Subscaling of a cytosolic RNA binding protein governs cell size homeostasis in the multiple fission alga Chlamydomonas

Coordination of growth and division in eukaryotic cells is essential for populations of proliferating cells to maintain size homeostasis, but the underlying mechanisms that govern cell size have only been investigated in a few taxa. The green alga Chlamydomonas reinhardtii (Chlamydomonas) proliferates using a multiple fission cell cycle that involves a long G1 phase followed by a rapid series of successive S and M phases (S/M) that produces 2n daughter cells. Two control points show cell-size dependence: Commitment in mid-G1 phase requires attainment of a minimum size to enable at least one mitotic division during S/M, and the S/M control point where mother cell size governs cell division number (n), ensuring that daughter distributions are uniform. tny1 mutants pass Commitment at a smaller size than wild type and undergo extra divisions during S/M phase to produce small daughters, indicating that TNY1 functions to inhibit size-dependent cell cycle progression. TNY1 encodes a cytosolic hnRNP A- related RNA binding protein and is produced once per cell cycle during S/M phase where it is apportioned to daughter cells, and then remains at constant absolute abundance as cells grow, a property known as subscaling (1). Altering the dosage of TNY1 in heterozygous diploids or through overexpression increased Commitment cell size and daughter cell size, indicating that TNY1 is a limiting factor for both size control checkpoints. Epistasis placed TNY1 function upstream of the retinoblastoma tumor suppressor complex (RBC) and one of its regulators, Cyclin-Dependent Kinase G1 (CDKG1) (2). Moreover, CDKG1 protein and mRNA were found to over-accumulate in tny1 cells suggesting that CDKG1 may be a direct target of repression by TNY1. Our data expand the potential roles of subscaling proteins outside the nucleus and imply a control mechanism that ties TNY1 accumulation to pre-division mother cell size. Author Summary Size control is a fundamental property of cells which requires balancing cell growth with cell division, but the mechanisms used by cells to achieve this balance are only partly understood. The best-characterized mechanisms for size control to date involve fixed amounts of nuclear- DNA-bound inhibitory factors which repress cell division until cells grow past a minimum size threshold to overcome the inhibition. The unicellular green alga Chlamydomonas and many other algae and protists use a non-canonical cell cycle where cells can grow by many-fold in size before dividing, and then undergo multiple fission which involves successive rapid divisions to produce a uniform-sized population of daughters. In Chlamydomonas an unknown size homeostasis mechanism couples mother cell size to division number such that larger mother cells divide more times than smaller mother cells. Here, we identified and characterized a key factor governing size control in Chlamydomonas, a cytoplasmic RNA-binding protein and division inhibitor, TNY1, that is produced in a fixed amount in daughter cells and does not increase with cell growth, a property called sub-scaling. We found that TNY1 represses production of a cell cycle activator, CDKG1, during multiple fission to control daughter cell size. TNY1 is the first example of a cytosolic cell cycle inhibitor that does not depend on nuclear DNA binding to govern sub-scaling.

functions to inhibit size-dependent cell cycle progression. TNY1 encodes a cytosolic hnRNP A-23 related RNA binding protein and is produced once per cell cycle during S/M phase where it is 24 apportioned to daughter cells, and then remains at constant absolute abundance as cells grow, 25 a property known as subscaling (1). Altering the dosage of TNY1 in heterozygous diploids or 26 through overexpression increased Commitment cell size and daughter cell size, indicating that 27 TNY1 is a limiting factor for both size control checkpoints. Epistasis placed TNY1 function 28 upstream of the retinoblastoma tumor suppressor complex (RBC) and one of its regulators, 29 Cyclin-Dependent Kinase G1 (CDKG1) (2). Moreover, CDKG1 protein and mRNA were found to 30 over-accumulate in tny1 cells suggesting that CDKG1 may be a direct target of repression by 31 TNY1. Our data expand the potential roles of subscaling proteins outside the nucleus and imply 32 a control mechanism that ties TNY1 accumulation to pre-division mother cell size. We next used epistasis experiments to determine the dependency tny1-1 phenotypes on other 150 size regulators. CDKG1 functions upstream of the RBC and cdkg1-2 null mutants cause a large-151 cell phenotype. cdkg1-2 tny1-1 double mutants had identical sizes as cdkg1-2 single mutants 152 indicating that TNY1 functions upstream of CDKG1 and the RBC, and does not appear to 153 control cell size homeostasis through an independent mechanism ( Figure 1B, C). Note that 154 Commitment sizes for cdkg1-2 and cdkg1-2 tny1-1 (~200 µm3) are very similar to the 155 Commitment size (~200 µm3) of a wild type strain ( Figure S1E, F), indicating that cdkg1-2 156 suppresses both the Commitment and the S/M size defects of tny1-1. 