Assembly and function of the amyloid‐like translational repressor Rim4 is coupled with nutrient conditions

Abstract

Amyloid‐like protein assemblies have been associated with toxic phenotypes because of their repetitive and stable structure. However, evidence that cells exploit these structures to control function and activity of some proteins in response to stimuli has questioned this paradigm. How amyloid‐like assembly can confer emergent functions and how cells couple assembly with environmental conditions remains unclear. Here, we study Rim4, an RNA‐binding protein that forms translation‐repressing assemblies during yeast meiosis. We demonstrate that in its assembled and repressive state, Rim4 binds RNA more efficiently than in its monomeric and idle state, revealing a causal connection between assembly and function. The Rim4‐binding site location within the transcript dictates whether the assemblies can repress translation, underscoring the importance of the architecture of this RNA‐protein structure for function. Rim4 assembly depends exclusively on its intrinsically disordered region and is prevented by the Ras/protein kinase A signaling pathway, which promotes growth and suppresses meiotic entry in yeast. Our results suggest a mechanism whereby cells couple a functional protein assembly with a stimulus to enforce a cell fate decision.

Amyloid‐like assembly of the yeast RNA‐binding protein Rim4, which is triggered by PKA inactivation, stimulates Rim4 target binding and translational repression.

Introduction

A distinctive property of amyloids is their remarkable stability caused by an ordered and repetitive structure that makes these protein aggregates resistant to ionic detergents in vivo and in vitro (Kryndushkin et al, 2003; Alberti et al, 2009; Newby & Lindquist, 2013; Boke et al, 2016). Amyloid‐like assemblies share these and other key biochemical features with amyloids (Alberti et al, 2009; Newby & Lindquist, 2013). Protein incorporation into amyloids or amyloid‐like assemblies causes conformational alterations and liquid‐to‐solid transitions, with possible consequent loss and/or gain of function altering cell structures and/or tissues. Moreover, the repetitive nature of these assemblies implies indefinite persistence, inappropriate ligand binding, and potential infectivity (Riek & Eisenberg, 2016). Therefore, amyloids and amyloid‐like assemblies have been long associated with toxic phenotypes. This paradigm has been challenged by evidence that some cells can form amyloid‐like assemblies in physiological contexts to modulate protein activity and perform specific tasks (Hou et al, 2011; Newby & Lindquist, 2013). Specifically, cells utilize these assemblies to sequester and inactivate select proteins (Saad et al, 2017; Cereghetti et al, 2021) or to obtain new structures with emergent functions (Raveendra et al, 2013; Khan et al, 2015; Boke et al, 2016; Hervas et al, 2020). A fundamental feature of functional amyloid‐like assemblies is their regulation (Cereghetti et al, 2018). Cells couple assembly with specific stimuli to ensure switch‐like responses (Newby & Lindquist, 2013). Moreover, some of these assemblies are completely reversible, revealing that cells have mechanisms to dismantle these stable structures, with potential implications for health (Boke et al, 2016; Saad et al, 2017; Carpenter et al, 2018; Cereghetti et al, 2021). Although we know that the formation of some amyloid‐like assemblies and other biomolecular condensates is triggered by stress (Franzmann & Alberti, 2019; Guzikowski et al, 2019), we lack a detailed understanding of the signal specificity and how signaling pathways promote or prevent assembly (Cereghetti et al, 2018). Moreover, despite their important roles in physiology, our understanding of how amyloid‐like assemblies can confer an emergent function is limited.

Budding yeast is a powerful organism to study function and regulation of amyloid‐like assemblies. Many known prions, most of which are amyloids, have been identified and characterized in this fungus (Liebman & Chernoff, 2012; Newby & Lindquist, 2013). Yeast use amyloid‐like switches to respond and adapt to stimuli and to regulate fate decisions (Caudron & Barral, 2013; Newby & Lindquist, 2013; Berchowitz et al, 2015; Riback et al, 2017; Saad et al, 2017). One example is Rim4, an RNA‐binding protein required for meiosis. In yeast, meiosis occurs in the context of sporulation, which is induced by starvation (Neiman, 2011). Rim4 amyloid‐like assemblies inhibit translation of select transcripts until completion of meiosis I (Berchowitz et al, 2013, 2015; Jin et al, 2015). At meiosis II onset, these assemblies are dismantled and cleared owing to multi‐site phosphorylation by the meiotic kinase Ime2 (Carpenter et al, 2018; Wang et al, 2020). Rim4 clearance allows consequent translation of previously repressed targets. This regulation ensures the correct temporal succession of events necessary for sporulation (Jin & Neiman, 2016). For example, translation of the B‐type cyclin CLB3 must be repressed during meiosis I to establish segregation of homologous chromosomes (Carlile & Amon, 2008). Rim4 assemblies repress CLB3 translation by binding its 5′ untranslated region (5′UTR) (Berchowitz et al, 2013, 2015). Although we know that Rim4 assembly is necessary to repress translation, we lack a mechanistic understanding of the assembly requirement for this process. Moreover, the signal that induces Rim4 assembly is unknown.

In this study, we investigate how assembly activates the repressive function of Rim4 and how cells couple this process with nutrient conditions. We found that assembled Rim4 binds RNA more efficiently than its monomeric form, a requirement for effective translational repression. Moreover, Rim4 represses translation when it binds the transcripts at their 5′ end, while it does not have any effect when it binds their 3′ end. These results reveal that the RNA‐protein structure architecture is fundamental for the function of this amyloid‐like assembly. Rim4 assembly depends exclusively on its intrinsically disordered region (IDR), which contains the regulatory elements to control this process. Assembly is prevented by protein kinase A (PKA) which promotes growth and suppresses meiotic entry. Although direct phosphorylation is not sufficient to place Rim4 assembly under the control of PKA, it contributes to assembly clearance and enables cells that have accumulated assembled Rim4 to efficiently resume exponential growth.

Results

Rim4 assembly and function can be transferred to starved mitotic cells

We set out to identify what controls Rim4 assembly and function. Meiotic entry and progression require specific starvation conditions that cannot be manipulated without disrupting the process itself. To circumvent these restrictions, we set up a method to study Rim4 in mitotic haploid cells. We engineered prototroph yeast strains carrying a synthetic gene expressing RIM4 under the control of the CUP1‐1 promoter (Thiele & Hamer, 1986) (Fig 1A and Appendix Fig S1A). Importantly, we deleted the endogenous RIM4 allele, to avoid possible interference with our synthetic construct. Our strains also harbored a β‐estradiol‐inducible (Ottoz et al, 2014) reporter gene containing the 5′UTR of known meiotic targets of Rim4 (Berchowitz et al, 2013; Jin et al, 2015) and encoding a destabilized yellow fluorescence protein (YFP) to ensure a dynamic readout (Gordon et al, 2007). After growing our strains exponentially in rich, glucose‐containing medium, we starved them and induced expression of RIM4 and the YFP reporter gene. We used sporulation medium to starve the cells, unless specified otherwise (Fig 1B, Appendix Fig S2 and Appendix Table S1). We tested the solubility of Rim4 by semi‐denaturing detergent agarose gel electrophoresis (SDD‐AGE), a technique that allows for the resolution of SDS‐resistant aggregates, a hallmark of amyloid‐like assemblies (Bagriantsev et al, 2006). We confirmed that, in starved mitotic cells, Rim4 formed SDS‐resistant particles similar to meiosis (Berchowitz et al, 2015) (Fig 1C). We monitored the ability of Rim4 to repress translation of the reporter gene by measuring yellow fluorescence by flow cytometry. Cells containing the reporter with the 5′UTRs of the Rim4 targets CLB3 or SPO20 did not accumulate YFP in starvation, despite the gene being transcribed (Fig 1D). This repression depended on both Rim4 and starvation (Fig 1D and Appendix Fig S1B). We did not observe translational repression when the reporter carried a third target of Rim4, the SPS4 5′UTR. We speculate that additional meiotic factors and/or RNA sequences may be required to repress this particular construct (Jin et al, 2017). Importantly, translation of reporter transcripts containing the 5′UTRs of genes that are not Rim4 targets, ACT1 and TDH3, was not fully repressed by Rim4, although the ACT1 5′UTR construct was slightly but significantly repressed. We speculate that this partial repression may result from unspecific RNA‐binding of Rim4 combined with the considerable length of this 5′UTR. Rim4 did not repress translation of the reporter gene in cells grown in rich, glucose‐containing medium, despite the presence of CLB3 or SPO20 5′UTRs (Fig 1D). In these growth conditions, Rim4 did not form SDS‐resistant particles, instead it existed mainly as a monomeric protein (Fig 1C and Appendix Fig S1C). These results show that we can reproduce the specific meiotic translational repression and assembly behavior of Rim4 in starved mitotic cells. Moreover, our data reinforce the correlation between the ability of Rim4 to repress translation and its assembly state, which depends on growth conditions.

Figure 1. Rim4 assembly and function can be transferred to starved mitotic cells.

1. Synthetic genes encoding RIM4 tagged with V5 and YFP (yellow fluorescence protein) preceded by a 5′UTR (untranslated region) targeted by Rim4 were integrated as single copies in the yeast genome. RIM4‐V5 transcription is controlled by copper (Cu⁺⁺) through the CUP1‐1 promoter. YFP transcription is controlled by a synthetic promoter (pLexO) recognized by the LexA‐ER‐B112 transcription factor, activated by β‐estradiol.
2. Schematics of the experimental workflow for RIM4‐V5 and YFP induction with 25 μM copper sulfate and 1 μM β‐estradiol, respectively, and media handling to study Rim4 assembly and function in mitotic cells.
3. An example of SDD‐AGE to evaluate the assembly state of Rim4. Strains used are B48 (meiotic control, G2‐arrested), B1565 (no tag, grown in YPD), and B1587. Note that the meiotic control lane is the same shown in Appendix Fig S1C, left panel.
4. Expression levels of the YFP reporter gene harboring different 5′UTRs in absence or presence of the RIM4‐V5 synthetic gene in YPD and SPO. Yellow fluorescence was measured by flow cytometry (n = 3). We plotted datapoints and their mean ± standard deviation. Statistical significance was assessed with a two‐sample Wilcoxon test. *Significant; n.s. = not significant; P‐values are stated in parentheses. YFP transcript levels were evaluated by Northern blot (n = 1). Direct staining of the ribosomal RNA (rRNA) is shown as a loading control. Strains are B1565, B1587, B1923, B1928, B2112, B2139, B2212, B2221, B2451, B2452, and B2317 (no reporter control, bg = fluorescence background).

