et al. (2004); Takahashi et al. (2010)
AtPAKRP1/Kin-12a, AtPAKRP1L/Kin-12bAt4g14150, At3g23670 Phragmoplast midzone(+)-end-directed kinesinLee & Liu (2000); Pan et al. (2004)
AtKRP125c/AtKin5C/LPH/RSW7At2g28620(Nt) TKRP125,(Dc) KRP120Phragmoplast MTs(+)-end-directed kinesinAsada et al. (1997); Barroso et al. (2000); Bannigan et al. (2007)
ATK5/Kin-5At4g05190 Spindle MTs; phragmoplast midzone(-)-end-directed kinesin; MT (+)-end tracking proteinAmbrose et al. (2005)
KCA2/KAC2At5g65460 Phragmoplast midzone(-)-end-directed kinesinVanstraelen et al. (2004, 2006); Malcos & Cyr (2011)
AtKCBP/ZWIAt5g65930NtKCBP, TvKCBP, GhKCBP, HvKCBPPPB MTs; phragmoplast MTs(-)-end-directed kinesin; Ca2+-calmodulin-regulated MT bundlingBowser & Reddy (1997); Smirnova et al. (1998); Vos et al. (2000); Preuss et al. (2003)
AtKinGAt1g63640NtKCH, OsKCH, GhKCHPPB MTs; phragmoplast MTs and MFs(-)-end-directed kinesin; KCH domain-mediated MF bindingXu et al. (2009); Frey et al. (2010); Buschmann et al. (2011); Klotz & Nick (2012)
ANP1, ANP2, ANP3At1g09000, At1g54960, At3g06030(Nt) NPK1Phragmoplast MTs; phragmoplast midzoneMAPKKK; activates ANQ1/MKK6Nishihama et al. (2001); Krysan et al. (2002)
ANQ1/MKK6At5g56580(Nt) NQK1/NtMEK1Phragmoplast MTs; phragmoplast midzoneMAPKK; activates MPK4 and MPK6Soyano et al. (2003); Beck et al. (2010, 2011)
MPK4, MPK6At4g01370, At2g43790(Nt) NRK1/NTF6PPB MTs (MPK6); phragmoplast MTs; phragmoplast midzone (MPK4)MAPK; activates MAP65-1, MAP65-2 and MAP65-3/PLECalderini et al. (1998); Beck et al. (2010, 2011); Kosetsu et al. (2010); Müller et al. (2010); Takahashi et al. (2010);
MAP65-1, MAP65-2, MAP65-3/PLEAt5g55230, At4g26760, At5g51600(Nt) MAP65-1a/bPPB MTs; phragmoplast MTs (MAP65-1, -2); phragmoplast midzone (MAP65-3)Phosphorylation-mediated regulation of NACK-PQR-targeted substrates; bundle MTsMüller et al. (2004); Sasabe et al. (2006, 2011); Beck et al. (2010)
TIO/AtFUAt1g50240 Phragmoplast midzoneKinase; regulates PAKRP1 and PAKRP1LOh et al. (2005, 2012)
RUK/EMB3013At5g18700 PPB; phragmoplastPhragmoplast organization/dynamicsKrupnova et al. (2009)
AUR1, AUR2At4g32830, At2g25880 CPKinaseVan Damme et al. (2011b)
Table 2. Proteins involved in membrane dynamics during plant cytokinesis
  1. a

    Known or predicted.

  2. b

    TGN/EE, trans-Golgi network/early endosome; CP, cell plate; MVB/PVC, multivesicular body/prevacuolar compartment; CDS, cortical division site; CW, cell wall; CME, clathrin-mediated endocytosis; PM, plasma membrane; ER, endoplasmic reticulum; MTs, microtubules; CCV, clathrin-coated vesicle; CDZ, cortical division zone.

KN/SYP111At1g08560Golgi; TGN/EE; CP vesicles; CP leading edget-SNARE: regulate vesicle and target membrane fusionWaizenegger et al. (2000); Chow et al. (2008)
ECHAt1g09330TGN/EEMaintain TGN/EE structure/functionGendre et al. (2011)
ELCAt3g12400TGN/EE; endosomeEndosomal sorting; stabilize MTsSpitzer et al. (2006)
RABA1b/BEX5At1g16920TGN/EEGTPase: regulate TGN/EE-to-CP vesicle transport/fusionFeraru et al. (2012)
RABA2a/AtRAB11c, RABA2b, RABA3At1g09630, At1g07410, At1g01200TGN/EE; CP leading edgeGTPase: regulate TGN/EE-to-CP vesicle transport/fusionChow et al. (2008)
RABA4bAt4g39990TGN/EEGTPase: regulate TGN/EE-to-PM vesicle transport/fusionPreuss et al. (2004, 2006)
RABF1/ARA6,RABF2b/ARA7At3g54840, At4g19640TGN/EE; MVB/PVC; CP; CP vesiclesGTPase: regulate TGN/EE-to-CP/PM vesicle transport/fusionUeda et al. (2004); Dhonukshe et al. (2006)
GN/VAN7/EMB30At1g13980TGN/EE; CPRegulate TGN/EE-to-CP vesicle transport/fusionDhonukshe et al. (2006)
SCD1At1g49040CCVs; CPCCV-associated Rab GEFFalbel et al. (2003); Korasick et al. (2010)
AtTRS33, AtTRS120, AtTRS130/CLUB (and others)At3g05000, At5g11040, At5g54440TGN/EE; CPTRAPPII (GEF for RabAs; i.e. regulate TGN/EE-to-CP trafficking) subunitsJaber et al. (2010); Thellmann et al. (2010); Qi et al. (2011); Qi & Zheng (2011)
EXO70A1 (and others)At5g03540CP; CDS; CWExocyst complex (vesicle-PM or vesicle-CP tether) subunitSynek et al. (2006); Fendrych et al. (2010)
SNAP33/SNP33At5g61210CPKN/VAMP721/722-interacting t-SNAREHeese et al. (2001)
VAMP721/VAMP7b, VAMP722/SAR1At1g04750, At2g33120CPRedundant, KN/SNP33-interacting v-SNAREsZhang et al. (2011)
NSPN11At2g35190CPKN-interacting v-SNAREZheng et al. (2002)
NSF, αSNAP1At4g04910, At3g56450CPKN/SNP33/VAMP721/722 trans-SNARE disassemblyRancour et al. (2002)
KEUAt1g12360CPPromote KN/SNP33/VAMP721/722 trans-SNARE complexAssaad et al. (2001);Heese et al. (2001)
CHC1, CHC2At3g08530, At3g11130CPw/CLC, forms clathrin coatVan Damme et al. (2011a)
CLC1, CLC2, CLC3At2g40060, At2g20760, At3g51890CPw/CHC, forms clathrin coatKonopka et al. (2008); Van Damme et al. (2011a,b); Ito et al. (2012); Fig. 2
AP180/Epsin (and other A/ENTH proteins)At1g05020CPPromote assembly of clathrin membrane patchesBarth & Holstein (2004);Ito et al. (2012);Song et al. (2006)
TPLATEAt3g01780CP; CDS flanks of PMAP-/COP1-like complex subunit; interacts with CHC1 and CLC2Van Damme et al. (2006, 2011a,b)
DRP1A/ADL1A/RSW9, DRP1C/ADL1C, DRP1E/ADL1E/EDR3, DRP2A/DLP6, DRP2B/ADL3At5g42080, At1g14830, At3g60190, At1g10290, At1g59610Sites of CME at PM; CP leading edgeGTPases: clathrin vesicle scission and CP membrane tubulationFujimoto et al. (2008, 2010); Konopka & Bednarek et al. (2008); Ito et al. (2012); Fig. 2
AtCDC48At3g09840CPAAA-ATPase chaperoneRancour et al. (2002)
SYP31/SED5At5g05760ER; Golgi, CPt-ER formation w/in CDZRancour et al. (2002)
RHD3/GOM8At3g13870ERGTPase: ER formation/dynamics Chen et al. (2011b)
PATL1At1g72150CPTGN/EE-derived vesicle formationPeterman et al. (2004)
PAS2/PEPAt5g10480CPElongate membrane phospholipidsBach et al. (2011)
FK/HYD2/ELL1, CPH/SMT1, HYD1/MAD4At3g52940, At5g13710, At1g20050CPBiosynthesize sterolsDiener et al. (2000); Jang et al. (2000); Schrick et al. (2000); Souter et al. (2002)
GSL8/MAS/CHORAt2g36850CPSynthesize calloseChen et al. (2011b); Guseman et al. (2010);
GCS1/KNF, RSW3/PSL5, CYT1/GMP1/SOZ1/VTC1, KOR/RSW2/IRX2/DEC/GH9a1/TSD1, PRC1/CESA6/IXR2, CESA1/RSW1, CSLD/SOS6At1g67490, At5g63840, At2g39770, At5g49720, At5g64740, At4g32410, At1g02730CP; CWSugar modification necessary to build CWsFagard et al. (2000); Boisson et al. (2001); Lane et al. (2001); Lukowitz et al. (2001); Burn et al. (2002); Gillmor et al. (2002);Hunter et al. (2012)

