Restoration of Proteostasis in the Endoplasmic Reticulum Reverses an Inflammation-Like Response to Cytoplasmic DNA in Caenorhabditis elegans

Abstract

Caenorhabditis elegans possesses a rudimentary innate immune response that serves as a model for various aspects of the human innate immune response. To date, a nematode response to pathogenic cytoplasmic DNA has not been identified...

Innate immune responses protect organisms against various insults, but may lead to tissue damage when aberrantly activated. In higher organisms, cytoplasmic DNA can trigger inflammatory responses that can lead to tissue degeneration. Simpler metazoan models could shed new mechanistic light on how inflammatory responses to cytoplasmic DNA lead to pathologies. Here, we show that in a DNase II-defective Caenorhabditis elegans strain, persistent cytoplasmic DNA leads to systemic tissue degeneration and loss of tissue functionality due to impaired proteostasis. These pathological outcomes can be therapeutically alleviated by restoring protein homeostasis, either via ectopic induction of the ER unfolded protein response or N-acetylglucosamine treatment. Our results establish C. elegans as an ancestral metazoan model for studying the outcomes of inflammation-like conditions caused by persistent cytoplasmic DNA and provide insight into potential therapies for human conditions involving chronic inflammation.

innate immunity, DNA sensing, inflammatory responses, Caenorhabditis elegans, DNase II

THE processes of sensing and responding to foreign DNA are important in a variety of contexts for various organisms, from bacteria to humans. In perhaps the simplest example, bacteria possess at least two mechanisms for responding to foreign DNA: restriction-modification systems (Vasu and Nagaraja 2013), which protect the bacteria from foreign DNA (especially during bacteriophage infection), and the clustered regularly interspaced short palindromic repeats-Cas9 system, which functions as a rudimentary adaptive immune response to foreign DNA (Ishino et al. 2018). Much more complex systems have evolved in higher eukaryotes (Gallucci and Maffei 2017; Gasser et al. 2017; Dhanwani et al. 2018). In mice and humans, a complex array of detection systems has been identified, and these systems are deployed to sense and respond to inappropriately localized DNA. Well-characterized DNA-sensing pathways include Toll-like receptor (TLR)-mediated sensing in endosomes (via TLR9), as well as direct sensing in the cytoplasm by the cGAS-STING-IRF3 and AIM2 inflammasome pathways. While these systems serve important functions in protecting the host organism from harmful effects of bacteria and viruses, failures in cellular processes that lead to chronic immune signaling can effectively turn these normally protective pathways against the host (Dhanwani et al. 2018).

TLR9 sensing of self-DNA can lead to detrimental effects due to chronic activation of inflammatory pathways. The injection of fetal DNA into the joints in mice, which is detected by TLR9 via its hypomethylation, can lead to complications and fetal loss via an inappropriate inflammatory response to the DNA (Scharfe-Nugent et al. 2012). When self-derived DNA is not cleared due to defects in DNase II-dependent degradation, several resulting inflammation-associated phenotypes have been described. For example, DNase II-deficient mice develop a polyarthritis-like disease that affects the joints due to the improper localization of host DNA in phagocytic cells, where it is detected by AIM2, which stimulates an autoinflammatory condition that ultimately leads to joint degeneration (Jakobs et al. 2015). A similar polyarthritis-like condition has been demonstrated in DNase II-deficient mice due to a failure in the degradation of self-chromosomal DNA during erythropoiesis, which leads to chronic TNFα production and joint degeneration (Kawane et al. 2006). Similar outcomes have been reported in mice after injection of mitochondrial or bacterial DNA directly into the joints (Collins et al. 2004). During autophagic clearance of damaged mitochondria, DNase II degrades the residual mitochondrial DNA (Kawane et al. 2006). Because of the similarities between mitochondrial and bacterial DNA, especially the lack of methylation, mitochondrial DNA can be recognized by host DNA-sensing pathways, including TLR9. In DNase II-deficient mice, failure to degrade the mitochondrial DNA during autophagic clearance of mitochondria in cardiomyocytes leads to a cell-autonomous inflammatory response that can ultimately lead to heart failure (Oka et al. 2012).

The common features of these processes are profound declines in tissue integrity and functionality that are generally ascribed to the induction of a chronic innate immune response. What is still lacking is a detailed understanding of the subcellular and molecular transactions that lead to these phenotypic outcomes. In an attempt to explore these unresolved questions in a simple metazoan model, we developed a method for directly injecting purified DNA into the cytoplasm of the intestinal cells of the 1-mm-long worm Caenorhabditis elegans. Via this approach, we could study the effects of foreign cytoplasmic DNA on tissue integrity. To expand our analysis of the effects of foreign DNA on tissues in a situation more likely to be encountered under environmental conditions, we developed a method to deliver foreign DNA into the cytoplasm of the worm’s intestinal cells via infection with a pathogenic Escherichia coli strain.

No DNA-sensing pathways have been discovered in C. elegans; however, this animal does possess a rudimentary, transcription-based innate immune response (Ermolaeva and Schumacher 2014). Several innate immunity-related regulators have been identified. The best characterized factors are FSHR-1 (Powell et al. 2009) and PMK-1 (Kim et al. 2002), the latter of which is the nematode homolog of the human p38 MAP kinase. The primary outcome of the activation of these pathways is robust transcriptional induction of many diverse genes (Shivers et al. 2008), most of whose functions have yet to be characterized. While it is clear that these pathways confer resistance to a broad range of pathogens, the mechanisms remain incompletely understood. However, what is becoming clear is that like the transcriptional responses associated with the induction of the human innate immune response, the activation of these pathways in C. elegans represents a trade-off between the protective effect of the response against the pathogen and the burdens caused by the massive upshift in protein production (Cheesman et al. 2016; Head et al. 2017).

Through our analysis, we demonstrated that C. elegans responds to foreign DNA, and that the outcome of persistent cytoplasmic DNA is a systemic decline in tissue integrity and functionality. We characterized the subcellular and molecular mechanisms underlying this phenotype, and we describe novel therapeutic interventions that can alleviate these detrimental effects. Our results might have implications for understanding and treating similar pathological outcomes in humans.

Materials and Methods

Strain maintenance

Unless otherwise indicated (see pathogen infection methods), all strains were maintained as described at 20° (Brenner 1974). The E. coli  OP50 strain was used for routine feeding. The N2 Bristol strain was used as the wild-type control. Other strains used in this study were CB1392  nuc-1(e1392) X and MT1902 nuc-1(n887) X. Except for where indicated in Figure 3, the nuc-1(e1392) allele was used throughout.

RNA interference treatment

RNA interference (RNAi) feeding vectors for fshr-1, pmk-1, and sams-1 were obtained from the Ahringer RNAi library (Source Bioscience, Nottingham, UK). RNAi feeding E. coli strains were cultured as follows: an overnight culture in LB liquid supplemented with ampicillin (0.1 mg/ml) was inoculated from a frozen stock and incubated for 37° overnight with vigorous aeration. This culture was diluted 1:1000 in fresh medium and grown to an OD₆₀₀ of ∼0.6 under similar conditions. At this time, IPTG was added to a final concentration of 2 mM and the cultures were incubated for an additional 3 hr. Next, 0.2 ml of this culture was added to NGM plates supplemented with the same concentration of IPTG and grown overnight at 37°. For sams-1, L2/L3 worms were fed RNAi for 10 hr prior to infection as L4 larvae.

Microinjection

Worms were picked the day before injection at the L4 larval stage and grown overnight at 20° on NGM plates seeded with OP50 until they became young adults. The general methodology followed is described in detail in Rieckher and Tavernarakis (2017). The next morning, the young adult worms were transferred into halocarbon oil (Sigma [Sigma Chemical], St. Louis, MO) on a 2% agarose pad for immobilization and injection. Microinjections were performed using an Axio Observer A1 (Zeiss [Carl Zeiss], Tornwood, NY) and FemtoJet (Eppendorf, Hamburg, Germany) microinjection system. Femtotips II (Eppendorf) were used for the injection and the fluorescent dye conjugate rhodamine B isothocyanate-dextran (10,000 M_r) (Sigma) was used to track the injection at a concentration of 6 mg/ml in egg salts buffer. DNA from the uropathogenic E. coli (UPEC) strain was purified using the Puregene Core Kit A (QIAGEN, Valencia, CA) and injected at a concentration of 8 μg/µl, and the synthetic CpG oligodeoxyribonucleotide (ODN) 2395 (Miltenyi Biotec, Bergisch Gladbach, Germany) was injected at a concentration of 100 ng/μl. All injections were made directly into a single intestinal cell. Following injection, the worms were washed with a mixture of M9 buffer and antibiotics (0.1 mg/ml streptomycin and 0.1 mg/ml gentamycin) to minimize contamination. First, 50 μl of the antibiotic solution was placed onto an empty NGM plate and up to 20 worms were placed in the liquid, which was allowed to completely soak into the agar. The worms were then transferred onto NGM plates seeded with E. coli  OP50. Next, the worms were examined using an Axio Zoom V16 (Zeiss) and only worms containing rhodamine B isothiocyanate-dextran in intestinal cells (and not in the pseudocoelomic space or intestinal lumen) were kept for further experiments. Each selected worm was placed on an individual NGM plate seeded with OP50 and maintained at 20 °.

