Mutation of C. elegans Integrator complex subunit 6 causes snRNA misprocessing and embryonic lethality

In a screen for embryonic lethal mutations in C. elegans, we noticed a striking set of developmental and transcriptional defects in worms and embryos homozygous for the t1903 mutation. t1903 was a thermosensitive allele of dic-1/INTS6, hereafter referred to as ints-6 owing to its homology to the human subunit 6 of the Integrator complex INTS6 gene. ints-6 encodes a protein of 869 amino acids that is highly conserved throughout the animal kingdom. The human homolog has been characterized as a member of the Integrator complex involved in the 3’-end processing of nascent snRNAs [7]. At the molecular level, INTS-6 is predicted to have a von Willebrand factor type A (vWA) domain at the N-terminal part that might serve as a surface for interaction with other proteins or molecules with which it forms complexes [35]. At its C-terminal end it features a COIL domain, which is a structural motif that in many proteins plays a fundamental role in subcellular infrastructure as a molecular ruler, positioning catalytic activities at fixed distances [36].

The t1903 mutation was a C to T substitution at position 3944 in the F08B4.1 gene that resulted in a swap from Ser to Phe in aa 850 of the protein (Fig 1A). At the permissive temperature (15°C), the worms were viable but exhibited some embryonic lethality (21.5%, n = 2153) and reduced offspring (215±23.5 descendats vs 262±20.2 of a WT, mean±sem, n = 10), whereas a shift to the restrictive temperature (25°C) resulted in full embryonic lethality (100%, n = 818) and reduced offspring (82±3.6 descendats vs 185±8.4 of a WT, mean±sem, n = 10) (S1 Fig). A knockout deletion in ints-6 (tm1615) (Fig 1A) [37] has a zygotic effect and homozygotes arrest at the L3 larval stage [38]. To determine the developmental consequences of ints-6 disruption, we performed four-dimensional (4D) microscope studies on embryos [39]. Analyses of thermosensitive ints-6 (t1903) mutant embryos recorded at the restrictive temperature showed morphogenesis defects leading to embryo death (Fig 1B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Characterization of ints-6 (t1903) mutant.

(a) Schematic representation of WT ints-6 gene compared to the knockout tm1615 and the thermo-sensitive t1903 mutant (b) DIC microscopy images from the two-cell stage show an embryonic lethal phenotype of ints-6 (t1903) whereas WT develops till hatched larvae. Scale bar: 10μm. (c) in vivo INTS-6 localization in the strain JCP378 (jcpSi19 [pJC56 (eft-3p::ints-6::3xFLAG::eGFP::ints-6UTR, unc-119(+))]II; unc-119(ed3)III), compared to a N2 WT embryo for autofluorescence (upper two panels). Lower three panels show GFP expression in the head, gonad and tail of the transgenic animals. Arrows indicate a nuclear localization of INTS-6. Scale bar: 20μm. (d) RNA deep sequencing reads aligned to the C. elegans genome in the region of U2 snRNA gene: W04G5.11, visualized on IGV software for each deep sequence technique, RibominusRNA-seq in or Poly(A)RNA-seq. N2 reads are shown in gray whereas ints-6 (t1903) mutant reads are in black. Underneath each graph, the C. elegans genome is represented in blue. The exons are shown as blue boxes and the introns as lines.

https://doi.org/10.1371/journal.pgen.1007981.g001

To further address the function of INTS-6, we generated transgenic animals expressing green fluorescent protein-tagged INTS-6 (INTS-6::3xFLAG::eGFP) under the control of its own ints-6 promoter and the eft-3 promoter (S2 Fig). Transgenes were assayed both as multicopy extrachromosomal arrays and single copy mosSCI integrated lines. Additional ints-6::3xFLAG tagged strains were generated by CRISPR [40,41,42]. Consistent with the localization and function of its human homolog [7,43], the protein appeared as a predominantly nuclear protein (Figs 1C, S3 and S4). Gene expression assessed by GFP detection and by anti-FLAG immunostaining was detected in all cells from embryos to adults. There was, however, a difference between soma and germline expression detected by immunofluorescence: whereas in somatic cells and oocytes INTS-6 was detected in the nucleus, in the embryonic germline it was detected both in the nucleus as well as in cytoplasmic granules (Figs 1C and S4).

To ascertain the role of ints-6 in 3’-end processing of the snRNAs, as suggested by its homology to the human Integrator complex subunit 6 and its nuclear localization, we performed deep-sequencing of total RNA obtained from ints-6 (t1903) mutant worms. Processing of the 3’ end of snRNAs was already defective at 15°C (permissive temperature) whereas mRNA termination appeared unaffected. Shifting of the ints-6 (t1903) mutant to 25°C (restrictive temperature) for 12h resulted in an increase in the amount of long transcripts derived from unprocessed snRNAs (Figs 1D and S5). 3’-end misprocessing was not restricted to U1, U2, U4 and U5 snRNA [7,29] but also affected SL1 and SL2 snRNAs families and certain small nucleolar RNAs (snoRNAs), but not other types of non-coding RNAs (ncRNAs) such as ribosomal RNAs (rRNAs), transfer RNAs (tRNAs) or lncRNAs (S6 Fig, S1 Table).

These results, homology to human subunit 6 of the Integrator complex, nuclear localization and its role in processing snRNAs, strongly suggest that F08B4.1/ints-6 is indeed a part of the C. elegans Integrator complex and prompted us to study the function of this complex in transcription and RNA metabolism during in vivo development of a complex organism.

Identification of the C. elegans Integrator complex

To define the polypeptide composition of the C. elegans Integrator complex, we focused on three defining features: its homology to Integrator subunits in other metazoans, the physical association of its subunits in a complex and the lack of 3’-end processing of snRNAs after disruption of the Integrator complex coding genes.

A search in the GeneBank database, using the BLAST algorithm, identified protein homologs for the various Integrator subunits in different species (Homo sapiens, Mus musculus, Gallus gallus, Danio rerio, Drosophila melanogaster and C. elegans). Virtually all Integrator subunits except INTS-14 in C. elegans are conserved throughout evolution (Fig 2A, S2 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. C. elegans Integrator complex identification.