II. The role of the cytoskeleton in cytokinesis

1. The preprophase band

Preprophase bands (Fig. 1a) form in somatic cells before mitosis through the selective depolymerization of non-PPB cortical microtubules (MTs), leaving behind a MT belt that surrounds the nucleus (Dhonukshe & Gadella, 2003). As the cell cycle progresses, cortical actin filaments assemble alongside PPB MTs and help condense and narrow the PPB into an area called the cortical division zone (CDZ) (Liu et al., 2011b); however, PPB narrowing is not absolutely required for cytokinesis (Eleftheriou & Palevitz, 1992; Yoneda et al., 2004). In fact, PPBs are not essential for cell division, as somatic cells can progress through mitosis and complete cytokinesis following experimental ablation of PPBs, albeit with misoriented spindles and misaligned CPs (Mineyuki et al., 1991; Murata & Wada, 1991; Marcus et al., 2005). Furthermore, PPBs are not required for endosperm cellularization or for meiotic and subsequent generative divisions during gametogenesis (Otegui & Staehelin, 2000; Azimzadeh et al., 2008).

Figure 1.

The arrangement of the cytoskeleton, Golgi apparatus and trans-Golgi network/early endosome (TGN/EE) during cytokinesis. (a) In late G2, microtubules (MTs; green) and actin microfilaments (MFs; red) surround the nucleus (N; blue) to form the Preprophase band (PPB), which predicts the future division plane (i.e. the cortical division zone, CDZ). The preprophase band (PPB) is disassembled in late prophase/early prometaphase. Cortical microfilaments are enriched alongside the PPB, forming the ‘MF twin peaks’, which will persist throughout mitosis and cytokinesis. Golgi (lavender-to-purple) also encircles the nucleus to form the ‘Golgi belt’. (b) Cytokinesis follows nuclear reformation and is initiated by the assembly of the phragmoplast perpendicular to the CDZ from MT and MF remnants of the mitotic spindle. Vesicles (Golgi-derived and recycled; orange) travel along the phragmoplast and fuse at its center to form the tubular-vesicular network (TVN) stage of the young cell plate (CP; brown, center). (c) In late cytokinesis, the cell plate (CP) and the phragmoplast have expanded toward the cortical division site of the parental plasma membrane. MT polarity within the phragmoplast is indicated by + (plus end) and – (minus end) signs in panels (b) and (c). Brackets in panels (b) and (c) denote the actin-depleted zone, which persists throughout mitosis and cytokinesis and marks the CDZ. (Note: proteins, most organelles and portions of the MF twin peaks have been omitted for clarity).

Proteins shown to localize to and promote the formation of PPBs include the Arabidopsis proteins TONNEAU1a (TON1a), TON1b, TON2, MICROTUBULE ORGANIZATION 1 (MOR1) and Arabidopsis thaliana KINESIN ungrouped clade, gene A (AtKINUa) (Traas et al., 1995; Whittington et al., 2001; Camilleri et al., 2002; Kawamura et al., 2006; Sakai et al., 2008; Azimzadeh et al., 2008; Malcos & Cyr, 2011). TON2/FASS and maize orthologs DISCORDIA1 (DCD1) and ALTERNATIVE DISCORDIA1 (ADD1) are putative regulatory subunits of PROTEIN PHOSPHATASE 2A (PP2A) (Camilleri et al., 2002; Wright et al., 2009), and TON2 genetically interacts with the PPB MT-binding proteins TON1a/b and MOR1 (Kirik et al., 2012). A possible model is that the putative phosphatase activity of TON2-associated PP2A may function to positively regulate TON1 and MOR1 binding to and stabilization of PPB MTs, wheras phosphorylation of TON1 and MOR1 by the kinase CDKA/CDC2aAt, which, following its activation by B-type cyclins in late prophase, associates with and promotes disassembly of mature PPBs (Imajuku et al., 2001; Weingartner et al., 2001), may inhibit their function. Similarly, PPB association of the Arabidopsis kinesin, AtKINUa, which contains a CDKA phosphorylation site (Azimzadeh et al., 2008; Malcos & Cyr, 2011), may be regulated by TON2-PP2A and CDKA activities. Additionally, AtKINUa contains a D-BOX motif, and thus regulation of its PPB association may also be dependent upon ubiquitination and proteasome-mediated degradation (Malcos & Cyr, 2011).

The PPB disassembles at late prophase/early prometaphase, well in advance of mitotic nuclear breakdown, yet its former position accurately foretells the site of the growing CP. The Arabidopsis TANGLED 1 (TAN1) MT-associated protein and Ran GTPase ACTIVATING PROTEIN 1 (RanGAP1) localize to the PPB and remain at the CDZ after PPB breakdown, serving as a persistent spatial marker throughout mitosis and cytokinesis (Walker et al., 2007; Xu et al., 2008). Although PPBs form in tan1 mutants and RanGAP1 RNAi lines, they are misoriented, resulting in distorted CPs and cell walls (CWs) (Smith et al., 1996; Cleary & Smith, 1998; Walker et al., 2007; Xu et al., 2008). Localization of TAN1 and RanGAP1 depends upon TON2 and the redundant pseudokinesins PHRAGMOPLAST-ORIENTING KINESIN1 (POK1) and POK2 (Müller et al., 2006; Walker et al., 2007; Xu et al., 2008). POK1 interacts with TAN1 and RanGAP1, and, consistent with this, pok1 pok2 double mutants mimic tan1 and RanGAP1 RNAi phenotypes (Müller et al., 2006; Walker et al., 2007; Xu et al., 2008).

The Arabidopsis AUXIN-INDUCED IN ROOT 9 (AIR9) protein localizes to PPBs, but this localization is not maintained throughout mitosis. Yet, at late stages of cytokinesis, AIR9 reappears at the leading edge of the CP, and, following CP insertion at the cortical division site (CDS; i.e. the site within the CDZ where the CP attaches to the parental CW), redistributes along the new CW (Buschmann et al., 2006). Callose, which is removed and replaced by cellulose in mature CWs (Samuels et al., 1995), is absent in the AIR9-positive, newly formed CWs (Buschmann et al., 2006), suggesting a role for AIR9 in recognition and maturation of CPs and CWs that join the parental PM at the CDS, which presumably possesses memory cues that mark the former PPB position.

