To study chaperone-chaperone and chaperone-substrate interactions, we perform synthetic interaction screens in Caenorhabditis elegans using RNA interference in combination with mild mutations or over-expression of chaperones and monitor tissue-specific protein dysfunction at the organismal level.
Correct folding and assembly of proteins and protein complexes are essential for cellular function. Cells employ quality control pathways that correct, sequester or eliminate damaged proteins to maintain a healthy proteome, thus ensuring cellular proteostasis and preventing further protein damage. Because of redundant functions within the proteostasis network, screening for detectable phenotypes using knockdown or mutations in chaperone-encoding genes in the multicellular organism Caenorhabditis elegans results in the detection of minor or no phenotypes in most cases. We have developed a targeted screening strategy to identify chaperones required for a specific function and thus bridge the gap between phenotype and function. Specifically, we monitor novel chaperone interactions using RNAi synthetic interaction screens, knocking-down chaperone expression, one chaperone at a time, in animals carrying a mutation in a chaperone-encoding gene or over-expressing a chaperone of interest. By disrupting two chaperones that individually present no gross phenotype, we can identify chaperones that aggravate or expose a specific phenotype when both perturbed. We demonstrate that this approach can identify specific sets of chaperones that function together to modulate the folding of a protein or protein complexes associated with a given phenotype.
Cells cope with protein damage by employing quality control machineries that repair, sequester or remove any damaged proteins1,2. Folding and assembly of protein complexes are supported by molecular chaperones, a diverse group of highly conserved proteins that can repair or sequester damaged proteins3,4,5,6,7. The removal of damaged proteins is mediated by the ubiquitin-proteasome system (UPS)8 or by the autophagy machinery9 in collaboration with chaperones10,11,12. Protein homeostasis (proteostasis) is, therefore, maintained by quality control networks composed of folding and degradation machineries3,13. However, understanding the interactions between the various components of the proteostasis network in vivo is a major challenge. While protein-protein interaction screens contribute important information on physical interactions and chaperone complexes14,15, understanding the organization and compensatory mechanisms within tissue-specific chaperone networks in vivo is lacking.
Genetic interactions are often used as a powerful tool to examine relationship between pairs of genes that are involved in common or compensatory biological pathways16,17,18. Such relationships can be measured by combining pairs of mutations and quantifying the impact of a mutation in one gene on the phenotypic severity caused by a mutation in the second gene16. While most such combinations do not show any effect in terms of phenotype, some genetic interactions can either aggravate or alleviate the severity of the measured phenotype. Aggravating mutations are observed when the phenotype of the double deletion mutant is more severe than the expected phenotype seen upon combining the single deletion mutants, implying that the two genes function in parallel pathways, together affecting a given function. In contrast, alleviating mutations are observed when the phenotype of the double deletion mutant is less severe than the phenotype seen with the single deletion mutants, implying that the two genes act together as a complex or participate in the same pathway16,18. Accordingly, diverse phenotypes that can be quantified, including broad phenotypes, such as lethality, growth rates and brood size, as well as specific phenotypes, such as transcriptional reporters, have been used to identify genetic interactions. For example, Jonikas et al. relied on an ER stress reporter to examine interactions of the Saccharomyces cerevisiae ER unfolded protein response proteostasis network using pairwise gene deletion analyses19.
Genetic interaction screens involve systematically crossing pairwise deletion mutations to generate a comprehensive set of double mutants20. However, in animal models, and specifically in C. elegans, this large-scale approach is not feasible. Instead, mutant strains can be tested for their genetic interaction patterns by down-regulating gene expression using RNA interference (RNAi)21. C. elegans is a powerful system for screens based on RNAi22,23. In C. elegans, double-stranded RNA (dsRNA) delivery is achieved by bacterial feeding, leading to the spread of dsRNA molecules to numerous tissues. In this manner, the introduced dsRNA molecules impact the animal via a rapid and simple procedure21. A genetic interaction screen using RNAi can, therefore, reveal the impact of down-regulating a set of genes or most C. elegans coding genes using RNAi libraries24. In such a screen, hits that impact the behavior of the mutant of interest but not the wild type strain are potential modifiers of the phenotype being monitored25. Here, we apply a combination of mutations and RNAi screening to systematically map tissue-specific chaperone interactions in C. elegans.