Source data are available online for this figure.

The Rim4 IDR controls assembly, enabling efficient RNA binding

We used our mitotic experimental setup to systematically analyze the role of Rim4's domains in assembly and function. Rim4 contains a C‐terminal IDR of 298 residues harboring a computationally predicted prion‐like domain with a poly‐asparagine tract (poly‐N) (Soushko & Mitchell, 2000; Alberti et al, 2009) (Fig 2A). This IDR is necessary for assembly and function during meiosis (Berchowitz et al, 2015). Rim4 contains three RNA recognition motifs (RRMs). The activity of RRM1 and RRM3 has been confirmed in meiosis (Soushko & Mitchell, 2000; Berchowitz et al, 2013, 2015). We tested the ability of a set of Rim4 mutants to assemble and repress translation of a reporter gene containing the CLB3 5′UTR (Fig 2B and C and Appendix Fig S3A). We inactivated all three RRMs simultaneously by substituting crucial phenylalanine or tyrosine residues with leucine. The resulting 3‐rrm mutant preserved its ability to assemble in starvation, confirming that the RRM activity is not necessary for Rim4 assembly (Berchowitz et al, 2015). This mutant was unable to repress translation, supporting the idea that Rim4 requires intact RRMs for function. A mutant lacking the C‐terminal 271 amino acids (Rim4ΔIDR) was mostly monomeric and failed to fully repress translation, confirming that assembly is required for function (Berchowitz et al, 2015). Next, we asked if the IDR alone is sufficient to drive assembly by fusing the C‐terminal 271 amino acids of Rim4 to a red fluorescence protein (RFP). RFP‐IDR^Rim4 displayed the same regulated assembly abilities of full‐length Rim4 but failed to repress translation of the reporter (Figs 2B and C, and EV1A and B; Note that the YFP levels in the strain expressing RFP‐IDR^Rim4 were generally lower compared to the control in both rich medium and starvation). We conclude that both the RRMs and the IDR are necessary for the repressive function of Rim4 in translation, but neither of them alone is sufficient for this task. Furthermore, the Rim4 IDR is a portable domain harboring the regulatory information to confer starvation‐mediated assembly to a protein.

Figure 2. Rim4 represses translation by forming assemblies that efficiently bind the 5′UTR of target mRNAs.

1. Domain structure of wild‐type Rim4 and the mutants analyzed in the Figure. Numbers indicate the first and the last amino acid residues of the constructs, excluding the V5 tag. The red triangles indicate phenylalanine‐ or tyrosine‐to‐leucine substitutions in the RNP1 and RNP2 sequence stretches to inactivate the RRMs. Note that RRM2 contains only RNP2.
2. SDD‐AGE of the Rim4 mutants and RFP in SPO. Strains used are B48 (meiotic control, G2‐arrested), B1565 (no tag, grown in YPD), B1587, B1949, B2058, B2345, and B2346 (n = 5).
3. Effects of the Rim4 mutants on the expression levels of the reporter gene encoding YFP harboring the CLB3 5′UTR in SPO. Yellow fluorescence was measured by flow cytometry (n = 7). We plotted the individual datapoints and their mean ± standard deviation. YFP transcript levels were evaluated by Northern blot (n = 2).
4. RNA‐IP followed by qPCR to assess the ability of the Rim4 mutants to bind YFP reporter mRNA. We plotted individual datapoints and their median and interquartile range (n = 4, except for B1587, whose n = 3). Strains used in (C) and (D) are B1565, B1587, B1949, B2058, B2345, and B2346.
5. Diagrams of the reporter genes to test the effects of the position of the CLB3 5′UTR on translational repression by Rim4. The short 5′UTR is a shortened version of the CLB3 5′UTR. All constructs contain a pLexO and a CYC1 terminator.
6. RNA‐IP followed by qPCR to assess the ability of Rim4 to bind YFP reporter mRNA encoded by the genes described in (E). We plotted individual datapoints and their median and interquartile range (n = 3).
7. Effects of the position of the CLB3 5′UTR within the transcript on the expression of the reporter gene encoding YFP in SPO. Yellow fluorescence was measured by flow cytometry (n = 4). We plotted the individual datapoints and their mean ± standard deviation. YFP transcript levels were evaluated by Northern blot (n = 2). Strains used in (F) and (G) are B1565, B1587, B1546, B1559, B3528, B3529. In (C) and (G) B2317 is the no reporter control. In the Northern blots, direct staining of the rRNA is shown as a loading control. In (C), (D), and (F) statistical significance was assessed performing a Dunn Kruskal–Wallis multiple comparison. In (G) statistical significance was assessed with a two‐sample Wilcoxon test; *Significant; n.s. = not significant; P‐values are stated in parentheses.

Source data are available online for this figure.

Figure EV1. The Rim4 IDR controls assembly, enabling efficient RNA binding.

1. SDD‐AGE of the Rim4 mutants and RFP in YPD. Strains used are B48 (meiotic control, G2‐arrested), B1565 (no tag, grown in YPD), B1587, B1949, B2058, B2345, and B2346 (n = 5).
2. Effects of the Rim4 mutants on the expression levels of the reporter gene encoding YFP harboring the CLB3 5′UTR in YPD. Yellow fluorescence was measured by flow cytometry (n = 7). We plotted the individual datapoints and their mean ± standard deviation. YFP transcript levels are evaluated by Northern blot (n = 2).
3. PNK assay to assess RNA binding of the Rim4 mutants described in Fig 2 in YPD and SPO. UV‐ and un‐crosslinked samples were run next to each other to evaluate RNA‐specific signal. Rich = rich, glucose‐containing medium; starv. = starvation; FL = full‐length (n = 3). In (A–C) strains used are B1565, B1587, B1949, B2058, B2345, and B2346.
4. RNA‐IP followed by qPCR to assess the RNA binding specificity of Rim4 in YPD and SPO. Strains are B1565 and B1587. We plotted individual datapoints and their median and interquartile range (n = 4, except for B1587 in SPO, whose n = 3). For (B) and (D), statistical significance was assessed performing a Dunn Kruskal–Wallis multiple comparison. *Significant; n.s. = not significant; P‐values are stated in parentheses.
5. Oligo dT retrotranscription followed by PCR to confirm the presence of the CLB3 5′UTR downstream of the stop codon in the short 5′UTR‐YFP‐CLB3 5′UTR reporter mRNA. For the PCR, we used primers annealing in the YFP ORF and in the CYC1 terminator to be able to amplify and compare the 3′ portions of all three reporter transcripts analyzed. We verified the presence of CLB3 5′UTR by inspecting the size of the PCR product. Strains analyzed are B1546, B1559, B1565, B1587, B3528, and B3529.
6. Effects of the position of the CLB3 5′UTR within the transcript on the expression levels of the reporter gene encoding YFP in YPD. Yellow fluorescence was measured by flow cytometry (n = 4). We plotted the individual datapoints and their mean ± standard deviation. Statistical significance was assessed with a two‐sample Wilcoxon test. YFP transcript levels were evaluated by Northern blot (n = 2). Strains used are B1565, B1587, B1546, B1559, B3528, B3529. In (B) and (F), direct staining of the ribosomal RNA (rRNA) is shown as a loading control. B2317 was used as a control to measure fluorescence background (bg).

Source data are available online for this figure.

We wanted to understand why assembly is necessary for Rim4 function. We tested if the assembly state affected the ability of Rim4 to bind RNA with a polynucleotide kinase assay (PNK), which allows to detect direct interactions of proteins with nucleic acids (Fig EV1C). Rim4 bound nucleic acids in starvation, where it was assembled, but not in rich, glucose‐containing medium, where it was monomeric. No signal was detected for RFP‐IDR^Rim4, ruling out the possibility that the IDR contributed to function by engaging directly with the RNA (Järvelin et al, 2016). Binding was abolished also in the 3‐rrm mutant. Taken together, these results show that the RRMs are the sole domains of Rim4 involved in RNA binding. Consistent with this conclusion, Rim4ΔIDR (which contains intact RRMs) exhibited some binding abilities in starvation. This result suggested that assembly is not required for RNA binding, because Rim4ΔIDR was unable to form amyloid‐like assemblies (Fig 2B). We investigated further the RNA binding abilities of Rim4 by RNA‐immunopurification (RNA‐IP) followed by qPCR. This assay allows to detect specific interactions with select transcripts, in contrast to the PNK assay, which detects protein‐RNA interactions unspecifically. Although binding differences measured by RNA‐IP were not statistically significant due to the large variance of the data, the signal trends were clear. We found that Rim4 bound the YFP reporter transcript containing the CLB3 5′UTR in starvation more efficiently than in rich, glucose‐containing medium (Fig EV1D). This interaction was specific, as a non‐target transcript, CUP1‐1, was not comparably bound in starvation. The RNA‐IP assay confirmed the inability of RFP‐IDR^Rim4 and the 3‐rrm mutant to bind the target RNA, but also showed a low signal for Rim4ΔIDR (Fig 2D). The discrepancy between the PNK and the RNA‐IP results relative to Rim4ΔIDR may reveal the ability of Rim4ΔIDR to bind nucleic acids unspecifically, and/or a set of specific but weak interactions that were captured by the PNK assay but not by the RNA‐IP assay (note that, in contrast with the PNK assay, the RNA‐IP assay did not include a crosslinking step with UV). We concluded that the Rim4 IDR is necessary for efficient binding of the target RNA. However, its role is indirect because it does not bind RNA itself. Therefore, we propose that Rim4 assembly, triggered by its IDR, stimulates RRM‐mediated binding of its target RNAs, a requisite for its function as a translational repressor.