UPEC infection assay

Plate preparation:

For the infection assays, OP50 and UPEC were streaked directly from the frozen stock onto sheep-blood agar (Oxoid/Thermo Fisher Scientific, Waltham, MA). This passage over blood agar was critical for maximum pathogenicity in C. elegans. Overnight cultures were then grown in standard LB liquid at 37 ° for ≤ 18 hr. Next, 0.480-ml aliquots of this culture was spread on 6 cm Petri plates containing 10 ml of high-peptone (HP) NGM, i.e., 3.5 g/liter tryptone/peptone from casein (Roth, Karsruhe, Germany) (compared to 2.5 g/liter for standard NGM). Higher pathogenicity was obtained using tryptone/peptone from casein compared to Bacto-Peptone (BD Biosciences, Franklin Lakes, NJ). The plates were incubated overnight at 37 ° for 18−24 hr, and OP50 and UPEC were treated identically to eliminate strain-dependent uncontrolled effects due to growth on richer medium.

Worm preparation:

Embryos were extracted from gravid young adults using sodium hypochlorite treatment and hatched overnight in M9 buffer with agitation. The newly hatched larvae were transferred to standard NGM plates seeded with OP50 within 18 hr of bleaching. These worms were then incubated at 20 ° until they reached the L4 larval stage (48 hr). The L4 worms were then transferred to UPEC infection plates or OP50 mock-infection plates, and incubated at 20 °. Worms were scored for survival at 24-hr intervals and death was defined as failure to respond to touch. Due to rapid declines in cuticle integrity, worms were not manipulated, i.e., transferred or moved, after the third day of infection as tissue damage and premature death could result.

Health parameters

Pharyngeal pumping was performed as described in Bansal et al. (2015). The sensitivity of this assay was remarkable as temperature differences of only 2–3° could obscure the results; thus, the experiment was performed at precisely 20°. Pharyngeal pumps were counted for 30 sec for each worm. Because of the smaller changes in the pumping rate after DNA injection, worms were monitored individually over 3 days. The pharyngeal pumping rates were measured for 30-sec intervals in triplicate for each worm with each measurement separated by ≥ 30 min to minimize behavioral effects or influences from other uncontrolled variables. For the infection experiments, the pumping rate for each individual animal is shown, summarized by the mean. For the injection experiments, each reported value represents the mean of all of the individual measurements at a given time point under each condition. For the crawling speed measurements (Figure 5), movies of worms were produced using a Zeiss Axio Zoom V16 microscope. The crawling speed was then calculated using the WormLab software package (mbf Bioscience, Williston, VT). For the fecundity experiment in Figure 5, starting from day 1 of adulthood, worms were transferred in groups of three to fresh plates daily for 5 days. The total number of eggs produced was determined by counting both the number of eggs remaining on the plate, as well as the newly hatched progeny from eggs laid earlier. For the pharyngeal bulb surface area measurement in Figure 5, analysis of the surface area of the pharyngeal bulbs was performed in Fiji (Schindelin et al. 2012) by drawing a circle around the pharyngeal metacorpus and posterior bulb to establish the regions of interest (ROIs), and then using the “Measure” function to determine their surface areas. The surface areas of the two ROIs in each worm were added, and the statistical analysis was performed on the combined values determined for each worm. The differential interference contrast (DIC) images of the pharynxes were collected using an AxioImager.M2 (Zeiss) and Zen Pro 2012 software (Zeiss). See also Supplemental Material, Figure S1B for further description of the egg-laying experiment in Figure 10, five worms for each test condition were placed on a plate and allowed to lay eggs per 5 hr. The total number of eggs after this period were counted.

Staining

Nile red (NR) powder (Sigma) was dissolved in acetone at a concentration 500 μg/ml. For treatment, this stock solution was diluted in 1× phosphate-buffered saline (PBS), and pipetted onto NGM plates that were plates previously seeded with either OP50 or UPEC (prepared as described in the UPEC infection experiment) to a final concentration of 0.05 μg/ml. Worms were transferred to these plates as L4 larvae (as in the UPEC infection experiment). NR staining was monitored by fluorescence microscopy using AxioImager.M2 (Zeiss) and Zen Pro 2012 software (Zeiss). For the staining of F-actin with phalloidin, worms were grown as indicated, fixed in 0.7% formaldehyde, reduced with 2-mercaptoethanol, and stained with a 1:100 dilution of Phalloidin-Atto 488 (Sigma).

Immunofluorescence and confocal imaging

After 72 hr of feeding on UPEC, wild-type and nuc-1 worms were picked into 1 ml of M9 buffer and rinsed to remove excess bacteria. The intestines were then isolated via decapitation in egg salts buffer with 0.2% Tween 20 (Roth). Tissue was fixed for 5 min in 3.7% formaldehyde (Roth) before being freeze cracked (Strome and Wood 1982). The fixed tissues were treated with ice cold 100% methanol for 10 min. Following rinsing in PBS + 1% Triton X-100 (Roth) (PBSTX) and PBS + 0.1% Tween 20 (PBST), the samples were blocked for 1 hr with normal donkey serum (NDS) (Sigma) diluted 1:10 in PBST before staining with antibody raised in mouse against α-tubulin (T6199; Sigma). The secondary antibody was donkey Alexa Fluor 488-conjugated anti-mouse (Life Technologies/Thermo Fisher Scientific). Both were diluted 1:1000 in NDS. Samples were mounted in DAPI Fluoromount-G (SouthernBiotech, Birmingham, AL). Confocal imaging was performed using a Zeiss Meta 510 laser scanning confocal system and Zen 2009 software. Intracellular DAPI-stained inclusions were scored by placing an intestinal cell nucleus in the center of a 50-μm² window. The number of DAPI⁺ inclusions in the cytosol of the intestinal cells (i.e., in proximity to the α-tubulin signal) was counted over the entire thickness of the intestinal cells.

TEM

Sample preparation was performed as in Hall et al. (2012). Briefly, worms were removed from food and fixed in 0.8% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.3. Worms were then treated with 0.8% OsO₄ in the same buffer. Worms were washed to remove glutaraldehyde and incubated in 2% OsO₄ in 0.1 M sodium cacodylate buffer for 24 hr at 4 °. Worms were then embedded in 2% agarose and dehydrated via treatment with increasing concentrations of ethanol (50−100%), followed by being embedded in Epon before preparation of the slices for imaging. Microscopy was performed using a Zeiss EM109 electron microscope.

Quantification of TEM phenotypes

To quantify the ER cisternae surface area, TEM slices were taken from two worms for each condition. The images were then partitioned into between 6 and 11 equal-sized boxes that contained only rough ER. The surface areas of the cisternae were calculated in Fiji (Schindelin et al. 2012) by drawing a region of interest around each cisternal space, followed by use of the “Measure” function to determine their surface areas. The medians of the measurements are reported as the metric “ER cisternae surface area.” The sample sizes are given in the figure. To quantify the ribosome distribution, the number of ribosomes located around well-defined cisternae were counted and divided by the circumference of the respective cisternal space (measured in Fiji). The medians of the measurements are reported as the metric “ribosome distribution index.” The sample sizes for each condition are given in figure 7.

Chemical treatments

Tunicamycin (Sigma) dissolved in DMSO was diluted to a final concentration of 0.5 μg/ml. For the treatment, freshly hatched L1 larvae were grown on normal NGM plates seeded with OP50 (no tunicamycin) until the L4 stage. Infections were carried out as described above, except that the HP plates contained 0.5 μg/ml tunicamycin or an equal concentration of DMSO. For injection experiments, worms were grown on OP50-seeded NGM plates until they were injected on the first day of adulthood (as described above). After the injections, the worms were transferred to new OP50-seeded NGM plates containing the same concentration of tunicamycin. Imaging of live worms was performed using a Zeiss Axio Zoom V16 microscope. Next, 4μ8c was obtained as 10 mM stock in DMSO (Glpbio, Montclair, CA). The solution was diluted in the experimental plates to a final concentration of 150 μM. The treatment was then carried out following the same procedure as that described for the tunicamycin treatment. DTT was added to the experimental plates at a final concentration of 5 mM, and the treatment was carried out following the same procedure as that described for the tunicamycin treatment. Survival was scored over 20 hr by counting the numbers of living and dead worms. N-acetylglucosamine (GlcNAc) (Sigma) was dissolved in water at a stock concentration of 500 mM. Worms were grown on OP50-seeded NGM plates until they were injected on the first day of adulthood as described above. After the injections, the worms were transferred to new OP50-seeded NGM plates containing a final concentration of 10 mM GlcNAc. Pumping was scored daily as described above.