157 The tny1-1 strain was found to contain a single insertion of the paroR marker in the first exon of 158 Cre07.g330300 (17) ( Figure 1A). tny1-1 was back-crossed to wild type CC-125 and random 159 progeny were chosen and scored for gamete cell size, mating type, and paromomycin 160 resistance. The paroR segregants were small, while the paroS segregants were wild-type size 161 indicating linkage between the paroR insertion and the tny1-1 phenotype (Methods, Figure  162 S1G). Rescue of the tny1-1 small cell defect was performed by transforming constructs that 163 contained either a full-length genomic fragment of wildtype Cre07.g330300 (gTNY1) or a 164 version with a C-terminal triple hemagglutinin epitope tag (gTNY-3xHA). In both cases, normal 165 daughter cell size was restored in transformants while no rescue was observed in control 166 transformants bearing an empty vector ( Figure 1D, Figure S1H, I). Rescue efficiency with either 167 of two constructs was somewhat low (~2%) but not atypical for Chlamydomonas rescues. 168 Immunoblotting of SDS-PAGE separated proteins from wild type, tny1-1, and rescued tny1-1 169 strains using polyclonal antibodies raised against recombinant TNY1 protein or anti-HA 170 antibodies detected proteins of the expected migration (~48 kDa) in wild type and rescued 171 strains showing that TNY1 expression was restored in those rescued lines ( Figure 1E).

9
Together these experiments confirm that disruption of Cre07.g330300 causes the tny1-1 173 phenotype. 174 175 TNY1 is predicted to encode a putative hnRNP A-related RNA 176 binding protein.

177
TNY1 is predicted to encode a protein with two N-terminal RNA recognition motifs (RRMs) and 178 a low complexity glycine-rich C-terminus (Figure 2A, Figure S2). This structure is found in 179 eukaryotic heterogeneous nuclear ribonucleoproteins (hnRNPs) and other related RNA binding 180 proteins that have diverse roles in nucleic acid regulation and metabolism, functioning as RNA 181 or DNA binding proteins (18,19). BLAST searching in different taxa was used to identify 182 proteins related to TNY1 in animals, plants, and algae. These sequences were curated and 183 used to estimate a maximum likelihood phylogeny which placed TNY1 in a clade of green algal 184 TNY1-like homologs, and this TNY1 clade was sister to a larger grouping of plant tandem RRM 185 hnRNP-like proteins suggesting a common origin at the base of the Viridaeplantae (Methods, 186  To determine the subcellular localization of TNY1, a genomic TNY1 construct with a C-terminal 195 fusion of mCherry (20) was used to rescue tny1-1 mutant cells and generate gTNY1-196 mCherry::tny1-1 strains with fusion protein expression confirmed by immunoblotting ( Figure S3). 197 Live cell confocal fluorescence microscopy revealed that TNY1-mCherry is detectable in the 198 cytosol throughout the vegetative cell cycle with a weak but significant signal detected at all 199 stages ( Figure 3A). Indirect immunofluorescence using anti-HA antibodies targeting tagged 200 TNY1-HA confirmed the cytosolic location and showed exclusion of TNY1 protein signal from 201 the nucleus ( Figure 3B). 202 203 TNY1 regulation and subscaling throughout the cell cycle.

204
To determine the accumulation pattern of TNY1 mRNA during the cell cycle wild-type cultures 205 were synchronized under a standard diurnal cycle (12hr:12hr light:dark) and RNA samples were 206 prepared from cells at different time points and used for quantitative RT-PCR. TNY1 mRNA was 207 present at very low levels during G1 phase and rose sharply to a peak toward the middle/end of 208 S/M phase, and then declined slowly in the dark phase after division ( Figure 4A top panel). This 209 experiment largely reproduced the results of (21) and (22), where the timing of TNY1 mRNA 210 accumulation coincided with that of many late mitotic and cilia-related genes. 211 The trigger for TNY1 mRNA accumulation is likely to be cell division, but we could not rule out 212 diurnal control or the light-to-dark transition as the drivers of TNY1 expression. To distinguish 213 these possibilities we used two alternative diurnal regimes where peak S/M phase (12-14 hrs 214 ZT) did not coincide with the light-to-dark transition (early dark regime -15hr:9hr light:dark; 215 extended light regime -9hr:15hr light:dark). In both alternative regimes, TNY1 mRNA peaked 216 with S/M phase and was not significantly shifted by the timing of the light-dark transition ( Figure  217 4A). 218 TNY1 mRNA declines gradually after S/M and becomes almost undetectable after the beginning 219 of the light period. To further determine if TNY1 mRNA turnover was facilitated by light, RNA 220 samples were collected under a shortened dark cycle, where cells were first synchronized under 221 12 that mother cell size or numbers of daughter nuclei may control TNY1 mRNA production ( Figure  248 4F, S4D, E). To determine if the subscaling of TNY1 is controlled by any feedback from size 249 control regulators, we examined its levels in cell size mutants. TNY1 protein levels were 250 determined in daughters produced from wild type, mat3-4, dp1-1 and cdkg1-2. Interestingly, 251 different sized daughter cells contained the same amount of TNY1 on a per cell basis ( Figure  252 4G). Therefore, TNY1 production in each cell cycle is governed independently of the mitotic size 253 control pathway, and its levels may instead be controlled by limiting factors that scale invariantly 254 with cell size such as genomic template for TNY1 transcription. 255 256 TNY1 is limiting for size control.

