NETWORKED2-subfamily proteins regulate the cortical actin cytoskeleton of growing pollen tubes and polarised pollen tube growth

in fertilisation To further characterise the role of Arabidopsis NET2 proteins in pollen development and fertilisation, we analysed the subcellular localisation of NET2A over the course of pollen grain development and investigated the role of the NET2 family using net2 loss-of-function mutants. (cid:1) We observed NET2A to localise to the F-actin cytoskeleton in developing pollen grains as it underwent striking structural reorganisations at speciﬁc stages of development and during germination and pollen tube growth. Furthermore, net2 loss-of-function mutants exhibited striking morphological defects in the early stages of pollen tube growth, arising from frequent changes to pollen tube growth trajectory. We observed defects in the cortical actin cytoskeleton and actin-driven subcellular processes in net2 mutant pollen tubes. (cid:1) We demonstrate that NET2 proteins are essential for normal actin-driven pollen development highlighting an important role for the NET2 family members in regulating pollen tube growth during fertilisation.


Introduction
The male reproductive cycle of angiosperms is an elaborate process, comprising the development and generation of highly specialised pollen grain cells within the anther locule; a process known as microgametogenesis. During fertilisation, the mature pollen germinates upon the stigma and extends a growing pollen tube, carrying nonmotile sperm cells through the pistil to the ovule, guided by extracellular signals secreted by the pistil (Kanaoka & Higashiyana, 2015).
The actin cytoskeleton is crucial for pollen development, germination and pollen tube growth (Gervais et al., 1994;Zonia et al., 1999;Gibbon et al., 1999;Vidali et al., 2001). During microgametogenesis actin regulates some processes including nuclear migration, mitotic divisions and cell expansion. Actin is also implicated in regulating specific reorganisations of cellular structure that occur in developing pollen, including fragmentation and dispersal of the pollen vacuole, and structuring of the plasma membrane, both of which are dependent on clathrinmediated endocytosis (CME; Kang et al., 2003;Yamamoto et al., 2003;Kim et al., 2013). In pollen germination and tube growth, actin regulates targeting of Golgi-derived secretory vesicles to the growing tip, cytoplasmic streaming and endocytosis (Vidali & Helper, 2000;Lee et al., 2008;Moscatelli et al., 2012;Rounds et al., 2014;Volger et al., 2015). During fertilisation, the actin cytoskeleton is believed to be controlled in response to extracellular signals secreted by the pistil to regulate pollen tube directional growth and navigation to the ovule. Each of these processes is mediated by highly organised and dedicated actin arrays, which are in turn regulated by a large number of actin-binding proteins.
We have recently identified one such protein, NETWORKED 2A (NET2A), a pollen-specific member of the NET superfamily of actin-binding proteins. We demonstrated that NET2A bound F-actin to the plasma membrane through interaction with the membrane-integral receptors, POLLEN RECEPTOR-LIKE KINASEs 4 and 5 (PRK4 and PRK5), to form physical links between the actin cytoskeleton and plasma membrane, so-called 'actin-membrane contact sites' (AMCSs; Duckney et al., 2017). PRKs are upstream regulators of extracellular signal transduction to the actin cytoskeleton during fertilisation (Tang et al., 2002(Tang et al., , 2004Lee et al., 2008;Wengier et al., 2010;Huang et al., 2014;Takeuchi and Highashiyama, 2016), suggesting important roles for NET2 subfamily proteins as downstream regulators of actin in response to extracellular guidance cues during fertilisation (Duckney et al., 2017). However, the importance of NET2 proteins in pollen has yet to be identified.
Here, we further investigated the potential roles for NET2 subfamily proteins in the male gametophyte using multiple approaches. Firstly, we analysed the subcellular localisation of NET2A to the actin cytoskeleton during pollen development, germination and pollen tube growth. We noted a dynamic association of NET2A with the actin cytoskeleton as it underwent structural reorganisations at specific stages in pollen development and tube growth, suggestive of its involvement in the regulation of actin in the male gametophyte. We next confirmed an important role for NET2 proteins in pollen tube growth through reverse genetic analysis of specific combinations of net2 mutants. We demonstrate that NET2 proteins are important in organising F-actin at the plasma membrane and regulating actin-driven subcellular processes including vesicle trafficking and cytoplasmic streaming in pollen tubes, potentially through the regulation of previously characterised AMCSs (Duckney et al., 2017).