Quantitative RT-PCR

Primers for the assessment of ER unfolded protein response (UPR^ER) activation were published previously (Richardson et al. 2010). Following total RNA extraction using the RNeasy Mini Kit (QIAGEN), cDNA was produced using the Superscript II protocol (Invitrogen, Carlsbad, CA). The quantitative RT-PCR (RT-qPCR) was carried out using Bio-Rad (Hercules, CA) CFX96 real-time PCR machines with SYBR Green I for amplification quantification (Sigma) and Platinum Taq polymerase (Invitrogen/Thermo Fisher Scientific). The specificity of the generated products was confirmed via melting curve analysis. For data analysis, the comparative C_T(2^−ΔΔC_T) method (Livak and Schmittgen 2001) was used, where 2^−ΔΔC_T = [(C_Tgene of interest – C_Tinternal control) sample A) – (C_Tgene of interest – C_Tinternal control sample B)]. The expression levels were normalized using the actin (act-1) and tubulin (tbg-1) genes as internal controls, and SDs were calculated as in Schmittgen and Livak (2008). The relative hsp-4 expression levels in Figure S4A were measured using the same method.

Statistics

The statistical methods used to evaluate the data are provided in the figure legends. All of the provided P-values are uncorrected. All experiments were repeated at least three times except for the injection experiments, which were all repeated twice.

Data availability

The authors affirm that all data necessary for confirming the conclusions of this article are represented fully within the article and its figures. Supplemental material available at Figshare: https://doi.org/10.25386/genetics.8326691.

Results

C. elegans is sensitive to persistent cytoplasmic DNA

To develop C. elegans as a model to study organismal effects of foreign cytoplasmic DNA, we devised a method for direct injection of foreign DNA into the cytoplasm of the worm’s intestinal cells. These cells were chosen for two reasons: (1) from a practical standpoint, the cells are large and facilitate reproducible injection, and (2) the intestine seems to be a center of immune function in the worm (Pukkila-Worley and Ausubel 2012; Ermolaeva and Schumacher 2014; Kim and Ewbank 2018; Yuen and Ausubel 2018) and therefore its cells are good candidates for mediating a response to foreign DNA. The intestine consists of a ring of four cells, followed by eight rings consisting of two cells each, which together form a tube structure with a central lumen. To stabilize the injected DNA, we used a classical nuc-1 mutant strain generated in Robert Horwitz’s laboratory (Wu et al. 2000) that is defective in DNase II activity in addition to the wild-type strain. In our first attempt, we injected purified E. coli genomic DNA into one cell of the second intestinal ring using a microinjection system (Figure 1A) (for detailed methods, see Materials and Methods). To monitor the location and success of the injection, the fluorescent compound rhodamine-isothiocyanate dextran (10,000 M_r) was included in the injection mixture, as it is commonly used for this purpose because it is generally inert in cells and has limited diffusion ability (Schmued et al. 1990).

Figure 1

Injection of DNA into intestinal cells causes declines in tissue functionality in DNase II-defective worms. (A) A cartoon showing the approach to intestinal injection. Ph, pharynx; Int I and Int II, intestinal rings I and II; the arrow indicates the approximate location of the injection. (B) Plot of the pharyngeal pumping rates of wt and nuc-1 worms injected with purified UPEC genomic DNA. (C) Plot of the pharyngeal pumping rates of uninjected wt and nuc-1 worms over the first 72 hr after reaching the young adult stage (starting 24 hr after L4). (D) Plot of the pharyngeal pumping rates of wt and nuc-1 worms injected with the fluorescent co-injection marker in the same buffer (rhodamine-dextran) without DNA. (E) Plot of the pharyngeal pumping rates of the same strains injected with synthetic CpG-containing ODN (ODN 2395). (F) Comparison of the sequences of ODN 2395 and the control, ODN 5328 (top). The underlined bases show the CpG sequences in ODN 2395 and the corresponding positions in the control ODN 5328. Plot of the pharyngeal pumping rates of nuc-1 worms injected with either ODN 2395 or ODN 5328 (bottom). Each plot shows the mean ± SD at each time point. The statistical significances of the differences in the trends between the control and experimental groups were assessed using two-way ANOVA (n.s.: P > 0.05, ** 0.01 > P > 0.001, and *** P < 0.001). n.s., not significant; ODN, oligodeoxyribonucleotide; UPEC, uropathogenic E. coli strain; wt, wild-type.

To monitor the consequences of DNA injection on the overall health of the animals, we measured the pharyngeal pumping rate, which is an established and highly sensitive approach for monitoring even small changes in tissue functionality during aging (Bolanowski et al. 1981; Huang et al. 2004; Chow et al. 2006), following induction of DNA damage (Wilson et al. 2017), and during infection (Tan et al. 1999; O’Quinn et al. 2001). In wild-type worms, we did not observe any change in the pumping rate following injection of E. coli genomic DNA; however, the same injection into DNase II-defective worms (due to a mutation in the nuc-1 gene) led to a statistically significant decline in the pumping rate over the course of 72 hr postinjection (Figure 1B). Importantly, such a difference in pumping rates was not observed in either uninjected animals or animals injected with the co-injection marker alone (Figure 1, C and D).

To confirm the phenotype observed with the purified genomic DNA, we used low-molecular weight CpG-rich synthetic ODNs. As shown in Figure 1E, injection of a CpG-containing ODN (ODN 2395) led to a similar pumping rate decline in the DNase II-defective worms. In contrast to the results obtained with E. coli genomic DNA, a slight reduction in the pumping rate of wild-type worms resulted after ODN injection. We attribute this difference to increased resistance of the ODNs to DNase digestion due to phosphorothioate linkages in the ODN backbone.

A common feature shared between E. coli genomic DNA and the synthetic ODNs is the presence of CpG sequences. In higher organisms, CpG-containing DNA introduced by bacterial pathogens can be sensed by innate immunity factor TLR9 to elicit an immune response, and we supposed that C. elegans might also respond specifically to CpG-containing DNA. To test this hypothesis, we compared the outcomes of injecting a CpG-containing ODN (ODN 2395) to those following injection of a non-CpG control ODN (ODN 5328) (Figure 1F, top). As shown in the plot in Figure 1F, injection of the control ODN had a milder effect on the pumping rate in the DNase II-defective background.

Cytoplasmic DNA accumulates in intestinal cells during UPEC infection and results in a systemic health decline

A limitation of our injection experiments was that we could introduce only small amounts of foreign DNA into the intestinal cells, likely preventing the detection of more robust effects on the animals that could result from higher loads of cytoplasmic DNA; therefore, a means of introducing larger amounts of DNA was needed. Some pathogenic E. coli, including some UPEC strains, can invade host cells (Lewis et al. 2016); therefore, we predicted that UPEC infection in the intestinal lumen of a DNase II-defective mutant worm strain might serve as a powerful tool for delivering larger amounts of DNA into the cytoplasm of the intestinal cells. We obtained a clinical UPEC isolate and established an intestinal infection by feeding the bacteria to the worms as their sole food source. To determine whether bacterial DNA accumulated in the cytoplasm of the intestinal cells of the UPEC-infected worms, we stained dissected intestinal cells with the DNA stain DAPI. Confocal imaging (Figure 2A) of stained animals revealed that, in addition to the bacterial DNA in the intestinal lumen (outlined with yellow dashed lines) of both wild-type and DNase II-defective worms, DAPI-stained bodies similar in size to the luminal bacterial DNA bodies preferentially accumulated in the cytoplasm of the intestinal cells of the DNase II-defective worms (some examples are indicated by white arrow heads) (Figure 2, A–C). The observation that both wild-type and DNase II-defective worms contain such bodies suggests that bacterial DNA enters the cytoplasm of the intestinal cells in both worm strains; however, this cytoplasmic DNA persists more robustly in the DNase II-defective background. Notably, similar DAPI-stained bodies were not observed in DNase II-defective animals grown on the nonpathogenic E. coli strain OP50, even when the DAPI signal from intestinal nuclei was increased beyond saturation (Figure S1A). Next, we were curious to investigate whether the persistent foreign DNA introduced via UPEC infection could induce a pumping rate decline in the DNase II-defective worms, as well as whether we could observe other declines in tissue integrity and functionality that occur preferentially in infected DNase II-defective worms. To address these questions, we examined several standard indicators of health in the worms: (1) pharyngeal pumping rate, (2) intestinal tissue integrity, (3) muscle integrity, and (4) reproductive function.

