A Super-Assembly of Whi3 Encodes Memory of Deceptive Encounters by Single Cells during Yeast Courtship Fabrice Caudron1 and Yves Barral1,* 1Institute of Biochemistry, Department of Biology, ETH Zurich, Schafmattstrasse 18, 8093 Zurich, Switzerland *Correspondence: [email protected] http://dx.doi.org/10.1016/j.cell.2013.10.046

SUMMARY

Cellular behavior is frequently influenced by the cell’s history, indicating that single cells may memorize past events. We report that budding yeast permanently escape pheromone-induced cell-cycle arrest when experiencing a deceptive mating attempt, i.e., not reaching their putative partner within reasonable time. This acquired behavior depends on superassembly and inactivation of the G1/S inhibitor Whi3, which liberates the G1 cyclin Cln3 from translational inhibition. Super-assembly of Whi3 is a slow response to pheromone, driven by polyQ and polyN domains, counteracted by Hsp70, and stable over generations. Unlike prion aggregates, Whi3 super-assemblies are not inherited mitotically but segregate to the mother cell. We propose that such polyQ- and polyN-based elements, termed here mnemons, act as cellular memory devices to encode previous environmental conditions. INTRODUCTION A large variety of signaling mechanisms are used by cellular systems to sense and respond to various extracellular stimuli. Often, these sensing pathways also support secondary responses, called adaptation, that adjust signal transduction and cellular responses to the intensity and kinetics of the signal. These adaptation strategies may combine short- and long-term memory effects. For example, photoreceptor cells rapidly attenuate the response of individual light receptors to allow perception of signal fluctuations. The same cells also mobilize long-term adaptation mechanisms to fit their overall sensitivity to ambient luminosity (Korenbrot, 2012). Similar combinations of shortand long-term adaptations underlie neuronal function and the processing and memorization of information in synapses (Kilian and Muller, 2002; Mayford et al., 2012). However, research generally focuses on how the primary signal is transduced, such that we rarely know how cells adapt to stimuli at different time scales. The response of budding yeast to mating pheromone has been an excellent model for characterizing G-protein-coupled recep1244 Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc.

tor signaling (reviewed in Wang and Dohlman [2004] and Chen and Thorner [2007]) and for studying signal attenuation and response-adaptation mechanisms (reviewed in Slessareva and Dohlman [2006]). Haploid yeast cells exist in two mating types: mating type a (MATa), which secretes the mating pheromone a factor, and mating type a (MATa), which produces a factor. Cells of opposite mating types sense each other’s presence through the pheromone receptors Ste2 (MATa cells; Jenness et al., 1983) and Ste3 (MATa cells; Hagen et al., 1986). These receptors of the seven transmembrane family signal through a G protein module and a mitogen-activated protein kinase (MAPK) cascade (Elion, 2000). The downstream mitogen-activated protein (MAP) kinase Fus3 in turn activates a transcription factor (Ste12; Tedford et al., 1997) and the cell-cycle inhibitor Far1 to promote arrest of both partner cells in the G1 phase (Chang and Herskowitz, 1990) and the formation of mating projections toward each other (Arkowitz, 2009). Successful mating culminates in fusion of the partners and the formation of a diploid yeast zygote. In theory, meiosis produces two MATa and MATa spores per ascus, which should suffice to form matching pairs upon spore germination. However, additional mechanisms further facilitate partner-match formation. For example, haploid mother cells regularly switch mating type to generate both MATa and MATa daughters that can subsequently mate with each other (reviewed in Haber [2012]). Also, when pheromone concentration remains low, yeast cells bud toward the source of the pheromone, bringing their daughters closer to distant mates (Erdman and Snyder, 2001). However, they also adapt to the pheromone signal over time (reviewed in Chen and Thorner [2007]), allowing them to resume vegetative proliferation when not finding a match, even if pheromone remains present (Moore, 1984). Adaptation mechanisms target the pheromone response pathway at many levels. First, MATa cells secrete the protease Bar1 to degrade a factor, lower its concentration in the medium, and sharpen the local pheromone gradient toward the nearest mating partner (Jin et al., 2011; Sprague and Herskowitz, 1981). Second, the regulator of G-protein-signaling protein, Sst2, attenuates signaling through interactions with both the receptor and the guanosine triphosphate-bound Ga protein (Dohlman, 2009). Third, upon pheromone response, the receptor is phosphorylated on its C-terminal tail and endocytosed (Jenness and Spatrick, 1986). Finally, the MAP kinase cascade is the target of feedback inhibition through both phosphorylation and degradation, allowing termination of the pheromone response

(Molina et al., 2010). Inactivating these attenuation mechanisms either causes hypersensitivity to pheromone or delays the return of the cell to vegetative proliferation upon pheromone withdrawal. Most studies on adaptation of the pheromone response have been carried out at the population level. In order to study the dynamics with which single yeast cells adapt to pheromone, we established microfluidics conditions in order to observe cells exposed to a controlled pheromone concentration for extended periods of time. RESULTS Budding Yeast Cells Escape Unproductive Pheromone Arrest Using a microfluidic device (see Experimental Procedures), we exposed wild-type MATa cells to a flow of constant a factor concentration for 16 hr. We varied medium flow and pheromone concentration to determine how these changes affected pheromone response (Figure 1A and Movie S1 available online). The proportion of cells escaping pheromone response and resuming division within the 16 hr decreased with pheromone concentration. At 5 nM, 48.2% of the cells escaped, compared to 21.6% and 1.9% at 6 nM and 8 nM, respectively (Figure 1B). At concentrations of 10 nM and 12 nM, no cells were observed to escape. To our surprise, the escape also depended on the pressure applied to drive the flow. At lower pressure (2.5 psi), all cells initially shmooed (5 nM and 12 nM); however, 74.5% and 13.5% of them, respectively, escaped during observation (Figure 1C). Upon redesign of the chip (Y04 chips), a pressure of 3.5 psi and 5–7 nM pheromone were chosen to study pheromone response and escape because 5 nM was close to the boundary of response (only 83.5% of the cells shmooed) and 7 nM was a sensitive condition for escape (average time to escape was 7.1 ± 3.2 hr; 6.5% of the cells failed to escape within 16 hr; Figure 1D). We next monitored whether cells that escaped and divided subsequently resumed shmooing as they returned to G1. The cell’s behavior depended on its identity. Mother cells kept dividing after finishing their first cycle, in spite of the pheromone. In up to 13 cell divisions, virtually none shmooed even once again (0.4% ± 0.7%; n = 217 cells). In contrast, the vast majority of their daughters shmooed (89.2% ± 2.9%; n = 414 cells), indicating that they were born naive (Figure 1A). These cells escaped later, with kinetics similar to their mothers (5.46 ± 2.38 hr over at least 12 hr; n = 34 cells; Figure 1D). Thus, yeast cells exposed to pheromone but failing to find a partner escape shmooing and enter a pheromone refractory state, in which they divide even in the presence of pheromone. This refractory state is maintained for virtually the remainder of the cell’s lifespan but is not inherited by daughters. We next asked whether maintenance of the refractory state required the presence of pheromone. We exposed naive cells to pheromone for 6.25 hr or 7.75 hr, removed the pheromone for 3.75 hr or 4.75 hr, respectively, reintroduced the pheromone, and monitored the cells’ reaction (Figure 1E). During the pheromone-free period, the cells divided at least twice and up to five times. Out of 22 initial cells, only one resumed shmooing