The location of the Rim4 binding sites within a transcript is important for translational regulation

Our results showed that assembly is necessary for Rim4 function because it is required for efficient binding to its mRNA targets. We next asked whether any Rim4 binding event is sufficient to repress translation or if the position of this interaction within the transcript also plays a role. To answer this question, we moved the region targeted by Rim4, the CLB3 5′UTR, downstream of the stop codon of the reporter gene. In order to ensure proper translational initiation of this construct, we placed a synthetic short 5′UTR in front of the start codon (Fig 2E). We confirmed that this transcript contained the entire sequence of the CLB3 5′UTR in its 3′ end by retrotranscription followed by PCR (RT–PCR) (Fig EV1E). Rim4 was able to bind the reporter transcript containing the CLB3 5′UTR independently of the position of this sequence within the construct (Fig 2F). As expected, Rim4 did not bind a control reporter transcript lacking the CLB3 5′UTR sequence. We next asked how the location of the CLB3 5′UTR within the reporter transcript influenced its translation. Strikingly, moving the CLB3 5′UTR downstream the stop codon did not result in translational repression of the reporter gene. Rather, the expression levels of this construct were comparable to the ones of the control reporter lacking a CLB3 5′UTR sequence (Figs 2G and EV1F and Appendix Fig S3B and C). The levels of the reporter mRNA containing the moved CLB3 5′UTR were higher in a strain co‐expressing RIM4 compared to its control strain lacking RIM4. Note this difference also in strains expressing the standard reporter gene (with the CLB3 5′UTR in its 5′ end), where translation was efficiently repressed (we often observed this effect, see for example also Fig 1D). Taken together, our results suggest that assembly is necessary to bind the target transcripts, but the ability of Rim4 to repress translation depends on the location of its target sites within the transcript. We propose that Rim4 assemblies prevent translation because they bind the 5′ region of a transcript, likely making the start codon invisible to ribosomes.

Rim4 assembly is controlled by the carbon source

We next wanted to explore why and how Rim4 assembly and function are stimulated by starvation. Yeast meiosis is triggered by the presence of a non‐fermentable carbon source, the absence of a fermentable one, and nitrogen starvation (Honigberg & Purnapatre, 2003). We sought to identify which of these nutrient limitations support Rim4 assembly (Fig 3A). Adding glucose to sporulation medium partially prevented assembly, as monomeric species could be visualized besides assemblies. Rim4 was fully assembled when glucose was replaced with a non‐fermentable carbon source in a rich medium containing abundant nitrogen compounds. Rim4 partially assembled when we combined nitrogen limitation with glucose, while it fully assembled when we combined nitrogen limitation with a non‐fermentable carbon source. We observed translational repression in all conditions where Rim4 was fully or partially assembled (Fig 3B and Appendix Fig S4A). Finally, Rim4 was exclusively monomeric during exponential growth in rich, glucose‐containing medium, but shifted progressively to an assembled state as the culture approached stationary phase (Figs 3C and EV2A). Taken together, our results indicate that the presence of a non‐fermentable carbon source combined with the absence of a fermentable one is the prime nutrient stimulus for Rim4 assembly.

Figure 3. Protein kinase A prevents Rim4 assembly.

1. SDD‐AGE of Rim4 in YPD, YPGly, YPAc, SDP, SGlyP, SAcP, SPO, and SPO + glucose. Strain is B1587 (n = 3).
2. Expression levels of the reporter gene encoding YFP harboring the CLB3 5′UTR in media used in (A). Yellow fluorescence was measured by flow cytometry (n = 4). We plotted the individual datapoints and their mean ± standard deviation. Statistical significance was assessed with a two‐sample Wilcoxon test. *Significant; n.s. = not significant; P‐values are stated in parentheses. YFP transcript levels were evaluated by Northern blot (n = 2 except for YPAc and SAcP, which were analyzed once). Direct staining of the rRNA is shown as a loading control. Strains are B1565, B1587, and B2317.
3. SDD‐AGE of Rim4 in B1587 at different growth stages in YPD (n = 1).
4. A simplified PKA signaling pathway illustration, highlighting crucial nodes discussed in the main text.
5. SDD‐AGE of Rim4 upon inactivation of PKA in a tpk1‐as strain. Strains are B1587 and B2611 (n = 3).
6. Effects of PKA inactivation in YPD on the reporter gene encoding YFP harboring the CLB3 5′UTR. Yellow fluorescence was measured by flow cytometry (n = 3). Data visualization and statistical significance were assessed as in (B). Strains are B1565, B1587, B2581, and B2611. In (D) and (E), strains were treated with 5 μM 1‐NM‐PP1 or just with vehicle (DMSO).
7. SDD‐AGE of Rim4 upon activation of PKA by overexpression of RAS2 ^G19V with 25 μM copper sulfate in SPO. Strains are B1587 and B2787 (n = 3).
8. Effects of RAS2 ^G19V overexpression with 25 μM copper sulfate in SPO on the reporter gene encoding YFP harboring the CLB3 5′UTR. Yellow fluorescence was measured by flow cytometry (n = 4). Data visualization and statistical significance were assessed as in (B). Strains are B1565, B1587, B2786, and B2787. In (A), (C), (E), and (G), G2‐arrested B48 is the meiotic control, and B1565 grown in YPD is the no tag control. In (B), (F), and (H), we measured the fluorescence background (bg) with B2317.

Source data are available online for this figure.

Figure EV2. Protein kinase A prevents Rim4 assembly – further strain characterization.

1. Top, growth curve of B1587 in YPD containing 25 μM copper sulfate (Cu⁺⁺). Cells were sampled to evaluate Rim4‐V5 levels and assembly (in Fig 3C) at optical densities marked in red. Bottom, SDS–PAGE for Rim4‐V5 levels at different growth stages in YPD. Direct staining of the total protein extracts (Ponceau) and blotting of Pgk1 are shown as loading controls (n = 1).
2. RNA‐IP followed by qPCR to assess the ability of Rim4 to bind the YFP reporter mRNA harboring the CLB3 5′UTR in a tpk1‐as strain. We plotted individual datapoints and their median and interquartile range (n = 3). Statistical significance was assessed with a Dunn Kruskal–Wallis multiple comparison. *Significant; n.s. = not significant; P‐values are stated in parentheses. Strains B1565, B1587, B2581, and B2611 were treated with 5 μM 1‐NM‐PP1.
3. Effects of PKA inactivation in a tpk1‐as strain in SPO on the reporter gene encoding YFP harboring the CLB3 5′UTR. Yellow fluorescence was measured by flow cytometry (n = 3). We plotted the individual datapoints and their mean ± standard deviation. Statistical significance was assessed with a two‐sample Wilcoxon test. Strains B1565, B1587, B2581, and B2611 were treated with 5 μM 1‐NM‐PP1 or just with vehicle (DMSO).
4. Two versions of a synthetic gene encoding HA‐RAS2 ^G19V were integrated in strains carrying the RIM4‐V5 and YFP constructs discussed in Fig 1A. In one version, HA‐RAS2 ^G19V transcription is controlled by LexA‐ER‐B112, activated by β‐estradiol, in the other version HA‐RAS2 ^G19V transcription is controlled by copper.
5. SDD‐AGE of Rim4 upon activation of PKA by overexpression of RAS2 ^G19V with 1 μM β‐estradiol in SPO. Strains used are B48 (meiotic control, G2‐arrested), B1565 (no tag, grown in YPD), B1587, and B2666 (n = 3).
6. Effects of RAS2 ^G19V overexpression with 1 μM β‐estradiol in SPO on the reporter gene encoding YFP harboring the CLB3 5′UTR. Yellow fluorescence was measured by flow cytometry (n = 4). Data visualization and statistical significance were assessed as in (C). Strains are B1565, B1587, B2665, and B2666. In (C) and (F), B2317 was used as a control to measure fluorescence background (bg).

Source data are available online for this figure.

PKA prevents Rim4 assembly

The observation that Rim4 assembly depends primarily on carbon source quality suggests a mechanism to couple Rim4 function with meiotic entry. In yeast, glucose activates the Ras/PKA signaling pathway, which promotes cell growth and prevents meiotic entry by blocking expression and activation of IME1, the master regulator of meiotic initiation (Honigberg & Purnapatre, 2003; Broach, 2012) (Figs 3D and EV3A). Moreover, inactive PKA is required for meiotic commitment (Gavade et al, 2022). We asked if active PKA prevents Rim4 assembly in our mitotic assay. To test this hypothesis, we used a tpk1‐as strain, where PKA can be inactivated by adding the ATP analog 1‐NM‐PP1 to the cell culture (Bishop et al, 2001; Weidberg et al, 2016). Rim4 assembled upon PKA inactivation in rich, glucose‐containing medium (Fig 3E), bound the reporter transcript (Fig EV2B), but did not repress its translation (Fig 3F). As a control, we verified that the tpk1‐as strain was able to repress translation in starvation in the presence of 1‐NM‐PP1 (Fig EV2C and Appendix Fig S4B). To further investigate the role of PKA in Rim4 function, we performed a complementary experiment, where we activated PKA during starvation. We engineered a β‐estradiol‐inducible RAS2 ^G19V synthetic gene, encoding a constitutively active protein analogous to the mammalian transforming mutant H‐ras ^G12V (Kataoka et al, 1984; Toda et al, 1985) (Fig EV2D and Appendix Fig S4C). Here, we induced RIM4 and RAS2 ^G19V in rich, glucose‐containing medium before starving the cells (Appendix Fig S2). This induction scheme ensured that cells already accumulated both Rim4 and Ras2^G19V before starvation, avoiding possible consequences of different induction kinetics of the expression of these two proteins. Cells expressing RAS2 ^G19V during starvation accumulated monomeric Rim4 (Fig EV2E). Strikingly, the reporter gene was de‐repressed in these cells (Fig EV2F and Appendix Fig S4D for results in rich, glucose‐containing medium). Note that the induction scheme we used for this experiment also caused the induction of the YFP reporter before starving the cells. Consequently, the magnitude of the repression levels of the wild‐type strain we used as a control in this experiment was lower compared to other experiments where the reporter gene was induced after shifting the cells to starvation. To avoid this shortcoming, we constructed strains where RAS2 ^G19V overexpression was controlled by copper (Fig EV2D and Appendix Fig S4E). Rim4 did not assemble nor repressed the translation of the reporter gene, confirming the data obtained with the β‐estradiol‐inducible RAS2 ^G19V allele (Fig 3G and H and Appendix Fig S4F). Taken together, our results demonstrate that the PKA signaling pathway transduces the carbon source input into Rim4 assembly. Moreover, prevention of assembly in starvation also prevents translational repression.

