Local and global Cdc42 GEFs for fission yeast cell polarity are coordinated by microtubules and the Tea1/Tea4/Pom1 axis

The conserved Rho-family GTPase Cdc42 plays a central role in eukaryotic cell polarity. The rod-shaped fission yeast Schizosaccharomyces pombe has two Cdc42 guanine-nucleotide exchange factors (GEFs), Scd1 and Gef1, but little is known about how they are coordinated in polarized growth. Although the microtubule cytoskeleton is normally not required for polarity maintenance in fission yeast, here we show that when scdl function is compromised, disruption of microtubules or the polarity landmark proteins Tea1, Tea4, or Pom1 leads to isotropic rather than polarized growth. Surprisingly, this isotropic growth is due to spatially inappropriate activity of Gef1, which is a cytosolic protein rather than a membrane-associated protein at cell tips like Scd1. Microtubules and the Tea1/Tea4/Pom1 axis counteract inappropriate Gef1 activity by regulating the localization of the Cdc42 GTPase-activating protein Rga4. Our results thus demonstrate coordination of “local” (Scd1) and “global” (Gef1) Cdc42 GEFs via microtubules and microtubule-dependent polarity landmarks.


INTRODUCTION
4 observed: mutations affecting MT nucleation and organization often lead to curved cells, 114 while mutations affecting landmark proteins tend to lead to bent or branched cells, 115 particularly after stress. Both of these phenotypes, however, contrast sharply with those 116 associated with mutations in the Cdc42 polarity module, which lead to round-or wide-cell 117 phenotypes (Chang et al., 1994;Kelly and Nurse, 2011;Miller and Johnson, 1994).

118
Collectively, these findings have led to a general model in which MTs and MT-dependent 119 landmark proteins play an important role in selecting sites for polarity establishment but are 120 not required for polarity establishment per se, or for the maintenance of polarized growth 121 (Chang and Martin, 2009;Sawin and Snaith, 2004). At the same time, these differences in

157
To characterize scd1∆ polarized growth in further detail, we used cdc2-asM17 cells,

184
we hypothesized that MTs might contribute specifically to polarized growth in scd1∆ cells.

186
BrB-PP1 treatment, both in the presence and absence of the MT-depolymerizing drug MBC 6 ( Fig. 2, Video 1). Inhibition of Cdc2-asM17 by 3-BrB-PP1 was particularly useful because it 188 allowed imaging of cell growth for several hours without intervening cell division. In the 189 absence of MBC, scd1∆ cdc2-asM17 grew in a polarized manner, as did control (scd1+) 190 cdc2-asM17 cells in the presence of MBC. Strikingly, after addition of MBC to scd1∆ cdc2-191 asM17 cells, Bgs4 no longer localized mainly to cell tips and instead formed transient, 192 mobile patches on the plasma membrane ( Fig. 2A). Accordingly, instead of growing in a 193 polarized manner, these cells grew isotropically, becoming increasingly round over time ( Fig.   194 2B,C). We conclude that MTs are critical for polarized growth in scd1∆ cells, but not in wild-195 type (scd1+) cells.

196
We hypothesized that MTs may contribute to polarized growth in scd1∆ cells via lacked detectable CRIB-3mCitrine at cell tips. We note, however, that other mutant 205 phenotypes (see below) indicate that some biologically relevant, functional Scd1 is produced 206 in these cells, albeit at very low levels.

230
These results indicate that Gef1 is not required for polarized growth in scd1 low cells 231 and therefore that the very low level of Scd1 expressed in scd1 low cells is sufficient for 232 viability and polarized growth. This in turn raised the question of why Tea1 and Tea4 are 233 required for polarized growth in scd1 low cells.

234
We hypothesized two possible roles for the Tea1/Tea4 system. The first possibility 235 was that Tea1 and Tea4 might enhance the intrinsic ability of Scd1 to serve as a GEF when 236 expressed at very low levels. The second possibility, which was motivated by the 237 observation that Gef1 overexpression causes cell rounding (Coll et al., 2003; 238 2012), was that rather than supporting Scd1 function directly, Tea1 and Tea4 may prevent or 239 counteract any inappropriate function of Gef1, which would otherwise somehow interfere 240 with the ability of low levels of Scd1 to promote polarized growth.

