COP9 signalosome component CSN-5 stabilizes PUF proteins FBF-1 and FBF-2 in Caenorhabditis elegans germline stem and progenitor cells

Abstract

RNA-binding proteins FBF-1 and FBF-2 (FBFs) are required for germline stem cell maintenance and the sperm/oocyte switch in Caenorhabditis elegans, although the mechanisms controlling FBF protein levels remain unknown. We identified an interaction between both FBFs and CSN-5), a component of the constitutive photomorphogenesis 9 (COP9) signalosome best known for its role in regulating protein degradation. Here, we find that the Mpr1/Pad1 N-terminal metalloprotease domain of CSN-5 interacts with the Pumilio and FBF RNA-binding domain of FBFs and the interaction is conserved for human homologs CSN5 and PUM1. The interaction between FBF-2 and CSN-5 can be detected in vivo by proximity ligation. csn-5 mutation results in the destabilization of FBF proteins, which may explain previously observed decrease in the numbers of germline stem and progenitor cells, and disruption of oogenesis. The loss of csn-5 does not decrease the levels of a related PUF protein PUF-3, and csn-5(lf) phenotype is not enhanced by fbf-1/2 knockdown, suggesting that the effect is specific to FBFs. The effect of csn-5 on oogenesis is largely independent of the COP9 signalosome and is cell autonomous. Surprisingly, the regulation of FBF protein levels involves a combination of COP9-dependent and COP9-independent mechanisms differentially affecting FBF-1 and FBF-2. This work supports a previously unappreciated role for CSN-5 in the stabilization of germline stem cell regulatory proteins FBF-1 and FBF-2.

Introduction

Stem cells are a population of unspecialized cells with the capability to both self-renew and preserve the stem cell pool or to differentiate and support tissue maintenance (Morrison and Kimble 2006). Stem cell function depends on faithful expression of essential developmental regulators (Morrison et al. 1997). A number of mechanisms controlling the expression of stem cell machinery at the levels of epigenetics, transcription, and RNA stability have been widely studied (for example, Lee et al. 2016; Chen et al. 2020; Samuels et al. 2020; Demarco et al. 2022; Vogiatzoglou et al. 2022). As the stem cell proteome is exquisitely controlled in both self-renewing state and through developmental transitions (Vilchez et al. 2014; Llamas et al. 2020), it is critical to understand the mechanisms regulating protein stability of key self-renewal factors.

We use the Caenorhabditis elegans germline to study stem cell regulators FBF-1 and FBF-2 (FBFs). The C. elegans gonad consists of 2 U-shaped arms, each with a population of germline stem and progenitor cells (SPCs) at the distal end (Fig. 1a; Kimble and White 1981). SPCs are maintained through the GLP-1/Notch signaling from the somatic niche cell (Austin and Kimble 1987; Berry et al. 1997; Kimble and Crittenden 2007). As the cells progress away from the niche, they begin to differentiate and enter meiosis before generating gametes at the proximal end (Hubbard 2007; Pazdernik and Schedl 2013; Voronina and Greenstein 2016). The hermaphrodite germlines produce sperm during larval development and switch to oogenesis in the young adults (Pazdernik and Schedl 2013).

Fig. 1.

CSN-5 interacts with FBFs. a) Schematic of adult C. elegans hermaphrodite gonad. Somatic distal tip cell is shown on the left, germline SPC nuclei are shown as hollow circles, crescent shapes represent cells in transition zone, dark circles represent cells in meiotic pachytene, and larger dark circles represent meiotic cells differentiating into oocytes (diplotene and diakinesis). Sperm is shown as smaller black dots. FBF protein expression is enriched in SPCs. COP9 subunit CSN-5 is expressed throughout oogenic germline (Smith et al. 2002). b) Interaction between CSN-5 and FBF-1/-2 is detected in a yeast 2-hybrid assay by growth on selective media (synthetically defined medium, SD) lacking histidine (H) and adenine (A) and supplemented with 3-AT after transformation with FBFs fused to Gal4 DNA-binding domain as baits and CSN-5 fused to Gal4 activation domain as prey. Empty prey vector is used as control and does not show growth on the selective media. Media lacking threonine (T) and leucine (L) are used to select for the presence of bait and prey plasmids. This experiment was performed in 2 replicates. c) GST pulldown assay to test direct interaction between GST::CSN-5 and both His₆::FBF-1 and His₆::FBF-2. Input is clarified FBF lysate that was added to GST pulldown. FBFs are detected by Western blot with anti-His. Total eluate protein (representative image from FBF-2 pulldown) was detected by stain-free chemistry (Taylor and Posch 2014) to confirm GST fusion retention on GSH beads; the position of full-length GST::CSN-5 is marked with a black arrowhead, and the lower band marked with a black dot is a degradation product of recombinant GST::CSN-5.

Paralogous FBF-1 and FBF-2 (FBFs) are RNA-binding proteins belonging to the highly conserved PUF (Pumilio and FBF) family (Wickens et al. 2002). FBFs are enriched in SPCs (Fig. 1a) and act by binding to and repressing their target mRNAs (Zhang et al. 1997; Crittenden et al. 2002; Prasad et al. 2016). FBFs are 89% identical at the protein sequence level and function in a largely redundant manner (Zhang et al. 1997; Crittenden et al. 2002). FBFs promote germline stem cell maintenance, and fbf-1  fbf-2 double mutant germline stem cells prematurely enter meiosis during late larval development (Crittenden et al. 2002; Wickens et al. 2002). Consistent with this function, many FBF targets encode developmental regulators, which are silenced in stem cells and then activated upon differentiation, such as the differentiation-promoting protein GLD-1 (Crittenden et al. 2002; Suh et al. 2009). FBFs also promote self-renewal of germline stem cells by repressing cki-2 (Kalchhauser et al. 2011), a Cyclin E/Cdk2 inhibitor that regulates the decision to enter/exit the cell cycle (Buck et al. 2009). Additionally, FBF-1 and FBF-2 regulate germline sex determination by facilitating the switch from spermatogenesis to oogenesis (Zhang et al. 1997) as fbf-1  fbf-2 double mutant gonads fail to undergo oogenesis, resulting in a masculinization of germline (Mog) phenotype where only sperm is produced (Crittenden et al. 2002). Despite these well-characterized functions of FBFs, the posttranslational mechanisms regulating the levels of FBF proteins in stem cells remain unknown.

One prominent cellular mechanism impacting protein stability and steady-state levels involves the constitutive photomorphogenesis 9 (COP9) signalosome, a highly conserved enzymatic complex composed of 8 subunits, denoted CSN1–CSN8 (Qin et al. 2020). Although initially discovered for its role in the regulation of transcription (Wei and Deng 1992), the COP9 signalosome's best-documented function is as a regulator of protein degradation (Wei et al. 1994; Chamovitz et al. 1996; Claret et al. 1996; Chamovitz 2009). COP9 impacts protein degradation by removing a ubiquitin-like protein, NEDD8, from cullin subunits of E3 ubiquitin ligases (Cope et al. 2002; Lingaraju et al. 2014). The balance between neddylation and deneddylation is essential to maintain the activity of cullin-based ubiquitin ligases and to facilitate the degradation of their substrates (Lyapina et al. 2001; Doronkin et al. 2003; Pintard et al. 2003; Wu et al. 2005). The overall architecture of COP9 resembles the 19S lid of the 26S proteasome, which is responsible for the majority of intracellular protein degradation (Enchev et al. 2012; Lingaraju et al. 2014). CSN subunits 1–4, 7, and 8 contain PCI (Proteasome, COP9 signalosome, translation Initiation factor) domains similar to the 6 subunits of the proteasome lid (Dessau et al. 2008; Hofmann and Bucher 1998). COP9 signalosome subunit 5 (CSN5) is the only catalytically competent component and contains a JAB1/MPN (Mpr1/Pad1 N-terminal) metalloprotease domain, which is responsible for the complex's deneddylating activity (Zhang et al. 2012; Echalier et al. 2013). The signalosome is inactive without the incorporation of CSN5, and likewise, the current literature suggests CSN5 is inactive on its own and unable to deneddylate cullins in isolation (Cope et al. 2002; Sharon et al. 2009; Echalier et al. 2013; Lingaraju et al. 2014). Furthermore, CSN5 must dimerize with the other MPN domain-containing COP9 subunit CSN6, to be incorporated into the COP9 holoenzyme and become activated (Birol et al. 2014; Lingaraju et al. 2014). Apart from COP9 subunits, CSN5 has been reported to interact with other cellular proteins (Tomoda et al. 1999; Smith et al. 2002; Yoshida et al. 2013; Shackleford and Claret 2010) and to promote the accumulation of several of its partners (Bae et al. 2002; Bemis et al. 2004; Orsborn et al. 2007; Liu et al. 2009; Lim et al. 2016). Furthermore, CSN5 interaction partners may be stabilized in a manner dependent on the entire COP9 complex (Wu et al. 2009) or independently of the COP9 holoenzyme (Bemis et al. 2004).

In a yeast 2-hybrid screen for FBF-binding partners, we identified C. elegans  CSN-5 as an interactor of FBF-2. This work investigates CSN-5's contribution to FBF function in the germline. We find that CSN-5 stabilizes both FBF-1 and FBF-2 in germline SPCs, correlating with effective SPC maintenance and promoting oogenesis. Our findings identify a previously unappreciated role for CSN-5, and we speculate its contribution to stem cell regulatory protein accumulation is relevant to a wide range of organisms.

Materials and methods

Nematode strains and culture

All C. elegans strains (Supplementary Table 1) used in this study were cultured on New Nematode Growth Medium (NNGM) plates seeded with Escherichia coli  OP50 as per standard protocols (Brenner 1974). csn(lf) mutants used in this study were obtained from the Caenorhabditis Genetics Center and outcrossed >8 times to wild type before analysis. The strains were maintained at 20°C, unless specified otherwise.

csn-2(ok1288) allele

In order to determine the molecular nature of csn-2(ok1288) allele, we conducted PCR analysis of genomic DNA as well as cDNA. Total RNA isolated from csn-2(ok1288) hermaphrodites was reverse-transcribed using SuperScript IV reverse transcriptase (Invitrogen) and amplified with the primers AGACCCAGGAAAAGTTCGGT and GAGACCATCATCCAAAATTGCGT. Our analysis revealed that the 1,680-nt genomic deletion removes most of intron 3, exon 4, and most of intron 4. The csn-2(ok1288) transcript is spliced from the end of exon 3 to the start of exon 5 resulting in a frameshift at amino acid 238 and a premature stop codon (removing 50% of ORF; Supplementary Fig. 1). The full-length CSN-2 contains a PCI domain consisting of helical repeats 1–8 and a winged-helix (WH) subdomain, which is followed by a C-terminal alpha helix. CSN-2 is incorporated into the COP9 signalosome through the association of the WH subdomain with the WH subdomains of other PCI proteins and through the integration of the C-terminal helix into a bundle containing helices of all COP9 subunits (Lingaraju et al. 2014). ok1288 deletion removes 3 C-terminal PCI repeats, the WH domain, and the C-terminal helix suggesting that the remaining fragment would not incorporate into COP9 signalosome.

