C. elegans monitor energy status to trigger innate immune responses via AMPK pathway against bacterial pathogens


 Pathogen recognition and triggering pattern of host innate immune system is critical to understanding pathogen-host interaction. Cellular surveillance systems have been reported as an important strategy for the identification of microbial infection. In the present study, using Bacillus thuringiensis-Caenorhabditis elegans as a model, we found a new approach for surveillance systems to sense the pathogens. We report that Bacillus thuringiensis produced Cry5Ba, a classical PFTs, leading mitochondrial damage and energy imbalance by causing potassium ion leakage, instead of directly targeting mitochondria. Interestingly, C. elegans can monitor intracellular energy status through the mitochondrial surveillance system to triggered innate immune responses against pathogenic attack via AMP-activated protein kinase (AMPK). Obviously, it is common that pathogens produce toxins to cause potassium leakage. Our study indicate that the imbalance of energy status is a common result of pathogen infection.Besides. AMPK-dependent surveillance system can act as a new stratege for host to recognize and defense pathogens.


Introduction
Animals encounter diverse pathogens in natural environment, and they have evolved different defense responses against pathogens for survival. The principal challenge for the host is how to sense the pathogens and trigger defense responses. The innate immune system is a universal and evolutionarily ancient part of such host defense response 1 .
It is generally accepted that hosts could discriminate pathogens from non-pathogenic bacteria in multiple ways. Firstly, the host can recognize microbe-associated molecular patterns (MAMPs) or endogenous danger-associated molecular patterns (DAMPs) by pattern-recognition receptors (PRRs) to induce immune signaling pathways 2 . This is also called pattern-triggered immunity (PTI), which is the most traditional way for hosts to identify pathogens. It is wellknow that MAMPs are highly conserved and common among microbes 3,4 . DAMPs are endogenous molecules secreted by damaged cells 5 , which also implies that they are not the premier immune-activating factor. However, the PRR ligands are not unique to pathogenic bacteria, but also found in the non-pathogenic bacteria 6 , making it hard to simply identify the pathogen in this way. Besides, the hosts could sense the existence or the damage caused by certain Results B. thuringiensis infection leads to cellular energy imbalance in C. elegans To investigate the intracellular physiological changes of C. elegans after pathogenic Bt infection, we conducted the C. elegans transcriptome analysis after infection by the nematicidal Bt strain BMB171/Cry5Ba, an acrystalliferous Bt mutant BMB171 transformed with toxin gene cry5Ba on the shuttle vector pHT304 30 . As a control, we compared the transcriptome to a non-nematicidal Bt strain BMB171/pHT304, BMB171 transformed with the empty vector pHT304. Enrichment pathway analyses highlighted several pathways strongly affected by the infection of nematicidal Bt strain ( Fig. 1A and Table S2). Interestingly, we found the energy metabolic related pathway was most strongly affected. This indicated that the regulators of the energy metabolic pathway may play important roles in host responses against Bt infection. To con rm this, we measured the concentrations of AMP and ATP by LC-MS 31 when wild-type C. elegans N2 fed with BMB171/Cry5Ba, the non-nematicidal control strain BMB171/pHT304 and the standard food strain E. coli OP50. The results showed that the AMP/ATP ratio had no signi cant difference after C. elegans N2 fed with BMB171/pHT304, but signi cantly increased after fed with BMB171/Cry5Ba, comparing with that of fed with E. coli OP50 (Fig. 1B). However, the AMP/ATP ratio showed no signi cant difference when the Cry5Ba-receptor mutant bre-5(ye17) worms were fed with either BMB171/Cry5Ba or BMB171/pHT304 (Fig. 1B). Next, we tested whether other nematicidal Bt can lead cell energy change, such as BMB171/Cry5Ca 30 , BMB171/Cry6Aa 32 , BMB171/Cry21Aa and nonnematicidal BMB171/Cry1Ac 33 . We found that nematicidal Bt strains can cause signi cant energy imbalance of C. elegans comparing with non-nematicidal Bt (Fig. S1). Taking together, we demonstrated that nematicidal Bt infection triggers a cellular energy imbalance of C. elegans, which was mainly attributed to the nematicidal toxins.

