Gary Ruvkun
Professor of Genetics, Harvard Medical School
Department of Molecular Biology, Massachusetts General Hospital
185 Cambridge Street, Boston, MA 02114
Major Research Interests: microRNA and RNA interference mechanisms, bacterial and animal genetic analysis of microbiome interactions, neuroendocrine control of detoxification, immunity, and aging, life on other planets
Small RNA research in the Ruvkun lab
Small RNA pathways regulate eukaryotic defense against invading genetic elements such as viruses. RNA interference in the animal C. elegans is highly ramified and represents the RNAi ancestral state common to plants and fungi, with particular Argonaute and RNA-dependent RNA polymerase proteins specialized for silencing of transposons and other integrated viruses or new invading viruses. The new genes we have discovered that strongly enhance RNAi reveal mechanisms by which siRNA production is naturally broadened to defend against novel viruses. Our saturation genetic analysis of antiviral pathways has revealed surprising new possibilities for how the genomic record of past viral infections can anticipate the constant variation in new viruses that are encountered, using DNA and RNA-based recombination between viral genomes archived in the C. elegans genome and RNA editing of those genomes. Our hypothesis that the retinoblastoma gene pathway mediates gene amplification of integrated viruses to produce substrates for unequal crossing over to generate siRNA diversity is very unorthodox but intersects with the aneuploidy common in tumor biology. The RNA editing we study uses ancient enzymes that enable wobble base pairing between tRNA and codons.
Somatic misexpression of germline meiotic recombination and RNA interference genes in dREAM complex mutants. Our comprehensive screens for RNAi defective and for enhanced RNAi mutants and gene inactivations have revealed an intersection between regulation of natural gene endoreduplication and RNAi in C. elegans. A large fraction of the gene inactivations that we discovered to enhance RNAi were also isolated as synMuv B genes in genetic screens in the Horvitz and Han labs for increased EGF signaling in vulval patterning (Clark, Horvitz 1994; Lu Horvitz 1998; Cui Han 2006; Harrison Horvitz 2006, 2007; Davison Horvitz 2005, 20011; Saffer Horvitz 2011). We discovered that all of these viable and quite healthy synMuv B mutants strongly enhance RNAi and cause misexpression of germline-specific genes in the intestine, including many genes implicated in RNAi such as pgl-1 and pgl-3 in P-granules (Wang, Ruvkun 2005; Wu, Ruvkun 2012). These dozen synMuv B genes encode homologues of the mammalian dREAM complex found in nearly all animals and plants, including tumor suppressor Rb and Rb complex components LIN-53 /RbAp48, LIN-37/MIP40, LIN-54/MIP120, DPL-1 /DP, LIN-9/MIP130. The Drosophila dREAM complex binds specifically at replication origins that flank the Drosophila chorion genes to control their local (about 50kb) endoreduplication (Lewis Botchan 2004; Bandura Calvi 2005). Gene expression analysis in a variety of C. elegans dREAM mutants reveals that many normally germline-specific genes are markedly upregulated in the soma (Kirienko Fay, 2006; Petrella Strome 2011; Wu Ruvkun 2012). The dramatic up-regulation of meiotic recombination and chromosome synapsis genes in the C. elegans dREAM complex mutants suggests an involvement of chromosome pairing and recombination in the function of endoreduplication; and these genes are activated in the intestinal cells (Petrella Strome 2011).
