Microtubule-associated ROP interactors both delimit and transduce ROP signaling and regulate microtubule dynamics

Evidence suggests that ICR proteins function as adaptors that mediate ROP signaling. Here, we studied the functions of ICR2 and its homologs ICR5 and ICR3. We showed that ICR2 is a microtubule-associated protein that regulates microtubule dynamics. ICR2 can retrieve activated ROPs from the plasma membrane, and it is recruited to a subset of ROP domains. Secondary cell wall pits in the metaxylem of icr2 and icr5 Arabidopsis single mutants and icr2/icr5 double and icr2/icr5/icr3 triple mutants were denser and larger than those in wild-type Col-0 seedlings, implicating these three ICRs in restriction of ROP function. The icr2 but not the icr5 mutants developed split root hairs further implicating ICR2 in restriction of ROP signaling. Taken together, our results show that ICR2, and likely also ICR5 and ICR3, have multiple functions as ROP effectors and as regulators of microtubule dynamics.

We have previously identified a family of ROP effectors that we designated "Interactors of 45 Constitutively active ROP" (ICRs) (Lavy et al., 2007). The ICRs are coiled-coil domain-46 containing proteins that do not contain additional known structural or catalytic domains 47 organization of ICR2mCh-labeled microtubules were found around ROP2 domains ( Figure  217 6A), and faint ICR2mCh labeling was detected around the ROP4 domains ( Figure 6B). 218 However, significant clustering of ICR2mCh-tagged microtubules was found around domains 219 containing ROP6, ROP9, and ROP10 ( Figure 6C-E) and weaker but clearly visible 220 microtubule reorganization was found around ROP11 domains ( Figure 6F). These results 221 suggest that ROPs may differ in their abilities to recruit ICR2-associated microtubules to 222 specific domains in the plasma membrane. We hypothesize that ICR2 has a dual role: On one 223 hand, it restricts ROP activity to domains in the plasma membrane, and on the other hand, it 224 functions as a scaffold for ROP interactions with microtubules and possibly with other 225 proteins.

ICR1 and ICR2 bind microtubules in vitro 227
Both ICR1 and ICR2 localize to microtubules in vivo (Hazak et  interaction with microtubules. To examine whether ICR1 and ICR2 are indeed MAPs, we 231 tested their interactions with microtubules in vitro using three independent assays. 232 Escherichia coli-expressed, affinity-purified ICR1-His 6 and ICR2-His 6 at concentrations 233 ranging between 1 to 10 µM were incubated with preformed taxol-stabilized microtubules. 234 The protein mixtures were precipitated by centrifugation at 100,000 x g, and the precipitated 235 proteins were separated by SDS-PAGE and visualized by Coomassie blue staining ( Figure  236 7A, 7-Supplement 1A). The levels of precipitated ICR1-His 6 and ICR2-His 6 were quantified 237 by densitometry of the relevant bands ( Figure 7B, 7-Supplement 1B). MAP65, a known 238 microtubule-interacting protein, was used a positive control. The binding of recombinant 239 ICRs to microtubules was saturated at stoichiometries of 0.4 mol ICR1-His 6 per mol of 240 tubulin and 0.85 mol ICR2-His 6 per mol of tubulin. This is in agreement with our findings 241 that both ICR1 and ICR2 interact directly with microtubules and that the binding of ICR2 242 with microtubules is stronger than that of ICR1. 243 Second, in vitro immuno-fluorescence assays were used to examine whether ICR1 and ICR2 244 colocalize with polymerized microtubules. To visualize microtubules, rhodamine-labeled 245 tubulin was mixed with non-labeled tubulin and polymerized into microtubules in the 246 presence of ICR1-His 6 or ICR2-His 6 . Visualized by in vitro immuno-localization established 247 that ICR1 and ICR2 are MAPs ( Figure 7C-E, 7-Supplement 2A-C). Incubation with 248 denatured ICR1-His 6 /ICR2-His 6 were used as negative controls ( Figure 7F-H, 7-Supplement 249 2D-F). ICR1 was distributed in individual punctae, whereas ICR2 was more evenly 250 distributed along microtubule filaments suggesting stronger binding, in line with results of 251 co-precipitation assays. 252 Third, we carried out an in vitro microtubule bundling assay. To this end, rhodamine-labeled 253 tubulin was polymerized into microtubules in the presence of increasing concentrations of 254 ICR1 or ICR2. With ICR1, microtubule bundling was detected only at the high concentration 255 of 5 µM, whereas ICR2 caused bundling even at 0.1 µM ( Figure 7I-K, 7-Supplement 3). 256 Taken together, the bundling assays further confirmed that that ICR1 and ICR2 are MAPs 257 and that binding of ICR1 to microtubules is weaker than ICR2 binding. 258 259 260