Transgenic animals

All transgene constructs were generated by Gateway cloning (Thermo Fisher Scientific). The 2 3xflag::csn-5 transgene constructs contain either the csn-5 promoter (1,110-bp upstream of the CDS) or gld-1 promoter (1,165-bp upstream of the CDS; Ellenbecker et al. 2019) and both contain csn-5 genomic coding and 3′UTR (733-bp downstream of the CDS) sequences in pCFJ150 (Frøkjaer-Jensen et al. 2008). The 3xflag::csn-5(D152N) transgene construct contains the gld-1 promoter, csn-5(D152N) mutant genomic coding sequence, and csn-5 3′UTR sequence in pCFJ150. Mutagenesis of csn-5(D152N) was performed using a Q5 site-directed mutagenesis kit per the manufacturer’s instructions (New England Biolabs). For transgenic lines used in proximity ligation assay (PLA), 3xflag::csn-5 was crossed with preexisting patcGFP::fbf or patcGFP lines (Wang et al. 2020). For all csn-5 transgenes, a single-copy insertion of the transgene was generated by homologous recombination into the universal Mos1 insertion site on LG II after Cas9-induced double-stranded break (Dickinson et al. 2013; Wang et al. 2016). Transgene insertion was confirmed by PCR.

Yeast 2-hybrid library screening and directed assays

For the yeast 2-hybrid screen, fbf-2 cDNA was cloned into the pGBTK vector (Clontech) and transformed into PJ68-4a yeast strain (MATa trp1-901 leu2-3,112 ura3-52 his3-200 gal4Δ gal80Δ LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ; James et al. 1996). The yeast 2-hybrid library was generated from the poly(A)+ mRNA isolated from mixed-stage N2 nematodes using the Matchmaker Kit (Clontech) as per the manufacturer's protocol and transformed into Y1HGold strain (MATα, ura3-52, his3-200, ade2-101, trp1-901, leu2-3, 112, gal4Δ, gal80Δ, met–, and MEL1). The bait and prey library strains were mated and cultured on selective plates SD/-Trp, Leu, His, Ade with 1 mM 3-AT; the total number of screened diploids was estimated at 2.32 × 10⁸. Out of 271 positive colonies, 99 (37%) were identified as csn-5 fragments by PCR and sequencing of prey plasmid inserts. For directed interaction assays, FBF-1 and FBF-2 in a bait vector pGBKT7 and CSN-5 in a prey vector pGADT7 were cotransformed in PJ69-4a. Empty prey vector was used as the control. Expression of c-myc-tagged FBF-1 or FBF-2 and HA-tagged CSN-5 proteins in yeast colonies was confirmed by Western blot (Supplementary Fig. 2); see Supplementary Table 2 for antibody information. Serially diluted yeast cultures (at OD600 0.2) expressing fbf bait and csn-5 prey (or empty prey) were spotted on control or the interaction selection plate SD/-Trp, -Leu, -His, -Ade (with 1 mM 3-AT to inhibit leaky expression of histidine reporter) and incubated at 30°C for 4 days.

GST pulldown assay

Expression constructs containing His₆::FBF-1 and His₆::FBF-2 have been described before (Wang et al. 2016). Truncated FBF expression constructs were generated by PCR from full-length FBF-1 or FBF-2 and inserted into pETDuet-1 plasmid (EMD Millipore). Full-length C. elegans  CSN-5 (amino acids 1–368) and CSN-6^MPN (amino acids 1–197) were amplified from N2 Bristol cDNA and cloned into pGEX-KG plasmid. Mutant CSN-5(D152N) was generated from full-length CSN-5 using a Q5 site-directed mutagenesis kit per the manufacturer’s instructions (New England Biolabs). Truncated CSN-5 constructs were made by PCR from full-length CSN-5 and inserted into pGEX-KG plasmids. Human CSN5^MPN (amino acids 1–257) was amplified from HEK293 cDNA and cloned into pGEX-KG plasmid. Human PUM1^RBD (isoform 1, amino acids 828–1176) and PUM2^RBD (isoform 3, amino acids 706–1055) were amplified from HEK293 cDNA and cloned into pETDuet-1 plasmid. All constructs were sequenced and transformed into E. coli strain BL21(DE3) for expression. Expression of His-tagged FBF protein constructs was induced with 0.1 mM IPTG at 15°C for 20–24 h. Expression of GST alone or GST-tagged CSN protein constructs was induced with 0.1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 37°C for 4.5 h. Pellets of induced bacteria were collected and resuspended in lysis buffer (20 mM Tris pH 7.5, 250 mM NaCl, 10% glycerol, 1 mM MgCl₂, and 0.1% Triton X-100) containing 10 mM beta-mercaptoethanol (BME), 1 mM phenylmethylsulfonyl fluoride (PMSF), 1× Roche protease inhibitor cocktail, 0.5-mg/ml lysozyme, and 6-µg/ml DNase I, rotated at 4°C for 1 h to lyse, and cleared by centrifugation at 1,110 g for 20 min. Clarified cell lysates of GST alone or GST-tagged CSN constructs were bound to glutathione beads in 20 mM Tris pH 7.5, 250 mM NaCl, 10% glycerol, 0.1% Triton X-100, 10 mM BME, 1 mM PMSF, and 1× Roche protease inhibitor cocktail for 1 h at 15°C. Binding reactions with clarified cell lysates of His₆-tagged preys were incubated at 15°C for 3 h and washed 4 times with 10 mM Tris pH 7.5, 150 mM NaCl, 0.1% NP-40, and 0.5 mg/ml BSA. For elution, beads were heated to 95°C for 5 min in sodium dodecyl sulfate (SDS) sample buffer and 10 mM dithiothreitol (DTT). To test if the FBF-2/CSN-5/-6-binding interaction is RNA dependent, 50-µg/ml RNase A was added to His₆::FBF-2 lysate before incubation with GST alone or GST::CSN-5/-6.

Western blot

Protein samples from GST pulldown assays were separated using Mini-PROTEAN TGX 4–20% precast gels (Bio-Rad) and visualized using either Coomassie, Western blotting, or stain-free chemistry. Stain-free chemistry allows for the visualization of proteins following electrophoresis by incorporating a trihalo compound that covalently binds to tryptophan residues, thus making the proteins fluorescent when exposed to UV light (Gürtler et al. 2013; Taylor and Posch 2014). For anti-His probing, equal volumes were loaded for all eluates. For anti-GST probing, GST-negative control eluate was diluted 8-fold further than GST::CSN constructs to avoid overexposure. For Western blotting, proteins were transferred to a 0.2-µm PVDF membrane (EMD Millipore). After transfer, membranes were blocked in 50 mM Tris pH 7.5, 180 mM NaCl, and 0.05% Tween 20 with 5% nonfat dry milk powder and probed with antibodies (see Supplementary Table 2 for antibody information) diluted in blocking solution. Membranes were developed using Luminata Crescendo Western HRP substrate (EMD Millipore) and visualized using Bio-Rad ChemiDoc MP Imaging System. The expression, and attachment to GSH beads, of all GST alone and GST-tagged constructs was confirmed for each pulldown experiment.

To assess relative GST::CSN-5^MPN and GST::CSN-6^MPN concentrations, total protein staining densitometry was performed using the Bio-Rad Chemi Doc MP Imaging System. Band intensities were quantified using Image Lab software version 5.1, and GST::CSN-5^MPN was set to 1×. GST::CSN-6^MPN was then serially diluted such that the total protein attached to GSH beads would range from 1× to 10× of GST::CSN-5^MPN.

Protein quantitation by immunoblotting

To analyze FBF protein levels in csn(lf) mutants, 50–100 adult worms of each genotype were picked from synchronous cultures and lysed in SDS–PAGE sample buffer containing 2.8% SDS and 75 mM DTT by boiling for 15 min prior to SDS–PAGE gel electrophoresis. Protein samples were separated and transferred, and membranes were blocked and probed with antibodies as above (see Supplementary Table 2 for antibody information). Band intensities were quantified using Bio-Rad Image Lab software version 5.1. The intensities of FBF bands were normalized to the intensity of tubulin for loading control and scaled to 1 in the control 3xv5::fbf-2 strain or rrf-1; 3xv5::fbf-2 on control RNA interference (RNAi) culture.

PLA

Duolink PLA was performed on dissected adult C. elegans gonads following a modified PLA protocol as described (Day et al. 2020, 2022; Wang et al. 2020). Fixation was as described below in Immunofluorescence section. Blocking steps included incubation in 1× PBS/0.1% Triton-X-100/0.1% BSA (PBS-T/BSA) for 2 × 15 min at room temperature, in 10% normal goat serum for 1 h at room temperature in a light-sealed humid chamber and in Duolink blocking buffer (Sigma-Aldrich) for 1 h at 37°C in a light-sealed humid chamber. Primary antibodies (see Supplementary Table 2) were diluted in PBS-T/BSA and incubated overnight at 4°C. Samples were then incubated with Duolink PLA probes (see Supplementary Table 2) for 1 h at 37°C in a light-sealed chamber. Next, slides were incubated in a light-sealed humid chamber for 30 min at 37°C for ligation and then for 100 min at 37°C for amplification before finally being mounted with Duolink Mounting medium with DAPI (Sigma-Aldrich). Images were acquired using a Zeiss 880 confocal microscope, and the ImageJ “Analyze Particles” plug-in was used to quantify PLA foci in germline images.

RNA extraction and RT-qPCR

C. elegans was synchronized by bleaching, and hatched L1 larvae were plated on NNGM plates with OP50 bacteria, grown at 20°C, and collected 24 h post-L4 stage. A total of 200–400 worms were collected per biological replicate and stored in TRIzol (Invitrogen) at −80˚C. Total RNA was extracted using either Monarch Total RNA Miniprep Kit (New England Biolabs) or Direct-zol RNA MiniPrep Kit (Zymo Research) as per the manufacturer’s protocols. RNA concentration was measured using Qubit (Thermo Fisher). cDNA was synthesized using the SuperScript IV reverse transcriptase (Invitrogen) using 300-ng RNA template per each 20-µL cDNA synthesis reaction. qPCR reactions were performed in technical triplicates per each input cDNA using iQ SYBR Green Supermix (Bio-Rad). Primers for all targets were as previously described: act-1 and fbf-2 (Chauve et al. 2020), unc-54 (Wang et al. 2020), and fbf-1 (Voronina and Seydoux 2010) all primers are listed in Supplementary Table 3. The abundance of each mRNA in csn(lf) mutant relative to the wild type was calculated using the comparative ΔΔCt method (Pfaffl 2001) with actin act-1 as a reference gene. After mRNA abundance of each tested gene was normalized to act-1, the fold change values from replicates were averaged.

Immunofluorescence

C. elegans was synchronized by bleaching, and hatched L1 larvae were plated on NNGM plates with OP50 bacteria, grown at 20°C, and collected 24 h post-L4 stage. Adult hermaphrodites were washed in M9, and germlines were dissected on poly-L-lysine treated slides, flash frozen, and fixed for 1 min in 100% methanol (−20°C) followed by 5 min in 2% paraformaldehyde/100 mM K₂HPO₄ (pH 6) at room temperature. The samples were then blocked in PBS-T/BSA for 30 min at room temperature. Next, samples were incubated with primary antibody (see Supplementary Table 2) diluted in PBS-T/BSA overnight at 4°C. Samples were then washed with PBS-T/BSA 3 times for 10 min per wash, before incubating with secondary antibody diluted in PBS-T/BSA for 2 h at room temperature in a dark humid chamber. Samples were washed again 3 times for 10 min per wash before adding 10-µL Vectashield with DAPI (Vector Laboratories) and coverslipping. Images were acquired with a Leica DFC300G camera attached to a Leica DM5500B microscope with a 40× LP FLUOTAR NA1.3 objective using LAS X software (Leica) and stitched together using Adobe Photoshop CS3.