Figure EV3. Rim4 assembly and function do not depend on Msn2, Msn4, Rim15, or Yak1.

1. A simplified illustration depicting how the PKA signaling pathway controls meiotic entry, highlighting crucial nodes discussed in the main text.
2. SDD‐AGE to evaluate the assembly state of Rim4 in MSN2 and MSN4 double deletion, or in RIM15, YAK1 single deletions strains in YPD and SPO. Strains used are B48 (meiotic control, G2‐arrested), B1565 (no tag, grown in YPD), B1587, B2707, B2960, and B3067 (n = 1).
3. Effects of MSN2 and MSN4 double deletion, or RIM15, YAK1 single deletions on the expression levels of the reporter gene encoding YFP harboring the CLB3 5′UTR in YPD (left) and SPO (right). Strains are B1565, B1587, B2704, B2707, B2960, B3008, B3066, B3067. B2317 was used as a control to measure fluorescence background (bg). Yellow fluorescence was measured by flow cytometry (n = 3). We plotted the individual datapoints and their mean ± standard deviation. Statistical significance was assessed with a two‐sample Wilcoxon test. *Significant; n.s. = not significant; P‐values are stated in parentheses.

Source data are available online for this figure.

The Rim4 IDR is a direct target of PKA

To understand how PKA controls Rim4 assembly, we tested the involvement of downstream targets of this kinase. We focused on those targets that are negatively regulated by PKA and promote meiotic entry or cell quiescence (Honigberg & Purnapatre, 2003; Broach, 2012) (Fig EV3A). Neither MSN2 and MSN4 double deletion, nor RIM15, nor YAK1 single deletions affected Rim4 assembly or function in our haploid strains (Fig EV3B and C and Appendix Fig S5A). We noticed that Rim4 contains six consensus sequences for PKA (R‐R/K‐X‐S*/T*) (Kennelly & Krebs, 1991), four of these located in its IDR (Fig 4A). We asked whether these sites were phosphorylated in media containing glucose. We immunopurified Rim4 and probed with an antibody that recognizes phospho‐PKA substrates. We found that at least one of these PKA sites was phosphorylated in a glucose‐dependent fashion (Fig 4B). Moreover, Rim4 was phosphorylated when PKA was ectopically activated, but it was not upon PKA inactivation, confirming that these were PKA‐dependent phosphorylation events (Fig EV4A and B). Finally, we observed phosphorylation only in media where Rim4 was monomeric or partially assembled. Our results indicate that Rim4 is phosphorylated by PKA and that this phosphorylation negatively correlates with assembly.

Figure 4. Rim4 is phosphorylated in a glucose‐dependent fashion.

1. Domain structure of Rim4 highlighting the position of critical serine and threonine residues that may be targeted by PKA. PKA consensus sites are in green. Non‐consensus sites whose phosphorylation correlates with PKA activity are highlighted in red, while those anti‐correlating are highlighted in blue.
2. Evaluation of Rim4 phosphorylation by PKA in YPD, YPGly, YPAc, SDP, SGlyP, SAcP, SPO, and SPO + glucose. Strains are B1565 and B1587. Gluc. = glucose; glyc. = glycerol; acet. = acetate; (biological replicates = 2 except for YPAc and SAcP, which were analyzed once).
3. Evaluation of phosphorylation by PKA of the Rim4 mutants described in Fig 2 in YPD and SPO. Strains are B1565, B1587, B1949, B2058, B2345, and B2346 (n = 3). In (B) and (C) Pgk1 was blotted to evaluate the quality of V5 immunopurification.
4. SDD‐AGE of Rim4 mutants carrying serine‐ or threonine‐to‐alanine or serine‐ or threonine‐to‐glutamic acid substitutions in potential PKA‐dependent sites in YPD and SPO. Strains are B1587, B2702, B2951, B2983, B3592, B3593, B3594, B3595, and B3596. Controls are B48 (meiotic, G2‐arrested) and B1565 (no tag, grown in YPD) (n = 3).
5. Yellow fluorescence of B1565, B1587, B2702, B2951, B2983, B3592, B3593, B3594, B3595, and B3596 strains was measured by flow cytometry (n = 3). We plotted the individual datapoints and their mean ± standard deviation. Statistical significance was assessed performing a Dunn Kruskal–Wallis multiple comparison; n.s. = not significant; P‐values are stated in parentheses. We used B2317 to measure fluorescence background (bg).

Source data are available online for this figure.

Figure EV4. Rim4 is phosphorylated in a PKA‐dependent fashion – further strain characterization.

1. Evaluation of Rim4 phosphorylation upon inactivation of PKA in a tpk1‐as strain in YPD and SPO. Strains were treated with 5 μM 1‐NM‐PP1 or just with vehicle (DMSO). Strains are B1565, B1587, and B2611 (n = 2).
2. Left, evaluation of Rim4 phosphorylation upon activation of PKA by overexpression of RAS2 ^G19V with 1 μM β‐estradiol in YPD and SPO. Strains are B1565, B1587, and B2666 (n = 2). Right, evaluation of Rim4 phosphorylation upon activation of PKA by overexpression of RAS2 ^G19V with 25 μM copper sulfate in YPD and SPO. Strains are B1565, B1587, and B2787 (n = 2).
3. Phosphorylation by PKA of Rim4 harboring serine‐to‐alanine or serine‐to‐glutamic acid substitutions in positions S429, S470, S525, and S612 (n = 3). Strains are B1565, B1587, B2951, and B2983. In (A–C) Pgk1 was blotted to evaluate the quality of Rim4‐V5 immunopurification.
4. List of peptides identified as differentially phosphorylated in a PKA‐dependent manner by mass spectrometry. Phosphorylated residues are highlighted in red, PKA consensus sites are underlined.
5. List of amino acid substitutions introduced on Rim4 to study the effect of PKA‐dependent phosphorylation events on Rim4 assembly and function.

Source data are available online for this figure.

We inspected phosphorylation of the Rim4 mutant constructs described in Fig 2, as they allowed us to differentiate between PKA sites located in the RRMs or in the IDR (Fig 4C). RFP‐IDR^Rim4, encompassing three out of four sites of the IDR (S470, S525, and S612), had the same phosphorylation patterns as wild‐type Rim4. In contrast, we did not observe phosphorylation of Rim4ΔIDR, which encompasses the T216, S367, and S429 sites. Moreover, inactivation of the RRMs did not alter the phosphorylation pattern of the 3‐rrm mutant, informing us that the RNA binding ability of Rim4 did not influence these modifications.

Our results suggested a possible direct role of PKA in preventing Rim4 assembly, because its IDR contains the elements necessary for regulated assembly and is phosphorylated in a PKA‐dependent manner. We substituted the serine of the PKA sites at position 429, 470, 525, and 612 with alanine (Rim4^4A) or glutamic acid (Rim4^4E), to prevent or mimic phosphorylation events, respectively. Neither Rim4^4A nor Rim4^4E were detected by the antibody recognizing phospho‐PKA substrates (Fig EV4C). These substitutions did not affect Rim4 behavior, suggesting that the PKA sites in the Rim4 IDR are not necessary nor sufficient to control Rim4 assembly and function in our mitotic assay (Fig 4D and E and Appendix Fig S5B). We obtained similar results for mutants where all six PKA sites where mutated (Rim4^6A and Rim4^6E). We asked if PKA regulates Rim4 by phosphorylating other sites outside the consensus sequences, or indirectly by controlling the activity of other proteins. Therefore, we performed mass spectrometry on immunopurified Rim4 from cells grown in rich, glucose‐containing medium, starvation, or carrying mutations to ectopically activate or inactivate PKA. We observed PKA‐dependent phosphorylation of the S470, S525, and S612 PKA sites located in the IDR (Dataset EV1). Some serine residues near to S470 and S525 (S473, S532, S534, and S535) were also phosphorylated in a PKA‐dependent manner (Fig 4A). Moreover, we identified two sites in the N‐terminus of Rim4, S10 and S16, whose phosphorylation patterns anticorrelated with PKA activity, suggesting a possible indirect role of PKA (Fig EV4D). We constructed two groups of Rim4 mutants, where all the “non‐consensus” sites detected by mass spectrometry were substituted with alanine or glutamic acid residues in combination with mutations in all six PKA sites (Rim4^2E+10A, Rim4^2A+10E) or only in the four PKA sites located in the IDR (Rim4^2E+8A, Rim4^2A+8E, Fig EV4E). No effects of these sets of mutations were observed on Rim4 assembly (Fig 4D) nor on Rim4 function, except for a slight reduction of reporter accumulation in cells expressing Rim4^2E+10A in glucose‐containing medium (Fig 4E and Appendix Fig S5B). We conclude that the PKA‐dependent phosphorylation events we detected by mass spectrometry are not sufficient nor necessary to control Rim4 assembly and function in the conditions tested. Therefore, we hypothesize that additional factors belonging to the Ras/PKA signaling pathway must be involved in Rim4 assembly regulation.

Rim4 phosphorylation by PKA contributes to Rim4 assembly clearance

Our results indicated that phosphorylation by PKA is not sufficient to prevent Rim4 assembly and function. Instead, we asked if phosphorylation is important for assembly clearance. PKA becomes active during return to growth, a developmental process occurring when yeast meiosis is interrupted upon exposure to glucose (Honigberg & Purnapatre, 2003; Simchen, 2009). We asked how phosphorylation of the PKA sites located in the Rim4 IDR impacted this process using diploid strains harboring the RIM4 ^4A or RIM4 ^4E alleles in the RIM4 endogenous locus. Strikingly, the Rim4^4A protein levels persisted longer compared to wild type or Rim4^4E upon transfer to a medium containing glucose (Fig 5A). Accordingly, Rim4^4A assemblies disappeared slowly compared to Rim4^4E or the wild type (Fig EV5A). The Rim4^4A strain exhibited a subtly delayed return to growth compared to the wild‐type and Rim4^4E strains, suggesting that persistent Rim4 assemblies negatively influence growth (Fig 5A). To test this possibility, we recorded growth curves of haploid cells after overexpressing RIM4 for 4.5 h in starvation (Fig 5B). The RIM4 ^4A mutant strain grew slowly compared to rim4Δ and the RIM4 wild‐type strains, while the RIM4 ^4E strain grew similarly to the wild type. Minor differential growth of the RIM4 ^4A strain was observed in control conditions, after RIM4 overexpression in rich, glucose‐containing medium (Fig EV5B). Taken together, these results show that phosphorylation by PKA contributes to efficient entry into exponential phase of meiotic or starved cells that accumulated Rim4, likely helping clearance of these assemblies.