241
To distinguish between these possibilities, we introduced gef1∆ into tea1∆ scd1 low 242 mCherry-Bgs4 cdc2-asM17 cells and imaged cells after scd1 repression and 3-BrB-PP1 243 addition. Remarkably, gef1∆ completely reversed the isotropic growth of tea1∆ scd1 low 244 mCherry-Bgs4 cdc2-asM17 cells, which now grew in a highly polarized manner, as seen 245 both by cell shape and by mCherry-Bgs4 enrichment at cell tips (Fig. 4A,B; Video 3). This 246 provides strong support for the second of the two possible roles proposed above.

247
In addition to a central catalytic Dbl homology (DH) domain, Gef1 contains an N-248 terminal region of unknown function and a C-terminal region that is proposed to contain a 249 Bin/amphiphysin/Rvs (BAR) domain (Das et al., 2015). Because gef1∆ removes all of these 250 domains, we wanted to determine whether the polarized growth seen in gef1∆ tea1∆ scd1 low 251 mCherry-Bgs4 cdc2-asM17 cells was specifically due to a loss of Gef1's GEF activity. We 252 therefore mutated conserved residues E318 and N505 in the Gef1 DH domain to generate a 253 mutant (E318A, N505A; termed gef1-EANA) that, based on previous structural and in vitro 254 biochemical analyses, should be folded properly but unable to bind Cdc42 (Aghazadeh et   with this, we found that gef1-EANA is a loss-of-function mutation, even though Gef1-EANA 257 protein localized in vivo identically to wild-type Gef1 (Fig. 4-figure supplement 1). In further 258 imaging experiments, we found that after scd1 repression, gef1-EANA tea1∆ scd1 low 259 mCherry-Bgs4 cdc2-asM17 cells were polarized both before and after 3-BrB-PP1 addition 260 (Fig. 4A,B; Video 3). This indicates that the reversal of isotropic growth seen in our 261 8 experiments can be attributed specifically to the loss of Gef1 GEF activity, rather than to the 262 absence of Gef1 protein more generally.

263
Collectively, these results suggest not only that Gef1 is not required for polarized 264 growth in scd1 low cells but also that preventing or counteracting Gef1 activity is a prerequisite 265 for polarized growth in scd1 low cells. According to this view, the main role of Tea1 (and Tea4) 266 in promoting polarized growth in scd1 low cells is to prevent isotropic growth caused by 267 inappropriate Gef1 activity; correspondingly, if Gef1 is not present, then Tea1 is no longer 268 required for polarized growth.

295
Targeting to cell tips converts Gef1 from a global to a local Cdc42 GEF

296
Together with our finding that gef1∆ and gef1-EANA mutations restore polarized 297 growth to tea1∆ scd1 low cells, the observation that Gef1 is normally cytosolic suggested that 298 9 the isotropic growth seen in scd1 mutant cells in the presence of MBC or in tea1∆ or tea4∆ 299 backgrounds is due to Gef1 acting on membrane-associated Cdc42 from a cytosolic pool, as 300 a "global" Cdc42 GEF. To support this view, we asked whether artificial targeting of Gef1 to 301 cell tips-that is, changing a "global" Cdc42 GEF into a "local" GEF--would convert it from an 302 activator of isotropic growth into an activator of polarized growth.

303
In one set of experiments, we used GFP and GFP-binding protein (GBP; (Rothbauer 304 et al., 2008)) to heterodimerize Gef1 with Tea1 (Fig. 5B). Fusion of Gef1-mCherry to GBP 305 rescued the synthetic lethality of gef1∆ scd1∆ cells, indicating that Gef1-mCherry-GBP is 306 functional. In gef1∆ scd1∆ cells expressing untagged Tea1, Gef1-mCherry-GBP was 307 cytosolic during interphase, and cells displayed the round morphology characteristic of 308 scd1∆ mutants. By contrast, in gef1∆ scd1∆ cells expressing Tea1-GFP, which is normally 309 localized to cell tips (Behrens and Nurse, 2002), Gef1-mCherry-GBP became localized to 310 cell tips, and cells displayed a normal, wild-type morphology. This demonstrates that 311 targeting Gef1 to cell tips is sufficient to promote highly robust polarized growth in scd1∆ 312 cells. In addition, the fact that upon dimerization Gef1 becomes localized to sites of Tea1 313 localization, rather than vice-versa, supports our finding that Gef1 normally has no specific 314 localization within the cytoplasm.

324
However, upon addition of rapamycin, CRIB-3mCitrine was quickly recruited to cell tips, and 325 morphology and polarized growth became similar to wild-type cells (Fig. 5C).