(a) Predicted C. elegans Integrator complex orthologs versus H. sapiens Integrator complex subunits. Protein domains were searched using the Pfam database. The length of each subunit is indicated in amino acids (aa). Abbreviations: DUF, domain of unknown function; HEAT, Huntingtin, Elongation factor 3, protein phosphatase 2A, and the yeast kinase TOR1; ARM, armadillo-like repeats; COIL, coiled coil domain; VWA, von Willebrand type A domain. β-lactamase/β-CASP (*indicates the presence of an inactive β- lactamase/β-CASP domain); RMMBL, Zn-dependent metallo-hydrolase, RNA specificity domain; 7TM GPCR srh: seven-transmembrane G-protein-coupled receptor, serpentine receptor class h; PHD, plant homeodomain finger; ISDCC, INTS6/SAGE1/DDX26B/CT45 C- terminus; TPR, tetratricopeptide repeats. (b) INTS-6 Immunoprecipitation scheme. Protein extracts were obtained from the JCP378 strain. The majority of the immunoprecipitate was analysed by LC-MS/MS. In addition, a small amount of the immunoprecipitate was run on SDS-PAGE and silver stained. The Integrator complex members detected by mass spectrometry are indicated on the silver stained gel according to their expected molecular weight. (c) Polypeptides detected from the Integrator complex predicted the orthologs in the four independent INTS-6 co-immunoprecipitation analyses.

https://doi.org/10.1371/journal.pgen.1007981.g002

Immunoprecipitation followed by mass spectrometry analyses (IP/MS) further confirmed that the proteins identified as C. elegans Integrator homologs are components of a multiprotein complex. To investigate the polypeptide composition of the C. elegans Integrator complex, we generated integrated transgenic worms expressing ints-6::3xFLAG::eGFP and purified INTS-6-associated proteins by anti-FLAG affinity purification (Figs 2B and S7). WT N2 animals were used as a control for nonspecific binding. The FLAG affinity eluate was separated on a polyacrylamide gel and stained with Coomassie blue dye. Bands of different molecular weight were excised from the gel and subjected to mass spectrometry analysis. This resulted in the identification of INTS-1 (C06A5.1), INTS-2 (ZC376.6), INTS-3 (Y92H12A.4), INTS-4 (W04A4.5), INTS-5 (Y51A2D.7), INTS-6 (F08B4.1), INTS-7 (D1043.1), INTS-8 (Y48G10A.4), INTS-9 (F19F10.12), INTS-11 (F10B5.8) and INTS-13 (R02D3.4) as components of an INTS-6-associated complex. INTS-10 (F47C12.3) and INTS-12 (T23B12.1) subunits were not detected in the mass spectrometry analysis, probably because they are the smallest subunits with molecular weights of 38.7 kDa and 25.9 kDa respectively (Fig 2).

To evaluate the functional contribution of the INTS proteins identified as subunits of the Integrator complex, we assessed 3’-end snRNA processing after RNA interference to knockdown each of them (S8 Fig). 3’-end snRNA processing was determined by RNA deep-sequencing, northern blot and retrotranscription, followed by PCR (RT-PCR) of regions downstream of the snRNA loci (Figs 3A, 4 and S9). U1, U2, U4, U5, SL1, SL2 snRNAs and certain snoRNAs revealed a significant accumulation of long transcripts beyond their 3’ end after depletion of the Integrator subunits INTS-1, -2, -4, -5, -6, -7, -8, -9, -11, whereas other types of ncRNAs, such as rRNAs, tRNAs or lncRNAs remained properly processed at their 3’ end (S1 Table). This indicates a lack of the 3’-end processing of the nascent snRNA transcripts. RNAi knockdown of the Integrator subunits phenocopied the snRNA processing defects observed in the ints-6 (t1903) mutant (Figs 1 and 3) confirming loss of activity of the complex. RNAP II read-through downstream of the snRNAs ranged between 1% and 6% of the total amount of U1 and U2 snRNAs. snRNA expression did not significantly change after depletion of the Integrator subunits and remained as high as in WT (S10 Fig). As a result, read-through transcription downstream of the snRNA loci reached the expression level of regulatory genes such as lit-1/NLK or daf-16/FOXO of the wnt and insulin pathways respectively (S11 Fig). Depletion of INTS-3, -10, -12 and -13 led only to a slight snRNA misprocessing (Figs 3A and 4). In all cases, mature snRNA transcripts remained at a high level after Integrator depletion (Figs 3 and S9). This is consistent with the reported long half-life of RNAP II-transcribed snRNAs [44]. In addition, stable Integrator complex could retain some activity after RNAi knockdown of single subunits. This explains why mRNA of coding genes is mostly properly spliced, with little intron retention, and polyadenylated after RNAi knockdown of any Integrator complex subunit, as assessed by RNA deep-sequencing and retrotranscription followed by PCR (RT-PCR) of intron-containing regions of coding mRNAs. Interestingly, splicing defects are prominent in the genes located directly downstream of the snRNA loci (15–18% of intron retention for ints-1, -8, -9, -11 subunit knockdown), whereas genes not affected by RNAP II read-through are only slightly affected (Figs 3A, 4 and S12).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. RNAi analysis of the C. elegans Integrator complex.

(a) RNA deep sequencing reads aligned to the C. elegans genome in the region of the U1 snRNA: H27M09.8 and the downstream in sense gene H27M09.5 visualized using IGV software for each RNAi Integrator member knockdown (logarithmic scale). L4440 control reads are shown in gray whereas Integrator complex RNAi reads are in black. Underneath each graph, the C. elegans genome is represented in blue. The exons are shown as blue boxes and the introns as lines. (b) Phenotypes of C. elegans Integrator complex predicted subunits after RNAi knockdown. L1 stage N2 worms were fed bacterial RNAi clones of each predicted member of the Integrator complex. Worms were grown at 15°C. Images were taken on the sixth day (adult stage). ints-2, -4, -5, -9, -11 show a larval arrest phenotype (lva), ints-1, -6, -7, -8 reached adulthood but had reduced offspring that died embryonically (emb) whereas ints-3, -10, -12, -13 showed non-obvious phenotypes.

https://doi.org/10.1371/journal.pgen.1007981.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Knockdown of C. elegans integrator subunits results in formation of long “chimeric sn-mRNAs” that can be translated to proteins in the case of “chimeric SL-mRNAs”.

