Xenobiotic and pathogen surveillance genetics and functional genomics

Our initial genetic analysis of the C. elegans response to natural chemical toxins reveals a complex genetic pathway for xenobiotic detection, aversion, and detoxification. We already know that many of the regulatory factors we have identified in our pilot studies are conserved in humans and distinct from those being studied in mammalian pharmacogenomics, which focuses on variation in cytochromes and other enzymes that modify drugs, rather than on the identification steps in the regulatory cascade from drug detection to detoxification. The human homologues of the genes we identify promise to explain how humans respond appropriately and inappropriately to drugs, or activate drug detoxification pathways in the absence of a triggering drug, perhaps inducing a false endocrine state of poisoning. Variation in xenobiotic detection and response pathways will be most obviously important for drug responses such as toxic responses to therapeutic drugs, for example, nausea or hypersensitivity, and addictive behaviors to drugs of abuse. But we believe that such variation will be the cause of diseases as diverse as anorexia nervosa, migraine, depression and developmental defects induced by reduction of function mutations in core cellular components such as mitochondrial and nuclearly encoded mitochrondrial genes.

We view the detection of drugs as developmental biologists, with key regulatory points and checkpoints, which is quite distinct from the standard model in pharmacology. Our hypothesis is that it is the decrease in the efficiency or activity of core cellular components by drugs or mutations that is detected, not the drugs themselves. This hypothesis explains our observations that inactivation of the core cellular components by RNAi mimics the drug detoxification inductions of drugs. The assessment of the decrease in function of the core conserved cell components triggers expression of drug modification and transport genes, as well endocrine signals to suspend feeding and move to a new nutritional source. We aim to identify all of the genetic components of the core cellular component assessment systems, the endocrine systems of spreading the detoxification signal systemically, and the coupling of these stress signals to upregulation of xenobiotic detoxification systems. Our genetic analysis of how toxins are detected and which endocrine signals are produced to make the animals feel sick may identify new endocrine pathways that have human homologues. Drug development targeting those pathways could serve as adjuvants for chemotherapeutic treatment of cancer, to suppress nausea, perhaps allowing the use of higher therapeutic doses, to mention one indication. Human variation in response to drugs and to the wide range of chemicals in the normal diet may be more common than previously appreciated. The most obvious manifestation would be toxic responses in a small percentage of humans to common doses of drugs or common foods. It is also possible that aberrant constitutive upregulation of drug detoxification pathways in the absence of a toxic trigger may induce feeding dysfunctions and drug detoxification behaviors in humans bearing variation in the xenobiotic response pathways we will discover. For example, inappropriate up-regulation of drug detoxification pathways may underlie migraine headaches and anorexia. These are good candidates for diseases of inappropriate xenobiotic responses because of their strong gender bias: 3 to 10 fold more women than men suffer from migraines and anorexia. Response to xenobiotics would be predicted to be more robust in females, where a short duration of poisoning can cause catastrophic fetal development defects, whereas in a male, it can be just a day of feeling poor. The headaches and feeding suppression in such disorders would be normal outputs of a xenobiotic detoxification program, if the patient was actually ingesting a poison. But in the absence of a toxin, the response is inappropriate and induces protective measures that decrease health. Similarly, the developmental arrest phenotypes induced by decrements in essential gene activities we have shown to be a genetic program subject to suppression by mutations in the program of arrest. That is, we believe that some of the some of the symptoms induced by reduction of function mutations in core cell components may represent a genetic program of developmental arrest normally triggered by natural xenobiotic chemicals that target these components. Abrogation of these xenobiotic detection programs may paradoxically improve the health and well-being of patients bearing such mutations. The genes we have so far identified and propose to comprehensively identify may constitute drug targets for the treatment of these diseases not heretofore considered treatable without gene therapy: mitochondrial disorders, cytoskeletal disorders and other mutations in the core components of cells that cause developmental defects. We do not propose to bypass the need for mitochrondria or ribosomes, only to suppress the phenotypes of weak mutations in such systems. Weak alleles are actually very common in these genetic diseases, where the null phenotype is embryonic lethality. Finally, the drug or essential gene inactivations in C. elegans inhibit feeding and induce an aversive program. The endocrine state of these aversively stimulated animals may be homologous to the endocrine state of humans who feel unwell. As we seek genetic suppressors of this aversive behavior, an endocrinology of feeling ill would emerge. Human homologues of the endocrine signals we identify could be used to develop antagonist drugs that might treat the malaise of being ill.