There is a common theme to all of the processes affected by the dREAM complex: the C. elegans intestine and hypodermis (Hedgecock White 1985), like the chorion of Drosophila (Royzman Orr-Weaver 1999, Bosco Orr Weaver 2001) undergo multiple programmed full genome endoreduplications that are probably independent of the dREAM complex but then a highly localized dREAM complex-mediated controlled amplification to dramatically increase gene dosage of only those 50 kb scale genomic regions. In the 32C polyploid intestinal cell that will never divide again, there is magnificent ammunition for natural generation of genetic diversity by unequal crossing over in repetitive elements. Thus we predict that the C. elegans dREAM complex mutants mediate localized amplification of client genomic regions such as the holocentromeric elements that we hypothesize mediate antiviral defense. Our full genome RNAi screens for gene inactivations that disable RNAi (Kim, Ruvkun, 2005; Wang Ruvkun 2005) revealed additional candidate regions for localized endoreduplication to increase RNA interference: we found that 6 of 36 viable RNAi-defective gene inactivations tested can suppress the enhanced RNAi and the misexpression of P granules in dREAM complex mutants: mes-4 (SET domain), sin-3, gfl-1 (ortholog of mammalian glioma amplified sequence 41, GAS41), zfp-1 (PHD domain, ortholog of mammalian AF10), M03C11.3 and ZK1127.3. Thus, the increased activity of all of these genes in the synMuv B mutants is required for the P granule misexpression and for the enhanced RNAi. Amazingly, those same RNAi-defective suppressors of enhanced RNAi in synMuv B mutants also suppressed the Multivulva phenotype (too much EGF signalling) from the hypodermis, a tissue that also endoreduplicates in C. elegans (Hedgecock White 1985). This suggested that the enhanced RNAi of the dREAM mutants is related to the expression of normally germline-specific RNAi components in the polyploid intestine and hypodermis. Our model is that the dREAM complex mutants, mes-4, zfp-1, gfl-1, sin-3, ZK1123.3, and M03C11.3 are amplified aberrantly and expressed at ultrahigh levels in the intestine and hypodermis. Inactivation of the synMuv suppressor genes that mediate the increase in RNAi may interfere with these amplifications or with the assembly of P granules and other germline components from these amplified gene products.
ADAR editing and the ERI-6/7 RNAi pathway silence endogenous viral elements and LTR retrotransposons. Long double-stranded RNAs are a step in RNA virus replication. Adenosine deaminases acting on double stranded RNA (ADARs) detect these RNA duplexes and edit the RNA sequence to destabilize the RNA duplexes. A double mutant in the C. elegans ADAR genes, adr-1 and adr-2 grows normally but if the animal also carries mutation in the ERI-6/7 RNA helicase or the ERGO-1 Argonaute protein that functions to silence newly acquired viruses, it is lethal (Reich Bass 2018; Fischer Ruvkun 2020). Single mutants for eri-6/7/helicase RNAi pathway are healthy but their RNAi pathway is enhanced. The eri-6/7; adr-1; adr-2 triple mutant shows a dramatic upregulation of the RNAi machinery and the unfolded protein response, consistent with the secretory stresses associated with viral replication and secretion (Reich Bass 2018; Fischer Ruvkun 2020). The strong induction of the UPR in C. elegans defective in ADAR editing and the eri-6/7 anti-viral RNAi pathway is likely a consequence of ER stress caused by overexpression of retrotransposons and viral proteins. These animals are suffering from the toxicity too much anti-viral RNA interference: the lethality of the eri-6/7; adr-1; adr-2 triple mutant is rescued by inactivation of many RNAi factors such as mut-16, rde-1, nrde-3 (Reich Bass 2018; Fischer Ruvkun 2020). A mutation in C. elegans drh-1, the C. elegans ortholog of the viral sensor protein RIG-I, also suppresses the lethality of the eri-6/7; adr-1; adr-2 triple mutant.
Phylogenetic profiling to discover genes that act with Argonaute and RNA dependent RNA polymerase genes to mediate antivirus programs. Argonaute proteins present small RNAs to their targets and are widely distributed across most eukaryotes; the basal eukaryote had Argonautes and was probably competent for RNAi. But Argonautes and RNAi have been lost in from many Ascomycota and Basidomycota and some protists. To identify other genes that were lost coincidently with small RNA pathway genes, we determined the phylogenetic profiles of all 20,000 C. elegans proteins in 85 animal, fungal, plant and protist genomes. The phylogenetic profile of Argonaute proteins shows a dramatic loss in the budding yeast, that has also decreased its introns from the average of about 10,000 in most fungi, to 100 in budding yeast. A hallmark of viruses is the lack of introns suggesting that intron surveillance is a logical mechanism to discern foreign viral genes in the background of so many more self- genes. Viable null mutations in U1 and U2-snRNP specific splicing factors causes defects in RNAi. Deep sequencing analysis showed that many siRNAs from intron exon junctions are correlated with both poor splicing motif quality and poor conservation of the host gene in the related nematode C. briggsae. Thus endogenous siRNAs are more abundant on poorly spliced mRNAs and those poorly spliced mRNAs tend to come from recently acquired, probably virally derived, foreign genetic elements. These RNAi-targeted transcripts are tightly bound to intact spliceosomes.