Left panel: schematic representation of the snRNA genes analysed. Red arrows represent the primer pairs used. Middle panel: gel electrophoresis showing RT-PCR analysis. RT minus controls do not show genomic DNA contamination. Double bands correspond to the spliced (lower band) and unspliced (upper band) mRNA fractions. Right panel: Western blot to check the translation of “chimeric sn-mRNAs” into proteins. (a) U1 type sn-mRNAs. a.1 PCR expected length: 396 bp (genomic: 439 bp). H27M09.5 estimated molecular weight: 68.8kDa. a.2 PCR expected length: 637 bp. F08H9.3 estimated molecular weight: 46.6 kDa. The U1 derived sn-mRNAs were not translated into proteins (b) U2 type snRNAs. b.1 PCR expected length: 782 bp (genomic: 840 bp). W04G5.8 estimated molecular weight: 72.7 kDa b.2 PCR expected length: 804 bp (genomic: 856 bp) INS-37 estimated molecular weight: 45.6 kDa. The U2 derived sn-mRNAs were not translated into proteins (c) SL-2 snRNA. PCR expected length: 358 bp (genomic 410bp). Y75B8A.23 estimated molecular weight: 39.5 kDa. The SL derived sn-mRNAs were translated into proteins. (d) Actin was used as the loading control: 41.8 kDa. PCR expected length: 156 bp (genomic: 208 bp).

https://doi.org/10.1371/journal.pgen.1007981.g004

Since the production of ectopic RNAs may have deleterious consequences, we extended the characterization of the phenotypic consequences of Integrator disruption to the level of the whole organism. RNAi of some Integrator subunits exhibited phenotypes ranging from severe larval arrest (ints-2, -4, -5, -9, -11) to reaching adulthood but having reduced offspring and subsequent embryonic lethality (ints-1, -6, -7, -8), whereas the subunits that resulted in only slight snRNA processing defects (ints-3, -10, -12, -13), showed non-obvious phenotypes (Fig 3B). These phenotypic differences correlate with the amount of long transcripts detected beyond the 3’ end of the snRNAs after knockdown of the Integrator complex subunits (Fig 3A).

In summary, these findings indicate that the C. elegans Integrator complex is likely comprised of at least 13 subunits, INTS-1 to INTS-13, involved in the 3’-end processing of snRNAs.

Integrator disruption causes transcription of genes located downstream of the snRNA loci

To understand the consequences of the lack of 3’-end processing of nascent snRNAs in the organism, we first studied the structure of the uncleaved snRNAs formed upon depletion of any member of the Integrator complex. The structure of the long unprocessed transcripts was assessed by deep-sequencing of total RNA obtained from ints-6 (t1903) and Integrator complex depleted worms (Figs 1D and 3A). This result was further confirmed by northern blot and RT-PCR using specific primers for snRNAs and their downstream genes (Figs 4 and S9). snRNA loci are present in multiple copies within the genome, either in intergenic regions or within coding genes and are oriented either in sense or antisense to the downstream gene (S1 Table).

For snRNA loci located in sense in the 5’ region upstream coding genes, the lack of snRNA processing resulted in the formation of long chimeric sn-mRNAs containing the snRNA sequence at the 5' end, followed by the sequence corresponding to the region between the snRNA and the gene downstream and the mature mRNA of the gene on the 3' end (Figs 1D, 3A and 4). In all these cases, intron retention was detected but most of these transcripts were processed by splicing and polyadenylated at the expected sites (S12 Fig). Thus, depletion of the Integrator complex caused upregulation of genes located downstream of the snRNA loci.

In contrast, for snRNA loci located downstream and opposite to coding genes, the lack of 3’-end processing resulted in the transcription of cis-antisense RNAs of the coding genes. Directional deep-sequencing of total RNA from WT and ints-6 (t1903) mutant worms revealed the existence of both types of transcripts: mRNA (in sense) derived from the endogenous promoter activity and antisense RNAs on the opposite strand, derived from the lack of processing of snRNAs located in antisense downstream of the gene. This suggests that these antisense RNAs might not be efficient at performing RNA silencing (S13 Fig). This is consistent with the fact that only double-stranded RNA has been shown to be substantially effective at producing RNA interference. Indeed, injection of purified single strand RNA has at most a modest effect on gene expression [45]. These results reveal the formation of antisense RNAs upon knockdown of the Integrator complex, although its putative function remains an open question.

Specific long chimeric sn-mRNAs generated upon Integrator disruption are translated into proteins

To establish whether long chimeric sn-mRNAs could be translated to proteins, we generated transgenic worms containing genes that had in sense snRNAs in their 5’ upstream region (from 1186 to 213 bp), tagged with FLAG and/or eGFP at the 3’-end of the coding gene in its genomic sequence. The 5 transgenes assayed contained two U1 snRNA genes (H27M09.8 and F08H9.10) in the 5’ upstream region of the tagged genes H27M09.5 and F08H9.3 respectively; two U2 snRNA genes (W04G5.11 and F08G2.9) in the respective 5’ region of the tagged W04G5.8 gene and ins-37, an insulin-like peptide; and finally, an SL-2 (sls-2.8) in the 5’ region of the tagged Y75B8A.23 gene. A single copy of these transgenes was integrated into the C. elegans genome by mosSCI [41]. Transgenic worms were treated with RNAi of each member of the Integrator complex or crossed with the ints-6 (t1903) mutant. Transgene expression was detected by RT-PCR and protein formation was assessed by western blot (Fig 4).

For the transgenes assayed, we concluded that the lack of 3’-end processing of sls-2.8 snRNA upon depletion of the Integrator complex resulted in transcription of a long chimeric RNA containing sls-2, the intergenic region and the downstream gene Y75B8A.23 mostly spliced and polyadenylated. This chimeric RNA was translated into a protein detected by western blot of the FLAG tag at its C-terminal end (Fig 4). Neither the sls-2 nor the intergenic region contained any ATGs. The first translation start codon that could be used in this long chimeric RNA was the initial ATG of the Y75B8A.23 gene. Y75B8A.23 is a hitherto uncharacterized nematode-specific gene that is highly expressed during spermatogenesis at late larval stages [46].

Lack of 3’-end processing of U1 and U2 snRNAs in the transgenes assayed, caused by RNAi of any Integrator subunit, led to transcription of long chimeric sn-mRNAs. However, these U1- and U2-derived chimeric RNAs were not translated into proteins (Fig 4). We observed that U1 and U2 snRNA genes contain several ATGs in their sequence (S3 Table) as well as a specific secondary structure [47]. To determine whether initial ATGs in these long chimeric sn-mRNAs could serve as start codons for translating peptides, we generated transgenes that contained HA and MYC tags in-frame with the 1^st and 2^nd ATG of the U1 snRNA gene (F08H9.10) and the U2 snRNA gene (W04G5.11) respectively. Depletion of the Integrator complex by RNAi did not result in peptide formation as determined by western blot using anti-HA and anti-MYC specific antibodies (S14 Fig).

These findings indicate that specific SL-derived, but not U1- or U2-derived, long chimeric sn-mRNAs generated upon Integrator complex downregulation may be translated into proteins.

Integrator disruption leads to a dramatic alteration of the transcriptomic profile

Downregulation of the Integrator complex has a direct effect on the transcription of genes located downstream of the snRNA loci. To decipher whether gene expression alteration is restricted to those genes or has a broad effect on the general expression profile, we analyzed the long-term transcriptomic profile of nematodes depleted for each member of the Integrator complex by RNAi.