ICR2 co-localizes with microtubules in all stages of the cell cycle 261
To characterize the subcellular localization of ICR2, we generated a marker composed of the 262 genomic sequence of ICR2 (including introns) with its promoter fused with the sequence for 263 3xYPet. To reduce potential steric hindrance a 33 amino acid linker was placed between 264 ICR2 and the 3xYPet tag. To avoid potential mis-localization due to overexpression, the 265 pICR2::ICR2 genomic :3xYpet construct was transformed into two icr2 T-DNA insertion mutants, 266 icr2-1 and icr2-2, that also express the microtubule marker RFP-MBD (icr2-1 X 267 In dividing cells, ICR2 was colocalized with microtubules during mitosis ( Figure 9A). ICR2-280 3xYPet colocalized with RFP-MBD in all mitotic stages including the preprophase band, the 281 spindle during metaphase and anaphase, and the expanding phragmoplast microtubules in 282 telophase ( Figure 9B). Interestingly, the localization of ICR2 on microtubules during cell 283 division coincided with the co-expression data, which showed high correlations with cell 284 division cytoskeleton genes (Table S1). 285 The pit phenotype of icr2 mutants prompted us to examine ICR2 localization in vascular  Figure 10A-D). Hence, ICR2 was expressed during relatively short 295 time window at the onset of dedifferentiation indicating that its likely has specific function 296 during xylem differentiation. The short time window in which ICR2 was expressed may also 297 explain why it was difficult to detect it in root vascular tissues. 298 In the dedifferentiating xylem cells, ICR2-3xYPet colocalized with lignified secondary cell 299 wall ( Figure 10E), whereas no fluorescence was detected in the pitted areas. Because the 300 ICR2-3xYPet fluorescence was detected only in few cells and was absent in most of the cells 301 that had secondary cell walls, it could not be a results of fluorescence channel spillover. 302 Furthermore, the images were generated using spectral separation to ensure the presence of 303 the YPet fluorescence. The strict colocalization of ICR2 with the secondary cell walls 304 strongly suggests that it was localized along cortical microtubules. The pit phenotype of the 305 icr2 mutant implicates ICR2 in restriction of pit formation and regulation of pit size. In the 306 developing MX, ICR2 may function by recruiting ROPs from the plasma membrane to 307 microtubules. 308 icr2 mutants display altered microtubule organization and dynamics 309 The localization of ICR2 to microtubules at all stages of the cell cycle suggested that it may 310 affect the organization and dynamics of microtubules. To test this, icr2-1 and icr2-2 plants 311 were crossed with UBQ10::RFP-MBD, and analysis of microtubule dynamics was carried out 312 on non-segregating double homozygous plants using high-frequency time-lapse imaging and 313 tracking of individual microtubule filaments. The tracking data ( Figure 11A) was used to 314 create kymographs ( Figure 11B), which were then used to calculate microtubule growth and 315 shrinkage rates, the time spent in each condition, transition times, and pauses in 316 growth/shrinkage. In root epidermal cells as well as root hairs, microtubule growth rates were 317 significantly slower in the icr2 mutants than Col-0 plants (p≤0.001) ( Figure 11C). In contrast, 318 shrinkage rates were lower only in the epidermis ( Figure 11C). Additionally, time spent at 319 pause was higher in mutant root epidermal cells than Col-0 plants ( Figure 11-Supplement 1), 320 and the transitions between filament growth, shrinkage, and pause occurred at higher 321 frequency in the icr2 mutants than in Col-0 plants ( Figure 11 that the function of the ICRs is more complex than previously thought. The mutant 336 phenotypes indicate that at least ICR2, ICR3, and ICR5 also function as regulators that 337 restrict ROP signaling, likely by recruiting ROPs from the plasma membrane to 338 microtubules. Furthermore, the phenotypic analysis showed that the function of ICR2 and 339 ICR5 is partially cell specific and that at least ICR2 may have ROP-independent functions in 340 the regulation of microtubule dynamics. The data also suggest that ICR2 functions as effector 341 for some ROPs. By recreating the naturally occurring active ROP domains in the plasma 342 membrane, we were able to visualize how a subset of plasma-membrane anchored, active 343 ROPs recruit the microtubule-associated ICR2 to these domains, thus relaying 344 spatiotemporally regulated signal transduction into the cell. 345