In situ protein quantitation by immunofluorescence

To analyze PUF-3 levels in csn-5(lf) and FBF levels in csn-5(lf) mutants treated with control or ned-8(RNAi), germlines were immunostained as above. Z-stack images spanning the thickness of the germline were taken under identical conditions across all samples in each experiment. PUF-3 and FBF protein levels within the SPC zone were analyzed using Fiji/ImageJ as previously described with some modifications (Brenner and Schedl 2016; Haupt et al. 2019). On the maximum intensity projection generated from each z-stack, the segmented line (width set to 45) tool was used to draw a freehand line from the distal tip through the length of the SPC zone (as indicated by costaining with REC-8 antibody, Hansen et al. 2004; or DNA morphology with DAPI) that bisected the gonad. Next, pixel intensity data for both V5 and FBF-1 channels were obtained using “Plot Profile” to generate raw intensity curves. To adjust for nonspecific background staining, we subtracted the intensity of the respective negative controls (N2 for V5::PUF-3 analysis and fbf-1(ok91) for FBF-1 and V5::FBF-2 analysis) from intensity curves of each corresponding genotype. The maximum background-subtracted values for each wild-type and mutant germline were determined and used in further analysis, using the first 18 cell diameters for PUF-3 (Brenner et al. 2022) and the first 23 cell diameters for FBFs. The maximum values of wild-type germlines were averaged, and the average value was used for the normalization of all values to set the wild-type levels to 1. To better visualize 3xV5::PUF-3 expression in the SPC zone, images were acquired using a Zeiss 880 confocal microscope under identical conditions, and V5 intensity was adjusted equally in Adobe Photoshop CS3 for 3xv5::puf-3 and csn-5(lf) 3xv5::puf-3 genetic backgrounds.

Germline SPC counts

C. elegans was synchronized by bleaching, and hatched L1 larvae were plated on NNGM plates with OP50 bacteria, grown at 20°C, and harvested 24 h post-L4. Adult gonads were dissected and stained for progenitor cell marker REC-8 (Hansen et al. 2004), and the number of SPCs was measured by counting the total number of cells positive for REC-8 staining using Cell Counter plug-in in Fiji (Schindelin et al. 2012) or Imaris 9.9 software, which produced comparable results.

EdU labeling

5-Ethinyl-2′-deoxyuridine (EdU) labeling was performed as previously described with some modifications (Wang et al. 2020). EdU bacterial plates were prepared by diluting an overnight culture of thymine deficient MG1693 E. coli (The Coli Genetic Stock Center; Yale University) 1/25 in 1% glucose, 1 mM MgSO₄, 5-µg/mL thymine, 6 µM thymidine, and 20 µM EdU in M9. This culture was grown at 37°C for 24 h, pelleted by centrifugation, resuspended in 10 mL M9, and plated on NNGM plates. Worm strains were synchronized by bleaching, and hatched L1 larvae were plated on NNGM plates with OP50 bacteria and grown at 20°C for 72 h to reach young adult stage when they were exposed to EdU-labeled bacteria for either 4 or 14 h. After feeding for the specified time, worms were immunostained as described above. After incubation with secondary antibody (see Supplementary Table 2 for antibody information), slides were washed 4 times for 15 min per wash in PBS-T/BSA. Next, the Click-iT reaction was performed according to manufacturer instructions with the exception that two 30-min Click-iT reactions were performed to increase the signal of the Alexa 488 dye. After incubation with the second Click-iT reaction, slides were washed 4 times for 15 min per wash in PBS-T/BSA before adding 10-µL Vectashield with DAPI (Vector Laboratories) and coverslipping.

Phenotypic analysis

For evaluation of oogenesis, C. elegans was synchronized by bleaching, and hatched L1 larvae were plated on NNGM plates with OP50 bacteria, grown at 20°C, and harvested after 24 h post-L4. Gonads were dissected, stained for DNA with DAPI, and scored for abnormal oocytes (defined as <3 oocytes per germline, uncondensed chromosomes, endomitotic oocytes, or increased number of DAPI spots indicating achiasmatic chromosomes), sperm only, and failure of gamete formation.

For scoring fertility in csn-5(ok1064) mutant strains (Table 1), C. elegans was synchronized by bleaching, and hatched L1 larvae were plated on NNGM plates with OP50 bacteria and grown at 20°C for 96 h before scoring fertility using a dissecting microscope. “Fertile” was classified by the presence of embryos in the uteri, and lack of embryos was classified as sterile.

Table 1.

Germline-expressed CSN-5 rescues fertility defect of csn-5(lf).

Genotype	% Fertile	n
3xv5::fbf-2(q932); csn-5(ok1064)	0	>500
csn-5p::3xflag::csn-5; csn-5(ok1064)	100	>500
gld-1p::3xflag::csn-5; csn-5(ok1064)	100	311
gld-1p::3xflag::csn-5(D152N); csn-5(ok1064)	40a	262

n indicates number of germlines scored.

^a Fertile animals were able to produce embryos that failed to hatch.

RNAi treatment

For RNAi, 2 knockdown methods were utilized. Feeding RNAi was utilized for pbs-6 and fbf-2 RNAi constructs in pL4440 (Kamath and Ahringer 2003). Empty vector pL4440 was used as a control in all RNAi experiments. All RNAi constructs were verified by sequencing and transformed into HT115 (DE3) E. coli. Three colonies of freshly transformed RNAi plasmids were combined for growth in LB/75 µg/mL carbenicillin media for 4 h and induced with 5 mM IPTG for 1–2 h at 37°C. RNAi plates (NNGM containing 100 µg/mL carbenicillin and 0.4 mM IPTG) were seeded with the pelleted cells. Synchronously hatched L1 larvae were plated directly on RNAi plates, grown at 24°C, and collected for analysis after reaching adulthood.

Soaking RNAi for ned-8 was performed as described by Yoon et al. (2012) with modifications. Templates for in vitro transcription of ned-8 flanked by T7 promoter sequences were amplified via PCR from RNAi constructs in pL4440 (Source BioScience RNAi library; Kamath and Ahringer 2003). Templates were then transcribed into double-stranded RNA (dsRNA) in vitro using a T7 MEGAscript Kit (Invitrogen) per the manufacturer's instructions at the 100-µL scale. dsRNA was then recovered using an RNA Cleanup Kit (New England Biolabs) per the manufacturer's instructions. Synchronously hatched L1 larvae (200–400) were then suspended in a soaking buffer containing 2.5× M9, 30 mM spermidine, and 0.5% gelatin with 1 µg/µL dsRNA for 48 h at 20°C. Larvae were then plated on OP50 plates for another 72 h and were collected for analysis after reaching adulthood.

Results

CSN-5 is a new interacting partner of FBFs

To identify proteins that associate with FBF-2, we performed a yeast 2-hybrid screen of a C. elegans cDNA library with FBF-2 as bait (see Materials and Methods for detail). Briefly, the yeast transformed with full-length FBF-2 fused to the Gal4 DNA-binding domain was mated to a yeast library of mixed-stage wild-type C. elegans cDNAs fused to the Gal4 activation domain. Resulting diploids were selected for growth on minimal media lacking histidine and adenine selecting for expression of HIS3 and ADE2, with the addition of 1 mM 3-AT (a competitive inhibitor of the HIS3 enzyme) for stringency. The screen of estimated 2 × 10⁸ yeast diploids identified a known FBF interactor SYGL-1 (Shin et al. 2017) as well as several potential new interactors (Supplementary Table 4), with one of the most abundant partners representing a fragment of COP9 signalosome subunit CSN-5. To validate this potential interaction and to test whether CSN-5 might bind FBF-1 as well as FBF-2, we performed a directed yeast 2-hybrid assay, in which full-length FBF-1 and FBF-2 fused to the Gal4 DNA-binding domain were cotransformed with full-length CSN-5 fused to the Gal4 activation domain. Robust growth on selective media was observed when CSN-5 was cotransformed with either FBF-1 or FBF-2, but not in controls (Fig. 1b). Yeast expression of c-myc-tagged FBFs and HA-tagged CSN-5 was detected by Western blot using anti-Myc and anti-HA, respectively (Supplementary Fig. 2). We conclude that CSN-5 interacts with both FBF-1 and FBF-2 in yeast. Additionally, we performed a GST pulldown assay with bacterially expressed GST-tagged CSN-5 and His₆-tagged FBFs, which detected direct interaction between these proteins (Fig. 1c).

The interaction between the RBD of FBFs and the MPN domain of CSN-5 is conserved in evolution

To elucidate the location of the interacting domains within FBFs and CSN-5, we generated several truncation constructs of FBF-1, FBF-2, and CSN-5 expressing distinct domains amenable to recombinant expression based on the previous studies (Fig. 2b and d; Bernstein et al. 2005; Echalier et al. 2013; Birol et al. 2014) and tested their association via GST pulldown assay. We observed that the conserved RNA-binding domains (RBDs) of both FBFs were sufficient for the interaction with CSN-5, and the binding between FBF-2 and CSN-5 was lost in the absence of FBF-2 RBD (Fig. 2a). Furthermore, the interaction between FBF-2 and CSN-5 was dependent on CSN-5^MPN domain (Fig. 2c). We then tested whether FBF-2 could interact with the other MPN domain of COP9 signalosome, contributed by CSN-6. The MPN domain of CSN-6 was delineated by homology to human CSN6 (Ma et al. 2014). GST pulldown assay suggested that the interaction of His₆::FBF-2 with GST::CSN-6^MPN is much weaker than that with CSN-5^MPN (Fig. 2e). We concluded that the FBF/CSN-5 interaction is mediated by the conserved, structured domains, with selectivity toward specific members of the domain families. We further tested whether FBF-2 interaction with MPN domains of CSN-5 and CSN-6 was RNA dependent and found that the binding was not disrupted by the RNase A treatment (Supplementary Fig. 3).

Fig. 2.

Interaction between FBF RBD and MPN domain of CSN-5 is conserved in evolution. a) GST or GST::CSN-5 were tested for binding His₆::FBF-1^RBD, His₆::FBF-2^RBD, and His₆::FBF-2^N-term constructs. Input is clarified FBF lysate that was added to GST pulldown. Proteins are detected by Western blot with anti-His and anti-GST; the position of full-length GST::CSN-5 (representative image from His₆::FBF-2^N-term pulldown) is marked with an arrowhead, and a lower band represents GST::CSN-5 degradation product (dot). b) FBF-1/-2 and PUM1/2 truncation constructs. Rectangle: RBD. Amino acid positions of the truncations are indicated above each construct. c) GST::CSN-5 truncation constructs were tested for binding His₆::FBF-2. Input is clarified FBF-2 lysate added to GST pulldown. Proteins are detected by Western blot with anti-His and anti-GST; the positions of GST-tagged constructs are marked with arrowheads, and C-terminally truncated fragments of GST::CSN-5 constructs are marked with dots. d) CSN-5, CSN-6, and CSN5 truncation constructs. Rectangle: core MPN domain. C-terminal helices (ovals) that incorporate into the helical bundle of the COP9 signalosome (Lingaraju et al. 2014). Amino acid positions of the truncations are indicated above each construct. e) GST pulldown of His₆::FBF-2 with GST::CSN-5^MPN and increasing concentrations of GST::CSN-6^MPN. FBF-2 is detected by Western blot with anti-His. Input is clarified FBF-2 lysate added to GST pulldown. Total eluate protein was seen by Coomassie to confirm GST retention on GSH beads; arrowheads mark GST::CSN constructs. Relative concentrations of GST-tagged constructs were determined by total protein staining densitometry of GST fusion bands in the eluates, where GST::CSN-5^MPN was set to 1×, and relative amounts of GST::CSN-6^MPN are indicated above the corresponding lanes. f) GST pulldown of human homologs, GST::CSN5^MPN with His₆::PUM1^RBD or His₆::PUM2^RBD. PUM proteins are detected by Western blot with anti-6xHis. Input is clarified PUM lysate added to GST pulldowns. Total eluate protein (representative image from His₆::PUM2^RBD pulldown) detected by stain-free chemistry to confirm GST fusion retention to GSH beads; arrowhead marks GST::CSN5^MPN.

