C-terminal domain (CTD) phosphatase links RhoGTPase signaling to Pol II CTD phosphorylationin Arabidopsis and yeastBo Zhanga,b, Guohua Yangb,1,2, Yu Chenb,1,3, Yihong Zhaoc, Peng Gaob,4, Bo Liud, Haiyang Wange,and Zhi-Liang Zhengb,f,g,5

aPlant Nutrient Signaling and Fruit Quality Improvement Laboratory, Citrus Research Institute, Southwest University, Beibei, Chongqing 400712, China;bDepartment of Biological Sciences, Lehman College, City University of New York, Bronx, NY 10468; cDivision of Biostatistics, Department of Child andAdolescent Psychiatry, New York University Langone Medical Center, New York, NY 10016; dDepartment of Plant Biology, University of California, Davis, CA95616; eBiotechnology Research Institute, Chinese Academy of Agricultural Sciences, Beijing 100081, China; fNational Citrus Engineering Research Center,Citrus Research Institute, Southwest University, Beibei, Chongqing 400712, China; and gBiology PhD Program, Graduate School and University Center, CityUniversity of New York, Bronx, NY 10016

Edited by Alberto R. Kornblihtt, University of Buenos Aires, Buenos Aires, Argentina, and approved November 4, 2016 (received for review April 12, 2016)

Rho GTPases, including the Rho, Cdc42, Rac, and ROP subfamilies,act as pivotal signaling switches in various growth and develop-mental processes. Compared with the well-defined role of cytoskel-etal organization in Rho signaling, much less is known regardingtranscriptional regulation. In a mutant screen for phenotypic en-hancers of transgenic Arabidopsis plants expressing a constitutivelyactive form of ROP2 (designated CA1-1), we identified RNA polymer-ase II (Pol II) C-terminal domain (CTD) phosphatase-like 1 (CPL1) as atranscriptional regulator of ROP2 signaling. We show that ROP2 acti-vation inhibits CPL1 activity by promoting its degradation, leading toan increase in CTD Ser5 and Ser2 phosphorylation. We also observedsimilar modulation of CTD phosphorylation by yeast Cdc42 GTPaseand enhanced degradation of the yeast CTD phosphatase Fcp1 byactivated ROP2 signaling. Taken together, our results suggest thatmodulation of the Pol II CTD code by Rho GTPase signaling repre-sents an evolutionarily conserved mechanism in both unicellularand multicellular eukaryotes.

Rho GTPase | ROP GTPase | Pol II | CTD code | CPL1

Rho family GTPases, including four subfamilies, Rho, Cdc42,and Rac in yeast and animals and ROP in plants, are key

plasma membrane-associated signaling switches (1–3). Geneticevidence has shown that these Rho GTPases are involved in awide range of growth and developmental processes, includingcell migration, division, differentiation, tissue morphogenesis,and organ development. In Arabidopsis, there are 11 members ofROP, each with unique and sometimes overlapping functions incell morphogenesis, cell growth, and response to hormones andvarious biotic and abiotic stresses (2–12). Recent studies haveshown that three closely related members, ROP2, ROP4, andROP6, function in cell polarity control through microtubule andactin cytoskeletal organization (3, 8, 10, 12, 13), but involvementof transcriptional regulation remains unknown.RNA polymerase II (Pol II) is a multiunit holoenzyme com-

plex critical for transcription in eukaryotic organisms. Its largestsubunit, RPB1, has a C-terminal domain (CTD) that containsvarious numbers of the heptad peptide (Y1S2P3T4S5P6S7) re-peats (14–17). These repeats undergo dynamic posttranslationalmodifications, such as phosphorylation of Tyr1, Ser2, Thr4, Ser5,and Ser7. Because of their combinatorial complexity and impor-tance in regulating gene expression, these modifications are col-lectively termed the “CTD code” (14–17). Dynamic modulation ofthe CTD code in Pol II along genes is essential for completingvarious key steps of transcriptional process by recruiting the CTD-associated proteins to the transcribing Pol II. A number of proteinkinases and phosphatases have been shown to regulate the CTDSer phosphorylation dynamics in yeast, plant, and animal cells (14–17). In Arabidopsis, there are several members of CTD phospha-

tases with sequence homology to yeast Fcp1 phosphatase (18–21).Among these members, CPL1 has been extensively studied in stressresponse and gene expression regulation (18, 19, 22–25). CPL1 hasbeen demonstrated to act as a Ser5 phosphatase in vitro, but itsupstream regulatory pathway remains to be revealed.Our work presented here establishes an unexpected link be-

tween Rho signaling and the Pol II CTD code modulation viaCTD phosphatase in both Arabidopsis and yeast systems. Using aforward genetic screen, we identified the CPL1 (CTD phospha-tase-like 1) gene as a transcriptional regulator of ROP2 signalingin the control of cell shape, cell size, and cell number. We haveshown that activation of ROP2 signaling stimulates the CTDSer5 and Ser2 phosphorylation status by inhibiting CPL1. Inaddition, we found that auxin, which is known to activate ROP2

Significance

Rho GTPase and polymerase II (Pol II), two key molecules involvedin cellular signaling and transcription in eukaryotic organisms, havebeen separately studied for more than 2 decades without evidenceshowing their functional linkage. We provide genetic and bio-chemical evidence linking these two molecules in an intracellularsignaling pathway. Rho GTPases in Arabidopsis and yeast canmodulate the phosphorylation status of the Pol II C-terminal do-main (CTD) by inhibiting the CTD phosphatases. Our finding rendersstrong support for a direct or “shortcut” model in transcriptionalcontrol. Compared with the classical transcriptional activator/repressor-mediated indirect model, this shortcut model of targetingthe core of Pol II likely provides an efficient transcriptional control torapidly bring about the broad changes in gene expression.

Author contributions: Z.-L.Z. conceived and supervised the project; B.Z., G.Y., Y.C., and Z.-L.Z.designed experiments; P.G. performed the CA1-1 mutagenesis and constructed UBCA1; G.Y.isolated cae2-1 CA1-1 and mapped the cae2 locus; Y.C. cloned cae2 and performed functionalcomplementation; B.Z. performed all other experimental work; Y.Z. performed statisticalanalysis; B.L. repeated the cell shape phenotypic characterization; H.W. provided expertisein protein degradation; and B.Z., B.L., H.W., and Z.-L.Z. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequence reported in this paper has been deposited in the NationalCenter for Biotechnology Information Gene Expression Omnibus database (accession no.GSE78118).1G.Y. and Y.C. contributed equally to this work.2Present address: Department of Infectious Diseases, St. Jude Children’s Research Hospital,Memphis, TN 38105.

3Present address: NGS Department, Genewiz, Suzhou, Jiangsu, PC 215123, China.4Present address: National Research Council of Canada, Saskatoon, SK S7N 0W9, Canada.5To whom correspondence should be addressed. Email: zhiliang.zheng@lehman.cuny.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1605871113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1605871113 PNAS | Published online November 28, 2016 | E8197–E8206

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

19,

202

0

signaling (10), increased CTD Ser5 and Ser2 phosphorylationin a ROP2-dependent manner. Moreover, we observed similarmodulation of the CTD code by Cdc42 GTPase activity in fissionyeast and similar protein degradation of a CPL1 homolog in thebudding yeast. Therefore, our results demonstrate that modu-lation of the Pol II CTD code by Rho GTPases via inhibitingCTD phosphatases represents an evolutionarily conservedmechanism in both unicellular and multicellular eukaryotes.

