A quantitative inventory of yeast P body proteins reveals principles of composition and specificity

  1. Wenmin Xing
  2. Denise Muhlrad
  3. Roy Parker
  4. Michael K Rosen  Is a corresponding author
  1. Department of Biophysics, Howard Hughes Medical Institute, UT Southwestern Medical Center, United States
  2. Department of Biochemistry, Howard Hughes Medical Institute, University of Colorado, United States

Abstract

P bodies are archetypal biomolecular condensates that concentrate proteins and RNA without a surrounding membrane. While dozens of P body proteins are known, the concentrations of components in the compartment have not been measured. We used live cell imaging to generate a quantitative inventory of the major proteins in yeast P bodies. Only seven proteins are highly concentrated in P bodies (5.1–15µM); the 24 others examined are appreciably lower (most ≤ 2.6µM). P body concentration correlates inversely with cytoplasmic exchange rate. Sequence elements driving Dcp2 concentration into P bodies are distributed across the protein and act synergistically. Our data indicate that P bodies, and probably other condensates, are compositionally simpler than suggested by proteomic analyses, with implications for specificity, reconstitution and evolution.

Introduction

Eukaryotic cells contain numerous compartments that concentrate specific sets of molecules without a surrounding membrane (Banani et al., 2017; Shin and Brangwynne, 2017). These structures, referred to as biomolecular condensates, are related to a variety of biological processes. Examples include cytoplasmic processing bodies (P bodies) associated with RNA metabolism (Decker and Parker, 2012); promyelocytic leukemia nuclear bodies (PML NBs) involved in transcription, DNA damage repair, and anti-viral responses (Lallemand-Breitenbach and de Thé, 2010); signaling clusters in T cell activation (Su et al., 2016); and HP1 clusters in heterochromatin organization (Larson et al., 2017; Strom et al., 2017). Many condensates form through self-assembly of multivalent molecules, including proteins composed of folded domains and/or disordered regions, RNA and DNA, and chromatin (Li et al., 2012; Kato et al., 2012; Nott et al., 2015; Su et al., 2016; Banani et al., 2017; Gibson et al., 2019). This process can lead to both liquid-like and solid-like structures (Banani et al., 2017; Shin and Brangwynne, 2017; Alberti and Dormann, 2019). Some condensates have been shown to respond sharply to changes in concentration of key components or regulators, salt and/or temperature, suggesting that they form through highly cooperative assembly mechanisms, including phase transitions (Beutel et al., 2019; Brangwynne et al., 2009Falahati and Wieschaus, 2017; Li et al., 2012; Riback et al., 2017; Saha et al., 2016; Smith et al., 2016; Wang et al., 2014; Weber and Brangwynne, 2015). The activities of biomolecular condensates are thought to derive from the assembly of specific collections of functionally related molecules into a unique physical environment. Thus, understanding condensates as chemical entities requires knowledge and understanding of their compositions.

Large-scale proteomics studies have been performed to determine comprehensive inventories of molecules that localize to different condensates, including stress granules (Jain et al., 2016; Markmiller et al., 2018; Youn et al., 2018), P bodies (Hubstenberger et al., 2017; Youn et al., 2018), and nucleoli (Andersen et al., 2002). There are also extensive studies on the localization of individual molecules to different condensates. Cumulatively, these studies suggest that condensates are biochemically complex, containing tens to hundreds of types of proteins and RNAs that show dense and complex patterns of molecular interactions. Although some transcriptomic studies and a few analyses of individual proteins have been quantitative (Khong et al., 2017; Klingauf et al., 2006; Leung et al., 2006; Wheeler et al., 2017), proteomic studies to date have been only qualitative. Therefore, both the relative and absolute concentrations of proteins in condensates, and their relationships to molecular connectivity, are largely unknown. Additionally, the relationships between concentration and dynamics of molecules in condensates have not been systematically explored. Moreover, while some components are shared between different condensates, many are uniquely concentrated in specific condensates. It is still unclear how the substantial specificity is determined in vivo.

To address these issues, we sought to examine the composition of a complex cellular condensate in a quantitative manner. We performed a systematic analysis of yeast P bodies, an archetypical biomolecular condensate. P bodies are protein- and mRNA-rich cytoplasmic condensates conserved from yeast to mammals (Ingelfinger et al., 2002; Sheth and Parker, 2003; van Dijk et al., 2002). They are thought to participate in RNA metabolism, modulating mRNA decay and acting as sites of RNA storage during cellular stress (Aizer et al., 2014; Bhattacharyya et al., 2006; Cougot et al., 2004; Sheth and Parker, 2003). Taking advantage of an available yeast GFP library, we used quantitative fluorescence microscopy to measure the absolute concentrations of 31 P body resident proteins within the condensates and in the surrounding cytoplasm. We also measured their dynamic properties using fluorescence recovery after photobleaching (FRAP). We find that P body proteins segregate into two groups based on their concentrations in P bodies and dynamics. Members of the first group (Dcp2, Edc3, Pat1, Xrn1, Lsm1, Dhh1, Upf1) are highly concentrated in P bodies (5.1–15 µM) and exchange slowly with the cytoplasm. Except for Xrn1, these all have high connectivity, interacting with multiple P body components. Partitioning of several of these highly concentrated (HC) proteins (Dcp2, Edc3, Pat1, and Xrn1) into P bodies is correlated, suggesting that they assemble cooperatively, consistent with their high connectivity. All proteins known to contribute strongly to P body formation through genetic studies are in this HC group. In contrast, members of the second less concentrated (LC) group are at substantially lower concentrations in P bodies (all except Sbp1, 0.7–2.6 µM) and exhibit faster dynamics. These all have low connectivity. In a molecular dissection of Dcp2, we find that the N-terminal domain, multivalent C-terminal domain, and central high affinity Edc3 binding site all contribute to partitioning of the protein into P bodies and to its dynamic exchange with the cytoplasm. Moreover, the N-terminal and C-terminal domains can act synergistically to promote recruitment of Dcp2 into P bodies. This highly distributive organization suggests a ready means of quantitatively modulating condensate composition during evolution. These data suggest that while a condensate may contain many components, only a small number are highly concentrated there.

Results

Strategies for quantification of P body proteins

In order to understand the chemical nature of a condensate it is important to both know its members, and also their concentrations. Two criteria must be met to accurately measure protein concentrations in a condensate based on fluorescence intensities alone: 1) the size of the condensate must be larger than the point spread function (PSF, i.e. the diffraction limit) of the microscope used in the analysis since the fluorescence intensities are diluted for small objects (Fink et al., 1998); 2) the composition of the condensate cannot change over time, allowing reliable comparisons among multiple condensates in multiple cells. P bodies in wild type S. cerevisiae are typically diffraction-limited in size under normal conditions. But they become larger when mRNA decay is decreased, for example when mRNA decay proteins such as Dcp1 are deleted or when cells respond to stresses such as glucose starvation (Teixeira et al., 2005; Teixeira and Parker, 2007). Initially we analyzed GFP-tagged proteins expressed in a dcp1Δ strain, because the effects of stress are time-dependent and thus more difficult to perform systematically measurements across a series of strains. We individually expressed 31 reported P body resident proteins tagged with GFP at their C-termini in their chromosomal location under their endogenous promoters (Figure 1—figure supplement 1A, and Supplementary file 1). We presume these puncta are P bodies because previous reports that the proteins localize to P bodies and in all cases tested we observed co-localization of the proteins with an mCherry-tagged P body marker, Edc3 (Figure 1—figure supplement 1B).

Using confocal microscopy, we limited our analyses to P bodies that were larger than the x-y PSF of our microscope (see Materials and methods), and corrected the diluting effect of the larger z-PSF based on an assumption that the structures were spherical. For each protein, we measured the absolute concentration in the P body and in the surrounding cytoplasm, based on cellular fluorescence intensity and independent calibration of fluorescence versus GFP concentration on our microscope (Figure 1—figure supplement 2)(see details in Materials and methods). We calculated the partition coefficient (PC) as the ratio of these two values.

We also used fluorescence recovery after photobleaching (FRAP) to measure the exchange between P bodies and the cytoplasm. Of the 31 proteins, 19 showed sufficiently punctate distributions in mid-log phase to permit analysis (PC >~2). We refer to these 19 proteins as regular P body proteins hereafter. The remaining 12 proteins were distributed relatively uniformly in the cytoplasm and their P body concentrations could not be analyzed; these may be stress or strain specific proteins that do not concentrate in P bodies under our experimental conditions (Supplementary file 1).

Two observations suggest that the GFP tag probably does not strongly affect protein behaviors. First, the PC values and dynamics of Edc3 and Dhh1 tagged with GFP and mCherry were similar (Figure 1—figure supplement 3A and B). Second, the PC values and dynamics of Dcp2 are nearly identical with an N-terminal or C-terminal GFP tag (Figure 1—figure supplement 3C). In a screen of this size it is not practical to validate each protein with multiple tags at multiple locations. It remains possible that some other proteins are affected by the tag.

Seven proteins/assemblies are highly concentrated in P bodies

We obtained PC for all 19 regular P body proteins, which revealed subclasses of these molecules. Average PC values of the 19 regular P body proteins had a wide range, with a maximum of 133 (Dcp2) steadily decreasing to a minimum of ~5 (Eap1, Ssd1) (Figure 1A and Supplementary file 2). Only a few proteins had high partition coefficients, including Dcp2 (133 ± 13), Edc3 (133 ± 8) and Pat1 (107 ± 12); the majority had PC values < 20. We also obtained absolute concentrations of each protein in P bodies, which ranged from ~0.7 µM to 15 µM and revealed two notable features of P bodies.

Figure 1 with 3 supplements see all
Seven proteins are highly concentrated in P bodies.

(A) Partition coefficients (PCs) of 19 P body proteins. Plots show PCs and mean values (blue lines)± standard error of the mean (SEM). (B) Absolute concentrations in P bodies of 19 P body proteins. Plots show absolute concentrations in P bodies and mean values (blue lines)± SEM. For (A) and (B) each dot represents an individual P body. Red, HC P body proteins. Black, LC P body proteins. One P body per cell was analyzed from 16 to 25 cells for each protein.

First, seven proteins, Dcp2, Edc3, Pat1, Xrn1, Lsm1 (likely representing the entire, constitutive Lsm1-7 assembly [Sharif and Conti, 2013]), Dhh1 and Upf1 – had average P body concentrations of 8.9–12 µM, with a slightly higher concentration (15 µM) for Dcp2 and slightly lower concentration (5 µM) for Upf1. We refer to these components as highly concentrated (HC) P body proteins.

A second striking observation was that there was a clear, sharp distinction between highly concentrated (HC) and less concentrated (LC) proteins (Figure 1B). After the HC proteins most other proteins have P body concentrations < 2.6 µM, with most ~1 µM. The only exception is Sbp1 (4.6 µM), which also has a very high cytoplasmic concentration, affording it a high P body concentration even with a small PC value (Figure 1, and Supplementary file 2). Sbp1 also is not highly enriched in glucose-starved wild type P bodies (see below); for these reasons we do not include it in the HC group. Our data are consistent with previously reported measurements since the total cellular concentrations that we measured are similar to a unified quantitative protein abundance database of S. cerevisiae (Ho et al., 2018Figure 1—figure supplement 3D).

Together, these data show that while yeast P bodies can contain many different proteins, only a relatively small number are highly concentrated in the compartment, and the majority of components are weakly concentrated.

Highly concentrated proteins tend to have slow dynamics

We also examined the dynamics of the 19 regular P body proteins using FRAP. For each GFP-fusion expressing strain, entire P bodies (0.4–0.8 µm) were photobleached and the fluorescence recovery curves were fit to single exponential. The exchange rate (k) and fractional recovery were used to assess their dynamics. This analysis revealed that the regular P body proteins exchange with the cytoplasm with very different rates and fractional recoveries (Figure 2—figure supplement 1 and Supplementary file 2). At one end of the distribution, Dcp2 and Edc3 showed no measurable recovery on a 150 s timescale. Proteins such as Pat1 and Upf1, showed intermediate rates and extents of recovery (k = ~0.01 s−1, recovery = ~50%). While at the other end of the spectrum, proteins such as Sbp1 and Eap1 recovered nearly 100% in 150 s (k = 0.07 s−1).

Since we observed heterogeneity of both partitioning and dynamics, we asked whether these parameters are related. For each protein, we plotted the exchange rate (k), and fractional recovery against P body concentration (Figure 2). Although exchange rate and recovery for both the HC proteins and LC proteins spanned broad ranges, the former (Dcp2, Edc3, Pat1, Xrn1, Lsm1, Upf1, and Dhh1) tended to have slower exchange rates and less fractional recoveries than the latter (Figure 2). On average, for the HC group, the exchange rate was five-fold smaller than the LC group, and recovery fraction was two-fold lower (Figure 2).

Figure 2 with 4 supplements see all
Highly concentrated proteins tend to have slow dynamics.

Exchange rate (A) or fractional recovery (B) was plotted as a function of absolute concentrations in P bodies, respectively (mean ± SEM). HC and LC proteins are indicated by red and black symbols, respectively. Graphs on the right show averaged exchange rate or fractional recovery (red lines) in HC and LC groups. Analysis of significance was calculated by Wilcoxon rank-sum test. *** and **** indicate p values less than 0.001 and 0.0005, respectively.

Three observations argue that the slower recovery dynamics of the HC proteins are not simply due to smaller fluorescent pools in cytoplasm, but rather reflect different interactions in the P body. First, neither exchange rate nor fractional recovery correlates with protein concentrations in cytoplasm (Figure 2—figure supplement 2A and B). Second, less than 50% of the total pool of each protein is localized to the P body (<8% for most), suggesting that there are still substantial fluorescent pools in the cytoplasm (Figure 4). Third, we also performed inverse FRAP (iFRAP), which is insensitive to the size of fluorescent pools, for several proteins. In iFRAP, the entire cytoplasm except one P body was bleached and the loss of fluorescence in the P body was followed over time. In each case, the exchange rate and fractional recovery measured by iFRAP was similar to those measured by FRAP (Figure 2—figure supplement 2C and D).

Partitioning of highly concentrated proteins into P bodies is correlated

We noticed that the PC values and P body concentrations of proteins varied in a broad range, for example, the PC of Dcp2 varied from 50 to 250 between different cells (Figure 1). In addition, the HC proteins, Dcp2:Edc3:Pat1:Lsm1:Dhh1:Xrn1, are present at roughly equimolar average levels (~10 µM), suggesting their concentrations may be correlated. To better understand the variability of partitioning, we measured the concentrations of pairs of HC proteins, differentially tagged with GFP or mCherry, in the same P bodies (Figure 3A). This revealed that the concentrations of paired Edc3 and Dcp2, Pat1 and Dcp2, and Xrn1 and Dcp2, are positively correlated as indicated by Pearson correlation coefficients of 0.6–0.7 (Figures 3B, C and D). Moreover, the ratios of the two measured protein concentrations have narrower ranges than the ratios calculated by randomizing the pairing (Figure 3A and E). These data indicate that partitioning of HC proteins into P bodies is correlated, suggesting they assemble cooperatively (see Discussion).

Partitioning of proteins into P bodies is correlated.

(A) Schematics of calculations of ratios of paired and randomized proteins. Paired ones are ratios of X-GFP to Dcp2-mCherry in the same P bodies where i = i’. Randomized pairing are achieved by randomly generating i and j in Excel where i ≠ j. The sample size is kept the same as paired ones. (B/C/D) Concentrations of Edc3 and Dcp2 (B), Pat1 and Dcp2 (C), and Xrn1 and Dcp2 (D) are correlated in P bodies. Each dot represents one P body in a cell, 86 cells (B), 64 cells (C), and 135 cells (D) were analyzed. Pearson correlation coefficients were calculated in Prism (GraphPad ). (E) Ratios of paired proteins have tighter ranges than ratios when randomizing pairing. Plots show ratios (black dots) and mean values ± standard deviation (red lines), values are shown. Fligner-Killeen test was used to test equality of variance in R. ***, p<0.001, ****, p<0.0005.

Figure 4 with 1 supplement see all
P bodies do not strongly sequester their resident proteins.