Retroviral elements are often the sequences that define centromeres; in fact, viruses appear to drive the surprising lack of conservation of centromere DNA sequences between closely related species (Malik Henikoff 2009). But nematodes add holocentromeres to the mix of possibilities: instead of one centromere for each of the six chromosomes, C. elegans has an estimated 1000 to 2000 point centromeres, hundreds per chromosome, precisely mapped by CHIP seq using KLE-2, the centromere kleisin or with HCP-3, the CENP-A histone of the centromere (Steiner Henikoff 2014). Only six of the holocentromeres recombine at meiosis, where one crossover per chromosome pair mediates chromosome cohesion at meiotic division I. But in the endoreduplicated intestinal nuclei which never again condense their chromosomes or undergo mitosis after the L2 stage but which undergo four rounds of probable full genome endoreduplication, we hypothesize that activated recombination between holocentromere elements in the synMuv B mutants (perhaps also in other enhanced RNAi mutants) could generate highly diverse siRNAs to confer added viral immunity. These recombinant siRNAs may anticipate the genetic variation that might arise from RNA-based recombination between viruses that C. elegans and all eukaryotes encounter. Diverse viruses are known to coinfect hosts and in the mixed infection, recombine with each other via RNA dependent RNA polymerase template choice, allowing acquisition of highly evolved genetic capacities (Kovalev Nagy 2019). We hypothesize that the enhanced RNAi phenotype of synMuv B mutants is due to this amped up viral siRNA generation in the intestine.
Delineating the siRNA signature of synMuv B mutants: The Eri phenotype of any of the synMuv B mutants is as strong as that of the rrf-3 RNA-dependent RNA polymerase mutant, the eri-1 RNase mutant, or the eri-6/7 RNA helicase mutant. All of these Eri mutants respond more intensely for example to histone his-44 dsRNA and recognize transgenes as foreign to silence them, most likely because of promiscuous production of siRNAs that now target other paralogous genes with some mismatch nucleotide homology. In the eri-6/7 helicase or ergo-1 “specialized for silencing endogenous integrated viruses” Argonaute mutant, the Eri phenotype is associated with the desilencing of about 100 recently acquired C. elegans genes, some with hallmarks of viral genomes (Gag, Pol, Env genes). Missing in the ergo-1, eri-6/7, eri-1 and rrf-3 mutants are 26G siRNAs, that are generated during embryogenesis from these foreign elements in the C. elegans genome (Fischer Ruvkun 2008, 2011, 2013). eri-1 and rrf-3 have the added defect of temperature sensitive spermatogenesis defects most likely because of sperm meiosis defects (Han Kim 2009). 26G siRNAs from viral genomes prime the RRF-3 production of secondary 22G siRNAs to amplify the silencing of these integrated viruses (Fischer Ruvkun 2011) that are also dramatically decreased in the eri-6/7, ergo-1, rrf-3, and eri-1 mutants. The lack of silencing of these viral genomes increases expression of these endogenous viruses, which may in turn trigger highly evolved antiviral immunity genes, which in C. elegans translates to RNAi factors. The heritable RNAi factor HRDE-1 Argonaute is also upregulated in eri-1, rrf-3, eri-6/7, ergo-1, and synMuv B mutants suggesting that heritable RNAi is induced (Wu Ruvkun 2012).
In collaboration with Tai Montgomery at Colorado State Univerisity, we have surveyed by deep sequencing of siRNAs the small RNA landscape of lin-35 and lin-54 amutants to determine which siRNAs are up and down in embryos and adults of each mutant strain compared to wild type. We expect to see siRNAs from integrated viruses archived in the C. elegans genome depleted, so that RNAi is enhanced as antiviral immunity is induced. A complete list of the C. elegans holocentromere locations is available at GSE44412 from the datasets of Steiner and Henikoff (2014). We expect that many of the 26G siRNAs that normally bind to ERGO-1 Argonaute derive from these holocentromeres so that the location of the CENP-A binding sites should be within kb of these integrated viruses. Normally, such holocentromeres may not be active for siRNA generation and recombination in the intestine, but in the dREAM complex mutants, many more 26G siRNAs may be generated from the holocentromeres.
