The BEACH Domain-Containing Protein SPIRRIG Modulates Actin-Dependent Root Hair Development in Coordination with the WAVE/SCAR and ARP2/3 Complexes

Root hairs are single cell protrusions that enable roots to optimize nutrient and water acquisition. They attain their tubular shapes by confining growth to the cell apex, a process called tip growth. The actin cytoskeleton and endomembrane system are essential for tip growth; however, little is known about how these cellular components coordinate their activities during this process. Here, we show that SPIRRIG (SPI), a BEACH domain-containing protein involved in membrane trafficking, and BRK1 and SCAR2, subunits of the WAVE/SCAR (W/SCR) and actin related protein (ARP)2/3 activation complexes, display polarized localizations to root hairs at distinct developmental stages. SPI accumulates at the root hair apex via post-Golgi vesicles and positively regulates tip growth by maintaining tip-focused vesicle secretion and filamentous-actin integrity. BRK1 and SCAR2 on the other hand, mark the root hair initiation domain to specify the position of root hair emergence. Live cell microscopy revealed that BRK1 depletion coincided with SPI accumulation as root hairs transitioned from initiation to tip growth. Furthermore, double mutant studies showed that SPI genetically interacts with BRK1 and ARP2/3. Taken together, our work uncovers a role for SPI in facilitating actin-dependent root hair development through pathways that intersect with the W/SCR and ARP2/3 complexes.


47
Root hairs are single cell tubular projections that emerge from root epidermal 48 cells. They increase the effective surface area of the root system by extending laterally 49 into soil pores, thus enabling the increased access to nutrients and water (Carminati et 50 al., 2017, Ruiz et al., 2020. Root hairs have been studied extensively by plant biologists 51 for decades because they serve as excellent models to unravel mechanisms by which 52 cell size and shape in plants is regulated (Grierson et al., 2014). To attain their 53 cylindrical shapes, root hairs undergo tip growth, a process in which expansion of the 54 cell is confined to its apical domain. Tip growth involves a balance between the directed 55 delivery of post-Golgi vesicles carrying protein complexes and cell wall building blocks 56 to the cell apex, and localized cell wall loosening and recycling of excess membranes.