We examined the gene expression profile of WT N2 worms synchronized at the first larval stage (L1) and grown on RNAi feeding plates for each member of the Integrator complex, for 6 days at 15°C. The gene expression data were normalized by a negative binomial distribution model using DESeq2 and EdgeR implementations and compared to a control grown under the same conditions using the empty L4440 vector as the RNAi clone. Three biological replicas of each analysis were performed. Raw sequence data generated in this study are available in the Gene Expression Omnibus (GEO) data repository (Accession number GSE111083). Quantitative analysis of differential expression was performed as described in materials and methods [48].

The Euclidean distance analysis of the global expression profile similarity of the three biological replicas of each Integrator subunit knock-down, plus the control, defined three groups (S15 Fig). These transcriptional groups broadly matched the different phenotypes observed for the depletion of each of the Integrator subunits with punctual replica exceptions (Figs 3, 5A–5B and S15). The transcriptomic profile of the RNAi-depleted worms for the catalytic unit homologs, ints-9 and ints-11, grouped together with ints-4 and ints-5. The highly uniform phenotypic and transcriptional response to the absence of any of these subunits suggests that they are functionally related and defines a category within the Integrator complex hereafter referred to as the Catalytic Class. Integrator subunits exhibiting this Catalytic Class transcriptional signature showed the strongest phenotype after depletion of any Integrator subunits. RNAi of ints-4, ints-5, ints-9 or ints-11 led to a strong lack of 3’-end snRNA processing and to larval arrest of the fed worms (Fig 3). Secondly, the transcriptomic profile of ints-1, ints-2, ints-6, ints-7, and ints-8 RNAi shared some common features with the Catalytic Class. However, their short Euclidean distance in the global expression profile similarity analysis grouped them together in a second transcriptional phenocluster that we hereafter refer to as the Holder Class. RNAi of any Holder Class subunits led to a clear lack of 3’-end snRNA processing. ints-2 RNAi fed worms arrested as larvae, whereas the rest of the Holder Class subunits (ints-1, ints-6, ints-7) reached adulthood but produced dead embryos in the next generation (Fig 3). Finally, far from the Catalytic and Holder Classes, the transcriptomic profiles of ints-3, ints-10-, ints-12-, ints-13-depleted worms did not show significant differences from the control and grouped with WT N2 worms fed the RNAi of the empty L4440 vector. This third transcriptional phenocluster, that we named the Auxiliary Class, makes only a mild contribution to 3’-end snRNA processing under the assayed conditions (Fig 4). In addition, Auxiliary Class subunits RNAi did not show an obvious phenotype (Fig 3). This indicates that under standard laboratory conditions it does not play a central role in snRNA processing and its function, under these specific conditions, is accessory.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Transcriptomic analysis of the C. elegans Integrator complex.

(a) Principal Component Analysis (PCA) plot based on DESeq2 regularized log2 transformation (rld) data shows functional clusters according to the similarity of the transcriptional profile. (b) Simplified diagram of Integrator complex members classified as Auxiliary Class, Holder Class or Catalytic Class based on the complete heatmap of the expression profile similarity shown in S15 Fig. (c) Venn diagrams of the Holder Class and Catalytic Class showing up-regulated genes. The gene ontology (GO) analysis of upregulated genes’ molecular functions (MF) is shown on the right. GO of MF vs number of genes within each category is shown as color bars, one bar per GO term. Bar length indicates the number of genes belonging to the different GO categories and color illustrates the statistical significance, from those with highly significant expression differences (red) to those with low expression differences (blue). GO functions are highly enriched in kinases and phosphatases. (d) Venn diagrams of the Holder and Catalytic Classes showing down-regulated genes. GO analysis of downregulated genes’ MF is shown on the right. GO of MF vs number of genes within each category is shown as color bars, one bar per GO term. Bar length indicates the number of genes belonging to the different GO categories and color illustrates the statistical significance, from those with highly significant expression differences (red) to those with low expression difference (blue).

https://doi.org/10.1371/journal.pgen.1007981.g005

To gain insight into the transcriptional role of the three Integrator complex phenoclusters, rather than the biochemical organization of the complex, we identified the overlapping set of genes significantly up- and down-regulated after knockdown of the Integrator subunits within each class (S4 Table). As expected, genes located in sense downstream of the snRNA loci that are directly affected by the lack of snRNA processing were upregulated both in the Catalytic and Holder Class RNAi groups. However, these genes constituted only a small fraction of the total altered genes. The rest of the upregulated genes do not have any snRNA in their upstream region as visualized in the RNAseq experiments. Therefore their upregulation is not caused by read-through of the RNAP II downstream of the snRNA. (S1, S4 Table).

In the Catalytic Class, common genes upregulated by knockdown of any subunit constituted, by far, the largest group of upregulated genes (1508 genes). Upregulated genes specific for the knockdown of any single subunit of the Catalytic Class represented less than 10% of the common response of the class; except for ints-5 that represented near 50% (Figs 5C and 6) and were enriched in hydrolase, lyase and chitinase activity suggesting a high catabolism rate (S5 Table). This transcriptomic signature highlights a major common function of these subunits. Genes upregulated after depletion of the Integrator Catalytic Class do not randomly fall within different GO molecular function categories. Instead, they are highly enriched in kinases and phosphatases involved in biological processes such as regulation of cell shape, cell proliferation, morphogenesis or signaling pathways. Interestingly, we did not detect activation of stress response pathways (S4 Table). The activation of these pathways depends on ATR kinase that senses blocked transcription elongation rather than DNA lesions directly. In fact, transcriptional and post-transcriptional activation of the stress response occurs when transcription elongation is blocked even in the absence of DNA damage [49,50]. This result strongly indicates that knockdown of the Integrator complex does not abrogate gene transcription. In global terms, upon depletion of the Integrator Catalytic Class, 23% of the total 438 C. elegans kinases are significantly upregulated (Representation Factor RF = 2.7 p<6.612e-21). Similarly, 38% of the total 206 C. elegans phosphatases are significantly upregulated (RF = 4.4 p<1.99e-31) (wormbook.org, nemates.org) (Fig 5C). These transcriptomic changes either cause or reflect the dramatic alteration of the signaling state of the cells upon lack of snRNA processing. In contrast to this gene upregulation response, gene downregulation does not show such a clear, common pattern within the Catalytic Class. The set of common genes downregulated by knockdown of any subunit of the Catalytic Class (397 genes) is smaller than the number of genes specifically downregulated by knockdown of ints-5 (557 genes) or ints-11 (487 genes). This common downregulation mainly affects peptidases involved in metabolism (Figs 5D and 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Representation of the up- and downregulated sets of genes in Integrator Complex disruption.