ResultsIsolation of an Arabidopsis cae2/cpl1 Mutant That Enhanced thePhenotypes of Transgenic Plants Expressing CA-rop2. To identifyregulators or effectors of ROP2 signaling, we mutagenized atransgenic line, designated CA1-1, which overexpresses a consti-tutively active form of ROP2 (CA-rop2 or ROP2G15V) (26) andisolated several recessive monogenic recessive ca-rop2 enhancers(cae). Among these, cae1 has a mutation in the kinesin gene andenhances the root hair phenotypes of CA1-1 (27), and cae2-1 ischaracterized here. As observed by others (13, 28), the cotyledonpavement cells in CA1-1 had a fat, hyperparallel, near-rectangularshape (Fig. 1A), compared with wild-type (WT), which exhibited atypical interdigitating puzzle-like cell shape. Although cae2-1 itselfdid not alter the cell shape, cae2-1 in the CA1-1 background (cae2-1CA1-1) had cotyledon pavement cells that were regularly spaced andnear-square shaped and lacked obvious lobes (Fig. 1A). Quantitativeanalysis using the Shape Factor (29), which has a value twofold higherthan Circularity used in another study (11), showed that CA1-1 hada higher Shape Factor (0.55) than WT and cae2-1 (0.22), but cae2-1CA1-1 had a cell Shape Factor of 0.76 (Fig. 1B), confirming thatcae2-1 CA1-1 is a CA1-1 enhancer. To determine whether other cellparameters were also affected by the enhancer mutation during cot-yledon growth, we analyzed cell images collected from 3- to 6-d-oldcotyledons (SI Appendix, Fig. S1A) using a different cell-clearingmethod (30). We found that CA1-1 cells at day 6 also had a slightlysmaller area on average than WT and that cae2-1 CA1-1 had thesmallest cells at all stages (SI Appendix, Fig. S1B). In contrast, whereasShape Factors inWT and cae2-1 similarly decreased during cotyledongrowth, CA1-1 kept larger Shape Factors and cae2-1 CA1-1 had thelargest ones at all stages. Consistently, the cae2-1 CA1-1 cotyledonhad more cells than all three other genotypes. Therefore, cae2-1greatly enhanced CA1-1 in terms of cell shape, size, and number.Through map-based cloning, we identified a G-to-A mutation in

the first nucleotide of the 11th intron of At4g21670 (Fig. 1C). Thisgene was first identified as CPL1 or FRY2, which encodes a Pol IICTD phosphatase involved in stress responses and acting as aregulator of gene expression (18, 19, 22–25). The cae2-1 mutationresulted in three types of splicing products of CPL1 mRNA incae2-1 CA1-1, and sequencing of the three cDNA moleculesshowed all of which differed from the WT CPL1 transcript(Fig. 1D; SI Appendix, Fig. S2 A and B). The enhanced cell-shapephenotype of cae2-1 CA1-1 was complemented in transgenic plantswith a 7.7-kb CPL1 genomic fragment (SI Appendix, Fig. S2C). Inaddition, crossing CA1-1 into fry2-1 phenocopied the pavement cellshape of cae2-1 CA1-1 (SI Appendix, Fig. S2D), demonstrating thatthe CPL1 mutation is responsible for the enhancer phenotype andthus that CAE2 is allelic to CPL1. As CPL1 is involved in tran-scriptional control, we investigated whether cae2-1 impacted ROP2expression. We found that all four genotypes exhibited a similarexpression level of WT ROP2, although the expression level of theCA-rop2 transgene in cae2-1 CA1-1 was twofold of that in CA1-1(SI Appendix, Fig. S3A). In addition, despite a role for CPL1 inregulating expression of osmotic stress-responsive genes (19), wefound that the enhancer phenotype of cae2-1 CA1-1 is not causedby osmotic stress (SI Appendix, Fig. S4).

Alteration of Ser5 and Ser2 but Not Ser7 Phosphorylation Status ofPol II CTD in CPL1 Mutants. CPL1 has been demonstrated to act asa specific protein phosphatase dephosphorylating the CTD Ser5

residue of RPB1 in vitro (31). To confirm that CPL1 acts in vivo,we examined the CTD Ser5 phosphorylation (Ser5P) status incae2-1. Results showed that Ser5P was up-regulated in cae2-1 by1.8-fold compared with WT without affecting the total RPB1protein level (Fig. 1 E and F). Surprisingly, Ser2 phosphorylation(Ser2P) was also up-regulated (1.4-fold) in cae2-1, whereasSer7P was not affected. Consistently, Ser5P and Ser2P levelswere also higher in cae2-1 CA1-1 than in CA1-1 (Fig. 1 E and F).Such differential Ser phosphorylation pattern was also observedin fry2-1 (Fig. 1G and H). Our result shows the function of CPL1as a CTD Ser5 phosphatase in vivo and reports an unexpectedconsequence of CPL1 mutations in Ser2 dephosphorylation.

Ser5 and Ser2 Phosphorylation Status Is Modulated by ROP2 Signaling.Interestingly, comparison of the Ser5P and Ser2P levels be-tween WT and CA1-1 indicated that CA1-1 had higher levels ofSer5P (3-fold) and Ser2P (1.8-fold) than WT, whereas Ser7Pand RPB1 total protein levels were not affected (Fig. 1 E andF). This difference was also observed by comparing cae2-1 andcae2-1 CA1-1. This indicates that the phosphorylation status ofSer5 and Ser2 but not Ser7 is modulated by ROP2 signaling. Toconfirm this finding, we examined Ser5P and Ser2P in theROP2 loss-of-function mutant. As ROP2 and ROP4 act re-dundantly in the formation of pavement cell shape (13), weused ROP2RNAi rop4-1, a ROP2 RNA interference (RNAi)transgenic line in the rop4-1 knockout mutant background,which greatly suppressed ROP2 expression (13). We foundthat both Ser5P and Ser2P levels decreased dramatically inROP2RNAi rop4-1, without affecting Ser7P and total RPB1protein (Fig. 1 I and J). Therefore, these genetic data consis-tently showed that activation of ROP2/4 GTPases leads tomodulation of the Pol II CTD phosphorylation at the Ser5 andSer2 positions rather than global CTD phosphorylation.Next, we tested whether a signal that activates ROP2/4 activity

also altered the CTD phosphorylation status. Auxin has beendemonstrated to activate Arabidopsis ROP2/4 GTPases (10) andtobacco ROP/Rac1 GTPases (5). In Arabidopsis, alteration of thecotyledon cell shape by auxin treatment is mediated by ROP2/4(10), and thus we investigated whether Ser5 and Ser2 phos-phorylation could be impacted by the exogenous auxin treat-ment. We found that levels of Ser5P and Ser2P, but not Ser7P,slightly increased in 1 μM naphthaleneacetic acid-treated WTcompared with the control, but ROP2RNAi rop4-1 did not exhibitsuch activation (SI Appendix, Fig. S5). This result shows thatelevation of Ser5P and Ser2P by auxin is dependent upon func-tional ROP2/4 GTPases, consistent with our finding that ROP2/4signaling modulates the specific Pol II CTD code.

CPL1 and CPL2 Have a Similar Function in CTD Phosphorylation andAct Redundantly in Cell Growth and Morphogenesis. Interestingly,comparison of the CTD phosphorylation status in CA1-1 andcae2-1 showed that cae2-1 has slightly lower Ser5P and Ser2Plevels than CA1-1 (Fig. 1 E and F). This might explain why cae2-1has not exhibited cellular phenotypes observed in CA1-1. Asfry2-1 has not shown any cotyledon pavement cell phenotypeeither, it is possible that CPL2, another CTD phosphatase ho-mologous to CPL1, has a redundant function with CPL1 andthat activation of ROP2 signaling suppresses both CPL1 and itsclose homolog. Earlier studies reported that CPL2, a close ho-molog of CPL1, plays a partially redundant role with CPL1, andthat their double mutant is lethal (22, 31). To test the functionalredundancy possibility, we characterize CPL2. Consistent withthe earlier reports, we found that, like cae2-1, cpl2-2 was alsoable to enhance the CA1-1 cell-shape phenotype after it wascrossed into CA1-1 (SI Appendix, Fig. S6A). Thus, the furtherincrease of Ser5P levels in cae2-1 CA1-1 compared with CA1-1(Fig. 1 E and F) could also be caused by the possibility that acti-vation of ROP2 signaling suppresses both CPL1 and CPL2. We

E8198 | www.pnas.org/cgi/doi/10.1073/pnas.1605871113 Zhang et al.