Preprophase band memory cues are likely established and maintained through the selective removal and/or exclusion of proteins from the CDZ. Despite their role in PPB formation, actin microfilaments (MFs) diminish at the CDZ as the PPB disassembles, and this actin-depleted zone persists throughout mitosis and cytokinesis (Hoshino et al., 2003; Panteris, 2008). In some cells, cortical actin accumulations are evident on either side of the CDZ, forming the so-called MF twin peaks, which is proposed to provide structural support to guide the expanding phragmoplast and growing CP toward the CDS (Cleary et al., 1992; Sano et al., 2005; Panteris, 2008). Consistent with this idea, depolymerization of cortical MFs before the appearance of the actin-depleted zone results in the formation of distorted CPs (Hoshino et al., 2003). The kinesin motor protein KINESIN CDKA-ASSOCIATED1 (KCA1) is excluded from the CDZ throughout mitosis and cytokinesis. This KCA-depleted zone forms only if intact PPB MTs are present; however, its maintenance following PPB disappearance is not MT- or MF-dependent (Vanstraelen et al., 2006). The localization of KCA1, like TON1 and AtKINUa, is negatively regulated by CDKA (Imajuku et al., 2001; Vanstraelen et al., 2006). Sustained exclusion of MFs and/or KCA1 from the CDZ throughout mitosis and cytokinesis suggests the cell has the ability to remember the former position of the PPB so that, later on, it can properly position the phragmoplast and CP.

2. The phragmoplast

Following completion of nuclear division in somatic plant cells, MFs and MT remnants of the mitotic spindle grow and reassemble into the phragmoplast, an antiparallel cytoskeletal array oriented perpendicular to the division plane, centered within the CDZ (Fig. 1b,c). Phragmoplast MTs and MFs coalign into a bipolar arrangement with their plus ends toward the phragmoplast midzone. It is upon this cytoskeletal scaffolding that vesicles are guided to the phragmoplast midzone wherein they fuse to build the CP. The phragmoplast is dynamic and must expand toward the cell cortex to allow the CP growing within it to expand centrifugally. This is accomplished as MTs and MFs depolymerize at the center of the phragmoplast and assemble at its leading edge (Liu et al., 2011a).

Early in vitro studies showing that fluorescently labeled tubulin polymerized onto extant MT plus ends, resulting in the appearance of a fluorescent band at the phragmoplast midzone, indicated that phragmoplast MTs are dynamic. This band widened and moved away from the midzone in the presence of excess unlabeled tubulin, suggesting that phragmoplast MT dynamics resulted from MT treadmilling and putative GTPase-dependent mechanoenzyme-mediated MT translocation toward the distal ends of the phragmoplast (Asada et al., 1991). However, subsequent studies examining phragmoplast MT turnover rates, and localization of EB1, a MT plus-end tracking protein, and various subunits of the negative-end-binding γ-tubulin ring complex (γ-TuRC) throughout the phragmoplast, were inconsistent with the phragmoplast MT treadmilling and translocation model (Liu et al., 1994; Hush et al., 1994; Dryková et al., 2003; Kumagai et al., 2003; Chan et al., 2003; Van Damme et al., 2004; Bisgrove et al., 2008; Zeng et al., 2009; Kong et al., 2010; Nakamura et al., 2010). Live cell imaging of EB1-GFP dynamics indicated that growing MT plus ends do not exclusively reside at the phragmoplast midzone, and phragmoplast MTs elongate both away from and toward the midzone (Ho et al., 2011). Furthermore, fluorescence recovery after photobleaching experiments and computer modeling using YFP-tubulin indicated that MT dynamic instability predominantly drives phragmoplast morphology, and that phragmoplast MT asymmetry is achieved by an increase in the number and rate of MTs that polymerize toward the CP rather than away (Smertenko et al., 2011).

Microtubule nucleation and growth are promoted by γ-TuRC, a complex of γ-tubulin and several γ-tubulin complex proteins, including GCP2 (Nakamura et al., 2010) and the GCP-WD ortholog NEDD1 (Zhu et al., 2011; Zeng et al., 2009). In animals and fungi, the γ-TuRC stabilizes MT minus ends to form the functional core of MT organizing centers (centrosomes and spindle pole bodies, respectively; Lüders & Stearns, 2007), assemblies that plant cells lack. Nonetheless, plant γ-TuRCs are capable of nucleating MTs on the sides of extant MTs (Murata et al., 2005). In nedd1 mutants, MTs fail to organize into a proper phragmoplast, leading to defective CP consolidation (Zeng et al., 2009). Plant orthologs of augmin complex subunits, which promote recruitment of γ-TuRC to spindles and polymerization of MTs in mammals (Zhu et al., 2011), are required for γ-tubulin recruitment, phragmoplast MT organization and CP formation (Zeng et al., 2009; Nakamura et al., 2010; Ho et al., 2011; Hotta et al., 2012). Thus, at least one way in which acentrosomal plant cells nucleate and organize mitotic MT arrays is through augmin-mediated association of γ-TuRC and MTs.

While MTs are the predominant cytoskeletal elements comprising the phragmoplast, MFs are also present (Hepler et al., 2002; Sano et al., 2005; Yu et al., 2006). Live cell imaging has shown that actin polymerization occurs at MF plus ends just outside the phragmoplast midzone and at the CP leading edge (Smertenko et al., 2010). The specific role of MFs in cytokinesis is unclear. Drug treatments that promote MF disassembly delay or inhibit CP formation (Schmit & Lambert, 1988; Valster et al., 1997; Kovar et al., 2006), whereas act2 and act2 act7 mutants have unaltered cell division (Baluška et al., 2001; Gilliland et al., 2002; Nishimura et al., 2003). However, it is difficult to assess the role of a particular population of MFs during cell division since actin mutants and MF-depolymerizing drugs have the potential to disrupt all actin-dependent cellular processes. Characterization of the formin proteins, which facilitate nucleation, capping, bundling and severing to regulate MF polymerization/depolymerization rates that influence MF dynamics (Staiger & Blanchoin, 2006; Blanchoin & Staiger, 2010), has been more informative of the role of actin in cytokinesis. The Arabidopsis formins AtFH5, AtFH8 and AFH14 are required for cytokinesis, but possess diverse MF-regulating activities necessary for CP formation (Ingouff et al., 2005; Kovar et al., 2006), mitotic progression (Xue et al., 2011), and phragmoplast MT and MF alignment (Li et al., 2010a; Wang et al., 2012), and thus may differentially regulate MF organization and/or function of during cytokinesis.

3. Cytoskeletal motor proteins

Higher plants use kinesin and myosin cytoskeletal motor proteins for ATP-dependent movement along MTs and MFs, respectively (Hepler et al., 2002). Relative to other eukaryotes, there appears to have been a major expansion in the number of kinesins encoded by plants (61 in Arabidopsis vs 45 in humans; Zhu & Dixit, 2008; Li et al., 2010b; Hotta et al., 2012), many of which may serve plant-specific functions at the PPB or phragmoplast for short-range movement of organelles and vesicles (Cai & Cresti, 2012). Myosins have been shown to drive the long-distance movement and repositioning of endomembrane compartments, including Golgi and Golgi-derived vesicles, during polar cell expansion; however, the role of myosins in plant cytokinesis is less clear. Studies using myosin-inhibiting drugs indicated that myosins influence phragmoplast alignment and CP expansion (Hepler et al., 2002), yet plant myosins are thought to preferentially associate with bundled MF cables, which are not prevalent during plant cell division (Smith & Oppenheimer, 2005; Thomas, 2012), raising questions about the necessity of myosins during cytokinesis.