This representation was obtained using the Upset web application (www.caleydo.org) and further supports their categorization in different transcriptional classes. Panels show the analysis of upregulated (left) and downregulated (right) genes after depletion of any Integrator subunit. Integrator subunits that significantly affect the expression of a set of genes are shown above. Queries are indicated below as groups of intersections that must (filled circle), may (dotted circle), or must not (empty circle) include a specific set. Histograms on the side represent the number of genes included in each query. Transcriptional catalytic class is highlighted in orange. (a) and (c) show an analysis of genes whose expression is affected by disruption of certain Integrator subunits but may also be affected by others. Therefore, they are present in specific groups of intersections but may also be present in others. (b) and (d) show a restrictive analysis of genes whose expression is specifically affected by disruption of certain Integrator subunits but not by others. Therefore, they are exclusively present in specific groups of intersections and not found in any other sets. Similar to the results shown in Fig 5, Holder and Catalytic classes (blue box) share an upregulated transcriptional signature. In addition to this set of upregulated genes, subunits grouped as Catalytic class (orange box) further show a closer transcriptional signature that affects the upregulation of a specific set of genes absent from the Holder class. In contrast to the level of shared upregulated genes, downregulation does not show such a clear transcriptional signature as indicated by the lower number of shared genes downregulated in the Holder and Catalytic classes. Among the individual Integrator subunits, ints-1 and ints-5 show a specific transcriptional signature, both in up- and down-regulated genes, indicating functions in the regulation of specific transcriptional programs. Access to the online data is described in Materials and Methods.

https://doi.org/10.1371/journal.pgen.1007981.g006

Regarding the Holder Class, the common genes upregulated by knocking down any subunit (473 genes) overlap in 90% with the Catalytic Class, indicating the involvement of this class in the major activity of the Integrator complex (Fig 5C). However, there is no significant common downregulation response to knockdown of the Holder Class. Among the members of this class, ints-1 shows a specific effect on the transcription of a subset of genes enriched in extracellular protein coding genes (Figs 5D and 6). Finally, RNAi depletion of Integrator subunits, grouped together as the Auxiliary Class, do not show significant differences from WT under laboratory conditions (S15 Fig).

In addition to the major functions of the Integrator classes, and specific subunits such as ints-1 and ints-5, certain pairs such as ints-1 and ints-6 share a fraction of up- and downregulated genes, suggesting a functional relationship between them (Figs 5C–5D and 6). Both sets are enriched in extracellular protein coding genes, indicating a specific function of these subunits in regulating extracellular matrix components. Moreover, although affecting different sets of genes, knockdown of Integrator subunits 1 or 11 causes downregulation of neuronal genes. Knockdown of Integrator subunits 6 and 9 causes downregulation of genes involved in mitochondrial activity. And finally, knockdown of Integrator subunits 7 and 8 causes downregulation of genes coding for gap junction structures (S5 Table). These data are available online for comparison and easy visualization of any of the multiple datasets by loading them onto the web version of the Upset application (http://caleydo.org/tools/upset/) (See Materials and Methods) (Fig 6).

Together, these data indicate that the Integrator complex has a major positive role in processing snRNAs and a negative function in the expression of genes involved in the regulation of signaling pathways by protein phospho-modification. Similar to what occurs in Drosophila and mammals, this effect likely reflects the direct role of the C. elegans Integrator complex on gene expression by regulating the RNAP II gene transcription cycle [8,31,32,33,34]. Additionally, formation of long, unprocessed sn-derived RNAs upon Integrator complex knockdown might have a cascade effect on the expression of other genes.

Worm strains and culture conditions

C. elegans strains were maintained on Nematode Growth Medium (NGM) agar plates seeded with a lawn of E. coli OP50. Bristol N2 was used as the WT strain. The nematodes were grown on these plates at 15°, 20° or 25°C, depending on the purpose of the experiment. The ints-6 (t1903) thermosensitive mutant was regularly grown at 15°C and shifted to 25°C 12h before the experiments, when required. The following strains were used:

JCP294 ints-6(t1903) IV

JCP301 jcpSi3 [pJC50 (ins-37p::ins-37::eGFP::ins-37UTR, unc-119(+))] II; unc-119(ed3) III

JCP341 jcpSi10 [pJC51 (ints-6p::ints-6::3xFLAG::eGFP::ints-6UTR, unc-119(+))] II; unc-119(ed3) III

JCP343 jcpSi12 [pJC55 (W04G5.8p::W04G5.8::eGFP::W04G5.8UTR, unc-119(+))] II; unc-119(ed3) III

JCP378 jcpSi19 [pJC56 (eft-3p::ints-6::3xFLAG::eGFP::ints-6UTR, unc- 119(+))]II; unc-119(ed3)III

JCP383 ints-6 (tm1615) IV; jcpSi10 [pJC51 (ints-6p::ints-6::3xFLAG::eGFP::ints-6UTR, unc-119(+))] II

JCP387 jcpSi24 [pJC57 (H27M09.5p::H27M09.5::3xFLAG::eGFP::H27M09.5UTR, unc-119(+))]II; unc-119(ed3)III

JCP394 jcpSi31 [pJC58 (Y75B8A.23p::Y75B8A.23::3xFLAG::eGFP::Y75B8A.23UTR, unc-119(+))]II; unc-119(ed3)III

JCP405 jcpSi37 [pJC60 (F08H9.3p::F08H9.3::3xFLAG::eGFP::F08H9.3UTR, unc-119(+)]II; unc-119(ed3)III

JCP462 ints-6(jcp1)[ints-6::3xFLAG]

JCP479 jcpSi53 [pJC63 (3-tags-in-3-frames (HA:MYC::TY) in the snRNA coding gene F08H9.10, unc-119(+))] II; unc-119(ed3) III

JCP504 jcpSi55 [pJC64 (3-tags-in-3-frames (HA::MYC::TY) in the snRNA coding gene W04G5.11, unc-119(+))] II; unc-119(ed3) III

JCP590 ints-2(jcp15)[ints-2::3xFLAG]

JCP614 ints-3(jcp8)[ints-3::3xFLAG]

JCP625 ints-9(jcp10)[ints-9::3xFLAG]

JCP626 ints-5(jcp21)[ints-5::3xFLAG]

JCP630 ints-7(jcp22)[ints-7::3xFLAG]

JCP643 ints-11(jcp31)[ints-11::3xFLAG]

JCP645 ints-13(jcp33)[ints-13::3xFLAG]

Differential interference contrast microscopy and fluorescence microscopy

Worms were monitored on NGM plates under a Leica dissecting microscope (MZ16FA). Gravid hermaphrodites were dissected. 2- to 4-cell stage embryos were mounted on 4% agar pads in water and sealed with Vaseline petroleum jelly. Imaging was performed at 25°C. Differential interference contrast microscopy (DIC) was performed on a motorized fluorescent Leica microscope (DM6000B) equipped with a Hamamatsu Orca-ER C10600 camera and fitted with DIC optics. The appropriate filters were selected for fluorescent microscopy. Images were captured using the open source Micro-manager software (www.micro-manager.org) and processed with XnView software and ImageJ software.