Figure 5. Rim4 phosphorylation by PKA contributes to Rim4 clearance.

1. Left, schematics illustrating the return to growth experiment. Right, SDS–PAGE to evaluate the levels of Rim4‐V5, Rim4^4A‐V5, and Rim4^4E‐V5. Direct staining of the total protein extracts (Ponceau) and blotting of Pgk1 are shown as loading controls. Strains used are B47 (no tag, meiotic control at hour 5.5 in SPO), B2947, B3235, and B3236 (n = 3). Bottom, growth curves of B2947, B3235, and B3236 upon transfer to YPD. Depicted are mean (full color) ± standard deviation (semi‐transparent color) of technical quadruplicates of one out of two biological replicates.
2. Left, schematics of RIM4‐V5 induction with 25 μM copper sulfate and media handling to study the effects of Rim4 phosphorylation by PKA on growth upon accumulation of Rim4 in starved mitotic cells. Right, growth curves of B1565, B1587, B2951, and B2983 in YPD were recorded in triplicates. Depicted are mean (full color) ± standard deviation (semi‐transparent color) of technical triplicates of one out of three biological replicates.
3. Rim4 phosphorylation during meiosis upon inactivation of PKA. Strains were treated with 5 μM 1‐NM‐PP1 or just with vehicle (DMSO) when transferred to SPO at hour 0. Pgk1 was blotted to evaluate the quality of Rim4‐V5 immunopurification. Strains used are B47 (no tag, meiotic control at hour 5.5 in SPO), B2947, and B3060 (n = 3).

Source data are available online for this figure.

Figure EV5. Characterization of the phosphorylation of the PKA sites of Rim4 IDR.

1. SDD–PAGE of Rim4‐V5, Rim4^4A‐V5, and Rim4^4E‐V5 during early sporulation and return to growth. Strains used are B48 (full‐length control, G2‐arrested), A35408 (Rim4ΔIDR, G2 arrested), B47 (no tag, 5.5 h in SPO), B2947, B3235, and B3236 (n = 1).
2. Left, schematics of transient RIM4‐V5 induction in YPD with 25 μM copper sulfate in mitotic cells. Right, growth curves of B1565, B1587, B2951, and B2983 in YPD were recorded in triplicates. Depicted are mean (full color) ± standard deviation (semi‐transparent color) of technical triplicates of one out of three biological replicates.
3. A synthetic gene encoding IME2 ^st ‐HA, induced by β‐estradiol with the LexA‐ER‐B112 transcription factor, was integrated in a strain carrying the copper‐induced RIM4‐V5 construct.
4. SDS–PAGE for the protein levels of Rim4‐V5 and Ime2^st‐HA. IME2 ^st ‐HA was overexpressed with 1 μM β‐estradiol. Direct staining of the total protein extracts (Ponceau) and blotting of Pgk1 are shown as loading controls. The asterisk indicates a residual Pgk1 signal from previous incubation with anti‐Pgk1 antibody. Strains are B1565, B1587, B3090, and B1 (no tag control) (n = 3).
5. Evaluation of the phosphorylation state of the PKA sites of Rim4 upon IME2 ^st ‐HA overexpression with 1 μM β‐estradiol. Note that cells expressing Ime2^st‐HA in glucose‐containing medium failed to phosphorylate Rim4 in a PKA‐dependent fashion probably as a consequence of the acute toxicity caused by the ectopic expression of IME2 ^st (Sari et al, 2008). Strains are B1565, B1587, and B3090. Pgk1 was blotted to evaluate the quality of Rim4‐V5 immunopurification (n = 3).
6. SDS–PAGE to evaluate the activity of PKA in a tpk1‐as strain during meiosis. B3060 was treated with 5 μM 1‐NM‐PP1 or just with vehicle (DMSO) upon transfer to SPO at hour 0. Note the general increase of signal in the B3060 + vehicle lanes from hour 6. The arrow at hour 6 indicates when vehicle‐treated cells start meiosis I. Direct staining of the total protein extracts is shown as a loading control (Ponceau) (n = 3).

Source data are available online for this figure.

Previous work showed that during meiosis 14‐3‐3 proteins bind to the phosphorylated S525 and S612 PKA sites of Rim4 and recruit the kinase Ime2, which facilitates assembly clearance (Herod et al, 2022). We wanted to identify which kinase is responsible for the phosphorylation of these sites. We asked if Ime2 phosphorylated the PKA sites of Rim4 by ectopically expressing the stabilized and constitutively active allele IME2 ^st (Sari et al, 2008) in our mitotic assay (Fig EV5C and D). We could exclude this possibility, as we did not detect any increase in Rim4 phosphorylation in starvation upon IME2 ^st overexpression (Fig EV5E). Instead, we observed an increase of PKA activity at meiosis I onset and a concomitant PKA‐dependent phosphorylation of Rim4 (Figs 5C and EV5F). Taken together, our results suggest that PKA‐dependent phosphorylation plays a role in Rim4 assembly clearance in both return to growth and meiotic progression.

Discussion

In this study, we investigated how yeast cells regulate amyloid‐like assembly of the meiotic protein Rim4 and why assembly is necessary for Rim4 function. We reconstituted Rim4 biology in mitotic haploid cells by ectopically expressing RIM4, demonstrating that no additional meiotic factors are required for its assembly and function. Failure to form assemblies, whether because of growth conditions or mutations, results in the inability of Rim4 to repress translation. Our results support a model where starvation‐induced assembly stimulates RNA binding (Fig 6A). How does assembly favor interactions with RNA, given that the Rim4 IDR does not bind RNA directly? The formation of an amyloid‐like structure may stimulate RNA binding through avidity or cooperativity because it increases the local concentration of the linked RRMs. Moreover, the highly ordered core of the assembly may constrain the RRMs into orientations that favor RNA binding (Riek & Eisenberg, 2016; Wu & Fuxreiter, 2016; Ottoz & Berchowitz, 2020; Hallegger et al, 2021). We also found that the ability of Rim4 to repress translation depends on where it binds in the transcripts. Therefore, Rim4 assemblies do not repress translation by simply “englobing” the RNA. Rather, repression depends on the formation of a polarized RNA‐protein structure where the orientation of the interactions between the amyloid‐like assembly and the substrate is crucial (Gu et al, 2021). Rim4 binds the 5′UTR of its target RNAs, suggesting that their cap is buried into the assemblies. Ribosomes cannot easily reach the start codon of these mRNAs because diffusion is limited in these dense aggregates. Moreover, in this binding configuration, the 3′ end of the transcripts protrudes in the aqueous phase disfavoring ribosome recycling upon termination and further contributing to translational repression (Gu et al, 2021). Our results raise the possibility that Rim4 assembly and function may become uncoupled upon ectopic PKA inactivation. Ectopic PKA inactivation causes acute toxicity and eventually cell death (Weidberg et al, 2016), with possible secondary effects on Rim4 function. Therefore, care should be taken when interpreting these results. Nevertheless, we do not want to exclude the possibility that assembly‐dependent RNA binding might not be sufficient for translational repression. Additional unknown factor(s) may be required for Rim4 function, as it has been demonstrated for Orb2, an amyloidogenic protein controlling translation implicated in long‐term memory (Khan et al, 2015).

Figure 6. Cells couple assembly and function of Rim4 with nutrient conditions.

1. A model for Rim4 amyloid‐like assembly formation and function. Intermediates and/or targets of the Ras/PKA signaling pathway and other growth‐dependent factors engage with Rim4 IDR, preventing assembly. When PKA is not active, Rim4 “collapses” into amyloid‐like assemblies that organize the target mRNAs into a polarized RNA‐protein structure inaccessible to the ribosomes, efficiently preventing translation.
2. A model for how PKA promotes Rim4 assembly clearance in the meiotic context. PKA gets activated at meiosis I onset and promotes 14‐3‐3 proteins recruitment to Rim4 by phosphorylating critical sites on the IDR. In turn, 14‐3‐3 proteins recruit Ime2 which phosphorylates Rim4 IDR at multiple sites, promoting assembly clearance and contributing to transition from meiosis I to II. If cells encounter glucose before meiotic commitment, phosphorylation by PKA facilitates Rim4 turnover, contributing to efficient entry in exponential phase.

The Rim4 IDR confers starvation‐dependent assembly to a fusion protein, similar to the IDRs of the yeast pyruvate kinase Cdc19 and the translational repressor Sbp1 (Saad et al, 2017). Signaling pathways control assembly of disordered proteins by modulating their IDR accessibility through post‐translational modifications (PTM) and/or interactions with other factors (Cereghetti et al, 2018; Hofweber & Dormann, 2019). Here, we established that the Ras/PKA signaling pathway negatively controls Rim4 amyloid‐like assembly. While Rim4 is phosphorylated in a PKA‐dependent fashion, our data do not support an essential role of this PTM in preventing Rim4 assembly. Instead, we propose that unknown target(s) of this kinase control this process, possibly by establishing protein–protein interactions (Fig 6A). Factors promoting stationary phase or the stress response, which are negatively regulated by PKA, are not necessary for Rim4 assembly in starvation. Therefore, we suggest that, when PKA is inactive, Rim4 may “collapse” into its assembled state by default. We showed that a mutant carrying mutations in all PKA consensus sites and PKA‐dependent phospho‐sites detected by mass spectrometry (Rim4^2E+10A) exhibits moderate repression abilities in glucose‐containing medium. Therefore, we do not exclude a minor role of PKA‐dependent phosphorylation in Rim4 function. Moreover, a recent work proposed that, during meiosis, 14‐3‐3 proteins compete with the mRNA for Rim4 binding via the two PKA sites (T216 and S367) located in the RRMs (Zhang et al, 2023). Our mitotic data do not support this model, as Rim4^6A, which carries all PKA sites mutated into alanine, does not repress translation in rich, glucose‐containing medium. However, in these conditions, Rim4^6A is not assembled, corroborating the idea that assembly is necessary for function. Therefore, further experiments are needed to better characterize the role of 14‐3‐3 in Rim4 function.