Our data establish an important role for NET2 proteins in the male gametophyte and in angiosperm fertilisation. We further demonstrate the importance of the previously characterised NET2-mediated connections between the actin cytoskeleton and the plasma membrane in the organisation of the cytoskeleton in pollen tube growth.

Molecular biology
To generate the pLat52:GFP-RabA4B construct, the cDNA of RabA4B was cloned from Arabidopsis Col-0 pollen using PCR and cloned into the binary vector pB7GWF52 using the Gateway cloning system (Invitrogen).
For pollen-specific expression of constructs under the pLAT52 promoter, the binary expression vectors pB7GW52 (for expression with no tag) and pB7GWF52 (for expression of N-terminal GFP) were generated from pB7GW2 and pB7GWF2 respectively, as previously described (Duckney et al., 2017). The pLAT52 promoter sequence (Twell et al., 1990) was PCR amplified with added 5 0 SacI and 3 0 SpeI sites. Excision of the CaMV 35S:promoter sequence was performed using SacI/SpeI double restriction enzyme digestion, and the 5 0 -SacI-pLAT52-SpeI-3 0 DNA fragment was ligated into the excision site using T7 DNA ligase (NEB).

Reverse-genetics
The net2a.1 (GABI_234H12), net2b.1 (GABI_486H12), net2b.2 (GABI_132E08), net2d.1 (SALK_042729) and net2d.2 (SALK_035401) T-DNA insertion mutants were ordered from NASC. Homozygosity was confirmed using the primers described in Supporting Information Table S1. Total RNA was extracted from inflorescences (Qiagen) and cDNA was synthesised using Superscript III (Invitrogen). RT-PCR was used to confirm the absence of full-length transcripts in the net2 T-DNA mutants using the primers listed in Table S1. RT-PCR amplification of the housekeeping gene EF1a was performed as a control, and equal amounts of Col-0 and mutant cDNA were used.

Live cell imaging
Live pollen grains and germinated pollen tubes were imaged using laser scanning confocal microscopy (LSCM) (Leica TCS SP5 microscope). Whole anthers were excised and dissected to permit visualisation of developing grains through the anther wall. Images were acquired in multitrack sequential mode with laser line switching when imaging co-localisation of multiple fluorophores. Rapid differential interference contrast (DIC) imaging was performed using the Applied Precision OMX BLAZE Super-Resolution microscope (GE Healthcare) in DIC operating mode at a rate of 286 MHz.

Pollen tube growth assays
In vitro germination of Arabidopsis pollen was performed on solid medium as previously described (Li et al., 1999). Germination medium consisted of 18% (w/v) sucrose, 0.01% (w/v) H 3 BO 4 , 1 mM Ca(NO 3 ) 2 , 1 mM MgSO 4 , 1 mM CaCl 2 and 0.5% (w/v) Agarose Type VII-A (Sigma), pH 7. Mature pollen from fully opened flowers was dusted onto solid germination medium and samples were incubated in a dark humid environment at 22°C for >4 h. Subsequently, germinated pollen was analysed using LSCM as described above. Mitotracker Red (Invitrogen) was used for visualisation of mitochondria in growing pollen tubes.
Aniline blue staining of fertilised pistils was performed as previously described (Deeks et al., 2007). Emasculated mature Col-0 pistils were fertilised with Col-0 and net2a/net2b/net2c/net2d mutant pollen grains and excised after either 6 h or 24 h after pollination before staining and imaging.