Dow

nloa

ded

by g

uest

on

July

19,

202

0

then constructed a homozygous double-mutant cae2-1 cpl2-2 andfound that it exhibited severe growth inhibition and fertility re-duction (SI Appendix, Fig. S6B). When the double mutant andcae2-1, which did not affect fertility, were crossed, only two or threeseeds were obtained in each silique using cae2-1 cpl2-2 as female

and no seeds were obtained from the reciprocal cross, indicatingthat male fertility was more severely compromised than female.However, at the end of the life cycle, a few seeds derived fromself-fertilization of each cae2-1 cpl2-2 plant were recoveredand subsequently used for biochemical analysis and cell-shape

Fig. 1. Characterization of the cae2-1 CA1-1 enhancer mutant and alteration of Ser5P and Ser2P levels in the CPL1 and ROP2 mutants. (A) Cotyledon pavement cellshapes in CA1-1, cae2-1, and cae2-1 CA1-1. WT, wild type. (B) Quantitative analysis of cell-shape factor (4π×area/perimeter2). Values are means ± SD (n = 60 cells fromfour plants), and different letters indicate a statistical difference with P < 0.001 (ANOVA). (C) Arabidopsis At4g21670 (CAE2/CPL1) gene structure and the location ofthemutation site in cae2-1. (D) RT-PCR analysis of CAE2/CPL1 expression inWT, CA1-1, and cae2-1 CA1-1. There was no difference in CPL1 expression level betweenWTand CA1-1, but three splicing products (CAE2-A, CAE2-B, and CAE2-C) were PCR-amplified using primer pairs CZP17 and CZP18 in cae2-1 CA1-1.ACT2 (Actin2) was usedas an internal control. (E and F) Alteration of Ser5P and Ser2P inWT, CA1-1, cae2-1, and cae2-1 CA1-1 (E) and the quantitative analysis (F). The proteins were analyzedby Western blot using Ser5P-, Ser2P-, and Ser7P-specific and anti-CTD antibodies. Tubulin was used as the protein-loading control detected with an anti-tubulinantibody. WT, wild type. (G and H) Alteration of Ser5P and Ser2P in fry2-1 (G) and the quantitative analysis (H). (I and J) Reduced Ser5P and Ser2P in ROP2RNAi rop4-1(I) and the quantitative analysis (J). Values in F, H, and J are means ± SD (n = 3 biological replicates) with different letters within the same category indicatinga statistical difference (P < 0.05; pair-wise t test).

Zhang et al. PNAS | Published online November 28, 2016 | E8199

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

19,

202

0

characterization. We found that, similar to cae2-1, cpl2-2 alsohad higher Ser5P and Ser2P levels than WT without affectingSer7P and total RPB1 levels and that cae2-1 cpl2-2 had thehighest levels of Ser5P and Ser2P (Fig. 2 A and B). Similar tocae2-1, cotyledon pavement cell shape, cell size, and cell numberin cpl2-2 were indistinguishable from WT, but cae2-1 cpl2-2exhibited a phenotype in all three growth and geometric param-eters (Fig. 2 C and D; SI Appendix, Fig. S7). Taken together, ourresults suggest that CPL1 and CPL2 are similarly involved in CTDSer5 and Ser2 dephosphorylation and likely act redundantly incotyledon pavement cell growth and morphogenesis.

ROP2 Signaling Inhibits CPL1 Activity by Promoting CPL1 Degradation.To investigate how ROP2 signaling regulates CPL1 function in thecontrol of CTD Ser5 and Ser2P phosphorylation, we first testedwhether CPL1 overexpression could suppress the elevated Ser5Pand Ser2P levels and cell-shape defect in CA-rop2 transgenic plants.To minimize transgene silencing, we used different promoters todrive the expression of CA-rop2 (UBQ10 promoter; designatedUBCA1) and CPL1 (CaMV 35S; designated 35S:CPL1). Indeed, wedid not observe any impact in CA-rop2 transgene expression intransgenic plants overexpressing CPL1 (SI Appendix, Fig. S3B).Interestingly, we found that 35S:CPL1 in UBCA1 decreased Ser5Pand Ser2P to a level close to that inWT without affecting Ser7P andtotal RPB1 levels (Fig. 3 A and B), consistent with our findings that

both Ser5P and Ser2P levels were higher in the cpl1 mutants (Fig. 1E–H; Fig. 2 A and B). Furthermore, overexpression of CPL1 inUBCA1 strongly, although not fully, restored the interdigitatingcell-shape defect caused by CA-rop2 overexpression (Fig. 3C).These results support the notion that CPL1 may act downstream ofROP2 signaling.We then tested whether ROP2 signaling inhibits CPL1 transcrip-

tion or protein stability. Gene expression analysis revealed that CA1-1and WT had a similar CPL1 mRNA level (Fig. 1D; SI Appendix,Fig. S8A). To determine whether ROP2 signaling affects CPL1protein stability, we generated transgenic lines expressing Flag-and Strep II-tagged CPL1 (designated CPL1-FAST). Expressionof 35S:CPL1-FAST strongly suppressed the UBCA1 cell-shapephenotype (SI Appendix, Fig. S8B), like 35S:CPL1 (Fig. 3C),indicating that the FAST tags did not interfere with the CPL1function. Using an in vitro protein degradation assay, we found thatCPL1-FAST was more rapidly degraded in CA1-1 extracts at 45 minthan that in WT (Fig. 3 D and E). At 90 min, whereas WT extractsshowed a substantial degradation of CPL1-FAST, it was almostcompletely degraded in CA1-1. However, addition of protein deg-radation inhibitors MG132 and lactacystin for 1.5 h reduced theCPL1-FAST degradation, indicating a potential involvement of theproteasome-mediated degradation mechanism. Taken together, ourgenetic and biochemical evidence suggests that ROP2 GTPasesignaling suppresses CPL1 activity by promoting its degradation.

Alteration of ROP2 and CPL1 Activities Impacts Expression of a CommonSubset of Genes.Given the role of CTD Ser5 phosphorylation in thePol II control of gene expression (18, 19, 22–24), we performedtranscriptome analysis to determine whether ROP2 and CPL1 af-fected expression of a similar set of downstream genes. RNA-sequencing analysis revealed that 1,333 genes were differentiallyexpressed (with a twofold cutoff) in cae2-1, CA1-1, or cae2-1 CA1-1compared with WT (SI Appendix, Fig. S9A). Approximately 43% ofthose differentially expressed genes in CA1-1 were also affected incae2-1 (SI Appendix, Fig. S9A), suggesting that a sizable commonsubset of genes was similarly controlled by CPL1 and ROP2.Furthermore, expression of a subset of CA1-1–affected genes wasfurther enhanced by the CPL1 mutation in cae2-1 CA1-1 (SI Ap-pendix, Fig. S9B). In addition, 49 genes, the products of which arepredicted to localize to the cell wall or be involved in cell-wallmodification, were also differentially expressed in these genotypes(SI Appendix, Fig. S9C). The cell wall has been recognized to play acritical role in cell-shape formation or ROP signaling (11, 32, 33).Consistently, we found that some of the cell-wall–related genes thatwere up-regulated in CA1-1 and/or cae2-1 were further enhancedin cae2-1 CA1-1, including those encoding pectin methylesterase 17(PME17), wall-associated kinase 1 (WAK1), and a putative invertase(At2g47550) (SI Appendix, Fig. S9C). The expression pattern ofthese three genes revealed by RNA sequencing was validated usingquantitative RT-PCR analysis (SI Appendix, Fig. S9D). Importantly,up-regulation of these genes by activated ROP2 was inhibited in the35S:CPL1 transgenic line (SI Appendix, Fig. S9E). Together, theseresults support that ROP2 signaling modulates the CTD code intranscriptional control by inhibiting CPL1.