57
Besides root hairs, tip growth is exhibited by other cell types such as pollen tubes, 58 fungal hyphae, and rhizoids of mosses, liverworts and algae (Bascom et al., 2018a). 59 Root epidermal cells called trichoblasts are the cell types that form root hairs. 60 Work in Arabidopsis thaliana has shown that trichoblasts are specified to become root 61 hair-forming cells early during root development through the patterned assembly of 62 protein complexes of transcriptional activators and repressors in different cell files 63 (Schiefelbein et al., 2014, Shibata et al., 2018. Upon establishing their identity, 64 trichoblasts undergo two developmental stages that lead to root hair outgrowth. The first 65 stage is the establishment of a root hair initiation domain (RHID) that eventually leads to 66 a conspicuous root hair bulge at the basal (root-tip oriented) end of the trichoblast and none of the subunits have been conclusively shown to exhibit clear polar 140 localization in these cell types. Therefore, the extent by which the W/SRC-ARP2/3 141 complex functions in tip growing cells of higher plants remains to be determined. 142 SPIRRIG (SPI) is one of eight genes that have been considered to be a member 143 of the DIS group, but compared to other DIS mutants, trichome phenotypes of spi are 144 less severe and the mutant does not display early stage cell swelling that is diagnostic 145 of the DIS group (Schwab et al., 2003). SPI was shown to encode a 3571 long amino 146 acid protein with N-terminally located armadillo (ARM) and concanavalin A (ConA)-like 147 lectin domains and C-terminally-located pleckstrin homology (PH), beige and Chediak 148 Higashi (BEACH) and WD40 repeat domains (Saedler et al., 2009). BEACH domain-149 containing proteins are highly conserved in eukaryotes and are known to function in 150 membrane dynamics, vesicle transport, apoptosis and receptor signaling. This family of 151 proteins are of clinical importance as they have been implicated in a variety of human 152 disorders such as cancer, autoimmunity syndrome and autism (Cullinane et al., 2013). 153 In addition to mild trichome defects, Arabidopsis SPI mutants have short root 154 hairs characterized by fragmented vacuoles suggesting that SPI, like other eukaryotic 155 BEACH domain-containing proteins, functions in membrane trafficking (Saedler et al.,156 2009). Additionally, an observation by Steffens et al. (2017) that showed SPI physically 157 interacting with proteins involved in endosomal sorting reinforces its role in membrane 158 remodeling. SPI was also demonstrated to associate with mRNA processing bodies (P-159 bodies) suggesting a novel role for SPI in post-transcriptional regulation (Steffens et al., 160 2015). Because SPI is one of the DIS genes, it was suggested that SPI might be 161 involved in actin-mediated cell developmental processes (Saedler et al., 2009), and 162 perhaps function in coordination with W/SCR and ARP2/3 complexes. However, SPI is 163 not a known W/SCR or ARP2/3 subunit; therefore, its relationship to ARP2/3 function is 164 unclear. Moreover, the uncertainty with regard to SPI function is confounded by the fact 165 that its subcellular localization in root hairs, which exhibit the most profound phenotype 166 when SPI is mutated, remains unknown. In this paper, we have addressed these 167 questions by showing that a functional SPI-fluorescent protein fusion accumulated at 168 the tips of rapidly growing root hairs. A SPI-tagged fluorescent protein fusion was not 169 detected in the RHID and in tips of non-growing mature root hairs. We further show that 170 the W/SCR complex subunits, BRK1 and SCAR2, localized to the RHID and exhibited 171 temporal dynamics opposite to that of SPI (i.e. functional fluorescent protein-tagged 172 BRK1 and SCAR2 declined as rapid root hair tip growth commenced). Double mutant 173 studies revealed that SPI and W/SCR-ARP2/3 genetically interact suggesting that one 174 function of SPI is to modulate actin-dependent root hair development, in part through 175 pathways that overlap with W/SCR and ARP2/3 complexes. We previously described a forward genetic screen that led to the isolation of three non-184 allelic recessive Arabidopsis mutants that were hypersensitive to the growth inhibitory 185 effects of LatB. The hypersensitive to LatB1 (hlb1) and hlb3 mutants have been 186 described previously (Sparks et al., 2016, Sun et al., 2019. hlb1 was disrupted in a 187 gene encoding a trans-Golgi Network-localized tetratricopeptide repeat protein involved 188 in actin-mediated membrane recycling (Sparks et al., 2016), whereas the genetic lesion 189 in hlb3 was found to encode the class II actin nucleator formin (Sun et al., 2019). Here,190 we report on hlb2, the third of these recessive mutants. Like hlb1 and hlb3, hlb2 primary 191 root growth was more severely inhibited by LatB when compared to wild type. In the 192 absence of LatB or at low (i.e. 25 nM) LatB concentrations, the primary root length of 193 hlb2 was similar to wild type. Differences in root length between wild type and hlb2 194 became apparent when seedlings were grown on a concentration of 50 nM LatB and 195 higher (Supplemental Figure 1A and B).