1. Preparation of nematode growth media plates for RNAi
2. Growing RNAi bacteria and seeding the plates
3. Non-stressful synchronization of embryos
4. Common phenotypic assays
5. Validation of protein knockdown
Using temperature-sensitive mutations in UNC-45 to screen for aggravating or alleviating interactions under permissive or restrictive conditions, respectively
Muscle assembly and maintenance offer an effective system to study tissue-specific chaperone interactions. The functional unit of contractile muscles, the sarcomere, presents a crystalline-like arrangement of structural and regulatory proteins. The stability of the motor protein myosin and its incorporation into the thick filaments of contractile muscle sarcomeres depends on cooperation of chaperones and UPS components30. An example of one such chaperone is the conserved and specialized myosin chaperone UNC-45 that is mainly expressed in body wall muscle31,32,33,34,35. Mutations in UNC-45 have been shown to induce myosin disorganization and severe motility defects in C. elegans31,36. UNC-45 tandem modules assemble into a multi-site docking platform37 that enforces collaboration between UNC-45, HSP-90 and HSP-1 and likely other chaperones and co-chaperones in myosin filament assembly25,36,37,38. To confirm known UNC-45 interactions and identify novel genetic interactions in muscle proteostasis, we established a strategy using C. elegans temperature-sensitive unc-45 mutations as a sensitized genetic background for tissue-specific chaperone interaction screening19,25.
Single amino acid substitutions in C. elegans UNC-45 (L822F and E781K, corresponding to the e286 and m94 alleles, respectively; unc-45(ts)) are responsible for temperature-dependent motility defects and myosin disorganization phenotypes when affected animals are grown under restrictive conditions (>22 °C). In contrast, these unc-45(ts) mutants show no movement or myosin organization defects at the permissive temperature (15 °C)31. In the proposed approach, age-synchronized unc-45(e286) animals at the first larval stage (L1) were depleted of different molecular chaperones by RNAi (97 genes) and then monitored for motility defects under permissive conditions (15 °C) (Figure 1). We confirmed the known interaction of UNC-45 and HSP-90 proteins in our genetic interactions screen and identified three hsp-90 co-chaperones, sti-1, ahsa-1, and daf-41, as specifically causing a synthetic movement defect in unc-45(ts) mutant animals but not in wild-type animals25. We went on to examine whether sti-1-, ahsa-1- or daf-41-associated synthetic phenotypes were caused by myosin disorganization by monitoring the subcellular arrangement of myosin heavy chain A (MYO-3), using established immuno-staining techniques39. Whereas treatment of wild type animals with sti-1, ahsa-1 or daf-41 RNAi did not affect myofilament organization, depletion of these genes in unc-45(e286) mutant animals resulted in complete disruption of sarcomeric structures and MYO-3 mislocalization, even under permissive conditions (15 °C). This effect was comparable to what was seen with unc-45(e286) single mutants grown at the restrictive temperature (25 °C)25. These results were confirmed using another unc-45(ts)-allele, namely unc-45(m94) mutant animals25.
To identify chaperones that destabilize UNC-45, we next screened for chaperones that improved the motility of unc-45(286) animals at 25 °C, relying on gene knockdown by RNAi. While unc-45(e286) mutant animals displayed severe movement defects at 25 °C, the motility of animals treated with RNAi against genes encoding four of the 97 chaperones screened was significantly improved. Here too, the results were confirmed using unc-45(m94) mutant animals. Thus, the use of temperature-sensitive mutations allows for establishing both aggravating and alleviating screens, depending on the temperature at which the screen is conducted.
Using tissue-specific over-expression of a chaperone to screen for aggravating interactions
We next utilized tissue-specific over-expression of a single chaperone as a mild perturbation of the muscle chaperone network for screening purposes. Specifically, we utilized animals over-expressing wild type dnj-24, encoding the C. elegans homolog of the Hsp40 protein DNAJB6 in the C. elegans body wall muscle (DNJ-24M). As above, L1 DNJ-24M animals were treated with RNAi for different chaperones and monitored for motility defects (20 °C) (Figure 1). While DNJ-24M animals showed no notable motility defects, three genes (of the 48 chaperone-encoding genes examined; 6%), namely hsp-1, rme-8, and dnj-8, specifically affected the motility of DNJ-24M-expressing animals but not wild type animals. It is of note that testing the specificity of the hits using animals over-expressing a different chaperone, namely HSP-90, in muscle (HSP90M), showed no effect on HSP90M motility upon hsp-1, rme-8, or dnj-8 RNAi treatment40. Taken together, the screening platform, employing mild perturbation to the chaperone network, such as the expression of metastable mutant proteins or tissue-specific over-expression, resulted in a highly specific hit rate (normally ~5%).