Activation of Pol II CTD Phosphorylation by Rho GTPase Is Conservedin Yeast. The finding in Arabidopsis that ROP2 signaling modu-lates the CTD phosphorylation led us to examine whethermodulation of the CTD code by Rho GTPases is evolutionarilyconserved. This was motivated by early observations that RasGTPase signaling was implicated in modulating Pol II in yeast(34) and rat cell culture (35). In budding yeast (Saccharomycescerevisiae), the Ras-PKA signaling pathway does not directlytarget Rpb1 CTD; instead, PKA directly phosphorylates Srb9, acomponent of Pol II holoenzyme (34). To test the hypothesis thatyeast may use Rho rather than Ras to modulate CTD phos-phorylation, we first used the fission yeast Schizosaccharomyces

Fig. 2. CPL1 and CPL2 have a similar function in CTD phosphorylation and actredundantly in cell morphogenesis. (A) Alteration of Ser5P and Ser2P in WT,cae2-1, cpl2-2, and cae2-1 cpl2-2. Tubulin, the protein loading control detectedwith anti-tubulin antibody. WT, wild type. (B) Quantitative analysis of Ser5Pand Ser2P in various genotypes. Values are means ± SD (n = 3 biologicalreplicates) with different letters indicating a statistical difference (P < 0.05;pair-wise t test). (C) Cotyledon pavement cells in WT, cae2-1, cpl2-2, and cae2-1cpl2-2. Representative cells were highlighted. (D) Quantitative analysis of cell-shape factor. Values are means ± SD (n = 60 cells from four plants), and dif-ferent letters indicate a significant difference (P < 0.001; ANOVA).

E8200 | www.pnas.org/cgi/doi/10.1073/pnas.1605871113 Zhang et al.

Dow

nloa

ded

by g

uest

on

July

19,

202

0

pombe, which shows a clear cell-shape change by activation of RhoGTPases such as Rho4 and Cdc42 (36–38). We expressedArabidopsisCA-rop2 with the thiamine-repressible nmt1 promoter in yeast cells.In the presence of thiamine, the CA-rop2 transformants did not alterthe normal rod shape, but after 48 h of removing thiamine theCA-rop2 transformants had acquired the bulbous shape that wasnot observed in the vector control (Fig. 4A). This indicates thatconstitutive activation of Arabidopsis ROP2 disrupted polarizedgrowth in yeast, as with similar yeast Rho1 and Cdc42 gain-of-function mutants (36, 39). CA-rop2 induction increased bothSer5P and Ser2P levels without altering the yeast Rpb1 level(Fig. 4 B and C). Furthermore, we found a dramatic increase inSer2P and Ser5P levels after only 1 h of CA-rop2 induction, atwhich time no morphological change in yeast cells could be ob-served (SI Appendix, Fig. S10 A–C). This suggests that the stim-ulation of CTD phosphorylation by ROP2 signaling is more likelya direct effect rather than an indirect consequence of morpholog-ical changes. To substantiate that Rho-GTPase-signaling–mediatedCTD phosphorylation is conserved in fission yeast, we used cdc42-1625, a loss-of-function Cdc42 mutant (40) that exhibited abnor-mal cell shapes (Fig. 4D). cdc42-1625 had greatly reduced Ser5pand Ser2P levels (Fig. 4 E and F), supporting that yeast Rhosignaling also modulates CTD phosphorylation in the control ofcell morphogenesis.We then investigated whether plants and yeast use a conserved

biochemical mechanism in modulating the CTD code by inhibiting

CTD phosphatase. Because the antibody for Saccharomycescerevisiae Fcp1, which acts as a Ser2 phosphatase (41) and is ho-mologous to CPL1 and likely a CPL3 ortholog (21), is available,budding yeast was analyzed. The Fcp1 protein level was lower inyeast transformants expressing CA-rop2 than in the vector controlafter 12 h of induction by galactose, and at the same time, theSer2P level increased (SI Appendix, Fig. S10 D and E). To betterdetect Fcp1, a budding yeast strain with the 3xFlag tags insertedin-frame into the C terminus of Fcp1 was used. We found thatwhen protein degradation inhibitors MG132 and lactacystin wereused, the Fcp1-3xFlag protein level in the CA-rop2–transformedstrain was recovered, and concomitantly the Ser2P level decreased(Fig. 4 G and H). Ser5P was similarly affected, consistent with anearlier finding that the fcp1 mutant also impacted Ser5 phos-phorylation (42). These results strongly suggest that a similar CTDphosphatase degradation mechanism operates in both Arabidopsisand yeast systems to regulate the Rho-GTPase-signaling–mediatedmodulation of Pol II CTD phosphorylation.Furthermore, we investigated whether induction of CA-rop2 in

fission yeast also mis-regulated expression of genes known tofunction in cell-shape formation. We found that CA-rop2 induc-tion in yeast up-regulates (threefold) Rho1 and down-regulates(eightfold) Rga1 gene expression (SI Appendix, Fig. S10F). Rga1encodes a GTPase-activating protein that negatively regulatesRho1 activity, and rga1 null cells showed a swollen, multiseptatedor branched shape (43), a phenotype similar to the cell expressing

Fig. 3. ROP2 inhibits CPL1 activities in the control of CTD phosphorylation and cell-shape formation. (A) CPL1 overexpression suppresses Ser5 and Ser2 CTDphosphorylation in UBCA1. WT, wild type; Tubulin, the protein loading control. (B) Quantitative analysis of Ser5P and Ser2P in CPL1 overexpression plants.Values are means ± SD (n = 3 biological replicates) with different letters within the same category indicating a statistical difference (P < 0.05; pair-wise t test).(C) CPL1 overexpression suppresses the UBCA1 phenotypes in pavement cell shape. Quantitative analysis of cell-shape factor was shown with values rep-resenting means ± SD (n = 60 cells from four plants) and P < 0.001 (ANOVA) for all pair-wise comparisons. (D) Enhanced CPL1 in vitro degradation in CA1-1.CPL1-FAST was purified from transgenic plants and mixed with an equal amount of protein extracts from WT and CA1-1 (as indicated by the RuBisCo loadingcontrol) in the presence of 1 mM ATP with or without the protein degradation inhibitors MG132 and lactacystin for the indicated time and then was detectedby an anti-Flag antibody. (E) Quantitative analysis of CPL1 degradation. Values are means ± SD (n = 3 biological replicates) with different letters for WT andCA1-1, respectively, indicating a statistical difference (P < 0.05; pair-wise t test).

Zhang et al. PNAS | Published online November 28, 2016 | E8201

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

19,

202

0

a constitutively active form of Rho1 (39). The Cdc42-Rac-Rho–signaling cascade has been reported (44), and our result indicatesthat ROP2 activation in fission yeast may up-regulate Rho1 at thetranscriptional level. In addition, we found that expression ofPmc5 was 10-fold lower in CA-rop2 transformants (SI Appendix,

Fig. S10F). Given that Pmc5 is a transcriptional regulator, alsocalled Med6 (mediator 6), and that pmc5 null cells are temperature-sensitive stubby, curved, septated, and misshapen (45), suppressionof Pmc5 expression by CA-rop2 induction indicates that Pmc5may be another transcriptional target regulated by activation of

Fig. 4. Regulation of CTD phosphorylation by Rho GTPase in yeast. (A) CA-rop2 affects cell shape in fission yeast. Vector, the vector only control; +VB1,uninduced (presence of thiamine/VB1); −VB1, induced (absence of thiamine/VB1) for 48 h. (B) CA-rop2 increases Ser5P and Ser2P in fission yeast. Tubulin, theprotein loading control. (C) Quantitative analysis of Ser5P and Ser2P in CA-rop2 transformants. (D) Yeast cdc42-1625 mutant shows abnormal cell shapes. WT,wild type. (E) cdc42-1625 mutant reduces Ser5P and Ser2P of Rpb1 CTD. (F) Quantitative analysis of Ser5P and Ser2P in cdc42-1625. (G) CA-rop2 promotesFcp1-3xFlag protein degradation in vivo. Cells were grown in the absence (−) or presence (+) of the protein degradation inhibitors MG132 and lactacystin for16 h of inducible expression of CA-rop2 by galactose in budding yeast. (H) Quantitative analysis of Fcp1-3xFlag degradation and Ser5P and Ser2P levels inCA-rop2 transformants. For C, F, and H, values are means ± SD (n = 3 biological replicates) with different letters within the same category indicating astatistical difference (P < 0.05; pair-wise t test).