Both FBFs and CSN-5 have homologous human proteins, PUM1/PUM2 and CSN5, respectively (Spassov and Jurecic 2002; Qin et al. 2020). We tested whether the RBDs and MPN domain of the corresponding human homologs (Fig. 2b and d) were able to interact. A GST pulldown assay of GST::CSN5^MPN with His₆::PUM1^RBD and His₆::PUM2^RBD revealed that the interaction we observed for the nematode proteins is conserved for human homologs CSN5 and PUM1, but not PUM2, suggesting an evolutionarily conserved protein complex (Fig. 2f). We considered whether CSN-5 was able to interact with other C. elegans PUF proteins including PUF-8, PUF-11, and PUF-3 but we were unable to produce soluble recombinant PUF proteins to test this possibility.

FBF-2 interacts with CSN-5 in vivo

After characterizing the interaction between FBFs and CSN-5 in vitro, we tested if this interaction is also observed in vivo. We utilized PLAs to detect in situ protein–protein interactions at distances < 40 nm (Söderberg et al. 2008; Day et al. 2020). PLA was performed on 3xflag::csn-5; gfp::fbf-1 and 3xflag::csn-5; gfp::fbf-2 animals using 3xflag::csn-5; gfp as a control for spurious proximity of coexpressed proteins. We observed a statistically significant (P < 0.0001; Fig. 3) increase in PLA signal within the SPC zone in 3xflag::csn-5; gfp::fbf-2 compared to both control and 3xflag::csn-5; gfp::fbf-1. By contrast, despite a number of CSN-5/FBF-1 germlines with higher PLA signal than in control, the increase in PLA signal was not statistically significant. It is possible that FBF-1 could transiently interact with CSN-5, and the PLA assay might not effectively capture those types of interactions. Nonetheless, these data suggest that FBF-2 interacts with CSN-5 in vivo.

Fig. 3.

FBF-2 interacts with CSN-5 in vivo. a–c) Confocal maximal projection images of the entire adult distal germline SPC zones with PLA foci (yellow) and DAPI (blue). Individual DAPI channels (aii, bii, and cii) are shown in grayscale for better contrast as maximal projections spanning half the depth of the germline. Germlines are outlined with dashed lines and vertical dotted lines indicate the beginning of the transition zone as indicated by DNA morphology. Genotypes are indicated with their respective images, where 3xflag::csn-5 transgenes were driven by the csn-5 promoter and gfp and gfp::fbf transgenes were driven by the gld-1 promoter. Scale bars: 10 µm. b) The PLA density (number of PLA foci per µm²) within the SPC zone was measured for germlines of each genetic background; each dot represents a single germline. Differences in PLA density for each protein pair were evaluated by 1-way ANOVA with Tukey's posttest. Asterisks denote statistical significance (****P < 0.0001) where FBF-2 had significantly greater PLA foci density than both FBF-1 and the control. Number of germlines scored (N) are indicated below the graph. Data are representative of 3 biological replicates. Mean group values are shown as lines and error bars denote SEM. All experiments were performed at 24°C.

COP9 is required to maintain FBF-1/2, but not PUF-3 protein levels in SPCs

In C. elegans, CSN-5 was previously identified as an interacting partner of germline proteins, GLH-1 and GLH-3 (Smith et al. 2002; Marnik et al. 2019), and found to promote the accumulation of GLH-1 (Orsborn et al. 2007). Therefore, we hypothesized that CSN-5 might similarly promote FBF accumulation and aimed to test FBF levels in the germlines where csn-5 function was disrupted. The effect of CSN-5 on FBFs may or may not depend on the COP9 complex (Tomoda et al. 1999; Wei et al. 2008; Yoshida et al. 2013). To distinguish COP9-dependent and independent roles of CSN-5, we documented FBF levels in mutants of 2 additional COP9 subunits, csn-2 and csn-6. If the whole complex was required for specific functions, we would expect to see the same, or very similar, phenotypes across all csn(loss-of-function, lf) mutants, as all mutations used in this study are expected to be null alleles (Brockway et al. 2014; Supplementary Fig. 1; Materials and Methods). The COP9-dependent deneddylation would be compromised to a similar extent in all 3 strains since without CSN2 and CSN6, COP9 complex cannot readily incorporate CSN5 and thus remains inactive (Lykke-Andersen et al. 2003; Birol et al. 2014; Lingaraju et al. 2014). Additionally, it is unlikely that adult csn(lf) germlines would have residual CSN protein from maternal contribution (Oron et al. 2002). We quantified the total FBF protein levels in adult csn(lf) mutant worms by Western blotting with antibodies to the endogenous FBF-1 or epitope-tagged endogenous 3xV5::FBF-2 (Shin et al. 2017) and used alpha-tubulin as a loading control for normalization as it is also expressed in the germline (Ortiz et al. 2014). Both FBF-1 and FBF-2 were significantly reduced in csn(lf) mutants; however, the degree of the effect varied. We found the most drastic reduction of FBF-1 protein level in csn-5(lf) (0.24-fold of the wild-type levels; P < 0.005; Fig. 4a and b), whereas csn-6(lf) and csn-2(lf) retained more FBF-1 (0.39-fold and 0.68-fold; P < 0.01 and P < 0.05, respectively; Fig. 4a and b). Furthermore, FBF-1 protein levels in the csn-5(lf) mutant were significantly lower than those in the csn-2(lf) mutant (P < 0.05; Fig. 4b). Immunostaining of csn(lf) mutants (Fig. 4d–g) confirmed reduction of FBF-1 in the distal germline, where FBFs are normally enriched. The stronger FBF-1 protein reduction in csn-5(lf) implies that CSN-5 might be promoting the accumulation of FBF-1 independently of the COP9 complex, since the relative amounts of FBF-1 would be similar across all mutants otherwise. Surprisingly and distinctly, we observed a significant reduction of 3xV5::FBF-2 in all csn(lf) mutants by Western blot (0.29-fold; P < 0.0001; Fig. 4a and c) and confirmed this observation by immunostaining (Fig. 4h–k), suggesting COP9 dependence. The reductions in FBF levels determined by immunoblotting whole worm lysates were consistent with those determined by quantification of immunofluorescent labeling in dissected germlines (see Fig. 8a and b). We conclude that CSN-5, as well as COP9, promotes FBF steady-state accumulation.

Fig. 4.

FBF-1/-2, but not PUF-3, protein levels are reduced in csn(lf) mutants. a) Western blot analysis of 3xv5::fbf-2(q932), referred to throughout the figure as wild type, and respective 3xv5::fbf-2(q932); csn(lf) adult mutant worm lysate reveals reduced levels of FBF-1 (top panels) and 3xV5::FBF-2 (bottom panels). Alpha-tubulin is used for normalization as a loading control. Endogenous FBF-1 and epitope-tagged endogenous 3xV5::FBF-2 are detected by anti-FBF-1 and anti-V5. b, c) Quantitation of total FBF protein level for each 3xv5::fbf-2(q932); csn(lf) mutant. A ratio of FBF/tubulin Western blot intensities in each strain was normalized to the ratio in 3xv5::fbf-2(q932). Differences in protein level were evaluated by 1-way ANOVA with Tukey's posttest. Gray asterisks denote statistical significance compared to wild type, and blue bracket and asterisk denote statistical significance comparing csn-2(lf) to csn-5(lf) (*P < 0.05; **P < 0.01; ***P < 0.005; ****P < 0.0001). Number of biological replicates (N) is indicated at the bottom of the graphs. Mean group values are shown as lines, and error bars denote standard deviation. d–k) Distal germlines of adult wild type (3xv5::fbf-2(q932)) and respective 3xv5::fbf-2(q932); csn(lf) mutants are dissected and stained with anti-FBF-1 (d–g) or anti-V5 (h–k) and DAPI to visualize protein levels of FBF-1/-2. The individual FBF and DAPI channels are also shown separately for better contrast. l, m) Distal germlines of adult 3xv5::puf-3 and csn-5(lf) 3xv5::puf-3 dissected and stained with anti-V5 and DAPI. Germlines are outlined with dashed lines. Scale bars: 10 μm. n) Peak 3xV5::PUF-3 protein levels in the distal germlines were determined by immunofluorescence and scaled to average peak levels in the 3xv5::puf-3 strain. Each data point represents a single germline. Differences in peak 3xV5::PUF-3 levels were evaluated by Student's t-test. Asterisks denote statistical significance (****P < 0.0001). Number of germlines scored (N) is indicated at the bottom of the graphs. PUF-3 abundance appeared higher in csn-5(lf) than in csn-5(+) across the distal 14 cell diameters (E.O., unpublished results). Mean group values are shown as lines, and error bars denote standard deviation. Data reflect 3 biological replicates.

Fig. 8.

ned-8(RNAi) rescues FBF-2 levels along with SPC numbers and oogenesis in csn-5(lf). a, b) Peak levels of FBF-1 a) and FBF-2 b) in distal germlines determined by immunostaining of 3xv5::fbf-2(q932), referred to here as wild type, and 3xv5::fbf-2; csn-5(lf), referred to here as csn-5(lf) treated with control or ned-8(RNAi). Endogenous FBF-1 and epitope-tagged endogenous 3xV5::FBF-2 detected by anti-FBF-1 and anti-V5. Peak FBF protein intensity in each germline was normalized to the average of wild-type control RNAi after background subtraction. Differences in protein level between control and ned-8(RNAi) were evaluated by 1-way ANOVA with Sidak's posttest, ns denotes not significant, and red asterisks denote statistical significance (****P < 0.0001). Number of germlines scored (N) is indicated at the bottom of the graphs. Mean group values are shown as lines, and error bars denote standard deviation. The data reflect 3 biological replicates. c–f) Representative images of wild-type and csn-5(lf) mutant germlines dissected and stained with anti-FBF-1 and anti-V5 to visualize protein levels of FBF-1/-2. Germlines are outlined with dashed lines. Scale bars: 10 μm. g) Quantification of SPCs in wild-type and csn(lf) mutants with ned-8(RNAi); each data point represents a single germline. Differences in number of SPCs between control and ned-8(RNAi) were evaluated by 1-way ANOVA with Sidak's posttest. ns denotes not significant, and asterisks denote statistical significance (****P < 0.0001). Number of germlines scored (N) are indicated at the bottom of the graph. SPC quantification reflects 4 biological replicates of csn-5(lf) control RNAi and 3 biological replicates of wild-type control RNAi, wild-type ned-8(RNAi), and csn-5(lf) ned-8(RNAi). Mean group values are shown as lines, and error bars denote standard deviation. h) Gametogenesis in ned-8(RNAi)-treated wild-type and csn-5(lf) mutant. Number of germlines scored (N) are indicated at the top of each column. Data are plotted in aggregate and reflect 3 biological replicates.