Using tissue-specific RNAi to screen for tissue-specific genetic interactions
Tissue-specific RNAi-sensitive strains allow for tissue-specific knockdown of genes while still using bacterial feeding for dsRNA delivery. These strains are mutant for the RDE-1 argonaute protein, a major component in the RNAi pathway required for effective gene silencing21. However, expressing wild type RDE-1 under the control of a tissue-specific promoter led to effective tissue-specific gene knockdown41,42. This tool thus allows for genetic interaction screens without using tissue-specific chaperones, such as UNC-45 or DNAJ-24M. For example, hsp-6 (mortalin) knockdown in wild type animals during development resulted in a strong developmental arrest (96±1% of the RNAi-treated animals). At the same time, hsp-6 knockdown in a strain expressing wild type RDE-1 in muscle did not cause developmental arrest, whereas expressing wild type RDE-1 in intestinal cells resulted in a strong developmental arrest phenotype (90±3%; Figure 2). Thus, HSP-6 function in intestinal cells is required for normal development. A mutation in hsp-6(mg585) that causes a mild growth delay can, therefore, be used to screen for aggravating or alleviating chaperone interactions by crossing that mutated gene into an intestinal-specific RNAi strain and screening the chaperone RNAi library.
Monitoring age-dependent changes in the folding environment using genetic interactions
Animals show an age-dependent decline in motility that is associated with sarcomeric disorganization43,44,45. Changes in protein folding capacity coincide with altered regulation and composition of the cellular proteostasis machinery28, including altered levels of UNC-45, CHN-1, and UFD-2, proteins of the muscle quality control machinery46. In agreement, myosin folding and degradation are affected by alterations in proteostatic capacity at the transition to adulthood47. We, therefore, asked whether such changes could impact chaperone interactions. For example, we monitored the impact of sti-1, ahsa-1 and daf-41 knockdown on motility over time. We found that although sti-1-, ahsa-1– and daf-41-RNAi treated animals showed reduced motility during larval development, motility strongly declined in unc-45(ts) adult worms. Moreover, MYO-3 organization in unc-45(ts) mutant animals treated with sti-1, ahsa-1 or daf-41 RNAi was similar to that of wild type worms at the fourth larval stage (L4), although the mutants exhibited disrupted sarcomeres after the animals reached adulthood (Figure 3). In contrast, both unc-45(ts) mutant animals treated with an empty vector control and wild type animals remained unaffected (Figure 3). Thus, proteostasis dynamics as a function of age or environmental conditions27,45,46,48,49 could critically impact chaperone interactions.
Figure 1: Setup of RNAi synthetic interaction screens using C. elegans carrying a mutation in a gene encoding a chaperone of interest. (A) Schematic representation of the basic setup of targeted chaperone interaction screens. (B) Hit validation requires confirmation of the genetic interaction using another chaperone mutant, as well as validating the specify of RNAi knockdown. (C-D) Simple readouts, such as motility defects, can be quantified to determine aggravating or alleviating interactions, using paralysis or thrashing assays, respectively. Please click here to view a larger version of this figure.
Figure 2: Tissue-specific RNAi of the mitochondrial chaperone hsp-6 can be used to examine genetic interactions in one tissue. Wild type intestine- and muscle-specific RNAi strains were treated with hsp-6 RNAi and (A) developmental delay was scored or (B) images were taken on the first day of adulthood. Data are mean ± SEM, N=6. Scale bar is 1 mm. Please click here to view a larger version of this figure.
Figure 3: Age-dependent effects of RNAi. (A) Motility with age. Wild type or unc-45(e286) embryos were placed on sti-1, ahsa-1 or daf-41 RNAi-seeded plates at 15 °C and scored for motility using a thrashing assay at each developmental stage, L1-L4, young adult and day 1 of adulthood. Data are mean ± SEM, N=15. (B) Confocal images of body wall muscle. Animals were treated as in A and fixed at the L4, young adult and day 1 of adulthood stages and immuno-stained with anti-MYO-3 antibodies. Scale bar is 10 µm. Please click here to view a larger version of this figure.