196
Through map-based cloning, we found that the mutation in hlb2 was confined to that hlb2 had root hairs that were about 80% shorter than wild type with some root hairs 205 forming only small bulges (Supplemental Figure 2B and C). Moreover, hlb2 had mild 206 trichome defects that were reminiscent of the phenotypes of previously isolated spi 207 mutant alleles (Supplemental Figure 2D; Saedler et al., 2009). To further verify if HLB2 208 is SPI, we obtained a mutant from the publicly available SALK collection 209 (SALK_065311), which had a T-DNA insertion in the 10 th exon of the SPI gene 210 (Supplemental Figure 2A; Alonso et al. (2003)). The SALK_065311 line we obtained is 211 the same as spi-3 mutant allele reported previously in Steffens et al. (2015). In addition 212 to having similar root hair and trichome defects as hlb2, primary roots of spi-3 were 213 hypersensitive to LatB (Supplemental Figure 2E), and a cross between hlb2 and 214 SALK_065311 failed to complement each other in the F1 hybrid. Taken together, these 215 results indicate that hlb2 is a new spi mutant allele. Based on earlier nomenclature 216 (Steffens et al., 2015), we renamed hlb2 to spi-5 (Supplemental Figure 2A) (Steffens et al., 2015, Steffens et al., 2017. However, these 223 constructs have not been shown to complement the defective root hair and trichome 224 phenotypes of spi. Because of the large size of SPI, we were unable to generate native  Brumos et al., 2020, Zhou et al., 2011 Once spi was transformed with a recombineered SPI-YPet construct, primary 233 root hypersensitivity to LatB, short root hairs and defective trichome phenotypes were 234 rescued, indicating that the construct was functional ( Figure 1A; Supplemental Figure   235 3). Transgenic complementation of spi with SPI-YPet provided additional evidence that 236 HLB2 is SPI. Confocal microscopy revealed that SPI-YPet signal was strongest in the showed that the intensity of SPI-YPet fluorescence in root hair tips is significantly 243 positively correlated to rapid root hair growth ( Figure 1D). These results indicate that the 244 SPI protein has functions related to root hair elongation and consistent with the short 245 root hair phenotypes of spi. The prominent SPI-YPet signal at the tips of elongating root hairs is reminiscent of the 250 localization patterns of post-Golgi markers such as RAB small GTPases, which are 251 known to function in tip-directed secretion (Preuss et al., 2004). We therefore 252 hypothesized that SPI is trafficked to the tips of root hairs via post-Golgi vesicles. To 253 test this hypothesis, seedlings expressing SPI-YPet were treated with brefeldin A (BFA).

254
BFA is a fungal toxin that is routinely used to investigate endomembrane dynamics 255 because it prevents vesicle formation for exocytosis by inhibiting ADP ribosylation factor 256 guanine nucleotide exchange factors (ARF-GEFs), while still enabling endocytosis and 257 some retrograde pathways to continue (Baluška, 2002, Doyle et al., 2015. In a study of other spi mutant alleles, the similarities in phenotypes between spi and 279 w/scr-arp2/3 suggests that SPI could function in actin-dependent cellular processes 280 (Saedler et al., 2009). However, because trichome defects of spi were mild compared to 281 other w/scr-arp2/3 mutants, no obvious actin phenotypes were observed in spi 282 trichomes (Schwab et al., 2003). To clarify the relationship between SPI and actin, we 283 focused on investigating actin organization in root hairs since they displayed the most 284 obvious growth defects in SPI-altered plants. 285 To study actin organization, we expressed the live F-actin reporter, UBQ10: 286 mGFP-Lifeact, in spi (Vidali et al., 2010). This particular F-actin reporter was selected 287 because it prominently labels the tip-focused F-actin meshwork typically observed in 288 root hairs that are rapidly growing (Sparks et al., 2016). In wild type, fine F-actin 289 networks were observed in the root hair bulge that had a weaker signal compared to the 290 thicker actin bundles in other regions of the trichoblast ( Figure 3A). As the root hair 291 bulge expanded and the root hair transitioned to rapid tip growth, the tip-focused F-actin 292 meshwork, which consisted of short filaments and dynamic puncta, became more  Figure 3F).

296
F-actin organization in the tips of spi root hairs was different from that of wild 297 type. As noted, some root hairs of spi were only able to form small bulges due to 298 premature termination of tip-growth (Supplemental Figure 2). In spi root hairs, the 299 distinct F-actin meshwork observed in wild-type root hairs was unable to form. Instead, 300 F-actin in these spi root hair bulges contained thick F-actin cables that resembled those 301 of wild-type root hairs that had terminated growth ( Figure 3G and H). However, the thick 302 F-actin cables in non-growing spiroot hair bulges were unstable as they dissipated 303 ( Figure 3I) and reformed again at a later time ( Figure 3J). In the small population of spi 304 root hairs that were able to undergo tip growth, a few exhibited the tip-focused F-actin 305 meshwork that resembled those observed in elongating wild-type root hairs ( Figure 3K).