Figure 2

Cytoplasmic DNA preferentially accumulates in the cytoplasm of intestinal cells in UPEC-infected DNase II-defective worms. (A) Top panels: single confocal Z-planes from UPEC-infected wt and nuc-1 worms showing two intestinal cells. The α-tubulin is in green and the DAPI (DNA) is in blue. The yellow lines mark the margin between the lumen and intestinal cells. Middle panels: only the DAPI signal is shown. Bottom panels: enlarged images of the DAPI channel of the same images. The arrows point to representative DAPI-stained cytoplasmic inclusions. (B) Three-dimension reconstructions from the confocal images confirming the position of the DAPI-stained bodies within the cytoplasm of the intestine. Bars, 5 μm (for all images). (C) Quantification of the number of DAPI-staining inclusions in UPEC-infected wt and nuc-1 worms. The data are presented as the mean ± SD. Statistical significance was assessed using an unpaired, two-tailed Student’s t-test (*** P < 0.001). L, intestinal lumen; N, nucleus; UPEC, uropathogenic E. coli strain; wt, wild-type.

The pharyngeal pumping rate was measured (as above) during the first 5 days after UPEC infection. The wild-type and DNase II-defective animals showed similar pumping rates when fed nonpathogenic E. coli  OP50 (Figure 3A). As shown in Figure 3B, both wild-type and DNase II-defective worms infected with UPEC showed more rapid declines in their pumping rates compared with the pumping rates of the animals mock infected with OP50; however, this effect was much more severe in the DNase II-defective animals. To exclude the possibility of strain-specific effects, we confirmed this effect using two standard nuc-1 alleles: e1392 and n887 (Figure 3, A and B). The more pronounced decline in pumping rate following infection in comparison to that due to single-cell DNA injections is likely due to a difference in the amount of cytoplasmic DNA delivered into the intestinal cells via each approach. The injected worms, which we calculated received less than 8 E. coli chromosome equivalents per injection in only a single cell, showed a slower pumping rate decline compared with that of the infected worms, which accumulated vastly more bacterial DNA in multiple intestinal cells.

Figure 3

UPEC infection induces a functional decline in the pharynx. Pharyngeal pumping rates of OP50 mock-infected (A) and UPEC-infected (B) wt and nuc-1 worms. Measurements for each individual worm are shown with the mean ± SD (C). Representative DIC images of pharynxes from OP50 mock-infected and UPEC-infected wt and nuc-1 worms on day 4 [corresponding to day 4 in (A and B)]. The pharynxes are highlighted in yellow. Bar, 100 μm. (D) Quantification of the combined surface areas of the pharyngeal metacorpora and posterior bulbs in OP50 mock-infected and UPEC-infected wt and nuc-1 worms on day 4. Tukey plots show the interquartile range and the median (see the Materials and Methods and Figure S1B). Statistical significance was assessed using unpaired, two-tailed Student’s t-tests; ** P < 0.01, indicates no significant difference. UPEC, uropathogenic E. coli strain; wt, wild-type.

Changes in pharyngeal pumping in C. elegans can be attributed either to behavioral changes [reviewed in Luedtke et al. (2010)] or to loss of tissue functionality. Therefore, it was important to confirm that the integrity of the pharynx was preferentially compromised in the UPEC-infected DNase II-defective worms. To address this point, we imaged worms using DIC microscopy 4 days postinfection (representative images are provided in Figure 3C) and then measured the combined surface area of the pharyngeal metacorpus and posterior bulb in each worm (explained in detail in Figure S1B). Indeed, we observed a pronounced and statistically significant decrease in pharyngeal surface area in the UPEC-infected DNase II-defective animals compared to that in the DNase II-defective animals mock infected with OP50 (Figure 3D). No significant difference was observed under the same conditions in the wild-type worms. These results demonstrate that the pumping rate declines in our experimental system likely result from pharyngeal atrophy and not from behavioral changes; therefore, we propose that the pumping rate is a reliable and quantitative metric for tissue functionality in our experimental system.

We next wanted to determine whether other tissues were also affected by UPEC infection. To examine the overall integrity of the intestine, we stained worms with the dye NR, which accumulates in punctate intracellular acidic compartments in healthy intestinal cells (Clokey and Jacobson 1986; O’Rourke et al. 2009). Dissolution of these punctate bodies is a hallmark of necrotic cell death and can be used as a marker for tissue integrity during pathogen infection (Zou et al. 2014). The intestines of representative wild-type and DNase II-defective worms mock infected with OP50 showed normal accumulation of NR in well-resolved punctate bodies (Figure 4A). In contrast, after 72 hr of UPEC infection, these compartments preferentially lost their punctate definition in the DNase II-defective animals (Figure 4B).

Figure 4

UPEC infection induces a loss of structural integrity in the intestine of DNase II-defective worms. (A) Nile red staining of age-matched wt and nuc-1 worms, 72 hr after OP50 mock infection (A) or UPEC infection (B) at the L4 stage. The images show the anterior (head) portion of the intestine. For each condition, a DIC image and an image of the Nile red channel are provided with a gray-scale image of the Nile red channel that highlights the punctate structures in the intestine. Bars, 100 μm. UPEC, uropathogenic E. coli strain; wt, wild-type.

As shown in Figure 5A, UPEC-infected DNase II-defective worms showed reduced crawling speeds compared to those of UPEC-infected wild-type worms over the first 96 hr postinfection. This effect could have been caused by either a behavioral change or by muscle degeneration. To distinguish these possibilities, we used phalloidin staining to examine the integrity of the F-actin organization in the body-wall muscles used for crawling (Wulf et al. 1979; Francis 1985). Consistent with a decline in tissue integrity rather than a behavioral change, the phalloidin staining revealed more pronounced F-actin disorganization and muscle atrophy in the UPEC-infected DNase II-defective worms, while the muscle cells of wild-type and DNase II-defective animals mock infected with OP50, as well as UPEC-infected wild-type worms, were similar (Figure 5B).

Figure 5

UPEC infection causes muscle degeneration and reproductive defects in infected DNase II-defective animals. (A) Relative crawling speeds of UPEC-infected wt and nuc-1 worms. (B) Representative images of phalloidin-stained body-wall muscles, in mock- or UPEC-infected wt and nuc-1 worms. The white dashed outlines show individual muscle cells. Bars, 20 μm. (C) Fecundity, scored as the number of eggs laid during the reproductive period, of wt and nuc-1 worms mock-infected with OP50 or infected with UPEC. The means ± SD are given, and the statistical significance was assessed via unpaired, two-tailed Student’s t-tests; ** P < 0.01. n.s., not significant; UPEC, uropathogenic E. coli strain; wt, wild-type.

Finally, we examined reproductive function by measuring the fecundity of infected worms. While wild-type and DNase II-defective worms have nearly identical brood sizes when mock infected with OP50 (Figure 5C, left), UPEC-infected DNase II-defective worms showed decreased egg production, especially during the first 2 days of egg laying (Figure 5C, right). Taken together, these results show that UPEC-infected DNase II-defective worms suffer from systemic declines in tissue integrity and functionality, which are largely absent in OP50 mock-infected and UPEC-infected wild-type worms, and OP50 mock-infected DNase II-defective worms. These findings suggest that functional DNase II protects the animals from detrimental effects caused by the accumulation of foreign DNA in the cytoplasm of the intestinal cells.

Cytoplasmic DNA-induced tissue degeneration requires the FSHR-1 innate immune signaling pathway

One explanation for the tissue degeneration phenotypes described above is aberrant induction of a toxic uncontrolled immune response to the cytoplasmic DNA, for which a precedent has been recently established following Pseudomonas aeruginosa infection (Cheesman et al. 2016). If this idea were true, then abrogating this response should alleviate the tissue declines in worms infected with UPEC or injected with ODNs. To address this possibility, we tested for the involvement of the two primary immune signaling pathways in C. elegans: the p38/PMK-1-mediated pathway and the FSHR-1-mediated pathway.