The reduction in FBF-1/2 protein levels may be caused by a decrease in fbf transcripts or by an effect on FBF proteins. To start distinguishing these possibilities, we tested if csn(lf) affects the steady-state levels of fbf-1/-2 mRNAs using qPCR. Using a largely somatic gene unc-54 (myosin heavy chain) to assess potential global downregulation of transcripts in the csn(lf) mutant backgrounds and normalizing mRNA abundance to a gender-neutral, germline-expressed reference gene act-1 (actin; Ortiz et al. 2014), we observed a reduction in fbf-1 mRNA abundance in each of the mutants, but the difference was not statistically significant in this experiment (ns, P = 0.09; Supplementary Fig. 4; however, in other experiments, we observed significant downregulation of fbf-1 mRNA in csn-5(lf) (Fig. 7e). Additionally, we observed a statistically significant reduction of fbf-2 mRNA in each csn(lf) mutant compared to wild type (P < 0.0001; Supplementary Fig. 4). By contrast, we did not detect a decrease in unc-54 mRNA in any csn(lf) mutant (ns, P = 0.13) indicating a specific effect on fbf mRNAs without global transcript downregulation (Supplementary Fig. 4). As the effects of all csn(lf) mutants on the levels of fbf transcripts were similar, we conclude that COP9 signalosome is required for normal steady-state levels of fbf mRNAs. However, distinct levels of FBF-1 protein in csn-2(lf) vs csn-5(lf) mutant backgrounds (Fig. 4a and b) suggest that CSN-5 might additionally promote FBF-1 protein accumulation.

Fig. 7.

csn-5(lf) destabilizes FBF proteins. a, c) FBF protein levels after treatment with a control RNAi and pbs-6(RNAi) in rrf-1(lf); 3xv5::fbf-2(q932), referred to as rrf-1(pk1417) here, and rrf-1(lf); 3xv5::fbf-2(q932); csn-5(lf), referred to as rrf-1(pk1417); csn-5(ok1064). a) Endogenous FBF-1 and c) epitope-tagged endogenous 3xV5::FBF-2 are detected by anti-FBF-1 and anti-V5. Alpha-tubulin is used for normalization as a loading control. b, d) Quantitation of total FBF protein levels. A ratio of FBF/tubulin Western blot intensities in each condition was normalized to wild-type control RNAi. Control RNAi data points for each genetic background are shown as circles, and pbs-6(RNAi) is shown as the triangles. Difference in FBF protein level was evaluated by 1-way ANOVA with Sidak's posttest. Red brackets with asterisks denote statistical significance comparing csn-5(lf) control RNAi to csn-5(lf) pbs-6(RNAi) (*P < 0.05; **P < 0.01); ns denotes not significant. Number of biological replicates (N) is indicated below the graph, mean value is indicated as a line, and error bars denote standard deviation. e) Steady-state levels of transcripts indicated on the X-axis are quantified by RT-qPCR and normalized to reference germline-expressed gene act-1; unc-54 is a somatic transcript that is not downregulated in csn-5(lf). Differences in relative mRNA abundance of fbf-1/-2 were evaluated by 1-way ANOVA with Dunnett's posttest. Asterisks denote statistical significance compared to wild-type control RNAi (**P < 0.01; ***P < 0.005; ****P < 0.0001). The data reflect 4 biological replicates of control RNAi for rrf-1(lf) and rrf-1(lf); csn-5(ok1064), 3 replicates of rrf-1(lf); pbs-6(RNAi), and 2 replicates of rrf-1(lf); csn-5(ok1064); pbs-6(RNAi). Mean values are plotted as lines with error bars representing standard deviation.

In addition to FBF-1 and FBF-2, C. elegans germline SPCs express 3 other members of PUF family RNA-binding proteins with regulatory roles in germline stem cell maintenance, PUF-3, PUF-11, and PUF-8 (Haupt et al. 2020; Ariz et al. 2009; Racher and Hansen 2012). Although we were unable to determine if CSN-5 interacts with other PUF proteins in vitro, we tested whether the levels of another PUF were affected by csn-5(lf). We investigated PUF-3, for which a tagged 3xv5 allele is available (Haupt et al. 2020). PUF-3 is expressed at low levels in the progenitor zone and reaches its strongest expression in the oocytes (Haupt et al. 2020; Spike et al. 2022). For direct comparison with FBFs, we focused on the progenitor zone and quantified 3xV5::PUF-3 across the distal 25 cell diameters (Fig. 4l and m). We observed a statistically significant increase in peak 3xV5::PUF-3 levels (1.8-fold; P < 0.0001; Fig. 4n) in the distal csn-5(lf) germline, opposite of what was seen for FBFs (Fig. 4b and c). We conclude CSN-5 contribution to protein accumulation is specific for FBFs in contrast to other PUF family members, such as PUF-3.

COP9 subunit mutants have reduced numbers of SPCs that continue to proliferate and enter meiosis

If CSN-5 facilitates FBF function in the germline, we expect csn-5(lf) mutation to cause developmental defects similar to those caused by fbf-1/2(lf). Furthermore, we tested whether greater residual FBF-1 levels in csn-2(lf) mutant compared to csn-5(lf) or csn-6(lf) correlated with milder phenotypes relevant to FBF function. Because FBFs regulate the switch from mitosis to meiosis in the C. elegans germline (Crittenden et al. 2002; Wickens et al. 2002) and therefore affect the size of its SPC population, we quantified the number of SPCs in csn(lf) mutants. We dissected and stained adult germlines with the antibody specific to proliferating cells in the progenitor zone, anti-REC-8, to visualize the SPC zone (Fig. 5ai–iv; Hansen et al. 2004). In general agreement with previous reports (Smith et al. 2002; Brockway et al. 2014), we observed that all csn(lf) mutants had reduced numbers of SPCs compared to the wild type (P < 0.0001; Fig. 5b). The csn-6(lf) and csn-5(lf) mutants had the most drastic reduction in SPCs, whereas the csn-2(lf) SPCs were less affected (P < 0.01; Fig. 5b). This supports the idea that the strongest effect on SPC number correlates with the strongest depletion of FBFs. Although a decrease in SPCs is not a phenotype unique to fbf(lf), it is consistent with compromised FBF function. To test whether csn-5 functions in the same genetic pathway as fbfs, we quantified the number of SPCs in csn-5(lf) following fbf knockdown with RNAi that targets both fbf-1 and fbf-2 (we were not able to generate a triple-mutant strain because of incompatibility between mIn1 and nT1 balancers). When cultured at higher temperatures (24°C), fbf(RNAi)-treated animals maintain reduced numbers of SPCs (Merritt and Seydoux 2010; Hansen and Schedl 2013). The efficacy of RNAi treatment was confirmed by germline masculinization of csn-5(lf)/nT1 as seen in fbf-1/-2(lf) mutants (Supplementary Fig. 5b; Crittenden et al. 2002) and by a reduction in SPC numbers in a wild-type strain (G. Weiss, unpublished results). We observed no additive effect on SPC number in csn-5(lf) fbf(RNAi) background (Supplementary Fig. 5a), supporting the hypothesis that csn-5 functions in the same genetic pathway as fbf-1 and fbf-2 in germline SPCs. We do not know whether the reduced SPC number in csn-5/nT1 (∼90 vs ∼250 in wild type) and its apparent insensitivity to fbf(RNAi) reflect haploinsufficiency of csn-5, an effect of nT1 balancer, or a combination of the two.

Fig. 5.

COP9 subunit mutants have reduced number of SPCs but sustain proliferation and meiotic entry. (ai–iv) Distal germlines dissected from adult 3xv5::fbf-2(q932), referred to as wild type, and respective 3xv5::fbf-2(q932); csn(lf) mutants were stained with anti-REC-8 to visualize SPC zone. Germlines are outlined with dashed lines, and vertical dotted lines indicate the SPC zone boundary determined by REC-8 staining. Scale bars: 10 μm. b) Quantification of SPCs in wild-type and csn(lf) mutants. Differences in number of SPCs were evaluated by 1-way ANOVA with Tukey's posttest. Gray asterisks denote statistical significance compared to wild type (****P < 0.0001), and blue asterisks denote statistical significance compared to csn-2(lf) (**P < 0.01). Each data point represents a single germline and number of germlines scored (N) are indicated at the bottom of the graph. Data reflect 6 biological replicates of wild type and 3 biological replicates for each csn(lf) mutant. ci–iii) Distal germlines dissected from adult 3xv5::fbf-2(q932), referred to as wild type, and respective 3xv5::fbf-2(q932); csn(lf) mutants after a 4-h pulse were stained for EdU (green) and anti-pH3 (red) to visualize cells in S-phase and identify mitotically dividing cells, respectively. Germlines are outlined with dashed lines. Scale bars: 10 μm. d) Germlines were scored for the presence of EdU incorporation post-4-h pulse, pH3(+) cells, both, or neither. Number of germlines scored (N) are indicated at the top of each column. Data are plotted in aggregate and reflect 2 biological replicates. e–g) Distal germlines from adult 3xv5::fbf-2(q932), referred to as wild type, and respective 3xv5::fbf-2(q932); csn(lf) mutants post-14-h pulse with EdU dissected and stained for EdU and REC-8 to visualize cells in S-phase and the SPC zone, respectively. Arrows indicate EdU(+)/REC-8(−) cells entering meiosis. Corresponding percent of germlines entering meiosis with number of germlines scored (N) indicated in overlay panels (iii), data reflect 2 biological replicates. Germlines are outlined with dashed lines and vertical dotted lines indicate the SPC boundary (by REC-8 staining). Scale bars: 10 μm.

Because csn-5(lf) and csn-6(lf) mutant germlines are noticeably smaller than wild type (Fig. 5aiii and iv, and full germline images in Fig. 6aiii and iv), we tested whether their germline SPCs arrested cell cycle progression. To test whether germline SPCs continue to proliferate and enter meiosis, we performed EdU incorporation assays combined with immunostaining with antibodies specific for mitotic M-phase or proliferating cells in the progenitor zone. First, young adult germlines were labeled with the thymidine analog EdU for 4 h and stained for phospho-histone H3 (pH3) to identify cells in S-phase and M-phase, respectively, and scored by how many germlines were positive for pH3, EdU incorporation, and both pH3 and EdU incorporation or had neither (Fig. 5c and d). We found that approximately 90% of csn-5/-6(lf) germlines had SPCs positive for EdU incorporation and/or pH3, together, indicating active cell cycle progression. Next, we tested if csn-5/-6(lf) SPCs were capable of meiotic entry by staining adult gonads that have been labeled with EdU for 14 h with REC-8, and scoring how many germlines had EdU-positive cells that were able to initiate meiosis and move beyond the SPC zone as detected by the loss of REC-8 staining (Fox et al. 2011; Kocsisova et al. 2018). With the longer 14-h EdU pulse (Fig. 5e–g), all germlines were positive for EdU labeling (E.O., unpublished results) suggesting the 10% of csn-5/-6(lf) germlines negative for both pH3 and EdU incorporation after the shorter 4-h EdU pulse (Fig. 5d) likely reflect slower cell cycle rather than cellular arrest. Approximately 98 and 95% of csn-5/-6(lf) germlines, respectively, exhibited progression of EdU-positive cells into meiosis (Fig. 5e–g). Additionally, we observed weaker EdU signal in the 2 csn(lf) mutants and found that EdU-positive cells accumulated slower than in the wild type despite identical EdU treatment conditions (E.O., unpublished results; see Materials and Methods for detail). We conclude that csn-5(lf) and csn-6(lf) germline SPCs sustain proliferation and meiotic entry, although both cell cycle and meiotic entry appear slower than in the wild type.