306
However, most of these slow growing spi root hairs lacked the tip-focused F-actin 307 meshwork ( Figure 3L and M; Supplemental Movie 4) or had thick F-actin bundles 308 protruding to the tip, a feature that was reminiscent of non-growing, mature wild-type 309 root hairs ( Figure 3N). The disruption of the tip-focused F-actin meshwork in spi was 3O). Our analysis showed that the fluorescence ratio in spi root hairs was significantly 314 reduced compared to wild-type root hairs, supporting visual observations that the tip-315 focused F-actin meshwork in spi root hairs is disrupted ( Figure 3P). 316 We next generated plants expressing both SPI-YPet and mRuby-Lifeact so we 317 could simultaneously observe SPI and F-actin in growing root hairs. In elongating root 318 hairs of dual labeled seedlings, SPI-YPet and the mRuby-labeled F-actin meshwork 319 overlapped at the tip ( Figure 3Q; Supplemental Movie 5). This provides support that root 320 hair tip-localized SPI is associated with the tip-focused F-actin meshwork, and as such 321 is involved in sustaining normal root hair elongation in coordination with actin.  Physcomitrella protonemal cells (Perroud and Quatrano, 2006, Perroud and Quatrano, 331 2008) raises the possibility that W/SRC-ARP2/3 maybe a root hair tip-localized 332 complex. In an attempt to link SPI with the W/SCR-ARP2/3 complex, we imaged roots showed that BRK1-YFP fluorescence is inversely proportional to root hair growth rate 345 ( Figure 4C).

346
Given that BRK1 stabilizes the entire family of SCAR proteins and is required for 347 functional W/SRC assembly (Le et al., 2006), we investigated if the SCAR protein 348 localized to the RHID, similarly to BRK1. To address this question, we imaged a 349 recombineered SCAR2-mCherry fusion expressed in the scar1 scar2 scar3 scar4 350 (scar1234) quadruple mutant. We found that like BRK1-YFP, SCAR2-mCherry marked 351 the RHID and dissipated when rapid root hair tip growth commenced ( Figure 4D).

352
The accumulation of BRK1-YFP and SCAR2-mCherry at the RHID led us to 353 hypothesize that brk1 and scar1234 might have defects in root hair initiation. One 354 parameter that has been studied extensively as an indicator of root hair initiation defects 355 is planar polarity, which is a measure of root hair position along the length of the 356 trichoblast (Nakamura and Grebe, 2018). We found that root hair position of brk1 and  Given that ARP2/3 is the known target of activated W/SRC, we hypothesized that 364 ARP2/3 might also localize to the RHID. To test this hypothesis, we imaged a line 365 expressing ARPC5-GFP, one of the subunits of the ARP2/3 complex (Yanagisawa et 366 al., 2015). This line has been confirmed to complement the arpc5 trichome defects. As 367 reported previously, ARPC5-GFP labeled the plasma membrane of initiating trichome 368 branches and the apex of elongating trichome branches (Yanagisawa et al. (2015); 369 Supplemental Figure 4A and B). Nevertheless, when observed in root hairs, the 370 ARPC5-GFP signal was distributed uniformly at various stages of development with no 371 distinct accumulation at the RHID or root hair apex (Supplemental Figure 4C and D).