Surveillance of essential cellular functions and xenobiotic detoxification in Caenorhabditis elegans

Naturally occurring antibiotics, or xenobiotics, are produced by a wide range of bacteria and fungi and target highly conserved proteins in eukaryotes and other microbes, such as the ribosome, mitochondria, and cytoskeleton. By targeting conserved RNAs and proteins, these antibiotics can disable a wide range of eukaryotic and prokaryotic competitors of the drug producing organisms. These xenobiotics cause developmental arrest or other toxic responses in many eukaryotes as well as other bacteria. In this chemical war game of measures and countermeasures, the target organisms respond to these xenobiotics through system-wide adaptive mechanisms and chemical detoxification.

We have done pilot genetic and functional genomic screens and propose more comprehensive RNAi and genetic screens and deep sequencing experiments to discover the regulatory cascade that mediate the surveillance of these toxins and their detoxification in animals. At the least ambitious level, understanding the genetic network of xenobiotic sensing and detoxification will provide insights into the pharmacokinetics of medicines, drug resistance, and responses to therapeutic chemicals as well as toxic chemicals. This is a field at its mezzanine level of sophistication. In mammals and insects, chemical detoxification pathways are activated to upregulate a battery of detoxification enzymes, including cytochrome P450, short-chain dehydrogenase/reductase, UDP-glucuronosyltransferase, and glutathione S-transferase genes. The best studied regulators of this response in mammals are the nuclear hormone receptors of the CAR and PXR class that are thought to use promiscous ligand binding domains to detect particular classes of xenobiotic chemicals to then activate these nuclear hormone receptors to directly regulate the enzymes listed above. The ligand binding domain of nuclear hormone receptors such as CAR are thought to detect suites of chemicals. This sort of model explains the detection of some chemical triggers but suffers from the problem of how the infinite number of possible chemical toxins are detected by the few NHR genes in mammals (there are 48 human NHRs).

Our hypothesis is that this mechanism for xenobiotic regulation is not general, and that instead it is the drug inhibition of core cellular components that is monitored and detected, not the drugs themselves. This model explains our observations outlined below that inactivation of the core cellular components by RNAi induces the same drug detoxification response induced by drugs. These data suggest that it is the assessment of the decrease in function of the core conserved cell components that is assessed when drugs are given, and serves to trigger drug detoxification gene expression. A signal from these core components, for example, a kinase cascade coupled to a transcription or chromatin factor, but also possibly activation of small RNAs, triggers expression of drug modification and transport genes, as well endocrine signals to not feed and move to a new nutritional source and systemically upregulate drug detoxification pathways in cells outside of the sentinel signaling cells.

How we came to the model that there is an endocrinology of core cellular component assessment: Our research over the past 15 years has revealed an insulin and TGF-beta endocrine system that controls whether the animals grow reproductively or arrest at the dauer diapause stage. Diapause arrest is an essential feature of many vertebrate and invertebrate life cycles, especially in regions with seasonal temperature and humidity extremes. Animals in diapause arrest survive temperature or humidity extremes or other stresses. For C. elegans, RNAi can be induced by feeding wild type or RNAi hypersensitive C. elegans strains any one of 19,000 isogenic E. coli strains that differ only in which of the C. elegans genes is expressed. Each dsRNA is expressed in the E. coli from a plasmid bearing converging T7 RNA polymerase promoters into which is cloned 1 kb of each C. elegans gene. There are 2700 distinct gene inactivations in C. elegans that cause developmental arrest or cessation of reproduction. The gene identity is annotated for each of these RNAi strains so that the molecular defect that causes developmental arrest is instantly revealed. The ribosome, the mitochrondrion, vacuolar H+-ATPases, the F0F1 proton channel ATPase and other core components of cells strongly emerged from this RNAi screen for developmental arrest.