ICR2 is a MAP 346
The in vitro assays confirmed the ICR2 is a MAP. Our in vitro system also enabled us to 347 compare the microtubule binding strength of ICR1 and ICR2. The co-sedimentation with 348 microtubules revealed that ICR2 has a higher affinity for microtubules than ICR1. Whereas 349 the binding of recombinant ICR2-His 6 to microtubules was not saturated until 0.85 mol per 350 mol of tubulin, saturation of ICR1-His 6 was reached at only 0.35 mol per mol of tubulin. 351 Similarly, the bundling assays showed that ICR1 caused microtubule bundling at 5 µM, 352 whereas ICR2 induced microtubule bundling at the much lower concentration of 0.1 µM. The 353 tighter binding of ICR2 to microtubules was also detected in the in vitro colocalization 354 assays, which revealed an even distribution of ICR2 along microtubules at 0.5 µM, whereas 355 ICR1 was detected as discrete punctae at 1 µM. It will be interesting to examine whether and 356 how the differential binding of ICR1 and ICR2 to microtubules affects their functions as 357 MAPs and as ROP signaling effectors. gRNA-tRNA plasmid, which contains a gRNA-tRNA-fused fragment, was used as a template 494 to synthesize the PTG construct. The gRNA scaffold fragment was amplified by PCR using a 495 pair of specific primers (Bsa-gRNA-F and gRNA-R), whereas the tRNA Gly fragment was 496 amplified as an overlapping fragment of the primers g-tRNA-F and tRNA-R. Then these two 497 fragments were fused as a gRNA-tRNA by overlapping extension PCR using primers Bsa-498 gRNA-F and tRNA-R. The overlapping PCR product was separated and purified from an agarose gel, and then inserted into pJET1.2 (Thermo Scientific) to generate the template 500 plasmid. The specific spacer sequences targeting ICR2, ICR3, and ICR5 were selected using 501 the CRISPR-PLANT database (www.genome.arizona.edu/crispr/). The PTG clones were 502 created using Golden Gate (GG) for the assembly of DNA fragments. In order to ligate 503 multiple DNA fragments in a desired order, GG assembly requires distinct 4-bp overhangs to 504 ligate two DNA fragments after digestion with BsaI. The gRNA spacers are the only unique 505 sequences in the PTG and were used for this purpose. Each part was amplified with spacer-506 specific primers containing the BsaI adaptor, except two terminal parts using gRNA spacer 507 primer and terminal specific primers containing BbsI site. These PCR fragments were ligated 508 together using GG assembly to produce the PTG with complete gRNA spacers targeting 509 ICR2, ICR3, and ICR5. The assembled product was amplified with short terminus specific 510 primers containing the BbsI adaptor. Next, using a second GG assembly step, the PTG 511 fragment was inserted into the BbsI digested pEntr_L1L2_AtU6gRNA. The PTG cassette was 512 than inserted into pMR294_pKGCAS9PLUS-1 by Gateway LR Clonase (Thermo Fisher 513 Scientific). The pEntr_L1L2_AtU6gRNA and pMR294_pKGCAS9PLUS-1 vectors were gifts 514 from Professor Gitta Coaker, University of California, Davis. 515 In all cloning, PCR-generated fragments were sequenced to verify that no PCR-generated 516 errors were introduced. In the cases of gene fusions, following cloning, the borders between 517 fragments were sequenced to verify that fragments were in frame. All primers and plasmids 518 used and generated in this work are listed in Tables S4-S6.  519 Sequencing. DNA sequencing was performed at the Tel Aviv University DNA sequencing 520 facility and was carried using the BigDye Terminator Cycle Sequencing Kit (Applied 521 Biosystems). 522

Plant genomic DNA isolation 523
Typically, 100 mg of liquid N 2 batch-frozen leaf tissue were ground with a mortar and pestle, 524 and genomic DNA was isolated using the GenElute Plant Gemomic DNA Kit (Sigma) 525 according to the manufacturer's protocol. 526

Total RNA isolation from plants 527
Arabidopsis seedlings were batch frozen using liquid N 2 , and tissue was ground with a mortar 528 and pestle. Total RNA was isolated from the ground material using the RNeasy SV total 529 RNA isolation kit (QIAGEN) according to the manufacturer's instructions. Agrobacterium tumefaciens strain GV3101/pMP90 was used for transient and stable 547 expression of recombinant genes in N. benthamiana and Arabidopsis as previously described 548 (Lavy et al., 2007). Growth media for bacteria was prepared as previously described 549 (Ausubel et al., 1995). For solid media, 1.5% w/v of agar was added to the medium. E. coli 550 cells were selected on 100 μg/mL ampicillin or 50 μg/mL kanamycin. Agrobacterium 551 tumefaciens GV3101/pMP90 was selected on 100 μg/mL gentamycin and 50 μg/mL 552 spectinomycin. 553