257
Size regulators that exhibit subscaling behavior are predicted to be dosage sensitive (23) (1, 6). 258 To determine whether TNY1 gene dosage might be limiting for size control, a set of isogenic 259 diploid strains was constructed with genotypes TNY1/TNY1, TNY1/tny1-1, and tny1-1/tny1-1 260 (Methods). Size profiles of daughters from synchronized diploid cultures of each strain were 261 determined and compared to each other and to haploid strains. Daughters from the two 262 homozygous strains were approximately twice the size of haploid daughters of the same 263 genotype, while the heterozygous TNY1/tny1-1 daughters were intermediate in size between 264 the homozygous mutant and wild-type diploid strains ( Figure 5A) and expressed less TNY1 265 protein than TNY1/TNY1 diploids ( Figure 5B). Further supporting the dosage sensitivity of 266 TNY1, we found that among the meiotic progeny of tny1-1::TNY1 (or tny1-1::TNY1-HA) rescued 267 strains backcrossed to a wild-type parental strain, those that inherited both the wild type TNY1 268 allele and the TNY1 transgene were larger than those that inherited only the wild type TNY1 269 allele or those that had the parental genotype of tny1-1 with a TNY1 rescuing construct ( Figure  270 S5A, B). 271 Besides altering gene dosage, we also generated a TNY1 transgene driven by a previously 272 characterized constitutive promoter/terminator from Chlamydomonas RPL23 gene (24). This 273 RPL23:gTNY1:RPL23 construct was transformed into a tny1-1 strain and transformants were 274 tested for size phenotypes along with control transformants that received an empty vector with a 275 selectable aph7 marker conferring hygromycin resistance (25). Among independent 276 RPL23:gTNY1::tny1-1 transformants, ~ 80% were rescued and were close to wild-type in mode 277 size ~ 80 µm 3 ( Figure S5C), while ~ 20% showed a large-cell phenotype with a modal cell size > 278 100 µm 3 that was never observed in rescue experiments using constructs with the controls 279 ( Figure 5C, Figure S5D). Large-sized RPL23:gTNY::tny1-1 populations are always larger than 280 the wild type strain throughout G1 ( Figure S5E), while passing Commitment and entering S/M 281 with similar timing as wild type and small-sized tny1-1 mutants ( Figure S5F). Taken together, 282 these data indicate that dosage and expression level of TNY1 impact mitotic cell size control 283 and are consistent with the subscaling behavior observed for TNY1 expression being an 284 important contributor to size-dependent cell cycle control. wild-type and tny1-1 strains were synchronized and CDKG1 mRNA levels were measured in 296 post-mitotic cells. Previously it was found that CDKG1 mRNA and CDKG1 protein super-scale 297 with mother cell size (2). In this experiment more CDKG1 message was detected in tny1-1 than 298 in wild type, even though comparing the pre-division populations, tny1-1 cells were smaller than 299 wild type ( Figure 6A). Since TNY1 is cytosolic, TNY1 is most likely to affect CDKG1 mRNA 300 levels by impacting message stability, though this finding does not rule out a possible role for 301 TNY1 in translational control of CDKG1. 302 To test the impact of TNY1 on CDKG1 protein abundance, tny1-1 was crossed into a rescued 303 cdkg1 strain expressing an HA epitope tagged allele HA-CDGK1 (2), so that CDKG1 protein

314
The finding that cytosolic TNY1 could inhibit accumulation of nuclear-localized CDKG1 protein 315 suggests a mechanism which might involve direct interaction of TNY1 with CDKG1 mRNA. We 316 first used native electrophoresis of whole cell extracts, and immunoblotting to determine if TNY1 317 might be part of a ribonucleoprotein complex (RNP). On native gels, TNY1 migrated near the 318 158 kDa marker which is slower than would be expected for free TNY1 (> 450 kDa). TNY1 319 migration was unchanged when extracts were pre-treated with DNAse, but when extracts were 320 treated with microcoocal nuclease which digests both DNA and RNA, or by increasing amounts 321 of RNAse, the TNY1 signal shifted to a slower moving complex migrating near the 450 kDa 322 marker. These results suggest that TNY1 is associated with RNA in vivo as an RNP, and that 323 the RNA component may contribute significantly to the negative charge state of the complex 324 and impacting its migration rate on native gels ( Figure 7A). 