Induction of RNA interference by C. elegans mitochondrial dysfunction via the DRH-1/RIG-I homologue RNA helicase and the EOL-1/RNA decapping enzyme . RNAi is an antiviral pathway that detects and cleaves foreign nucleic acids. In mammals, mitochondrially-localized proteins such as MAVS, RIG-I, and MDA5 mediate antiviral responses. We found that mutations in many different C. elegans mitochondrial components robustly enhance RNA interference. As shown below, we use a variety of E. coli strains that express dsRNAs that cause no phenotype in wild type C. elegans but are lethal or show a novel phenotype in an array of mutants that have enhanced RNA interference. The enhanced RNAi responses to mitochondrial dysfunction depend on the RIG-I homologue, the DRH-1 RNA helicase. Comparing the C. elegans transcriptional response of a mitochondrial mutant and infection with the Orsay RNA virus, we found a striking induction of multiple members of C. elegans pals- genes implicated in anti-viral and anti-pathogen response pathways, and the eol-1/DXO RNA decapping enzyme gene. eol-1 transcription is induction is DRH-1 dependent, and an eol-1 null mutation in strongly suppresses the antiviral RNAi response normally induced by mitochondrial dysfunction. EOL-1 protein forms foci in the cytosol only if the mitochondrion is stressed, and the production of these foci are dependent on production of RNA from the mitochondrial genome. This may be mechanistically related to the central role that dsRNA released from the mitochondria plays in mammalian antiviral response pathways. A decrement in mitochondrial function is one of the most potent mechanisms to increase longevity in a variety of species. Mutations in eol-1 or drh-1 suppress the increase in longevity caused by mitochondrial dysfunction. Thus, enhanced RNA interference and antiviral activity is a key output from mitochondria for anti-aging. The dramatic increase in frailty in human old age may reflect such an increase in viral vulnerability.
Electron transport chain genetics: Suppression of oxygen sensitivity of C. elegans electron transport chain mutants by mutations in the PPGN-1 membrane protease or the CMTR-1 RNA methyl transferase. We have been working with a collection of viable electron transport amino acid substitution and late stop codon mutations in various protein subunits of complex I, complex II, complex III, and complex IV. Mutations in distinct Complex I subunit genes of the electron transport chain, nduf-7 and nduf-2/gas-1, are viable at atmospheric (20%) oxygen but are inviable at 50% oxygen, which does not inhibit wild type growth. By selecting for survival in 50% oxygen, a population of more than 100,000 F2 progeny of nduf-7 or nduf-2/gas-1 animals after a random EMS mutagenesis, we identified dozens of mutations in the nuclearly-encoded mitochondrial membrane protease gene ppgn-1, an orthologue of bacterial ftsH, and a cluster of gain of function mutations in the RNA methyltransferase gene cmtr-1, an orthologue of bacterial ftsJ, that suppress mutations in complex I subunits. Amazingly, ftsH and ftsJ are located in the same operon in many species of bacteria, including E. coli, and are located in the same operon as the Complex I of the electron transport chain in the bacterial genera Thioalkalivibrio, Alkalilimnicoli, and Legionella. Thus, the suppression of the oxygen sensitivity of two different C. elegans complex I mutations identified mutations in distinct classes of C. elegans nuclear genes that were located in the same operon in the bacteria endosymbiont 2 billion years ago that became the mitochondria.
Protein editing by deglycosylation: Our comprehensive dissection of how proteasomal challenges are detected and compensated generated the surprising discovery that the SKN-1A ER-localized transcription factor is N-glycosylated, and then deglycosylated by PNG-1/NGLY1, editing particular N-glycosylated asparagines to aspartic acid has transformed our understanding of the pathways that mediate proteasomal response. The N to D editing of SKN-1A that we proved with genome edits of N to D of the skn-1 gene were actually first predicted by comparison of SKN-1A protein sequences between distantly related nematodes: we saw clear evidence that N to D substitutions in SKN-1A across nematodes. Our 2021 bioRxiv paper showing phylogenetic evidence for N to D editing in Covid Spike and RNA dependent RNA polymerase proteins, the viral resistance of NGLY1 homozygous patients, and the RNA virus resistance of mammalian cells after NGLY1 gene inactivation motivates our current screen for defective viral responses in the png-1, ddi-1, and skn-1a C. elegans proteasomal pathway mutants. We discovered that the NGLY1/PNG-1 deglycosylation enzyme mediates the N to D protein editing of the SKN-1A transcription factor to control proteasomal capacity. This PNG-1/NGLY1 protein editing, and DDI-1 protein cleavage of the ER-localized N-glycosylated SKN-1A transcription factor also applies to the human homologue transcription factor NRF1 in tumors: the same aneuploid tumors are sensitive to the loss of NGLY1, DDI2, and NRF1, exactly the pathway that emerged from our worm genetics. This pathway may mediate aneuploid cell responses to the protein assembly and degradation challenges of aneuploidy in tumors.