Results
Dynamic rearrangement of NET2A subcellular localisation patterns during pollen grain development and germination The pollen-specific NETWORKED protein, NET2A, has recently been implicated in mediating AMCSs in growing pollen tubes (Duckney et al., 2017). In this study, we investigated whether NET2A may also play a role in pollen development and germination. NET2A subcellular localisation was investigated in live Arabidopsis pollen grains stably expressing native promoter-driven NET2A-GFP. NET2A-GFP was noted to exhibit distinct localisation patterns at specific stages of microgametogenesis, undergoing dynamic reorganisations throughout the pollen developmental programme. To observe successive changes in NET2A-GFP subcellular localisation over the course of pollen grain development, serial analysis was performed on pollen grains taken from each bud on an individual inflorescence. As previously described (Regan & Moffatt, 1990;Sanders et al., 1998), the sequential position of each bud on the inflorescence represents a unique developmental stage of each bud at that point in time and, therefore, the specific developmental stage of the pollen within. DAPI staining was performed to identify the developmental stage at which the specific NET2A arrays occurred.
During microgametogenesis, NET2A-GFP was first observable at discreet, static punctae within developing bicellular grains ( Fig. 1a-i). These NET2A-GFP punctae did not appear to be specifically associated with the pollen plasma membrane, as described previously in pollen tubes (Deeks et al., 2012;Duckney et al., 2017), but alternatively distributed throughout the cytoplasm. In older flowers with further developed bicellular pollen, NET2A-GFP subsequently localised to a notably distinct pattern, decorating highly organised, transversely orientated cables at the cell cortex, and a finer meshwork of filaments extending throughout the cytoplasm ( Fig. 1a-ii). This NET2A-GFP subcellular localisation pattern was designated as the 'NET2A transverse array'. The transverse array displayed marked polarity within the bicellular grains; with the cortical actin cables consistently aligning transversely to the cell longitudinal axis and the exine aperture (Fig. 1b). In late-bicellular and early-tricellular pollen, NET2A-GFP decorated a highly filamentous network at the cell cortex, which extended throughout the cytoplasm, connecting to a basket structure surrounding the vegetative cell nucleus (VCN) ( Fig. 1a-iii). This NET2A-GFP-decorated network appeared to be static under observation using time-lapse LSCM. We termed this NET2A-GFP localisation pattern the 'NET2A filamentous array'. Subsequently, before dehiscence, NET2A-GFP was observed to localise to a dense network of fine filaments at the cell cortex, interconnected with a sparser filamentous network throughout the cytoplasm. This localisation pattern was named the 'NET2A cortical array' . Finally, in mature dehisced pollen grains NET2A-GFP localisation was fully cytosolic and did not decorate filaments or punctae. Dehiscence occurs over a period of 6 h (Smyth et al., 1990) therefore narrowing the change in NET2A localisation to this timeframe.
Therefore, we demonstrate that NET2A exhibits distinctive localisation patterns at different stages of pollen development. The NET2A-GFP-decorated punctae and filamentous systems were static under observation for periods of 1 min and it is, therefore, likely that the highly distinct NET2A-GFP arrays observed are a result of extensive reorganisations at specific stages of pollen development.

NET2A localises to F-actin during microgametogenesis
NET2A is implicated in the regulation of actin in growing pollen tubes (Duckney et al., 2017) and it is now possible that NET2A may also play a role in the regulation of actin-dependent processes in developing pollen. Therefore, we investigated whether NET2A may localise to F-actin in the developing grains. Co-localisation of NET2A and F-actin was visualised in live pollen grains of stable Arabidopsis transgenic lines co-expressing native promoter-driven NET2A-GFP and the actin marker construct, FABD2-RFP, expressed under the pollen-specific promoter, pLAT52 (Ketelaar et al., 2004;Duckney et al., 2017). Here, NET2A-GFP colocalised with F-actin in each of the NET2A arrays described above. Furthermore, we observed F-actin to undergo a series of developmentally coordinated rearrangements in tandem with NET2A.