Although phosphorylation by PKA does not prevent Rim4 assembly, we demonstrated that this PTM becomes relevant for assembly clearance during meiosis (Fig 6B). Inactive PKA promotes meiotic entry and ensures assembly as soon as Rim4 gets expressed. If cells encounter glucose before meiotic commitment, they exit the meiotic program and return to growth (Simchen, 2009). Here, cells need to rapidly modify their proteome to enter the cell cycle. We demonstrated that Rim4 assemblies represent a burden for these cells and that phosphorylation by PKA is instrumental in facilitating their clearance and entry into exponential phase. Phosphorylation by PKA is relevant also during meiotic progression once the cells are committed. Early work reported that cAMP levels increase at the onset of the meiotic divisions, suggesting a concomitant increase of PKA activity (Uno et al, 1985). Accordingly, we observed that Rim4 becomes phosphorylated in a PKA‐dependent fashion. Similarly, a recent report showed that Rim4 is phosphorylated by PKA during meiosis (Zhang et al, 2023). We propose that PKA is the kinase responsible for the priming phosphorylation events at positions S525 and S612 required to recruit 14‐3‐3 proteins, which in turn recruit Ime2 (Herod et al, 2022). Therefore, PKA indirectly facilitates Rim4 phosphorylation by Ime2, contributing to Rim4 disassembly and clearance at meiosis II onset (Carpenter et al, 2018).

In this work, we established that starvation stimulates the formation of Rim4 amyloid‐like assemblies as a consequence of the Ras/PKA signaling pathway inactivation. Several biomolecular condensates, including Cdc19, processing bodies, and stress granules form during starvation (Saad et al, 2017; Franzmann & Alberti, 2019; Guzikowski et al, 2019). So far, PKA has been directly involved in processing bodies regulation (Ramachandran et al, 2011). How many other condensates and IDRs are controlled by this master regulator of growth? Functionally related IDRs exhibit conserved molecular features, such as PTM and/or binding sites (Zarin et al, 2019). The identification of the conserved features in the IDRs and their interactions with Ras/PKA and other nutrient signaling pathways will extend our general understanding of the relationship between starvation and the formation, clearance, and function of biomolecular condensates.

Materials and Methods

All chemicals, unless stated, were purchased from Sigma Aldrich.

Plasmid construction

We used Escherichia coli DH5α for plasmid preparation (Life Technologies, Invitrogen). We obtained a destabilized YFP reporter by fusing the ADH1 tail to CitrineA206K (Gordon et al, 2007). The copper‐inducible RIM4‐V5 constructs, the LexA‐ER‐B112 transcription factor, and the β‐estradiol‐inducible YFP reporter genes were cloned in the pRG integrative vectors (Gnügge et al, 2016). The inducible HA‐RAS2 ^G19V and IME2 ^st ‐HA synthetic genes were cloned into p275, an integrative vector for the TRP1 locus we constructed for this study. In our SK1 wild‐type strains, the region of the TRP1 ORF between +255 and +285 (where +1 is A of the start codon) is substituted with hisG. Our p275 vector is integrated into the TRP1 locus with a double cross‐over recombination mechanism, similar to the pRG integrative vector series, and contains a multicloning site 383 nt downstream of the TRP1 stop codon (where 1 is G of the TAG stop codon). The verified all our constructs by Sanger sequencing (Genewiz, Azenta Life Sciences). Appendix Table S2 lists all plasmids used in this work.

Yeast strain construction

All strains used in this work were derived from Saccharomyces cerevisiae SK1 (Kane & Roth, 1974) with standard transformation methods (Gietz et al, 1995), crosses, and dissections. Each yeast modification introduced by transformation was confirmed by colony PCR. All transformants were backcrossed. All strains constructed for this work are flo8Δ and prototroph. FLO8 ORF was deleted in a marker‐less fashion. We substituted the FLO8 ORF with the Candida albicans URA3 cassette from pRG206MX. This knock‐out strain was transformed with a PCR product fusing together the FLO8 promoter and terminator. Marker‐less flo8Δ transformants were isolated by counterselection on 5′ FOA. Prototrophies for HIS3, LEU2, and URA3 were obtained by integrating the pRG vectors, while for TRP1 we integrated a PCR product reconstituting the Trp+ phenotype or our p275‐derived constructs. All haploid strains used for the mitotic experiments are rim4Δ, where the RIM4 ORF was deleted in a marker‐less fashion like FLO8. For the meiotic time courses, we introduced RIM4‐V5, RIM4 ^4A‐V5, and RIM4 ^4E‐V5 in the RIM4 locus in a marker‐less fashion using CRISPR/Cas9. We transformed a strain where the RIM4 ORF was substituted with Candida albicans URA3 with a pCas plasmid derived from Ryan et al (2016) and the repair template. The pCas plasmid contained the sequence 5′‐ACCACCAACCAAGAGCCAAG‐3′ to target the URA3 ORF. These modifications were confirmed by Sanger sequencing. TPK3, MSN2, MSN4, RIM15, and YAK1 knock‐out strains were obtained by substituting the ORFs with the kanMX6 and NATMX6 cassettes, encoding resistance to G418 (Geneticin) and Nourseothricin, respectively. For the tpk1‐as strains, we derived the tpk1::tpk1M164G and tpk2::kanMX6 alleles from (Weidberg et al, 2016), but we constructed a new TPK3 knock‐out allele with a more convenient marker for our experiments (tpk3::NATMX). Appendix Table S3 lists all yeast strains used.

Growth conditions

We grew our strains at 30°C, liquid cultures were shaken vigorously. Appendix Fig S2 illustrates our mitotic induction schemes. Haploid strains were grown on YPGly plates overnight and cultured in liquid YPD to reach saturation. Saturated cultures were diluted in fresh YPD to resume exponential growth. After two doublings, cells were washed with water once and transferred to fresh medium at a defined OD, aiming at getting all cultures at similar ODs at sampling. One hour after transfer to fresh medium, we induced the cultures with 25 μM copper sulfate and 1 μM β‐estradiol. After 3 h of induction, we sampled the cells. This induction scheme was modified for experiments evaluating ectopic PKA or Ime2^st activity. For experiments with strains containing the β‐estradiol‐inducible HA‐RAS2 ^G19V or IME2 ^st ‐HA alleles, or copper‐inducible HA‐RAS2^G19V , we induced with 25 μM copper sulfate and 1 μM β‐estradiol or only with 25 μM copper sulfate, respectively, after about one doubling time after diluting the saturated YPD pre‐cultures in fresh YPD. Therefore, cells were induced in YPD before the water wash and transfer to fresh YPD or SPO containing 25 μM copper sulfate and 1 μM β‐estradiol. We sampled 4 h after media transfer. For mitotic experiments involving the tpk1‐as strains we induced only with 25 μM copper sulfate before media transfer. After washing the cells with water, we transferred them to fresh YPD or SPO containing 25 μM copper sulfate, 1 μM β‐estradiol, and 5 μM 1‐NM‐PP1 or the corresponding volume of DMSO (vehicle). We sampled 3 h after media transfer. See Appendix Tables S1 and S4 for media recipes. To evaluate Rim4 assembly during exponential growth in YPD, B1587 was first grown on YPGly plates overnight, then cultured in liquid YPD overnight, making sure that cells were in exponential phase all the time. The next morning, the exponential culture was diluted to OD 0.25 in YPD and induced with 25 μM copper sulfate. We monitored growth by measuring the OD of the culture every 15 min until it reached saturation.

For meiotic time courses and return to growth, diploid strains were grown on YPGly plates overnight, transferred to YPD plates containing 4% glucose overnight, and cultured in liquid YPD overnight until they reached saturation. Strains were then transferred to BYTA, grown overnight, washed once with water, and transferred to SPO at OD 1.8. For experiments involving the tpk1‐as strain, we added 5 μM 1‐NM‐PP1 or the corresponding volume of DMSO at hour 0 in SPO. Cells were sampled every hour. After 24 h, we verified that all our strains underwent meiosis by checking for the presence of tetrads. For return to growth experiments, after 3.5 h in SPO, cells were washed once with water and transferred to fresh YPD at OD 1.8.

Flow cytometry

Approximately 2 OD of cells were fixed in 1% formaldehyde in PBS for 30 min at room temperature, washed once, and resuspended in PBS. We used a BD LSRFortessa cell analyzer (BD Biosciences), using a 488 nm excitation laser and a 525/50 nm emission filter. We recorded 10,000 events per sample using the FACSDiva software (BD Biosciences). We used the software R (www.r‐project.org) with the Bioconductor package (www.bioconductor.org) to analyze the data. We gated for un‐budded and budded cells together, which were identified by visualizing the signal area of forward scatter and side scatter plots (Ottoz et al, 2014). We measured the yellow fluorescence background levels in all growth conditions tested by analyzing a strain lacking the YFP reporter. For each replicate, we calculated the median fluorescence level of the gated fluorescence events. We then calculated the mean and the standard deviation of the medians of the biological replicates. In the figures, we represented the means as bar plots and the standard deviation as error bars.

Protein methods

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblots

We resuspended pellets containing approximately 7.2 OD of cells in 5% Trichloroacetic acid (TCA, Fisher Chemical), washed once with acetone, and air‐dried. We added SDS–PAGE lysis buffer (1× TE (10 mM Tris HCl pH 8, 1 mM EDTA), 1× Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific), 2.75 mM DTT, 10 mM Tris HCl pH 11) and glass beads, acid washed (425–600 μm dia.) to the pellets and processed them in a FastPrep‐24 Homogenizer (MP Biomedical) machine for 45 s at 6.5 m/s. We added 3× SDS–PAGE loading buffer (187 mM Tris HCl pH 6.8, 30% glycerol, 9% SDS, 6% β‐mercaptoethanol, 0.05% Bromophenol Blue) to a final 1× concentration, adjusted pH with Tris HCl pH 11, boiled samples for 5 min, and spun for 5 min at 16,000 g.

We performed SDS–PAGE using BioRad materials and systems and Criterion TGX Precast Gradient Gels 4–15%. We used Precision Plus Protein Dual Color Standard as ladder. We transferred proteins with a Trans‐Blot Turbo System using Trans‐Blot Turbo Transfer Stacks and 0.2 μm nitrocellulose soaked in 1× Trans‐Blot Turbo Transfer Buffer with 20% ethanol.