Plots show fractions in visible P bodies (black dots) and mean values (red lines)± SEM in mid-log phase without any cellular stress. Each dot represents fraction of protein in visible P bodies corresponding to an individual cell. 16–25 cells were analyzed for each protein.

Partitioning and dynamics of proteins are not strongly affected by Dcp1 deletion

Since the above analysis was all performed in a dcp1∆ strain, we wanted to measure PC and P body dynamics under a different condition to evaluate if the Dcp1 deletion made a substantial impact on P body composition or dynamics. For this experiment, we analyzed wild type strains after 30–60 min of glucose starvation. By several criteria, we observed similar results to our analysis in dcp1∆ strains.

First, similar to the dcp1∆ strains, during glucose deprivation, Dcp2, Edc3, Pat1, Lsm1, Xrn1, Upf1, and Dhh1, remain the most concentrated proteins in wild type P bodies, although Pat1 and Lsm1 dropped compared to the dcp1∆ strains (Figure 2—figure supplement 3A and B). Two additional proteins, Dcp1 and Pby1, also partition strongly into wild type P bodies under glucose starvation. Dcp1 binds with high affinity to Dcp2, and Pby1, in turn, binds to Dcp1, explaining its absence in dcp1Δ P bodies (Krogan et al., 2006). Only two members of the LC group, Hek2 and Sbp1, partition sufficiently to permit analysis suggesting that the partitioning of LC proteins is lower under glucose starvation.

We also observed that the dynamics of all proteins under glucose starvation were qualitatively similar to the dcp1Δ strains. The HC proteins have slower dynamics, although there is no clear distinction between HC and LC groups because of the limited number of the latter (Figure 2—figure supplement 3C and D). Nevertheless, these data suggest that partitioning and dynamics of proteins are qualitatively similar in the wild type strains under glucose starvation and the dcp1Δ strains, especially for the HC proteins. We thus performed all subsequent experiments in dcp1Δ strains, unless stated otherwise.

P bodies do not strongly sequester their resident proteins

The biological importance of concentrating certain proteins into P bodies is unclear. One proposed function is that the P body could sequester molecules, inhibiting their activities in the cytoplasm. Similarly, P bodies have been suggested to store mRNAs or proteins under cellular stress, which could then be returned to the cytoplasm when the stress is resolved (Aizer et al., 2014; Bhattacharyya et al., 2006; Brengues et al., 2005). To examine the efficiency of protein sequestration, we quantified the fractions of each P body protein in the compartments (FP). We first measured FP in mid-log phase without stress. Under these conditions, the most concentrated proteins, Dcp2 and Edc3, are on average ~30% sequestered in observable P bodies (Figure 4). For the other highly concentrated proteins (Pat1, Xrn1, Lsm1, Upf1, and Dhh1), about 10% is sequestered in visible P bodies. FP is even smaller for less concentrated proteins such that only about 5% of each is in visible P bodies.

We further asked if efficiency of P body sequestration changes under different conditions. We first measured the degree of P body sequestration in dcp1Δ strains after 4 hr of glucose starvation (Teixeira and Parker, 2007). Although FP increased for most of the proteins, visible P bodies still sequester less than 40% of each of the concentrated proteins (Figure 4—figure supplement 1A). We next grew cells to stationary phase, and again, less than 50% of each protein except Dcp2 and Edc3 is in P bodies (Figure 4—figure supplement 1B) .

One possible explanation for the low degree of sequestration is that many P bodies are too small to be observed by standard confocal microscopy (Rao and Parker, 2017). This would give artifactually low FP values, as a significant fraction of P body-associated protein would not be accounted for in the integrated P body fluorescence intensity. Nevertheless, we can estimate an upper limit of FP for most proteins by assuming that all P bodies in a given cell have identical compositions independent of size, and that the FP value for the most sequestered protein (Dcp2), is, in fact one when small P bodies are properly accounted for. With this assumption, the estimated maximum FP (FP,max) for each protein would be FP,max = FP * 1/FP(Dcp2). Even with this conservative estimate, most proteins are only sequestered to <20% in P bodies under all conditions.

Together, our quantifications indicate that P bodies do not strongly sequester their resident proteins under the conditions examined. Our data of course do not rule out the possibility that sequestration could be higher under different conditions, nor do they speak to sequestration/storage of RNA, which could have even higher PC values than proteins, perhaps due to non-equilibrium processes (Ditlev et al., 2018; Hubstenberger et al., 2017).

Elements controlling Dcp2 partitioning and dynamics are distributed across the protein

We have shown that proteins concentrate into P bodies to different degrees and with distinct dynamic behaviors. We next sought to understand what molecular features could control these properties, using Dcp2, one of the most highly concentrated and least dynamic P body components, as an example. We divided Dcp2 into three distinct regions (Figure 5A). The N-terminal domain (NTD) of the protein possesses decapping activity and binds to Dcp1 and mRNAs (Deshmukh et al., 2008; She et al., 2008). The multivalent C-terminal domain (CTD) contains multiple short helical leucine rich motifs (HLMs) that bind to Edc3. Finally, near the center of Dcp2, the first HLM (HLM1) appears to bind Edc3 with appreciably higher affinity than all other HLMs (Charenton et al., 2016), and mutations to this motif impair Dcp2 partitioning into P bodies (Harigaya et al., 2010). We expressed different Dcp2 variants in dcp1Δdcp2Δ strains, in which P bodies are still formed, but they are smaller than in yeast expressing full length Dcp2 (Figure 5—figure supplement 1A),. The expression levels of these Dcp2 variants are similar to wild type Dcp2 as assessed by western blotting (Figure 5—figure supplement 1B). The analysis of these variants identified three molecular elements of Dcp2 that affect its partitioning into P bodies.

Figure 5 with 1 supplement see all
Elements controlling Dcp2 partitioning and dynamics are distributed across the protein, and contribute to specific recruitment to biomolecular condensates.

(A) Schematics of domain architecture of Dcp2 FL and mutants. Red, N-terminal domain (NTD). Orange, HLM1. Blue, other 10 HLMs in C-terminal domain. Grey, inactivated HLMs. (B/C) Partition coefficients (B) and dynamics (C) of GFP tagged Dcp2 FL (black), Dcp2ΔN (red), Dcp2 ΔH1 (orange), and Dcp2 300 (grey) in dcp1Δdcp2Δ strain. One P body per cell was analyzed from 19 cells. PCs are shown as mean values (bold lines)± SEM (B). Recovery curves are shown as average of 19 P bodies ± SEM (C). Significance was calculated by the Wilcoxon rank-sum test. *** and **** indicate p values less than 0.001 and 0.0005, respectively. (D) Representative images showing dcp1Δdcp2Δ yeast strains expressing GFP tagged Dcp2 300 ΔH1, Dcp2C Δ5H and Dcp2ΔH1 Δ5H. Edc3-mCherry was also expressed as a P body marker. Scale bar, 5 µm. (E) Fractions of cells exhibiting puncta formed by each mutant. n(Dcp2 300 ΔH1)=163, n(Dcp2C Δ5H)=188, n(Dcp2ΔH1 Δ5H)=204. (F) Partition coefficients of GFP-tagged Dcp2 300 ΔH1, Dcp2C Δ5H and Dcp2ΔH1 Δ5H, and mean values (red lines)± SEM. 100 P bodies were analyzed. Significance was calculated by the Wilcoxon rank-sum test. **** indicates p values less than 0.0005. (G) Specific recruitment to P bodies can be achieved by distributing elements across Dcp2, even when they recognize distinct ligands within a condensate.

First, we found that N-terminally truncated Dcp2 (Dcp2ΔN) partitions into P bodies less efficiently than the full-length protein (Dcp2 FL), with PC values of 72 ± 8 and 124 ± 9, respectively. FRAP analysis revealed that Dcp2ΔN had an appreciably higher recovery fraction, 0.35 over 150 s, compared to Dcp2 FL, which does not recover at all in this timeframe (Figure 5B and C). This demonstrates that the N-terminal domain of Dcp2 promotes P body accumulation.

Second, mutating HLM1 (Dcp2ΔH1) to alanine in the full-length protein strongly decreased the PC to ~26, and increased the recovery after photobleaching to 0.65 over 150 s. Thus, both the NTD and HLM1 contribute to concentrating Dcp2 into P bodies and decreasing its exchange with the cytoplasm.

Third, additional data suggest that the other HLMs in Dcp2’s C-terminal extension can contribute to P body targeting when HLM1 is missing. This is based on the observations that while truncation of the CTD (Dcp2 300) had no significant effect on the PC and dynamics of Dcp2 (Figure 5B and C), removing the C-terminal domain from Dcp2 ΔH1 (Dcp2 300ΔH1) impaired recruitment to P bodies (Figure 5D).

Since the N-terminal domain and HLM1 are required for efficient partitioning and maintaining the characteristic slow dynamics of Dcp2, and the C-terminal domain can compensate when HLM1 is lacking, we conclude that elements controlling protein partitioning and dynamics are distributed across the protein.

Recruitment elements of Dcp2 act synergistically

We next asked whether these regions can act synergistically to promote P body recruitment. We analyzed three Dcp2 mutant fragments in dcp1Δdcp2Δ strains: 1) Dcp2 300ΔH1, which can only interact with RNA; 2) a C-terminal Dcp2 fragment, Dcp2C Δ5H, in which five out of nine HLMs have been inactivated. This construct partitions into P bodies more weakly than the wild type C-terminal domain, Dcp2C, affording a larger dynamic range (Figure 5—figure supplement 1C); and 3) Dcp2ΔH1 Δ5H, fusion of Dcp2 300ΔH1 and Dcp2C Δ5H, which interacts with both RNA and Edc3 (Figure 5A and G). The three mutants expressed at similar levels in dcp1Δdcp2Δ strains (Figure 5—figure supplement 1D). We analyzed recruitment of the three proteins into P bodies using Edc3 as a P body marker.

We observed synergistic effects between these two regions of Dcp2. Specifically, Dcp2 300ΔH1 and Dcp2C Δ5H are measurably recruited into microscopic P bodies in only ~20% and~30% of cells, respectively (see Materials and methods). However, the fusion of the two fragments, Dcp2ΔH1 Δ5H, is recruited into P bodies in ~95% of cells (Figure 5D and Figure 5E). Moreover, partition coefficients of Dcp2 300ΔH1 and Dcp2C Δ5H are 2.5 and 3.5, respectively, while the partition coefficient of Dcp2ΔH1 Δ5H is 31 (Figure 5E). Since the product of the individual partition coefficients (8.75, which is exponentially related to the sum of the free energies of partitioning) is less than that of the fusion protein (31), we conclude that the elements act synergistically in the fusion to promote accumulation into P bodies.

Thus, while Dcp2 300ΔH1 and Dcp2C Δ5H are each recruited only weakly to P bodies, their fusion, Dcp2ΔH1 Δ5H, is recruited strongly. This indicates that recruitment elements can act synergistically when fused in cis, even when they recognize distinct ligands (in this case, Edc3 and RNA) within a condensate (Figure 5G). This behavior is likely mechanistically similar to avidity effects in canonical molecular interactions, where high affinity can be achieved through multivalent binding.

Partitioning of Dcp2 into P bodies can be saturated

To further examine whether partitioning of proteins can be enhanced by adding control elements, we added HLM1 to Dcp2 FL and DcpC Δ5H, referring to these variants as Dcp2 FL-H1, and DcpC Δ5H-H1, respectively (Figure 6A). For the weakly partitioning DcpC Δ5H, adding HLM1 significantly increased its concentration in P bodies to ~16 µM, comparable to Dcp2 FL, and increased its partition coefficient to ~69 (Figures 6B, C and D), although Dcp2C Δ5H expressed higher than Dcp2C Δ5H-H1 (Figure 6—figure supplement 1). In contrast, adding HLM1 to Dcp2 FL, which already partitions into P bodies strongly, does not significantly further enhance either the absolute concentration or partition coefficient. Thus, binding of HLM1 to Edc3 can contribute strongly to Dcp2 partitioning. However, the interaction between the two wild type proteins appears to be effectively saturated, consistent with their strong correlation in concentrations in P-bodies (Figure 3), and increasing their affinity by adding a second HLM1 element to Dcp2 does not draw more of the protein into P bodies. In general, while partitioning into a condensate can be increased by strengthening weak interactions between a pair of components, the magnitude of such an effect is limited as binding nears saturation.

Figure 6 with 1 supplement see all
Partitioning of Dcp2 into P bodies can be saturated.

(A) Schematics of domain architecture of Dcp2 FL, Dcp2 FL-H1, Dcp2C Δ5H, and Dcp2C Δ5H-H1. (B) Representative images showing dcp1Δdcp2Δ yeast strains expressing GFP tagged Dcp2 FL, Dcp2 FL-H1, Dcp2C Δ5H, and Dcp2C Δ5H-H1, with Edc3-mCherry co-expressed as a P body marker. Scale bar, 5 µm. (C/D) P body concentrations/PCs of Dcp2 FL (n = 49), Dcp2 FL-H1 (n = 79), Dcp2C Δ5H (n = 63), and Dcp2C Δ5H-H1 (n = 52), and mean values (red lines)± SEM.

RNA binding and turnover affect Dcp2 partitioning and dynamics

As the N-terminal domain of Dcp2 possesses both RNA binding and decapping activities, we further asked how these two interactions with RNA, an important scaffold of P bodies, affect partitioning and dynamics. Starting with the Dcp2 300 fragment, we mutated previously reported RNA binding residues, R170, K212, K216, and R229, to alanine, generating Dcp2 300 AAAA (Figure 7ADeshmukh et al., 2008). Dcp2 300 AAAA does not partition into P bodies, despite being expressed similarly to Dcp2 300 (Figure 7—figure supplement 1A), suggesting that RNA binding is important for Dcp2 partitioning (Figure 7B).

Figure 7 with 1 supplement see all
RNA binding and turnover affect Dcp2 partitioning and dynamics.

(A) Schematics of domain architecture of Dcp2 300, Dcp2 300 AAAA, Dcp2ΔH1 and Dcp2ΔH1 WD. (B) Representative images of dcp1Δdcp2Δ yeast strain expressing GFP tagged Dcp2 300 and Dcp2 300 AAAA. Scale bar, 5 µm. (C) Representative images of dcp2Δ yeast strain expressing GFP tagged Dcp2ΔH1 and Dcp2ΔH1 WD under normal and glucose starvation conditions. Scale bar, 5 µm. (D) Inhibition of RNA turnover promotes P body formation. Number of P bodies formed by Dcp2ΔH1 (blue, n = 115) and Dcp2ΔH1 WD (red, n = 91) under normal conditions, Dcp2ΔH1 (light blue, n = 119) and Dcp2ΔH1 WD (light red, n = 144) under 30–60 min glucose starvation. (E) Inhibition of RNA turnover promotes more Dcp2ΔH1 to partition into P bodies. Total fractions of Dcp2ΔH1(blue, n = 20) and Dcp2ΔH1 WD (red, n = 20) without stress, and Dcp2ΔH1(light blue, n = 23) and Dcp2ΔH1 WD (light red, n = 32) with glucose starvation, in P bodies. Numbers show mean values ± SEM. Significance was calculated by the Wilcoxon rank-sum test. ****, p<0.0005 (F) Inhibition of RNA turnover slows Dcp2ΔH1 exchange rate by increasing amount of RNA. Top, FRAP recovery curves of Dcp2ΔH1 (blue, n = 20) and Dcp2ΔH1 WD (red, n = 20) without stress. Bottom, FRAP recovery curves of Dcp2ΔH1 (light blue, n = 23) and Dcp2ΔH1 WD (light red, n = 32) under glucose starvation.

To investigate the effect of catalytic activity on Dcp2 partitioning and dynamics, we made mutations in the Dcp2ΔH1 construct, affording a better dynamic range than the wild type protein (Figure 5B and C). We mutated W50 and D54, which were previously implicated in RNA cap recognition and hydrolysis, to alanine, giving Dcp2ΔH1 WD (Figure 7ACharenton et al., 2016; Floor et al., 2010). We then analyzed Dcp2ΔH1 and Dcp2ΔH1 WD in a dcp2Δ strain, where the former protein should reconstitute significant mRNA decapping activity while the latter should not.