372
Although there was no clear polarized localization of ARPC5-GFP in root hairs, it is 373 possible that ARP2/3 is locally activated in RHID or root hair apex, but that active pool is 374 obscured by a large soluble pool of the complex in this cell type.  Figure 5D). In brk1, SPI-YPet signal was weak during early root hair 394 bulge formation and intensified as root hair tip growth accelerated ( Figure 5D). Taken 395 together, these results indicate that SPI is antagonistic to BRK1 during the transition 396 from root hair initiation to rapid tip growth.  However, it remains to be determined whether SPI genetically interacts with W/SRC-405 ARP2/3 pathway components. To address this question, we generated double spi brk1 406 mutants and compared their root hair phenotypes with single mutants. We first 407 investigated the genetic relationship between SPI and BRK1 based on length of mature 408 root hairs. We found that at low magnification, root hair length of brk1 appeared 409 morphologically similar to wild type ( Figure 6A). Quantification of root hair length, 410 however, uncovered a small but significant reduction in root hair length in brk1 when 411 compared to wild type ( Figure 6B). On the other hand, spi brk1 had short root hairs that 412 were similar to spi ( Figure 6A and B). We next examined whether SPI and BRK1 413 genetically interact in specifying root hair planar polarity. We found that spi displayed a 414 more shoot ward (apical)-shifted root hair position when compared to wild type and 415 brk1. Quantification of root hair planar polarity showed that that root hair position of spi 416 was identical to spi brk1 ( Figure 6C). Taken together, our results indicate that SPI is 417 epistatic to BRK1 for root hair planar polarity and tip growth.

418
Epistasis between SPI and BRK1 prompted us to expand our genetic interaction 419 studies to other genes in the ARP2/3 pathway. We therefore generated spi arp2, spi interaction between SPI and ARP2/3 with regard to root hair tip growth. 438 We next investigated if SPI and ARP2/3 genetically interact in specifying the 439 position of the RHID. We found a mild but statistically significant shoot ward shift in the 440 position of the RHID in arp2, arp3 and arpc5 when compared to wild type. The extent to 441 which the position of the RHID shifted towards the shoot in single arp2/3 complex 442 mutants was less than that of spi and brk1 (compare Figure 6C and 7D). On the other 443 hand, the position of the RHID in spi arp2/3 was more similar to single arp2/3 indicating 444 that ARP2/3 is epistatic to SPI for root hair planar polarity ( Figure 7C and D). In many 445 cases we observed that the aborted root hair bulges in spi arp2/3 emerged very close to 446 the basal (root ward) end wall of the trichoblast ( Figure 7C). The synergistic and 447 epistatic relationship between SPI and ARP2/3, for root hair length and planar polarity, 448 respectively, indicate that SPI also genetically interacts with ARP2/3 during root hair 449 development. However, the nature of this interaction appears to differ from that of 450 BRK1.

451
Genetic interaction between SPI and W/SRC-ARP2/3 was also analyzed for 452 other spi phenotypes such as primary root hypersensitivity to LatB (Supplemental 453 Figure 1). For LatB sensitivity, we found that arpc5 primary roots were inhibited to the 454 same extent as wild type when grown on LatB (Supplemental Figure 6B). Although 455 primary roots of arp2, arp3 and brk1 exhibited some hypersensitivity to LatB, the 456 response of these mutants to the compound was not as severe as spi and leaned more 457 toward wild-type sensitivity. By contrast, the hypersensitivity of primary roots of spi arp2, 458 spi arp3, spi arpc5 and spi brk1 to LatB was more similar to spi (Supplemental Figure   459 6A and B) suggesting genetic interactions also occurs between SPI and W/SCR-460 ARP2/3 for this phenotype.

461
Genetic interaction between SPI and W/SRC-ARP2/3 was also observed in 462 trichomes (Li et al., 2003, Schwab et al., 2003. Consistent with 463 previous reports, arp2, arp3, arpc5 and brk1 had severe trichome shape defects when 464 compared to wild type, while those of spi were mild when viewed with a scanning 465 electron microscope. One particular trichome defect is branch length, in which arp2, 466 arp3, arpc5 and brk1 trichomes have distinctively short branches when compared to 467 wild type (Supplemental Figure 7). On the other hand, trichomes of spi were wavy and 468 crooked, but not as short as those of arp2, arp3, arpc5 and brk1. We found that 469 trichomes of all the double mutants resembled those of spi (Supplemental Figure 7A). Our work uncovers new insights underlying actin-mediated root hair development. A 484 major result from our studies is the revelation that SPI is a root hair tip-localized protein.