325 A simple model for regulation of CDKG1 by TNY1 is direct binding of TNY1 to the CDKG1 326 mRNA which has an unusually long (1.7kb) and uridine-rich (27% U versus 19% average for 327 Chlamydomonas genes) 3' UTR (26)-both rare features in Chlamydomonas mRNAs that tend 328 to have shorter 3' UTRs and overall GC-rich nucleotide composition. We attempted to detect 329 TNY1 binding to CDKG1 mRNA in vivo using RNA crosslinking and immunoprecipitation (RIP) 330 (27) but were unable to amplify an enriched signal due to high background. Instead, we 331 developed an in vitro assay where radiolabeled CDKG1 mRNA fragments were used as a probe 332 for binding to GST-TNY1 fusion protein or GST immobilized on a membrane (Methods) (28). 333 Radiolabeled CDKG1 mRNA was synthesized in two fragments, with the 5' region including the 334 5'UTR and CDS in one fragment, and the 3' UTR in a second fragment. After incubation of 335 radiolabeled RNA with membrane-bound GST1-TNY1 or GST1 and washing, the signal was 336 detected only for the 3' UTR fragment binding to GST1-TNY1 ( Figure 7B). These data indicated 337 that TNY1 protein can bind RNA with sequence specificity, including sequences in the 3' UTR of 338 its likely target gene CDKG1. ). An appealing property of subscaled proteins is that their absolute abundance in a cell can 370 act as a denominator for perceiving changes in cell size whose proxy is a protein or other 371 molecule whose cytoplasmic concentration is constant. Interestingly, in the above examples 372 subscaling could be directly tied to nuclear function via DNA or chromatin (6). In budding yeast, 373 nuclear Whi5 protein binds to and inhibits the DNA bound transcription factor SBF, a key 374 activator of S phase transcription. While some regulation of Whi5 abundance may occur based 375 on synthesis of Whi5, it is also limited by chromatin binding (8, 29, 30). Similar findings were 376 made for the RB protein in mammalian cells which is a functional analog of Whi5 for S phase 377 transcription (10). In plants, chromatin binding by the CDK inhibitor KRP4 coupled with 378 elimination of excess unbound KRP4 allows daughter cells to be apportioned with a fixed 379 amount of KRP4 that acts as a concentration dependent inhibitor of the cell cycle in the 380 subsequent G1 phase and ensures that S phase entry occurs at a constant average cell size 381 regardless of daughter cell sizes (7). Here we found that subscaling can also occur for a 382 cytosolic protein, TNY1, that has no direct connection to the nucleus or chromatin. This finding 383 raises the question of how TNY1 synthesis is controlled and how its levels can be modulated so 384 that daughters always contain the same amount of TNY1. One way to achieve a fixed dose of 385 TNY1 per cell would be if production of TNY1 mRNA is limited by TNY1 gene copy number in 386 daughters and not influenced by cell size related factors (e.g. transcription factor abundance, 387 co-activator abundance) (23), but this remains to be determined. Supporting this idea, TNY1 388 absolute abundance in daughters was not influenced by cell size mutants that caused 389 production of large or small daughters. To date, TNY1 is the only cell cycle regulatory protein in 390 Chlamydomonas known to subscale. The RB complex is downstream of TNY1 in 391 Chlamydomonas, but MAT3/RBR increases in abundance during G1 phase (2, 16) and does not 392 show dosage sensitivity for size control as its mammalian homolog RB and its yeast counterpart 393 Whi5 do. Thus, the systems-level target for subscaling of size control is not conserved between 394 algae and these two members of the opsithokont phylum. 