We observed that formation of P bodies was promoted by expressing Dcp2ΔH1 WD (Figure 7C), which is consistent with previous observations that mutations that block decapping catalysis lead to increased P bodies (Sheth and Parker, 2003; Teixeira and Parker, 2007). Dcp2ΔH1 WD formed six P bodies per cell on average, compared to two P bodies per cell for Dcp2ΔH1 (Figure 7D), despite being expressed at similar levels (Figure 7—figure supplement 1B). Changing the catalytic rate of decapping also changed the accumulation of Dcp2 in P bodies and altered its dynamics. Specifically, ~10% of Dcp2ΔH1 WD was concentrated in visible P bodies compared with ~4% for Dcp2ΔH1 (Figure 7E). The exchange rate also decreased from 0.041 s−1 to 0.017 s−1 and fractional recovery decreased from 0.85 to 0.65 in the WD mutant (Figure 7F).

To further test whether these changes are due to increasing amounts of cellular mRNA or to disrupting catalytic activity of Dcp2 per se, we analyzed these two variants after 30–60 min of glucose starvation, conditions in which mRNA is increased due to translation repression. We observed that under these conditions Dcp2ΔH1 and Dcp2ΔH1 WD showed similar dynamics (exchange rates of 0.021 s−1 and 0.013 s−1, respectively, and fractional recoveries of 0.57 and 0.60, respectively, Figure 7F) and a similar distribution of the number of P bodies per cell (Figure 7D), suggesting that the differences observed in the non-starved conditions were due to differences in mRNA concentration. However, even in the starved conditions, the Dcp2ΔH1 WD mutant still shows a larger total fraction in P bodies than the Dcp2ΔH1 protein (Figure 7E), indicating that catalytic activity does control some aspects of P body formation. Together these data suggest that changes in the partitioning and dynamics of the Dcp2 WD mutant are caused by a combination of increase mRNA concentration and loss of catalytic activity.

Discussion

This work presents the first quantitative description of an RNP granule, which is representative of the broader class of non-membrane bound organelles referred to as biomolecular condensates. This analysis has exposed features of yeast P bodies that should be generalizable to other condensates.

Two classes of P body components

A major contribution of this work is to demonstrate two distinct classes of proteins within yeast P bodies. Members of one class, which we name the highly concentrated (HC) P body proteins, are highly enriched, with PC ≥30 and P body concentrations > 5 µM. These include Dcp2, Edc3, Pat1, the Lsm1-7 complex, Xrn1, Dhh1, and Upf1 (colored red in Figure 1). On average, these proteins show slower dynamics of exchange from P bodies and a small fraction of exchangeable molecules. In contrast, the less concentrated (LC) P body proteins, Not2, CCR4, Pop2, Upf2, Upf3, Hek2, Eap1, Edc1, Bre5, Psp2, Sbp1 and Ssd1, have PC <15, lower concentrations in P bodies, and show rapid exchange with the cytoplasm and a large fraction of exchangeable molecules. Thus, despite their apparent complexity from the literature, yeast P bodies are made up of a group of highly concentrated proteins, with much lower concentrations of additional factors. This illustrates a likely general principle that biological condensates that appear quite complex by qualitative analyses of their components, may in fact have a much simpler primary complexity and organization.

The differences between HC and LC P-body proteins are probably closely related to their intermolecular connectivity patterns within P bodies, which are listed in Supplementary file 3 based on literature data. With the exception of Xrn1 (see below), the HC proteins all have high valency of interaction (number of interacting molecules) and high connectivity to other P body proteins and RNA (≥4 directly interacting molecules). This connectivity should afford increased partitioning into the condensate and reduced exchange rates and fractional recovery with the surrounding cytoplasm. It should also enhance the ability of the HC proteins to assemble, and thus contribute significantly to P body formation. In contrast, all LC proteins have low valency and low connectivity, and usually interact with RNA and/or only one HC protein. These features should generally afford lower P body concentrations and more rapid and complete exchange. Since many of the LC components are RNA binding proteins (Supplementary file 3 and Figure 2—figure supplement 4), we suggest they are recruited into P bodies by the high local concentration of RNA. The LC components with the highest PC (Pop2, Ccr4, Upf2, Upf3) all also have direct interactions with a HC P body component (Pop2 interacting with Dhh1, Upf2 interacting with Upf1), consistent with the idea that partitioning is determined by cooperative interactions of a given protein with multiple components within a condensate (see below). Thus, connectivity, and its molecular underpinning, valency of interaction, is likely to provide significant predictive power in understanding the composition of condensates.

We note that while connectivity is significantly correlated with P body concentration and dynamics, the correlation is not perfect, likely because affinity and energy-consuming processes, in addition to connectivity, can play a significant role. For example, the PC of the 1–300 fragment of Dcp2, which includes the HLM1 high affinity binding site for Edc3, is 132. In contrast, the PC of the Dcp2∆H1 ∆5H, where HLM1 is replaced by lower affinity binding sites for Edc3 in the C terminal extension of Dcp2, is only 31. Similarly, the HC protein Xrn1 is concentrated in P bodies to ~10 µM even though it only has two connections to other P body components. This is presumably due to the high affinity of the protein for RNA (Chang et al., 2011; Banani et al., 2016). Finally, the exchange dynamics of Dhh1 appear to be governed by its rate of RNA-dependent ATP hydrolysis (Mugler et al., 2016). Thus, while connectivity patterns can be a useful guide to condensate composition and behavior, other molecular details must be considered to develop a complete understanding.

Assembly of P bodies through cooperative interactions

Our data also provide insights into the nature of P body assembly. The positive correlations between HC protein concentrations (Figure 3) suggest that these molecules partition into P bodies cooperatively. This cooperativity likely arises from the high connectivity among the HC proteins, such that partitioning of one molecule promotes partitioning of others that interact with it directly and indirectly. Given the high concentration and multivalency of HC P body components one would predict that they would all be able to contribute to P body assembly. Indeed, this is what is seen in the literature. The three most enriched P body components, Dcp2, Pat1, and Edc3, are the three major proteins shown to affect P body assembly (Buchan et al., 2008; Decker et al., 2007; Teixeira and Parker, 2007). Other HC proteins, Dhh1, Upf1, and the Lsm1-7 complex can contribute to P body assembly as well (Decker et al., 2007; Sheth and Parker, 2006; Hondele et al., 2019; Rao and Parker, 2017). In contrast, none of the LC components are required for efficient P-body assembly. This illustrates a second principle that interactions that produce condensates are distributed across their highly-valent components. These act with differing degrees of cooperativity to promote formation of the larger assembly. This principle can explain why smaller P-bodies can still assemble in cells lacking one HC protein component (Rao and Parker, 2017).

It is notable that in the dcp1Δ strains, the HC components, Dcp2:Edc3:Pat1:Lsm1:Dhh1:Xrn1, are present at roughly equimolar concentrations (~10 µM), with the exception of Upf1, which is ~2 fold lower. This suggests that perhaps these components form a discrete, stoichiometrically defined RNP complex which then assembles to higher order to form the condensate. However, our data speak against this extreme of high cooperativity. First, the ratio of Dcp2 to Edc3 ranges from 0.5 to 2.3 in our analysis, inconsistent with a strictly defined ratio expected from a stoichiometric complex on every mRNA. Cast differently, the cooperativity in P body concentrations illustrated in Figure 3 is significant, but not as high as expected from a discrete complex. Further, the P body concentrations of Pat1 and the Lsm1-7 complex drop about two fold in wild type strains under glucose starvation conditions relative to the dcp1Δ strains, again speaking against a discrete assembly and suggesting that the interaction network of the HC proteins could be altered under different conditions.

An unanswered question is whether intermolecular RNA-RNA interactions contribute to P body assembly. This possibility is suggested by the robust self-assembly of RNA, the roles of intermolecular RNA-RNA interactions in stress granule assembly (Van Treeck et al., 2018; Tauber et al., 2020), the observation that RNA is required for P body formation (Teixeira et al., 2005), and the fact that every P body protein interacts with RNA (Supplementary file 3 and Figure 2—figure supplement 4). P body formation clearly requires protein interactions since Dcp2, Edc3, Dhh1, Lsm1-7 complex and Pat1 have all been genetically shown to promote P body formation (Decker et al., 2007; Teixeira and Parker, 2007; Rao and Parker, 2017; Hondele et al., 2019). In some cases, the assembly of P bodies can be directly attributed to specific protein-protein interactions, such as the requirement for Edc3 dimerization to promote P body formation (Ling et al., 2008). Moreover, several P body proteins either in isolation, or in mixtures, can undergo LLPS to create P body like assemblies in vitro (Fromm et al., 2014; Schütz et al., 2017). Whether the mRNAs in P bodies simply serve as a platform to facilitate interactions between HC P body components, or also form intermolecular RNA-RNA interactions contributing to P body formation remains to be seen.

General principles of scaffolds and clients in natural condensates

We previously proposed that the components of biomolecular condensates could be classified into two groups, scaffolds and clients (Banani et al., 2016; Ditlev et al., 2018). Scaffolds are defined as components that are required for condensate formation, while clients are not necessary for integrity of the condensate, but are recruited by interacting with scaffolds. As corollaries, the model posited that there should generally be few scaffolds and many clients in natural condensates, and that scaffolds should have high valency of interaction elements, while clients should have lower valency. In addition, scaffolds should have higher concentration within the condensate than clients, since the former recruit the latter. The model was generated based largely on the behaviors of engineered multivalent macromolecules, where the distinction between scaffold and client was stark by design. In this view, the yeast P body scaffolds include RNA (Teixeira et al., 2005) as well as Dcp2, Edc3, and Pat1, Lsm1-7 and Dhh1 (Decker et al., 2007; Teixeira and Parker, 2007; Hondele et al., 2019), since deletion of these factors impairs formation of the condensate. Notably, all of these proteins are in the HC group and have high valency of interaction and high connectivity. Under the same conditions, no molecules in the LC group are known to impair P body formation (although when cells are at high density, deletion of Sbp1 produces smaller P bodies [Segal et al., 2006]). Thus, information on the relative concentrations of proteins in a condensate (the HC vs LC distinction) is likely to aid prediction of molecules that play significant roles in generating the structure (and be imperfectly correlated with connectivity, as described above).

Yet a simple scaffold/client classification does not account for the differences in effect of the various deletions; for example loss of RNA completely eliminates P bodies, whereas loss of the other factors has only a partial effect. Thus, rather than a black-and-white classification, we have come to believe that scaffold and client are better used as descriptors, where condensate components are more scaffold-like or more client-like depending on the degree to which their deletion affects the cooperative assembly of other components. Scaffold-like components contribute more strongly to condensate formation; their deletion should decrease the size and number of condensates. Client-like components should have lesser or no such effects. In the context of phase transitions, the descriptor characterizes the degree by which a given condensate component influences the multi-dimensional phase boundary of the system, both the position of the boundary and its shape, since effects of deletion are unlikely to be evenly distributed across all molecules in a condensate (see below). Deletion of scaffold-like molecules will produce larger changes and deletion of client-like molecules will produce more subtle effects.

Intuitively, scaffold-like components should have higher connectivity among molecules in the condensate, and will be more central in the interaction network, whereas client-like molecules should have low connectivity and be more peripheral in the network. A similar concept has been described in the context of stress granules recently (Yang et al., 2020). In addition to affecting formation of the condensate, highly connected molecules are expected to have greater influence on the composition of the structure. Deletion of scaffold-like molecules will change partitioning of numerous components, while deletion of client-like molecules will only affect partitioning of their immediate interacting components in the network. Importantly, since connectivity in a condensate is not homogeneous (different molecules have different connectivity), deleting molecules is likely to have heterogeneous effects on composition, changing the relative ratios of components. Thus, the thermodynamics of forming a condensate and the composition of the resulting structure will be coupled, based on the connectivity patterns in the interaction network. Further exploration of these ideas will require large-scale, systematic analysis of condensate composition in the presence of individual deletions of multiple components.

Compositional specificity

A major question regarding biomolecular condensates is how the composition of the compartments is determined in vivo. Our data suggest that recruitment of Dcp2 into P bodies is distributed across a large number of interaction elements, including a high affinity Edc3-binding motif, an RNA binding domain and weak multivalent interactions with Edc3. This organization provides several insights into compositional specificity that should be generalizable to other molecules.

First, simple mass action will concentrate components that bind with high affinity to scaffolds. For example, adding the high affinity HLM1 to the weakly partitioning Dcp2C Δ5H promotes its partitioning into P bodies to a level similar to that of wild type Dcp2. High affinity binding to RNA likely also explains the strong partitioning of Xrn1 into P bodies, even though it makes few known interactions with other components and its loss does not have deleterious effects to formation of the condensates (i.e. it is more client-like than scaffold-like in its properties).

Second, increasing affinity for a condensate component does not increase partitioning if affinity is already high. We found that adding a second HLM1 element to full length Dcp2, which should increase affinity for Edc3, does not increase its partition coefficient. The upper bound may be determined by limiting concentrations of scaffold-like interaction partners. This also indicates that partitioning may be more readily tuned when it is mediated by weak multivalent interactions rather than by high affinity interactions.

Third, protein elements that bind scaffold-like components weakly will not partition strongly into a condensate, but when two such elements are fused together, even if they bind different scaffold-like components, they can be recruited strongly. We found that two fragments of Dcp2, which bind RNA and Edc3, respectively, partition only weakly into P bodies individually. Yet when fused together, they are recruited strongly. This behavior is analogous to previous observations that cooperativity between IDRs and folded domains can promote recruitment into both phase separated droplets in vitro and P bodies in cells (Lin et al., 2015; Protter et al., 2018). Such effects, similar to avidity effects in canonical molecular interactions, will greatly narrow the specificity of recruitment, even for scaffolds that individually bind ligands promiscuously. This mechanism also provides ready routes for evolution of new clients through genetic rearrangements that fuse together multiple low-affinity interaction elements. Distributing recruitment across a large number of interaction elements would also render composition less susceptible to mutations, which could lead to evolutionary selection. A similar mechanism may be applied to RNA partitioning as well, since a predominant metric for mRNA partitioning into P bodies or stress granules is length, which may be a simple proxy for the number for interactions (Khong et al., 2017; Matheny et al., 2019).

Quantitative considerations of condensate function

Quantitative analyses as we have performed play an important role in assessing the functions of condensates. For example, several condensates, including P bodies, have been proposed as sites for sequestration/storage of biomolecules, in part because they appear as qualitatively bright puncta by microscopy, and in part because their disruption can activate certain processes (Arimoto et al., 2008; Decker and Parker, 2012; Li et al., 2000). However, when total condensate volume (typically <1–2% of cytoplasm/nucleoplasm), and partition coefficients (2 ~ 200) are quantified, it is evident that only a small fraction of most molecular species are sequestered within condensates. (Figure 4, Leung et al., 2006; Rao and Parker, 2017; Wheeler et al., 2017). While in some processes small changes in the amount of available species in the cytoplasm/nucleoplasm could have functional consequences, in others, different mechanisms must be considered. For example, rather than sequestration, an inhibitory catalytic function within a condensate could also explain the activation of a process upon condensate disruption. Such considerations will further advance the already significant impact of biochemical reconstitutions of phase separation on understanding the functions of biomolecular condensates (Fromm et al., 2014; Schütz et al., 2017; Woodruff et al., 2017).

Conclusions

Together, our work suggests that condensates may generally be organized around a relatively small number of highly concentrated, less dynamic scaffold-like components. This construction would provide a relatively simple route for a condensate to appear during evolution, in that only a small number of proteins would need to develop the ability to assemble cooperatively. Once this structure was established, other proteins could evolve the ability to interact with the scaffold-like components and consequently be recruited into the compartment to different degrees determined by their connectivity, interaction affinity and cooperativity among different interaction elements. A composition of this nature also indicates that condensates are compositionally less complicated than suggested by proteomics studies, where tens to hundreds of proteins have been annotated as residents of particular condensates. Our quantitation indicates that most components are present in only small amounts, and the majority of the protein mass derives from only a few types of molecules. Such an understanding greatly simplifies efforts to reconstitute condensates in vitro, and can frame models of their biochemical functions.