Consistent with the arrangement of antiparallel MT plus ends at the phragmoplast midzone, it has been hypothesized that MT plus-end-directed kinesins facilitate the transport of vesicles along phragmoplast MTs to the midzone to deliver material for CP construction. Immunolocalization and live cell imaging of fluorescent fusion proteins have shown that many of the kinesins encoded by Arabidopsis (Li et al., 2010b), tobacco (Asada et al., 1997), rice (Lee & Liu, 2000), cotton (Preuss et al., 2003; Xu et al., 2009), and carrot (Barroso et al., 2000) localize to the phragmoplast in vivo. Of these, however, only the Arabidopsis PHRAGMOPLAST-ASSOCIATED KINESIN-RELATED MOTOR PROTEIN 2 (PAKRP2) decorates phragmoplast MTs in a punctate fashion, consistent with localization on CP transport vesicles (Lee et al., 2001).

Other plus-end-directed kinesins have been shown to localize to and/or be required for the construction, maintenance or dynamics of the phragmoplast. The tobacco NPK1-ACTIVATING KINESIN-LIKE PROTEIN 1 (NACK1) and NACK2 (Nishihama et al., 2002), Arabidopsis AtNACK1/HIK and AtNACK2/STD/TES (Strompen et al., 2002; Tanaka et al., 2004) and rice DBS1/OsNACK (Sazuka et al., 2005), as well as the Arabidopsis PHRAGMOPLAST-ASSOCIATED KINESIN-RELATED PROTEIN 1 (AtPAKRP1) and AtPAKRP1L (Lee & Liu, 2000; Pan et al., 2004; Lee et al., 2007), localize to the phragmoplast midzone and are directly implicated in phragmoplast organization and/or dynamics necessary for CP expansion. However, the function of the TOBACCO KINESIN-RELATED PROTEIN 125 (TKRP125) protein, and its orthologs in carrot, DcKRP120, and Arabidopsis, AtKRP125c, which localize along phragmoplast MTs (Asada et al., 1997; Barroso et al., 2000; Bannigan et al., 2007), remains to be determined.

Arabidopsis MT minus-end-directed ATK5, KCA1 and KCA2 localize to the phragmoplast midzone (Vanstraelen et al., 2004, 2006; Ambrose et al., 2005), which seems paradoxical as this site was thought to be occupied exclusively by MT plus ends. However, according to the dynamic instability model for phragmoplast MT dynamics, MT minus ends also likely reside at the phragmoplast midzone (Smertenko et al., 2011). Furthermore, kinesins not only function in directional transport, but can regulate organization of MT arrays through translocation, depolymerization, bundling or crosslinking to other cellular components (Zhu & Dixit, 2008). Consistent with this, localization of ATK5 and KCA1 at or near the phragmoplast midzone is independent of motor domain activity (Vanstraelen et al., 2004, 2006; Ambrose et al., 2005). The distribution, levels and localization of KCA1 and KCA2 are also likely mediated by their phosphorylation status (Vanstraelen et al., 2004, 2006), and ubiquitin-mediated degradation (Malcos & Cyr, 2011).

Various other minus-end-directed plant-specific kinesins that function in cytokinesis contain domains that modulate their activity during cell division such as calpain homology and calmodulin-binding protein domains. KINESIN-LIKE CALMODULIN-BINDING PROTEINs (KCBPs) from various plant species, including Arabidopsis, tobacco, spiderwort, cotton, and blood lily, localize to the PPB and phragmoplast where they likely mediate Ca2+-calmodulin-regulated MT bundling (Bowser & Reddy, 1997; Smirnova et al., 1998; Vos et al., 2000; Preuss et al., 2003). Arabidopsis, tobacco and rice members of the KINESIN WITH CALPAIN HOMOLOGY DOMAIN (KCH) family associate along the entire length of phragmoplast MTs (Frey et al., 2010; Buschmann et al., 2011), or the phragmoplast midzone (Xu et al., 2009), and can bind and bundle both MTs and MFs (Preuss et al., 2004; Xu et al., 2009; Umezu et al., 2011; Klotz & Nick, 2012), likely coupling these two cytoskeletal networks during cytokinesis.

Given the importance and number of kinesins involved in phragmoplast cytoskeletal dynamics and vesicle transport, it is not surprising that there exist numerous diverse potential modes of kinesin regulation, including the action of kinases. As discussed earlier, the kinase CDKA regulates the activity of a number of proteins, including PPB- and phragmoplast-associated kinesins. Kinesins themselves also regulate kinase signaling cascades. In tobacco, NACK1 and/or NACK2 kinesin motor proteins activate the NACK-PQR kinase cascade, comprising the mitogen-activated protein kinase kinase kinase (MAPKKK), NUCLEUS- AND PHRAGMOPLAST-LOCALIZING KINASE 1 (NPK1), the MAPKK, NQK1, and the MAPK, NRK1 (Calderini et al., 1998; Nishihama et al., 2001, 2002; Soyano et al., 2003; Takahashi et al., 2004; Komis et al., 2011). NRK1 in turn regulates a number of substrates; including the MT bundling protein MAP65-1, which is inactivated by phosphorylation, allowing for phragmoplast and CP expansion (Sasabe et al., 2006; Smertenko et al., 2006). In Arabidopsis, AtNACK1 and AtNACK2 kinesins activate an orthologous NACK-PQR cascade, which is required for cytokinesis, consisting of the MAPKKKs, ANP1, ANP2 and ANP3 (Krysan et al., 2002), the MAPKK ANQ1 (Beck et al., 2010, 2011), and several downstream MAPKs, including MPK4a and MPK6, and target substrates, including MAP65-1, -2, and -3 (Müller et al., 2004, 2010; Smertenko et al., 2004; Sasabe et al., 2006, 2011; Sasabe & Machida, 2006; Beck et al., 2010, 2011; Takahashi et al., 2010; Kosetsu et al., 2010; Komis et al., 2011; Bögre, 2011).

The TWO-IN-ONE (TIO) kinase interacts with the kinesins PAKRP1 and PAKRP1L and likely modulates their localization and/or activity at the phragmoplast midzone via phosphorylation (Oh et al., 2012). RUNKEL (RUK), a MT-associated protein with a putative kinase domain, localizes to PPBs, phragmoplast midzones and the expanding CP. However, the kinase activity of RUK is not required for cytokinesis (Krupnova et al., 2009), thus its function during cytokinesis is unclear. There are several additional kinases that influence cell division through their regulation of cell cycle progression or of cell polarity to regulate asymmetric divisions necessary for proper tissue patterning, including Arabidopsis AURORA 1 (AUR1); however, this kinase affects cytokinesis only indirectly, perhaps through its ability to phosphorylate HISTONE H3 (Kurihara et al., 2006; Demidov et al., 2009; Van Damme et al., 2011b).

III. Membrane dynamics during cytokinesis

1. Endomembrane organization

During division, the plant secretory pathway undergoes numerous organizational changes to facilitate the initiation, consolidation and maturation of the CP and segregation of endomembranes into daughter cells. The trans-Golgi network (TGN) and endosomes are the major compartments through which newly synthesized and endocytosed proteins, lipids and CW polysaccharides traffic to and from the CDZ and developing CP. In plants, the TGN, also known as the partially coated reticulum (Pesacreta & Lucas, 1984; Tanchak et al., 1988), mediates delivery of secretory proteins to the PM and vacuole. Originally thought to be distinct from early endosomes (EEs), more recent studies demonstrate that the TGN can move independently of the Golgi apparatus (Staehelin & Kang, 2008; Toyooka et al., 2009; Viotti et al., 2010), and is the initial compartment to which endocytosed PM proteins and the dye FM4-64, a marker of bulk PM endocytosis, are delivered (Dettmer et al., 2006; Lam et al., 2007; Chow et al., 2008; Viotti et al., 2010). Together these data support the current view of the TGN as a transiently mobile TGN/EE hybrid compartment through which PM- and vacuole-destined secretory and endocytic cargo pass.