Expression of pLAT52 promoter-driven FABD2-RFP was first observable in bicellular pollen grains exhibiting the NET2A transverse array. F-actin localised to highly organised cables at the cell cortex, aligned perpendicular to the cell longitudinal axis, forming an 'actin girdle' structure ( Fig. 2a). NET2A-GFP colocalised with the actin girdle, suggestive that NET2A may decorate the actin cytoskeleton in the transverse array. In latebicellular and early-tricellular pollen grains, striking colocalisation between NET2A-GFP and F-actin was evident in the filamentous array (Fig. 2b). F-actin localised to a cortical filamentous network, interconnected with cytosolic filaments, continuous with an actin basket surrounding the VCN. NET2A-GFP co-localised with F-actin at the VCN nuclear basket, cytosolic filaments and cortical actin networks. Finally, NET2A also localised to the actin cytoskeleton in the NET2A cortical array (Fig. 2c). A dense network of fine actin filaments was present at the cell cortex, continuous with a sparse network of fine actin filaments in the cytoplasm. NET2A-GFP co-localised with F-actin at both the cell cortex and in the cytoplasm.
Therefore, the actin cytoskeleton reorganises successively into highly distinctive, structures at specific stages of Arabidopsis pollen grain development, to which NET2A localises.

Changes in NET2A distribution occur during pollen germination
The actin cytoskeleton has a fundamental role in pollen germination (Gibbon et al., 1999), during which it undergoes highly dynamic rearrangements in structure (Volger et al., 2015). Similarly, NET2A localises to drastically different patterns before and after pollen germination. Whilst NET2A is cytosolic in mature pollen grains before germination (Fig. 3a), it localises to focal AMCSs at the shank plasma membrane of growing pollen tubes (Deeks et al., 2012;Duckney et al., 2017). We therefore investigated how NET2A localisation patterns are redistributed during pollen germination, as the actin cytoskeleton undergoes dynamic changes in structure. Fig. 3a shows the cytosolic localisation of NET2A-GFP in mature pollen before germination. Here, NET2A-GFP did not localise to filamentous networks or punctae. In pollen grains undergoing germination, NET2A-GFP was largely cytosolic, however discreet NET2A-GFP foci had formed at the cell cortex in the vicinity of the bulge site, where the nascent pollen tube germinates (Fig. 3b). In newly germinated pollen tubes, NET2A-GFP localised to punctae at the plasma membrane of the base and shank of the growing tube, similar to that found in fully germinated pollen tubes as previously described ( Fig. 3c; Duckney et al., 2017). Taken together, it appears that NET2A is recruited from the cytosol to punctae at the site of germination and, during pollen tube growth, is deployed at the plasma membrane of the newly established shank region as the maturing membrane becomes increasingly distal from the growing tip (Fig. 3d). NET2A punctae are sites of actin-membrane interactions in growing pollen tubes (Duckney et al., 2017). Further to this, these NET2A punctae are also present at the pollen grain germination site before and after germination. Therefore, we propose that NET2A may also mediate actinmembrane interactions at discreet foci at the site of pollen tube outgrowth during germination. We have shown that NET2A localises to F-actin in developing pollen and establishes actin-membrane contacts sites at the PM of germinating and growing tubes (Duckney et al., 2017). We therefore hypothesised a function for NET2 proteins in the regulation of actin during pollen development and tube growth. To investigate the role of NET2 in pollen, we generated loss-offunction mutants of each member of the NET2 protein family. We obtained T-DNA insertion mutants for NET2A, NET2B and NET2D, (Fig. 4a,b) and used CRISPR/Cas9 technology to generate loss-of-function mutant alleles of NET2C (Fig. 4c,d). Double, triple and quadruple mutants of each combination of net2 mutant alleles were generated and analysed for defects in pollen development and tube growth (Fig. S1). We observed phenotypic defects in specific triple mutant combinations and defective pollen tube morphology was