Embryo and worm immunostaining

Synchronized populations of C. elegans eggs, larvae or adults were freeze-cracked, fixed with -20°C cold methanol for 2 min and cold acetone for 4 min. Next, samples were dried at RT and recovered by adding a drop of PBS containing 0.1% Tween (PBST) for 5 min. Once eggs or worms were properly prepared, they were blocked for 15 to 30 min in 1% BSA PBST blocking solution. Subsequently, samples were incubated either O/N at 4°C or at RT using FLAG antibody in 1% BSA PBST (F1804 Sigma, 1:500 dilution) followed by 1 or 2 hours incubation with the secondary antibody Alexa Fluor 633 in 1% BSA PBST (Invitrogen A-21050, 1:500 dilution). Finally, samples were mounted using VECTASHIELD Antifade Mounting Medium (H-1000 Vector laboratories) with DAPI (1 μg/ml).

Immunocytochemistry

U2OS cells were cultured and incubated on pre-treated poly-L-Lysine coated coverslips (Sigma) to improve adhesion. Cells were washed with PBS/Ca²⁺Mg²⁺ (1 mM) and fixed with 4% paraformaldehyde in PBS/Na⁺K⁺ (1 mM) for 30 min at RT with gentle agitation. Next, two washes with PBS (1 mM) were performed for 5 min each and then cells were permeabilized with 0.1% Triton X-100/PBS for 10 min at RT with gentle shaking.

Next, cells were washed with 0.2% PBS-BSA for 10 min at RT. Subsequently, cells were incubated with Ints6 antibody (Bethyl Laboratories, 1:50 dilution) in 0.2% PBS-BSA for 1 h at RT in a humid chamber. Afterwards, cells were washed with PBS (3 times, 7 min each) and incubated with the secondary antibody (CyTM3-conjugated AffiniPure Goat Anti-Rabbit IgG, 115-165-003, Jackson ImmunoResearch Laboratories, 1:500 dilution) in 0.2% PBS-BSA for 30 min at RT in the dark. After this incubation, cells were washed with PBS (3 times, 5 min) and incubated with DAPI (2 μg/ml) in PBS for 5 min, all in the dark. Finally, cells were washed with Milli-Q water (3 times) and coverslips were mounted on the slides using the SlowFade Antifade kit (Invitrogen).

Immunoprecipitation

To immunoprecipitate INTS-6, and co-immunoprecipitate its interactors, protein extracts from the JCP378 strain (jcpSi19[pJC56(eft-3p::ints-6::3xFLAG::eGFP::ints-6UTR,unc-119(+))]II;unc-119(ed3)III) were used and extracts from N2 worms were the negative control. IPs/Co-IPs were performed with ANTI-FLAG M2 Magnetic Beads (Sigma).

Protein extracts were filtered through a 5.0 μm filter and subsequently through 0.45 μm filters to remove any remaining cell debris and particulates that could interfere with protein binding. In each IP/Co-IP reaction, 150 μl of the ANTI-FLAG M2 Magnetic Beads were incubated with 30 mg of protein extract.

Immediately afterwards, the beads were equilibrated with TBS buffer (375 μl per IP/Co-IP reaction: 50 mM Tris HCl, 150 mM NaCl, pH 7.4). This step was repeated, leaving the beads with a small amount of buffer. Then, the protein extract was incubated with the equilibrated beads for 3–4 h or O/N, always at 4°C in a rotating rack with gentle mixing. Once the binding step was complete, the beads were collected and the supernatants were removed, followed by the washing steps. The beads were washed with TBS buffer (1500 μl per IP/Co-IP reaction) three sequential times for 10 min on a rotating rack at 4°C to remove all non-specifically bound proteins.

INTS-6::3xFLAG::eGFP fusion protein and consequently its interacting proteins were eluted from the magnetic beads either by boiling samples in SDS-PAGE sample buffer (100 μl) or by competitive elution with 3xFLAG peptide (400 ng/μl 3xFLAG peptide in TBS) for 1 h at RT in a rotating rack with gentle mixing). Finally, eluates were precipitated using TCA.

Mass spectrometry

Eluted IPs were run on SDS-PAGE (Mini-PROTEAN™ TGX™ Precast Gels, Any kDa). Next, gels were stained with Coomassie Blue and bands were excised. The CIC bioGUNE proteomics platform (https://www.cicbiogune.es/org/plataformas/Proteomics) performed the proteomic analysis. Proteins were digested with Trypsin from each gel band and analyzed by LC-MS/MS: Liquid Chromatography-Mass Spectrometry/Mass Spectrometry.

Protein extraction

Worms (usually from 8 to 10 NGM plates) were harvested with M9 buffer and collected in 50 ml Falcon tubes. They were washed several times, allowing them to settle to the bottom between washes. After the last washing step, as much supernatant as possible was removed. Then, a double volume of lysis buffer (50 mM Tris HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100), containing 1x protease inhibitors (Complete EDTA-free Protease inhibitor, Roche) and 1x phosphatase inhibitors (PhosSTOP, Roche) was added to each worm pellet. Next, samples were ground in liquid nitrogen using a pre-chilled mortar and pestle. Ground worms were thawed on ice followed by centrifugation at 4°C (15000 rpm, 15 min) to eliminate any non-soluble tissue or cellular remains. The supernatants were transferred to fresh Eppendorf tubes.

SDS polyacrylamide gel electrophoresis (SDS-PAGE)

30 μg of protein extract per sample were loaded onto the polyacrylamide gels (Mini-PROTEAN TGX Precast Gels) after boiling for 5 min in Laemmli buffer (80 mM Tris-HCl pH 6.8, 5 mM DTT, 2% SDS, 7.5% glycerol, 5 mM EDTA, 0.002% bromophenol blue). Samples were run using the Mini- PROTEAN Tetra Cell or the Criterion™ Cell electrophoresis system (Bio-Rad) at a constant voltage of 120 V in SDS-PAGE running buffer (25 mM Tris, 200 mM glycine, 0.1% (w/v) SDS) until the tracking dye reached the bottom of the gel. Precision Plus Protein Dual Color Standards (Bio-Rad) were used as the size reference.