In early mitosis, Golgi and TGN/EEs accumulate near the PPB (Dixit & Cyr, 2002) and in a subcortical ring, the ‘Golgi belt’, surrounding the future site of CP formation (Nebenführ et al., 2000) (Fig. 1a). Subsequently, vesicles carrying newly synthesized proteins and polysaccharides, including xyloglucans and arabinogalactans (Samuels et al., 1995; Seguí-Simarro et al., 2004), traffic along phragmoplast MTs to initiate and provide material for CP construction. In particular, the cytokinesis-specific syntaxin-related SNARE, KNOLLE (KN), which is required for CP formation (Lauber et al., 1997), traffics through the Golgi and TGN/EE before appearing at the division plane (Chow et al., 2008). Similarly, newly synthesized PM syntaxins, SYP121, SYP112, and SYP132, are targeted to the CP when expressed under control of the M-phase-specific KN promoter (Müller et al., 2003; Reichardt et al., 2007, 2011), indicating that trafficking of biosynthetic secretory cargo from the TGN/EE is, perhaps by default (Touihri et al., 2011), polarized toward the division plane rather than the PM during cytokinesis.

In addition to the delivery of newly synthesized cargo, mature CW-derived pectins and xyloglucans (Baluška et al., 2005), and FM4-64 (Dhonukshe et al., 2006; Dettmer et al., 2006; Reichardt et al., 2007) localize to the CP during cytokinesis. Direct fusion of late endosomes (i.e. multivesicular bodies/prevacuolar compartments; MVBs/PVCs) with the CP was postulated to mediate delivery of this internalized material to the CP (Dhonukshe et al., 2006). However, the TGN/EE marker vacuolar H(+)-ATPase subunit a1 (VHA-a1), and markers of the late endocytic MVB/PVC compartments, do not localize to the CP, bringing into question the generality of this model (Dettmer et al., 2006; Chow et al., 2008; Lam et al., 2008). Indeed, echidna (ech) mutants, defective for a protein required for proper TGN/EE structure and function, show mislocalization of VHA-a1 to the CP and tonoplast (Gendre et al., 2011), suggesting that endosomes may fuse with CPs, albeit under abnormal conditions. More likely, internalized CW components and proteins reach the CP during cytokinesis following their endocytosis and delivery to the TGN/EE. Analysis of other PM proteins that undergo constitutive or regulated endocytosis and cycling between the PM and TGN/EE, and the use of the inhibitor, Endosidin1, which selectively affects the recycling of PM proteins (e.g. the brassinosteroid receptor BRI1 and the auxin carriers PIN-FORMED 2 (PIN2) and AUX1, but not PIN1; Robert et al., 2008) could be used to further address this. Such studies may also address the question of whether newly synthesized and recycled CW components and PM proteins are delivered to the CP via a distinct or a common class of TGN/EE-derived vesicles.

Although endocytic cargo localizes to developing CPs, whether its delivery to the division plane is essential for cytokinesis is controversial. Based on the finding that the late endosomal trafficking inhibitor, wortmannin, did not affect CP formation, it was concluded that endocytosis was not required for cytokinesis (Reichardt et al., 2007). However, wortmannin has other less well-defined effects, including inhibition of clathrin dynamics (Ito et al., 2012), and thus it is best to exercise caution when interpreting its effects on CP formation. By contrast, elch mutants, defective in a subunit of the MVB ESCRT-I sorting complex (Spitzer et al., 2006), and expression of a dominant negative form of the MVB/PVC Rab GTPase, RABF2b(S24N) (Dhonukshe et al., 2006), result in cytokinesis defects. Further experiments are therefore necessary to reconcile these contradictory findings about the necessity for delivery of internalized CW components and PM proteins to CP formation.

2. Cell plate vesicle trafficking machinery

The formation, targeting and fusion of transport vesicles within each branch of the plant exocytic and endocytic secretory pathway are regulated by a large number of evolutionarily conserved cytosolic and membrane-associated factors (Bassham et al., 2008). Formation of secretory and endocytic vesicles involves the assembly of distinct coat protein complexes (e.g. COPII, COPI and clathrin) that drive membrane budding and the selection of cargo proteins (Hwang & Robinson, 2009). Despite the morphological data that indicate that CP- and PM-destined vesicles are transported from the Golgi and TGN/EE, the molecular machinery involved in their formation and loading with cargo remains to be more clearly defined. Clathrin-coated vesicles (CCVs) have been implicated in the transport of proteins, including vacuolar proteins (Harley & Beevers, 1989; Song et al., 2012), from the Golgi and TGN/EE; however, this view was recently challenged (Scheuring et al., 2011). Perhaps CCVs may thus function in the trafficking of newly synthesized and/or recycled material from the TGN/EE to the PM and CP. In addition to CCVs, other vesicle types have also been observed to form from the TGN/EE, including dense vesicles containing vacuolar storage proteins (Hohl et al., 1996; Hinz et al., 1999), and secretory vesicles (Staehelin & Kang, 2008). TGN/EE-derived secretory vesicle clusters have been implicated in the delivery of proteins and polysaccharides (Toyooka et al., 2009) to the PM and CP; however, their formation remains to be determined.

While limited understanding of their formation exists, we have a better understanding of the cadre of proteins involved in directional transport, docking and fusion of CP-destined vesicles. Rab/Ypt proteins are monomeric small GTPases, which, in their GTP-bound form, recruit divergent effectors to coordinate the formation, transport, tethering and fusion of secretory and endocytic vesicles and organelles throughout the biosynthetic and endocytic trafficking pathways (Nielsen et al., 2008; Woollard & Moore, 2008; Hutagalung & Novick, 2011). Of the 57 Rabs encoded by Arabidopsis, 26 have been annotated to encode RabA-type GTPases, which are most closely related to members of the mammalian Rab11 subclass that function at recycling endosomes (Van Ijzendoorn, 2006) and are critical for later stages of cytokinesis (Skop et al., 2001). Likewise, in Saccharomyces cerevisiae, the Ypt31/Ypt32 pair of RabA/Rab11-related proteins are required for polarized secretion of TGN-derived vesicles (Jedd et al., 1997; Lipatova et al., 2008) and recycling from the PM through EEs to the Golgi (Chen et al., 2011a).

Plant RabA proteins have been divided into six subclasses (RABA1-6) (Rutherford & Moore, 2002), with various members, including NtRAB11, RABA1b, RABA2a and RABA4b, functioning in the delivery of material from the TGN/EE to the PM (Preuss et al., 2004, 2006; De Graaf et al., 2005; Chow et al., 2008; Feraru et al., 2012). Similarly, RABA2 and RABA3 colocalize with VHAa1 and internalized FM4-64 within a subcompartment of the TGN/EE and are PM-associated, suggesting that they function in exocytosis as well (Chow et al., 2008). During cytokinesis, RABA2 and RABA3 colocalize extensively with KN in both the TGN/EE and at the CP leading edge, which is embedded within a ribosome-excluding CP assembly matrix (CPAM) (Otegui & Staehelin, 2004) and is the site to which CP-destined vesicles are delivered and fuse with the expanding CP. GTPase-inactive or GTP-bound-mimic mutant RABA2 proteins also accumulate at the CP (Chow et al., 2008) and result in cytokinesis defects, including abnormal cell files and multinucleated cells with incomplete CWs. Together these data indicate that RabAs likely regulate the trafficking of TGN/EE-derived vesicles containing newly synthesized, and possibly recycled, material to the division plane.