evident in net2a/net2b/net2c and net2a/net2c/net2d mutants, which exhibited a 'zig-zagging' profile ( Fig. 4e). We also noted that these mutant pollen tubes exhibited periodic depolarisation along the length of the tube, consistent with tip growing cells of mutants of other actin regulatory proteins such as CAP1 (Deeks et al., 2007). net2a/net2b/ net2c/net2d quadruple mutants appeared to exhibit no increased severity in pollen tube growth defects (Fig. S1). No phenotype was observed in single or double net2 mutant combinations. The data demonstrate important roles for the NET2 protein subfamily in pollen. (c) NET2A localises to punctae at the shank plasma membrane of newly germinated pollen tubes, similar to as described in growing pollen tubes (Deeks et al., 2012;Duckney et al., 2017). NET2A-GFP punctae in the pollen tube shank (indicated as NET2A) are distinguishable from exine autofluorescence (labelled as 'exine'). Bars, 10 µm, with an exception for (b ii), for which bar, 5 µm. Pollen tube growth is restricted to the cell tip, indicating that defects in pollen tube morphology are likely to arise from defective tip growth. We therefore used time-lapse microscopy to study the progression of tip growth in net2a/net2b/net2c mutant pollen over time. Whereas Col-0 pollen tubes largely extend in a uniform direction, the pollen tubes of net2a/net2b/net2c mutants appear unable to maintain a constant growth trajectory (Video S1), resulting in the observed morphological defects. The percentage of net2a/net2b/net2c pollen tubes exhibiting this morphological defect was observed to be 92.6%, compared with 1.0% in Col-0 pollen (Fig. 4f). We also noted a slight, but statistically significant, decrease in the growth rate of net2a/net2b/net2c mutant pollen tubes compared with that of Col-0 pollen tubes ( Fig. 4g; P < 0.05). Interestingly, a more pronounced reduction in pollen tube growth rate was observed during pollen tube growth through the stigma in semi-in vivo growth assays and in vivo during fertilisation (Figs S2, S3). We could observe no reduction in pollen tube germination in net2a/net2b/net2c mutant pollen (Fig. 4h), leading us to speculate that the net2 loss-offunction phenotype had not become apparent until after pollen germination. We concluded that the NET2 protein subfamily has important roles in the regulation of pollen tube growth.

New Phytologist
The cortical actin cytoskeleton is disorganised in net2 lossof-function mutants We have demonstrated previously that NET2 proteins bind cortical longitudinal actin cables to the plasma membrane in growing pollen tubes (Duckney et al., 2017), suggesting an important role in actin-driven pollen tube growth. To investigate a role for NET2 proteins in the regulation of actin in growing pollen tubes, we investigated the actin cytoskeleton of net2a/net2b/net2c/net2d mutant pollen to identify potential defects in the pollen tube cytoskeleton.
To visualise the actin cytoskeleton in growing pollen tubes, we expressed the actin marker, Lifeact (Smertenko et al., 2010) linked to the 5 0 -end of the fluorescent protein, mNeonGreen, under the pLAT52 promoter. The actin cytoskeleton of the net2 quadruple mutant pollen tubes was observed to be disrupted, (however no defects in actin organisation were observed in developing net2a/net2b/net2c/net2d mutant pollen grains expressing Lifeact-mNeonGreen; Fig. S4). Using CLSM we revealed that Col-0 pollen tubes displayed thick, longitudinally aligned actin cables extending over much of the length of the growing pollen tube. By contrast, net2 quadruple mutant pollen tubes exhibited a sparse array of short and fine actin filaments (Fig. 5a). To accurately compare the structure of the actin cytoskeleton of Col-0 and net2 quadruple mutant pollen tubes, we used rapid DIC imaging to capture the 3D structures of the pollen tube cytoskeleton in high temporal resolution (Fig. 5b). The data demonstrate that the actin cytoskeleton was disorganised in the net2 quadruple mutant tubes, appearing fragmented and frequently misaligned to the longitudinal axis of the growing pollen tube. Furthermore, the prominent longitudinal actin cables of the Col-0 pollen tubes were markedly absent from the net2 quadruple mutants.