Materials and methods

Yeast strains

Request a detailed protocol

Yeast strains used in this study are listed in Supplementary file 1. GFP or mCherry tagged proteins used to generate Figures 14 are expressed from their endogenous locus. Yeast strains carrying plasmids were constructed using lithium acetate-based transformation (Gietz and Schiestl, 2007).

Plasmid construction

Request a detailed protocol

Plasmids used in this study are listed in Supplementary file 4. All Dcp2 mutants were expressed under the DCP2 promoter on a low-copy centromeric plasmid (pRP1902) as previously reported (Harigaya et al., 2010). Dcp2 point mutations were made by site-directed mutagenesis using KOD Xtrem Hot start DNA Polymerase followed by Dpn1 digestion. Dcp2C ∆5H, and Dcp2∆H1 ∆5H were constructed by Gibson assembly into the vector used for Dcp2 FL, pRP1903 (Harigaya et al., 2010). N-Dcp2 was also constructed from pRP1903.

Yeast growth conditions dcp1Δdcp2Δ strains and dcp2Δ strains expressing GFP tagged Dcp2 mutants were grown in synthetic medium lacking uracil but containing 2% glucose. dcp1Δdcp2Δ strains expressing both GFP tagged Dcp2 mutants and Edc3-mCherry were grown in the same media also lacking lysine. Glucose starvation in Figure 2—figure supplement 3Figure 4—figure supplement 1A and Figure 7C was performed by exchanging with the corresponding synthetic medium lacking 2% glucose for durations indicated in the text. Stationary stage in Figure 4—figure supplement 1B was achieved by growing cells for 5 days and OD600 >6. For imaging, cells were grown at 30°C until OD600 = 0.4 ~ 0.6, and then immobilized on concanavalin-A (Sigma-Aldrich) coated glass bottom dishes (MatTek).

Image acquisition and analysis

Request a detailed protocol

All images were acquired using a Leica SP8 Laser Scanning Confocal Microscope using a 100 × 1.4 NA oil immersion objective. Images were analyzed using Fiji.

Identification of P bodies

Request a detailed protocol

For strains expressing only GFP tagged protein, P bodies were identified by thresholding the GFP fluorescence intensity using the MaxEntropy algorithm in Fiji. For strains expressing both GFP tagged proteins and Edc3-mCherry Figure 5 and Figure 6, Edc3-mCherry signals were thresholded (MaxEntropy) to identify P bodies, and created masks. Absolute concentrations of each Dcp2 mutants in P bodies and partition coefficients (P bodies within masks created by Edc3-mCherry signals) were analyzed as described below, using Edc3-mCherry signals to quantify size of P bodies. To determine fractions of cells having GFP puncta in Figure 5E, PC greater than two was chosen arbitrarily that punctate localization could be observed in GFP channel. Because the formation of and partitioning into biomolecular condensates are sensitive to protein expression levels, we eliminated cells with low expression (bottom 10% of the populations) and high expression (top 10% in the populations) of GFP and mCherry.

Measurements of absolute concentrations in P bodies and cytoplasm, and partition coefficients

Request a detailed protocol

To quantify the absolute concentrations of GFP-tagged proteins in P bodies, we assume that P bodies are spherical (based on the similar diameters in x and y when measured), and correct their measured intensities based on the point spread function (PSF) of our microscope (Fink et al., 1998). We determined the PSF using 0.2 µm fluorescent microspheres (Invitrogen) imaged with the same optics, filters, zoom settings and pinhole settings used throughout our study. We then modeled (Matlab) the intensity-diluting effect of the PSF when imaging spheres of different sizes through convoluting the PSF with the sphere. This yielded a correction curve relating sphere diameter to the fraction of true maximum intensity actually measured in the image (Figure 1—figure supplement 2Fink et al., 1998), assuming all fluorescence intensity derived from the sphere and none derived from the surroundings. We limited our cellular analyses to P bodies with measured diameter >0.33 µm (most are 0.4–0.8 µm in dcp1Δ strains), which is 1.1 times larger than the x-y PSF, and thus the size of P body can be accurately measured as the full width at half maximum intensity (FWHM) of the object. From the measured diameter in the x-y dimension, an assumption of spherical shape, and the correction curve, we determined the calibration factor (CF) for the P body intensity.

To determine P body intensity (IPbody,measured), we first measured the maximum intensity of P body and then drew a one pixel circle around it to find the surrounding pixels. Nine pixels including the maximal one were averaged to get the IPbody,measured. Since the correction was based on the assumption that all fluorescence intensity derived from the object, we applied it only to the incremental intensity of the P body over the cytoplasm (IPbody,measured-Icyto); cytoplasmic intensity also contributes to IPbody, measured, but is homogenous across the cell and should not be corrected for the PSF effect. The real maximum intensity of P body(IPbody) was thus calculated as [(IPbody,measured-Icyto)/CF + Icyto]. Cytoplasm intensity was calculated by averaging the mean intensities of three ROIs the same size as P bodies in the cytoplasm.

We used standard curves of the fluorescence intensities of GFP solutions imaged with identical parameters as yeast to convert IPbody and Icyto to absolute concentrations in P bodies (CPbody) and the cytoplasm (Ccyto). Because the intensities of P bodies marked by different proteins have a large dynamic range, to avoid saturation of our camera, we imaged them using different laser powers and gain settings, and generated different GFP standard curves accordingly. Partition coefficients were calculated as CPbody/Ccyto.

Total fractions in P bodies, average cellular concentration and number of molecules per cell

Request a detailed protocol

We collected z-stacks of yeast cells with a 0.22 µm step size. To calculate the number of molecules in cytoplasm (Ncyto), we first measured the cell volume. Diameters in x, y and z directions were measured manually with Otsu thresholding to determine the cell boundaries in Fiji. Cell volume was calculated as Vcell = 4/3*π*(x/2)*(y/x)*(z/2), assuming that yeast cells are ellipsoidal. Previous studies have shown that the cytoplasmic volume of a yeast cell is about 67% of the total cell volume (Uchida et al., 2011; Yamaguchi et al., 2011). To calculate the number of molecules in P bodies (NPbody), we measured the x-y diameter of each P body in the cell and calculated its volume by assuming a spherical shape. The concentrations in the cytoplasm (Ccyto) and P body (CPbody) were determined as described above. Ncyto and NPbody were calculated as Ncyto = Vcell x 0.67 x Ccyto x NA (6.02 × 1023). NPbody = (VPbody x CPbody)  x NA (6.02 × 1023). Average cellular concentration = ((VPbody x CPbody)  + Vcell x 0.67 x Ccyto)/Vcell x 0.67. The fraction of molecules in P bodies is thus NPbody/(Ncyto + NPbody).

Fluorescence recovery after photobleaching (FRAP)

Request a detailed protocol

Selected P bodies were bleached using an 0.5 W 488 nm laser at 60% laser power. Images were collected from a single plane using a 2.5 airy unit pinhole at 5 s intervals for 150 s. Fluorescence intensities were analyzed manually in Fiji. Background intensities (Ibackground) were first subtracted. Because yeast cells are small, the cytoplasm may be bleached slightly while bleaching the P bodies. We thus measured the average fluorescence intensities of cytoplasm (excluding the bleached P body) before bleaching (Icytobefore) and in the first frame after bleaching (Icytoafter) to account for this effect. An unbleached P body was used to correct for the photo-bleaching during image acquisition in the recovery phase (Iunbleached). The corrected recovered intensities (Irecovery) were normalized to the intensities pre-bleaching (Ipre-bleaching).

Irecovery=(I-Ibackground)(Iunbleached-Ibackground)×(Icytoafter-Ibackground)(Icytobefore-Ibackground)
It=Irecovery(Ipre-bleaching-Ibackground)

Normalized intensities were fitted to a single exponential recovery (one-phase association function in Prism) (GraphPad Software).

I(t)=I+(I0-I)e-kt

where I0, I and k were fit as intensity immediately after bleach, intensity at long times and the rate constant for recovery, respectively. The fractional recovery was calculated as:

F= I-I01-I0

inverse FRAP (iFRAP)

Request a detailed protocol

For iFRAP, the whole cytoplasm except one P body was bleached three times for a total of 1.5 s using a 0.5 W 488 nm laser at 100% laser power. Fluorescence intensities were analyzed as in the FRAP experiments above. Because intensities of the unbleached P body were likely affected during bleaching, we normalized the intensity to the intensities of P body in the first frame after photo-bleaching (Ipre).

It=(I-Ibackground)(Iunbleached-Ibackground)×1(Ipre-Ibackground)

Normalized intensities were fitted to a single exponential decay (one-phase decay function in Prism) (GraphPad Software).

I(t)=I+(I0-I)e-kt

where I0, I and k were fitted parameters. The fractional decay was calculated as:

F=I0-I

Construction of connection map and calculation of centrality

Request a detailed protocol

Protein-protein and protein-RNA interactions were summarized from literatures as shown in Supplementary file 3. When counting number of interactions each protein makes to others, Lsm1-7 complex was treated as a whole entity, and all the other proteins were treated as individual molecules. The connection network was generated using Cytoscape (Shannon et al., 2003). Eigenvector centrality of each node was calculated using CytoNCA in Cytoscape (Tang et al., 2015), method in which connections to highly connected nodes contribute more to score of the questioned node than equal number of connections to less connected nodes. The size of nodes and the distance of other nodes to RNA node were manually adjusted to reflect eigenvector centrality.

Western blot

Request a detailed protocol

Yeast total extracts were prepared as previously described (Knop et al., 1999). 1.5 × 108 cells from OD600 = 0.4–0.6 cultures were resuspended in 1150 µl lysis buffer (0.24 M NaOH, 1% β-mercaptoethanol, 1 mM EDTA, 1 mM PMSF, 5 µM Pepstatin A, 10 µM Leupeptin). After incubation on ice for 20 min, 150 µl 55% trichloroacetic acid (TCA) was added to precipitate proteins on ice for 20 min. The mixture was centrifuged at 16100 rpm at 4°C for 10 min. The pellet was resuspened in 250 µl HU buffer (8 M urea, 5% SDS, 200 mM Tris-HCl [pH 6.8], 1 mM EDTA, 5% β-mercaptoethanol, and 1% bromophenol blue) and incubated at 65°C for 10 min, followed by 16100 rpm centrifugation at RT for 5 min. The supernatant was used for subsequent analyses. Immunoblotting was performed with primary antibodies: rabbit-anti-GFP (1:2000) (Abcam), and mouse-anti-PGK1 (1:1000) (Abcam). Mouse-anti-rabbit-IgG (1:10,000) (Santa Cruz) and goat-anti-mouse-IgG (1:10,000) (Santa Cruz) were used as secondary antibodies.

Quantifications and statistical analysis

Request a detailed protocol

Detailed statistics including number of cells analyzed, mean value, standard deviation and standard error of the mean are indicated in each figure legend. The Wilcoxon rank-sum test was performed using GraphPad Prism (GraphPad software). The Fligner-Killeen test was performed using R. Significance was determined as: ***, p<0.001; ****, p<0.0005.

Data availability

All data have been submitted to Dryad (https://doi.org/10.5061/dryad.02v6wwq0q).

The following data sets were generated
    1. Xing W
    2. Muhlrad D
    3. Parker R
    4. Rosen MK
    (2020) Dryad Digital Repository
    Data from: A quantitative inventory of yeast P body proteins reveals principles of composition and specificity.
    https://doi.org/10.5061/dryad.02v6wwq0q

References

Decision letter

  1. Robert H Singer
    Reviewing Editor; Albert Einstein College of Medicine, United States
  2. James L Manley
    Senior Editor; Columbia University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The work represents an effort to analyze the enrichment and dynamic properties of approximately 20 P body proteins in yeast. The descriptive part of the study is comprehensive (31 proteins included in the initial survey, 19 whose abundance was quantified) and suggest the existence of two groups: proteins strongly enriched in P bodies – several of which are also required for P body assembly – and proteins that are less enriched – fewer of which affect P body assembly. Overall the picture is one of a dynamic condensate.

Decision letter after peer review:

[Editors’ note: the authors were asked to provide a plan for revisions before the editors issued a final decision. What follows is the editors’ letter requesting such plan.]

Thank you for sending your article entitled "A quantitative inventory of yeast P body proteins reveals principles of compositional specificity" for peer review at eLife. Your article is being evaluated by 3 peer reviewers, and the evaluation is being overseen by a Reviewing Editor and James Manley as the Senior Editor.

Given the list of essential revisions, including new experiments, the editors and reviewers invite you to respond within the next two weeks with an action plan and timetable for the possible completion of the additional work. We plan to share your responses with the reviewers and then issue a binding recommendation.

Summary from a reviewer: This manuscript reports on a quantitative analysis of P body composition in yeast. The authors characterize the concentration, enrichment relative to cytoplasmic pool, and dynamics of 31 reported P body proteins tagged with GFP. They identify 19 that were sufficiently concentrated in P bodies under the conditions tested and divide them into two classes: a "core group" comprising 7 proteins with high enrichment and low dynamics, and the remainder 12 proteins with low enrichment and high dynamics. The authors argue that these observations support a scaffold/client model for condensate assembly where the number of interaction partners for a given protein (valency) can be used to predict its enrichment in condensates/P bodies.

All reviewers had similar issues about the work that revolve around novelty, quantitation, the fit of the data to their model and the lack of experimental testing. If you feel that you can address these concerns expeditiously, we will consider a revised manuscript.

The main concerns of the reviewers can be summarized as follows:

1. The novelty of the work is not emphasized. Clearly it is more comprehensive than previous work of these authors, but it needs some more perspective and additional testing.

2. The authors need to consider other models for assembly than just client/scaffold. The discrepancies between the data and their model are not resolved. There are several experiments detailed that could do more to enlighten the murky conclusions (see below). For instance, rather than tagging with a single fluorophore, can they estimate composition and variation better with a two fluorophore pairwise analysis.

3. Can they provide any additional data or insight to shed light on the discrepancy that they point out between their model and the results for DCP2?

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "A quantitative inventory of yeast P body proteins reveals principles of compositional specificity" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The reviewers feel that they cannot support a resubmission of this work. The reasons for this are best summarized by one of the reviewers, who states:

"I am not in favor to invite a resubmission. Unless I misunderstood something, the authors are not planning to address concerns relating to significance and the validity of their model.

The authors state that: "As the reviewers suggest, we believe that A) RNA is a key scaffold of P bodies, B) binding affinity of proteins for each other and for RNA is important and C) valency of interactions is also important".

But offer no experiments/data to support these points.

They also state that: "We did not intend for the data in Figure 7 to represent a test or validation of a valency-based scaffold/client model, or a result stemming from the previous data in the manuscript. Rather, we intended it simply as an empirical correlation that could potentially be useful in making predictions about other condensates where quantitative imaging data are not readily obtainable".

This does not seem to match the article's title, which promises "compositional principles". Their offer to image two components are the same time will not add much beyond confirming what they have already shown?

Finally, the statement that "our data show that only a handful of the many proteins in P bodies are highly concentrated in the structure. This information is key to conceptualizing the constitution of condensates. It speaks against most prevailing models of condensate composition, which cast the compartments as intractably complex, with hundreds of components, all implicitly weighted equally in terms of functional significance. Our data suggest the compartments are much simpler and provide the first routes to reliable biomimetic reconstitution".

Biomimetic reconstitution of RNA granules has already begun, and it is already been shown that few components can be sufficient to mimic condensate structure (Feric et al., 2016, Putnam et al., 2019). The statement also seems to imply that knowledge of the concentration of each factor will help identify such key components, a concept that remains untested."