The leucine-rich repeat (LRR)-containing protein FSHR-1 regulates immune responses to several different pathogens (Powell et al. 2009). We tested for a role of FSHR-1 in our experimental system via RNAi-mediated depletion of FSHR-1 in wild-type and DNase II-defective worms. To validate the RNAi, we first ensured that it did not affect the baseline pumping rates when the worms were grown on OP50 (Figure S2A). We also confirmed the efficacy of the RNAi by showing that it induced sensitivity to the well-characterized C. elegans pathogen P. aeruginosa strain PA14 (Tan et al. 1999) (Figure S2B). Worms were then infected with UPEC (Figure 6, A and B) or injected with CpG ODNs (Figure 6, C and D), and the pharyngeal pumping rates were scored for the indicated times. At the early time points post-UPEC infection, both wild-type; fshr-1(RNAi) and nuc-1; fshr-1(RNAi) worms showed decreased pumping rates, consistent with the possibility that FSHR-1 controls a beneficial response to the pathogen (Figure 6, A and B). FSHR-1 depletion had no effect on UPEC-infected wild-type worms at 72 and 96 hr postinfection (Figure 6A). In contrast, UPEC-infected nuc-1; fshr-1(RNAi) worms showed robust improvements in their pumping rates at 72 and 96 hr postinfection (Figure 6B). In fact, FSHR-1 depletion restored the pumping rates of the infected DNase II-defective worms to those of the infected wild-type worms at the later time points (Figure 6B), suggesting that an FSHR-1-mediated response is likely the sole cause of the pumping rate declines in the infected DNase II-defective animals. FSHR-1 depletion also eliminated the pumping rate decline caused by CpG ODN injection in the DNase II-defective worms (Figure 6D), while it had no effect on the pumping rate in the wild-type worms (Figure 6C).

Figure 6

Tissue degeneration due to cytoplasmic DNA depends entirely on FSHR-1. (A and B) Plots of the pharyngeal pumping rates of wt and nuc-1 worms infected with UPEC, with and without RNAi against fshr-1. (C and D) Plots of the pharyngeal pumping rates of CpG ODN-injected wt and nuc-1 worms treated with RNAi against fshr-1. In all cases, the error bars show the mean ± SD. For (A and B), the statistical significance was assessed via unpaired, two-tailed Students t-tests. For (C and D), the statistical significance of the differences in the trends between the control and experimental groups were assessed using two-way ANOVA. In all cases, n.s. P > 0.05, * P < 0.01, ** 0.01 > P > 0.001, and *** P < 0.001. n.s., not significant; ODN, oligodeoxyribonucleotide; RNAi, RNA interference; UPEC, uropathogenic E. coli strain; wt, wild-type.

The C. elegans p38 MAP kinase homolog PMK-1 is another important mediator of the worm’s innate immune responses to a variety of pathogens (Kim et al. 2002; Kamaladevi and Balamurugan 2015). While our data in Figure 6 suggest that our phenotypes can be entirely attributed to an FSHR-1-dependent effect, we also wanted to confirm that PMK-1 is not involved. To this end, we depleted PMK-1 in wild-type and DNase II-defective worms via RNAi. We validated the pmk-1 RNAi as described in the Materials and Methods and as shown in Figure S3, A–C. As expected, RNAi-mediated PMK-1 depletion did not rescue the pumping rate decline observed in DNase II-defective worms injected with CpG ODNs (Figure S3D). Thus, we conclude that it is unlikely that p38/PMK-1 plays a role in the tissue degeneration phenotypes reported here.

Chronic UPEC infection leads to a disruption of protein homeostasis

We next set out to clarify the basis of the increased sensitivity of the DNase II-defective worms to persistent cytoplasmic DNA. Immunity-associated stress responses in C. elegans result in the induction of a large number genes encoding putative secreted peptides (Shivers et al. 2008), and dysregulation of this type of response can be detrimental to C. elegans (Cheesman et al. 2016). One hypothesis to explain the preferential tissue degeneration observed in DNase II-defective worms is that the persistent cytoplasmic DNA leads to a prolonged upshift in the production of secreted proteins that strains the animal’s protein quality control mechanisms, to the point that proteotoxic effects cause cellular damage. The plausibility of this hypothesis is supported by the previous observation that the UPR^ER is required by worms to tolerate a prolonged innate immune response to P. aeruginosa (Richardson et al. 2010) and that the UPR^ER may assist in the activation of the immune response (Dai et al. 2015). To test this hypothesis, we examined several indicators of the proteostasis status of the worms.

We used TEM to directly examine the structure of the ER and the ribosome distribution on the rough ER in the intestinal cells. As shown in Figure 7A (left), in wild-type worms, UPEC infection (bottom) did not alter the overall structure of the ER, as it appeared similar to that of OP50 mock-infected worms (top). In contrast, the ER of the UPEC-infected DNase II-defective worms (Figure 7A, right) showed enlarged cisternal spaces (bottom), a widely accepted hallmark of ER stress in C. elegans (Richardson et al. 2010; Tillman et al. 2018). We quantified this phenotype and report it as the ER cisternae surface area (see the Materials and Methods) in Figure 7B. Interestingly, the OP50 mock-infected DNase II-defective worms showed mild enlargement of the cisternal spaces compared to those of the OP50 mock-infected wild-type worms (Figure 7B). This effect could be explained by the previously reported observation that DNase II-defective worms induce a mild but persistent immune response, even in the absence of a pathogen (Yu et al. 2015). Such a chronic response could place a baseline strain on the ER that is reflected by this phenotype. On an even finer scale, there was clear disruption of the regular ribosome spacing around the rough ER cisternae in the UPEC-infected DNase II-defective worms (Figure 7A, right). Some examples of regular spacing are shown by the blue arrows, and some examples of disruptions are indicated by the red arrows. We quantified this phenotype and report it in Figure 7C as the ribosome distribution index (for details, see the Materials and Methods).

Figure 7

UPEC infection leads to ER disruption in DNase II-defective worms. (A) Transmission electron micrographs showing the ER in OP50 mock-infected and UPEC-infected wt (left) and nuc-1 (right) worms. The magnifications are indicated. Bars, 500 nm. The yellow boxes highlight the areas shown with higher magnification on the right. The open blue arrows indicate representative regions of the ER with normal ribosome distribution, and the filled red arrows indicate representative regions with disrupted ribosome distribution. (B) Plot showing a quantification of the ER cisternal surface areas. (C) Plot showing the ribosome distribution index. For (B and C), the Tukey plots show the interquartile range and the median, and the median and sample size is given for each condition. The statistical significance was assessed via unpaired, two-tailed Students t-tests; n.s. P > 0.05, * P < 0.01, ** 0.01 > P > 0.001, *** P < 0.001. n.s., not significant; UPEC, uropathogenic E. coli strain; wt, wild-type.

To further confirm the presence of a proteostasis defect via an entirely different approach, we tested the sensitivity of the worms to the proteotoxic agent dithiothreitol (DTT) using a well-established assay (e.g., Labbadia and Morimoto 2015; Roux et al. 2016). The rationale for this approach was that animals with baseline proteostatic strains due to endogenous sources are more sensitive to DTT than are animals with normal proteostasis. We infected L4-stage larvae with UPEC and transferred them to plates containing 5 mM DTT 72-hr later. We then scored their survival over 20 hr (the experiment is outlined in Figure 8A). As expected, the UPEC-infected DNase II-defective animals were profoundly sensitive to DTT, while the UPEC-infected wild-type animals showed only a mild effect (Figure 8B). Taken together, the imaging data and the results of our proteotoxicty assay independently and consistently confirm the presence of a proteostatic defect that is most severe in the UPEC-infected DNase II-defective animals.

Figure 8

UPEC infection causes proteostasis defects in DNase II-defective worms. (A) A diagram of the timeline for the experiment shown in (B). Worms were infected at the L4 stage and then transferred to DTT 72-hr later. The survival was scored over the next 20 hr (B) Kaplan–Meier survival plot of DTT-treated worms. The statistical significance was calculated via the log-rank test; ** 0.01 > P > 0.001 and **** P < 0.0001. (C) Schematic of the C. elegans UPR^ER pathway. The abbreviations correspond to those in (D). (D) Plot showing the relative levels of the different forms of the xbp-1 transcript 24 and 48 hr postinfection: xpb-1 total, xbp-1 pre-mRNA (unspl.), xbp-1 spl., and the XBP-1 regulatory target hsp-4. The data are shown as the mean ± SD. No statistically significant differences were observed based on two-tailed Student’s t-tests. unspl., unspliced; UPEC, uropathogenic E. coli strain; spl., spliced; UPR^ER, ER unfolded protein response; wt, wild-type.