E8202 | www.pnas.org/cgi/doi/10.1073/pnas.1605871113 Zhang et al.

Dow

nloa

ded

by g

uest

on

July

19,

202

0

ROP2 in fission yeast. These gene expression analyses indicatethat Rho GTPase signaling alters transcription of yeast cell-shape–related genes.Finally, we tested whether Pol II Rpb1 CTD phosphorylation

is important for the fission yeast cell-shape change induced byCA-rop2. Fission yeast has four imperfect heptad repeats fol-lowed by 25 consensus heptad repeats, and studies have shownthat retaining the four imperfect repeats plus four consensusrepeats will lead to lethality (46, 47). Therefore, several studiestoward understanding the CTD code in fission yeast have reliedon changing the amino acid residues in 12–14 consensus repeats(47–49). We obtained the strains carrying S2A (all Ser2 in 14repeats changed to Ala), (S5)3(S5A)11 (Ser5 in 11 repeats ischanged to Ala while it remains the same in 3 other repeats), and(S5)2(S5A)12 (Ser5 in only 12 repeats is changed to Ala) muta-tions and their WT (48, 49) for transformation with the nmt1promoter driven CA-rop2. These mutants had a very similarshape to their WT when CA-rop2 was not induced in the pres-ence of VB1 (Fig. 5). Similar to an earlier observation (Fig. 5A),the CA-rop2 induction caused the WT cells to develop into theoval or round shapes; however, S2A mutants partially suppressedthe CA-rop2 effect, with many cells retaining the rod shape al-though being slightly shorter than the uninduced S2A cells. Al-though (S5)3(S5A)11 mutant cells were similar to S2A, (S5)2(S5A)12mutants showed almost complete suppression except the occur-rence of slightly swollen cells compared with the uninduced control(Fig. 5). It is unlikely that the suppression of the CA-rop2 cell-shapephenotype was caused by inhibition of CA-rop2 expression becausethese S2A and S5A mutants had similar CA-rop2 protein levels asWT when induced in the absence of thiamine (SI Appendix, Fig.S10G). Therefore, this genetic interaction result suggests that CTDS2P and S5P are important targets for constitutively activatedROP2 signaling in cell-shape formation, with S5P playing a pre-dominant role. Taken together, our genetic, biochemical, and geneexpression evidence shows that Rho GTPase signaling in yeasttargets the Pol II CTD code using an evolutionarily conserved CTDphosphatase degradation mechanism.

DiscussionRho GTPase signaling and the Pol II CTD code modulationhave been separately studied in yeast, plants, and animals formore than 2 decades. We present genetic and biochemical evi-dence in Arabidopsis to establish a surprising link between thesetwo key molecules in intracellular signaling. In addition, we havefound that Cdc42 GTPase signaling in yeast similarly modulatesPol II CTD Ser5 and Ser2 phosphorylation. Therefore, our re-sults suggest that modulation of the Pol II CTD code by RhoGTPase signaling represents an evolutionarily conserved mech-anism in intracellular signaling. This has led us to propose a

model of the Rho GTPase modulation of the Pol II CTD code ineukaryotes (Fig. 6). In this model, a positive cue, such as auxin inplants, activates ROP2 signaling, leading to inhibition of CPL1,which in turn relieves its inhibition on CTD phosphorylation.Consequently, CTD becomes hyperphosphorylated at Ser5 andSer2 sites. Such a CTD code modulation can quickly and stronglyaffect expression of those genes important for cell growth andshape formation. An important future direction is to understandhow Rho GTPase signaling inhibits CPL1 function. One of theinhibitory mechanisms could be promotion of CPL1 protein deg-radation. Given that ROP has been shown to promote degrada-tion of auxin-signaling–related proteins such as AUX in thenucleus (7), proteasome-based protein degradation could be an-other common mechanism in ROP signaling. Indeed, we haveshown that activation of ROP2 signaling could lead to enhanceddegradation of a CPL1 homolog in yeast (Fcp1) in vivo, indicatingthat yeast and plant cells likely employ a conserved biochemicalmechanism to regulate the function of CTD phosphatases in Rhosignaling. Future work toward understanding how Rho signalingmediates the proteasome-mediated CPL1 or Fcp1 degradationwill fill in the gaps of this Rho GTPase-Pol II signaling pathway.Our unexpected finding that links Rho GTPase signaling to

the Pol II CTD code modulation has broad implications in thecontrol of cell morphogenesis or cell growth in plants and othereukaryotes. In animals, regulation of Pol II by distinct signalingpathways (50) or its role in embryo development (51, 52) hasbeen reported, but no Rho GTPase has been implicated in modu-lating Pol II activities. In plants, Rho signaling has been demon-strated to control cytoskeletal organization in cell morphogenesis,but no transcriptional regulator has been identified in plant cell-shape formation. Thus, our finding that ROP2 signaling can alsolead to Pol II CTD modulation is exciting. However, currently, wecannot distinguish whether cell shape, cell size, and number ofphenotypes observed in CA1-1, cae2-1 CA1-1, and cae2-1 cpl2-2 aredirectly or indirectly caused by the changes in the Pol II CTDphosphorylation status. After controlling for the cell-size factor,cae2-1 CA1-1 still showed a statistically significant difference inShape Factor compared with WT (SI Appendix, Fig. S1B), indicat-ing that the cell-shape phenotype of cae2-1 CA1-1 cannot becompletely explained by the cell-size effect. Thus, although at pre-sent we cannot distinguish whether the pavement cell-shape alter-ation in cae2-1 CA1-1 is the cause or the consequence of cellnumber and/or size change due to the complex interactions betweencell division, expansion, and morphogenesis (38, 53, 54), it is pos-sible that cae2-1 and CA1-1 impact both a common cell size andshape formation pathway and another pathway in the control of cellshape unrelated to cell size. In addition, it remains unknownwhether CPL1 or Pol II also indirectly alters cytoskeleton organi-zation or vesicle transport in the cotyledon cell growth and shape

Fig. 5. Fission yeast Rpb1 CTD S2A and S5A mutants suppress the cell-shape phenotype caused by CA-rop2 induction. Yeast mutants carrying S2A and S5Amutations and transformed with the inducible CA-rop2 construct were grown for 48 h. WT, wild type; +VB1, uninduced (presence of thiamine/VB1); −VB1,induced (absence of thiamine/VB1).