upon pheromone re-exposure. All other cells kept dividing (Figure 1E). Therefore, the pheromone refractory state is maintained even in absence of pheromone. In contrast, all cells born after the first exposure to pheromone schmoo as naive cells (n > 60). In order to determine whether the refractory state depended on escape, we monitored the behavior of cells that were still shmooing upon pheromone withdrawal in the above conditions. Remarkably, upon re-exposure, most of these cells proved to be refractory to pheromone (67 cells out of 75). Upon varying the duration of the first exposure, we determined that, after 3 hr, virtually all cells had reached the refractory state (83.8% ± 6.7%; Figures 1F and 1G; Movie S2), although none had escaped pheromone response. Thus, the refractory state was established roughly 4 hr prior to escape. Thus, haploid yeast cells exposed to pheromone but failing to mate enter within 3 hr into a pheromone refractory state and eventually escape cell-cycle arrest. The refractory state is then stable, possibly throughout the remaining life of the cell, even in the absence of pheromone. Therefore, yeast cells ‘‘memorize’’ unproductive mating attempts and become refractory. However, daughters do not inherit this adaptation. Establishment of the Refractory State Precedes Inactivation of the Response Pathway Next, we asked whether establishment of the refractory state involved the inactivation of the mating pathway. To test whether refractory cells were fully resistant to pheromone, we monitored how cells that had escaped at 5 nM pheromone reacted to increasing pheromone concentrations (3-, 6-, or 10-fold). Remarkably, a growing fraction of the cells resumed shmooing with increasing pheromone concentrations (up to 78% of the cells at 50 nM; n = 67 cells; Figures 2A and 2B; Movie S3). Thus, cells had kept the ability to sense pheromone upon escape but increased their response threshold. Accordingly, levels of the a factor receptor, Ste2, remained unchanged at the cell surface (Figure 2C). Upon pheromone binding, Ste2 is endocytosed and targeted to the vacuole (Schandel and Jenness, 1994; Jenness and Spatrick, 1986). To determine whether Ste2 still bound pheromone after escape, we measured the amounts of internalized Ste2-GFP present in pheromone-treated cells prior to and after escape (Figures 3A and 3B). As expected, the levels of internal Ste2-GFP increased over time in treated compared to nontreated cells. However, after 7.5 hr, the dividing and the still responding cells showed no significant difference (5.0 ± 2.2 arbitrary units [a.u.], n = 50 cells versus 5.4 ± 1.9, n = 53 cells). In contrast, the internal levels of Ste2 rapidly dropped upon pheromone withdrawal (2.5 ± 0.6 a.u.; n = 58 cells 2 hr after withdrawal; Figures 3A and 3B). Thus, consistent with previous studies (Jenness and Spatrick, 1986), the cells continue to bind pheromone even after escape. To further characterize pheromone signaling during escape, we next monitored signaling events at the level of the MAP kinase Fus3 and of the FUS1 promoter, downstream of Fus3 (Roberts et al., 2000). In response to pheromone, Fus3 enters the nucleus (Blackwell et al., 2003, 2007) and the nuclear-tocytoplasmic ratio of Fus3-GFP increases, as observed by Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc. 1245

A

0h

2h

5.5h

8.5h

14.5h

d2

d2

d3

d3

d1 d1

d5

[α-factor] (nM)

d4

7 0

0

8 Cycling Shmooing

C % of cells shmooing

% of cells shmooing

5.5 Psi Y02 100

50

0 0

8 12 Time (hours) 5 nM 6 nM

d2

4

d3

d4

50

16

0 0

d7

d8

3.5 Psi Y04 100

50

4

8 12 16 Time (hours) 2.5 Psi 5.5 Psi

8 nM

d6

D

5 nM Y02

100

d5

% of cells shmooing

B

d1

16 Time (hours)

0

0

4

12 8 Time (hours) 5 nM 7 nM

16

7 nM daughters

E

0h

2h

6h

10h

14h d3 d3

d7

d1

d1 d2

d2 d5

[α-factor] (nM)

d6

5 0

0

4

8

Cycling Shmooing

F

d1

d2

d3

2h

0h

12

d4

d5

3.5h

d6

16

d7

d8

d9

6.5h

Time (hours)

d10

10h d4 d5d6 d3

d3 d2 d1

d2

[α-factor] (nM)

d1

9 0 0

4 Cycling Shmooing

12

8

d1

d2

d3

d4

d5

d6

d7

d8

16 Time (hours) d9

G % of cells shmooing at second exposure

100

50

0 0

1

2

3

(legend on next page)

1246 Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc.

0h

2h

3.5h

5h

B

11h d3 d1

d4 d1 d5

[α-factor] (nM)

d2

50 0

0

4 Cycling Shmooing

C

12 Time (hours)

8

d1

d2

d3

d4

100 % of cells resuming shmooing after escape

A

60

20 0

20

30

40

50

Second pheromone concentration after escape at 5nM (nM)

d5

Time in pheromone (hours) 0

4h

5h30

DIC

Escaped cell

Ste2-GFP

Non Escaped cell

Figure 2. After Escape, the Cells Still Respond but Only to Higher Pheromone Concentration (A) A wild-type cell exposed to 5 nM pheromone escaped and resumed shmooing upon switching to 50 nM pheromone. See also Movie S3. (B) Percent of cells resuming shmooing after escape at indicated pheromone concentrations (n = 82, 34, and 67 cells, respectively). (C) Single focal plane images of cells expressing Ste2-GFP and exposed to 5 nM pheromone during the indicated time. Arrowheads point to Ste2-GFP at the plasma membrane. The scale bars represent 5 mm. DIC, differential interference contrast.

time-lapse microscopy (Figures 3C and 3D). Whereas this ratio decreased upon pheromone withdrawal (Figures 3E and 3F), it continued to increase in cells stimulated with pheromone, even after they had escaped (Figures 3F–3H). In contrast, using a GFP reporter (PFUS1-UBI-YDkGFP*; Houser et al., 2012; Figures 3I–3M), we observed that the FUS1 promoter was rapidly turned off as the cells restarted dividing in the presence of pheromone. However, inactivation of the FUS1 promoter followed escape rather than preceded escape. Therefore, neither escape nor the refractory state involves inactivation of the pheromone response pathway.

Whi3 Inactivation and Release of Cln3 Activity Promote Escape The G1 cyclin Cln3 plays a key role in triggering the expression of the other G1 cyclins and thereby in promoting start and entry into a new budding cycle (Dirick and Nasmyth, 1991; Tyers et al., 1993). Furthermore, the cln3D mutant cells resume division inefficiently after pheromone arrest (Courchesne et al., 1989). Therefore, we investigated whether Cln3 promotes escape from pheromone-induced cell-cycle arrest. Supporting this idea, only few cln3D mutant cells escaped pheromone arrest after 12 and 16 hr (10.5% and 29.6%, respectively; Figure 4A).

Figure 1. Yeast Cells Become Refractory to Pheromone after an Unsuccessful Mating Attempt (A) Escape of a wild-type cell exposed to 7 nM pheromone. See also Movie S1. (B–D) Percentage of initial cells still shmooing after the indicated time in the indicated microfluidic setups. (E) Refractory state of a cell allowed to escape from pheromone arrest (5 nM pheromone) and released in pheromone-free medium for 4.75 hr. (F) A wild-type cell exposed to 9 nM pheromone for 3 hr, switched to a pheromone-free medium for 2.5 hr, and re-exposed to a medium containing 9 nM pheromone for the remaining 10.5 hr. See also Movie S2. (G) Percent of initial cells shmooing upon second exposure to 9 nM pheromone (mean ± SD). The scale bars represent 5 mm. Time axis (bottom) indicates the timing of the budding events and the cell-cycle state (d, daughter cell). The blue bar under each image indicates the presence and concentration of pheromone.

Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc. 1247

A

B 5h30 + 2h in YPD

Ste2-GFP

7h30 Escaped cell

Ste2-GFP levels (a.u.)

7h30 Non escaped cell

0h

10

5

0 4h

0h

7h30 Non escaped cells

mean Fus3-GFP intensity

+ pheromone until t = 1.75 h 0h 1.5h

Fus3-GFP nuclear to cytoplasmic ratio

5.5h

7h

2.5h

5.5h

7h

0h

1.5h

2.5h

5.5h

7h

Fus3-GFP 4

4

2 0

3.5 7.00 Time (hours)

9 0 10.5

2 0

3.5 7.00 Time (hours)

10.5

9 0

2 0

3.5 7.00 Time (hours)

10.5

9 0

2 0

-3.5

0 3.5 Time (hours)

9 0

h

pe

h

ca

25

es

5.

5

0

h

4

***

***

pe

4

*** ***

3

6

+

6

pe

6

10

ca

6

M

escape

h

8

0

8

ca

8

0

5.

L

8

3.50

es

K

-3.50 -1.75 0 1.75 Time (hours)

h

7.00

75

3.50 5.25 Time (hours)

1.