Mammalian NGLY1 peptide:N-glycanase orthologous to C. elegans PNG-1 had been known for decades to remove an amine from the glycosylated asparagine as it deglycosylates the client protein to leave behind an aspartic acid residue in the deglycosylated client protein NxS/T glycosylation site (Suzuki 2002). But the glycosylation field was focused on the removal of glycosylation, not on the protein editing. N-glycosylation is a common feature of viral envelope proteins; Covid-19 Spike protein has 22 N-glycosylation sites, some located in the region that binds to the ACE2 receptor and mediates viral entry into host cells (Zhang 2020). Using the pattern of N to D substitutions in NxS/T N-glycosylation sites in phylogenetic comparisons related coronaviruses, we found 57 D-substituted N-glycosylation sites in Covid-19 protein orthologues in other coronaviruses that are candidates for N to D editing, including 13 in the Spike protein. A particularly striking example is N657 of the Spike protein of Covid-19, where the D substitution occurs often in the Bat SARS virus. This phylogenetic analysis suggests that N to D editing of viral proteins by NGLY1. Some of these N to D sequence changes have been observed in mass spectroscopy of viral peptides, in many cases, bound to HLA. NGLY1 deficient patients show aberrantly increased antibody titers toward rubella and/or rubeola following vaccination, and their parents report few cold or flu infections (Lam 2017). An RNAi screen for human gene inactivations that confer immunity to Enterovirus 71, an RNA virus, identified NGLY1 as a gene activity needed for viral replication (Wu 2016).
Surveillance of translation in immunity and aging The daf-2/insulin signaling pathway emerged from comprehensive C. elegans genetic analysis of aging as the most potent regulator of lifespan, and insulin-like signaling is also a significant aging axis in mammals. Our genome-wide RNAi screens for increased longevity revealed pathways that surveil the ribosome for deficits and couple to longevity control. This is not a quirk of C. elegans: it is related to the mTOR/rapamycin regulation of longevity in mammals. We test whether the mutations that fail to activate detoxification and defense pathways when translation is targeted by pathogens have consequences on the lifespan and measures of aging of the animal. Our discovery of the coupling of longevity control to detoxification and immunity provides a satisfying model for why women live longer than men: placental females may trigger these detoxification, innate immunity, and longevity pathways at lower levels of perceived toxins to protect the developing fetus. The dramatic gender differences in anorexia nervosa and autoimmunity are explained by induction of detoxification and innate immunity programs to inhibit feeding and activate immune responses to a falsely detected pathogen (autoimmunity). The genetic variation in these pathways may have been selected by resistance in the distant past to pathogenic bacteria that often target the ribosome.
We have discovered a remarkable detector of ribosomal dysfunction, a frameshift between open reading frames of the ZIP-2 transcription factor. ZIP-2 detects ribosomal elongation stalling by frameshifting from a 146 aa upstream open reading frame (oORF) to the main open reading frame (mORF) that bears a bZip DNA binding domain to activate the pgp-5 ABC transporter detoxification response to translational elongation defects. Gene inactivation by RNAi of core functions such as translation triggers immune responses as well as a dramatic increase in longevity (Curran, 2007; Melo, 2012). These responses may have evolved to cease reproduction, increase longevity, and activate aversive responses to pathogenic bacterial toxins and virulence factors. This surveillance of key pathways for dysfunction is heretical to traditional toxicology or innate immunity models---it does not actually chemically detect the particular chemical toxin or the particular protein virulence factor. Rather only the inhibition of, for example, translation, is detected, and it serves as a trigger to induce detoxification and innate immunity defenses and trigger a cessation of reproduction and delay of aging (Jones, 2006). Rather than a toxin or virulence factor being directly detected, instead its potency for its cellular target, for example the ribosome, is detected by the surveillance by a sensor for a decreased rate of translation. The detection of pathogens by this pathway can be agnostic to the chemical mechanism of their toxins or virulence factors, so that entirely novel virulence factors and toxins can be detected by their inhibition of core cellular processes. Such a eukaryotic system can misinterpret its own host mutations in for example translation, as pathogen attacks. We have discovered during the past granting cycle that the translational frameshift of the ZIP-2 transcription factor is a fantastic sensor for attacks on the ribosome that directly couples to defense. It is signal transduction in by a single component, a sensor/transcriptional regulator.