326
Taken together with the results above, these results suggest that during interphase,

327
Gef1 is normally localized to the cytosol, where it is active as a global Cdc42 GEF.

343
To determine whether isotropic growth after Pom1 inhibition depends on Gef1, we

11
In parallel, we introduced rga4∆ into scd1 low and gef1∆ scd1 low cells (Fig. 7B,C). The 373 rga4∆ scd1 low cells were compromised in polarized growth, becoming wider and rounder 374 than control scd1 low cells, although this was not as extreme as in rga4∆ scd1∆ or tea1∆ 375 scd1 low cells (see Discussion). Importantly, however, these polarity defects were almost 376 completely rescued in rga4∆ gef1∆ scd1 low cells, indicating that the defects associated with 377 rga4∆ in scd1 mutants are indeed mediated through Gef1.

378
The rescue of rga4∆ wide/round morphological defects by gef1∆ in a scd1 low 379 background appeared to conflict with a previous report that rga4∆ gef1∆ double mutants 380 were wider than either rga4∆ or gef1∆ single mutants (Kelly and Nurse, 2011). We therefore

397
We have shown that Gef1 is a cytosolic, "global" Cdc42 GEF, unlike Scd1, which is a      tagged beta-glucan synthase Bgs4, whose localization normally correlates precisely with 446 polarized growth (Cortes et al., 2005). An extended interphase allowed us to unambiguously 447 distinguish polarized vs. isotropic growth in scd1 mutants, which, because of their short/wide 448 shape, do not normally elongate very much during a single cell cycle. In addition, imaging 449 during extended interphase can circumvent problems that may arise when strains have 450 abnormal phenotypes associated with cytokinesis (e.g. pom1∆; (Bahler and Pringle, 1998)

474
In this context, it is also interesting to compare polarity phenotypes of scd1 low rga4∆ 475 with scd1 low tea1∆, because scd1 low tea1∆ cells grow completely isotropically (as do 476 scd1 low tea4∆, and scd1 low pom1∆), even though they express enough Scd1 to maintain 477 viability in a tea1∆ genetic background. We can imagine two non-exclusive explanations for 478 this difference in phenotype. First, in addition to regulating Rga4, the Tea1/Tea4/Pom1 axis 479 could have a separate role in either bolstering scd1 low function or countering gef1 function.

480
Second, the different phenotypes could be due to the presence vs. the absence of Rga4.

481
That is, in scd1 low tea1∆ cells, the GAP activity of Rga4 will be distributed essentially evenly

546
We note that when grown on solid PMG medium without thiamine, scd1 low tea1∆, 547 scd1 low tea4∆, and scd1 low pom1∆ double mutants formed colonies that were noticeably 548 smaller than wild-type cells and scd1 low single mutants. Under these conditions, the double 549 mutants also showed some defects in septum positioning and in completion of cytokinesis.