Fig. 6.

Mutations in COP9 subunits disrupt oogenesis. ai–iv) Gonads of adult 3xv5::fbf-2(q932), referred to as wild type, and respective 3xv5::fbf-2(q932); csn(lf) mutants indicated in each panel dissected and stained for DNA with DAPI. Arrows indicate gametes. Germlines are outlined with dashed lines. Scale bars: 10 μm. b) Gametogenesis in wild-type and csn(lf) mutants. Number of germlines scored (N) are indicated at the top of each column. Data are plotted in aggregate and reflect 6 biological replicates of wild type, 4 replicates of csn-6(lf), 3 replicates of each csn-2(lf) and csn-5(lf), and 2 replicates of csn-2(lf); fbf-1(lf). c, d) Gonads of adult 3xv5::puf-3 and csn-5(lf) 3xv5::puf-3 mutant indicated in each panel dissected and stained for DNA with DAPI, V5::PUF-3 with anti-V5 (c), or sperm with anti-MSP (d). Arrows indicate gametes and embryos. Germlines are outlined with dashed lines. Scale bars: 10 μm. e) Oogenesis in csn-6(lf) is partially restored by an extra copy of csn-5 gene in 3xflag::csn-5; csn-6(lf) strain. A number of germlines scored (N) are indicated at the top of the graph. Data are plotted in aggregate and reflect 3 biological replicates.

Defective oogenesis in COP9 subunit mutants

FBFs play a role in germline sex determination by regulating the spermatogenesis to oogenesis switch (Zhang et al. 1997; Crittenden et al. 2002), so we assessed if germline sex determination was affected in the csn(lf) mutants. Again, we hypothesized that greater residual FBF-1 levels in csn-2(lf) mutant compared to csn-5(lf) or csn-6(lf) would correlate with milder phenotypes relevant to FBF function. The majority of csn-6(lf) and csn-5(lf) mutants failed to produce oocytes and generated only sperm, in agreement with previous work utilizing csn-5(RNAi) (Smith et al. 2002; Fig. 6aiii and iv and b). This raised the possibility that downregulation of FBF protein in these csn(lf) genetic backgrounds might be secondary to germline masculinization, as previously observed for GLD-1 (Jones et al. 1996). To address this, we compared FBF levels in the male vs hermaphrodites of the same genetic background and observed no significant difference in FBFs (Supplementary Fig. 6a), suggesting that the decrease in FBF protein levels is not a secondary consequence of any defect in germline sex determination. Although germline masculinization can result from a variety of genetic causes (reviewed in Ellis 2022), it is consistent with a decrease in fbf function (Zhang et al. 1997; Crittenden et al. 2002). Curiously, the csn-2(lf) mutant had a clearly distinct phenotype where most germlines could still form oocytes (Fig. 6aii and b). As we observed significantly more FBF-1 protein in csn-2(lf) than in csn-5(lf) (Fig. 4b), we next combined csn-2(lf) with fbf-1(lf) to test if the maintenance of oogenesis in csn-2(lf) was dependent on FBF-1. We found the failure of oogenesis in csn-2(lf); fbf-1(lf) germlines increased to 82%, similar to those observed in csn-5(lf) (83%; Fig. 6b). This is consistent with the interpretation that the reduction of FBFs underlies disrupted oogenesis in csn-5/-6(lf) mutants.

Despite the lack of oocytes by DAPI staining, csn-5(lf) and csn-6(lf) germlines do not produce excess of sperm observed in many masculinized genetic backgrounds (Ellis 2022). This may be a result of cell cycle disruption interfering with continued gametogenesis or a block in the progression of oogenic germ cells from pachytene to diplotene. To distinguish between these possibilities and investigate molecular markers of oogenesis in csn-5(lf) germlines, we examined the expression of PUF-3 that is highly produced in proximal pachytene at the start of oogenesis and throughout oocytes (Haupt et al. 2020; Spike et al. 2022; Fig. 6ci). We observed the absence of 3xV5::PUF-3 at the proximal end of csn-5(lf) germlines in the presence of sperm, suggesting a failure of oogenesis (Fig. 6cii and dii). This supports our initial interpretation that csn-5(lf) mutant germlines fail to promote oogenesis, rather than arrest oogenic pachytene cells at the transition to diplotene.

CSN-5's role in oogenesis is largely independent of COP9 and cell autonomous

We hypothesized that the similar germline SPC and disrupted oogenesis phenotypes observed in csn-5(lf) and csn-6(lf) mutants were caused by a decrease in the levels of CSN-5 protein in csn-6(lf), as previously observed after csn-6(RNAi) (Miller et al. 2009). Indeed, we observed that the levels of 3xFLAG::CSN-5 transgene in a csn-6(lf) background decreased by approximately 60% compared to the control (Supplementary Fig. 6b). We aimed to test whether CSN-5-dependent and COP9-independent phenotypes might be rescued by increasing csn-5 copy number in the csn-6(lf) background. To this end, we first tested whether the 3xflag::csn-5 transgene was able to complement the csn-5(lf) mutant and observed complete rescue of fertility (Table 1; Supplementary Fig. 7a). Next, we introduced the 3xflag::csn-5 transgene into csn-6(lf) and found that an extra copy of csn-5 was able to partially rescue oogenesis of csn-6(lf) (Fig. 6e). However, it was unable to rescue the low numbers of SPCs in csn-6(lf) (E.O., unpublished results). Ectopic CSN-5 in csn-6(lf) would be unable to rescue COP9 function because in the absence of CSN-6, CSN-5 fails to incorporate into the COP9 holoenzyme and the complex remains enzymatically inactive (Birol et al. 2014; Lingaraju et al. 2014). We conclude that the role of CSN-5 in oogenesis does not require COP9 signalosome.

Because CSN-5 is expressed in both germline and somatic tissues (Smith et al. 2002; Miller et al. 2009), we tested whether its function was required in the soma or the germline. We generated a 3xflag::csn-5 transgene using the gld-1 promoter, which is expected to restrict expression to the germline (Ellenbecker et al. 2019; see Materials and Methods for details), and still observed 100% rescue of fertility when incorporated into the csn-5(lf) mutant background (Table 1; Supplementary Fig. 7b). This suggests that the effects of csn-5 on oogenesis are likely cell autonomous within the germline. We next tested whether the csn-5 function in promoting oocyte development was dependent on its deneddylating activity by generating a catalytically inactive CSN-5(D152N) transgene, based on a previously reported catalytic mutant (Cope et al. 2002; Peth et al. 2007). CSN-5(D152N) is incorporated into COP9 but fails to promote deneddylation (Cope et al. 2002); therefore, it would only rescue the COP9-independent function of csn-5. We find that 3xFLAG::CSN-5(D152N) transgene can partially rescue oogenesis in csn-5(lf) (Table 1; Supplementary Fig. 7c), suggesting that the effect of csn-5 on gametogenesis is not entirely dependent on its proteolytic activity. However, CSN-5-mediated deneddylation by COP9 is required for meiotic synapsis (Brockway et al. 2014) and embryonic cell division (Pintard et al. 2003). Accordingly, the oogenic germlines of 3xflag::csn-5(D152N); csn-5(lf) had extended persistence of leptotene/zygotene chromosome morphology (Supplementary Fig. 7c), and all embryos produced by the 3xflag::csn-5(D152N); csn-5(lf) were dead. Additionally, inactive CSN-5(D152N) maintains its ability to bind FBF-2 in vitro via GST pulldown (Supplementary Fig. 7d) as expected if its effect depended on interaction with FBF-2.

CSN-5 maintains FBF protein levels through multiple mechanisms

To begin investigating a mechanism behind the compromised FBF accumulation and function in csn-5(lf) mutants, we tested whether CSN-5 stabilizes FBF proteins by interfering with their proteasome-mediated degradation. If true, we expect that disruption of the proteasome function in csn-5(lf) mutants would restore FBF protein levels. To test this hypothesis, we knocked down pbs-6, a catalytic subunit of the 20S proteasome, in the rrf-1(lf) genetic background to preferentially direct the RNAi knockdown to the germline (Sijen et al. 2001; Kumsta and Hansen 2012). By Western blotting, we observed a complete rescue of both steady-state FBF-1 (P < 0.01) and 3xV5::FBF-2 (P < 0.05) protein levels in the adult worm lysates of rrf-1(lf); csn-5(lf) following pbs-6(RNAi) after normalization to tubulin (Fig. 7a–d). Due to the highly pleiotropic effects of proteasome disruption by pbs-6(RNAi), we were unable to assess phenotypic effects of this FBF protein rescue. Since we previously observed a decrease in fbf-1/2 transcript levels in csn-5(lf) background (Supplementary Fig. 4), we next determined whether the rescue of FBF protein accumulation upon proteasome disruption was associated with an increase in fbf mRNA abundance. Quantification of steady-state fbf mRNA levels by qRT-PCR revealed that pbs-6(RNAi) did not increase fbf transcript levels (Fig. 7e) despite the complete rescue of FBF proteins. At this time, the role of COP9 in transcription remains somewhat enigmatic (for review, see Chamovitz 2009), and the effect of COP9 on fbf transcription is the subject of future research. We conclude that FBF is subject to proteasomal-mediated degradation, which is counteracted by CSN-5.

The main enzymatic activity of the COP9 signalosome is deneddylation of cullin subunits of Skp1-cullin 1-F-box (SCF) ubiquitin ligases to promote turnover of the SCF targets (Lyapina et al. 2001; Cope et al. 2002; Pintard et al. 2003). Therefore, some phenotypes of csn(lf) mutants could result from the increased neddylation. Notably, a knockdown of C. elegans NEDD8 homolog, ned-8, has been shown to partially suppress meiotic defects seen in csn(lf) mutants (Brockway et al. 2014). By contrast, the effects of mammalian CSN5 on the stability and function of its interactors HIF-1a and E2F1 are independent of its deneddylase activity (Bemis et al. 2004; Hallstrom and Nevins 2006). To investigate whether csn-5 is affecting FBF stability via deneddylation, we performed ned-8(RNAi) on the csn-5(lf) mutant (we were unable to generate a ned-8; csn-5 mutant strain for analysis due to larval arrest observed in the homozygous ned-8 mutant; Brockway et al. 2014; E.O., unpublished results). We quantified in situ expression of FBF-1 and V5::FBF-2 in distal mitotic cells simultaneously by costaining the treated germlines with respective antibodies. Interestingly, we observed a partial rescue of peak FBF-2, but not FBF-1, levels in ned-8(RNAi)-treated csn-5(lf) mutants (increased to 0.76-fold on average; Fig. 8a–f). We conclude that CSN-5 contributes to FBF-1 and FBF-2 maintenance through 2 mechanisms, stabilizing FBF-1 independent of the deneddylating activity and promoting FBF-2 accumulation through a combination of deneddylation-dependent and deneddylation-independent mechanisms. Consistent with a partial rescue of FBF-2 function, we find that ned-8(RNAi) treatment of csn-5(lf) partially rescues SPCs numbers (Fig. 8g). Additionally, we observed that ned-8(RNAi) completely rescues oogenesis (Fig. 8h). However, interpretation of this result is complicated by the fact that neddylation is essential for the function of cullin-based E3 ubiquitin ligases (Duda et al. 2008; Saha and Deshaies 2008; Boh et al. 2011), and a cullin-based CUL-2/FEM-1/FEM-2/FEM-3 ubiquitin ligase is required for spermatogenesis (Starostina et al. 2007). Therefore, it is unclear whether the rescue of oogenesis is solely due to an increase in FBF-2 or also to the downregulation of CUL-2/FEM-1/FEM-2/FEM-3 activity.