Western blot analysis

For antibody-specific detection of proteins, samples were separated by SDS- PAGE and transferred to a 0.45 μm nitrocellulose membrane (Protran BA 85, GE Healthcare) in transfer buffer (25 mM Tris, 192 mM glycine, 10% methanol) for 90 min at 4°C and a constant voltage of 90 V using the Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad) or Criterion Blotter. After transfer, the membrane was blocked in TBS-T (49 mM Tris base 102 mM NaCl, 5.4 mM KCl, 0.05% (v/v) Tween-20, pH 8) with 5% (w/v) non-fat dry milk (Sveltesse Nestlé) for 60 min with gentle rocking. To detect the protein of interest, the membrane was incubated with the primary antibody (HA 6E2, Cell Signaling 1:1000; GFP Living Colors GFP Monoclonal 632381 Clontech 1:1000; FLAG F1804 Sigma 1:1000; TY1 SAB4800032 Sigma 1:1000; MYC 9B11 Cell Signaling 1:1000; actin (I-19) sc-1616 Santa Cruz Biotechnology 1:1000) diluted in TBS-T milk for 120 min at RT or O/N at 4°C. Afterwards, the membrane was washed three times with TBS-T for 10 min each and then incubated with the respective horseradish peroxidase (HRP) conjugated secondary antibody in TBS-T milk (α-mouse HRP linked GE Healthcare NA931 1:2500; α-goat IgG HRP linked 805-035-180 Jackson ImmunoResearch 1:5000) for 60 min at RT. The membrane was washed twice more with TBS-T for 10 min each time and then once with TBS only.

To chemiluminescently detect the protein of interest, ECL Blotting Detection Reagents (GE Healthcare) and Amersham Hyperfilms ECL (GE Healthcare) were used according to the manufacturer’s instructions.

Films were developed manually: 1 min in developing solution (Agfa developer G153), 1 min in fixing solution (Agfa fixer G-345) and then rinsed in water.

Coomassie blue and silver staining

Polyacrylamide gels were stained with Coomassie blue using the Colloidal Blue Staining Kit (Invitrogen) according to the manufacturer’s instructions. Polyacrylamide gels were silver stained using Silver Stain for Mass Spectrometry (Thermo Scientific) according to the manufacturer’s instructions.

RNA interference

Plates seeded with the corresponding RNAi clones were used to feed synchronized WT L1 worms. RNAi clones were obtained from either the ORFeome Library [62]: ZC376.6 (ints-2), Y51A2D.7 (ints-5) T23B12.1 (ints-12) R02D3.4 (ints-13) or the Ahringer Library [63]: F47C12.3 (ints-10).

The RNAi clones C06A5.1 (ints-1), Y92H12A.4 (ints-3), W04A4.5 (ints-4), F08B4.1 (ints-6), D1043.1 (ints-7), Y48G10A.4 (ints-8), F19F10.12 (ints-9), F10B5.8 (ints-11) were cloned from cDNA using the following primers:

ints-1 (Fw 5’-AAACCACGAGTTGGACAAGG-3’,

Rv 5’-TCAAATCAATCGGCATTTCA-3’),

ints-3 (Fw 5’-TTCGCCAAAATGTGAAACAA-3’,

Rv 5’-AGACGTAGGTCAGCGAGGAA-3’)

ints-4 (Fw 5’- CGGATCCCAGAAGAATCGTA-3’,

Rv 5’-CGTCATCACTTGCATCATCC-3’)

ints-6 (Fw 5’- CTCGTTTGAATCCACAAGCA-3’,

Rv 5’-TGAGCTTTTGAGGCATGTTG-3’)

ints-7 (Fw 5’- TGTGAATGCGATGCTTCTTC-3’,

Rv 5’ ACATGTACGGGCAGTTGTCA-3’)

ints-8 (Fw: 5’-TTACTAAGCTTCCATAGATCGCCGTAATCGT-3’,

Rv: 5’- TTACTCTCGAGGTGAGTGGGCCGTGAAGTAT-3’),

ints-9 (Fw: 5’-TATATCAAAGCCCGCGAATC-3’,

Rv: 5’- GGTCTCATCCGGTTTTCAA-3’)

ints-11 (Fw: 5’- AAAAAGGTTGTCGGATGTGC-3’,

Rv: 5’- GCTTCGGTTGAGCAGAAATC-3’)

All RNAi clones were verified by sequencing.

5 ml LB medium containing ampicillin (100 μg/ml) was inoculated with a single bacterial colony and incubated at 37°C for 8 h with constant shaking. 400 μl of the bacterial culture was spread on 90 mm NGM RNAi feeding plates (NGM plates with 100 μg/ml ampicillin, 12.5 μg/ml tetracycline, 1 mM IPTG) and incubated O/N at RT to grow a bacterial lawn and induce dsRNA expression. The next day, synchronized L1 populations were transferred to RNAi feeding plates. The WT strain fed with a clone carrying the empty L4440 vector was used as an RNAi negative control. Functioning of RNAi was assessed by detection of RNA-dependent RNA polymerase mediated amplification of the gene transcripts subjected to RNAi silencing, and western blot of representative subunits of the three transcriptional classes (S8 Fig).

RNA extraction

Worms from 3 to 5 plates were washed off with M9 buffer and collected in 50 ml Falcon tubes, allowing them to settle to the bottom of the tube. Worm pellets were subsequently washed with M9 buffer until no bacterial remains were visible. Next, worm pellets were transferred to Eppendorf tubes and as much supernatant as possible was removed. RNA extractions were performed using the mirVana™miRNA isolation kit (Ambion) following the manufacturer’s protocol for total RNA isolation. Worm tissue homogenization was carried out with the assistance of a polytron pre-chilled with liquid nitrogen.

Northern blot

7 μg of total RNA was denatured for 5 minutes at 95°C and loaded onto denaturing 10% polyacrylamide gels containing TBE-Urea (Bio-rad). Electrophoresis run for about 150 min at 150 V. RNA was transferred to positively charged nylon membranes (Hybond-N+, Roche). After brief washing using 2 × SSC, the transferred blots were cross-linked under short-wave UV light. After prehybridization at 50°C for 1 hour in Church Buffer (0.36M Na2HPO4, 0,14M NaH2PO4, 7% SDS, 1mM EDTA), the blots were subjected to hybridization with Digoxigenin-labeled DNA probes overnight at 50°C. The membranes were washed as follows: twice for 5 minutes at room temperature in 2× SSC, three times 10 minutes at 50°C in 2× SSC containing 0.4% SDS, 10 minutes at room temperature in 1× Washing Buffer (DIG Wash and Blocking Buffer Set, Roche). Membranes were blocked for 30 min at room temperature in 1× Blocking Buffer (DIG Wash and Blocking Buffer Set, Roche), then incubated with anti-Digoxigenin AP Fab fragments (Roche) diluted (1/10000) in 1× Blocking Buffer at room temperature for 30 min and then, they were washed twice (15 minutes each time) at room temperature in 1× Washing Buffer. Membranes were equilibrated in 1× detection buffer (DIG Wash and Blocking Buffer Set, Roche) incubate with several drops of CDP-star (Roche).