1.75

es

J

0

***

1

h

7.00

9 0

0

3.50 5.25 Time (hours)

1.6

h

1.75

9 0

2.

I

0

1.6

5

7.00

9 0

0

3.50 5.25 Time (hours)

1.6

3.

1.75

9 0

2.0

h

1.6

2.0

***

2

h

2.0

2.0

*** ***

0

2.4

0

2.4

5.

2.4

H

escape

FUS1-UBI-YΔkGFP* mean

G

[α-factor] (nM)

F

FUS3-GFP nuclear to cytoplasmic ratio

E

P

2.4

0

FUS1-UBI-YΔkGFP* mean

1.5h

[α-factor] (nM)

FUS3-GFP nuclear to cytoplasmic ratio

D

P

0h

+ pheromone

Fus3-GFP

2.5h

=

Fus3-GFP

mean Fus3-GFP intensity

- pheromone

Fus3-GFP

C

5h30 + 2h in YPD

7h30 Escaped cells

Figure 3. The Pheromone Response Pathway Remains Functional during Escape (A) Localization of Ste2-GFP in cells (sum projection) exposed to pheromone for the indicated time and released in pheromone-free medium for 2 hr (right). Note that the plasma membrane localization is blurred due to the sum projection. (B) Quantification of Ste2-GFP fluorescence intensity (arbitrary units) in cells (n > 50) as in (A). Mean ± SD. (C) Schematic of the Fus3-GFP nuclear to cytoplasmic ratio measurement assay and localization of Fus3-GFP in wild-type cells upon exposure and removal after 1.75 hr of pheromone (bottom left), no exposure to pheromone (top right), or constant exposure to pheromone (bottom right). (D–F) Fus3-GFP nuclear/cytoplasmic average intensity ratio (mean ± SEM) in cells exposed to pheromone for 1.75 hr (before the dashed line) and then to a pheromone-free medium (after the dashed line; D), in cells dividing in pheromone-free medium (E) or in cells in constant pheromone (F; arrows indicate escape events). (G) Same data as in (F), but single traces were synchronized by the time of escape (dashed line) before averaging. n = 40 cells for each condition. (H) Statistical analysis of the values of (D), (E), and (G) at indicated times. (I–K) PFUS1-UBI-YDkGFP* average intensity (mean ± SEM) for cells exposed to pheromone for 1.75 hr (before the dashed line) and then to a pheromone-free medium (after the dashed line; I), cells in pheromone-free medium (J), or cells in constant pheromone (K; arrows indicate escape events). (L) Same data as in (K), but single traces were reaveraged as in (G). n = 50 cells for each condition. (M) Statistical analysis of the values of (I), (J), and (L) at indicated times. Three asterisks indicate p < 0.001 using one-way ANOVA.

This observation prompted us to assess the involvement of the mRNA-binding protein, Whi3 (Garı´ et al., 2001), a translational inhibitor of Cln3 expression. The whi3D mutant cells fail to inhibit CLN3, progress faster from G1 into S phase, are small compared to wild-type (Figure 4C; Nash et al., 2001), and arrest poorly in response to pheromone (Figure 4A). Remarkably, the whi3D mutant cells that succeeded in shmooing escaped three times faster than wild-type cells (2.65 ± 1.25 hr versus 7.09 ± 3.17 hr). Cells expressing a whi3 allele lacking its RNA recognition motif (RRM), whi3-DRRM, behaved similarly (Figure 4A). 1248 Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc.

Thus, Whi3 and its RNA-binding activity promotes cell-cycle arrest in response to pheromone and delayed the subsequent escape process. To test whether Whi3 promoted cell-cycle arrest through inhibition of CLN3, we next investigated the behavior of the whi3D cln3D double-mutant cells. These cells responded efficiently to pheromone and most of them (66%) failed to resume budding within 16 hr of treatment, like the cln3D single-mutant cells. Furthermore, mutant cells expressing a CLN3 mRNA not recognized by Whi3 but encoding wild-type Cln3 (CLN3-mGCAU;

A

B

C 1.5 WHI3

50

PolyQ

whi3-ΔpQ

3HA

PolyN

% of counts

% of cells shmooing

100

RRM

RRM

PolyN

1.0

0.5 0

whi3-ΔpN 0

4

8 12 Time (hours)

D

RRM

0.0

4

6

8

10

3HA

RRM

wild-type

whi3-ΔpN

whi3-ΔpQ

whi3-ΔpQpN

whi3Δ

2h

5.5h

8.5h

14.5h

[α-factor] (nM)

whi3-ΔpQ

0h

3HA

Cell size (μm)

whi3-ΔQpN

whi3Δ cln3Δ whi3Δ CLN3-mGCAU

wild-type cln3Δ whi3-ΔRRM

PolyQ

16

7 0 0h

8

16 Time (hours)

Cycling Shmooing

E

F 7 nM

50

0

0

4

8 12 Time (hours)

50

0

16

50

0 0

4

8 Time (hours)

7 nM

100 % of cells shmooing

100 % of cells shmooing

% of cells shmooing

100

G 5 nM

12

16

0

4

8 12 Time (hours)

wild-type

whi3-ΔpQ

wild-type

whi3-ΔpQ

whi3-ΔpN cln3Δ

whi3-ΔpN

whi3-ΔpQpN

whi3-ΔpN

whi3-ΔpQpN

whi3-ΔpQ cln3Δ

16

whi3-ΔpQpN cln3Δ whi3-ΔpQpN CLN3-mGCAU Time in pheromone (hours) 0 Cln3-3HA wild type Pgk1

Cln3-3HA whi3-ΔpQpN

Pgk1

2

3

4

I 5

5 Cln3-3HA levels (a.u.)

H

wild type whi3-ΔpQpN

4 3 2 1 0 0

1

2 3 Time (hours)

4

5

Figure 4. PrD-Dependent Repression of Whi3 and Release of Cln3 from Inhibition Drives Escape (A) Percentage of cells of indicated genotypes shmooing in 7 nM pheromone over time, as in Figure 1B. (B) Schematic representation of WHI3 alleles. (C) Size distribution for cells of indicated genotypes. (D) Escape of a whi3-DpQ mutant cell exposed to 7 nM pheromone, as in Figure 1A. The scale bar represents 5 mm. See also Movie S4. (E–G) Plots as in (B) for cells of indicated genotypes in the presence of 7 nM (E and G) or 5 nM (F) pheromone. (H) Western blot analysis of in vivo levels of Cln3-3HA in wild-type and whi3-DpQpN mutant cells exposed to pheromone for the indicated times. (I) Quantification of Cln3-3HA from (H). See also Data S1.

Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc. 1249

Colomina et al., 2008) phenocopied the whi3D and whi3-DRRM single mutants (Figure 4A). Thus, Whi3 promoted pheromoneinduced cell-cycle arrest and delayed escape mainly through binding the CLN3 mRNA and inhibiting CLN3 function. We conclude that escape involves releasing CLN3 from Whi3dependent inhibition. Like many other RNA-binding proteins, Whi3 contains stretches of amino acids predicted to promote its aggregation (PrDs; Alberti et al., 2009) and specifically two PrDs conserved in Whi3 orthologs (Data S1; Lee et al., 2013), which we named polyQ and polyN due to their enrichment in glutamine and asparagine residues, respectively. Intrigued by these domains, we tested whether they contributed to Whi3 regulation. We replaced either the polyQ or the polyN by three copies of the hemagglutinin (HA) tag to keep roughly the same size as wild-type Whi3 (Figure 4B). Mutant cells lacking either the polyQ (whi3-DpQ) or the polyN (whi3-DpN) responded to pheromone as efficiently as wild-type cells and grew to similar sizes (Figure 4C), suggesting that these WHI3 alleles are functional. In contrast, cells lacking both the polyQ and the polyN (whi3-DpQpN) grew larger than wild-type cells, indicating that Whi3-DpQpN was an overactive protein (Figure 4C). Therefore, the polyQ and the polyN domains do not mediate Cln3 repression. We next monitored how cells lacking these domains escaped from pheromone arrest. Removing the polyQ, and to a lesser extent the polyN, substantially delayed escape (Figures 4D–4F; Movie S4; at 5 nM pheromone, 13.4% of the whi3-DpQ MATa mutant cells, n = 230, escaped shmooing after 12 hr versus 81.2% of the wild-type cells, n = 232). Deletion of both PrDs enhanced that phenotype (Figure 4F). Thus, the polyQ and polyN domains synergistically promote escape, i.e., the release of the CLN3 mRNA from Whi3-dependent inhibition. Together, these observations raised the possibility that the PrDs contribute to the progressive inhibition of the protein in response to pheromone. Consistent with the Whi3-DpQpN protein being overactive in repressing the CLN3 mRNA, whi3DpQpN CLN3-mGCAU double-mutant cells were able to escape (Figure 4G) to similar levels as the CLN3-mGCAU single-mutant cells (Figure 4A). Furthermore, the cln3D whi3-DpN, cln3D whi3DpQ, and cln3D whi3-DpQpN double-mutant cells all behaved similarly to the cln3D single-mutant cells (Figure 4G). Thus, the Whi3-DpQpN protein delayed escape from cell-cycle arrest through increased binding and repression of the CLN3 mRNA. Consistent with this idea, during a 5 hr pheromone treatment, Cln3 levels progressively increased in wild-type but remained very low in the whi3-DpQpN mutant cells (Figures 4H and 4I). Thus, the polyQ domain of Whi3, helped by the polyN, contributes to the gradual release of CLN3 from Whi3-dependent inhibition in pheromone-treated cells. This, in turn, promotes the termination of the pheromone-mediated G1 arrest and return of the cells to proliferation. The polyQ and the polyN Domains of Whi3 Are Required for the Maintenance of the Pheromone Refractory State As established above, cells acquire the refractory state upon transient exposure to pheromone. Remarkably, under this treatment, most whi3-DpQpN and whi3-DpQ and many of the whi3DpN mutant cells reresponded to pheromone like naive cells 1250 Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc.

(Figures 5A and 5B; Movie S5). Furthermore, whereas throughout our movies wild-type cells that escaped pheromone arrest maintained the refractory state as long as we could follow them, many of the whi3-DpN and whi3-DpQ mutant cells resumed shmooing, particularly at low pheromone concentration (5 nM; Figures 5C and 5D; Movie S6). This effect was even stronger in the whi3-DpQpN mutant cells (Figure 5E). Thus, prior to escape, the PrDs contribute to both establishment and maintenance of the refractory state. Whi3 Forms Amorphous Cytoplasmic Super-Assemblies in Response to Pheromone Altogether, our data indicate that the PrDs of Whi3 mediate Whi3 inactivation and the establishment and maintenance of a stable pheromone-refractory state. Because PrDs are known to induce the formation of multimeric structures, such as prion-forming, amyloid fibers (Kabani and Melki, 2011) and dynamic heteromeric amyloid-like fibers that form hydrogels (Kato et al., 2012), we contemplated the possibility that the PrDs of Whi3 promote its integration into higher-order structures in response to pheromone. Total extracts of pheromone-treated (5.5 hr) or untreated cells were fractionated over 0%–40% sucrose gradients. After pheromone treatment, the distribution of Whi3-tandem affinity purification (TAP), but not Whi3-DpQpN-TAP, was shifted toward higher-density fractions (Figure S1). Thus, the PrDs indeed facilitated incorporation of Whi3 into larger structures upon pheromone treatment. Moreover, when total cell extracts of pheromone-treated cells (4.5 hr) were analyzed by semidenaturating detergent agarose gel electrophoresis (SDDAGE), sample boiling was necessary for Whi3-TAP to efficiently enter the gel (Figure 6A). This was not necessary when Whi3 was extracted from nontreated cells. Furthermore, pheromone treatment made Whi3, but not Whi3-DpQpN, partially resistant to proteinase K (Figures 6B and 6C). Thus, we conclude that, in response to pheromone, the PrDs of Whi3 switch conformation and promote the protein to assemble into a larger and stable structure. We next wondered if we could detect Whi3 super-assemblies within cells. We tagged endogenous Whi3 with 3GFP and observed its localization. In exponentially growing cells, Whi33GFP localized diffusely throughout the cytoplasm and to brighter granules, as reported for mRNA-bound Whi3 (Garı´ et al., 2001; Figures 6D and S2A, arrowheads). These granules typically measured one to two pixels across and were found in only one z-plane (z sections were 400 nm apart). Upon pheromone treatment, Whi3-3GFP progressively formed an amorphous cytoplasmic super-assembly (Figure 6D, arrows and 6E, green and red lines). Super-assemblies were several pixels across and found in at least two focal planes (Figure S2C). They had no specific shape and varied in brightness, but unlike the granules, they were resistant to formaldehyde fixation (Figure S3). After 3 hr in pheromone, cells containing a super-assembly (>70.0% ± 5.0%) started to lose Whi3 granules (Figure 6E, red lines). Therefore, concomitantly with the emergence of the pheromone-refractory state, the cells contained generally one Whi3-3GFP super-assembly in the cytoplasm. Consistent with the super-assemblies contributing to this state, the daughters formed by cells that had escaped arrest

6.5h

[α-factor] (nM)

B

10h

d1

d2

d1

d2

9

100 *** *** 50

0

0 0

4 Cycling Shmooing

C

d1

0h

1h

12

8

16 Time (hours)

2.5h

4.5h

D

11h

d4 d2

5 4

12

8

16 Time (hours)

*

wh i3 -Δ p wh Q i3 -Δ pN

-ty d3

d4

d5

d6

d7

d8

ld

d2

7 nM *

40

0

0

3

whi3-ΔpQ whi3-ΔpQpN

80 **

30

wi

d1

2

E

5 nM

pe

[α-factor] (nM)

d5

% of cells resuming shmooing at least once in 13 hours after escape

whi3-ΔpN

d3

d1

Cycling Shmooing

1 wild-type whi3-ΔpN

d1

0

0

d2

60

0

**

0%

wi ld -ty wh pe i3 -Δ p wh Q wh i3-Δ pN i3 -Δ pQ Δp N

3.5h

% of cells resuming shmooing at least once in 8 hours after escape

2h

whi3-ΔpQ

0h

% of cells shmooing at second exposure

A

Figure 5. The Refractory State and Its Maintenance Depend on the PrDs of Whi3 (A) A whi3-DpQ mutant cell exposed to 9 nM pheromone for 3 hr, switched to a pheromone-free medium for 2.5 hr, and re-exposed to 9 nM pheromone for the remaining 10.5 hr. See also Movie S5. (B) Percent of cells of indicated genotypes shmooing upon re-exposure to 9 nM pheromone. (C) A whi3-DpN mutant cell escaped and resumed shmooing during constant exposure to 5 nM pheromone. See also Movie S6. (D) Quantification of cells shmooing at least once in 13 hr after escape of the indicated genotypes. (E) Quantification of cells of indicated genotypes that reshmooed at least once in 8 hr after escape. For (A) and (C), scale bars = 5 mm. For (B), (D), and (E) (mean ± SD), stars indicate p values obtained from one-way ANOVA comparing to wild-type.

generally had none (87.9% ± 2.4%; Figures 6F and S2B), which fits with the fact that the vast majority of these cells successfully responded to pheromone (89.2% ± 2.9%; see above). In agreement with these super-assemblies being driven by the PrDs of Whi3, the frequency of the cells exhibiting super-assemblies was reduced 1.2-fold, 2.1-fold, and 4.0-fold in the whi3DpN-3GFP, whi3-DpQ-3GFP, and whi3-DpQpN-3GFP mutant cells, respectively (Figures 6E and S3). However, neither the polyQ nor polyN domain contributed to the formation of Whi3 granules. This correlates well with the granules being the functional units of Whi3 and the polyQ-less and polyN-less alleles successfully arresting the cell cycle in response to pheromone. The Hsp70 Chaperone Ssa1 Inhibits the Escape The formation of polyQ- and polyN-dependent structures, such as prions, is generally under the control of protein chaperones. For example, Hsp70 (Ssa1 and Ssa2) and Hsp100 (Hsp104) proteins counteract conversion to [PSI+] (Sweeny and Shorter, 2008) and generally colocalize with amyloid aggregates (Kaganovich et al., 2008; Lee et al., 2010). Unlike amyloids, Whi3-3GFP super-assemblies did not colocalize with Hsp104mCherry granules (Figures 7A and 7B) and were not juxtanuclear (Figure S4), suggesting that Whi3 did not partition to the JUNQ or IPOD, two depository sites for misfolded proteins (Kaganovich et al., 2008). In contrast, Ssa1-mCherry formed super-assemblies in response to pheromone, which colocalized with Whi33GFP super-assemblies (in 62.5% ± 9.0% of the cells; Figures 7C and 7D), but not with granules (Figures 7E and 7F).