Toxins or mutations that disrupt translational elongation up-regulate pgp-5 by frame-shifting the translation of the ZIP-2 transcription factor between two open reading frames. The C. elegans ABC transporter gene pgp-5 is induced 100x to 1000x by exposure to the translation elongation inhibitor hygromycin (produced by Streptomyces hygroscopicus) or Pseudomonas aeruginosa ToxA which ribosylates elongation factor 2 (EF2) (Dunbar 2012; Estes, 2010; Govindan, 2015; McEwan, 2012). pgp-5 is one of 60 C. elegans ABC transporters that may mediate the cellular excretion of toxins, but pgp-5 is induced specifically by gene inactivations that disrupt translation (Govindan, 2015). The bZip transcription factor ZIP-2 was identified in an RNAi screen for gene inactivations that disrupt the induction of irg-1 (infection response gene) and pgp-5 by the translation inhibitors (Estes, 2010; Dunbar, 2012). A zip-2 null mutant does not activate pgp-5::GFP on toxins, gene inactivations, or mutations that disrupt translation. Translation of the ZIP-2 transcription factor increases in response to treatments with ribosomal toxins or mutations in genes affecting translation, but the mRNA level of zip-2 does not increase (Dunbar, 2012). ModEncode data shows that ZIP-2 binds to the pgp-5 promoter.
To dissect the particular step in protein translation that is linked to zip-2/pgp-5::GFP up-regulation, we tested pgp-5::GFP against a panel of chemical toxins targeting either the initiation, elongation or termination steps in translation. Kasugamycin and linezolid inhibit translational initiation but did not up-regulate zip-2/pgp-5::GFP. Hygromycin, gentamycin, paromomycin, anisomycin, and blasticidin that inhibit the elongation step of translation cause an up-regulation of zip-2/pgp-5::GFP. Hygromycin binds to the tRNA decoding site of the ribosome (Ganoza, 2001; Borovinskaya 2008). No pgp-5::GFP was induced in, for example, hygromycin treated zip-2(tm4248) null mutant; pgp-5::GFP. Puromycin inhibits the peptide release step of translation and did not up-regulate zip-2/pgp-5::GFP. These toxins are lethal to C. elegans at high dose, but we tested sub-lethal toxin doses.
To explore the C. elegans mutations that can activate the zip-2 translational sensor pathway, we did a large EMS mutagenesis screen for mutations that up-regulate pgp-5::GFP. Fifty mutants expressing pgp-5::GFP with translation inhibitor were identified, and ten of these mutants were full genome sequenced. Mutations identified in this analysis included the aminoacyl-tRNA synthetases cars-1, mars-1, and yars-1. A previously isolated viable and backcrossed point mutation in the rars-1 arginyl-tRNA synthetase, rars-1(gc47) (Anderson, 2009), activates pgp-5::GFP expression. pgp-5::GFP expression in the rars-1(gc47) mutant was suppressed by zip-2(RNAi). We found that RNAi of any of the dozens of AA-tRNA synthetase genes also induces pgp-5::GFP (as well as larval arrest due to the translational defects). A deficiency of charged tRNAs may force ribosomes to pause and frameshift during the elongation step of translation. This suggests that zip-2 mRNA translation is responsive to changes in peptide elongation on the ribosome. Hygromycin induces pgp-5::GFP at a concentration at which no lethality, infertility, or growth inhibition is observed, as if the activation of ZIP-2 is sensitive to the initial low dose of a toxin an animal first encounters as it approaches or consumes a pathogen. Frameshifted ZIP-2 may activate both a decrease the toxin dose via PGP-5 ABC transporter detoxification as well as behavioral aversion to the bacteria that triggered this ZIP-2 upregulation, a likely microbial source of a ribosomal toxin.