Discussion

In this study, we have identified the COP9 signalosome component CSN-5 as a new interacting partner of PUF family RNA-binding proteins FBF-1 and FBF-2 and documented both COP9-independent and COP9-dependent CSN-5 functions in promoting SPC proliferation as well as oogenesis in C. elegans germline. We find that FBF-1 and FBF-2 are destabilized in the csn-5(lf) mutant germlines resulting in a significant reduction of their protein levels. We propose that destabilization of FBF proteins accounts for a number of phenotypes observed in csn-5(lf) mutant and that CSN-5 maintains steady-state levels of FBF-1 and FBF-2 in the C. elegans germline by distinct mechanisms (Fig. 9).

Fig. 9.

Model for CSN-5 activity stabilizing FBFs. a) CSN-5 promotes the stability of FBFs both in the context of COP9 signalosome and independent of COP9, thereby allowing FBFs proteins to maintain the SPC zone and promoting oogenesis. b) Intrinsically unstable FBF proteins are readily degraded by the proteasome resulting in a germline with fewer SPCs and no female gametes in the absence of CSN-5.

CSN-5/FBF interaction is mediated by MPN and PUF domains

We found the interaction between CSN-5 and FBFs in a yeast 2-hybrid screen and confirmed it with GST pulldowns of recombinant proteins (Fig. 1b and d). CSN-5 has not been reported in the prior published yeast 2-hybrid screen using FBF-1 as a bait (Kraemer et al. 1999; Eckmann et al. 2002), but it might be a result of differences between the assays. For example, fusion of the bait protein to the LexA DNA-binding domain rather than Gal4 used in our study may have sterically prevented some interactions. Additionally, the prey library cDNA in our study was generated with an oligo-dT primer, while the previous studies used random priming; these methods are expected to produce distinct transcript representation biases. We found that the direct interaction between PUF family proteins (FBF-1, FBF-2, and PUM1) and CSN5 homologs is mediated by conserved structured domains, the RBD and the MPN domain, respectively (Fig. 2a, c, and f). However, not all PUF RBDs bind CSN5 and not every type of MPN domain binds PUF proteins. At this time, the determinants of PUF/MPN binding are enigmatic. PUF domains of PUM1 and PUM2 are very similar to each other (over 90% identical; Spassov and Jurecic 2002), yet they exhibit distinct interaction properties with the CSN5^MPN (Fig. 2f). On the other hand, FBF-1 and FBF-2 belong to a separate, nematode-specific clade of PUF RBDs, with less than 30% identity to PUM RBDs (Wickens et al. 2002; Spassov and Jurecic 2003), yet both are able to interact with CSN-5^MPN. This raises the question whether additional PUF proteins out of 10 described in C. elegans also bind to CSN-5. In addition to FBFs, germline SPCs express PUF-3 and PUF-11 that cluster into a nematode-specific subfamily related to FBFs (Haupt et al. 2020) and PUF-8 that is closer to human and Drosophila homologs (Ariz et al. 2009; Racher and Hansen 2012). We were not able to test the association between CSN-5 and PUF domains of PUF-3, PUF-11, and PUF-8 due to difficulties with recombinant PUF expression. If these PUFs interact with CSN-5, it is possible that some of the effects of csn-5(lf) are due to the downregulation of additional germline PUFs. However, we do not observe a downregulation of PUF-3 in csn-5(lf) SPCs (Fig. 4l–n), suggesting specificity toward FBFs.

Similar to CSN-5, CSN-6 contains an MPN domain (Birol et al. 2014). However, CSN-6's MPN domain could only bind FBF-2's RBD at much higher concentrations than CSN-5 (Fig. 2e). Taken together, this suggests a selective association between a subset of PUF RNA-binding proteins and their specific MPN domain-containing partners. In the future, these selective interactions can help elucidate the sites mediating binding between these proteins.

We further find in vivo association between CSN-5 and FBF-2 by PLA (Fig. 3). Although the in vivo proximity signal between CSN-5 and FBF-1 did not reach statistical significance, it is possible that the interaction of CSN-5 with FBF-1 might be unstable or transient and thus remain undetectable with PLA. In line with this, coimmunoprecipitation has not recovered CSN-5 as an FBF-1 interactor (Friend et al. 2012). One explanation of the higher prevalence of CSN-5/FBF-2 interaction is that both CSN-5 and FBF-2 are present in P granules and thus might interact more frequently compared to FBF-1 that shows only minor localization to P granules (Smith et al. 2002; Voronina et al. 2012; Marnik et al. 2019). Alternatively, FBF-1 might be indirectly regulated by CSN-5. Nonetheless, here, we demonstrate that both FBFs require CSN-5 for proper accumulation and function.

CSN-5 promotes FBF accumulation

CSN-5 and its homologs promote the stabilization of their interaction partners (Bemis et al. 2004; Orsborn et al. 2007). We demonstrate that CSN-5 is similarly involved in stabilizing FBFs, as both FBFs are significantly reduced in csn(lf) mutants (Fig. 4a–k). We find that FBF-1 is most significantly reduced in csn-5(lf), compared to csn-6(lf) or csn-2(lf) (Fig. 4a and b), suggesting that csn-5 can promote FBF-1 accumulation independent of the COP9 complex. Conversely, we find that FBF-2 is depleted similarly in the csn(lf) mutants (Fig. 4a and c), suggesting the entire complex might promote FBF-2 protein accumulation. Interestingly, we also observed a reduction of fbf mRNAs in the csn(lf) mutants (Supplementary Fig. 4; Fig. 7e), indicating that the entire COP9 complex is needed to maintain steady-state transcript levels. The statistical significance of the decrease in fbf-1 mRNA varied between these experiments, which may stem from the genetic background (wt vs rrf-1(lf)) and bacterial diet (OP50 vs HT115); each of these variables may have contributed to distinct outcomes. Previous work found that FBF proteins bind fbf mRNAs and thus FBFs have been speculated to autoregulate and cross-regulate each other (Lamont et al. 2004; Prasad et al. 2016). However, the reduction in fbf mRNAs observed in csn mutants is unlikely to result from a disruption in FBF autoregulation, since the rescue of FBF proteins with pbs-6(RNAi) did not restore fbf mRNA levels (Fig. 7). As reviewed in Chamovitz (2009), a clear understanding of the COP9 complex role in regulating transcription, outside the realm of stabilizing transcription factors and thus affecting steady-state transcript levels, remains elusive. The effect of COP9 on regulating steady-state levels of fbf mRNAs remains the subject of future research. Taken together, we conclude that CSN-5 regulates FBF protein levels through both COP9-independent and COP9-dependent mechanisms. This function of CSN-5 in FBF protein accumulation, regardless of COP9-dependence, reveals a previously unappreciated role for CSN-5 and also provides the first insight regarding the posttranslational regulation of PUF family members FBF-1 and FBF-2.

By contrast, we find that CSN-5 does not promote the accumulation of a related PUF-3, as we observed a significant increase of PUF-3 protein in SPCs of the csn-5(lf) mutant germlines (Fig. 4l–n). The levels of PUF-3 and PUF-11 in SPCs are restricted by a TRIM-NHL protein NHL-2 (Brenner et al. 2022), and their degradation at the oocyte-to-embryo transition is mediated by the ubiquitin-proteasome system (Spike et al. 2022). The effect on PUF-3 accumulation is distinct from that seen with FBFs indicating CSN-5 specifically regulates FBFs and phenotypes observed in csn-5(lf) are not due to a downregulation of multiple PUF proteins. As PUF-3 functions in parallel to FBFs in germline stem cell maintenance (Haupt et al. 2020), upregulation of PUF-3 may serve to compensate for FBF downregulation in csn-5(lf) and sustain SPC function in csn-5(lf).

CSN-5 promotes FBF protein stability

In this work, we uncovered destabilization of FBF proteins in csn-5(lf) mutant background suggesting that csn-5 promotes FBF accumulation through protecting them from degradation. FBF protein levels in the csn-5(lf) genetic background were rescued by disrupting proteasome-mediated degradation without affecting fbf transcript levels (Fig. 7), suggesting a posttranslational regulatory role of CSN-5 counteracting proteasomal-mediated degradation of FBFs. Since proteasome inactivation causes numerous germline defects (Fernando et al. 2022 and E.O., unpublished results for pbs-6(RNAi)), it is unclear whether the rescue of FBF levels following pbs-6(RNAi) is sufficient to alleviate csn-5(lf) phenotype. Interestingly, we find that FBF-2, but not FBF-1, levels are partially rescued in the csn-5(lf) genetic background with ned-8(RNAi) (Fig. 8a–f). This suggests that CSN-5 likely regulates FBF-1 in a COP9-independent manner, whereas regulation of FBF-2 might involve a combination of COP9-dependent and COP9-independent mechanisms. These distinct mechanisms of FBF-1/-2 regulation by CSN-5 are also supported by the observation that FBF-1 levels were most drastically reduced in csn-5(lf) whereas FBF-2 levels were equally reduced among all mutants (Fig. 4). As cullins are the only known targets of COP9-mediated deneddylation (Qin et al. 2020), it is not surprising we have not observed an accumulation of neddylated FBF-2 by Western blot analysis of csn(lf) mutant lysates.

CSN-5 stabilizes its partners by several mechanisms

CSN-5/CSN5 interacts with a number of proteins beyond components of the COP9 signalosome (for example, Tomoda et al. 1999; Bae et al. 2002; Smith et al. 2002; Yoshida et al. 2013) and promotes the accumulation of some of its interacting partners (Bemis et al. 2004; Orsborn et al. 2007). Our work defines a novel class of CSN-5 interactors, the PUF family FBF proteins that are stabilized by CSN-5 thus revealing a new mechanism of CSN-5 contribution to SPC maintenance in C. elegans. Based on our detection of binding between the human homologs, we speculate that CSN5 might similarly stabilize PUM1. Both this work and previous reports suggest that CSN-5 may stabilize its partners through several mechanisms. One mechanism documented in both fission yeast and mammalian cells is the recruitment of a deubiquitinating enzyme (DUB) Ubp12p/USP15 to the COP9 signalosome substrates (Zhou et al. 2003; Hetfeld et al. 2005). Importantly, in several cases, DUB recruitment was shown to depend on CSN5 (Groisman et al. 2003; Wee et al. 2005; Liu et al. 2009). If FBFs are substrates of CSN-5-associated DUBs, a transient in vivo association between FBF-1 and CSN-5 would not be surprising. Similarly, recruitment of DUBs would be consistent with the ability of catalytically dead CSN-5(D152N) to rescue FBF function. Alternatively, CSN5 was reported to directly interact with HIF1α and to stabilize HIF1α in a COP9-independent manner (Bemis et al. 2004). Stabilization of HIF1α was mediated by preventing proline hydroxylation of HIF1α and by interfering with the association of hydroxylated HIF1α with pVHL ubiquitin ligase (Bemis et al. 2004). As a third option, CSN5 metalloprotease was proposed to directly deubiquitinate its binding partner PD-L1 (Lim et al. 2016; Liu et al. 2020). Overall, CSN-5 might be able to protect FBFs from ubiquitination and thus promote their stabilization in C. elegans germline SPCs through either of these mechanisms.