In addition to the RabA GTPases, the MVB/PVC Rabs, RABF1 and RABF2b (Ueda et al., 2004), localize to CPs, and expression of a GTPase mutant form of RABF2b results in cytokinesis defects (Dhonukshe et al., 2006). However, localization of RABF2b and other late edosomal/MVB/PVC markers, for example, BP80 and the ARF guanine nucleotide exchange factor (GEF), GNOM, to the CP was notconfirmed (Chow et al., 2008). Furthermore, in contrast to the preferential association of RABA2a/A2b/A3 and KN with the CP leading edge (Chow et al., 2008), RABF1 and RABF2b were distributed throughout the division plane (Dhonukshe et al., 2006), inconsistent with a role in membrane delivery and addition to the growing CP. Thus further work is necessary to understand whether there is a direct role for late endocytic Rabs in CP formation.

Rabs are activated by RabGEFs, inactivated by GTPase activating proteins (RabGAPs), and recycled by GDP dissociation inhibitors (RabGDIs). At the present time, little is known about the identity and function of plant RabGAPs and RabGDIs in membrane trafficking to the PM and CP. However, the putative late secretory pathway RabGEFs, TRAPPII (discussed later) and SCD1, are required for CP formation and cell expansion. Loss-of-function and missense mutations in the SCD1 N-terminal DENN domain cause defects in cytokinesis, cell expansion and innate immunity responses to pathogens (Falbel et al., 2003; Korasick et al., 2010). DENN domains are constituents of proteins that associate with Rabs (Levivier et al., 2001) and function as animal Rab GEFs (Sato et al., 2008; Allaire et al., 2010; Yoshimura et al., 2010). In particular, Connecdenn/DENND1A/RME-4 functions as a CCV-associated GEF for Rab35 (Allaire et al., 2010), a Rab required for CCV trafficking, endosomal recycling and cytokinesis in mammalian cells (Kouranti et al., 2006; Patino-Lopez et al., 2008). As plants lack Rab35 homologs, SCD1 may function as a GEF for members of the significantly expanded and divergent plant RabA/Rab11 family (Rutherford & Moore, 2002) or other Rab(s) that function in late secretory membrane trafficking during cytokinesis and cell expansion.

Following their formation and trafficking to their final destination, vesicles are brought into close proximity to their target membrane through the action of Rabs and cognate tethering factors. In yeast, the multisubunit TRAPPII and exocyst tethering complexes function in the late Golgi and mediate TGN/EE to PM trafficking, respectively (Cai et al., 2007). TRAPPII consists of the seven subunits of the TRAPPI complex, which is required for endoplasmic reticulum (ER) to Golgi transport, and three additional subunits, Trs65, Trs120, and Trs130 (Sciorra et al., 2005). Both yeast TRAPPI and TRAPPII function as Rab GEFs. TRAPPII serves as a GEF for Ypt31/32, the yeast Rab11 orthologs, and is required for Rab11 localization and cleavage furrow ingression during cytokinesis in Drosophila (Robinett et al., 2009). Mammalian TRAPPII, however, localizes to the early Golgi and activates the early secretory pathway Rab protein, Rab1 (Yamasaki et al., 2009).

By electron microscopic (EM) tomography, putative tethering factors have been observed to be associated with fusing CP vesicles (Otegui & Staehelin, 2004; Seguí-Simarro et al., 2004) and recent studies reveal that TRAPPI, TRAPPII, and the exocyst are required for cytokinesis (Fendrych et al., 2010; Thellmann et al., 2010; Qi et al., 2011). Disruption of the Arabidopsis TRAPPI/TRAPPII-subunit-encoding AtTRS33, AtTRS120 or AtTRS130 genes results in defective CP formation and cell expansion (Jaber et al., 2010; Thellmann et al., 2010; Qi et al., 2011). Further localization and genetic interaction studies suggest that the Arabidopsis TRAPPII complex likely serves as a GEF for RabAs in the trafficking of TGN/EE-derived vesicles containing newly synthesized and endocytosed proteins to the PM and CP (Qi et al., 2011; Qi & Zheng, 2011).

Vesicle fusion at the PM and developing CP is aided by the exocyst complex, an evolutionarily conserved assembly that is localized to sites of active exocytosis on the PM in polarized mammalian cells (Cai et al., 2007). Plants encode multiple genes for each of the exocyst subunits, the most extreme example being the Arabidopsis Exo70 subunit, which is encoded by a family of 23 genes (Synek et al., 2006) that are primarily expressed in cells that require active exocytosis (Hála et al., 2008; Li et al., 2012). Arabidopsis and maize exocyst subunit mutants display multiple defects in plant growth and development, cell expansion, and incomplete epidermal cytokinesis (Wen et al., 2005; Cole et al., 2005; Hála et al., 2008; Fendrych et al., 2010; Li et al., 2012). The exocyst complex is associated with the leading edge, maturing regions and parental PM fusion sites of the CP (Fendrych et al., 2010), suggesting that the exocyst mediates the addition and consolidation of CP membrane at all stages of CP formation.

Whether additional tethering complexes function directly in the tethering of transport vesicles required for CP assembly remains to be determined. Characterization of Arabidopsis mutants expressing defective genes for subunits of the GARP complex, which functions in endosome to late-Golgi trafficking in yeast (Cai et al., 2007) and is required for pollen tube elongation (Wang et al., 2008), did not reveal a role for this tethering complex in CP formation (Thellmann et al., 2010); however, it cannot be ruled out that other tethering complexes (e.g. TRAPPII) could compensate for the role of GARP in membrane retrieval from the developing CP.

Ultimately it is the assembly of specific trans-SNARE protein complexes between transport vesicles and target membranes that drives membrane bilayer fusion. These complexes comprise four α-helical SNARE domains that are contributed by multiple (Q-type) target membrane t-SNAREs and one (R-type) v-SNARE on the vesicle/donor membrane (Südhof & Rothman, 2009). Formation of the initial trans-SNARE complex between opposing membranes is regulated by a host of proteins, including their activation by Sec1/Munc18-like (SM) proteins, and N-ethylmaleimide sensitive factor (NSF)- and α-SNAP-mediated ATP-dependent disassembly. The most likely SNARE complex involved in CP vesicle fusion comprises the t-SNAREs, KN (Lauber et al., 1997) and SNAP33 (Heese et al., 2001), and the functionally redundant t-SNAREs, VAMP721 and VAMP722, all of which localize to and are required for the formation of the CP (Kwon et al., 2008; Zhang et al., 2011). Another t-SNARE, NSPN11, shows CP localization and KN interaction (Zheng et al., 2002); however, as nspn11 mutants show no detectable cytokinesis defects, it is unknown whether this SNARE is involved in CP membrane fusion. KN interacts with NSF/α-SNAP (Rancour et al., 2002) and the Arabidopsis SM protein, KEULE, which is essential for CP formation (Waizenegger et al., 2000; Assaad et al., 2001; Park et al., 2008). Although SM proteins are generally thought to interact with the closed/inactive form of syntaxin-type SNAREs to promote SNARE complex formation (Südhof & Rothman, 2009), KEULE appears to interact preferentially with a fusion-competent form of KN (Park et al., 2008), raising the interesting question of whether CP vesicle fusion is mechanistically similar to, or distinct from, other SNARE-mediated membrane fusion processes.

Of the SNAREs shown to be required for CP vesicle fusion, only KN specifically functions in cytokinesis. In nondividing cells, SNAP33 and VAMP721/722 interact with other members of the SYP1 t-SNARE family in the trafficking of CW material and PM proteins (Kwon et al., 2008; Zhang et al., 2011). In particular, VAMP721 and VAMP722, which localize to the VHA-a1-positive TGN/EE, function as v-SNAREs in the targeting and fusion of exocytic TGN/EE-derived vesicles with the PM (Zhang et al., 2011).