The main open reading frame (mORF) of zip-2 encodes a 308 amino acid transcription factor with a bZip DNA binding domain in the C terminal aa 242-305. An upstream zip-2 overlapping open reading frame (oORF) starts 73 base pairs upstream of the mORF AUG; the oORF encodes a 146 amino acid protein out of frame but partially overlapping (101 amino acids) with the N-terminal region of ZIP-2 mORF. Using CRISPR editing of the zip-2 chromosomal locus, we removed the ATG start codon from oORF in a pgp-5::GFP reporter strain. This oORF initiation codon mutant constitutively translated the zip-2 mRNA to a functional ZIP-2 protein to induce pgp-5::GFP with no hygromycin. Similarly, CRISPR-engineered two early stop codons in oORF also induced pgp-5::GFP with no hygromycin. pgp-5::GFP expression in both CRISPR-engineered oORF mutants was suppressed by zip-2(RNAi). Thus, normal translation of oORF inhibits translation of the downstream zip-2 mORF. The activated ZIP-2 protein in the absence of oORF start codons or with nonsense codons in oORF could either be only the downstream mORF or a fusion protein of oORF frameshifted to the mORF DNA binding domain protein. Because only drugs that inhibit ribosome elongation, but not drugs that inhibit initiation of translation activate the ZIP-2 transcriptional regulation of pgp-5; this interaction between the upstream ORF and downstream ORF is very different from the action of upstream ORFs on for example GCN2 (Hinnebusch 2016). Our most surprising result is that the zip-2 ATG-start codon in the mORF is not required for translation of a functional ZIP-2 transcription factor: using CRISPR, we edited the zip-2 chromosomal locus mORF ATG to ACC. This mutant induces zip-2/pgp-5::GFP on hygromycin but does not express pgp-5::GFP with no hygromycin. These data suggest that a failure to initiate translation or elongate translation on mORF can be compensated by the normal translation of oORF. The most parsimonious model that also weaves in our discovery that ribosomal toxins specific for translational elongation activate ZIP-2, is that translation initiated on oORF can frameshift to mORF if oORF translational elongation is inhibited.
Consistent with a function of the oORF-encoded protein sequence, oORF encodes many acidic and S/T amino acid residues that may be phosphorylated by the kinase cascades we have identified. Comparing the zip-2 oORF sequence between C. elegans and C. briggsae, separated by 30 million years of evolution (using the vertebrate fossil record for years per aa substitution), we can see conservation of the oORF protein sequence as well as of the DNA sequence in that region.
The engine of the Ruvkun lab has been genetic analysis, genome analysis, and functional genomics. We were early users of full genome RNAi libraries for surrogate genetics; this has been displaced in the lab over the past five years by full genome sequencing of newly generated mutations in large-scale genetic screening.In this era of full genome sequences, genetic screening has become literally 50x more productive. Recent Ruvkun lab students and postdocs have discovered entire genetic pathways in this manner.A decade ago, it was common for a single student to figure out one gene during a PhD or postdoc.We no longer endure a year of linkage mapping and chromosome walking.This is a sea change: in a period of weeks after the screen, we now identify lesions in dozens of different genes. And notice also that our genetic pathway discovery papers of 2016 to 2022 are two to three author papers; these are not hundred author Big Science papers. Genetics constantly enlarges our interests to generate new projects and perspectives. There are always many different projects, kinds of questions and approaches in the lab, which provides excellent training, intellectual flexibility and confidence. Instead of a lab where students and the PI divide increasingly smaller slices of the same pie, students and postdocs generate their own projects which they take with them to their own labs.
For the past 20 years, in collaboration with Mike Finney (MJ Research, Vaxart and other biotechs), Maria Zuber (MIT), and Chris Carr (MIT/MGH, now at Georgia Tech), we have been developing a small nucleic acid sequencing instrument to send to Mars or other bodies to detect and sequence DNA. This instrument seeks to test the theory that nucleic-acid-based life exchanged between Earth and Mars early in solar system history. We have also explored the evidence for lateral transfer of microbial life to the early Earth.