Zhang et al. PNAS | Published online November 28, 2016 | E8203

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

19,

202

0

formation process. However, consistent with a critical role of Pol IIin transcriptional control, we found a dramatic transcriptomicchange in the cae2-1 CA1-1 enhancer mutant and the suppressionof several cell-wall–related genes by CPL1 overexpression. Thisfinding indicates a potentially important role for Pol II transcrip-tion in cell morphogenesis and growth. In corroboration with theimportance of transcriptional control in ROP2-mediated cellmorphogenesis, we found that expression of three cell-shape–relatedgenes in fission yeast (Rho1, Rga1, and Pmc5) was altered by in-ducible expression of CA-rop2. Therefore, if an essential role forRho-Pol II signaling pathway-exerted gene expression can be dem-onstrated in cell-shape determination, it will significantly expand themechanistic understanding of cell growth and geometry determina-tion beyond the well-studied cytoskeletal control.It should also be noted that the Pol II CTD code modulation

as measured by Western blots using Ser2P- and Ser5P-specificantibodies does not account for all of the cell geometric differ-ences observed in CA1-1, cae2-1 CA1-1, and cae2-1 cpl2-2. Forexample, cae2-1 has slightly higher Ser5P and Ser2P levels thanWT, but they are not sufficient to cause a cell-shape change; alarger increase in Ser5P and Ser2P in CA1-1 leads to a dramaticchange in cell shape, and a further increase in Ser5P and Ser2Plevels in cae2-1 CA1-1 causes the most dramatic cell-shapechange (Fig. 1 A and B; SI Appendix, Fig. S1). This stronglysuggests that the highly phosphorylated CTD is the cause for the

cell-shape change. However, cae2-1 cpl2-2 has even higher Ser5Pand Ser2P levels than CA1-1, but the double mutant does notshow a stronger cell-shape change than CA1-1. Although thedouble-mutant cae2-1 cpl2-2 has similar Ser5P and Ser2P levelsand cell sizes as cae2-1 CA1-1, they still exhibited different cellshapes than the enhancer cae2-1 CA1-1. This discrepancy could bedue to the fact that ROP2 and CPL1/2 impact on slightly differentdistributions of the phosphorylated Ser5 and/or Ser2 in variousheptad peptide repeats within the CTD. In fission yeast, thecombination of Ser2 and Ser5 phosphorylation status in the CTDcode theoretically would lead to 4n distinct primary structures,where n is the number of heptad repeats, and thus “reading” thislarge number of distinct and complex sets of the CTD codebecomes critical in relaying upstream biological information totranscriptional control (48, 49). Our finding that the (S5)2(S5A)12mutant almost completely suppressed the cell-shape phenotypecaused by the CA-rop2 induction and the S2A and (S5)3(S5A)11mutants exhibited only partial suppression suggests that thenumber of the Ser5-phosphorylated heptad repeats in the CTD iscritical for Rho-GTPase–mediated cell-shape control. Therefore,it is necessary to determine whether Rho signaling modulatesdistinct CTD codes in each heptad repeat dynamically in responseto both external factors and internal cues. Alternatively or addi-tionally, the discrepancy could be due to the possibility that ROP2signaling also targets other factors. For example, CTD kinases(55) might be potential targets, as we have found that over-expression of CPL1 could not fully, although it could strongly,restore the cell-shape defect in UBCA1. Other CPL1 substrates orinteracting proteins, such as HYL1 (22) and the nonsense-mediateddecay factors eIF4AIII and UPF3 (56), might also be regulated byROP2 signaling. Therefore, it is important to investigate all of thesepossibilities for a better understanding of how Rho GTPase regu-lates gene expression.Our findings in yeast and plant systems on the modulation of

the CTD code by Rho signaling render a strong support for anevolutionarily conserved “shortcut” model in the control of PolII transcription for both unicellular and multicellular eukaryotes.Prior work in yeast showing the interaction of Snf1 kinase orRas-PKA signaling with components of the Srb/Mediator com-plex has led to the proposal of a shortcut, “bypass,” or directmodel in transcriptional control (34, 57). Compared with theclassical transcriptional activator/repressor-mediated indirectmodel, this shortcut or direct model, by targeting the Pol IIholoenzyme, has been argued to provide an efficient control oftranscription to rapidly bring about the broad changes in geneexpression (34, 57). Our work suggests that eukaryotes likelyshare a transcriptional regulatory mechanism by which Rhosignaling targets the Pol II CTD code itself. Nevertheless, theevidence that Rho signaling directly activates a nuclear protea-some pathway is needed to demonstrate the Rho-CTD shortcutmodel. There are at least two possibilities for such regulation. Oneis a Rho-signaling molecule that can directly phosphorylate CTD,such as PKA-mediated phosphorylation of the Srb/Mediatorcomplex in yeast Ras signaling (34), or that can translocate intothe nucleus to activate the CPL1 or Fcp1 degradation. The otherpossibility is that Rho itself could be translocated into the nucleusand acts upon the proteasome to degrade CTD phosphatases. Itshould be noted that such nuclear localization or nucleocytoplasmicshuttling has been reported for Rac1 GTPase (58). Clearly, oneimportant future direction is to dissect the mechanism by whichRho signaling directly exerts its effect in nuclear degradation ofCTD phosphatases or regulation of CTD kinases. If the proposedRho-Pol II shortcut model of transcriptional control can be con-vincingly demonstrated, it will shed light on the complex, and yetefficient cellular regulatory systems such as this model will enableeukaryotes to rapidly respond to various internal and external cuesduring growth and development, thereby increasing the environ-mental fitness of eukaryotes.

Fig. 6. A model of Rho GTPase modulation of the Pol II CTD code viainhibiting CTD phosphatases. In response to signals (such as plant hormoneauxin), Rho GTPases (such as ROP2 in Arabidopsis or Cdc42 in yeast) are ac-tivated, leading to the stimulation of proteasome activity and, consequently,the degradation of CTD phosphatases (such as CPL1 and Fcp1). This leads toan increase of CTD Ser5 and Ser2 phosphorylation levels, and thus the CTDcode is modulated, impacting Pol II transcription and ultimately affecting cellgrowth and morphogenesis.

E8204 | www.pnas.org/cgi/doi/10.1073/pnas.1605871113 Zhang et al.

Dow

nloa

ded

by g

uest

on

July

19,

202

0

Materials and MethodsExperimental details are provided in SI Appendix, SI Materials and Methods.

Plant Materials and Growth. Arabidopsis thaliana Columbia WT and cpl2-2(SALK_059753) were obtained from The Arabidopsis Information Resource.For cotyledon cell-shape and protein degradation analyses, soil-grownseedlings were used if not otherwise specified. For protein analysis, liquid-cultured seedlings were used. For transgenic plants first described in thispaper, homozygous plants of T3 or T4 generations were used in most phe-notypic characterizations and biochemical or gene expression studies. Atleast four independent homozygous lines were used for initial Western blotand phenotypic studies, and one of these that showed similar results toother lines was chosen as a representative for presentation in all figures.

Map-Based Cloning and Complementation Test. cae2-1 CA1-1 was crossed toecotype Landsberg erecta for mapping the CAE2 gene. To complement thecae2-1CA1-1 phenotype, a 7.7-kb CPL1 genomic fragment was transformedinto cae2-1 CA1-1.

RNA-Sequencing Data Analysis. Total RNA was extracted from shoots of 7-d-old seedlings of WT, CA1-1, cae2-1, and cae2-1 CA1-1, each genotype withtwo biological replicates. RNA sequencing was performed via IlluminaHiSeqTM 2000 and raw data (deposited in the National Center for Bio-technology Information Gene Expression Omnibus database; accession no.GSE78118) were preprocessed by Beijing Genomics Institute Tech SolutionsCo. Differentially expressed genes between two genotypes were identifiedusing the exact test for negative binomial models implemented in EdgeR(59). Statistically significant genes had at least twofold changes and a falsediscovery rate value less than 0.1.

Plasmid Construction. The plant expression vectors for overexpressing CA-rop2(UBCA1; GZ47), CPL1 (BZ01), and CPL1-FAST (BZ02) and for inducible expres-sion of CA-rop2 in yeast (BZ04 and BZ06) were constructed by PCR-basedcloning as described in detail in SI Appendix, SI Materials and Methods.

Regular and Quantitative RT-PCR. RT-PCR analysis of the Arabidopsis CAE2 tran-script and quantitative real-time PCR analysis of PME17, WAK1, and At2g47550/invertase genes were performed using gene-specific primers (SI Appendix, TableS1). For quantitative PCR analysis, gene expression values were normalized toACT2. For quantitative PCR analysis of fission yeast Rho1, Rga1, and Pmc5 ex-pression, primers were included in SI Appendix, Table S1, and normalization wasdescribed in SI Appendix, SI Materials and Methods.