Physical disruption of the ER would likely be associated with defects in protein expression and folding, and such effects normally lead to compensatory activation of the UPR^ER. To evaluate the status of the UPR^ER activation in our worms, we measured the relative levels of the unspliced and spliced forms of the xbp-1 transcript (Yoshida et al. 2001). Splicing of the xbp-1 primary transcript by the specific endoribonuclease IRE-1 leads to the expression of the active form of XBP-1, a transcription factor that acts as the central activator of the UPR^ER (outlined in Figure 8C). We collected RNA from UPEC-infected wild-type and DNase II-defective worms, 24 and 48 hr after UPEC infection, and, as shown in Figure 8D, we found that the level of spliced xbp-1 transcript was not significantly increased under any condition. Importantly, the expression level of hsp-4, a prototypical XBP-1 target gene (Figure 8, C and D), was also not induced. We conclude from these results that a response specific to the DNase II-defective animals during UPEC infection strains the protein quality control pathways, ultimately leading to proteotoxic effects that result in the tissue degeneration observed in DNase II-defective animals. These results suggest two mutually exclusive possibilities: (1) despite the ER disruption, there was no compensatory UPR^ER activation, or (2) that the response is activated but is not sufficiently sustained over the course of the infection to provide a long-term protective effect. These results are consistent with the previously reported toxicity of an aberrantly activated innate immune response in P. aeruginosa-infected C. elegans (Cheesman et al. 2016). We next explored therapeutic interventions that might alleviate the proteotoxic effects of the infection and intestinal DNA injection.

Therapeutic improvement of ER protein homeostasis alleviates the tissue functionality decline caused by persistent cytoplasmic DNA

If a chronic, unresolved defect in ER homeostasis due to a failure to induce or sustain the UPR^ER does indeed drive the tissue degeneration in the presence of persistent cytoplasmic DNA, we wondered if we could exogenously drive UPR^ER activation or enhance the protein-folding capacity in the ER to alleviate the tissue degeneration phenotypes. We used three very different approaches that could lead to compensatory proteostasis assurance responses: (1) ectopic activation of the UPR^ER via mild, subtoxic, chemically induced protein-folding stress (Figure 9A); (2) therapeutic enhancement of protein-folding capacity in the ER (Figure 9D); and (3) activation of the UPR^ER independent of protein-folding stress (Figure S4D). We first used a continuous low-dose (0.5 μg/ml) treatment with tunicamycin during the course of UPEC infection and after DNA injection (Figure 9, A–C). Tunicamycin inhibits protein glycosylation, leading to UPR^ER activation (King and Tabiowo 1981) (Figure 9A). Wild-type worms treated with tunicamycin during UPEC infection showed no improvement in pharyngeal pumping over 96 hr of infection, although they still showed the expected infection-dependent health decline (see Discussion) (Figure 9B, left). In stark contrast, tunicamycin treatment of UPEC-infected DNase II-defective worms reverted their decreased pumping rate to a level comparable to that of the infected wild-type worms (compare the open circles in Figure 9B, left and right), while tunicamycin treatment had no effect in OP50 mock-infected DNase II-defective worms (Figure S4F). Tunicamycin treatment also completely eliminated the pumping rate decline caused by ODN injection into the intestine of DNase II-defective animals (Figure 9C). Based on these results, we concluded that exogenous, therapeutic UPR^ER activation could prevent the proteotoxic effects of the chronic response to cytoplasmic DNA.

Figure 9

Restored protein homeostasis alleviates the tissue functionality declines in DNase II-defective worms caused by UPEC infection and ODN injection. (A) Schematic of UPR^ER activation by Tun. (B) Plots of the pharyngeal pumping rates of UPEC-infected wt and nuc-1 worms infected with UPEC, with and without treatment with 0.5 μg/ml Tun. (C) Plot of the pharyngeal pumping rates of ODN-injected nuc-1 worms, with and without treatment with 0.5 μg/ml Tun. (D) Schematic of ER stabilization by GlcNAc. (E) Plots of the pharyngeal pumping rates of UPEC-infected wt and nuc-1 worms, with and without treatment with 10 mM GlcNAc. (F) Plot of the pharyngeal pumping rates of ODN-injected nuc-1 worms with and without treatment with 10 mM GlcNAc. (G) Outline of the experiment shown in (H and I), where the colors correspond to data points in the following plots. Plots of the pharyngeal pumping rates of UPEC-infected wt (H) and nuc-1 (I) worms, with and without treatment with 0.5 μg/ml Tun at the time points shown in (G). All plots show the mean ± SD. For (B, E, H, and I), the statistical significance was assessed via unpaired, two-tailed Student’s t-tests; n.s. P > 0.05, * P < 0.01, ** 0.01 > P > 0.001, and *** P < 0.001. For (C and F), the statistical significance was assessed via two-way ANOVA (***P < 0.001). GlcNAc, N-acetylglucosamine; n.s., not significant; Tun, tunicamycin; ODN, oligodeoxyribonucleotide; UPEC, uropathogenic E. coli strain; spl., spliced; UPR^ER, ER unfolded protein response; wt, wild-type.

To further verify this conclusion, we tested two fundamental aspects of the underlying biology: (1) that the low dose of tunicamycin does indeed lead to UPR^ER activation (Figure S4A); and (2) that the IRE-1 axis is required for the beneficial effect of tunicamycin (Figure S4, B and C). To confirm the UPR^ER activation, we measured the relative levels of hsp-4 expression in untreated and treated UPEC-infected DNase II-defective worms via RT-qPCR. As shown in Figure S4A, treatment with 0.5 μg/ml led to a twofold increase in the relative hsp-4 mRNA abundance 48 hr postinfection. To confirm the involvement of the IRE-1 axis, we used the specific IRE-1 inhibitor 4μ8c (CAS 14003-96-4), the effectiveness of which in inhibiting the activation of the C. elegans UPR^ER in response to ER stress has been confirmed and carefully characterized (Roux et al. 2016). We chose to use chemical inhibition for two important reasons. First, the UPR^ER has important roles during larval development; thus, application of the inhibitor only during the interval of interest avoids intractable effects from inhibiting the UPR^ER during development (Richardson et al. 2010). Second, 4μ8c specifically inhibits the ribonuclease domain of IRE-1, which prevents xbp-1 mRNA splicing, but does not affect its kinase activity, which is likely involved in functions uncoupled from xbp-1 splicing (Safra et al. 2013; Roux et al. 2016). In this experiment, we infected DNase II-defective L4 larvae with UPEC, and treated them with tunicamycin and 4μ8c, alone and in combination, and then scored the pumping rates 48-hr later (see the experimental outline in Figure S4B). As shown in Figure S4C, we found that 4μ8c alone had no statistically significant effect on untreated infected worms, while it completely abrogated the beneficial effect of tunicamycin treatment. From these experiments, we concluded that the beneficial effect of tunicamycin in UPEC-infected DNase II-defective worms likely depends on UPR^ER activation via the IRE-1 pathway.

Because tunicamycin interferes with post-translational processing and can be toxic under high concentrations (King and Tabiowo 1981; Denzel et al. 2014), we wanted to further confirm that its beneficial effect was not due to ER-independent factors. To this end, we sought an alternative experimental strategy to activate the UPR^ER without affecting the protein homeostasis systems. Loss of the S-adenosylmethionine synthase SAMS-1 leads to induction of the UPR^ER via induction of lipid disequilibrium in the ER with no induction of protein-folding stress (Hou et al. 2014) (Figure S4D). Mirroring the tunicamycin results, RNAi-mediated sams-1 depletion rescued the pumping rate decline in UPEC-infected DNase II-defective worms, and provided little or no benefit to UPEC-infected wild-type worms (Figure S4E). While sams-1 has been previously assigned a function in immunity (Ding et al. 2015), an effect of sams-1 depletion on baseline immunity is unlikely since it did not alter the sensitivity of the wild-type worms to UPEC infection. From the data in Figure 8, A–C and Figure S4E, we concluded that mild ectopic UPR^ER induction can relieve the proteotoxic effects caused by an upshift in protein synthesis associated with the worm’s stress response pathways.

Tunicamycin treatment and SAMS-1 depletion activate the UPR^ER via the disruption of normal cellular processes (protein glycosylation and lipid homeostasis, respectively). We next wondered whether protection of ER function without such side effects could also provide protection against tissue degeneration in UPEC-infected or ODN-injected DNase II-defective worms. Treatment of C. elegans with GlcNAc slows aging and suppresses the proteotoxic effects of protein aggregation by augmenting the activity of the hexosamine pathway, which enhances protein folding and improves ER homeostasis (Denzel et al. 2014) (Figure 9D). To test whether GlcNAc treatment protects DNase II-defective worms from the detrimental effects of cytoplasmic DNA, we treated DNase II-defective animals with 10 mM GlcNAc during UPEC infection or following intestinal ODN injection. As shown in Figure 9, E and F, sustained GlcNAc treatment of DNase II-defective animals following UPEC infection and ODN injection specifically eliminated the pumping rate declines in the DNase II-defective animals; thus, enhancement of the ER protein-folding capacity via this nontoxic pharmacological approach offers a similar benefit to those offered by tunicamycin treatment and SAMS-1 depletion.