Reviewer #1:

In this study, Xing et al., quantitatively investigated the composition of yeast P body using an imaging-based method. They found that only a small number of proteins, the core components, are highly enriched in P bodies. The authors demonstrated that the protein concentration within the P body is inversely correlated with the protein cytoplasmic exchange rate. Then the authors further characterized the elements that determine the concentration of Dcp2p in P bodies. By calibrating their system to measure absolute protein concentration in cells, the authors provide a method that could be used to study the composition of other types of cellular granules. The manuscript aim and structure are clear and straightforward. While many of the observations about granule dynamics were previously addressed by the Rosen's and other groups, this work represents a concerted and comprehensive attack on the near-total protein composition and their partitioning behavior in the P-body. To further improve the novelty, we suggest additional questions that could be addressed:

1) A major advance of this study is the precise concentration measurement of GFP-fusion proteins, which is done by comparing the cellular GFP intensity with standard curves. All conclusions depend on these standards and background subtraction. Even though the Rosen lab previously published a paper using a similar approach (Banani et al., 2016), the details of these calibrations and background subtraction should be reported in the supplemental materials.

2) To estimate the protein concentrations in P-bodies, the authors made a fundamental assumption: that all P-bodies have the same and constant composition. However, in Figure 1, the authors showed that the core components of P-bodies, which are also the most abundant proteins, have strong (4-5-fold) variability in concentration and partitioning coefficient. It is unclear how this variability affects the relative quantifications reported in this manuscript (Figure 3). A control experiment that could address this question would be to tag several P-bodies components with spectrally distinct fluorescent proteins (2 or 3 at the time) and verify that the expected relative ratios are recapitulated.

3) Along the same line, it is unclear whether the GFP-tagged proteins, which were used to quantify the P-body components in the dcp1∆ background, were expressed from the endogenous locus (Table 1: YRP1936, yRP2254, yRP2237, yRP2246, yRP2230, yRP1840, yRP1736, yRP2269, yRP1844, yRP1916, yRP1842). The authors should specify this since it would have significant consequences on the quantifications if there were a mixed population of tagged and untagged proteins.

4) In Figure 2, The authors found a negative correlation between the fraction recovery rate and P-body concentration. However, the molecular mechanism behind this correlation is not discussed. The authors should expand the Discussion and link it to the interaction valency.

5) In Figures 4 and 5, the authors measured the partition coefficients of several Dcp2 mutants. It would be important to demonstrate that these mutations do not significantly affect the protein stability and expression by a western blot of these mutants.

6) In Figure 5, the authors suggest that the decapping activity of Dcp2 affects the partition coefficient and recovery rate. However, as the authors mentioned in the text, it is not clear whether this is due to an increase in cytoplasmic mRNA or a direct result from Dcp2 catalytic activity. This issue may be resolved by expressing Dcp2 variants in wild type cells and inducing P-bodies with starvation.

7) In Figure 7, the authors found a correlation between P-body partitioning and protein interaction valency. This finding would be more convincing if the authors could demonstrate that when the interaction valency of a protein increases, this protein will become more concentrated in P-bodies. For example, this can be done by fusing a client protein with a domain of a scaffold protein, like Dcp2C ∆5H. The fusion protein should have more interaction valencies, and the fusion protein would be expected to be more concentrated in the P-bodies. This experiment will also enhance the conclusion made in Figure 6.

Reviewer #2:

This manuscript reports on a quantitative analysis of P body composition in yeast. The authors characterize the concentration, enrichment relative to cytoplasmic pool, and dynamics of 31 reported P body proteins tagged with GFP. They identify 19 that were sufficiently concentrated in P bodies under the conditions tested and divide them into two classes: a "core group" comprising 7 proteins with high enrichment and low dynamics, and the remainder 12 proteins with low enrichment and high dynamics. The authors argue that these observations support a scaffold/client model for condensate assembly where the number of interaction partners for a given protein (valency) can be used to predict its enrichment in condensates/P bodies.

This model was derived by the Rosen group in a previous study that examined the in vitro phase separation properties of an artificial set of proteins with varying numbers of low-affinity binding motifs (valency). The model predicts that "scaffolds" (proteins essential for condensate assembly) exhibit high enrichment and high valency, whereas "clients" exhibit low enrichment and low valency. In accordance with the model, the authors note a rough correlation between valency and enrichment among the 19 P body proteins examined (but see below). To test the model directly, they deleted binding motifs in one predicted "high-valency" P body protein (Dcp2). Surprisingly, they found no strong correlation between valency and enrichment (Many of the Dcp2 mutants do not lie on the correlations in Figure 7. It is not clear to us how to resolve this discrepancy). They also find that some known clients (proteins not essential for P body assembly) exhibit strong enrichment, contrary to the model. Despite these discrepancies, the authors continue to argue that their findings support the scaffold/client model.

It is very surprising that the authors do not consider other models that might fit their data better. One possibility is that RNA functions as the true scaffold for P bodies and that proteins are recruited to P bodies by virtue of their affinity for RNA or for other proteins that binds RNA. In this regard, it is interesting that components of multiprotein complexes are found in roughly stoichiometric amounts in P bodies, suggesting that complexes are preserved in the condensates. In this alternative model, high affinity for RNA by one protein in the complex would be sufficient to enrich the entire complex, even though none of the other subunits in the complex bind to other P body components.

The authors also claim that their analyses suggest that a relatively small number of core proteins (7) make up most of the protein content in P bodies. However, since other P body components may have yet to be discovered, this statement can only by an approximation at this time. Furthermore, even if true, the significance of this hypothesis is unclear – proteins present in low concentrations in P bodies may still have important roles.

In conclusion, while this survey documents concentrations and dynamics for a significant number of P body proteins, it does not appear to provide significant insights into the "rules" that regulate P body composition. The model put forth by the authors is not consistent with the data and other more plausible models are not considered.

Reviewer #3:

This article by the Parker and Rosen labs investigates P bodies components with the aim to provide a quantitative description. P bodies are cytoplasmic foci first described in yeast that are sites where RNA decapping and degradation can occur. P bodies are dynamic, require RNA and are sensitive to the translational status of the cell, where translational repression and stress favor P body growth. Many years of study in the Parker lab and by others has identified a number of critical components of P body assembly and provided evidence that there is not a single factor that is absolutely necessary for P body assembly and that P body assembly can be influenced by multiple and different protein-protein and/or protein-RNA interactions. In this study, the Parker and Rosen lab have used an existing set of 31 GFP tagged P body components and compared their relative abundance and physical properties in P bodies. They find that 7 proteins are highly concentrated and less mobile, they therefore suggest that these are scaffold proteins. Multiple interaction domains apparently provide structure to these components within the granule, while other components which are less enriched are more mobile.

This is a nicely executed study of limited new insight.

- Limits to novelty. It was already known that these 9 proteins are the major proteins of the P body that also play some role in P body function. These earlier studies had also already shown that several of these proteins contain multiple interaction domains through which P body components interact.

- Limits to quantitative analysis. Analysis with confocal has clear limits for spatial resolution and the quantitative measurements are based on fluorescence intensity, not single molecule measurements. Since all components are tagged with the same fluorophore (GFP) no pairwise or higher order analysis could be used to determine variation between granules, protein occupancy, correlation between proteins, to test, as example, whether all components are in a granule at same time, and with similar concentration.

- Unclear relevance of DCP2 analysis. The analysis of DCP2 seems not add much to the conclusions of the paper. Rather than showing that this protein is multivalent as has been shown by domain dissection of other “scaffold” proteins in the past, it would have been informative if the authors had provided whether the “client” proteins have indeed fewer interaction domains.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "A quantitative inventory of yeast P body proteins reveals principles of composition and specificity" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by James Manley as the Senior Editor The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

Summary

This work analyzes the enrichment and dynamic properties of ~20 P body proteins in yeast. The descriptive part of the study is reasonably comprehensive (31 proteins included in the initial survey, 19 whose abundance was quantified) and suggest the existence of two groups: proteins strongly enriched in P bodies – several of which are also required for P body assembly – and proteins that are less enriched – fewer of which affect P body assembly.

The authors examine the domains in DCP2 required for P body localization and identify multiple domains that appear to function partially redundantly. They conclude that a collection of synergistic interactions drive P body assembly, a conclusion reached by other studies as well.

Revisions:

The reviewers felt the manuscript and work makes a contribution to the field, but some revisions are necessary to take into account caveats in the interpretation of the data. For instance, reviewer 2 feels that the term "cooperativity" is not appropriately supported by the data and that mRNA is not considered as a component of the P bodies. Reviewer 3 feels that the functional properties are not sufficiently addressed, particularly in view of specificity (why are some mRNAs and RBPs excluded?).

In sum, the manuscript should be modified to take into account the limitations of the study as indicated by the two reviewers, and their suggestions for revision as indicated below.

Reviewer #2:

In this largely revised manuscript, the authors report on a comprehensive quantification of component proteins of a biological condensate, the Dcp1 depleted P bodies of yeast. Through this they identified seven “core” P body proteins that are partitioned substantially, have a high concentration and slow mobility when enriched in P bodies (Figure 1 and 2). Concentration of these seven proteins in P bodies is positively correlated, consistent with previous data showing that these proteins interact biochemically (Figure 3). The reported enrichment parameters seem similar when Dcp1 depleted P bodies are compared with wild type P bodies under glucose starvation, however Dcp1 and its binding partner Pby1 are missing and RNA degradation is defective in the P body model used throughout the analysis. Thus, while at the level of their biophysical analysis the Dcp1 depleted P-bodies seem similar to P bodies under physiological conditions, functionally these “P bodies” are distinct. This reduces the relevance of this study. In Figure 4 the authors observe that core protein concentrations in P bodies are largely independent of stress conditions, however the authors only considered the protein components, while the regulated entity of stress P bodies are mRNAs and not proteins.

Next the authors dissect DCP2 P body localization in a Dcp1, Dcp2 double null mutant, which still makes some P body (although Figure S1A only shows the presence of small “P body like” structures but no quantitative comparison to control P bodies, in wild type under glucose starvation and Dcp1 depleted). In deletions studies the authors find that various domains of Dcp2 contribute to P body partitioning and show some redundancy between the H1 domain and other helical domains in the CTP. The authors conclude that elements in Dcp2 act “cooperatively”. However, their data show additivity or at most synergy but not cooperativity. In order to use the term cooperativity, the authors need to demonstrate that the initial binding event increases the binding affinity of subsequent binding events. The authors use structure-function study of DCP2 to argue that connectivity, affinity and catalytic activity are three major factors for DCP2 P body recruitment and P body size and number. However, more specific experiments would need to be performed to strengthen this argument or to generalize the finding to other P body components or other types of RNP bodies: 1) doing domain/binding site deletion study on Edc3 or other core proteins; 2) adding valency or high affinity binding site to a client-like protein to turn it into a scaffold (sufficiency). 3) Testing the model with stress granule scaffold proteins.

Overall, this study is technically well done but rather limited in its take home message or novelty. A true functional readout is missing as the experimental set up does not relate to P body function (such as the level of RNA in different granules and decapping activity). The manuscript could easily be condensed into four figures, each making one point for the characterization of a particular type of condensate: Figure 1 and 2: enrichment of certain factors in P bodies, Figures 3 and 4 relationship of concentration between components and under differing conditions, and Figures 5 and 6 structural analysis of one component, suggesting partial redundancy between the CTD helices and Figure 7 with studies addressing RNA binding.

Reviewer #3:

In this revised manuscript, Xing, et al. analyze the enrichment and dynamic properties of ~20 P body proteins in yeast. The descriptive part of the study is reasonably comprehensive (31 proteins included in the initial survey, 19 whose abundance was quantified) and suggest the existence of two groups: proteins strongly enriched in P bodies – several of which are also required for P body assembly – and proteins that are less enriched – fewer of which affect P body assembly. From this the authors suggest that P bodies are "biochemically simpler than suggested by proteomics". This is at best a conjecture and at worst a gross oversimplification since abundance is not necessarily predictive of function – especially for enzymes.

In the second part of the study, they authors examine the domains in DCP2 required for P body localization and identify multiple domains that appear to function partially redundantly. They conclude that a collection of synergistic interactions drive P body assembly, a conclusion reached by several other studies (reviewed in Lu Na and Slavoff, 2018) including in vitro analyses (Shutz et al., 2017).

Overall, we are left with a study that will be a useful reference for future P body studies, but that does not deliver on the promise of "principles of composition and specificity". The question of specificity, in particular (why don't all RNA-binding proteins localize to P bodies?) is not examined directly. The authors only perform structure-function studies on one P body protein and never address what excludes other RNA-binding proteins from P bodies.

Additional comments:

1) The authors are not consistent in the description of the CTD. They suggest first that "N-terminal domain, HLM1, and C-terminal domain are all required for efficient partitioning and maintaining the characteristic slow dynamics of Dcp2". However, later they say that "deleting the CTD has no effect on PC of full length Dcp2". The data presented suggest that the C-terminal domain is not required (300-end) but can compensate when HLM1 is absent.

2) The intact C-terminal domain is never tested; instead the authors used a CTD with 5 of the HLM domains mutated. While the authors imply that this version of the CTD has weaker association with P bodies, no data is shown, and it is unclear from the text why this version of the CTD was used.

3) For the data in Figure 7, the authors use DCP2 constructs with reduced binding to Edc3 to investigate the role of RNA binding and decapping activity in P body association. However, it is not clear why RNA binding was studied using the Dcp2 300 mutant and decapping was studied in the Dcp2ΔH1 mutant.

4) A figure comparing the P body concentration/partition coefficient and exchange rates for the different DCP2 mutants (similar to Figure 2A and B) might be a more intuitive way to compare all the different DCP2 mutants used in this study.

5) The DCP2 structure function studies identify domains in DCP2 that promote P body localization. Whether these domains are sufficient to drive a non-P body protein into P bodies is not addressed. Without such "sufficiency" experiments, whether the results for DCP2 are generalizable for other proteins remains unclear. For example, they suggest that Xrn1 is stably recruited because of its high RNA binding affinity, however, this hypothesis does not address why other high affinity RNA binding proteins are excluded from P bodies.

6) At the end of the Discussion the authors introduce the concept of avidity where multiple interactions result in strong binding. It might be useful to introduce this concept earlier.

https://doi.org/10.7554/eLife.56525.sa1

Author response

[Editors’ note: what follows is the authors’ plan to address the revisions.]

The main concerns of the reviewers can be summarized as follows:

1. The novelty of the work is not emphasized. Clearly it is more comprehensive than previous work of these authors, but it needs some more perspective and additional testing.

On the issue of novelty, we feel our work adds significant new knowledge about P bodies and biomolecular condensates in general. First, our work is the only report of the absolute or relative concentrations of the major proteins within any condensate. Such concentrations are essential knowledge in considering the nature and biochemical functions of P bodies (and by extension, other condensates). For example, models of condensates as storage compartments are cast in doubt by our data in Figure 3, showing that only a small fraction of most proteins are present in P bodies. In addition, models of condensates as sites of high biochemical activity must contend with the fact that most non-core components are only concentrated < 10-fold relative to the surrounding cytoplasm and that most of the molecules are still in the cytoplasm. Condensate functions thus more likely arise from the core components, or from collections of the non-core components (e.g. in a reaction cascade). These are important concepts for the field. Second, our data show that only a handful of the many proteins in P bodies are highly concentrated in the structure. This information is key to conceptualizing the constitution of condensates. It speaks against most prevailing models of condensate composition, which cast the compartments as intractably complex, with hundreds of components, all implicitly weighted equally in terms of functional significance. Our data suggest the compartments are much simpler and provide the first routes to reliable biomimetic reconstitution.

Finally, while we have more work to do to understand the correlations of Figure 7, as an empirical predictor of the degree to which full length proteins are likely to be concentrated within a condensate, they are quite valuable to the field.

2. The authors need to consider other models for assembly than just client/scaffold. The discrepancies between the data and their model are not resolved. There are several experiments detailed that could do more to enlighten the murky conclusions (see below). For instance, rather than tagging with a single fluorophore, can they estimate composition and variation better with a two fluorophore pairwise analysis.

3. Can they provide any additional data or insight to shed light on the discrepancy that they point out between their model and the results for DCP2?