Western Blot. Seedlings were grounded in liquid nitrogen to extract totalprotein. For yeast protein experiments, cells were collected and mechanically

disrupted with glass beads. Standard Western blot procedure was used,and proteins were detected using Ser2P-, Ser5P-, and Ser7P-specific anti-bodies and anti-Rpb1 and anti-tubulin antibodies (SI Appendix, SI Materialsand Methods). Results were visualized by the chemiluminescent method(Thermo, 34096).

Protein Degradation. For in vitro degradation, protein purified from 35S:CPL1-FAST transgenic plants was mixed with an equal amount of crude extracts ofWT and CA1-1, respectively, in the presence of 1 mM ATP with and withoutthe proteasome inhibitors MG132 and lactacystin. After incubation, proteinswere detected using an anti-Flag antibody. For in vivo Fcp1 protein degra-dation in yeast, nuclear extracts from the budding yeast transformantscarrying the CA-rop2 expression vector were detected using an anti-Fcp1antibody or an anti-Flag antibody for the 3xFlag-tagged Fcp1 strain with orwithout MG132 and lactacystin.

Cotyledon Cell Imaging and Quantitative Analysis. Images of cotyledon cellswere collected using either agarose gel imprinting (60) or an ethanol/aceticacid treatment-based method (30). The Shape Factor was determined usingthe equation (4π × area/perimeter2) as described (29). Statistical analysis wasperformed using one-way ANOVA.

Yeast Materials, Growth, and Transformation. The S. pombe cdc42-1625, rpb1-S2A, (S5)3(S5A)11, and (S5)2(S5A)12 mutants and their corresponding WTstrains were cultured in YPAD (yeast extract, peptone, adenine, and dex-trose) medium or Edinburgh minimal medium (EMM). The S. cerevisiaeW303-1B WT and the Fcp1-3xFlag strains were cultured in YPAD medium orSC (Synthetic Complete) minimal medium. The electroporation method wasused for yeast transformation. Cell shape and Rpb1 CTD S2P and S5P levelswere analyzed in the transformants or mutants.

ACKNOWLEDGMENTS. We thank Dr. Zhenbiao Yang for insightful discus-sion of this project; Drs. Thomas Leustek and Alice Cheung for their criticalcomments on the manuscript; Dingquan Huang for extraction of RNA sam-ples used for transcriptome profiling; Dr. Yuh-Ru J. Lee for help with the cell-shape phenotypic validation; Dr. Shanjin Huang for help with a cell-clearingtechnique; Drs. Yong Ding and Michael Fromm for advice on the Pol II an-tibody selection; Dr. Rongmin Zhao for suggestion on the use of proteasomeinhibitors; Drs. Yiji Xia for providing the FAST vector; Dr. Liming Xiong forfry2-1; Drs. Jian-Qiu Wu and Sophie G. Martin for fission yeast cdc42-1625;Dr. Jack Greenblatt for Fcp1 antibody; Drs. Maria Aristizabal and MichaelKorbo for the Fcp1-3xFlag–tagged budding yeast strain; and Drs. StewartShuman and Beate Schwer for the fission yeast rpb1-S2A and S5A mutantstrains. This work was initially supported by National Institute of General MedicalSciences Grant 3S06GM008225-20S1 (to Z.-L.Z.) and partly supported by Na-tional Science Foundation Grant IOS-1121551 (to Z.-L.Z.).

1. Jaffe AB, Hall A (2005) Rho GTPases: Biochemistry and biology. Annu Rev Cell Dev Biol

21:247–269.2. Yang Z (2008) Cell polarity signaling in Arabidopsis. Annu Rev Cell Dev Biol 24:

551–575.3. Craddock C, Lavagi I, Yang Z (2012) New insights into Rho signaling from plant ROP/

Rac GTPases. Trends Cell Biol 22(9):492–501.4. Yang Z (2002) Small GTPases: Versatile signaling switches in plants. Plant Cell

14(Suppl):S375–S388.5. Tao LZ, Cheung AY, Wu HM (2002) Plant Rac-like GTPases are activated by auxin and

mediate auxin-responsive gene expression. Plant Cell 14(11):2745–2760.6. Xin Z, Zhao Y, Zheng ZL (2005) Transcriptome analysis reveals specific modulation of abscisic

acid signaling by ROP10 small GTPase in Arabidopsis. Plant Physiol 139(3):1350–1365.7. Tao LZ, Cheung AY, Nibau C, Wu HM (2005) RAC GTPases in tobacco and Arabidopsis

mediate auxin-induced formation of proteolytically active nuclear protein bodies that

contain AUX/IAA proteins. Plant Cell 17(8):2369–2383.8. Lavy M, et al. (2007) A novel ROP/RAC effector links cell polarity, root-meristem

maintenance, and vesicle trafficking. Curr Biol 17(11):947–952.9. Jeon BW, et al. (2008) The Arabidopsis small G protein ROP2 is activated by light in

guard cells and inhibits light-induced stomatal opening. Plant Cell 20(1):75–87.10. Xu T, et al. (2010) Cell surface- and rho GTPase-based auxin signaling controls cellular

interdigitation in Arabidopsis. Cell 143(1):99–110.11. Oda Y, Fukuda H (2012) Initiation of cell wall pattern by a Rho- and microtubule-

driven symmetry breaking. Science 337(6100):1333–1336.12. Xu T, et al. (2014) Cell surface ABP1-TMK auxin-sensing complex activates ROP GTPase

signaling. Science 343(6174):1025–1028.13. Fu Y, Gu Y, Zheng Z, Wasteneys G, Yang Z (2005) Arabidopsis interdigitating cell

growth requires two antagonistic pathways with opposing action on cell morpho-

genesis. Cell 120(5):687–700.14. Hsin JP, Manley JL (2012) The RNA polymerase II CTD coordinates transcription and

RNA processing. Genes Dev 26(19):2119–2137.

15. Egloff S, Dienstbier M, Murphy S (2012) Updating the RNA polymerase CTD code:Adding gene-specific layers. Trends Genet 28(7):333–341.

16. Hajheidari M, Koncz C, Eick D (2013) Emerging roles for RNA polymerase II CTD inArabidopsis. Trends Plant Sci 18(11):633–643.

17. Eick D, Geyer M (2013) The RNA polymerase II carboxy-terminal domain (CTD) code.Chem Rev 113(11):8456–8490.

18. Koiwa H, et al. (2002) C-terminal domain phosphatase-like family members (AtCPLs)differentially regulate Arabidopsis thaliana abiotic stress signaling, growth, and de-velopment. Proc Natl Acad Sci USA 99(16):10893–10898.

19. Xiong L, et al. (2002) Repression of stress-responsive genes by FIERY2, a novel tran-scriptional regulator in Arabidopsis. Proc Natl Acad Sci USA 99(16):10899–10904.

20. Jin YM, et al. (2011) AtCPL5, a novel Ser-2-specific RNA polymerase II C-terminaldomain phosphatase, positively regulates ABA and drought responses in Arabidopsis.New Phytol 190(1):57–74.

21. Li F, et al. (2014) Modulation of RNA polymerase II phosphorylation downstream ofpathogen perception orchestrates plant immunity. Cell Host Microbe 16(6):748–758.

22. Manavella PA, et al. (2012) Fast-forward genetics identifies plant CPL phosphatases asregulators of miRNA processing factor HYL1. Cell 151(4):859–870.

23. Jiang J, et al. (2013) The Arabidopsis RNA binding protein with K homology motifs,SHINY1, interacts with the C-terminal domain phosphatase-like 1 (CPL1) to repressstress-inducible gene expression. PLoS Genet 9(7):e1003625.