3. Cell plate membrane recycling

The flux of membrane to the division plane during cytokinesis is balanced by retrograde transport. Based on changes to the surface area of the CP throughout cytokinesis, it is estimated that c. 70% of the membrane delivered to the division plane is ultimately recycled (Samuels et al., 1995; Otegui et al., 2001). Recent studies have implicated clathrin-dependent membrane transport in the recycling of proteins and lipids during positioning, maturation and fusion of the CP with the parental membrane, as well as their redistribution following cytokinesis. While MFs have been shown to be integral to clathrin-mediated endocytosis (CME) in mammals and yeast (Mooren et al., 2012), the role of MFs and/or MTs in clathrin-dependent endocytosis from the PM or membrane recycling from the CP is not well understood.

Clathrin-dependent membrane transport is likely required for localized changes in the protein and/or lipid composition of the PPB-associated PM/CW domain before mitosis, which serve as cortical memory cues for CP fusion with the CDS. Support for this hypothesis came from EM analysis of chemically fixed dividing guard mother cells, which showed prominent CW thickenings, indicative of polarized secretion of PM and CW material to the CDS, and the presence of coated vesicles at the CDZ (Galatis et al., 1984). Although these thickenings are likely a unique aspect of guard cell cytokinesis, the use of electron tomographic analysis of high-pressure frozen/freeze-substituted samples and live cell imaging has provided convincing evidence for the enrichment of endosomes, endocytic clathrin-coated pits and CCVs at the CDZ in other cell types (Dhonukshe et al., 2005; Karahara et al., 2009; Van Damme et al., 2011a).

Electron microscopic studies of high-pressure frozen/freeze-substituted plant cells undergoing cytokinesis have also highlighted the importance of clathrin in the recycling of CP membranes during CP maturation. CCVs have been detected budding from maturing regions of the CP concomitant with the appearance of MVB/PVCs, organelles to which the CCVs are believed to deliver recycled membrane and cargo from the CP (Samuels et al., 1995; Otegui et al., 2001; Otegui & Staehelin, 2004; Seguí-Simarro et al., 2004; Dhonukshe et al., 2005; Seguí-Simarro & Staehelin, 2006; Tahara et al., 2007). Indeed, KN is recycled from the CP as it matures, and accumulates in MVBs/PVCs before being degraded in the vacuole (Tse et al., 2004; Reichardt et al., 2007; Boutté et al., 2010). Also, the PIN proteins are targeted to developing CPs during cytokinesis and recycled via clathrin-dependent membrane trafficking during CP maturation (Baluška et al., 2005; Dhonukshe et al., 2006, 2007; Mravec et al., 2011).

Recent studies have demonstrated that endocytosis, and particularly CME, is essential for plant growth, development and signaling (Dhonukshe et al., 2007; Robert et al., 2010; Kitakura et al., 2011; Adam et al., 2012). In mammalian cells, CME and the formation of CCVs at the TGN involves the coordinated interplay of accessory and regulatory proteins that is initiated by binding of the cargo adaptor protein (AP) complexes, AP-2 and AP-1, at the PM and TGN, respectively. Subsequently, the polymerization of clathrin triskelia, composed of heavy chain (CHC) and light chain (CLC) subunits, and the membrane remodeling activities of accessory proteins, are necessary for the invagination and release of the budding CCVs (Kirchhausen, 2000). Many evolutionarily conserved structural and regulatory proteins involved in CCV formation, such as CLC, CHC, AP-2 adaptins, dynamins and AP180/Epsin N-terminal homology (A/ENTH) domain-containing proteins (involved in membrane curvature), have been identified in plants (Blackbourn & Jackson, 1996; Holstein, 2002; Barth & Holstein, 2004; Kotchoni et al., 2009; Bednarek & Backues, 2010; Song et al., 2006). Live cell imaging has also demonstrated that clathrin, the adaptin-like TPLATE, the plant dynamin-related proteins and A/ENTH proteins localize to the CP (Konopka et al., 2008; Konopka & Bednarek, 2008; Fujimoto et al., 2010; Van Damme et al., 2011a; Ito et al., 2012; Song et al., 2006), suggesting that clathrin-mediated CP membrane recycling is mechanistically related to CCV formation at the TGN and during CME.

Dynamin and DYNAMIN-RELATED PROTEINS (DRPs) are a structurally similar but functionally diverse group of large GTPases. In plants, two distinct DRP families, DRP1 and DRP2, play essential and nonredundant roles in plant endomembrane trafficking and dynamics (Gu & Verma, 1997; Kang et al., 2003a,b; Collings et al., 2008; Backues et al., 2010; Taylor, 2011; Mravec et al., 2011). DRP1 and DRP2 colocalize together with CLC2 at the PM in dynamic foci that likely correspond to sites of CME (Konopka et al., 2008; Konopka & Bednarek, 2008; Fujimoto et al., 2010) and within maturing regions of the expanding CP (Fig. 2; S. K. Backues & S. Y. Bednarek, unpublished) (Kang et al., 2003a; Konopka et al., 2008; Fujimoto et al., 2008, 2010). drp1 and drp2 mutants also display defects in CP formation, such as incomplete CWs (Gu & Verma, 1997; Kang et al., 2003a; Collings et al., 2008; Mravec et al., 2011) or highly convoluted CP membranes (Backues et al., 2010), respectively. More specifically, in drp1 mutants, KN is mislocalized to the PM following completion of CP formation (Boutté et al., 2010), and clathrin-mediated recycling of PIN proteins from the developing CP is defective (Mravec et al., 2011).

Figure 2.

Time-lapse laser scanning confocal microscopy images of DYNAMIN-RELATED PROTEIN 1A (DRP1A)-GFP (left panels; green) and CLATHRIN LIGHT CHAIN 2 (CLC2)-mOrange (right panels; magenta) at the cell plate. Horizontal lines indicate the leading edges of the cell plate. Bar, 5 μm.

Imaging studies reveal that DRP1/2 also localize to the leading edge of the forming CP (Kang et al., 2003a; Konopka et al., 2008; Fujimoto et al., 2010) before the appearance of clathrin and A/ENTH proteins (Ito et al., 2012; Song et al., 2006) (Fig. 2; S. K. Backues & S. Y. Bednarek, unpublished). Indeed, DRP1 and possibly DRP2 ring complexes have been observed by EM to form constrictions in the early syncytial-type and somatic CP tubular networks in the absence of clathrin-coated buds (Otegui et al., 2001; Seguí-Simarro et al., 2004). Based on current models, dynamins are recruited to sites of CCV formation via their interaction with clathrin, lipids, the AP complex and A/ENTH proteins. Hence the localization of DRPs to the CP leading edge in the absence of clathrin and A/ENTH domain-containing proteins (Ito et al., 2012; Song et al., 2006) is likely to be distinct from their recruitment in CCV formation. Putative functions for the DRP rings at the CP leading edge include membrane tubulation, which may promote CP tubule fusion through the generation of regions of high curvature, and tethering or restricting the lateral diffusion of enzymes (e.g. callose synthases) and other proteins required for CP formation and maturation.

The plant-specific TPLATE protein contains a domain similar to regions found within the large subunit of the AP complexes and the β, γ1 and γ2 subunits of the COP1 coatamer protein complex (Van Damme et al., 2006), suggesting it functions in membrane dynamics. TPLATE colocalizes with CLC2 to growing CPs and later to the flanks of the CDS as the CP attaches to the parental PM (Van Damme et al., 2011a). RNAi-mediated knockdown of TPLATE perturbs this attachment, resulting in incomplete CWs (Van Damme et al., 2006). TPLATE was also shown to interact with CHC1 and CLC2. Together these data implicate TPLATE in clathrin-mediated membrane reorganization/recycling at the CP and/or parental PM necessary for integration of the CP at the CDS (Van Damme et al., 2011a).