Solution | Preparation instructions | Storage |
1 M CaCl2 (1 L) | Add 147.01 g CaCl2·2H2O | Store at RT |
Add dH2O to 1 L | ||
Autoclave or filter (0.22 µm) | ||
1 M KH2PO4, pH 6.0 (1 L) | Add 136.09 g KH2PO4 | Store at RT |
Add 800 mL dH2O | ||
Mix using magnetic stirrer until dissolved | ||
Titrate pH using KOH | ||
Add dH2O to 1 L | ||
Autoclave or filter (0.22 µm) | ||
1 M MgSO4 (1 L) | Add 248.58 g MgSO4·7H2O | Store at RT |
Add dH2O to 1 L | ||
Autoclave or filter (0.22 µm) | ||
Cholesterol solution (50 mL) | Add 250 mg cholesterol to a 50 mL Falcon tube | Store at -20 °C |
Completely dissolve in 40 mL ethanol | ||
Add ethanol to 50 mL | ||
1 M IPTG (50 mL) | Add 11.9 g IPTG (isopropyl-β-D-thiogalactopyranoside) to a 50 mL Falcon tube | Store in the dark at -20 °C |
Completely dissolve in 40 mL dH2O | ||
Add dH2O to 50 mL | ||
Filter (0.22 µm), aliquot 1mL tubes | ||
Ampicillin stock (50 mL) | Add 5 g ampicillin to a 50 mL Falcon tube | Store at -20 °C |
Completely dissolve in 40 mL dH2O | ||
Add dH2O to 50 mL | ||
Filter (0.22 µm), distribute 1 mL aliquot into Eppendorf tubes | ||
Tetracycline stock (50 mL) | Add 250 mg tetracycline to a 50 mL Falcon tube | Store at -20 °C |
Completely dissolve in 40 mL ethanol | ||
Add ethanol to 50 mL | ||
Aliquot 1 mL tubes | ||
M9 buffer (1 L) | Add 5.8 g Na2HPO4·7H2O | Store at RT |
Add 3 g KH2PO4 | ||
Add 5 g NaCl | ||
Add 0.25 g MgSO4·7H2O | ||
Add dH2O to 1 L | ||
Filter (0.22 µm) | ||
1X PBS-T (pH 7.4) ( 1 L) | Add 8 g of NaCl | Store at RT |
Add 200 mg of KCl | ||
Add 1.44 g Na2HPO4·7H2O | ||
Add 240 mg KH2PO4 | ||
Completely dissolve in 800 mL dH2O | ||
Titrate pH using KOH tp pH 7.4 | ||
Add 500 µL Tween-20 | ||
Add dH2O to 1 L | ||
5x sample buffer | Add 6.8 mL dH2O | Store at -20 °C |
Add 2 mL 0.5M Tris pH 6.8 | ||
Add 3.2 mL Glycerol | ||
Add 1.6 mL 20% SDS | ||
Add 0.8 mL ß-mercaptoethanol | ||
1% bromophenol blue |
Table 1: Solution Recipes
An integrated picture of the proteostasis network reflecting how it is organized and functions in different metazoan cells and tissues remains lacking. To address this shortcoming, specific information on the interactions of various components of this network, such as molecular chaperones, in specific tissues during the course of development and aging is required. Here, we showed how the use of tissue-specific perturbations enabled us to examine the chaperone network in a given tissue. To explore tissue-specific chaperone genetic interactions, three different approaches were considered. In the first approach, UNC-45, a chaperone that is highly expressed in muscle cells, was used to screen for chaperone interactions via feeding-RNAi25. While the use of a specialized chaperone allows for discerning tissue-specificity, it can only report on that highly focused sub-network to which it contributes. Note that most C. elegans neurons are resistant to feeding-based RNAi delivery50 and thus using this approach to identify genetic interactions in neuronal cells requires crossing the mutant chaperone examined with an RNAi-enhanced strain51. In a second approach, a tissue-specific promoter was used to drive over-expression of a chaperone in muscle and thus specifically affect the muscle folding environment40. However, over-expressing a single chaperone could be generally beneficial to the folding environment and thus mask disruption caused by the two chaperones together. The third approach relied on tissue-specific RNAi knockdown to examine chaperone interactions in a given a tissue41. One advantage of this approach is that it allows for targeting neuronal cells that are resistant to RNAi delivery via feeding42. Still, this approach requires the use of a mild mutant (although not specialized), as well as that this mutant be crossed into a rde-1 null mutant carrying a tissue-specific rde-1 rescue gene. Importantly, these approaches can be combined so as to potentially modulate chaperone function in a single tissue or even a single cell.