485
Although SPI fused to fluorescent proteins was reported to localize to endosomes and 486 P-bodies, such studies have been limited to transient expression assays in biolistically-487 bombarded leaves (Steffens et al., 2015;Steffens et al., 2017). As such, a mechanistic  (Perroud and Quatrano, 2008). Like BRK1-YFP, an ARPC4-GFP construct labeled 542 the tips of caulonemal cells (Perroud and Quatrano, 2008) whereas no clear tip-focused 543 labeling of ARPC5-GFP was observed in root hairs. The observation that BRK1-YFP 544 prominently localized to the RHID, but dissipated during active root hair tip growth was 545 surprising given that the Arabidopsis BRK1 complemented the defective filamentous 546 growth of Physcomitrella BRK1 knockouts (Perroud and Quatrano, 2008). This suggest  However, the epistatic relationships between SPI and W/SCR-ARP2/3 for trichome 560 morphology and primary root hypersensitivity to LatB as reported here continue to point 561 to functional relationships between these genes in diffusely-growing cells. The 562 observation that phenotypes of spi w/scr-arp2/3 tilt more toward spi indicate that SPI 563 might negatively regulate an actin pathway that is parallel to ARP2/3.

564
The weakening of BRK1-mRuby3 fluorescence coinciding with SPI-YPet 565 accumulation, and the persistence of SPI-YPet signal in spi root hair tips, provides 566 indirect evidence that SPI might play a role in mediating BRK1 stability or localized 567 clustering at the plasma membrane of the RHID. Because spi brk1 root hair phenotypes 568 resemble spi, ectopic expression of BRK1 does not appear to contribute to the spi 569 phenotypes. Although, it is unknown why BRK1-YFP signal persists in spi, it is tempting 570 to speculate that SPI might modulate BRK1 via protein degradation pathways. This 571 possibility is supported by studies in mammals pointing to a role for BEACH domain-572 containing proteins in protein degradation via the ubiquitination pathway. In mouse 573 models for example, the BEACH domain-containing protein, WDR81, was shown to be 574 essential for removal of autophagy-dependent ubiquitinated proteins (Liu et al., 2017).

575
In this regard, it is worth noting that the W/SCR-ARP2/3 complex was demonstrated to 576 function in stress-induced autophagy (Wang et al., 2016) and proteasome inhibitors 577 stabilized SCAR during dark-induced primary root growth inhibition (Dyachok et al., 578 2011). It is possible that SPI-mediated proteolytic pathways and BRK1 operate 579 antagonistically to specify the levels of W/SCR at the RHID that enables the transition to 580 actin-dependent rapid tip growth. However, such as scenario is complicated by the 581 observation that the formation of tip-directed SPI-YPet does not appear to require 582 BRK1. This suggest that SPI's appearance at the root hair tip is regulated by other 583 factors besides W/SCR.

584
The functional links between SPI and W/SCR-ARP2/3 during root hair 585 development as suggested by live cell imaging is supported by the double mutant 586 studies reported here. We found that SPI is epistatic to BRK1 for RHID and root hair tip an NPF-independent function of ARP2/3. Future studies will probe deeper into the 597 regulatory relationships between SPI and W/SCR-ARP2/3 that underlie these 598 unexpected genetic interactions.

599
In summary, our work provides new data that contribute to our understanding of SPI and W/SCR-ARP2/3, the precise nature of such relationships remain unclear. One 614 plausible scenario is that SPI negatively regulates an actin pathway that is parallel to 615 W/SCR-ARP2/3. The persistence of BRK1-YFP signal in spi suggests that one 616 mechanism for negative regulation could occur at a node in which SPI and W/SCR 617 converge. For the future, it will be important to determine whether SPI physically 618 interacts with actin or subunits of the W/SCR-ARP2/3 complex to better explain the 619 functional links between SPI and W/SCR-ARP2/3.

656
To identify the HLB2 gene, homozygous hlb2 (Col-0 ecotype) was out-crossed to 657 the Landsberg erecta ecotype to generate seeds for map-based cloning because 658 attempts to identify the responsible mutation for hlb2 phenotype using TAIL-PCR had 659 been unsuccessful. Segregating F2 seedlings were surface sterilized as described 660 above and grown for three days. These seedlings were then transferred to MS media 661 containing 50 nM LatB and root lengths were marked on the plates to track root growth.