We next asked whether Ssa1 interferes with the formation of Whi3 super-assemblies. In the absence of pheromone, 10.0% ± 1.7% of ssa1D, but not ssa2D, mutant cells already contained Whi3-3GFP super-assemblies (1.1 ± 1.0 in wild-type cells; Figures 7G, 7H, and S5A). Furthermore, in these cells, but not in ssa2D mutant cells, Whi3 super-assembled faster upon pheromone treatment than in wild-type cells (Figure 7H). Moreover, the accelerated assembly of Whi3 in ssa1D mutant cells depended on Whi3’s PrDs; deleting SSA1 in whi3-DpQpN mutant cells did not restore Whi3 super-assembly (Figure S5A). In contrast, Whi3 super-assembly was slower in hsp104D mutant cells (Figure S5A). Therefore, Ssa1 counteracts the super-assembly of Whi3, whereas Hsp104 promotes it. The observation that chaperone mutations modulated super-assembly of Whi3 allowed us to ask whether this superassembly is functionally relevant for establishing the pheromone-refractory state. Consistent with this possibility, ssa1D, but not ssa2D, mutant cells escaped faster than wild-type cells (Figure 7I), and about 10% (n = 190 cells) of ssa1D cells did not shmoo, in fitting with 10.0% containing a Whi3-3GFP super-assembly prior to pheromone treatment. The hsp104D mutant cells escaped and became pheromone-refractory at a slightly slower rate than wild-type cells (Figures 7I and S5B), consistent with Hsp104 promoting super-assembly. In fitting with Ssa1 inhibiting escape through Whi3, the ssa1D whi3D double-mutant phenocopied the whi3D single-mutant cells with respect to pheromone response and escape (Figure 7J), and the SSA1 deletion did not hasten escape of cln3D mutant cells Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc. 1251

37°C

25°C

37°C

25°C

Treatment prior to loading

B

+ 100°C

-

Pheromone

C

Whi3-TAP

-

Pheromone 100°C

A

Proteinase K treatment (min) 100

0

+

0.5 1 2

0

0.5

Whi3-ΔpQpN-TAP

-

Pheromone 1

Proteinase K treatment (min) 100

2

75

75

50

50

0

+

0.5 1 2

0 0.5

1

2

Whi3-TAP anti-TAP western SDD-AGE

E

Time in pheromone 0h

2h

F

3h wild-type

% of cells

100

Whi3-3GFP

anti-TAP western

anti-TAP western

100

50

whi3-ΔpQ

Whi3-3GFP

D

37

37

50

0 0

1

2

3

0 0

1

2

3

100 % of cells

whi3-ΔpQ-3GFP

100

50

% of cells

whi3-ΔpN

100

whi3-ΔpQpN

50

50

0 0 0

1

2

3

Time in pheromone (hours)

0

0

1

2

3

Time in pheromone (hours)

Figure 6. Whi3 Forms Asymmetrically Inherited Super-Assemblies in Response to Pheromone (A) Behavior of Whi3-TAP from cells ± pheromone treatment on SDD-AGE. (B and C) Proteinase K sensitivity of Whi3-TAP (B) and Whi3-DpQpN-TAP (C) extracted from cells treated or not with pheromone. Arrowheads indicate resistant species. (D) Localization of Whi3-3GFP ± pheromone treatment (7 nM). Arrowheads, granules; arrows, super-assembly. (E) Quantification of Whi3-3GFP super-assemblies (mean ± SD). Blue indicates cells with no super-assembly and many granules, green cells with a superassembly and many granules, and red cells with a super-assembly and very few or no granules. Transparent lines correspond to the wild-type values. (F) Asymmetric segregation of Whi3-3GFP super-assemblies in cells exposed to pheromone for 3 hr and released in a pheromone-free medium for 1.5 hr. See also Figures S1, S2, and S3.

(Figure 7J). Finally, Ssa1 function required the PrDs of Whi3, because the ssa1D mutation had little effect on escape when WHI3 was replaced by the whi3-DpQpN allele (Figures 7K and S5C). Therefore, we conclude that super-assembly drives the establishment of the pheromone-refractory state and escape. DISCUSSION A Case for Cellular Memory How biological systems acquire and encode memories of past events is a tantalizing question in biology. The stability over time of the pheromone-refractory state described here indicates that yeast cells are able to maintain long-term encoding of a past experience. This is akin to memory, if we define memory as what underlies the ability to modulate a response as a function of whether a similar stimulus has been encountered in the past. Under this definition, our studies underline the emerging idea (Kilian and Muller, 2002; Mayford et al., 2012) that the mechanisms of memory are not unique to neurons but may have originally emerged in unicellular organisms. Thus, these simple systems may provide powerful models for studying the molecular mechanisms of memory. 1252 Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc.

Here, we show that yeast cells treated with pheromone mount an adaptation response to resume proliferation if no successful mating takes place in a reasonable amount of time. Whereas the primary response to pheromone, leading to cell-cycle arrest, takes place within minutes, adaptation is slow and leads to escape only after hours. These timings open a finite time window for mating. The adaptation process starts early, as the refractory state is reached hours before the cells actually escape. The primary response to pheromone includes the arrest of the cells in the G1 phase of the cell cycle. This arrest involves the inactivation of the CLN3 gene. The Far1 protein binds and keeps the Cln3 protein unable to promote cell-cycle entry (Tyers and Futcher, 1993), whereas the Whi3 protein represses the CLN3 mRNA (Garı´ et al., 2001) and others. Our data suggest that adaptation follows the following scenario. The PrDs of Whi3 slowly mediate the transition of the protein from a soluble and active to a partially insoluble, super-assembled and inactive state. Hence, within the first 3 hr of cell-cycle arrest, the CLN3 mRNA is released from inhibition and Cln3 starts accumulating again. At this point, the cells have acquired the refractory state, i.e., if the pheromone is removed, they will not respond to subsequent stimulation.

A

Hsp104-mCherry

B

merge

1.0

Fluorescence intensity (a.u.)

Whi3-3GFP

0.6

0.2

C

Ssa1-mCherry

D

merge

Ssa1-mCherry

merge

F

0

Time in pheromone 2h

Ssa1-mCherry Whi3-3GFP

0.7

0

1 2 3 Distance (μm)

H

3h

% of cells

4

ssa1Δ

100

Whi3-3GFP ssa1Δ

0h

1 2 3 Distance (μm)

1.0

0.4

G

6

0.6

Fluorescence intensity (a.u.)

Whi3-3GFP

2 4 Distance (μm)

Ssa1-mCherry Whi3-3GFP

0.2

E

0

1.0

Fluorescence intensity (a.u.)

Whi3-3GFP

Hsp104-mCherry Whi3-3GFP

50

0 0 1 2 3 Time in pheromone (hours)

I

J

100

0

4

8 Time (hours)

12

wild type

ssa1Δ

hsp104Δ

ssa1Δ ssa2Δ

16 ssa2Δ

% of cells shmooing

50

0

K 100

% of cells shmooing

% of cells shmooing

100

50

0

0

4

8 12 Time (hours)

ssa1Δ

whi3Δ

ssa1Δ cln3Δ

16 cln3Δ

ssa1Δ whi3Δ

50

0

0

4 ssa1Δ

8 12 Time (hours)

16

whi3-ΔpQpN

whi3-ΔpQpN ssa1Δ

Figure 7. The Hsp70 Chaperone Ssa1 Inhibits Whi3 Assembly and the Escape (A) Lack of colocalization of Hsp104-mCherry with Whi3-3GFP super-assemblies in a wild-type cell treated with pheromone (7 nM) for 3 hr. (B) Whi3-3GFP and Hsp104-mCherry fluorescence intensities profiles of the line indicated in (A). (C and D) Colocalization (C) and quantification (D) of the intensities of Whi3-3GFP and Ssa1-mCherry super-assemblies in a wild-type cell treated with pheromone. (E and F) Lack of colocalization (E) and quantification (F) of the intensities of Ssa1-mCherry with Whi3-3GFP granules in a wild-type cell treated with pheromone. (G) Localization of Whi3-3GFP ± pheromone treatment (7 nM) in ssa1D mutant cells. Arrowheads, granules; arrows, super-assemblies; scale bars for (A), (C), (E), and (G) = 5 mm. (H) Quantification of Whi3-3GFP assembly (mean ± SD), as for Figure 6E. (I–K) Percentage of cells with indicated genotypes shmooing in the presence of 7 nM pheromone. See also Figures S4 and S5.