As NET2 proteins are implicated in the regulation of AMCSs at the pollen tube plasma membrane, we next investigated potential roles for NET2 in the organisation of F-actin at the cell cortex. Analysis of the actin cytoskeleton at the cell cortex of net2 quadruple mutant pollen tubes revealed a decrease in cortical Factin compared with Col-0 pollen tubes (Fig. 5c-i). We recorded a density of 0.97 actin filaments lm À1 at the cell cortex of Col-0 pollen tubes, whereas the density of cortical F-actin in net2 quadruple mutants was significantly lower, at 0.57 actin filaments lm À1 . Furthermore, net2 quadruple mutants also exhibited a reduced proportion of cortical actin filaments (within 1 lm of the cell cortex) relative to core actin filaments (>1 lm of the cell cortex). Orthogonal cross-sections of 3D confocal images of growing pollen tubes expressing Lifeact-mNeonGreen revealed the positions of longitudinal actin cables in the cell (Fig. 5d-i,dii). We observed a cortical F-actin ratio of 2.1 in Col-0 pollen tubes and a significantly reduced cortical F-actin ratio of 0.6 in net2 quadruple mutant pollen tubes. Taken together, our data indicate that net2 quadruple mutants exhibited a reduction in cortical F-actin at the pollen tube plasma membrane. This observation was consistent with the hypothesis that NET2 proteins may serve as linkers or anchors of F-actin at the plasma membrane and demonstrated an important function for the NET2 protein subfamily in the regulation of the pollen tube actin cytoskeleton.

Polar transport and positioning of organelles is impaired in net2 mutant pollen tubes
The actin cytoskeleton regulates cytoplasmic streaming, organelle positioning and tip-focussed transport of secretory vesicles (SVs) in growing pollen tubes (Vidali & Helper, 2000). To better understand how NET2 proteins may regulate actin-driven pollen tube growth, we proceeded to investigate whether NET2 proteins have important functions in regulating tip-focussed vesicle transport and cytoplasmic streaming.
To examine tip-focussed vesicle transport in growing pollen tubes, we analysed the subcellular localisation of SVs in pollen tubes stably expressing the secretory vesicle marker GFP-RabA4B (Zhang et al., 2010) under the pLAT52 promoter. The localisation pattern of GFP-RabA4B in net2a/net2b/net2c/net2d mutant pollen tubes was observed to be markedly different from that of Col-0 ( Fig. 6a-i). Whilst Col-0 pollen tubes exhibited a classical inverted cone-shaped localisation pattern at the pollen tube tip (Zhang et al., 2010), the net2 quadruple mutant pollen tubes displayed a massive tip-focussed accumulation of GFP-RabA4B and loss of the ordered inverted cone distribution pattern. Col-0 pollen tubes were noted to possess a 2.7-fold ratio of GFP-RabA4B-labelled SVs at the pollen tube tip compared with the distal shank region of the cell. In comparison, the tip : distal ratio of SVs at the tip of mutant pollen tubes was 6.9, constituting a statistically significant increase (P < 0.01). The data indicate a role for NET2 proteins in the regulation of tip-focussed vesicle transport.
A role for NET2 proteins in the regulation of cytoplasmic streaming was also investigated. Visualisation of cytoplasmic streaming was performed by monitoring the transport of mitochondria in pollen tubes labelled by Mitotracker TM Red. We observed defects in transport and positioning of mitochondria in net2a/net2b/net2c/net2d mutant pollen tubes ( Fig. 6b; Video S2). Whilst Col-0 pollen tubes exhibit rapid anterograde and retrograde transport of mitochondria, the longitudinal movement of mitochondria in net2 quadruple mutant pollen tubes was perturbed, with reduced mitochondrial velocity and frequent short circular trajectories clearly observable. Furthermore, in net2 quadruple mutants, abnormal accumulations of mitochondria in the pollen tube subapex was also evident, but were not observed in Col-0 tubes. It can therefore be concluded that NET2 proteins are important in the organisation of cytoplasmic streaming and organelle positioning.
Taken together, our data indicate an association of NET2 proteins with F-actin during pollen tube development and pollen tube growth, leading us to speculate a role for NET2 in the regulation of the pollen cytoskeleton. We demonstrated that NET2 proteins are necessary for normal pollen tube growth, organisation of the pollen tube cytoskeleton and actin-dependent subcellular processes such as polar vesicle trafficking and cytoplasmic streaming.