Concerns about the scaffold/client model appear to arise in part from the reviewers’ misperceptions about the model itself, and about our intent in presenting the valency correlations in Figure 7. In retrospect, we understand how these misperceptions arose, and believe that an extensive revision of the text, coupled with additional data, should lead to a much crisper presentation and more accurate model. In brief, we did not intend to portray a model where valency of interactions is the only parameter important in dictating condensate composition, nor where RNA is not important. Rather, as the reviewers suggest, we believe that A) RNA is a key scaffold of P bodies, B) binding affinity of proteins for each other and for RNA is important and C) valency of interactions is also important. We also mistakenly equated “core” proteins with “scaffold” proteins and “non-core” with “client” in the sentence near the start of the Discussion quoted by the second reviewer, rather than describing these as correlated but not identical, which was our intent through the rest of the Discussion. This set off a series of concerns that can be addressed in a revised introduction and discussion.

Relatedly, we did not intend for the data in Figure 7 to represent a test or validation of a valency-based scaffold/client model, or a result stemming from the previous data in the manuscript. Rather, we intended it simply as an empirical correlation that could potentially be useful in making predictions about other condensates where quantitative imaging data are not readily obtainable. This evidently did not come across correctly, and our presentation framed the model as exclusively valency-based, which was not our intent. This can be corrected in a revision. Nevertheless, we agree with the reviewers that it is important for us to understand why the Dcp2 mutants do not fall on the same correlation as the various full-length proteins and have designed an additional set of mutants to understand this better.

The second reviewer also makes an interesting point that the ~1:1 stoichiometry of many of the core proteins suggests that these may be forming a stereotypical assembly that binds RNA and then recruits other proteins, leading to a somewhat different model for composition. We have several thoughts on this issue. First, we agree that the relative stoichiometry is striking. In fact, an earlier version of the manuscript discussed this in some detail, similarly to the reviewer’s comments. However, we eventually deleted this discussion because A) the relative stoichiometry is different in the glucose-starved wild type strains, and B) the wide range of P body concentrations of each component and lack of multi-component correlations in individual yeast cells made it hard to claim a specific assembly. Nevertheless, we agree with the reviewers that our understanding of P body composition would be appreciably strengthened by acquisition of 2-color data, where we could quantify the absolute concentrations of 2 proteins simultaneously in individual cells. While it is impractical to acquire such data on all possible P body protein pairs, we will generate strains to examine key correlations within the core group and between the core and non-core proteins. These data will allow us to distinguish the different models for P body composition.

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

The reviewers feel that they cannot support a resubmission of this work. The reasons for this are best summarized by one of the reviewers, who states:

"I am not in favor to invite a resubmission. Unless I misunderstood something, the authors are not planning to address concerns relating to significance and the validity of their model.

The authors state that: "As the reviewers suggest, we believe that A) RNA is a key scaffold of P bodies, B) binding affinity of proteins for each other and for RNA is important and C) valency of interactions is also important".

But offer no experiments/data to support these points.

We apologize that our initial response did not state clearly enough that our intent was to heavily revise our model, better illustrate its significance, and test it further. We have done all of these things in the revision.

As a brief summary here, we no longer present the work as testing a scaffold/client model for condensate assembly. Rather, we now provide a much more nuanced view of P body assembly, emphasizing the role of high connectivity (related to, but distinct from valency) between both proteins and RNA in producing the condensate, but also discussing the importance of binding affinity and active processes. We also discuss several other models and explain how these are inconsistent with data from our labs and others, both in this paper and the literature. As suggested by the reviewers, and discussed with the editor in April, we have now acquired two-color data (examining correlations between two P body components), and data examining the role of high-affinity binding elements in driving P body recruitment, which both prove instrumental in comparing different models.

Finally, we are now are much more explicit in describing the novelty of our findings, which includes, among 5 points detailed in the accompanying document: A) the discovery that despite the large number of molecules present in P bodies, only a small number (7) are highly concentrated there with large partition coefficients and B) the finding that intermolecular connectivity plays a key role in governing the concentrations of molecules in P bodies.

They also state that: "We did not intend for the data in Figure 7 to represent a test or validation of a valency-based scaffold/client model, or a result stemming from the previous data in the manuscript. Rather, we intended it simply as an empirical correlation that could potentially be useful in making predictions about other condensates where quantitative imaging data are not readily obtainable".

This does not seem to match the article's title, which promises "compositional principles". Their offer to image two components are the same time will not add much beyond confirming what they have already shown?

We have removed the quantitative correlations between valency and physical features of P bodies (original Figure 7), which was problematic for a variety of reasons. Our two-color experiments have provided important tests of potential models for condensate formation. Further, as stated above, we have extensively revised our model.

We believe that the manuscript provides several principles regarding the composition of P bodies. These are described in the revised Discussion section of the paper, which is completely rewritten. These include: A) the concept (supported by evidence) that biological condensates that appear quite complex by qualitative analyses of their components, may in fact have a much simpler primary complexity and organization when considered quantitatively; B) the idea that differences between highly concentrated and weakly concentrated P-body proteins are probably closely related to their intermolecular connectivity patterns within P bodies; C) the idea that interactions that produce condensates are distributed across their highly-valent components, these act with differing degrees of cooperativity to promote formation of the larger assembly; and D) the prediction that thermodynamics of forming a condensate and the composition of the resulting structure should be coupled, based on the connectivity patterns in the interaction network; i.e. deletion of a highly connected molecule should affect both the concentrations of other factors needed to form the condensate as well as the relative concentrations of the components in the condensate.

Finally, the statement that "our data show that only a handful of the many proteins in P bodies are highly concentrated in the structure. This information is key to conceptualizing the constitution of condensates. It speaks against most prevailing models of condensate composition, which cast the compartments as intractably complex, with hundreds of components, all implicitly weighted equally in terms of functional significance. Our data suggest the compartments are much simpler and provide the first routes to reliable biomimetic reconstitution".

Biomimetic reconstitution of RNA granules has already begun, and it is already been shown that few components can be sufficient to mimic condensate structure (Feric et al., 2016, Putnam et al., 2019). The statement also seems to imply that knowledge of the concentration of each factor will help identify such key components, a concept that remains untested."

While the work cited in Feric and Putnam is elegant, both utilize only two components in nucleoli and P granules, respectively. Thus, the degree to which they capture the properties of the cellular structures is unknown. Moreover, to our knowledge, it has not been established that the components used in these reconstitutions are, quantitatively, the dominant components of the cellular structures, or whether other components are equally concentrated. Thus, again, the degree to which these reconstituted structures reproduce the biological structures is uncertain. In our view, a critical advance in biochemical reconstitution of condensates that closely resemble cellular condensates is knowledge of the dominant (i.e. most concentrated) components, and their dynamic properties in vivo. With this information, one can combine the appropriate molecules in vitro at their total cellular concentrations and learn whether they form condensates of appropriate component concentrations (both relative and absolute) and dynamics. If they meet these criteria for multiple components, then one can have some confidence that the biochemistry truly is biomimetic. Without this prior information, one is making an educated guess about which proteins to combine biochemically and how well the reconstituted structure mimics the cellular structure.

Summary from a reviewer: This manuscript reports on a quantitative analysis of P body composition in yeast. The authors characterize the concentration, enrichment relative to cytoplasmic pool, and dynamics of 31 reported P body proteins tagged with GFP. They identify 19 that were sufficiently concentrated in P bodies under the conditions tested and divide them into two classes: a "core group" comprising 7 proteins with high enrichment and low dynamics, and the remainder 12 proteins with low enrichment and high dynamics. The authors argue that these observations support a scaffold/client model for condensate assembly where the number of interaction partners for a given protein (valency) can be used to predict its enrichment in condensates/P bodies.

We have extensively revised the text, and no longer present the work as testing a scaffold/client model for condensate assembly. In fact, based on our data we now present a new, more nuanced view of the scaffold/client nomenclature that better captures the behaviors of natural condensates and will be more useful to the field. We also have removed the analyses describing quantitative relationships between P body enrichment and valency of interactions. We do argue, however, that there is a meaningful qualitative relationship between connectivity in the P body interaction network and molecular behaviors, with a variety of caveats that are now detailed in the Discussion. We say more about each of these issues below.

The main concerns of the reviewers can be summarized as follows:

1) The novelty of the work is not emphasized. Clearly it is more comprehensive than previous work of these authors, but it needs some more perspective and additional testing.

We have extensively revised the Discussion to highlight the new principles that we feel derive from our data. First, our work is the first comprehensive, quantitative analysis of the composition and dynamics of any natural biomolecular condensate. Second, this quantitation has revealed that despite the large number of molecules present in P bodies, only a small number (7) are highly concentrated there with large partition coefficients. This is a large reduction in complexity when considering the formation, regulation and function of P bodies. Third, the quantitation has revealed that only a small fraction of most proteins are present in P bodies, which has important implications for functions of the condensates. Fourth, most (6 of 7) of the highly concentrated proteins/complexes are highly connected in the P body interaction network, and none of the weakly concentrated proteins are highly connected. These correlations suggest that connectivity plays an important role in governing the composition of the structures. It is important to note, though, that other factors, including binding affinity and active processes also contribute to molecular behaviors, and we now explain such complexities in our discussion. Finally, based on our data and other data in the literature, we have revised the notion of scaffolds and clients in biomolecular condensates. We argue that these terms should not be used to classify molecules into binary groups, as we had done before, but rather as qualitative descriptors of the degree to which a molecule contributes to formation and composition of a condensate. Thus, a molecule should be described as more scaffold-like or more client-like depending on whether has greater or lesser effects, respectively. This use of the terms better reflects experimental data, while still capturing the idea that some molecules contribute more than others to the formation and composition of a condensate.

2) The authors need to consider other models for assembly than just client/scaffold. The discrepancies between the data and their model are not resolved. There are several experiments detailed that could do more to enlighten the murky conclusions (see below). For instance, rather than tagging with a single fluorophore, can they estimate composition and variation better with a two fluorophore pairwise analysis.

We now provide a much more nuanced view of P body assembly, emphasizing the role of highly-connected molecules (both proteins and RNA) in producing the condensate, but also discussing the importance of binding affinity and other parameters. As requested by the reviewers we have also performed 2-color imaging to examine correlations between the enrichment between different molecular pairs (new Figure 3). These data revealed significant correlations (Pearson’s R of 0.6-0.7) between Dcp2 enrichment and that of Edc3, Pat1 and Xrn1. The Dcp2-Xrn1 correlation is particularly interesting, as the two proteins are not known to bind each other directly, suggesting correlations mediated by indirect connectivity in the condensate (likely through RNA or other core P-body components). The correlations are not so strong, though, to suggest that a stoichiometrically-defined complex is at the heart of P body formation (a model suggested by one of the reviewers). Moreover, as we argue below, existing data speak against a model where RNA is the only scaffold-like molecule in P bodies (a second model suggested in the review), as a number of proteins have also been shown genetically to play important roles in generating the condensates. These issues – cooperativity in recruitment, the possibility of a stoichiometric complex, the roles of RNA-RNA and RNA-protein interactions, and other factors – are now considered in some detail in the Discussion section of the manuscript. We hope this more nuanced view of condensate formation will appeal to the reviewers.

3) Can they provide any additional data or insight to shed light on the discrepancy that they point out between their model and the results for DCP2?

We have addressed this in two ways. First, we have made additional Dcp2 mutants to understand better how its binding to Edc3 controls it enrichment in P bodies (new Figure 6). These lead to an important conclusion that when affinity between two condensate proteins is low, increasing affinity can increase enrichment, but when affinity is already high, increasing it further does not increase enrichment. Thus, tuning of enrichment likely occurs through alterations in the low/modest-affinity regime. Additionally, as described above, the new discussion provides a more nuanced view of recruitment, as deriving from a combination of connectivity, affinity and active processes. When considered together, these explain the behaviors of our Dcp2 mutants. Essentially, not all regions of Dcp2 contribute equally to enrichment; deleting a high-affinity binding element (to either RNA or Edc3) has a much more pronounced effect on enrichment than deleting a low-affinity element.

Reviewer #1:

[…] To further improve the novelty, we suggest additional questions that could be addressed:

1) A major advance of this study is the precise concentration measurement of GFP-fusion proteins, which is done by comparing the cellular GFP intensity with standard curves. All conclusions depend on these standards and background subtraction. Even though the Rosen lab previously published a paper using a similar approach (Banani et al., 2016), the details of these calibrations and background subtraction should be reported in the supplemental materials.

We have added a detailed description of the image acquisition (shown in Figure 1—figure supplement 2) and analysis procedures in the Materials and methods section.

2) To estimate the protein concentrations in P-bodies, the authors made a fundamental assumption: that all P-bodies have the same and constant composition. However, in Figure 1, the authors showed that the core components of P-bodies, which are also the most abundant proteins, have strong (4-5-fold) variability in concentration and partitioning coefficient. It is unclear how this variability affects the relative quantifications reported in this manuscript (Figure 3). A control experiment that could address this question would be to tag several P-bodies components with spectrally distinct fluorescent proteins (2 or 3 at the time) and verify that the expected relative ratios are recapitulated.

We note that in measuring the absolute concentrations of the various species, we have not made any assumptions about whether P bodies have constant or variable compositions. Assuming that GFP-tagging does not alter the concentrations of a given protein (which our data in Figure 1—figure supplement 3 support) our data capture the range of concentration values sampled by each component across a population of cells. These data show that compositions have substantial variability, as the reviewer notes.

In interpreting the data, the question of relative variability becomes important. Here, we thank the reviewer for suggesting multi-color experiments. In a new Figure 3 we have now used 2-color imaging to simultaneously quantify the concentrations of three pairs of proteins in individual cells, Dcp2/Edc3, Dcp2/Pat1 and Dcp2/Xrn1. We find that in all cases, the P body concentrations of the pairs are significantly correlated (Pearson’s R of 0.6-0.7). Thus, the concentrations of these components fluctuate together (albeit with still some remaining variability, as the R values are not 1). Given the high connectivity of interactions between P body proteins, we believe that many components likely show similar correlations, but a more extensive analysis of this point is beyond the scope of this already lengthy study.

Whether there could be multiple types of P bodies, with different compositional profiles, in a cell population is a difficult question to address. We do not see strong evidence for multimodal distributions in the concentration profiles, which speaks against this possibility. But again, addressing this possibility comprehensively would require vastly more data to assess whether the concentration distributions could be fit better to single, or multiple populations.

3) Along the same line, it is unclear whether the GFP-tagged proteins, which were used to quantify the P-body components in the dcp1∆ background, were expressed from the endogenous locus (Table 1: YRP1936, yRP2254, yRP2237, yRP2246, yRP2230, yRP1840, yRP1736, yRP2269, yRP1844, yRP1916, yRP1842). The authors should specify this since it would have significant consequences on the quantifications if there were a mixed population of tagged and untagged proteins.

All GFP-tagged and mCherry tagged proteins used to generate Figures 1-4 were expressed from their endogenous locus. This is stated in the main text and in the Materials and methods.

4) In Figure 2, The authors found a negative correlation between the fraction recovery rate and P-body concentration. However, the molecular mechanism behind this correlation is not discussed. The authors should expand the discussion and link it to the interaction valency.

As suggested by the reviewer, in the Discussion we now link connectivity in the network (the number of different molecular types a given protein interacts with, which is related to interaction valency, the number of different molecules a given molecule interacts with) to the FRAP behavior. We state “With the exception of Xrn1 (see below), the core proteins all have high valency of interaction (number of interacting molecules) and high connectivity to other P body proteins and RNA (≥ 4 directly interacting molecules). […] These features should generally afford lower P body concentrations and more rapid and complete exchange.”

5) In Figures 4 and 5, the authors measured the partition coefficients of several Dcp2 mutants. It would be important to demonstrate that these mutations do not significantly affect the protein stability and expression by a western blot of these mutants.

In Figure 5—figure supplement 1B and 1C, and Figure 7—figure supplement 1A and 1B, we now show western blots for nearly all Dcp2 mutants examined by microscopy. These data show that proteins compared in the same figure express at approximately the same level. Mutants not examined by western blotting were examined by fluorescence imaging to compare total expression level in the cells analyzed (Figure 6—figure supplement 1). The Dcp2 mutants compared in Figure 6 do express at somewhat different levels, such that Dcp2C∆5H, which partitions weakly into P bodies, expresses ~30% higher than Dcp2C∆5H-H1, which partitions much more strongly (Figure 6 and Figure 6—figure supplement 1). However, the direction of this difference actually strengthens our conclusion that the lower partitioning of Dcp2C∆5H was not because of reduced expression.