We visualized total transferred protein extracts using a Ponceau S staining solution (Thermofisher). Membranes were then cut to incubate different size ranges with different antibodies. We blocked the membranes in 3% milk 1× TBS‐T (10 mM Trizma Base, 150 mM NaCl, 0.1% Tween 20, pH 7.6) for 30 min at room temperature. Appendix Table S5 lists the antibodies and dilutions used to blot constructs tagged with V5, HA‐tagged Ras2^G19V, HA‐tagged Ime2^st, Pgk1, YFP, and phospho‐PKA substrates. We visualized protein signal using the Amersham ECL Prime Western Blotting Detection Reagent (Cytiva) and an Amersham Imager 600 (GE Healthcare).

Semi‐denaturing detergent agarose electrophoresis (SDD‐AGE)

We adapted our SDD‐AGE protocol from Bagriantsev et al, 2006 and Halfmann & Lindquist, 2008. All steps were performed at 4°C. Snap frozen pellets containing approximately 3.6 OD of cells were resuspended in SDD‐AGE lysis buffer (100 mM Tris HCl pH 8.0, 20 mM NaCl, 2 mM MgCl₂, 1% Triton‐X, 50 mM β‐mercaptoethanol, 2× Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific)) and zirconia beads (0.5 mm dia. Zirconia/Silica, BioSpec Products) and processed in a FastPrep‐24 Homogenizer machine for 45 s at 6.5 m/s. We spun the protein extracts at 2,500 g for 10 min, transferred the supernatant to a fresh tube, spun again for 5 min, transferred the supernatant, added 4× SDD‐AGE loading buffer (0.5× TAE, 20% glycerol, 8% SDS, and 0.05% bromophenol blue, where 10× TAE is 0.4 M Trizma Base, 1.14% glacial acetic acid, 10 mM EDTA) to a final 1× concentration, and incubated the extracts for 10 min at room temperature. We loaded a 1.7% agarose gel in 0.5× TAES (0.5× TAE with 0.1% SDS) and run at low voltage overnight at 4°C.

We performed capillary transfer overnight at room temperature on an Amersham Protran 0.45 μm nitrocellulose membrane (Cytiva) using Blotting Paper 703 (VWR) soaked in 1× TBS (10 mM Trizma Base, 150 mM NaCl) and a sandwich of Pro‐Series wipers.

Immunopurification (IP) from denatured protein extracts

We used agarose beads conjugated with anti‐V5 antibody (Sigma, A7345) to pull down constructs tagged with V5. Prior to usage, beads were equilibrated with NP‐40 lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl, 2 mM MgCl₂, 1% NP‐40, 10% glycerol) and kept at 4°C. Pellets containing approximately 7.2 OD of cells were resuspended in 5% TCA, washed once with acetone, and air‐dried. Pellets were then resuspended into SDS–PAGE lysis buffer and glass beads and processed in a FastPrep‐24 Homogenizer for 45 s at 6.5 m/s. We added SDS to a final 1% concentration and boiled the extracts for 5 min. We added 9 volumes of NP‐40 lysis buffer and clarified by spinning 15 min at 16,000 g at 4°C. We took an input aliquot of clarified protein extract, added 3× SDS–PAGE Loading buffer to a final 1× concentration, and boiled for 5 min. We added the rest of the clarified protein extract to the beads and incubated for 2 h at 4°C with rotation. We pelleted the beads by spinning at 400 g for 30 s and aspirated the supernatant (unbound fraction). We washed twice with a high salt buffer (50 mM Tris–HCl pH 7.5, 1 M NaCl, 1 mM EDTA, 1% NP‐40, 0.1% SDS, and 0.5% sodium deoxycholate) and twice with a low salt buffer (20 mM Tris–HCl pH 7.5, 10 mM MgCl₂, 0.2% Tween‐20). We resuspended the beads in NP‐40 lysis buffer containing 1× SDS–PAGE Loading buffer and boiled for 5 min.

We analyzed the IP by SDS–PAGE, by loading about 0.18% of input and 19% of IP in Criterion TGX Precast Gradient Gels 4–15% from BioRad. For each experiment, we run one gel to visualize V5 and Pgk1, and one gel to visualize the phospho‐PKA substrates.

Mass spectrometry

We performed mass spectrometry (MS) to identify PKA‐dependent phosphorylation events on Rim4 in a wild‐type strain (B1587) induced in YPD or SPO, a tpk1‐as strain (B2611) induced in YPD + 5 μM 1‐NM‐PP1, and a RAS2 ^G19V strain (B2666) induced in SPO. We immunopurified Rim4‐V5 from denatured protein extracts obtained from 24 ml cultures (about 48 OD of cells) and resuspended in NP‐40 lysis buffer containing 1× SDS–PAGE Loading buffer without Bromophenol Blue. We reduced disulfide bonds with 5 mM DTT for 45 min at room temperature, and subsequently alkylated cysteines with 10 mM iodoacetamide in the dark for 45 min at room temperature. Proteins were precipitated onto magnetic SP3 beads by adding ethanol, resulting in a sample that was 50% organic solvent, and by shaking for 8 min at room temperature (Hughes et al, 2019). Beads were washed three times with 1 ml 80% ethanol and reconstituted in 100 μl ammonium bicarbonate. Because conventional trypsin digestion would not digest Rim4 into fragments that can be easily detected by MS and that can be confidently mapped back to the sequence, we decided to split each sample into two parts and digest them using either Promega sequencing grade modified trypsin and Promega sequencing grade glu‐c or Promega sequencing grade chymotrypsin in an enzyme‐to‐substrate ratio of 1:50. After 16 h of digestion, samples were taken off the magnetic beads, acidified to a final concentration of 1% formic acid, dried down using a Thermo Savant SpeedVac, and reconstituted in 3% Acetonitrile/0.2% formic acid for injection into the mass spectrometer.

About 1 μg of total peptides was analyzed on a Waters M‐Class UPLC using a 15 cm IonOpticks Aurora Elite column (75 μm inner diameter; 1.7 μm particle size; heated to 50°C) coupled to a benchtop Thermo Fisher Scientific Orbitrap Q Exactive HF mass spectrometer. Peptides were separated at a flow rate of 400 nl/min with a 90‐min gradient, including sample loading and column equilibration times. Data were acquired in data‐independent mode using Xcalibur 4.5 software. MS1 Spectra were measured with a resolution of 120,000, an AGC target of 2e4 and a mass range from 350 to 1,650 m/z. Per MS1, 23 equally distanced, sequential segments were triggered at a resolution of 30,000, an AGC target of 3e6, a segment width of 56 m/z, and a fixed first mass of 200 m/z. The stepped collision energies were set to 22.5, 25, and 27.

Raw data were analyzed with Spectronaut software version 17.2 (Bruderer et al, 2015) using directDIA analysis methodology and a UniProt database (Saccharomyces cerevisiae, UP000004932, where we substituted the Rim4 sequence with a Rim4‐V5 sequence from SK1 background, see Appendix Fig S6 for overall Rim4‐V5 coverage obtained). Carbamidomethylation on cysteines was set as a fixed modification. Oxidation of methionine, protein N‐terminal acetylation, and phosphorylation on serine/threonine/tyrosine were set as variable modifications. Samples digested with different enzymes were searched separately by setting Trypsin/P and glu‐C or Chymotrypsin as the digestion enzymes for the relevant samples. We obtained two biological replicates for each condition measured. We normalized the Rim4 peptides intensities by the total Rim4 signal intensity and converted them in parts per million. For each phosphorylated peptide surveyed, we calculated the total phospho and non‐phospho signal intensities and their ratio. We evaluated consistency among the two biological replicates by comparing the total phospho and non‐phospho signals for each phosphorylated peptide. We excluded from further analysis those peptides whose replicate ratios were larger than 10. To identify which peptides were differentially phosphorylated in a PKA‐dependent fashion, we compared the phospho/non‐phospho ratios of wild‐type in YPD versus SPO, wild‐type in YPD versus tpk1‐as, and RAS2 ^G19V versus wild‐type in SPO. For each comparison, we calculated the log₂ ratio. We then applied a sequence of three rules. First, we considered only the peptides for which the log₂ ratio of the phospho/non‐phospho signals in wildtype YPD versus SPO was > |1|. Second, we compared just the phospho signals in wildtype YPD versus SPO. This comparison helped to confirm if the difference identified in the first comparison was truly driven by a difference in phosphorylation signal itself and not by differences in the amount of non‐phospho signal in YPD and SPO. Practically, we checked if the log₂ ratio of the phospho/phospho signals in YPD versus SPO had the same sign as the log₂ ratio of the phospho/non‐phospho signals in YPD versus SPO and was > |1|. Third, we evaluated if the phosphorylation patterns were consistent along the three comparisons by inspecting the sign of the log₂ ratio of the phospho/non‐phospho signals of wild‐type in YPD versus SPO, wild‐type in YPD versus tpk1‐as, and RAS2 ^G19V versus wild‐type in SPO. We mutated the sites found in the peptides whose phosphorylation patterns satisfied all three rules above unless they contained a PKA site (Dataset EV1).

RNA methods

Northern blots

Snap frozen pellets containing approximately 3.6 OD of cells were resuspended in TES buffer (10 mM Tris HCl pH 7.5, 10 mM EDTA, 0.5% SDS), zirconia beads, and acid‐Phenol:Chloroform pH 4.5 (with IAA 125:24:1, Ambion), and incubated for 30 min at 65°C shaking vigorously. We added 1 ml 100% ethanol and 40 μl 3 M NaAc (pH 5.5) to the supernatant and precipitated for 30 min at 4°C. RNA was then pelleted, washed once with 80% ethanol, and air dried. We performed a 30 min treatment with 20 U Turbo DNase (Invitrogen) at 37°C and precipitated again. RNA pellets were washed once, air dried, and resuspended in water.