The inactivation of Whi3 drives the cells in a second step of adaptation; the cells now escape from the cell-cycle arrest and resume proliferation. Through some unclear mechanism, the

escape event also turns off pheromone-induced transcription of the FUS1 gene. Remarkably, however, the pheromone receptor and the MAP kinase cascade remain at least partially active, Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc. 1253

allowing cells to resume shmooing if pheromone concentration increases significantly. Strikingly, the PrDs of Whi3 keep the pheromone-refractory state stable for many cell cycles, possibly for the cell’s entire life. At the same time, as long as we monitored it, the Whi3 super-assembly remained stable in the escaped mother cell. Thus, we propose that super-assembly promotes both escape, through the release of Cln3 from inhibition, and the long-term encoding of the refractory state, i.e., memory. Our data indicate that the encoding of the refractory state involves a conformational change driven by the PrDs of Whi3 and the concomitant super-assembly of Whi3. Furthermore, the failure of daughters to inherit the refractory state from their mothers also correlates with their failure to inherit super-assemblies. Finally, the PrDs are required for the stability of the refractory state over time. Thus, the super-assembly process may ensure both its own stability and its discrete mode of inheritance at mitosis at the same time as it turns off the primary function of Whi3 in translation control. By analogy to prions and hydrogels (Halfmann and Lindquist, 2010; Kato et al., 2012), the superassembled form of Whi3 might be self-promoting by facilitating the further recruitment of newly synthesized Whi3 to it, even after the assembly trigger has disappeared. Thus, one important feature of the mechanism of refraction is its virtual irreversibility, promoting the maintenance of memory beyond the lifetime of single molecules. Whi3 Super-Assemblies At the biochemical level, the mechanism of Whi3 switching to its inhibited, memory-promoting state is reminiscent of protein aggregation in hydrogels and prions (Halfmann and Lindquist, 2010; Kato et al., 2012). Inhibited Whi3 was present in higherdensity fractions of sucrose gradients and required boiling to be detected by SDD-AGE. However, in these gels, we could not detect higher molecular weight species typical of prions or amyloid assemblies. Also, unlike prions, Whi3 super-assemblies were not shared with daughter cells but remained in mother cells. They partially colocalized with the chaperone Ssa1, but not with Hsp104. Therefore, Whi3 behaved similarly, but not exactly, like prions, both at the biochemical and cellular levels. Like other mRNA-binding proteins containing low-complexity domains, Whi3 might rather engage into hydrogel-like structures (Han et al., 2012; Kato et al., 2012). However, unlike hydrogels, Whi3 super-assemblies appeared to liberate their target mRNA. Therefore, Whi3 assemblies might represent yet another type of structure, with its own dynamics and properties. Importantly, removing both PrDs of Whi3 largely, but not completely, abolished super-assembly. Furthermore, whi3-DpQpN cells were still able to mount a refractory state, albeit very inefficiently. Therefore, other factors may contribute to the establishment of the refractory state and be part of the super-assembly. Due to the differences between Whi3 and classical prions and because of its function in memory encoding, we propose to term it ‘‘mnemon.’’ We define a mnemon as a protein that super-assembles in a controlled manner to encode memory of a past event and segregates discretely, being inherited asymmetrically at mitosis or kept in a specific subcellular compartment in nondividing cells. 1254 Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc.

A Refractory State against Pheromone-Induced Cell-Cycle Arrest The idea that the refractory state is a form of cellular memory implies that it must have a physiological relevance in nature. Adaptation was especially efficient in cells exposed to low and intermediate pheromone concentrations. In the wild, such levels probably signal an unproductive mating attempt. Indeed, as partners come closer to each other, they become exposed to rapidly increasing pheromone concentrations (Achstetter, 1989; Strazdis and MacKay, 1983). Therefore, we speculate that the establishment of a stable refractory state and its confinement to the mother cell may have two purposes in the wild. First, long-term adaptation might prevent opening a niche for cheaters. Indeed, when cells arrest in response to pheromone, the few cells that would not do so, yet keep secreting pheromone, would pre-empt the resources of the environment for their progeny, at the cost of their neighbors. Conversely, the ability to escape inconclusive arrest may give a selective advantage to the corresponding genome because it promotes its maintenance and proliferation in the mating population. Second, the refractory state allows mother cells to increase the chances of their progeny to reach a partner when those are at a distance (Erdman and Snyder, 2001). Obviously, this strategy only works if the refractory state is not passed on to the daughters. Thus, a strategy of long-term adaptation may have arisen because it helps individual cells to hedge bets when stopping their proliferation to search for a mate, at the risk of environmental resources being overtaken by competitors. A Family of Mnemons? The involvement of low-complexity domains and protein super-assembly in the establishment and the maintenance of a memory trait is highly reminiscent of the long-term plasticity of neurons mediated by CPEB in Aplysia and Orb2 in Drosophila (Keleman et al., 2007; Si et al., 2003). CPEB and Orb2 have prion-like domains and are mRNA-binding proteins regulating translation. CPEB forms higher-order structures, leading to its activation and to memory maintenance (Heinrich and Lindquist, 2011; Si et al., 2003, 2010). Ironically, in Drosophila, the PrD of Orb2 is specifically required for longterm courtship memory (Keleman et al., 2007). When males encounter a fertilized female, they learn that this female is refractory to mating and remember it when encountering other fertilized females. Together, these observations reveal striking similarities between memory processes and their mechanisms in multi- and unicellular organisms. Indeed, under our definition, CPEB and Orb2 should potentially be considered themselves as mnemons. In silico surveys identified more than 150 proteins with PrDs in the yeast genome (Alberti et al., 2009). Some formed prion-like aggregates and behaved as such, but most of them did not. Many of these PrD-containing proteins are RNA-binding proteins, a feature conserved in higher eukaryotes (Alberti et al., 2009; King et al., 2012). Therefore, many of these proteins might be mnemons and yeast might memorize additional events. Similarly, in the syncytia of A. gossypii, Whi3 (especially its polyQ)