A common theme to many of these cellular components is that many are targets of antibiotics produced by a range of fungi and microbes that nematodes encounter in the environment. We hypothesize that as a larvae or adult enters an environment with an antibiotic, there may be signaling pathways that detect, for example, antibiotic-induced ribosomal deficiency to trigger cessation of reproductive developmental trajectory, arrest at a particular developmental point, as well as xenobiotic protective pathways in all of the cells of the animal via an endocrine relay system. The induced stress adaptation and survival pathways would ensure that the animal could escape the antibiotic and resume reproductive development in a less toxic environment or feeding on less toxic microbes. Inhibition of translation by RNAi of translation factors may mimic the ribosomal function deficiency induced by antibiotics in the normal C. elegans ecosystem, and trigger the physiological response of developmental arrest and cessation of reproduction. Other gene inactivations also causing arrest are a vacuolar ATPase and the mitochondrial ATP synthase, also targets of natural antibiotics.

C. elegans aversion to drugs and essential gene inactivations, a nausea response.
C.elegans tend not to leave an E coli lawn or other desirable bacteria from their normal habitat which they feed on unless the animals are stressed by some aversive stimulus. As shown in the figure to the left, we noticed when working with the gene inactivations that cause developmental arrest that the animals tended to leave the lawn of bacteria targeting the core components of cells that are essential genes, unlike control E. coli strains expressing dsRNAs corresponding to other C. elegans genes that do not induce arrest. This suggested that the developmental arrest is coupled to a behavioral repertoire that includes “feeling sick” and a calculation that the source of that stress is the ingested food. Because bacteria and fungi produce a large array of xenobiotic secondary metabolites, generally products of polyketide synthase genes or non-ribosomal peptide synthesis genes which constitute 5-10% of many microbial genomes, the conclusion that any drug was likely to come from a bacteria or fungus on which the animal was feeding is reasonable. The physiological calculus of the animal may be that when, for example, the ribosome is not functioning at 100%, the most likely cause in the soil would be feeding on a bacterium or fungus that synthesizes an antiribosomal drug. Therefore cessation of feeding at that instant, and induction of dispersal behavior seeking a different source of nutrition is of high selective value.

To test if the dispersal behaviors induced by essential gene inactivation are congruent with the behaviors induced by drugs, we challenged C.elegans on a benign E. coli food source with antibiotics that inhibit distinct core components and assayed dispersal behavior. We tested drugs that target translation (Geneticin), the electron transport chain (antimycin), proteasome (Bortezomib), vacuolar ATPase (concanamycin A), ER stress (tunicamycin), oxidative stress (paraquat) and DNA damage (zeocin). Each drug induced animal dispersal from an otherwise benign bacterial lawn, reiterating the response elicited by RNAi against the corresponding protein targets. As shown in the figure to the left, these drugs stimulated dispersal within the first 4-6 hours of exposure, and as early as 1 hour.

We observed dispersal behavior in roughly one-third of the 2700 essential gene inactivations tested. Only 1%, or 30x lower frequency, of a random selection on non essential RNAi clones induced dispersal. In many cases arrested animals move significantly faster than animals on control RNAi. Animal speed and tortuosity (a metric for the degree of turning within an object’s trajectory) were quantitated using object-tracking software. Animals grown on repellant RNAi clones travelled much faster and exhibited straighter trajectories than those grown on control RNAi. These trajectories are consistent with dispersal behavior and are reminiscent of the bacterial chemotactic changes in tumbling vs straight trajectories when attracted or repelled by nutrients. We tested for the induction of a stress reporter genes by the drugs and by RNAi of genes that cause developmental arrest because these GFP fusion inductions could be used to advantage in genetic screens for poison blind mutants. Many of essential gene inactivations that induce dispersal also induce the GFP reporters responsive to ER stress (hsp-4::GFP), mitochondrial stress (hsp-6::GFP), oxidative stress (sod-3::GFP), xenobiotic and oxidative stress (gst-4::GFP) or osmotic stress (gpdh-1::GFP). There was ~20-fold enrichment for induction of stress reporters in the test gene set, with 40% of the genes tested exhibiting strong activation of at least one of the reporters, while ~1.9% of a random clone set strongly activated a reporter. These data suggest that behavioral aversion to drugs is coregulated with drug detoxification pathways indicated by the fusion genes.