The quality control machinery can impact the function of many gene products by perturbing proteostasis. This is a major challenge when using genetic interactions to explore the quality control network18. For example, chronic expression of aggregation-prone proteins or proteostasis collapse in aging resulted in phenotypic aggravation of many unrelated metastable proteins in C. elegans and yeast45,52,53. Moreover, clathrin-mediated endocytosis in mammalian cells was inhibited upon functional sequestration of Hsc70 to protein aggregates. Yet, it was shown that endocytosis could be rescued by Hsc70 over-expression, while aggregation could not54,55. Likewise, a genetic screen in Drosophila designed to uncover regulators of the heat shock response identified a missense mutation in flight muscle actin that constitutively activated the heat shock response56. Taken together, perturbation of the proteostasis network can expose metastable proteins or induce stress that can impact result specificity. Nonetheless, analyzing genetic interactions can yield highly specific and functional insight. For example, epistatic analyses of yeast genes required for folding in the endoplasmic reticulum identified specific genetic interactions between molecular chaperones that were subsequently validated19. Likewise, an aggravating screen for chaperones that enhance the toxicity of two aggregation-prone models (as measured by motility) identified a specific subset of 18 chaperones, orthologs of which impacted Huntingtin aggregation in human cells26. Here, we showed that various perturbations of the proteostasis network can uncover specific and functional chaperone interactions.
The main advantage of using RNAi feeding-based genetic interaction screens is the relative simplicity of the method. Even employing a general behavioral output, such as motility, can reveal novel genetic interactions (Figure 3). However, variability and partial effects of expression knockdown can limit the robustness and specificity of the results57. Moreover, genetic interactions are not indicative of physical interactions and the relationship between two genes could thus be indirect. Exploring the nature of the interactions and discarding non-specific interactions can be time consuming57,58,59. This is a concern that needs to be addressed in the screen setup and validation. For example, using null alleles in genetic screening allows for determining whether these genes function in the same or distinct pathways in a given biological process. However, using partial loss of gene function, hypomorphic alleles, such as temperature-sensitive alleles, or RNAi-dependent down-regulation of expression, results in residual activities that can yield aggravating or alleviating phenotypes, regardless of whether the genes act in the same or in parallel pathways59. Thus, the nature of any interaction requires further analysis. However, using hypomorphic alleles and RNAi could identify a broad range of interactors, including genes in the same protein complex or pathway or in a redundant pathway59. While this larger scope of possible interactions can yield more hits in a genetic screen, it can also lead to non-specific interactions, such as, for example, the non-specific exposure of temperature-sensitive alleles in proteostasis collapse45,53.
Hit validation and non-specific interactions can be examined in several ways. The number of hits should be low. For example, down-regulation of chaperone expression in an unc-45 mutant background resulted in a small percentage of hits (4%), with most chaperone gene down-regulation not showing any effect on motility. A similar rate was observed when DNJ-24M was over-expressed (6%). As noted above, a screen for chaperone interactions with aggregation-prone proteins identified 18 chaperones of 219 screened (8%).
Several mutant alleles, over-expression lines or disease models should be used. The use of different mutant alleles that have different impacts on chaperone function, as well as different strains with different genetic backgrounds, can support the specificity of any screen hits. Alternatively, using mutation or over-expression of a different chaperone that does not lead to similar perturbation can serve as a negative control. For example, both unc-45(e286) and unc-45(m94) showed aggravating behavior when sti-1, ahsa-1 or daf-41 were down-regulated. Moreover, a similar interaction was observed when animals carrying the ahsa-1(ok3501) deletion mutant were treated with hsp-90 RNAi25.