RT-PCR

First, RNA was treated with DNase to eliminate any DNA contamination. In each sample, a total reaction of 10 μl contained: 500 ng RNA, 1μl RQ1 RNase-Free DNase (Promega), 1x RQ1 DNase 10X Reaction Buffer and DEPC water. Reactions were incubated for 30 min at 37°C. Reactions were stopped by adding 1 μl STOP solution (Promega) and incubating them for 10 min at 65°C. cDNA synthesis was performed using Transcriptor First Strand cDNA Synthesis Kit (Roche) from 500 ng of total RNA according to the manufacturer’s instructions. Samples were stored at 4°C for immediate use or at -20°C for longer periods. To study 3’ end processing, PCR amplification was performed using the GoTaq DNA Polymerase (Promega). The primers used were:

sls-2.8 Fw 5’-GCTGTCGTTTCGATCTCTCG-3’;

Y75B8A.23 Rv 5’-TGTCGTGAGTAGGTGTGCAA-3’;

H27M09.8 Fw 5’-GTGTGGCAGTCTCGAGTTGA-3’;

H27MO9.5 Rv 5’-TTGAACCTTTTCGTCGGAAC-3’;

F08G2.9 Fw 5’-TGGAACCTAGGGAAGACTCG-3’;

ins-37 Rv 5’-TTGAACTTGTCCGGGATTCT-3’;

W04G5.11 Fw 5’- ATTTTTGGAACCCAGGGAAG-3’;

W04G5.8 Rv 5’-GTGGAGATTTCTGCGACACA;

F08H9.10 Fw 5’- TGACCTATGTGGCAGTCTCG-3’;

F08H9.3 Rv 5’- TCGACAATCTCATTCCGACA-3’;

act-1 Fw 5’- CCAGGAATTGCTGATCGTATG-3’;

act-1 Rv 5’-GGAGAGGGAAGCGAGGATAG-3’

The final concentrations in each PCR reaction were: 1x GoTaq Reaction Buffer (1.5 mM MgCl₂), 0.2 mM dNTPs, 0.2 μM upstream primer and downstream primer, 2.5 units GoTaq DNA Polymerase plus the required amount of DNA template (<500 ng) in each case. For the upstream primers that target the snRNAs, the concentration used was 0.8 μM.

Reactions were performed using GeneAmp PCR System 9700 thermal cyclers (Applied Biosystems). PCR conditions were adjusted in each reaction based on the DNA fragment to be amplified and the primer pairs used, but all reached 40 amplification cycles.

cDNA Library Preparation and Ultrasequencing

Sequencing libraries were prepared by following the TruSeq RNA Library (LS) Preparation Kit v2 instructions (Illumina Inc., San Diego, CA) from 1 ug of total RNA that was previously depleted using the RiboZero (Human/Mouse/Rat) protocol. All libraries were run in a HiSeq1500 PE100 lane in Rapid mode, pooled in equimolar amounts to a 10 nM final concentration. The library concentration was measured via qPCR using the KAPA library quantification kit for Illumina sequencing platforms (Kapa Biosystems, Wilmington, MA) before high throughput sequencing.

Bioinformatic analysis was performed as described [48]: The quality of RNAseq results was initially assessed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The raw reads were trimmed, filtered for those with a Phred quality score of at least 25 and all adapters were removed with Trimmomatic software [64].

Clean reads were aligned versus the N2 Caenorhabditis elegans reference genome (release WBcel235.85, http://www.ensembl.org/Caenorhabditis_elegans/Info/Index) using Tophat2 [65] with default parameters. Resulting alignment files were quality assessed with Qualimap2 [66] and sorted and indexed with Samtools software [67]. After taking a read count on gene features with the FeatureCounts tool [68], quantitative differential expression analysis between conditions was performed both by DESeq2 [69] and edgeR [70] implementations to compare the groups in pairs. Both implemented as R Bioconductor packages and performed read-count normalization by following a negative binomial distribution model. In order to automate this process and facilitate all group combination analysis, the SARTools pipeline [71] was used. All resultant data was obtained as HTML files and CSV tables, including density count distribution analysis, pairwise scatter plots, cluster dendrograms, Principal Component Analysis (PCoA) plots, size factor estimations, dispersion plots and MA- and volcano plots. Resulting tables, including raw counts, normalized counts, Fold-Change estimation and dispersion data for each of the analysis methods (DESeq2 and edgeR) were annotated with additional data from Biomart (https://bioconductor.org/packages/release/bioc/html/biomaRt.html), WormBase (http://www.wormbase.org) and org.Ce.eg.db (https://bioconductor.org/packages/release/data/annotation/html/org.Ce.eg.db.html) databases. Final tables also include the associated gene name, Ensembl Transcript and protein information, GO Term ID and names, EntrezID, UniprotTrEMBL information and Human ortholog ID and gene name data.

In order to control the False Discovery Rate (FDR), p-values were amended by Benjamini-Hochberg (BH) multiple testing corrections [72]. Those features showing corrected p-values below the 0.05 threshold were considered up- or down-regulated genes. To reinforce downstream analysis and discard false-positive over/under-expressed genes, common up- and down-regulated features were extracted from DESeq2 and edgeR tables. The CRAN packages eVenn (https://www.rdocumentation.org/packages/eVenn/versions/2.4) and pheatmap (https://CRAN.R-project.org/package=pheatmap) were used to graphically represent Venn diagrams and heatmap plots showing these common features.

Gene Ontology enrichment analysis was performed for common up/down regulated genes by using the clusterProfiler package [73] through its enrichGO tool. This tool uses a hypergeometric BH model to obtain adjusted q-values. Each GO category (Biological Process–BP-, Molecular Function–MF, and Cellular Component–CC-) was represented in bar plots, showing its relative abundance and associated q-value. Similarly, KEGG and Reactome pathway analysis was conducted using clusterProfiler (enrichKEGG) and ReactomePA [74] tools. The KEGG pathway maps were obtained with the Pathview package [75].

By tacking DESeq2 expression values, the regularized log2 transformation (rlog) data was represented as a 2-dimensional Principal Component Analysis (PCA) plot. Sample-to-sample Euclidean distances were calculated from the rld data and represented in a heatmap, showing the adjacent clustering information [48].

Lists of up- and downregulated genes are available online for comparison and easy visualization by loading the following datasets onto the web version of the Upset application (http://caleydo.org/tools/upset/):

Upregulated genes:

https://raw.githubusercontent.com/CharoLopez/upset-data/master/Up_regulated_all.json

Downregulated genes:

https://raw.githubusercontent.com/CharoLopez/upset-data/master/Down_regulated_all.json