709
For the BRK1-mRuby3 construct, the fluorescent protein 3x-mRUBY3 was tagged with 710 a C-terminal linker (10 Alanine, Glycine) using Thermo Fisher Scientific GeneArt to 711 include BamH1 and XbaI sites at its 5' and 3' ends, respectively. Codon optimization for 712 Arabidopsis was performed on the 3x-mRUBY3 and internal linkers prior to synthesis.

713
The 3X-mRuby3 fragment was inserted as a BamH1/XbaI into the plasmid   To quantify root hair length, fully-grown root hairs from a region of the primary 761 root located between 5 to 15 mm from the root tip were photographed with a Nikon 762 SMZ1500 stereomicroscope. For planar polarity measurements, images of trichoblasts 763 in which the apical and basal end walls were clearly visible were acquired with a Nikon 764 Eclipse TE300 inverted microscope using a 20x objective. Root hair lengths were 765 extracted from digital images using ImageJ (v1.51) software (https://imagej.nih.gov/ij/).

766
For planar polarity, the distance between the basal trichoblast wall and base of the 767 emerging root hair (a) and the length of the trichoblast (b) were obtained using Image J. Live cell imaging of root hairs using confocal microscopy was performed on 4 or 5-day-781 old seedlings grown on the 48 mm × 64 mm coverslip system described above.

782
Coverslips with the seedlings were placed horizontally on the stage of an inverted Leica was drawn and mean fluorescence within this area was acquired using Image J.

801
Fluorescence was expressed as the ratio of mean fluorescence within the root tip ROI 802 to background fluorescence. For SPI-YPet, the background used was the region 803 adjacent, but outside the root hair tip ( Figure 1C) while for SEC-RFP, the background 804 used was an area on the sub-apical region of the root hair tip ( Figure 2F). For BRK1-805 YFP, a ROI was drawn along the apical-most root hair tip that was about 20 pixels-wide 806 using the selection brush tool of image J. The ratio of fluorescence within this area to 807 background fluorescence was obtained ( Figure 2C). Root hair growth rate data was 808 derived from the same root hair images in which fluorescence images were acquired.

809
For the former, an image of a root hair was taken at time 0 and every 5 minutes 810 thereafter. The growth rate refers to the displacement of the root hair tip in μm divided 811 by time elapsed (min).

815
Transverse sections of the root hair tip were obtained from Z-stacks using the surpass 816 view interface of the Imaris software and exported as 8-bit TIFF files. From these 817 images, an ROI spanning the circular area of the root hair tip was drawn using image J 818 and mean fluorescence was extracted ( Figure 3O). The ratio of the tip fluorescence to 819 background was obtained from 18-21 root hairs.

820
For SPI-YPet and BRK1-YFP, a scatter plot to determine the relationship 821 between growth rate and tip-focused fluorescence was done in R (R Core Team, 2019) 822 using ggplot function in ggplot2 package (Wickham, 2016). Linear regression analysis 823 were performed using lme4 package on R software (Bates et al., 2015, R Core Team, 824 2019).

825
Trichomes were imaged using a Hitachi Table top TM3000   for data in Figure 7D on root hair planar polarity in wild-type, spi, arp and spi arp.   Based on their common location at the root hair initiation domain (RHID), ROP/RAC-GTPase and ROP-GEF3 activate the W/SCR complex. This in turn activates a pool of ARP2/3 in the RHID to ensure the correct position of root hair emergence. Genetic interaction studies point to SPI and W/SCR-ARP2/3 functioning in parallel pathways during root hair initiation and early root hair bulge formation. SPI, through a yet to be determined mechanism, is recruited to the tips of root hair outgrowths via post-Golgi vesicles. SPI enrichment at the root hair tip then negatively regulates the W/SCR-ARP2/3-actin nucleation pathway as the root bulge enlarges by displacing or degrading BRK1. The antagonistic effect of SPI on BRK1 leads to the depletion of the active pool ARP2/3. Actin-mediated rapid root hair tip growth then commences predominantly through a SPI-dependent mechanism.