B. thuringiensis infection leads to mitochondria damage
Mitochondria produces most of the cell's ATP through oxidative phosphorylation and the tricarboxylic acid cycle, and plays a vital in uence on cell metabolism 34 . Mitochondria are constantly in a state of fusion and division, which are essential for maintaining mitochondrial respiration and homeostasis, and even cell death 35,36 . It has been proved that PFT from pathogens can caused serious mitochondrial disruption in C. elegans 37 . Also, toxins from Pseudomonas pathogens can target mitochondria and lead to mitochondria fragmentation (MF) and it dysfunctional 16,17 . The MF phenomenon can result in bioenergetics defects, which cause cell energy imbalance 38 . To assess how nematicidal Bt causes cellular energy imbalance. We tested several physiological and biochemical aspects of mitochondrial damage including MF, mitochondrial membrane potential (ΔΨm), and mitochondrial DNA (mtDNA) content. We used the transgene worms SJ4143(zcIs17[P ges−1 ::GFP mt ]) as MF reporter which stably expressed GFP in mitochondria matrix of intestinal cells to detect the mitochondrial morphology 39 . When worms fed with nematicidal Bt BMB171/Cry5Ba, 76.67% of worms showed MF ( Fig. 2A and 2B). However, when worms were fed with non-nematicidal Bt strain BMB171/pHT304 or the standard food strain E. coli OP50, the mitochondria morphologies of most worms kept tubular, only less than 15% of worms showed MF ( Fig. 2A and 2B). These results showed that nematicidal Bt infection leads to MF. Next, we tested whether BMB171/Cry5Ba infection can cause changes in mitochondrial membrane potential (ΔΨm) and mitochondrial DNA (mtDNA) content. Using wild type N2 worms, we found BMB171/Cry5Ba infection can lead to a signi cant reduction in mitochondrial membrane potential (ΔΨm) and mtDNA content comparing with non-nematicidal Bt BMB171/pHT304 (Fig. 2C, 2D and S2).
When the Cry5Ba receptor related gene bre-5 was silenced by RNA interference (RNAi) in transgene worms SJ4143(zcIs17 [P ges−1 :: GFPmt]), only 12.22% of worms showed MF phenomenon after fed with BMB171/Cry5Ba ( Fig. 2A and 2B), indicating that the Cry protein toxin Cry5Ba was the key factor that caused the MF phenomenon during Bt infection. However, when Cry5Ba-receptor null mutant bre-5(ye17) worms fed with BMB171/Cry5Ba in the same condition, the ΔΨm and mtDNA content were similar to bre-5(ye17) worms fed with BMB171/pHT304 ( Fig. 2C-D). We conclude that Cry5Ba toxin is the key factor for Bt to manipulate host cell mitochondria and cause serious mitochondrial dysfunction.
Mdivi-1 is an e cient inhibitor to attenuates mitochondrial division by inhibiting the mitochondrial division dynamin, and it also suppresses mitochondrial outer membrane permeabilization 40 . To assess the relationship between cell energy imbalance and mitochondrial dysfunction after Bt infection, we checked whether mitochondrial division inhibitor mdivi-1 can recover mitochondria damage. We observed that mdivi-1 can effectively reduce Cry5B-mediated MF ( Fig. 2A-B). Besides, our results showed that when wild type N2 worms were fed with BMB171/Cry5Ba adding mdivi-1, the AMP/ATP ratio was signi cantly reduced, the ratio of mtDNA/nDNA returned to normal levels ( Fig. 2D-E). However, we found mdivi-1 cannot recover the reduction of mitochondrial membrane potential (Fig. 2C). In general, these results indicate that mitochondrial damage is responsible for the increase in intracellular AMP/ATP.