Discussion
The actin cytoskeleton has an essential role in pollen development, germination and pollen tube growth (Gervais et al., 1994;Gibbon et al., 1999;Zonia et al., 1999;Vidali et al., 2001). Our data suggest an important role for NET2 proteins in angiosperm fertilisation and the regulation of the pollen cytoskeleton during gametogenesis, pollen germination and tube growth. The observation of dynamic NET2A localisation to the actin cytoskeleton during pollen development and tube growth suggests its involvement in the regulation of actin-driven processes occurring during the Arabidopsis male reproductive cycle, whilst phenotypic defects observed in net2 mutant pollen confirm the crucial role for NET2 in the normal regulation of actin-driven pollen tube growth.
The actin cytoskeleton has not yet been characterised during the course of pollen grain development and our data provide a novel illustration of the dynamic structure of the actin cytoskeleton in live developing Arabidopsis pollen grains, in previously unobserved detail. Our data demonstrate that NET2A-GFP reorganises to unique structures at specific stages of microgametogenesis; localising to the actin cytoskeleton over the course of pollen grain development. We first observed NET2A-GFP in bicellular pollen as punctae distributed throughout the cytoplasm, differing from previous observations of plasma membrane-localised NET2A in pollen tubes (Duckney et al., 2017). This is possibly because in growing pollen tubes, PRK4 and PRK5 recruit NET2A to the plasma membrane (Duckney et al., 2017), but are not abundantly expressed in early pollen development (Winter  et al., 2007). Later in bicellular pollen development, NET2A was observed to localise to F-actin in a so-called 'transverse array', in which a polarised, transversely aligned belt of actin cables was observed at the cell cortex. Following this, in late-bicellular and tricellular pollen NET2A-GFP localised to a filamentous actin array, connected to the nuclear envelope of the VCN, with filamentous networks of actin extending to a network of cortical Factin filaments. In mature tricellular grains, NET2A-GFP decorated a cortical array of actin filaments, consisting of a dense meshwork of fine filaments at the cell cortex. Our data drew significant similarities to, and extends, previous studies conducted on the actin cytoskeleton of fixed in vitro-cultured pollen grains from N. tabacum and B. napus. It is therefore, possible to infer the potential functions of each actin structure and consequently those with which NET2A may be involved. Similar actin structures have been hypothesised to regulate pollen cell elongation, nuclear positioning and restructuring of the plasma membrane (Gervais et al., 1994;Zonia et al., 1999;Kang et al., 2003;Yamamoto et al., 2003) and it is possible that NET2 proteins may contribute to the robustness of pollen grain developmental processes through an association with F-actin. Notably, NET2A-GFP was observed to bind to actin at the cell cortex of developing pollen grains, where it may serve to establish AMCSs, as reported in growing pollen tubes (Duckney et al., 2017) and was also observed to localise to F-actin at the VCN. The actin cytoskeleton is known to be physically tethered to the outer nuclear envelope in Arabidopsis (Tamura et al., 2013;Zhou et al., 2014). As such, it is possible that NET2 proteins may mediate connections between the actin cytoskeleton and nuclear envelope in Arabidopsis pollen grains.
During germination, NET2A was observed to be recruited from the cytosol to membrane-localised foci, specifically at the site of pollen tube outgrowth. Following pollen germination, these highly localised NET2A punctae at the VC cortex are continuous with the NET2A punctae of previously characterised AMCSs at the pollen tube shank (Deeks et al., 2012;Duckney et al., 2017). This led us to propose that the NET2A punctae observed in germinating pollen were also AMCSs. As the growing tube extends, cytosolic NET2A is likely to be recruited to membrane foci to establish AMCSs as the membrane becomes increasingly distal to the growing tip and gains shank identity. Actin has a principle role in pollen germination (Gibbon et al., 1999), during which F-actin cables reorganise to align longitudinally to the germination site, protruding into the nascent tube and later forming the longitudinal actin cables of the pollen tube shank (Volger et al., 2015). It is tempting to speculate that NET2A may bind actin to the plasma membrane at the site of germination, perhaps to support targeted pollen tube outgrowth and later organise longitudinal actin cables at the cell cortex of the shank as the pollen tube extends.