The identity of the chaperone hits and their known interactions should be examined. For example, chaperones identified in a unc-45 aggravating screen are a highly specific set of chaperones required for the HSP-90 ATPase cycle, including those encoding a client recruiter (STI-1), a remodeling co-chaperone (AHSA-1), and a client maturation co-chaperone (DAF-41). In fact, this set of co-chaperones forms a complete HSP-90 folding cycle60. Likewise, a DNJ-24M screen identified HSP-1, the main chaperone partner of Hsp40s40.
Changes in interactions over the lifespan of the animal should also be examined. For example, chaperones identified in unc-45 aggravating screen strongly impacted motility and myosin organization in adulthood but had a milder effect during development. This could be due to changes in the proteostasis network in adulthood28 or to changes in myosin folding requirements between myo-fiber folding and maintenance.
Use complementary biochemical approaches to directly examine interactions between the proteins, as well as their localization in the cell. For example, the HSP-90 co-chaperones, STI-1, AHSA-1 and DAF-41, are localized to the sarcomere where they interact with myosin25.
C. elegans is a well-established metazoan model for monitoring quality control. It is often used to monitor cellular and organismal proteostasis using a variable toolkit of cell biology, biochemical and genetic approaches. Here, we employed genetic screening approaches and available tools57,58,59, such as a mutant bank, available RNAi libraries22,23 and tissue-specific RNAi strains41,42, to monitor chaperone interactions in a living animal during development and aging. The use of simple behavioral assays, such as motility, simplify the screen of many possible gene pairs to explore novel genetic interactions. This can then serve as a platform to further explore chaperone localization and physical interactions using biochemical tools to mechanistically study their potential interactions in vivo and in vitro. The protocol described here has been successfully used to identify novel chaperone interactions in C. elegans body wall muscle25,40.
The authors have nothing to disclose.
We thank the Caenorhabditis Genetics Center, funded by the NIH National Center for Research Resources (NCRR), for some of the nematode strains. Monoclonal antibodies developed by H.F. Epstein were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the Department of Biology, University of Iowa. This research was supported by a grant from the Israel Science Foundation (grant No. 278/18) and by a grant from the Israel Ministry of Science & Technology, and the Ministry of Foreign Affairs and International Cooperation, General Directorate for Country Promotion, Italian Republic (grant No. 3-14337). We thank members of the Ben-Zvi laboratory for help in preparing this manuscript.
12-well-plates | SPL | BA3D16B | |
40 mm plates | Greiner Bio-one | 627160 | |
60 mm plates | Greiner Bio-one | 628102 | |
6-well plates | Thermo Scientific | 140675 | |
96 well 2 mL 128.0/85mm | Greiner Bio-one | 780278 | |
Agar | Formedium | AGA03 | |
Ampicillin | Formedium | 69-52-3 | |
bromophenol blue | Sigma | BO126-25G | |
CaCl2 | Merck | 1.02382.0500 | |
Camera | Qimaging | q30548 | |
Cholesterol | Amresco | 0433-250G | |
Confocal | Leica | DM5500 | |
Filter (0.22 µm) | Sigma | SCGPUO2RE | |
Fluorescent stereomicroscope | Leica | MZ165FC | |
Glycerol | Frutarom | 2355519000024 | |
IPTG | Formedium | 367-93-1 | |
KCl | Merck | 104936 | |
KH2PO4 | Merck | 1.04873.1000 | |
KOH | Bio-Lab | 001649029100 | |
MgSO4 | Fisher | 22189-08-8 | Gift from the Morimoto laboratory |
Myosin MHC A (MYO-3) antibody | Hybridoma Bank | 5-6 | |
Na2HPO4·7H2O | Sigma | s-0751 | |
NaCl | Bio-Lab | 001903029100 | |
Peptone | Merck | 61930705001730 | |
Plate pouring pump | Integra | does it p920 | |
RNAi Chaperone library | NA | NA | |
SDS | VWR Life Science | 0837-500 | |
ß-mercaptoethanol | Bio world | 41300000-1 | |
stereomicroscope | Leica | MZ6 | |
Tetracycline | Duchefa Biochemie | 64-75-5 | |
Tris | Bio-Lab | 002009239100 | |
Tween-20 | Fisher | BP337-500 |