B. thuringiensis -induced mitochondrial damage is the result of intracellular potassium leakage The Cry5Ba toxin in Bt is a typical pore-forming toxin (PFT) which can form pores in the cell membrane, causing the unbalance in selectively permeable of the plasma membrane 43,44 . Mitochondria face various intracellular stressors, to explain the mechanism of Cry5Ba caused mitochondrial damage, we determine the direct stressors of mitochondria in response to Cry5Ba. We rst check whether Cry5Ba can target mitochondrial directly, our results showed Cry5Ba can be translocated into epithelial cells, but can't colocalize with mitochondria (Fig. 3A). In addition to target mitochondria directly, it had been proved PFTs can lead to severe ion dysregulation 45 . The osmotic imbalance caused by ion dysregulation has been demonstrated important to mitochondria 46 . Next, we assess the cellular Ca 2+ and K + levels after Bt infection. We found that Bt infection caused a signi cant decrease in potassium ion concentration, but not calcium ion. In a K + -free environment, the leakage of potassium is more serious. When the potassium concentration increased, the leakage of potassium would be alleviated. But in bre-5(ye17) worms, potassium content did not change at any condition ( Fig. 3A-3B and S5-S6). So we can conclude that during the Bt infection, Cry5Ba caused intracellular potassium leakage.
To verify our hypothesis that potassium leakage is the direct reason for mitochondrial damage. We detect the MF phenomenon during different potassium concentrations, we found during the Bt infection, the restoration of intracellular potassium can also reduce the MF phenotype, but serious potassium leakage caused a greater proportion of MF phenotypes (Fig. 3C). At the same time, the mitochondrial membrane potential could also return to normal level with potassium addition (Fig. 3D). As we found, the MF phenotype caused by Cry5Ba toxin is responsible to energy imbalance. More importantly, our results showed the restoration of potassium concentration can restore the content of AMP/ATP (Fig. 3E). Therefore, we concluded the leakage of potassium ions caused by Bt infection directly causes mitochondrial stress,which subsequently leads to mitochondrial damage and energy imbalance.
Cell energy imbalance mediated by mitochondria damage activates the AMP-activated protein kinase The AMP-activated protein kinase (AMPK) is a sensor of energy status that maintains energy homeostasis and can be activated by a decrease in energy levels 47 . The synthesis and catabolism of ATP are largely regulated by AMPK 47 . AMPK is activated via phosphorylation of Thr172 on the α catalysis subunit (AAK-2 protein) 47 . It is well known that increasing of AMP/ATP proportion is the classical way to activate the AMPK 47 . We observed that aak-2 was signi cantly up-regulated and the AMP/ATP ratios were signi cantly increased when worms are infected by nematicidal Bt (Fig. 1B and  S7). Therefore, we speculated that nematicidal Bt infection may activate AMPK. To con rm this hypothesis, we performed western blotting to analyze the Thr172 phosphorylation of AAK-2 protein in worms. The results showed that the Thr172 of AAK-2 was phosphorylated when C. elegans fed with BMB171/Cry5Ba but not in the control strain BMB171/pHT304 treatment (Fig. 4A). What's more, the Thr172 of AAK-2 protein was not phosphorylated when Cry5Ba-receptor null mutant bre-5(ye17) worms were fed with BMB171/Cry5Ba under the same conditions (Fig. 4A). Inhibition of mitochondrial fragmentation using mdivi-1 could signi cantly suppress the Thr172 phosphorylation of AAK-2 (Fig. 4A). Besides, restore the leakage of potassium ions caused by Bt infection can also suppress the Thr172 phosphorylation of AAK-2 (Fig. 4B). These results demonstrated that the AMPK of worms is activated by nematicidal Bt infection via the phosphorylation of the core subunit AAK-2.
To assess the relationship between cell energy imbalance and the activating of AMPK after Bt infection, we knock down the aak-2 transcription levels by RNAi in the mitochondria reporter worms SJ4143(zcIs17[P ges−1 :: GFP mt ]). Then, the non-RNAi and aak-2 RNAi worms were fed with BMB171/Cry5Ba or BMB171/pHT304, respectively. The RNAi-aak-2 worms also showed a high level of MF when worms fed with BMB171/Cry5Ba, most mitochondria keep tubular when worms fed with control strain BMB171/pHT304 (Fig. 4C-4D).