The drugs also induce these fusion genes. Tunicamycin is a natural product the bacterium Streptomyces Iysosuperficus that C. elegans might naturally encounter in the soil. Tunicamycin affects the ER by inhibiting an enzyme required for N-linked glycosylation and strongly induces HSP-4, a component of the ER unfolded protein response. Antimycin is a natural product Streptomyces bacteria in the soil and inhibits the mitochondrial electon transport complex 3 by binding the cytochrome b subunit. This blocks the transfer of electrons between subunit b and subunit c, preventing ATP generation. Antimycin induces hsp-6, a component of MT UPR. One trivial explanation for the dispersal behavior of animals that are arrested is that any arrest has an associated non-feeding response. Two approaches have been used to assess this. First, if the arrest phenotype can be suppressed, it strongly favors a program of arrest and dispersal rather just a default arrest state. We have found that mutations in daf-18, the PTEN homologue of C. elegans, strongly suppress the disperal behavior of many of the arrested animals, without actually suppressing the arrest itself, pointing to an insulin pathway for the disperal behavior but not for the arrest. Secondly, our genetic analysis below identifies gene inactivations that suppress the arrest induced by drugs, again favoring a genetic program for arrest. These have not yet been tested for suppression of the dispersal behavior.

Inactivation of essential cellular activities stimulates pathogen and detoxification responses

Because bacterial pathogens are known to stimulate food-avoidance behavior in C. elegans, the observation that inactivation of cellular processes using toxins or RNAi produced a similar behavioral response suggests an adaptable mechanism for toxin and pathogen detection mediated by cellular surveillance pathways in the host. We reasoned that if animals use internal physiologic cues to recognize the presence of pathogens, then perhaps RNAi of essential cellular components would trigger physiologic defenses associated with pathogen infection or toxin exposure – in the absence of pathogens or toxins.

We examined the expression of several GFP reporters previously shown to represent activation of innate immune programs: clec-60 (a C-type lectin/CUB domain protein induced by S. aureus and M. nematophilum), F35E12.5 (a CUB domain protein induced by Y. pestis, M.nematophilum and P. aeruginosa) , and nlp-29 (a conserved glycine/tyrosine-rich antimicrobial peptide induced by D. coniospora, S. marcescens and wounding). Finally, irg-1::GFP is a reporter induced by P. aeruginosa and cadmium poisoning whose expression in response to pathogen exposure requires ZIP-2, a bZIP transcription factor. Induction of each pathogen reporter was analyzed against a panel of ~100 gene inactivations representing the major functional classes identified in our aversion screen and a random panel of genes from the whole genome RNAi library. For all 4 pathogen-response genes, we observed a significantly elevated frequency of activation by the aversion gene set relative to the control set. In general, the three intestinal reporters were induced by RNAi against most functional classes, with most potent induction following disruption of protein synthesis, metabolic and mitochondrial functions.

If host surveillance pathways detect the presence of chemical toxins by monitoring core cellular activities, then RNAi-mediated disruption of these activities might also stimulate drug detoxification responses, such as cytochrome P450 and glutathione S-transferase genes. We assayed GFP transcriptional reporters for induction of xenobiotic detoxification enzymes, cyp-35B1 (an intestinally-expressed cytochrome P450 oxidase) and gst-4 (a glutathione S-transferase induced by drugs and toxins). Each of these reporters was induced at a frequency ~4-fold above background (~40% of aversion-inducing essential gene inactivations tested induced cyp-35B1::GFP and ~20% induced gst-4::GFP, p<0.0001, chi-squared test).

An endocrine system for assessment of core cellular components
Our analysis suggests that the integrity of these core cellular components may be assessed in cells that tend to be the most exposed to the environment, and that a signaling pathway to endocrine control of development and reproduction may operate. For example, we have found that there are sentinel signaling cells: the intestine where microbes are first encountered in an intimate way and sensory neurons that might “probe” the antibiotic environment before the rest of the cells of the animal are inhibited. These sensory cells are likely to be endocrine beacons to the rest of the animal. We used a collection of strains in which gene knockdowns can be accomplished within a single, restricted tissue using the same technique of feeding animals bacteria expressing dsRNA against target genes. The strain backgrounds used for these experiments are fully defective for RNAi because they lack the necessary gene (either rde-1 or sid-1). Only the tissue(s) in which gene function is restored (using tissue-specific promoters to rescue rde-1 or sid-1 expression) are competent for gene knockdown when fed RNAi against a given target. In addition, we a mutant strain proficient for RNAi in the germline but not in somatic tissue (rrf-1) due to differing mechanisms of RNAi in these tissue types. Inactivation of core cell components in each tissue tested – the hypodermis, the intestine, the gonad and body wall muscle – is competent to trigger dispersal. This suggests the existence of an endocrine system for response to essential gene inactivation.