6) In Figure 5, the authors suggest that the decapping activity of Dcp2 affects the partition coefficient and recovery rate. However, as the authors mentioned in the text, it is not clear whether this is due to an increase in cytoplasmic mRNA or a direct result from Dcp2 catalytic activity. This issue may be resolved by expressing Dcp2 variants in wild type cells and inducing P-bodies with starvation.

We thank the reviewer for this suggestion. We have now performed this experiment, and the results are summarized in a new Figure 7. In brief, under starvation conditions the dynamics and distribution of number of P bodies are the same for the two mutants, suggesting that these parameters are mainly responding to mRNA levels when catalysis is impaired. However, the total fraction of material in P bodies is still higher for the catalytic mutant during starvation. Thus, this parameter of P bodies appears to be responding to the loss of catalytic activity. We conclude from these data that both the increase in mRNA levels and loss of catalytic activity account for the behavior of the WD mutant.

7) In Figure 7, the authors found a correlation between P-body partitioning and protein interaction valency. This finding would be more convincing if the authors could demonstrate that when the interaction valency of a protein increases, this protein will become more concentrated in P-bodies. For example, this can be done by fusing a client protein with a domain of a scaffold protein, like Dcp2C ∆5H. The fusion protein should have more interaction valencies, and the fusion protein would be expected to be more concentrated in the P-bodies. This experiment will also enhance the conclusion made in Figure 6.

We have removed Figure 7 from the paper, and no longer discuss quantitative relationships between valency and partitioning. The problem with invoking valency alone is that it ignores affinity (an issue that we mentioned in the previous text, but did not emphasize strongly enough), and affinity plays an important role in the partitioning of Dcp2. In the revised manuscript, we have addressed this in two ways. First, we now explicitly state that data here, and in the literature, indicate that RNA binding by the N-terminal domain and Edc3 binding by HLM1 both occur with high affinity, and contribute strongly to Dcp2 partitioning, while the C-terminal HLM elements are lower affinity and contribute less (Figure 5). Second, we have done experiments analogous to those suggested by the reviewer. We added the high-affinity Edc3-binding motif, HLM1, to both Dcp2 ∆5H, which partitions weakly, and to Dcp2 wild type, which partitions strongly. We find that addition of HLM1 only increases the partitioning of the former protein (Figure 6D). We conclude from these data that when affinity for Edc3 is low (Dcp2 ∆5H), adding an additional high affinity binding site can increase partitioning of Dcp2. But when affinity is already high (Dcp2 wild type), increasing it further has no effect. This result is likely general in considering recruitment of proteins into condensates.

Reviewer #2:

[…]

This model was derived by the Rosen group in a previous study that examined the in vitro phase separation properties of an artificial set of proteins with varying numbers of low-affinity binding motifs (valency). The model predicts that "scaffolds" (proteins essential for condensate assembly) exhibit high enrichment and high valency, whereas "clients" exhibit low enrichment and low valency. In accordance with the model, the authors note a rough correlation between valency and enrichment among the 19 P body proteins examined (but see below). To test the model directly, they deleted binding motifs in one predicted "high-valency" P body protein (Dcp2). Surprisingly, they found no strong correlation between valency and enrichment (Many of the Dcp2 mutants do not lie on the correlations in Figure 7. It is not clear to us how to resolve this discrepancy). They also find that some known clients (proteins not essential for P body assembly) exhibit strong enrichment, contrary to the model. Despite these discrepancies, the authors continue to argue that their findings support the scaffold/client model.

We have extensively revised the text and our analyses of our data to address these critiques. Most significantly, we have removed Figure 7 and eliminated all discussion related to a quantitative correlation between valency and P body concentration. As the reviewer states, valency alone cannot account for our data on Dcp2 mutants. We now present a much more nuanced view of how concentration and dynamic properties of P body components appear to be determined. We believe that the connectivity of interactions (the number of molecule types contacted by a particular molecule) plays an important role, as most (6 of 7) of the highly concentrated proteins are also highly connected (≥4 interaction partners), and all of the weakly concentrated proteins have low connectivity (≤2 interaction partners). Further, most (the same 6 of 7) of the highly connected molecules have also been shown to contribute to P body assembly, and none of the low connectivity molecules have been shown to do so. Nevertheless, connectivity is not the only factor, and binding affinity and active processes can also be very important. In this light, we now discuss how interactions of the Dcp2 N-terminal domain with RNA and of the HLM1 element with Edc3 are of higher affinity than other Dcp2 interactions, and thus play more important roles in determining the concentration of the protein in P bodies (explaining the deletion data). Similarly, Xrn1 binds RNA with high affinity. Thus, even though its connectivity is low, it is strongly concentrated in P bodies and shows moderate dynamics. In terms of active processes, ATP hydrolysis by Dcp2 also contributes to the dynamics of the protein. Thus, connectivity is important, but not the whole story in defining the behavior of condensate molecules. These issues are discussed in the new Discussion.

Through these considerations we have come to a different view of how the notions of scaffold and client should be used. Rather than classifying molecules as either scaffold or client, we now feel that these terms should be used as descriptors. A molecule is more scaffold-like if it plays a greater role in P body assembly, and is more client-like if it plays a lesser role. In this way, we can account for the fact that different molecules have different effects upon deletion. E.g. RNA appears to be the most scaffold-like component of P bodies, as its elimination by RNAse destroys the condensates (consistent with the fact that all 19 P body components quantified here bind RNA), while Edc3 and Pat1 are less scaffold-like, since their deletion strains retain some P bodies, albeit fewer and smaller than wild type strains. In contrast, all of the 12 non-core proteins we studied are client-like, since none has been shown to significantly decrease P body assembly when deleted. We note that connectivity (and probably network centrality) does appear to play a role in determining whether a molecule is more scaffold-like or more client-like, as the most connected molecules are all scaffold-like (and more central in the P body interaction network) and molecules with few connections are all client-like (and more peripheral in the network). These issues are discussed in the new Discussion in a new section titled “General principles of scaffolds and clients in natural condensates.”

It is very surprising that the authors do not consider other models that might fit their data better. One possibility is that RNA functions as the true scaffold for P bodies and that proteins are recruited to P bodies by virtue of their affinity for RNA or for other proteins that binds RNA. In this regard, it is interesting that components of multiprotein complexes are found in roughly stoichiometric amounts in P bodies, suggesting that complexes are preserved in the condensates. In this alternative model, high affinity for RNA by one protein in the complex would be sufficient to enrich the entire complex, even though none of the other subunits in the complex bind to other P body components.

As we understand this comment, the reviewer is suggesting that P-bodies form through interactions between mRNAs and then those mRNAs recruit the P-body core proteins by high affinity RNA binding. However, several observations in the literature argue against this model. For example, as we point out in the manuscript, P body formation clearly requires protein-protein interactions since Dcp2, Edc3, Dhh1, Lsm1-7 complex and Pat1 have all been genetically shown to promote P-body formation (e.g. Decker et al., 2007; Sheth and Parker, 2006; Hondele et al., 2019; Rao and Parker, 2017). Moreover, the mechanisms by which proteins promote P-body formation can be directly linked to specific protein-protein interactions, such as the dimerization of Edc3 (Ling et al., 2008). In addition, some proteins require other P-body components for their recuritment to P-bodies. For example, Dcp1 requires Dcp2 to be recruited to P-bodies and the Lsm1-7 complex requires Pat1 (Teixeira and Parker, 2007). Thus, P-body formation, and recruitment of core P-body components to P-bodies, requires protein interactions.

We agree with the reviewer that RNA contributes to P-body formation. Clear evidence suggests that a pool of untranslating mRNAs is required for P-body formation (Teixeira et al., 2005). One clear role for RNA is to provide binding sites for interacting proteins thereby allowing P-body assembly. In this role, we agree that the RNA is functioning in a scaffold-like manner. Whether RNA also contributes to P-body formation through intermolecular RNA-RNA interactions remains to be established. Regardless, a model where RNA is the sole scaffold, which simply binds the protein components of P bodies is unlikely. To clarify this issue in the manuscript, we have added this argument in the new Discussion.

Moreover, as described above, we now present a much more detailed and nuanced view of P body assembly that does not hinge on a simple, binary classification of molecules as either scaffolds or clients. The revised Discussion also addresses the second model proposed here by the reviewer, that the highly concentrated proteins form a stoichiometric complex, which then assembles on RNA to produce the larger structure. A stoichiometric complex would show extremely high cooperativity in recruitment into P bodies, since all elements would enter or exit together. While new Figure 3 does show correlations between the P body concentrations of Dcp2 with Edc3, Pat1 and Xrn1, the correlations are not sufficiently high (Pearson’s R values of 0.6-0.7) to suggest a stoichiometrically defined assembly, even though the proteins have roughly equal average concentrations in the condensate. Further, the P body concentrations of Pat1 and the Lsm1-7 complex drop about two fold in wild type strains under glucose starvation conditions relative to the dcp1Δ strains, again speaking against a discrete assembly. These arguments are presented in a new paragraph in the Discussion.

The authors also claim that their analyses suggest that a relatively small number of core proteins (7) make up most of the protein content in P bodies. However, since other P body components may have yet to be discovered, this statement can only by an approximation at this time. Furthermore, even if true, the significance of this hypothesis is unclear – proteins present in low concentrations in P bodies may still have important roles.

We have removed this claim from the text.

In conclusion, while this survey documents concentrations and dynamics for a significant number of P body proteins, it does not appear to provide significant insights into the "rules" that regulate P body composition. The model put forth by the authors is not consistent with the data and other more plausible models are not considered.

As described in our response to the first summary comment at the top of this review, we feel that our work has revealed a number of new and important principles regarding the formation and composition of P bodies. The new text presents a much more nuanced description of P body assembly that is now consistent with all of our data, and considers a variety of different mechanistic issues and possibilities.

Reviewer #3:

[…]

- Limits to novelty. It was already known that these 9 proteins are the major proteins of the P body that also play some role in P body function. These earlier studies had also already shown that several of these proteins contain multiple interaction domains through which P body components interact.

As described in response to Summary point 1 at the top of this letter, we feel that our work is novel in many respects, including being the first comprehensive, quantitative analysis of any biomolecular condensate. Through this quantitation we revealed two distinct classes of molecules based on concentration in the structures, something that could not have been inferred from previous work, and also the important idea that for nearly all components, only a small fraction of the molecules are present in P bodies. Analysis of these data in light of known molecular interactions also revealed correlations between the physical behaviors of P body components and their molecular connectivity, binding affinity and ATP hydrolysis activity. Again, these could not have been inferred from previous data, as they depend inherently on our quantitative analyses. If the reviewer can point us to relevant literature documenting that the core P-body proteins we identify as being highly enriched have higher partition coefficients than other identified P-body components, we would be happy to revise our manuscript accordingly.

- Limits to quantitative analysis. Analysis with confocal has clear limits for spatial resolution and the quantitative measurements are based on fluorescence intensity, not single molecule measurements. Since all components are tagged with the same fluorophore (GFP) no pairwise or higher order analysis could be used to determine variation between granules, protein occupancy, correlation between proteins, to test, as example, whether all components are in a granule at same time, and with similar concentration.

As detailed above, we have now performed 2-color imaging of several pairs of P body proteins, and find substantial cooperativity between their enrichment. This has led us to a more detailed view of P body assembly, which is described in a virtually completely rewritten Discussion section of the manuscript.

- Unclear relevance of DCP2 analysis. The analysis of DCP2 seems not add much to the conclusions of the paper. Rather than showing that this protein is multivalent as has been shown by domain dissection of other “scaffold” proteins in the past, it would have been informative if the authors had provided whether the “client” proteins have indeed fewer interaction domains.

Our analyses of Dcp2 have dissected the mechanisms by which it is enriched within P bodies. Although the notion of multivalency has been demonstrated for other proteins as well, this dissection provides useful information about Dcp2. Further, in studies of Dcp2 we have found two previously unrecognized properties that are likely relevant to components of other condensates. First, we find that two fragments of the protein that partition only weakly into P bodies individually, partition strongly and cooperatively when fused together. This occurs even though the fragments bind to different P body components, and thus is not the same as simple avidity. This result was far from obvious to us when we initiated the work. Second, in new data presented in new Figure 6, we find that adding a high affinity Edc3-binding element to a Dcp2 protein only increases partitioning into P bodies when the Dcp2 has low affinity for Edc3. When affinity is already high, adding more binding elements has no effect. Thus, partitioning of a condensate component can be saturated, likely due to limiting concentrations of binding partners. Again, this was not an obvious result when we began these experiments.

Regarding proteins with client-like behaviors (see response to Summary point 1 for our new way of using this term), as detailed above, we show in Table 3 and Figure 2—figure supplement 3 that all molecules that are weakly concentrated in P bodies have low connectivity to other P body proteins and are not known to contribute strongly to P body assembly. So indeed, client-like behavior is correlated with limited interactions with other P body components. The correlation is not universally true in the opposite direction. That is, there is small number of proteins that have only few connections to other P body components but yet are highly concentrated in the condensate (Xrn1, Dcp1 and Pby1). These proteins are known to have high affinity for another highly concentrated P body component, which recruits them. These issues are now covered in detail in our revised Discussion.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Revisions:

The reviewers felt the manuscript and work makes a contribution to the field, but some revisions are necessary to take into account caveats in the interpretation of the data. For instance, reviewer 2 feels that the term "cooperativity" is not appropriately supported by the data and that mRNA is not considered as a component of the P bodies.

We have changed our description of the data in Figure 5F, the source of this claim, to “synergistic”, which states that the effect of combining two components together results in a larger effect than the sum of the components individually. In the case of Dcp2, the Dcp2 300ΔH1 construct has a PC of 2.5, and the Dcp2C Δ5H construct has PC of 3.5, while the of Dcp2ΔH1 Δ5H construct, which combines Dcp2 300ΔH1 and Dcp2C Δ5H has PC of 31, which is > 2.5*3.5. Thus, the system meets the definition of synergy. We now note that this effect may result from cooperative binding to molecules in the P body.

The manuscript mentions RNA as an important P body component. We feel that this is sufficient, especially given that the paper is focused on an inventory of P body proteins, not RNAs.

Reviewer 3 feels that the functional properties are not sufficiently addressed, particularly in view of specificity (why are some mRNAs and RBPs excluded?).

Our study does not attempt to address the function of P bodies. This is a complicated issue that would require much more and different data than we can realistically put in the manuscript. Our work does address the specificity of protein recruitment in a significant way through the idea of synergistic recruitment of Dcp2 to P bodies. That is, proteins that are resident often contain multiple weak recruitment elements, and by implication proteins that contain few such elements are recruited poorly if at all. RNA recruitment may follow similar principles, and we now mention this idea explicitly.

Reviewer #2:

In this largely revised manuscript, the authors report on a comprehensive quantification of component proteins of a biological condensate, the Dcp1 depleted P bodies of yeast. Through this they identified seven “core” P body proteins that are partitioned substantially, have a high concentration and slow mobility when enriched in P bodies (Figure 1 and 2). Concentration of these seven proteins in P bodies is positively correlated, consistent with previous data showing that these proteins interact biochemically (Figure 3). The reported enrichment parameters seem similar when Dcp1 depleted P bodies are compared with wild type P bodies under glucose starvation, however Dcp1 and its binding partner Pby1 are missing and RNA degradation is defective in the P body model used throughout the analysis. Thus, while at the level of their biophysical analysis the Dcp1 depleted P-bodies seem similar to P bodies under physiological conditions, functionally these “P bodies” are distinct. This reduces the relevance of this study.

The reviewer is correct that we have studied P bodies under 2 different conditions, only one of which is wild type. We note that this is not wholly different, however, from most studies in the field. For example, most analyses of stress granules induce them with arsenic, a toxin that is only rarely encountered in natural biology. Many other studies involve overexpression of individual components, again a non-physiologic situation. There is nevertheless substantial value in such analyses in understanding the principles by which condensates are assembled, and we feel that our study is in a similar vein. While Dcp1-/- P bodies and glucose starved wild type P bodies may have different functions, what is striking is that their compositions are closely related. Moreover the key patterns of composition are retained – a small number of highly concentrated proteins and a larger number of weakly concentrated proteins – which is an important feature of these cellular structures that has not to our knowledge been examined elsewhere (even the 3 recently published Cell papers on stress granules only reported partition coefficients, which lacks information relative to the absolute concentrations reported here).