We were next interested in whether tunicamycin treatment could be used as a curative intervention to treat the effects of ongoing infection. To test this possibility, we transferred UPEC-infected worms from plates containing only UPEC to identical plates supplemented with 0.5 μg/ml tunicamycin at regular intervals up to 72 hr postinfection and measured the pumping rates on the following days up to 120 hr postinfection (see the experimental outline in Figure 9G). DNase II-defective worms transferred to tunicamycin plates as late as 72 hr postinfection showed robust and statistically significant improvements in their pumping rates up to 120 hr postinfection (Figure 9I). No obvious differences in the pumping rates of the treated wild-type worms (either uninfected or infected) were observed (Figure 9H). These results further support the conclusion that infected wild-type worms do not experience the same unresolved ER stress present in the UPEC-infected DNase II-defective worms.

As discussed above, the pumping rate can be influenced both by changes in tissue functionality and behavioral effects; thus, it was important to confirm that the beneficial effects of tunicamycin were not restricted to the pumping rate decline. Thus, we examined the effects of tunicamycin treatment on other tissue degeneration phenotypes described above. The most striking demonstration of the beneficial effect of tunicamycin was provided by imaging of populations of treated and untreated UPEC-infected worms 72 hr postinfection. The arrow heads in Figure 10A, bottom left) indicate representative UPEC-infected DNase II-defective worms suffering from severe tissue degeneration. Such worms were not observed in the UPEC-infected untreated and treated wild-type worms Figure 10A, left and right), as these worms showed the typical sinusoidal shape of healthy worms. Upon tunicamycin treatment, the UPEC-infected DNase II-defective worms resembled the UPEC-infected wild-type worms, as the full-body degeneration was generally absent (Figure 10B, bottom right). To quantify this effect, we assigned the worms to three groups (see examples for wild-type and DNase II-defective worms in Figure 10, C and D): (1) sinusoidal, i.e., worms with a healthy appearance; (2) intermediate, i.e., worms with milder full-body degeneration; and (3) defective, i.e., worms with severe full-body degeneration. As shown in Figure 10E, tunicamycin treatment lead to a decrease in the number of worms in the defective class in the UPEC-infected DNase II-defective worms but offered no advantage to UPEC-infected wild-type worms, which generally fell into the sinusoidal class both with and without tunicamycin treatment. As shown above in Figure 5C, UPEC-infected DNase II-defective worms showed a defect in reproductive capacity. We tested whether tunicamycin treatment could improve the reproductive ability of these worms, and indeed there was a beneficial effect on the egg laying of UPEC-infected DNase II-defective animals (Figure 10F). Taken together, these results confirm that tunicamycin offers a robust and specific therapeutic benefit for UPEC-infected DNase II-defective worms that can likely be attributed to improved proteostasis. Based on the combined results from Figure 9 and Figure 10, we conclude that mild UPR^ER induction can provide a protective effect even during a chronic infection.

Figure 10

Tunicamycin treatment improves overall tissue stability and functionality in UPEC-infected worms. Images of populations of UPEC-infected wt (A) and nuc-1 (B) worms, with (left) and without (right) tunicamycin treatment. The arrows indicate representative worms suffering from full-body tissue degeneration. (A and B) Bar, 1 mm. Representative images of wt (C) and nuc-1 (D) worms, showing the sin, int, and def phenotypes used for the quantification in (E). (C and D) Bar, 250 μm. (E) Quantification of the phenotypes shown in (C and D). The statistical significance was determined via Pearson’s chi-square test. The P-values are provided in the figure. (F) Egg-laying assay for wt (circle markers) and nuc-1 (square markers) worms, mock infected with OP50 (left) or infected with UPEC (right), with (open markers) and without (closed markers) Tun treatment. The data are given as the mean ± SD of the number of eggs laid per worm over a 5-hr period. def, defective; int, intermediate; sin, sinusoidal; Tun/tun, tunicamycin; UPEC, uropathogenic E. coli strain; wt, wild-type.

Discussion

Many open questions remain about the mechanisms of sensing and responding to cytoplasmic DNA. Perhaps more enigmatic are the mechanisms through which chronic innate immune responses lead to tissue damage. Furthermore, our repertoire of therapeutic options for alleviating these detrimental outcomes is limited. In this study, our goal was to establish C. elegans as a simple metazoan model to further explore how animals respond to cytoplasmic DNA, and to better understand the effects of chronic activation of the downstream responses at the cellular and organismal levels. Our ultimate goal was to develop interventions to preserve and improve tissue functionality under such conditions.

The nematode innate immune response has been extensively investigated, and several key regulators that mediate systemic responses to bacteria and fungi have been well documented (Kim et al. 2002; Powell et al. 2009; Ermolaeva and Schumacher 2014; Kim and Ewbank 2018). However, what is entirely lacking in the C. elegans field are data on whether the worm can sense and respond to cytoplasmic DNA. None of the known canonical DNA-sensing pathways identified in higher eukaryotes are conserved in C. elegans. While the C. elegans genome does encode one TLR (TOL-1) (Pujol et al. 2001), it appears to have a limited role in immunity (Tenor and Aballay 2008; Galbadage et al. 2016; Rangan et al. 2016; Battisti et al. 2017); furthermore, the immune functions of TOL-1 may operate mainly via a developmental function (Brandt and Ringstad 2015). The absence of these systems suggests that C. elegans may possess alternative and novel pathways to fulfill this important defense function. In this study, we demonstrated that C. elegans can indeed mount a response to foreign cytoplasmic DNA, and that when this response is chronically activated it can lead to severe multisystem declines in tissue stability and functionality. After identifying the subcellular effects responsible for these declines, we successfully developed simple interventions that alleviate these effects, which might also be amenable for use in higher organisms, including humans.

Upon the introduction of foreign DNA into the cytoplasm of worm intestinal cells, either by direct microinjection (Figure 1) or via infection with a clinically isolated UPEC strain (Figure 2), we observed only a mild effect after UPEC infection and almost no effect after DNA injection in the wild-type worms. The pathogenic effects of the UPEC strain are likely due to adverse effects of the intestinal colonization by the bacteria or incomplete host-mediated immunity. However, a different story emerged in a strain lacking DNase II, encoded by nuc-1 in C. elegans. The introduction of pathogenic DNA into the cytoplasm of the intestinal cells in these worms led to profound declines in tissue stability and functionality that worsened over the course of several days, particularly in the intestine (Figure 4). An unexpected outcome of the introduction of foreign DNA was that tissues distal to the intestinal, including the pharynx (Figure 1 and Figure 3), the muscles (Figure 5, A and B), and the reproductive system (Figure 5C), were severely affected. One explanation for these phenotypes is that the expression levels of many secreted factors increase during the nematode innate immune response (Kato et al. 2002; Mallo et al. 2002; Schulenburg et al. 2004) and that these factors can, when chronically expressed, have detrimental systemic effects. We propose that these effects are reminiscent of the tissue damage that occurs during chronic inflammatory responses in mammals, including humans.

Dysregulation of the p38/PMK-1-mediated response to P. aeruginosa, in the form of chronic induction, is toxic to worms (Cheesman et al. 2016). Moreover, in developing worms, the UPR^ER is required for survival during P. aeruginosa infection (Richardson et al. 2010). These observations prompted us to investigate whether the chronic response to pathogenic DNA could disrupt protein homeostasis. To explore this question, we used TEM to examine the ER structure (Figure 7), and molecular and cellular techniques to examine the function of the UPR^ER (Figure 8 and Figure S4). We observed that not only was the superstructure of the ER disrupted [a previously characterized feature of impaired protein homeostasis (Richardson et al. 2010)], but that the UPR^ER was also not active during the chronic exposure to cytoplasmic DNA. The ability of worms to activate the UPR^ER appears to be stronger in young animals [REF]; therefore, at the time points used in our assays, the challenges posed by DNA injection and UPEC infection might be insufficient to activate a robust response. Another possible explanation, which we favor, is that the UPR^ER may be activated during the initial phase of the response, but cannot be maintained when the immune response becomes chronic. The lack of this ongoing compensatory response, as well as the burden of the enhanced gene expression, might lead to proteotoxic effects that form the subcellular basis for the observed tissue degeneration. To further support the contribution made by enhanced gene expression due to a genetically controlled immune response, we depleted two of the key innate immune response regulators (FSHR-1 and PMK-1) and then challenged the worms with pathogenic cytoplasmic DNA. Loss of the G protein-coupled receptor FSHR-1 (Powell et al. 2009) fully alleviated the phenotype in the DNase II-defective worms to the level of the wild-type worms under the same conditions (Figure 6). Through this experiment, we confirmed that chronic activation of the FSHR-1-mediated immune response indeed underlies the tissue degeneration phenotype entirely, and we propose that FSHR-1 is the likely mediator of the response to pathogenic cytoplasmic DNA. These conclusions were further supported by the finding that loss of PMK-1, at least via RNAi-mediated depletion, in the DNase II-defective worms had no beneficial effect on the tissue degeneration phenotype (Figure S3). Given the structure of the LRRs in FSHR-1, which are also present in mammalian TLRs, it is conceivable that this immune signaling component could function as direct sensor of cytoplasmic DNA; however, this speculation requires further experimental confirmation.