We separated RNA in denaturing 1.3% agarose gels containing 1× MOPS (where 10× MOPS is 0.2 M MOPS, 0.05 M NaAc, 0.01 M EDTA, pH 7) and 16.6% formaldehyde. We performed capillary transfer overnight at room temperature on an Amersham Hybond‐N+ membrane (GE Healthcare) soaked in 10× SSC (1.5 M NaCl, 0.15 M Trisodium citrate dihydrate). Membranes were crosslinked with 120,000 μJ/cm² in a Spectrolinker XL‐1000 UV crosslinker (Spectronics corporation) and stained with a Methylene Blue solution (0.03% methylene blue, 0.4 M NaAc pH 5.5) to visualize ribosomal RNA. Membranes were then incubated in 20 ml hybridization buffer (0.25 M Sodium Phosphate pH 7.2, 0.25 M NaCl, 1 mM EDTA pH 8, 7% SDS, and 5% Dextran Sulfate) with 300 μl of a 10 mg/ml stock of boiled sheared salmon sperm DNA (Invitrogen) at 65°C rotating for 4 h. We prepared radioactive probes labeled with 50 μCi [α‐³²P]dCTP using the Amersham MegaPrime Labeling kit (Cytiva) on a PCR product obtained from p269 with 5′‐AATGGGATCCAAAGGTGAAGAATTATTCAC‐3′ and 5′‐TTTGTACAATTCATCCATACCATGGGT‐3′. Membranes were incubated with 20 ml hybridization buffer and the radioactive probe at 65°C rotating overnight, washed twice with low stringency wash (2× SSC, 1% SDS) and once with high stringency wash (0.1× SSC 1% SDS), dried, exposed to a phosphor‐screen (GE Healthcare), and scanned with a Typhoon Trio scanner (Amersham) or FLA 9000 (GE Healthcare).

Polynucleotide kinase (PNK) assay

We adapted our PNK assay protocol from Beckmann et al, 2015. We UV‐crosslinked cultures with 960,000 μJ/cm² and pelleted approximately 9 OD of cells. Pellets were resuspended in 5% TCA, washed once with acetone, and air‐dried. We performed an immunopurification from denatured protein extracts with the following modifications. After clarification, we treated the whole protein extracts with 8 ng/μl RNase A and 4 U Turbo DNase for 15 min shaking at 37°C. After immunopurification, beads were washed twice with high salt buffer, twice with low salt buffer, and twice with PNK buffer (50 mM Tris HCl pH 7.5, 50 mM NaCl, 10 mM MgCl₂, 0.5% NP‐40) containing 5 mM DTT. We resuspended the beads in PNK buffer + 5 mM DTT, 0.1 μCi/μl [γ‐³²P]ATP (Perkin Elmer), 1 U/μl T4 PNK (New England Biolabs), and labeled for 15 min at 37°C. We washed with 1 ml PNK buffer without DTT and resuspended the beads in PNK buffer without DTT and 1× SDS–PAGE loading buffer. We boiled the samples for 5 min and performed an SDS–PAGE using Criterion TGX Precast Gradient Gels 4–15% from BioRad. Upon transfer to a 0.2 μm nitrocellulose membrane, we exposed the membrane to a phosphor‐screen and scanned with a Typhoon FLA 9000 (GE Healthcare). The membrane was then re‐hydrated in 1× TBS‐T, blotted, and imaged with an Amersham Imager 600 (GE Healthcare).

RNA‐immunopurification (RNA‐IP) followed by quantitative PCR (qPCR)

We adapted our RNA‐IP qPCR protocol from Görnemann et al, 2011 and Mukherjee et al, 2021. We used agarose beads conjugated with anti‐V5 antibody (Sigma, A7345) to pull down constructs tagged with V5. All steps were performed at 4°C. Prior to usage, beads were equilibrated with IP Buffer (50 mM Tris HCl pH 7.5, 125 mM KCl, 0.1% NP‐40 with 1× Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific), 1 mM DTT, and 10 U/ml Recombinant RNase Inhibitor (Takara, 2313B) added freshly). Snap frozen pellets containing approximately 3.6 OD of cells were resuspended in 200 μl IP Buffer and zirconia beads (0.5 mm dia. Zirconia/Silica, BioSpec Products) and processed in a FastPrep‐24 Homogenizer machine for 45 s at 6.5 m/s. After adding 800 μl IP Buffer to reach a total volume of 1,000 μl, we spun the whole cell lysate at 2,500 g for 10 min, transferred the supernatant to a fresh tube, spun again at 2,500 g for 5 min. We took 50 μl cell lysate as input sample for protein analysis and another 50 μl (that is, 5% of the total cell lysate) as input sample for RNA analysis. We transferred the rest of the cell lysate (900 μl) to the beads and incubated for 2 h at 4°C with rotation. We pelleted the beads by spinning at 400 g for 30 s and aspirated the supernatant (unbound fraction). We washed the beads four times with IP Buffer, incubating for 10 min at 4°C with rotation in between. We resuspended the beads in 900 μl IP buffer and took 50 μl as IP sample for protein analysis. We used the remaining 850 μl (that is, 85% volume of the original cell lysate) as IP sample for RNA analysis. Input and IP RNA samples were resuspended in 400 μl TES buffer and 400 μl acid‐Phenol:Chloroform pH 4.5 and treated as described in the Northern blots section. We performed one or two Turbo DNase treatments with a precipitation step in between. After air drying the RNA pellets, we resuspended both input and IP samples in 11 μl water.

We used oligo dT and the SuperScript IV Reverse Transcriptase kit from ThermoFisher (18091050) to obtain cDNA. We retrotranscribed 5.5 μl of input RNA and used the remaining 5.5 μl in a retrotranscription reaction lacking the enzyme (‐RT) to assess DNA contaminations. We also retrotranscribed 5.5 μl IP RNA of the no tag sample. Since we wanted to compare the RNA binding efficiency of different Rim4 mutants, or of wild‐type Rim4 in different growth conditions, we needed to assess differences in immunopurification efficiencies and retrotranscribe equivalent ratios of IP RNA for each sample. Therefore, we performed immunoblots of the IP protein samples and quantified the protein signal along each lane using the ImageJ software. We drew a rectangle along the complete lane and quantified the mean intensity signal. For each lane, we considered the same area. We then subtracted the background signal, which we obtained from the no tag lane. We calculated the equivalent ratio of each lane by dividing the signal of the least abundant lane by the signal of all other lanes. We retrotranscribed 5.5 μl IP RNA of the least abundant sample and equivalent ratios of IP RNA of all other samples. For each IP RNA sample, we also set up ‐RT reactions.

We performed qPCR on a BioRad C1000 Touch Thermal Cycler CFX96 Real Time System. We used the PowerUP Sybr Green MasterMix (A25741, Applied BioSystems) and a standard cycling mode (activation: 50°C for 2 min followed by 95°C for 2 min; amplification: 45 cycles of 95°C for 15 s, 56°C for 15 s, 72°C for 1 min). We measured Citrine and CUP1‐1 cDNAs and performed three technical replicates for each retrotranscription reaction. We also performed qRT–PCR on ‐RT reactions to assess DNA contaminations. We used 5′‐GGTTGAATTAGATGGTGATGTTA‐3′ and 5′‐GGCAATTTACCAGTAGTACAAA‐3′ to detect Citrine (Ottoz et al, 2014), and 5′‐ATCACATAAAATGTTCAGCGA‐3′ and 5′‐CTTGGTTTCTTCAGACTTGTTA‐3′ to detect CUP1‐1.

We used the percent input method used for ChIP analysis to quantify RNA binding. First, we calculated the mean Cq for each technical triplicate. Second, we adjusted both input (ADJin) and IP (ADJIP) Cq values of each sample to 100% by subtracting the number of cycles corresponding to the dilution factors of input and IP, respectively (that is, Cq_Dil = log_{primer efficiency}(dilution factor)). Finally, the % input was calculated as 100 × (primer efficiency)^{(ADJin−ADJIP)}.

To evaluate the presence or absence of the CLB3 5′UTR in the 3′ end of the YFP reporter gene (Fig EV1E), we performed a PCR on the RT and ‐RT reactions of input RNA extracted from starved strains with 5′‐GGTATTACCCATGGTATGGATG‐3′ and 5′‐GCGTGAATGTAAGCGTGACA‐3′.

Growth curves

We inoculated 10,000 cells in 200 μl fresh YPD in each well of a 96‐well plate Nunclon Delta Surface (Thermo Scientific). For each strain and condition, we prepared triplicates or quadruplicates. We incubated the plate in a TECAN Infinite M200 NanoQuant reader (Tecan group) at 30°C, shaking. Absorbance at 600 nm was measured every 6 min. We calculated the mean of the absorbance and the standard deviation of each triplicate/quadruplicate for each time point.

Statistical tests

We used the software R for statistical analysis. Two‐sample Wilcoxon rank sum tests were performed to evaluate the significance of the difference of fluorescence signals between strains expressing the same YFP allele but differing by the absence (control) or presence (positive) of a RIM4‐V5 allele. We tested the null hypothesis that the control strain exhibited equal or less levels of YFP than the positive strain, and the alternative hypothesis that the YFP levels in the control were greater than in the positive strain. Two‐sample Wilcoxon rank sum tests were also performed when evaluating the effects of Rim4 on YFP levels in different growth conditions. We set the significance level α to 0.05 and rejected the null hypothesis if the P‐value was ≤ α. We performed a Dunn Kruskal–Wallis multiple comparison to evaluate the effects of the RIM4‐V5 mutants on YFP expression levels measured by flow cytometry and the RNA‐binding abilities of these constructs measured by RNA‐IP. P‐values were adjusted with the Benjamini‐Hochberg method. We set the significance level α to 0.05 and rejected the null hypothesis if the P‐value was ≤ α/2.

Author contributions

Diana SM Ottoz: Conceptualization; resources; data curation; formal analysis; supervision; validation; investigation; visualization; methodology; writing – original draft; project administration; writing – review and editing. Lauren C Tang: Data curation; formal analysis; validation; investigation; methodology. Annie E Dyatel: Investigation. Marko Jovanovic: Resources; supervision; funding acquisition; methodology; project administration. Luke E Berchowitz: Conceptualization; resources; supervision; funding acquisition; methodology; project administration; writing – review and editing.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Dataset EV1

Source Data for Expanded View

PDF+

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

The EMBO Journal (2023) 42: e113332

Data availability

The mass spectrometry dataset produced in this study is available in the MassIVE database (web link is https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=025be480eb8e44e48e4974276ca55a22; ftp address for download is ftp://MSV000092632@massive.ucsd.edu).