In Figure 4 the authors observe that core protein concentrations in P bodies are largely independent of stress conditions, however the authors only considered the protein components, while the regulated entity of stress P bodies are mRNAs and not proteins.

We state that our data do not address changes in mRNA content during starvation: “… nor do they speak to sequestration/storage of RNA…”. We also note that the main point of Figure 4 is to show that for most proteins, P bodies do not substantially deplete the amount of molecules from the cytoplasm, not to compare Dcp1-/- and glucose starved wild type P bodies. If the reviewer is interested, recent results from mammalian cells suggest the composition of P-body mRNAs appears to be largely regulated by their translation rate, perhaps with an additional input from the poly(A) tail length (Matheny et al., 2019).

Next the authors dissect DCP2 P body localization in a Dcp1, Dcp2 double null mutant, which still makes some P body (although Figure S1A only shows the presence of small “P body like” structures but no quantitative comparison to control P bodies, in wild type under glucose starvation and Dcp1 depleted).

We are not entirely sure what the reviewer is asking for here, as we have quantified the PC and dynamics for the various re-expressed Dcp2 mutants. We have now quantified the size distribution of the Edc3 puncta in the Dcp1, Dcp2 double null mutant shown in Figure S5A (Figure S1A is unrelated to these experiments; we assume a typo from the reviewer), and report that in Figure 5—figure supplement 1A.

In deletions studies the authors find that various domains of Dcp2 contribute to P body partitioning and show some redundancy between the H1 domain and other helical domains in the CTP. The authors conclude that elements in Dcp2 act “cooperatively”. However, their data show additivity or at most synergy but not cooperativity. In order to use the term cooperativity, the authors need to demonstrate that the initial binding event increases the binding affinity of subsequent binding events.

As stated above we have changed our description of the data in Figure 5F, the source of this claim, to “synergistic”, which states that the effect of combining two components together results in a larger effect than the sum of the components individually. In the case of Dcp2, the Dcp2 300ΔH1 construct has a PC of 2.5, and the Dcp2C Δ5H construct has PC of 3.5, while the of Dcp2ΔH1 Δ5H construct, which combines Dcp2 300ΔH1 and Dcp2C Δ5H has PC of 31, which is > 2.5*3.5. Thus, the system meets the definition of synergy. We now note that this effect may result from cooperative binding to molecules in the P body.

The authors use structure-function study of DCP2 to argue that connectivity, affinity and catalytic activity are three major factors for DCP2 P body recruitment and P body size and number. However, more specific experiments would need to be performed to strengthen this argument or to generalize the finding to other P body components or other types of RNP bodies: 1) doing domain/binding site deletion study on Edc3 or other core proteins; 2) adding valency or high affinity binding site to a client-like protein to turn it into a scaffold (sufficiency). 3) Testing the model with stress granule scaffold proteins.

We agree that our arguments will be strengthened by additional experiments, and these are planned for additional projects and publications in the future.

Overall, this study is technically well done but rather limited in its take home message or novelty. A true functional readout is missing as the experimental set up does not relate to P body function (such as the level of RNA in different granules and decapping activity). The manuscript could easily be condensed into four figures, each making one point for the characterization of a particular type of condensate: Figure 1 and 2: enrichment of certain factors in P bodies, Figures 3 and 4 relationship of concentration between components and under differing conditions, and Figures 5 and 6 structural analysis of one component, suggesting partial redundancy between the CTD helices and Figure 7 with studies addressing RNA binding.

We agree that we have not attempted to address P body function in this work. We believe that such analyses would require a publication in their own right. Given the lack of figure constraints in eLife, we have not compressed the data into a smaller number of figures, as we fear this would make the paper harder to read and understand (e.g. Figure 4 is not meant to compare different conditions, but rather to make the point that most molecules are not present in P bodies for most species. Hence its title “P bodies do not strongly sequester their resident proteins.”

Reviewer #3:

In this revised manuscript, Xing, et al. analyze the enrichment and dynamic properties of ~20 P body proteins in yeast. The descriptive part of the study is reasonably comprehensive (31 proteins included in the initial survey, 19 whose abundance was quantified) and suggest the existence of two groups: proteins strongly enriched in P bodies – several of which are also required for P body assembly – and proteins that are less enriched – fewer of which affect P body assembly. From this the authors suggest that P bodies are "biochemically simpler than suggested by proteomics". This is at best a conjecture and at worst a gross oversimplification since abundance is not necessarily predictive of function – especially for enzymes.

For clarity, we have changed the conclusion to “compositionally simpler than suggested by proteomics”. This should address the reviewer’s principal concern. We note, however, that our data in Figure 4 speak against the reviewer’s assertion that abundance does not predict function. That figure shows that for nearly all proteins enriched in P bodies, the large majority of the molecular species are not in the body but rather are present in the cytoplasm. For all but the top 6 molecules, less than 20% is in the body, even when conservatively correcting for non-observable sub-diffraction P bodies (without this correction the value drops to 8%). Thus, it is hard to claim that the amount in the P body makes a large difference to total activity of that species in the cell. There are ways around this problem involving, for example a huge increase in specific activity (activity per molecule) within the P body versus the cytoplasm, or co-concentration of multiple molecules within a cascade. But these require additional assumptions, and the first order view in our opinion is that abundance does play an important role in framing the biochemical functions that arise from condensates, and since only a small number of molecules are highly concentrated, the compartments are, in fact, likely to be biochemically simpler than one might imagine from a proteomics study that does not determine concentrations and thus essentially weights all components equally.

In the second part of the study, they authors examine the domains in DCP2 required for P body localization and identify multiple domains that appear to function partially redundantly. They conclude that a collection of synergistic interactions drive P body assembly, a conclusion reached by several other studies (reviewed in Lu Na and Slavoff, 2018) including in vitro analyses (Shutz et al., 2017).

Overall, we are left with a study that will be a useful reference for future P body studies, but that does not deliver on the promise of "principles of composition and specificity". The question of specificity, in particular (why don't all RNA-binding proteins localize to P bodies?) is not examined directly. The authors only perform structure-function studies on one P body protein and never address what excludes other RNA-binding proteins from P bodies.

As stated above, we feel that our work does address the specificity of protein recruitment in a significant way through the idea of synergistic recruitment of Dcp2 to P bodies. That is, proteins that are resident often contain multiple weak recruitment elements, and by implication proteins that contain few such elements are recruited poorly if at all. RNA recruitment may follow similar principles, and we now mention this idea explicitly.

Our work also suggests to us that other RNA binding proteins do accumulate in P-bodies, but their level simply reflects the number of available binding sites in P-body mRNAs. Specifically, while the MS2 RNA binding protein does not typically accumulate in P-bodies, the addition of multiple binding sites for this protein in an mRNA resident in P-bodies can lead to the accumulation of MS2 in P-bodies. Second, we do observe other RNA binding proteins in P-bodies, they just partition less effectively into P-bodies than the core components (Figure 1).

We note that exclusion of molecules from a condensate (PC < 1) will be very hard to observe and meaningfully quantify in cells, given the small size of the structures. Further, it is more likely that most molecules will simply not be recruited (PC ~1) than be overtly excluded; we are not aware of molecules that are excluded from P bodies. In general, the physical mechanisms that do afford exclusion from condensates in vivo are not understood at all, and are well beyond the focus of the present study.

We also note that to our knowledge, this is the first time that synergy has been quantitatively demonstrated for recruitment of a protein into a condensate in vivo. The Sprangers work mentioned above is interesting but is purely in vitro. In cells, to claim synergy one would need to delete the protein and then quantify the partition coefficients for different fragments when reintroduced individually and together, as we have done. We are not aware of that level of quantitative dissection having been done elsewhere.

Additional comments:

1) The authors are not consistent in the description of the CTD. They suggest first that "N-terminal domain, HLM1, and C-terminal domain are all required for efficient partitioning and maintaining the characteristic slow dynamics of Dcp2". However, later they say that "deleting the CTD has no effect on PC of full length Dcp2". The data presented suggest that the C-terminal domain is not required (300-end) but can compensate when HLM1 is absent.

We thank the reviewer for catching this discrepancy. We have changed the first statement to “ Since the N-terminal domain and HLM1 are required for efficient partitioning and maintaining the characteristic slow dynamics of Dcp2, and the C-terminal domain can compensate when HLM1 is lacking, we conclude that elements controlling partitioning and dynamics are distributed across the protein.”

2) The intact C-terminal domain is never tested; instead the authors used a CTD with 5 of the HLM domains mutated. While the authors imply that this version of the CTD has weaker association with P bodies, no data is shown, and it is unclear from the text why this version of the CTD was used.

We had these data, and now show them in Figure 5—figure supplement 1C.

3) For the data in Figure 7, the authors use DCP2 constructs with reduced binding to Edc3 to investigate the role of RNA binding and decapping activity in P body association. However, it is not clear why RNA binding was studied using the Dcp2 300 mutant and decapping was studied in the Dcp2ΔH1 mutant.

These were essentially historical features of the work, relating to when we obtained various constructs. We do not feel that the differences here impact our conclusions.

4) A figure comparing the P body concentration/partition coefficient and exchange rates for the different DCP2 mutants (similar to Figure 2A and B) might be a more intuitive way to compare all the different DCP2 mutants used in this study.

We feel that while such a plot would help in some respects, it would detract in others, and have not included it. The principle detraction is that in Figure 2 we are specifically highlighting correlations between P body concentration and exchange rates. But the points in Figures 5 and 6 is different, and focused on understanding different aspects of Dcp2 recruitment.

5) The DCP2 structure function studies identify domains in DCP2 that promote P body localization. Whether these domains are sufficient to drive a non-P body protein into P bodies is not addressed. Without such "sufficiency" experiments, whether the results for DCP2 are generalizable for other proteins remains unclear. For example, they suggest that Xrn1 is stably recruited because of its high RNA binding affinity, however, this hypothesis does not address why other high affinity RNA binding proteins are excluded from P bodies.

This comment addresses two points. First, whether domains within Dcp2 that promote P-body localization can act in a dominant manner. Essentially, we have demonstrated this phenomenon with GFP as the foreign protein. A related question, but one beyond the scope of this manuscript is whether the P-body targeting domain of Dcp2 could override features of other RNA binding proteins that either exclude them from P-bodies or target those RNA binding proteins to other condensates. We agree this would be an interesting question and will consider it in future work.

The second point the reviewer raises is that we do not understand why other RNA binding proteins, which can bind RNA with high affinity, do not partition strongly into P-bodies. Two observations suggest to us that other RNA binding proteins do accumulate in P-bodies, but their level simply reflects the number of available binding sites in P-body mRNAs. Specifically, while the MS2 RNA binding protein does not typically accumulate in P-bodies, the addition of multiple binding sites for this protein in an mRNA resident in P-bodies can lead to the accumulation of MS2 in P-bodies. Second, we do observe other RNA binding proteins in P-bodies, they just partition less effectively into P-bodies than the core components (Figure 1).

6) At the end of the Discussion the authors introduce the concept of avidity where multiple interactions result in strong binding. It might be useful to introduce this concept earlier.

We have now introduced the idea of avidity in the Results section on Figure 5.

https://doi.org/10.7554/eLife.56525.sa2

Article and author information

Author details

  1. Wenmin Xing

    Department of Biophysics, Howard Hughes Medical Institute, UT Southwestern Medical Center, Dallas, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6445-0615
  2. Denise Muhlrad

    Department of Biochemistry, Howard Hughes Medical Institute, University of Colorado, Boulder, United States
    Contribution
    Resources, Investigation
    Competing interests
    No competing interests declared
  3. Roy Parker

    Department of Biochemistry, Howard Hughes Medical Institute, University of Colorado, Boulder, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8412-4152
  4. Michael K Rosen

    Department of Biophysics, Howard Hughes Medical Institute, UT Southwestern Medical Center, Dallas, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Project administration, Writing - review and editing
    For correspondence
    michael.rosen@utsouthwestern.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0775-7917

Funding

Howard Hughes Medical Institute

  • Wenmin Xing
  • Denise Muhlrad
  • Roy Parker
  • Michael K Rosen

Welch Foundation (I-1544)

  • Michael K Rosen

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Saumya Jain for discussions on performing FRAP; Salman Banani, Jon Ditlev and Allyson Rice for helpful discussions on image analysis; Simon Currie and Jon Ditlev for critical reading of the manuscript; and members of the Rosen lab for helpful discussions. This work was supported by the Howard Hughes Medical Institute (MKR and RP) and Welch Foundation (grant I-1544, MKR).

Senior Editor

  1. James L Manley, Columbia University, United States

Reviewing Editor

  1. Robert H Singer, Albert Einstein College of Medicine, United States

Version history

  1. Received: March 31, 2020
  2. Accepted: June 18, 2020
  3. Accepted Manuscript published: June 19, 2020 (version 1)
  4. Version of Record published: July 21, 2020 (version 2)

Copyright

© 2020, Xing et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 5,280
    Page views
  • 795
    Downloads
  • 62
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Wenmin Xing
  2. Denise Muhlrad
  3. Roy Parker
  4. Michael K Rosen
(2020)
A quantitative inventory of yeast P body proteins reveals principles of composition and specificity
eLife 9:e56525.
https://doi.org/10.7554/eLife.56525

Further reading

    1. Cell Biology
    2. Developmental Biology
    Simon Schneider, Andjela Kovacevic ... Hubert Schorle
    Research Article

    Cylicins are testis-specific proteins, which are exclusively expressed during spermiogenesis. In mice and humans, two Cylicins, the gonosomal X-linked Cylicin 1 (Cylc1/CYLC1) and the autosomal Cylicin 2 (Cylc2/CYLC2) genes, have been identified. Cylicins are cytoskeletal proteins with an overall positive charge due to lysine-rich repeats. While Cylicins have been localized in the acrosomal region of round spermatids, they resemble a major component of the calyx within the perinuclear theca at the posterior part of mature sperm nuclei. However, the role of Cylicins during spermiogenesis has not yet been investigated. Here, we applied CRISPR/Cas9-mediated gene editing in zygotes to establish Cylc1- and Cylc2-deficient mouse lines as a model to study the function of these proteins. Cylc1 deficiency resulted in male subfertility, whereas Cylc2-/-, Cylc1-/yCylc2+/-, and Cylc1-/yCylc2-/- males were infertile. Phenotypical characterization revealed that loss of Cylicins prevents proper calyx assembly during spermiogenesis. This results in decreased epididymal sperm counts, impaired shedding of excess cytoplasm, and severe structural malformations, ultimately resulting in impaired sperm motility. Furthermore, exome sequencing identified an infertile man with a hemizygous variant in CYLC1 and a heterozygous variant in CYLC2, displaying morphological abnormalities of the sperm including the absence of the acrosome. Thus, our study highlights the relevance and importance of Cylicins for spermiogenic remodeling and male fertility in human and mouse, and provides the basis for further studies on unraveling the complex molecular interactions between perinuclear theca proteins required during spermiogenesis.

    1. Cell Biology
    2. Structural Biology and Molecular Biophysics
    Bronwyn A Lucas, Benjamin A Himes, Nikolaus Grigorieff
    Research Advance

    Previously we showed that 2D template matching (2DTM) can be used to localize macromolecular complexes in images recorded by cryogenic electron microscopy (cryo-EM) with high precision, even in the presence of noise and cellular background (Lucas et al., 2021; Lucas et al., 2022). Here, we show that once localized, these particles may be averaged together to generate high-resolution 3D reconstructions. However, regions included in the template may suffer from template bias, leading to inflated resolution estimates and making the interpretation of high-resolution features unreliable. We evaluate conditions that minimize template bias while retaining the benefits of high-precision localization, and we show that molecular features not present in the template can be reconstructed at high resolution from targets found by 2DTM, extending prior work at low-resolution. Moreover, we present a quantitative metric for template bias to aid the interpretation of 3D reconstructions calculated with particles localized using high-resolution templates and fine angular sampling.