395 Interestingly, TNY1 shares some similarity to budding yeast Whi3, an RNA binding protein and 396 negative cell cycle regulator that functions in part by restricting expression of the limiting G1 397 cyclin Cln3 (31, 32). In budding yeast, Whi3 represses the function of Cdc28-Cln3 by retaining 398 Cdc28-Cln3 complexes in the cytoplasm in G1 phase (33). Whi3 does not impact the 399 abundance of Cdc28 but does represses CLN3 mRNA abundance and translational efficiency 400 Replicator (, which delivers ~3 μL hanging drop) on a light shelf at 25°C for 6 days. 439 Approximately 1/3 of each stamped spot was removed with a toothpick and resuspended in 440 nitrogen-free HSM in a new 96 well plate to create a gamete suspension. Gametes were then 441 checked for cell size using a Coulter Counter. Confirmed mutants were then crossed to wild-442 type strain CC125, and progeny were tested for linkage of the suppressor phenotype to the 443 pSI103 insertion. The tny1-1 insertion site was determined by sequencing junction fragments 444 from ligation mediated PCR (45), and the insertion site was confirmed using genotyping primers 445 for tny1-1 (Supplementary Table).   Trizol-like reagent following the method of (15) then digested with RNase-free Turbo DNase 534 following the manufacturer's protocol. 4 µg total RNA was reverse transcribed with oligo dT and 535 random hexamers (9:1) using Thermo Script Reverse Transcriptase at 25°C for 10 min, 42°C for 536 10 min, 50°C for 20 min, 55°C for 20 min, 60°C for 20 min, 85°C for 5 min. SYBR-Green based 537 qPCR reactions in two technical duplicates of two biological replicates were performed and 538 quantitated in a Bio-Rad CFX96 system. Each 10 μL reaction contained 0.1 µL cDNA, 1x 539 Invitrogen Taq buffer, 3.5 mM MgCl2, 0.5x SYBR Green I, 0.05% Tween 20, 0.05 mg/mL BSA, 540 5% DMSO, 200 µM dNTPs, 0.3 µM primers, and 5U of Invitrogen Taq DNA polymerase. 541 Expression was normalized against GBLP (GenBank NC_057009.1) as an internal control. The 542 melting curve was examined for each reaction to ensure that no primer dimers or non-specific 543 PCR products were present. qPCR primers for CDKG1, TNY1, and GBLP can be found in the 544 Light microscopy 547 Chlamydomonas cells were fixed in 0.2% glutaraldehyde final concentration. Cells were 548 mounted on slides and imaged with a Leica DMI 6000 B microscope with a 63* oil objective and 549 DIC optics with images taken using a Photometrics Coolsnap HQ2 CCD camera.

backbone. 581
Chlamydomonas codon-optimized mCherry was amplified using a primer set of BamH1 582 mCherry F and XbaI mCherry R (Supplementary Table) from pLM006 (20), digested with 583 BamH1 and Xba1, then used to replace GFP in the plasmid pTNY-GFP digested with BamHI 584 and XbaI to create plasmid pTNY1-mCherry. pTNY1-mCherry was transformed into tny1-1 and 585 rescued transformants were identified by measuring gamete sizes as described above and then 586 confirmed by immunoblotting and measuring sizes of vegetative daughter cells. for TNY1-587 mCherry using α TNY1 antibody.  Table). 32 P labeled RNA was generated/ transcribed in vitro using 622 a Maxiscript kit in the presence of α-32 P-CTP (NEN) according to manufacturer instructions. 623 Each 25 μL reaction had the following components: DNA template 0.5ug, 10x Transcription 624 buffer 2 μL, 0.5 mM ATP, 10mM GTP 1 μL, 10mM UTP 1 μL, 500uM CTP 1 μL, 32 P-CTP 2 μL 625 (10 mCi/mL), 2 μL T7 RNA polymerase. After 1hr reaction at 30°C, the mixture was treated with 626 DNAseI (ambion) and purified with Sigma post reaction clean-up columns SigmaSpin™ to 627 remove unincorporated nucleotides. RNA integrity was visualized by separating a sample of the 628 RNA on a urea denaturing 4% polyacrylamide gel followed by autoradiography. and stained with Ponceau S. The membrane was incubated at 4°C overnight with re-naturation 652 buffer: 50 mM tris-HCl pH 7.5, 100 mM KCl, 1% Triton X-100 and 10% glycerol. After re-653 naturation, the membrane was incubated for 1hr with reactivation buffer (Tris-HCl pH 7.5, 0.1 % 654 triton X-100, 10% glycerol) at room temperature, blocked for one hour with yeast tRNA (80 655 μg/mL) in reactivation buffer followed by incubation with 32 P labeled RNA in reactivation buffer 656 for 3 hrs. Membranes were washed 4 times with reactivation buffer and exposed to X-ray film 657 for 2 days at -80°C before development.