We also explored interventions that, if successful in alleviating these detrimental effects, would substantiate the functional role of the UPR^ER and potentially precipitate novel anti-inflammatory therapeutic routes. Based on our observation of ER disruption and a lack of sustained UPR^ER induction (Figure 9 and Figure S4), we used three mechanistically independent approaches that improve protein homeostasis with the goal of reducing the toxicity of the chronic response. We exposed the worms to a low, subtoxic level of tunicamycin, a chemical that robustly activates the UPR^ER by interfering with N-linked glycosylation of newly translated proteins. This treatment induced a profound rescue of the tissue degeneration phenotype in the DNase II-defective worms exposed to pathogenic DNA (Figure 9 and Figure 10). We suggest that this treatment constitutively activates and maintains the UPR^ER during the entire course of the chronic response, and that this ongoing compensatory response rescues the tissue degeneration phenotype. As an alternative genetic approach to activate the UPR^ER, we depleted the S-adenosylmethionine synthase SAMS-1 via RNAi (Figure S4, D and E). Loss of SAMS-1 is known to enhance protein homeostasis without inducing proteotoxic stress (in contrast to tunicamycin). Similarly, SAMS-1 depletion also rescued the tissue degeneration phenotype. As both tunicamycin exposure and SAMS-1 depletion inhibit fundamental cellular processes to induce the UPR^ER, we wanted to explore a completely nontoxic approach. Treatment with GlcNAc enhances protein homeostasis by increasing the protein-folding capacity of the ER. Similar to the other two interventions, GlcNAc treatment also rescued to the time-dependent tissue degeneration phenotype (Figure 10, D and E). Finally, in addition to the studies by Richardson et al. (2010) and Dai et al. (2015) cited above, protein homeostasis has been further identified to be important component in the response to infection in C. elegans (Richardson et al. 2010). Furthermore, enhanced ER homeostasis has recently been shown to enhance the animals’ tolerance of the innate immune response (Tillman et al. 2018). These outside observations further substantiate our model, which is outlined in Figure 11 (see the discussion in the figure legend).

Figure 11

Model for the tissue functionality declines caused by persistent cytoplasmic DNA. (A) Cytoplasmic DNA triggers an immune response via FSHR-1 that provides early protection against infection; however, when the foreign DNA persists as a result of DNase II-deficiency, the immune response becomes chronic and results in an inflammation-like response that leads to disruption of the ER, leading to declines in tissue functionality and ultimately reduced survival. Therapeutic activation of the ER unfolded protein response via tunicamycin treatment or augmentation of protein folding by GlcNAc can rescue these tissue declines. (B) Both wild-type and DNase II-defective animals suffer from general pathogenic effects of UPEC infection, represented by the health decline marked with “x.” The UPEC-infected DNase II-defective worms suffer from an additional decline marked by “y,” which is absent in the wild-type animals. Our novel therapeutic interventions can rescue the y component, but not the x component, supporting the conclusion that the DNase II-dependent phenotype can be partitioned into effects shared with the wild-type worms, which are likely DNA-independent, and effects that only occur in the absence of DNase II (NUC-1), which are likely due to specific effects of undigested cytoplasmic DNA. GlcNAc, N-acetylglucosamine; ODN, oligodeoxyribonucleotide; UPEC, uropathogenic E. coli strain.

We believe that the identification of therapeutic approaches that alleviate the outcomes of the chronically activated innate immune response may have the most importance from the perspective of human health. Treatment with a low concentration of tunicamycin [which has been shown to be well tolerated by mice (Wang et al. 2015)], or GlcNAc, had robust protective effects in the DNase II-defective worms following UPEC infection or intestinal DNA injection. The dependence of this rescue on the absence of DNase II, together with our ability to entirely rescue the DNase II phenotype to the wild-type level via disruption of the FSHR-1 innate immune signaling pathway, strongly supports the idea that these outcomes are, indeed, a direct result of a systemic response to the pathogenic DNA (Figure 11). Thus, we believe that we have demonstrated for the first time that C. elegans can mount a response to cytoplasmic DNA. Furthermore, we propose that the tunicamycin-induced protection occurs via low-level UPR^ER activation (Figure S4, A–C), and that the enhanced protein-folding capacity afforded by GlcNAc helps to maintain ER functionality. The outcome of each intervention is an amelioration of the proteotoxic effects resulting from the chronic immune response, without alleviation of the response itself. These findings may have important implications for the understanding of DNASE2-related inflammatory conditions in humans. ER stress has been shown to be a mediator of kidney injury, as reviewed in Inagi (2009), and tunicamycin injection reduced the severity of inflammatory kidney disease in a mouse model for mesangioproliferative glomerulonephritis (Inagi et al. 2008). These observations provide additional support for our hypothesis that mild UPR^ER induction may provide protection in the face of the significant upshifts in protein expression involved in inflammatory responses. Therefore, we propose that therapeutic protection of ER functionality in patients with inflammatory conditions, such as rheumatoid arthritis and inflammation-induced heart failure, may offer a similar benefit.

Innate immune responses also put significant strain on the host’s cells, and the ensuing inflammation can result in tissue damage (Wallach et al. 2014). Inflammation can disrupt tissue functionality and, when chronically active, can contribute to the aging process (Pawelec et al. 2014). Inflammatory processes are associated with a wide range of human disorders including bowel disease (Hanauer 2006), arthritis (Firestein 2003), and heart failure (Oka et al. 2012). Thus, it is of pivotal importance that we gain a better understanding of the molecular mechanisms through which innate immune responses result in inflammation. The characterization of this new tissue degeneration phenotype in worms, which we propose can serve as a model for inflammation-associated tissue degeneration in higher animals, as well as our identification of a previously unknown response to persistent cytoplasmic pathogenic DNA, establishes C. elegans as an ancestral model for investigating inflammatory responses and for determining the role of protein quality mechanisms in counteracting the detrimental consequences of inflammation. We suggest that further exploration of these processes in this metazoan model will provide new perspectives into the cellular and organismal responses to pathogenic DNA, and persistent challenges to the innate immune system.

Acknowledgments

We are grateful to Jennifer Engelmeyer for outstanding technical support in the laboratory, and to Astrid Schauss and Beatrix Martiny [Cologne Excellence Cluster for Cellular Stress Responses in Aging-Associated Diseases (CECAD) Imaging Facility] for sample preparation for the TEM. Worm strains were provided by the Caenorhabditis Genetics Center (funded by the National Institutes of Health National Center for Research Resources) and the National Bioresource Project. The UPEC strain was provided by Olaf Utermöhlen (Institute for Medical Microbiology, Immunology and Hygiene, Medical Center, University of Cologne). B.S. acknowledges funding from the Deutsche Forschungsgemeinschaft (SCHU 2494/3-1, SCHU 2494/7-1, CECAD, SFB 829, SFB 670, KFO 286, and KFO 329), the European Research Council (ERC) (Starting grant 260383), the Deutsche Krebshilfe (70112899), the The European Cooperation in Science and Technology (COST) action (BM1408), and the Bundesministerium für Forschung und Bildung (Sybacol FKZ0315893). The authors declare no competing interests.

Author contributions: A.B.W. designed and led the project, performed experiments, and, with B.S., wrote the manuscript. J.-E.M. performed experiments and assisted with data analysis. F.H. performed the injection experiments and assisted with data analysis. W.B. performed the TEM and assisted with image analysis. B.S. supervised the project and collaborated on writing the manuscript. All of the authors were involved in critically reviewing and editing the manuscript.

Footnotes

Supplemental material available at Figshare: https://doi.org/10.25386/genetics.8326691.

Communicating editor: M. Schuldiner