CSN-5 promotes oogenesis independent of COP9 signalosome

In this study, we found that csn-5 contributes to oogenesis through the maintenance of FBF protein levels. The failure of oogenesis in csn-5(lf) and csn-6(lf) correlated with a stronger depletion of FBF-1 protein compared to csn-2(lf) mutant (Fig. 6a and b). By contrast, oocytes formed in the majority of csn-2(lf) mutant germlines (Fig. 6aii and b), and this was dependent on residual FBF-1 as oogenesis failure in csn-2(lf); fbf-1(lf) increased to levels seen in csn-5(lf) (Fig. 6b). The function of csn-5 in gametogenesis appears to be cell autonomous as the 3xflag::csn-5 transgene driven by the gld-1 promoter (Ellenbecker et al. 2019) and expected to only be expressed in the germline rescued oocyte formation to 100%, similar to the 3xflag::csn-5 transgene under the control of csn-5 promoter (Table 1; Supplementary Fig. 7b).

Does the mutation of csn-5 compromise other subunits of the COP9 complex? Previously published work in human cells, Drosophila larvae, and with recombinant proteins suggests that the loss of CSN5 leaves the remainder of COP9 subunits largely intact and assembled, except unable to carry out deneddylation of cullin substrates (Oron et al. 2002; Groisman et al. 2003; Yun et al. 2004; Peth et al. 2007; Sharon et al. 2009). Therefore, we interpret csn-5(lf) phenotypes as reflecting the loss of CSN-5 protein and deneddylation, but not affecting the other COP9 subunits. By contrast, loss of several other COP9 subunits results in downregulation of CSN5, as observed for knockdowns of human CSN1 and CSN3 (Peth et al. 2007) as well as for knockdown or mutation of C. elegans csn-6 (Miller et al. 2009; Supplementary Fig. 6b). Therefore, care must be taken to distinguish whether any phenotypes of csn-6(lf) result from the secondary loss of CSN-5. We approached this by introducing an additional copy of csn-5 into csn-6(lf) genetic background, which was sufficient to rescue oogenesis (Fig. 6e). This suggests that defective oogenesis in csn-6(lf) results from the secondary loss of CSN-5. Additionally, this result argued that CSN-5 is sufficient to promote oogenesis in the absence of CSN-5/CSN-6 heterodimer, since csn-6(lf) is a null allele where maternal protein stockpile is expected to be depleted (Brockway et al. 2014). Overall, we conclude that CSN-5 is able to promote oogenesis outside of COP9 holoenzyme.

Finally, several lines of evidence suggest that CSN-5 can promote oogenesis independent of its deneddylating activity. First, csn-2(lf) is expected to disrupt COP9-dependent deneddylation, yet it still allows oogenesis in ∼70% of mutant germlines. Second, CSN6 is critically required for COP9-dependent deneddylation (Birol et al. 2014), yet 3xflag::csn-5 rescues oogenesis in csn-6(lf) genetic background. Third, csn-5 can support considerable oogenesis independent of its deneddylating activity as we observe a rescue of fertility to ∼40% by the catalytically inactive, germline-specific transgene, 3xflag::csn-5(D152N); csn-5(lf) (Table 1; Supplementary Fig. 7c). An observation that appears in conflict with this conclusion is the 100% rescue of oogenesis by ned-8(RNAi) in csn-5(lf) background (Fig. 8h). However, ned-8(RNAi) did not affect FBF-1 levels and rescued peak FBF-2 amounts only in ∼63% germlines (Fig. 8a and b), suggesting that the rescue of oogenesis is independent of FBF levels in at least 37% of germlines. This FBF-independent oogenesis might be explained by a reduction in the activity of cullin-based E3 ligase CUL-2/FEM-1/FEM-2/FEM-3 resulting from the downregulation of neddylation via ned-8 knockdown. Neddylation is required for the function of all cullin-based E3 ligases (Duda et al. 2008; Saha and Deshaies 2008; Boh et al. 2011), and CUL-2/FEM-1/FEM-2/FEM-3 is required for spermatogenesis to degrade TRA-1 (Starostina et al. 2007). The CUL-2/FEM-1/FEM-2/FEM-3 complex functions downstream of FBFs in the germline sex determination pathway (Ellis 2022) and fem-3 is epistatic to fbfs (Zhang et al. 1997). Therefore, it is possible that ned-8(RNAi) indirectly circumvents the requirement for csn-5 in oogenesis and that the contribution of COP9-mediated deneddylation to germline sex determination is through maintaining the balance of neddylated/deneddylated CUL-2/FEM-1/FEM-2/FEM-3.

CSN-5 contributes to SPC maintenance by safeguarding FBFs

We find that both CSN-5 and COP9 signalosomes are required for germline SPC maintenance in C. elegans and identify FBFs as a specific target of CSN-5 regulation relevant for this process. COP9 function in supporting germline stem cells has been reported in Drosophila ovary (Pan et al. 2014) and in C. elegans germline (Brockway et al. 2014), and our current results provide a mechanistic explanation for the prior observations. We observed a significant loss of SPCs in csn(lf), with the most drastic decrease for csn-5/-6(lf) (Fig. 5a and b), which correlated to the most severe decrease in FBF-1 and FBF-2 (Fig. 4a–c). Although this phenotype is consistent with fbf(lf), it could be caused by csn(lf) affecting other stem cell regulators. However, we found that csn-5 functions within the same genetic pathway as fbfs when we performed fbf knockdown in csn-5(lf) mutant and found no further reduction in the number of SPCs (Supplementary Fig. 5a). Additionally, csn-5(lf) did not reduce the levels of PUF-3 (Fig. 4l–n), further supporting the notion that it affects SPCs through the regulation of FBF levels. One limitation of this study is that we were not able to test whether the stabilization of FBFs is sufficient to suppress the effects of csn-5(lf) due to pleiotropic defects resulting from proteasome disruption. Nevertheless, we find that a partial rescue of FBF-2 levels in csn-5(lf) by ned-8(RNAi) leads to a partial recovery of SPC numbers (Fig. 8b and g), supporting our overall conclusions. In the future, the identification of specific ubiquitin ligases controlling FBF stability would provide additional tools to address this question.

It is still unclear whether CSN-5's function in SPC maintenance depends on the presence of other COP9 subunits as we were unable to rescue SPC numbers in csn-6(lf) mutant by providing an extra copy of 3xflag::csn-5. This result might be due to the insufficient levels of the transgenic protein (Supplementary Fig. 6b). Alternatively, it remains possible that CSN-5 needs to form a heterodimer with CSN-6 to promote SPC maintenance. We find a partial rescue of SPC numbers in csn-5(lf) ned-8(RNAi), which is consistent with a partial rescue of FBF-2 function, along with a partial rescue of FBF-2 protein levels (Fig. 8b and g). Therefore, we conclude that deneddylation activity of CSN-5 contributes to SPC maintenance, possibly through the regulation of FBF-2 levels. We considered whether csn-5(lf) ned-8(RNAi) produced a germline state reminiscent of fbf-1(lf) since we observed an increase in FBF-2 protein without a rescue of FBF-1. fbf-1(lf) mutation results in a smaller progenitor zone compared to the wild-type germlines (Lamont et al. 2004). However, the number of SPCs in csn-5(lf) ned-8(RNAi) (∼0.62× of the wild type on average; Fig. 8g) is smaller than that of fbf-1(lf) mutant (∼0.85× of the wild type; Wang et al. 2020), suggesting that the partial rescue of FBF-2 levels is not sufficient to recapitulate its full functionality.

Several previously characterized mutants of splicing factors exhibit a combination of defective female fate specification as well as underproliferated germlines, reminiscent of those observed in csn-5 and csn-6 mutants (Puoti and Kimble 1999, 2000; Belfiore et al. 2004; Konishi et al. 2008; Mantina et al. 2009; Zanetti et al. 2011; Wang et al. 2012). This raised the question of whether underproliferation observed in csn-5 and -6(lf) was linked to the failure of oogenesis. However, we could not rescue a number of SPCs (E.O., unpublished results) with 3xflag::csn-5; csn-6(ok1604) despite being able to rescue oogenesis (Fig. 6e), suggesting that the reduction in SPCs in csn-6(lf) does not result from a defective oogenesis. Additionally, male germlines maintain a similar SPC population to the hermaphrodite germlines, suggesting no intrinsic sex-specific differences in SPC numbers (Morgan et al. 2010).

The SPCs of csn-5/-6(lf) germlines continue to proliferate and enter meiosis (Fig. 5c–g) albeit slower than the control germlines. Slower proliferation might result from the direct impact of csn(lf) on cell cycle machinery since homologs of COP9 complex subunits have previously been found regulating proteins involved in cell cycle progression (Tomoda et al. 1999; Yang et al. 2002; Doronkin et al. 2003; Yoshida et al. 2013). At this time, it remains unknown if the balance between proliferation and differentiation is disrupted in these csn(lf) mutants, and further research would be required to understand their cell cycle dynamics. In summary, we conclude that csn-5 promotes C. elegans SPC maintenance through the regulation of FBF protein accumulation.

We speculate that the regulatory relationship between CSN-5 and PUF family proteins is conserved and has implications for understanding mechanisms of stem cell maintenance in other biological contexts. Intrinsic instability of PUF family proteins counteracted by CSN-5 might reflect a general mechanism regulating the abundance of essential stem cell factors and their effects on stem cell population dynamics. Additionally, the function of CSN-5 in FBF stabilization may be relevant for understanding cancer as CSN5 is upregulated in many human cancers (Lee et al. 2011; Liu et al. 2020). Similarly, a number of cancers overexpress PUM1 and/or PUM2 (Guan et al. 2018; Gor et al. 2021; Shi et al. 2021; Smialek et al. 2021). This PUM protein overexpression in cancer might be borne out through its stabilization via CSN5 and therefore reflect an unappreciated mechanism promoting cancer cell proliferation.

Data availability

This study did not generate large data sets or codes, but raw data/images are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables. All alleles and strains (Supplementary Table 1) are available upon request. Antibody information is provided in Supplementary Table 2, and qPCR primer sequences are in Supplementary Table 3. Supplementary Table 4 lists candidate FBF-2 interactors identified in the yeast 2-hybrid screen.

Supplemental material available at GENETICS online.

Funding

This work was supported by National Institute of Health grants R01GM109053 to EV, P20GM103546 (S. Sprang, PI; E.V. Pilot Project PI), and P20GM103474 (B. Bothner, PI; E.V. infrastructure support awardee), and Toelle-Bekken Family Memorial Fund Award to EO. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.