Our investigation of net2 loss-of-function mutants clearly confirmed an important role for NET2 proteins in the male gametophyte and we have noted that each member of the NET2 protein family subclade functions in the regulation of pollen tube growth. Interestingly, NET2 did not appear to be necessary for pollen grain development as we could not identify any major defect in actin organisation in the net2 quadruple mutant, perhaps due to functional redundancy with other NETs known to be expressed in pollen grains, such as NET3 . The 'zig-zag' morphology phenotype of mutant pollen tubes was determined to result from a loss of ability of mutant tubes to maintain a constant growth trajectory. Pollen tube growth trajectory is maintained by homeostatic signalling cascades, modulated in response to extracellular signals (Guan et al., 2013;Takeuchi & Higashiyama, 2016;Schleible & McCubbin, 2019). We have recently demonstrated that NET2 interacts with PRKs, known regulators of actin-driven pollen tube growth and targeting in response to extracellular signals (Takeuchi and Higashiyama, 2016). From our data, it could be speculated that NET2 may play a role in signalling to the actin cytoskeleton downstream of PRKs to maintain normal pollen tube growth trajectory. In support of this hypothesis the phenotypic growth defects of net2 pollen appeared more pronounced during growth through the stigma, indicating a potential role in extracellular signal transduction to the cytoskeleton in the regulation of actin-driven growth. The observed 'zig-zag' phenotype is a curious one and, to our knowledge, mutations of no other cytoskeletal-regulatory proteins result in a similar pollen tube defect. Our data therefore demonstrate a unique and crucial role for NET2 proteins in normal pollen tube growth, perhaps in signal transduction during fertilisation.
The defective pollen tube growth phenotype can be attributed to loss of AMCSs at the pollen tube plasma membrane. Consistent with the hypothesised role for NET2 in anchoring actin to the plasma membrane (Duckney et al., 2017), we observed a reduction of cortical F-actin in net2 quadruple mutant pollen tubes that is likely to representing unpinning of actin cables from the membrane. The actin cytoskeleton was disorganised, misaligned and fragmented in mutant pollen tubes, indicating a role for NET2 at AMCSs in structural organisation of the pollen tube cytoskeleton, which may serve as points of stabilisation and structural orientation of longitudinal actin cables. The importance of NET2 in the regulation of the pollen tube cytoskeleton was further evidenced in defects in actin-driven subcellular processes including cytoplasmic streaming and vesicle transport. Longitudinal actin cables are important in rapid, long-distance transport of organelles (Vidali et al., 2001), so it is therefore unsurprising that mitochondrial movement was impaired and with loss of defined trajectory in mutant pollen. Taken together our data indicate that net2 loss of function indirectly affects mitochondrial movement as a result of impairment of the actin cytoskeleton. Disruption to cytoplasmic organisation was also evident in mutant tubes, with abnormal accumulations of mitochondria in the pollen tube subapex and aberrant distribution patterns of SVs at the tip. We propose that impairment of actin-driven streaming in mutant pollen tubes prevents active mixing of organelles leading to a 'phase separation' small SVs at the extreme apex and shank-restricted large organelles. Pollen tube growth occurs only at the tip through targeted apical secretion. The abnormal transport of SVs in mutant pollen tubes is likely to have been causative of the defective pollen tube growth here observed.
Taken together, we demonstrate dynamic localisation of NET2 to the actin cytoskeleton throughout each stage of pollen development, pollen germination and pollen tube growth, indicative of an involvement of NET2 in the organisation of actin in the male gametophyte. In particular, we demonstrate an essential role for NET2 proteins in normal pollen tube growth, with functions in the organisation of F-actin and organelle transport.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.    Table S1 Primers used in this study.
Video S1 net2 mutant pollen tubes are unable to maintain a constant growth trajectory.