645
Quantification of the percentage of cells with polarized mCherry-Bgs4 or GFP-Bgs4 signals 646 on cell tips (Fig 2, 3, 4 and 6 Figure 4-figure supplement 1. gef1-EANA-3mCherry is a gef1 loss-  is not detected at cell tips, and Cdc42-mCherry SW is not enriched at cell tips, even though it remains associated with the plasma membrane. Unlike the high-contrast CRIB reporter, these additional reporters do not show any nuclear localization; therefore, their absence from cell tips in scd1∆ cells cannot be attributed to an increased nuclear localization. In bottom panels, cells were treated with 3-BrB-PP1 for 5 hr prior to imaging. Scale bar, 10 µm.   genotypes (all are in scd1 low cdc2-asM17 mCherry-bgs4 background), at the indicated times after start of imaging. scd1 expression was repressed with thiamine for 24 hr before imaging. 3-BrB-PP1 was added 30 minutes before imaging. Diagrams show cell outlines at beginning and end of movies; outlines were aligned slightly to account for limited cell movement. Note that newborn daughter cells often have less mCherry-Bgs4 at cell tips (e.g. in scd1 low cdc2-asM17 mCherry-bgs4 tea1∆ gef1-EANA cells). (B) Percent cells containing polarized mCherry-Bgs4 at cell tips, from movies of the type shown in Fig. 4A and Fig. 3; n indicates total number of cells scored (polarized plus non-polarized). mCherry-Bgs4 polarization was measured during the first four hours of imaging, when mCherry-Bgs4 signal remains strong. Pairwise differences relative to control were highly significant for all strains except gef1∆, (p <0.0001; Fisher's exact test, with correction for multiple comparisons). Scale bar, 10 µm. See also Figure 4--figure supplement 1 and Video 3. Localization of Gef1-3mCitrine in wild-type and scd1∆ cells. In both cases, Gef1 is not at cell tips but is present at the septum. (B) Ectopic targeting of Gef1-mCherry-GBP to cell tips by coexpression of Tea1-GFP, in scd1∆ background. In cells lacking tagged Tea1 (left-hand panels), Gef1-mCherry-GBP remains cytosolic, and cells are round. In tea1-GFP cells (middle panels), Gef1-mCherry-GBP is at cell tips, and cells become strongly polarized. Right-hand panels show absence of bleed-through from GFP channel to mCherry channel. (C) Time-resolved targeting of Gef1-Frb to cell tips by rapamycin-induced dimerization in gef1-Frb tea1-FKBP scd1∆ CRIB-3mCitrine cells. Rapamycin was added just after the 0 min timepoint. Red arrowheads indicate examples of CRIB-3mCitrine appearance at cell tips after rapamycin addition. Note that after rapamycin addition, cells grow polarized, and eventually CRIB-3mCitrine also appears at opposite cell tips (green asterisks, 56 min), indicating transition from monopolar to bipolar growth. Scale Bars, 10 µm. See also Figure  Pom1-as1-tdTomato and GFP-Bgs4, in scd1∆ cdc2-asM17 backgrounds after 3-BrB-PP1 treatment. 3-BrB-PP1 inhibits activity of both Cdc2-asM17 and Pom1-as1-tdTomato and was added just after the 0 hr time-point. Diagrams show cell outlines at beginning and end of movies; outlines were aligned slightly because of limited cell movement. Note that 3-BrB-PP1 treatment depolarizes Pom1-as1-tdTomato, and this leads to isotropic growth. (B) Percent cells containing polarized GFP-Bgs4 at cell tips, from movies of the type shown in (A). Differences were highly significant (p<0.0001; Fisher's exact test). (C) Time courses from movies showing cell morphology and distribution of Pom1-tdTomato or Pom1-as1-tdTomato, and Rga4-3GFP, in scd1∆ cdc2-asM17 backgrounds after 3-BrB-PP1 treatment. Scale bars, 10 µm. See also Videos 6 and 7.  is a plasma membrane-associated "local" Cdc42 GEF at cell tips and maintains a focused polarity zone via positive feedback. 2) Gef1 (pink) is a cytosolic, "global" Gdc42 GEF. 3) Microtubules (MTs; green) target the Tea1/Tea4/Pom1 axis (green) to cell tips. 4) This restricts Cdc42 GAP Rga4 (blue) to the plasma membrane in the cell middle. 5) Rga4 on the membrane locally counters cytosolic Gef1 activity, preventing net GEF activity in the cell middle (represented by large arrows pointing toward cell tips but small arrows pointing elsewhere). (B) The model as applied to scd1∆ and scd1∆ rga4∆ cells. In scd1∆ cells, there is no strong focused polarity zone, but Rga4 can still locally counter global Gef1 activity, leading to greater "net" Gef1 activity in the region of cell tips, as in wild-type cells. Therefore, cells are polarized but significantly wider than wild-type. In scd1∆ rga4∆ cells, absence of Rga4 means that Gef1 is not locally countered anywhere and is therefore active isotropically. Distribution of MTs and Tea1/Tea4/Pom1 will also ultimately be abnormal, due to round cell shape. (C) The model as applied to scd1 low , scd1 low tea1∆/tea4∆/pom1∆, and scd1 low tea1∆/tea4∆/pom1∆ gef1∆ cells. In scd1 low cells, only a very limited amount of the "local" Cdc42 GEF (i.e. Scd1) is present at cell tips, and thus the polarity zone is not focused as in wild-type. However, "net" Gef1 activity remains greater in region of cell tips, and thus Gef1 cooperates with Scd1. In scd1 low tea1∆/tea4∆/pom1∆ cells, Rga4 is no longer spatially restricted. "Net" Gef1 activity is therefore isotropic, competing with Scd1 and overwhelming the contribution that low Scd1 levels would otherwise make to polarized growth. In scd1 low tea1∆/tea4∆/pom1∆ gef1∆ cells, competition from Gef1 is alleviated, allowing the limited amount of Scd1 to support polarized growth.