drives the individualization of the different nuclei in the same cytoplasm (Lee et al., 2013), possibly memorizing their history. Higher eukaryotic genomes also encode these types of proteins, suggesting that mnemons will be found in multicellular organisms, where they may contribute to diverse forms of memory, including cell-fate determination. Thus, the list of mnemons is likely to expand in the near future. Remarkably, it did not escape our attention that the Whi3 mnemon segregates with the aging mother cell, whereas daughter cells are born naive. Thus, at least in this case, memory segregates with age in yeast, as it does in metazoans. EXPERIMENTAL PROCEDURES Strains Media and genetic methods are described (Guthrie and Fink, 2004). All strains used in this study are bar1D and derivatives of S288c or were backcrossed four times to the S288c background. Fluorescent proteins were tagged at endogenous loci (Knop et al., 1999). PrDs of Whi3 were replaced by the 3xHA tag as generically described (Schneider et al., 1995). Resulting clones were subsequently confirmed by sequencing. The polyQ was replaced from nucleotide 738 to nucleotide 868 and the polyN from nucleotide 1140 to nucleotide 1444. The whi3-DpQpN allele is the whi3DQR-2 allele (Nash et al., 2001) in which the polyN is replaced by 3xHA. Microscopy All images were acquired either on a Personal Deltavision (Applied Precision) equipped with a CCD HQ2 camera (Roper) and 250W Xenon lamps controlled by Softworx or a Core Deltavision (Applied Precision) equipped with a CCD HQ camera (Roper) and solid-state light-emitting diodes controlled by Softworx. Fluorescein isothiocyanate and tetramethylrhodamine isothiocyanate filters were used for imaging GFP and mCherry fluorescence. Microfluidics Most of the experiments were carried out with the ONIX microfluidic perfusion platform with Y04C microfluidic plates (CellAsic). Early experiments and switch experiments from Figures 1B, 1C, 1E, 1F, 2A, and 2B were carried out with the ONIX microfluidic flow control panel (CellAsic) and Y02 plates. Medium was yeast extract peptone dextrose (YPD) containing 5 nM or 7 nM a factor for escape experiments or 9 nM for switch experiments. For fluorescence imaging, we used synthetic depleted without tryptophane (SD-TRP) medium supplemented with 20 mg/ml casein and 9 nM a factor, which gave similar escape dynamics as YPD + 7 nM pheromone. Quantification of Whi3-3GFP Super-Assemblies and Granules Cells grown in YPD ± a factor were briefly centrifuged at 600 g, resuspended in SD-TRP medium, placed between slide and coverslip, and imaged directly. Images were analyzed after deconvolution with Softworx software. Three clones with total n = 180 cells were observed for each point. N = 120 cells for Ssa1-mCherry and Whi3-3GFP colocalization. For fixed samples, cells grown in YPD ± a factor were centrifuged, resuspended with 100 ml of 4% paraformaldehyde, and incubated at room temperature (RT) for 15 min. Cells were washed once in 1 ml 100 mM potassium phosphate pH 7.5 and 1.2 M sorbitol and resuspended in 50 ml KPO4/sorbitol. The proportion of cells containing a Whi3 super-assembly was determined upon imaging. N = 540 cells for each time point. Quantification of Shmooing Shmooing or budding states were inspected visually. During microfluidic experiments, images were taken every 15 min. Unbudded cells that showed a polarized growth were counted as shmooing. Unbudded cells undergoing isotropic growth were counted as G1 cells. Usually these cells soon started forming a bud. Samples consisted of three independent clones and n > 207 cells. For Figures 4D and 4E, three clones were used and a total of 91 (3 hr), 45 (2 hr 30 min–2 hr 45 min), and 22 cells (1 hr 45 min–2 hr 15 min).

Sucrose Gradients Sucrose gradients were formed with 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, pH 7.5, 100 mM NaCl, 10 mM CaCl2, 0.5 mM dithiothreitol, and protease inhibitors (Complete EDTA-free protease inhibitor cocktail, Roche). Five 2.2 ml layers of 0%, 10%, 20%, 30%, and 40% sucrose were deposited and were allowed to equilibrate at 4 C overnight. Yeast cells were diluted to optical density (OD) 0.2 in YPD + a factor (40 nM) or to OD 0.1 in YPD and shaken at 30 C for 5.5 hr. Cells were harvested by centrifugation, resuspended in 400 ml cold lysis buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 10 mM CaCl2, 1% (v/v) Triton X-100) containing protease inhibitor (Complete EDTA-free protease inhibitor cocktail, Roche) and lysed by mechanical disruption using glass beads (5 3 1 min vortexing at 4 C). After pelleting cell debris, 400 ml of the supernatant was applied to each gradient. Gradients were centrifuged at 4 C, 35,000 rpm for 140 min. One milliliter fractions were collected and proteins were precipitated using trichloroacetic acid and separated by SDS-PAGE. Whi3-TAP was detected with a Peroxidase Anti-Peroxidase antibody (Sigma). Membranes were stripped and reprobed with an antiPgk1 antibody (Molecular Probes). Proteinase K Treatment Yeast extracts were prepared as above. Protein concentration was determined (RC DC Protein Assay Kit; Bio-Rad) and adjusted for all samples. Proteinase K (from Tritirachium album, Sigma) was added to each sample (1% of the total protein content), incubated at RT for the indicated times, and inactivated by adding 23 Laemmli buffer and immediate boiling (5 min). SDD-AGE Seventy-five microliters of cleared yeast extracts prepared as above were transferred to a tube containing 25 ml of 43 sample buffer (23 Tris acetate EDTA, 20% glycerol, 4% SDS, 0.04% bromophenol blue) and incubated at 25 C, 37 C, or 95 C for 5 min. Samples were then loaded on a 1.5% agarose/0.1% SDS gel and run at 50 V for 2 to 3 hr. Transfer onto nitrocellulose membranes was performed as described (Halfmann and Lindquist, 2008). SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, five figures, one data file, and six movies and can be found with this article online at http://dx.doi.org/10.1016/j.cell.2013.10.046. ACKNOWLEDGMENTS We would like to thank R. Dechant and M. Peter for help with cell-size measurements; M. Bayer for the 3GFP construct; B. Futcher, E. Garı´, and B. Errede for strains and plasmids; the Light Microscopy and Screening Center of ETH Zu¨rich for microscopy support; U. Kutay, P. Meraldi, R. Kroschewski, C. Weirich, and A.M. Farcas for critical reading of the manuscript; and S. Alberti, P. Picotti, Y. Feng, M. Krzyzanowski, and the whole Barral group for discussions. F.C. and Y.B. are supported by a grant of the European Research Council to Y.B. Received: April 16, 2013 Revised: August 2, 2013 Accepted: October 25, 2013 Published: December 5, 2013 REFERENCES Achstetter, T. (1989). Regulation of alpha-factor production in Saccharomyces cerevisiae: a-factor pheromone-induced expression of the MF alpha 1 and STE13 genes. Mol. Cell. Biol. 9, 4507–4514. Alberti, S., Halfmann, R., King, O., Kapila, A., and Lindquist, S. (2009). A systematic survey identifies prions and illuminates sequence features of prionogenic proteins. Cell 137, 146–158. Arkowitz, R.A. (2009). Chemical gradients and chemotropism in yeast. Cold Spring Harb. Perspect. Biol. 1, a001958.

Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc. 1255

Blackwell, E., Halatek, I.M., Kim, H.-J.N., Ellicott, A.T., Obukhov, A.A., and Stone, D.E. (2003). Effect of the pheromone-responsive G(alpha) and phosphatase proteins of Saccharomyces cerevisiae on the subcellular localization of the Fus3 mitogen-activated protein kinase. Mol. Cell. Biol. 23, 1135–1150. Blackwell, E., Kim, H.-J.N., and Stone, D.E. (2007). The pheromone-induced nuclear accumulation of the Fus3 MAPK in yeast depends on its phosphorylation state and on Dig1 and Dig2. BMC Cell Biol. 8, 44. Chang, F., and Herskowitz, I. (1990). Identification of a gene necessary for cell cycle arrest by a negative growth factor of yeast: FAR1 is an inhibitor of a G1 cyclin, CLN2. Cell 63, 999–1011. Chen, R.E., and Thorner, J. (2007). Function and regulation in MAPK signaling pathways: lessons learned from the yeast Saccharomyces cerevisiae. Biochim. Biophys. Acta 1773, 1311–1340. Colomina, N., Ferrezuelo, F., Wang, H., Aldea, M., and Garı´, E. (2008). Whi3, a developmental regulator of budding yeast, binds a large set of mRNAs functionally related to the endoplasmic reticulum. J. Biol. Chem. 283, 28670– 28679. Courchesne, W.E., Kunisawa, R., and Thorner, J. (1989). A putative protein kinase overcomes pheromone-induced arrest of cell cycling in S. cerevisiae. Cell 58, 1107–1119. Dirick, L., and Nasmyth, K. (1991). Positive feedback in the activation of G1 cyclins in yeast. Nature 351, 754–757. Dohlman, H.G. (2009). RGS proteins the early days. Prog. Mol. Biol. Transl. Sci. 86, 1–14. Elion, E.A. (2000). Pheromone response, mating and cell biology. Curr. Opin. Microbiol. 3, 573–581.