24. Chen T, et al. (2013) A KH-domain RNA-binding protein interacts with FIERY2/CTDphosphatase-like 1 and splicing factors and is important for pre-mRNA splicing inArabidopsis. PLoS Genet 9(10):e1003875.

25. Jeong IS, et al. (2013) Arabidopsis C-terminal domain phosphatase-like 1 functions inmiRNA accumulation and DNA methylation. PLoS One 8(9):e74739.

26. Li H, Shen JJ, Zheng ZL, Lin Y, Yang Z (2001) The Rop GTPase switch controls multipledevelopmental processes in Arabidopsis. Plant Physiol 126(2):670–684.

27. Yang G, Gao P, Zhang H, Huang S, Zheng ZL (2007) A mutation in MRH2 kinesinenhances the root hair tip growth defect caused by constitutively activated ROP2small GTPase in Arabidopsis. PLoS One 2(10):e1074.

Zhang et al. PNAS | Published online November 28, 2016 | E8205

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

19,

202

0

28. Fu Y, Li H, Yang Z (2002) The ROP2 GTPase controls the formation of cortical fine F-actinand the early phase of directional cell expansion during Arabidopsis organogenesis.Plant Cell 14(4):777–794.

29. Kirik V, et al. (2007) CLASP localizes in two discrete patterns on cortical microtubulesand is required for cell morphogenesis and cell division in Arabidopsis. J Cell Sci 120(Pt24):4416–4425.

30. Yang K, et al. (2014) Requirement for A-type cyclin-dependent kinase and cyclinsfor the terminal division in the stomatal lineage of Arabidopsis. J Exp Bot 65(9):2449–2461.

31. Koiwa H, et al. (2004) Arabidopsis C-terminal domain phosphatase-like 1 and 2 areessential Ser-5-specific C-terminal domain phosphatases. Proc Natl Acad Sci USA101(40):14539–14544.

32. Ivakov A, Persson S (2013) Plant cell shape: Modulators and measurements. FrontPlant Sci 4:439.

33. Kawasaki T, et al. (2006) Cinnamoyl-CoA reductase, a key enzyme in lignin biosyn-thesis, is an effector of small GTPase Rac in defense signaling in rice. Proc Natl AcadSci USA 103(1):230–235.

34. Chang YW, Howard SC, Herman PK (2004) The Ras/PKA signaling pathway directlytargets the Srb9 protein, a component of the general RNA polymerase II transcriptionapparatus. Mol Cell 15(1):107–116.

35. Abdellatif M, et al. (1998) A Ras-dependent pathway regulates RNA polymerase IIphosphorylation in cardiac myocytes: Implications for cardiac hypertrophy. Mol CellBiol 18(11):6729–6736.

36. Miller PJ, Johnson DI (1994) Cdc42p GTPase is involved in controlling polarized cellgrowth in Schizosaccharomyces pombe. Mol Cell Biol 14(2):1075–1083.

37. Nakano K, Mutoh T, Arai R, Mabuchi I (2003) The small GTPase Rho4 is involved incontrolling cell morphology and septation in fission yeast. Genes Cells 8(4):357–370.

38. Hachet O, Bendezú FO, Martin SG (2012) Fission yeast: In shape to divide. Curr OpinCell Biol 24(6):858–864.

39. Nakano K, Arai R, Mabuchi I (1997) The small GTP-binding protein Rho1 is a multi-functional protein that regulates actin localization, cell polarity, and septum forma-tion in the fission yeast Schizosaccharomyces pombe. Genes Cells 2(11):679–694.

40. Martin SG, Rincón SA, Basu R, Pérez P, Chang F (2007) Regulation of the formin for3pby cdc42p and bud6p. Mol Biol Cell 18(10):4155–4167.

41. Kobor MS, et al. (1999) An unusual eukaryotic protein phosphatase required fortranscription by RNA polymerase II and CTD dephosphorylation in S. cerevisiae. MolCell 4(1):55–62.

42. Cho EJ, Kobor MS, Kim M, Greenblatt J, Buratowski S (2001) Opposing effects of Ctk1kinase and Fcp1 phosphatase at Ser 2 of the RNA polymerase II C-terminal domain.Genes Dev 15(24):3319–3329.

43. Nakano K, Mutoh T, Mabuchi I (2001) Characterization of GTPase-activating proteinsfor the function of the Rho-family small GTPases in the fission yeast Schizosaccharomycespombe. Genes Cells 6(12):1031–1042.

44. Chant J, Stowers L (1995) GTPase cascades choreographing cellular behavior: Move-

ment, morphogenesis, and more. Cell 81(1):1–4.45. Hayles J, et al. (2013) A genome-wide resource of cell cycle and cell shape genes of

fission yeast. Open Biol 3(5):130053.46. Schneider S, Pei Y, Shuman S, Schwer B (2010) Separable functions of the fission yeast

Spt5 carboxyl-terminal domain (CTD) in capping enzyme binding and transcription

elongation overlap with those of the RNA polymerase II CTD. Mol Cell Biol 30(10):

2353–2364.47. Hoffman K, Yoo H, Karagiannis J (2013) Synthetically engineered rpb1 alleles altering

RNA polymerase II carboxy terminal domain phosphorylation induce discrete mor-

phogenetic defects in Schizosaccharomyces pombe. Commun Integr Biol 6(3):e23954.48. Schwer B, Shuman S (2011) Deciphering the RNA polymerase II CTD code in fission

yeast. Mol Cell 43(2):311–318.49. Schwer B, Sanchez AM, Shuman S (2012) Punctuation and syntax of the RNA poly-

merase II CTD code in fission yeast. Proc Natl Acad Sci USA 109(44):18024–18029.50. Danko CG, et al. (2013) Signaling pathways differentially affect RNA polymerase II

initiation, pausing, and elongation rate in cells. Mol Cell 50(2):212–222.51. Lagha M, et al. (2013) Paused Pol II coordinates tissue morphogenesis in the Dro-

sophila embryo. Cell 153(5):976–987.52. Saunders A, Core LJ, Sutcliffe C, Lis JT, Ashe HL (2013) Extensive polymerase pausing

during Drosophila axis patterning enables high-level and pliable transcription. Genes

Dev 27(10):1146–1158.53. Coelho CM, Leevers SJ (2000) Do growth and cell division rates determine cell size in

multicellular organisms? J Cell Sci 113(Pt 17):2927–2934.54. Cadart C, Zlotek-Zlotkiewicz E, Le Berre M, Piel M, Matthews HK (2014) Exploring the

function of cell shape and size during mitosis. Dev Cell 29(2):159–169.55. Hajheidari M, Farrona S, Huettel B, Koncz Z, Koncz C (2012) CDKF;1 and CDKD protein

kinases regulate phosphorylation of serine residues in the C-terminal domain of

Arabidopsis RNA polymerase II. Plant Cell 24(4):1626–1642.56. Cui P, et al. (2016) The RNA polymerase II C-terminal domain phosphatase-like protein

FIERY2/CPL1 interacts with eIF4AIII and is essential for nonsense-mediated mRNA

decay in Arabidopsis. Plant Cell 28(3):770–785.57. Kuchin S, Treich I, Carlson M (2000) A regulatory shortcut between the Snf1 protein

kinase and RNA polymerase II holoenzyme. Proc Natl Acad Sci USA 97(14):7916–7920.58. Navarro-Lérida I, et al. (2015) Rac1 nucleocytoplasmic shuttling drives nuclear shape

changes and tumor invasion. Dev Cell 32(3):318–334.59. Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: A Bioconductor package for

differential expression analysis of digital gene expression data. Bioinformatics 26(1):

139–140.60. Mathur J, Koncz C (1997) Method for preparation of epidermal imprints using agarose.

Biotechniques 22(2):280–282.

E8206 | www.pnas.org/cgi/doi/10.1073/pnas.1605871113 Zhang et al.

Dow

nloa

ded

by g

uest

on

July

19,

202

0