Yeast two-hybrid assays 554
Saccharomyces cerevisiae strain PJ69-4a was used as host. Plasmids for expression of ROPS 555 Arabidopsis on plates, plates contained 0.5X Murashige Skoog (MS) medium (Duchefa) 584 titrated to pH 5.7 with MES and KOH and 0.8% plant agar (Duchefa). In some cases, the 585 medium was prepared with 1% sucrose. The seeds were then moved to a growth chamber and 586 placed vertically in most cases, or horizontally for germination assays to grow under long-587 day conditions (16 h light/8 h dark-light intensity 100 µE·m -2 ·s -1 ) at ~22 °C. Prior to the 588 transfer to the growth chamber, the sown seeds were stratified at 4 °C for 48 h in darkness. In 589 both cases, seeds were surface sterilized by evaporation of HCl (6 mL) in sodium 590 hypochlorite (100 mL) in a closed container for 1 h. 591 Stable transformation in Arabidopsis. Transformation was performed by the floral dip 592 method (Clough and Bent, 1998).

Vascular Cell Induction Culture System Using Arabidopsis Leaves (VISUAL) 596
Vascular cell dedifferentiation was carried out with the VISUAL system as previously 597 described (Kondo et al., 2016) with the following modifications. In brief, 10-15 seedlings 598 were grown in 10 ml liquid growth media and (2.2 g/L MS Basal Medium, 10 g/L sucrose 599 and 0.5 g L -1 MES adjusted to pH to 5.7 using KOH) in 6-wells-plates at 25 ºC under a long- tracking data were used to create kymographs, which were then used to calculate microtubule 621 growth and shrinkage rates, the time spent at each condition, as well as the transitions 622 between them and pauses in growth/shrinkage. This analysis of imaging data was performed 623 using the KymoToolBox ImageJ plugin (Zala et al., 2013). Typically, 5-10 microtubule 624 filaments were analyzed per cell and five cells, each from a different plant, were analyzed for each genotype and cell type. Overall, the number microtubule filaments analyzed was 77-626

Secondary cell wall pits area and pit density per area 628
Analysis of secondary cell wall of the MX pits was carried out on seedling roots at 8 DAG. 629 Roots were imaged using differential interference contrast (DIC) light microscopy after 630 clearing with chloral hydrate:lactic acid (2:1) for 1-3 days. To quantify the area of secondary 631 cell wall pits, pits were manually selected in DIC images and analyzed using ImageJ. Pit 632 density was calculated as the number of secondary cell wall pits divided by the area of MX 633 vessel cells and expressed as the number of pits per 1000 m 2 . Two or three cells were 634 imaged for each plant, and four or five plants were analyzed for each genotype. 635

Protoxylem lignification 636
Roots at 7 DAG were imaged for lignin autofluorescence by excitation at 405 nm. Emission 637 was detected with a spectral detector set between 410 nm and 524 nm. Z-stacks were taken of 638 6-10 focal planes, and maximum intensity images were created. Analysis was carried out in 639 the maturation zone of the root on maturing PX cells, which at this region have a well-640 defined spiral pattern. No MX differentiation was detected. Mean distance between lignified 641 spirals was measured using the semi-automated Cell-o-Tape macro for ImageJ (Fiji). Five 642 roots were analyzed for each genotype, and in each plant two PX cells were imaged and 643 quantified. 644

Analysis of root hairs deformation 645
Root hairs were observed in seedlings grown on 0.5X MS agar medium at 8 DAG. Ten 646 seedlings for each genotype were compared, and the percent of normal and deformed root 647 hairs was scored. 648

Root hair measurements 649
Root hairs in seedlings at 7 DAG were imaged and measured as previously described 650 (Denninger et al., 2019). The first visible swelling of the cell outline was defined as first 651 bulge, and distance to root tip was measured. Root hair density was analyzed in the next 652 2 mm. Root hair length was measured in a region 3-6 mm away from the root tip. 653

Light and confocal laser scanning microscopy 654
Brightfield and Nomarsky DIC imaging was performed with an Axioplan-2 Imaging 655 microscope (Zeiss) equipped with an Axio-Cam and a cooled charge-coupled device camera using either 10X, 20X dry, or 63X water immersion objectives with numerical aperture 657 values of 0.5, 0.9, and 1.2, respectively. Laser scanning confocal microscopy and associated 658 brightfield and DIC imaging was performed using LSM 780-NLO confocal laser scanning 659 microscope (Zeiss) with 10X and 20X air objectives and 40X and 63X water immersion 660 objectives with numerical apertures of 0.3, 0.8, 1.2, and 1.15. Fluorescein was visualized by 661 excitation with an argon laser at 488 nm; emission was detected between 493 and 556 nm.