Jin, M., Errede, B., Behar, M., Mather, W., Nayak, S., Hasty, J., Dohlman, H.G., and Elston, T.C. (2011). Yeast dynamically modify their environment to achieve better mating efficiency. Sci. Signal. 4, ra54. Kabani, M., and Melki, R. (2011). Yeast prions assembly and propagation: contributions of the prion and non-prion moieties and the nature of assemblies. Prion 5, 277–284. Kaganovich, D., Kopito, R., and Frydman, J. (2008). Misfolded proteins partition between two distinct quality control compartments. Nature 454, 1088–1095. Kato, M., Han, T.W., Xie, S., Shi, K., Du, X., Wu, L.C., Mirzaei, H., Goldsmith, E.J., Longgood, J., Pei, J., et al. (2012). Cell-free formation of RNA granules: low complexity sequence domains form dynamic fibers within hydrogels. Cell 149, 753–767. Keleman, K., Kru¨ttner, S., Alenius, M., and Dickson, B.J. (2007). Function of the Drosophila CPEB protein Orb2 in long-term courtship memory. Nat. Neurosci. 10, 1587–1593. Kilian, A.E., and Muller, B.S. (2002). Life-like learning in technical artefacts: biochemical vs. neuronal mechanisms. Proceedings of the 9th International Conference on Neural Information Processing (ICONIP’O2), 1, 296–300. King, O.D., Gitler, A.D., and Shorter, J. (2012). The tip of the iceberg: RNAbinding proteins with prion-like domains in neurodegenerative disease. Brain Res. 1462, 61–80. Knop, M., Siegers, K., Pereira, G., Zachariae, W., Winsor, B., Nasmyth, K., and Schiebel, E. (1999). Epitope tagging of yeast genes using a PCR-based strategy: more tags and improved practical routines. Yeast 15 (10B), 963–972.

Erdman, S., and Snyder, M. (2001). A filamentous growth response mediated by the yeast mating pathway. Genetics 159, 919–928.

Korenbrot, J.I. (2012). Speed, sensitivity, and stability of the light response in rod and cone photoreceptors: facts and models. Prog. Retin. Eye Res. 31, 442–466.

Garı´, E., Volpe, T., Wang, H., Gallego, C., Futcher, B., and Aldea, M. (2001). Whi3 binds the mRNA of the G1 cyclin CLN3 to modulate cell fate in budding yeast. Genes Dev. 15, 2803–2808.

Lee, R.E.C., Brunette, S., Puente, L.G., and Megeney, L.A. (2010). Metacaspase Yca1 is required for clearance of insoluble protein aggregates. Proc. Natl. Acad. Sci. USA 107, 13348–13353.

Guthrie, C., and Fink, G.R. (2004). Guide to Yeast Genetics and Molecular and Cell Biology (San Diego: Elsevier), p. 933.

Lee, C., Zhang, H., Baker, A.E., Occhipinti, P., Borsuk, M.E., and Gladfelter, A.S. (2013). Protein aggregation behavior regulates cyclin transcript localization and cell-cycle control. Dev. Cell 25, 572–584.

Haber, J.E. (2012). Mating-type genes and MAT switching in Saccharomyces cerevisiae. Genetics 191, 33–64.

Mayford, M., Siegelbaum, S.A., and Kandel, E.R. (2012). Synapses and memory storage. Cold Spring Harb. Perspect. Biol. 4, a005751.

Hagen, D.C., McCaffrey, G., and Sprague, G.F., Jr. (1986). Evidence the yeast STE3 gene encodes a receptor for the peptide pheromone a factor: gene sequence and implications for the structure of the presumed receptor. Proc. Natl. Acad. Sci. USA 83, 1418–1422.

Molina, M., Cid, V.J., and Martı´n, H. (2010). Fine regulation of Saccharomyces cerevisiae MAPK pathways by post-translational modifications. Yeast 27, 503–511.

Halfmann, R., and Lindquist, S. (2008). Screening for amyloid aggregation by Semi-Denaturing Detergent-Agarose Gel Electrophoresis. J. Vis. Exp., 2008, 838.

Moore, S.A. (1984). Yeast cells recover from mating pheromone alpha factorinduced division arrest by desensitization in the absence of alpha factor destruction. J. Biol. Chem. 259, 1004–1010.

Halfmann, R., and Lindquist, S. (2010). Epigenetics in the extreme: prions and the inheritance of environmentally acquired traits. Science 330, 629–632.

Nash, R.S., Volpe, T., and Futcher, B. (2001). Isolation and characterization of WHI3, a size-control gene of Saccharomyces cerevisiae. Genetics 157, 1469–1480.

Han, T.W., Kato, M., Xie, S., Wu, L.C., Mirzaei, H., Pei, J., Chen, M., Xie, Y., Allen, J., Xiao, G., and McKnight, S.L. (2012). Cell-free formation of RNA granules: bound RNAs identify features and components of cellular assemblies. Cell 149, 768–779. Heinrich, S.U., and Lindquist, S. (2011). Protein-only mechanism induces selfperpetuating changes in the activity of neuronal Aplysia cytoplasmic polyadenylation element binding protein (CPEB). Proc. Natl. Acad. Sci. USA 108, 2999–3004.

Roberts, C.J., Nelson, B., Marton, M.J., Stoughton, R., Meyer, M.R., Bennett, H.A., He, Y.D., Dai, H., Walker, W.L., Hughes, T.R., et al. (2000). Signaling and circuitry of multiple MAPK pathways revealed by a matrix of global gene expression profiles. Science 287, 873–880. Schandel, K.A., and Jenness, D.D. (1994). Direct evidence for ligand-induced internalization of the yeast alpha-factor pheromone receptor. Mol. Cell. Biol. 14, 7245–7255.

Houser, J.R., Ford, E., Chatterjea, S.M., Maleri, S., Elston, T.C., and Errede, B. (2012). An improved short-lived fluorescent protein transcriptional reporter for Saccharomyces cerevisiae. Yeast 29, 519–530.

Schneider, B.L., Seufert, W., Steiner, B., Yang, Q.H., and Futcher, A.B. (1995). Use of polymerase chain reaction epitope tagging for protein tagging in Saccharomyces cerevisiae. Yeast 11, 1265–1274.

Jenness, D.D., and Spatrick, P. (1986). Down regulation of the alpha-factor pheromone receptor in S. cerevisiae. Cell 46, 345–353.

Si, K., Lindquist, S., and Kandel, E.R. (2003). A neuronal isoform of the aplysia CPEB has prion-like properties. Cell 115, 879–891.

Jenness, D.D., Burkholder, A.C., and Hartwell, L.H. (1983). Binding of alphafactor pheromone to yeast a cells: chemical and genetic evidence for an alpha-factor receptor. Cell 35, 521–529.

Si, K., Choi, Y.-B., White-Grindley, E., Majumdar, A., and Kandel, E.R. (2010). Aplysia CPEB can form prion-like multimers in sensory neurons that contribute to long-term facilitation. Cell 140, 421–435.

1256 Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc.

Slessareva, J.E., and Dohlman, H.G. (2006). G protein signaling in yeast: new components, new connections, new compartments. Science 314, 1412–1413.

Tedford, K., Kim, S., Sa, D., Stevens, K., and Tyers, M. (1997). Regulation of the mating pheromone and invasive growth responses in yeast by two MAP kinase substrates. Curr. Biol. 7, 228–238.

Sprague, G.F., Jr., and Herskowitz, I. (1981). Control of yeast cell type by the mating type locus. I. Identification and control of expression of the a-specific gene BAR1. J. Mol. Biol. 153, 305–321.

Tyers, M., and Futcher, B. (1993). Far1 and Fus3 link the mating pheromone signal transduction pathway to three G1-phase Cdc28 kinase complexes. Mol. Cell. Biol. 13, 5659–5669.

Strazdis, J.R., and MacKay, V.L. (1983). Induction of yeast mating pheromone a-factor by alpha cells. Nature 305, 543–545.

Tyers, M., Tokiwa, G., and Futcher, B. (1993). Comparison of the Saccharomyces cerevisiae G1 cyclins: Cln3 may be an upstream activator of Cln1, Cln2 and other cyclins. EMBO J. 12, 1955–1968.

Sweeny, E.A., and Shorter, J. (2008). Prion proteostasis: Hsp104 meets its supporting cast. Prion 2, 135–140.

Wang, Y., and Dohlman, H.G. (2004). Pheromone signaling mechanisms in yeast: a prototypical sex machine. Science 306, 1508–1509.

Cell 155, 1244–1257, December 5, 2013 ª2013 Elsevier Inc. 1257