4. The endoplasmic reticulum

In addition to the close association of Golgi, TGN/EE and MVBs with the division plane, maturation and fusion of the CP at the CDS are accompanied by the recruitment of ER membranes to the CDZ and the formation of a tightly associated reticular network surrounding the CP (Hepler, 1982; Schopfer & Hepler, 1991; Cutler & Ehrhardt, 2002; Otegui & Staehelin, 2004; Seguí-Simarro et al., 2004). This ER network may be the precursor to the cortical ER network that lies in close proximity to the PM in all plant cells (Lichtscheidl & Hepler, 1996; Staehelin, 1997).

Potential roles for the CP-associated ER are direct lipid transfer to the CP and/or establishment and maintenance of an appropriate ionic environment (e.g. supplying Ca2+) (Lichtscheidl & Hepler, 1996; Otegui & Staehelin, 2004; Seguí-Simarro et al., 2004) to support CW biosynthesis and CP membrane fusion and consolidation. Tubular elements of the CP-associated ER network also fuse and become entrapped, forming the desmotubule that traverses plasmodesmata (Hepler, 1982) between adjacent daughter cells. Coordinate assembly of the CP and expanding PM with ER membrane networks likely involves active membrane trafficking and fusion within the division plane.

Endoplasmic reticulum formation within the CDZ may be mediated by the abundant and conserved hexameric AAA (ATPases associated with diverse cellular activities) molecular chaperone, CDC48/p97, which is structurally related to NSF (Hanson & Whiteheart, 2005). Like NSF, CDC48/p97 functions in a variety of SNARE-dependent membrane fusion pathways including homotypic fusion of ER, transitional-ER (t-ER), and mitotic Golgi fragments (Latterich et al., 1995; Rabouille et al., 1995; Patel et al., 1998; Roy et al., 2000).

In plant cells, CDC48/p97 localizes to the cytoplasm, nucleus, PM and ER during interphase (Aker et al., 2007; Park et al., 2012) and at the phragmoplast midzone during late cytokinesis, and its function is required for cytokinesis (Park et al., 2012). However, unlike NSF, AtCDC48 does not interact with KN (Rancour et al., 2002), indicating that its chaperone activity is not required for the initiation of CP membrane fusion. Rather, based on its colocalization and interaction with the Golgi/ER t-SNARE, SYP31 (Rancour et al., 2002; Bubeck et al., 2008), an ortholog of the syntaxin5 SNARE protein required for assembly of t-ER (Roy et al., 2000) and ER network formation in mammals (Uchiyama et al., 2002), AtCDC48 and SYP31 may be required for ER formation within the CDZ critical for CP maturation. However, as CDC48/p97 activity is required for many other cellular processes, including ER-associated protein degradation (Meyer et al., 2012) and endocytosis (Ritz et al., 2011; Ramanathan & Ye, 2012), other roles for AtCDC48 in cytokinesis cannot be ruled out. Further characterization of ER membrane reticulons (Sparkes et al., 2010) and the GTPase RHD3 (Chen et al., 2009), which mediate tubular ER membrane formation and dynamics, may provide additional insight into the dynamics and role of ER membrane assembly in CP formation.

5. Cell wall and lipid requirements

In addition to their structural roles in cellular membranes, phosphotidylinositides (PIs) function as signaling molecules and influence the localization of PI-binding proteins. In dividing cells, the PI3P-specific probe YFP-2xFYVE localizes in punctate structures at or near the CP leading edge (Vermeer et al., 2006), while the PI4P marker, YFP-PHFAPP1, labels the growing CP and STmd-positive TGN/EEs (Vermeer et al., 2009), suggesting a role in CP expansion/maturation. PI3P/PI4P may facilitate the localization of factors required for CP formation, such as the DRP2 proteins, which contain a PI3P-/PI4P-binding PH domain (Bednarek & Backues, 2010). Similarly, Arabidopsis PATELLIN (PATL) localizes to growing CPs and contains a PI3P-binding domain homologous to that of Sec14, which functions in the formation of TGN-derived vesicles (Peterman et al., 2004).

Other lipids, including very long fatty acyl chain (VLFA) of phospholipids and sterols, are required for cytokinesis. Mutations within the Arabidopsis PASTICCINO2 (PAS2) gene, which encodes an elongase necessary for synthesis of VLFAs, delays or halts CP expansion concomitant with mislocalization of KN and RABA2a (Bach et al., 2011). The CP membrane sterol composition is also critical for maintenance of KN within the CDZ via endocytosis during late, but not early, cytokinesis (Boutté et al., 2010) and for the repolarization of PIN2 proteins after cytokinesis that localize to the CP during division (Men et al., 2008). Sterol-rich microdomains, which are crucial for cellulose biosynthesis (Schrick et al., 2000, 2002, 2004, 2012), may also be critical for maintaining the structural integrity and enzyme activities involved in CW formation within the developing CP. Sterol biosynthesis mutants fackel (fk), cephalopod1 (cph1) and hydra1 (hyd1) display cell division, cell expansion and embryonic patterning defects; specifically, typical cellulose-deficient phenotypes are observed, including multinucleated cells, incomplete CWs and CW thickenings as a result of ectopic callose and lignin deposition (Diener et al., 2000; Jang et al., 2000; Schrick et al., 2000, 2002; Souter et al., 2002).

During cytokinesis, vesicles carrying newly synthesized and recycled pectic polysaccharides, hemicelluloses, arabinogalactan proteins, and biosynthetic enzymes are delivered to the lumen of the growing CP to initiate CW formation (Staehelin & Moore, 1995; Baluška et al., 2005). Callose is the major CW material found in young CPs, and is later replaced by cellulose in the mature primary CW (Samuels et al., 1995). The Arabidopsis callose synthase mutant gsl8 displays multinucleated epidermal cells with incomplete CWs, implicating the necessity of GSL8 for successful cytokinesis (Töller et al., 2008; Chen et al., 2011b; Guseman et al., 2010). These phenotypes are shared by many sugar modification mutants, such as knopf (knf), radially swollen 3 (rsw3), cytokinesis-defective1 (cyt1), and korrigan (kor), and by several cellulose biosynthetic mutants, such as procuste1 (prc1), cellulose synthase A1 (cesA1) and cellulose synthase-like D1 (cslD), implying that their wild-type gene products may contribute to late CP formation and/or strengthening of the daughter CW (Zuo et al., 2000; Fagard et al., 2000; Lukowitz et al., 2001; Boisson et al., 2001; Lane et al., 2001; Gillmor et al., 2002; Burn et al., 2002; Hunter et al., 2012).

IV. Conclusion

Higher plant cytokinesis depends upon cytoskeletal dynamics and membrane trafficking for the demarcation of division orientation and the temporally and spatially distinct processes required for CP initiation, maturation and integration with the CDS. As highlighted in this review, in recent years there has been a vast increase of our knowledge of the molecular players and their roles in PPB formation, phragmoplast MT and MF dynamics, and CP membrane trafficking and fusion. However, there remains a number of significant questions regarding the mechanisms by which cytoskeletal dynamics and membrane trafficking are coordinated across time and space during cell division to execute cytokinesis. In particular, major questions to be addressed are how the PPB is initially defined and what molecular cues reside at or are excluded from the CDZ that direct CP fusion at the CDS. Likewise, the specific phragmoplast cytoskeletal dynamics required for, and the roles of endosomes and endocytic trafficking in, delivery and recycling of protein, lipid and CW material to and from the CP are poorly understood. As our inventory of the molecular players involved in cytoskeletal and membrane dynamics necessary for cytokinesis continues to increase, additional imaging and biochemical approaches, including the use of in vitro system(s) that reconstitute various stages of phragmoplast and CP biogenesis and function, will likely address some of the controversial issues and questions that we raised in this review and provide a more comprehensive understanding of plant cytokinesis.


The authors would like to thank past/current members of our laboratory for their contributions and helpful discussions in preparing this review. We also apologize to colleagues whose work we may not have discussed or cited. We gratefully acknowledge support from the National Science Foundation (grant no. 1121998) to S.Y.B.