The molecular mechanism of xenobiotic detection to induce xenobiotic protection programs and developmental arrest: We interpret developmental arrest differently than most who work on essential genes. Usually such arrests or sterility are considered the consequence of a loss of an essential cellular component; that the animal cannot develop past a developmental milestone in the absence of the particular gene product. However, another view of such developmental arrests is that they are a “programmed” response to a deficiency in a key function, and that active signaling pathways mediate the arrest point, as a sort of “developmental checkpoint”. This view is motivated by the finding that arrest at the dauer stage is an active process and can be abrogated by mutations in dauer defective genes such as the FoxO factor DAF-16 or the Smad factor DAF-3 or the nuclear hormone receptor DAF-12. Importantly such an arrest program interpretation depends on two key attributes: 1. The arrest should be reversible. We have observed that when drug is removed in many cases, the arrest is reversible. 2. There should be mutations that disable the arrest program. This is also established below.

We have used the failure to induce the ER or mitochondrial or other stress GFP fusion genes under drug or stressed conditions to derive strains of C. elegans that are “blind” to mitochrondrial dysfunction or ribosomal dysfunction, and therefore do not induce these GFP fusion genes (outlined in the figure to the left). We can then ask if a failure to induce these genes makes the strains more or less sensitive to these drugs, that is disrupts the normally associated induction of drug resistance, allowing us to study the genetic pathways of xenobiotic surveillance and detoxification. In this way, we will discern the signaling pathway, both within cells and between cells for xenobiotic responses. We may find that animals deficient in these stress sensory or response execution pathways have major defects in the detoxification of xenobiotics or in the responses that allow survival when intoxicated.

We first did pilot screens starting with cherry picked libraries of sublibraries, and propose to ramp up to full genome screens in this proposal. Our first candidate genes for signaling from essential gene inactivation to xenobiotic resistance emerged from a survey of gene inactivations from four 100 to 300 gene candidate libraries. So far from these pilot screens, shown in the table to the left, we have identified 20 gene inactivations that fail to activate the drug induced reporter genes. These genes come from all three candidate libraries. The genes identified by these inactivations are candidates for the genetic pathways that transduce deficiencies in ribosomal or mitochrondrial function to endocrine pathways for developmental arrest.

The key difference between how we have come to view xenobiotic detection and how the mammalian or insect drug detoxification/insecticide resistance field views the problem: Our focus on drug resistance is distinct from the usual interest in resistance or sensitivity one drug or pathway. We are most interested in mutations or gene inactivations that broadly disable response to inactivation of the ribosome or of the mitochondion but not in mutations that make the target protein of the drug compromised for binding, for example by mutating a key amino acid residue for drug binding. Find the drug receptor is the usual goal of drug resistance genetics, though the insecticide resistance field has broadened the goals to discover gene amplifications that mediate resistance (2). In contrast, we wish to identify the signaling components that sense the ribosome or mitochrondrial or cytoskeletal integrity and signal from these assessments to generalized drug detoxification pathways. So our expectation is that many of the genes we identify will make animals hypersensitive or resistant to more than one class of drug.

A signal from these core components, for example, most likely a kinase cascade to a transcription or chromatin factor, but also possibly activation of small RNAs, triggers expression of drug modification and transport genes, as well endocrine signals to not feed and move to a new nutritional source. This is different than the problem posed by most of the field of toxicology: how do the transcriptional regulatory factors such as the nuclear hormone receptor CAR recognize the infinite number of possible chemicals in the toxin world. If toxins are recognized by their inhibition of the ribosome or mitochondrion or other core cell components, the only recognition needed is that they are drugs that cause a decrement in core component function. That is only the function of the core elements must have a checkpoint control.

Revised on 2013-02-13 17:57:38 UTC