We also measured the concentrations of AMP and ATP when wild-type C. elegans N2 and RNAi-aak-2 worms were fed with BMB171/Cry5Ba or BMB171/pHT304. The AMP/ATP ratio was signi cantly increased when RNAi-aak-2 worms were treated with BMB171/Cry5Ba compared to BMB171/pHT304 ( Fig. 4E). However, the increasing of AMP/ATP ratio was no signi cant difference when RNAi-aak-2 and N2 worms were treated with BMB171/Cry5Ba (Fig. 4E). These results demonstrate that knockdown of the aak-2 does not affect the level of MF and energy imbalance during Bt infection, and suggest that AMPK activation is a result rather than a cause of the above MF phenomenon. Thus, we concluded that Btmediated cell energy imbalance activates the AMPK in C. elegans.
We next asked whether the AMPK subunit α2 knockout phenotype is speci c for Bt infection. The sensitivity of the wild type N2 and aak-2(ok524) mutant worms were analyzed to the heavy metal copper sulfate and oxidative stress (hydrogen peroxide). Our results showed that it is no signi cant difference in these treatments between the wild type N2 and the mutant aak-2(ok524) worms (t-test, p > 0.05) (Fig. S10A and S10B).
In addition, we activated AMPK using 5-Aminoimidazole-4-carboxamide 1-ß-D-ribofuranoside (AICAR), a typical activator stimulating AAK-2 kinase activity of AMPK via phosphorylation of Thr172 on α catalysis subunit (AAK-2 protein) 51 . Western blotting results showed that the AMPK is activated by 50 µM of AICAR via AAK-2 Thr172 phosphorylation (Fig. 3A). We then compared the sensitivity of the AICAR-treated with non-treated N2 worms to BMB171/Cry5Ba infection by survival assays and growth assay, respectively. The results showed that the AICAR-activating worms showed more signi cant resistance to BMB171/Cry5Ba infection comparing with the no-activating N2 worms ( Fig. 4D and 4E). However, compared with the wild type N2 worms, the resistance to BMB171/Cry5Ba infection was not signi cantly different for aak-2 mutant either in AICAR-activating or no-activating conditions ( Fig. 4D and 4E). Taking together, our results demonstrated that AMPK plays an important role in C. elegans defense against BMB171/Cry5Ba. AMPK activity in the intestine is required for C. elegans resistance to Bt infection To independently con rm the importance of AAK-2 in defense to BMB171/Cry5Ba infection, we constructed the transgenic strain aak-2(ok524) (P aak−2 ::aak-2), which expresses the aak-2 under the control of its native promoter P aak−2 to rescue the aak-2 function under the background of mutant aak-2(ok524) worms. The growth assay and survival assay results showed that aak-2 expression driven by its own promoter P aak−2 can completely alleviate the hypersensitivity of aak-2(ok524) mutant to BMB171/Cry5Ba infection ( Fig. 6A and 6B), supporting that AAK-2 is independently important for worms to defense BMB171/Cry5Ba infection.
C. elegans apparently lacks professional immune cells, but can rely on epithelial cells for immune defenses 52 . Cry5Ba can attack the intestine of worms and forming pores in the membrane of the intestine cell 53,54 . We therefore hypothesized intestinal-speci c activity of AMPK regulates immune responses to Bt infection immediately. We drove the aak-2 expression under different tissue-speci c promoters, including the intestine-speci c promoter P vha−6 55 , the muscle-speci c promoter P myo−3 56 , and the neuron-speci c promoter P rab−3 57 . We found that only aak-2 expression under the intestine-speci c promoter P vha−6 alleviated the hypersensitivity of aak-2(ok524) mutant to BMB171/Cry5Ba infection ( Fig. 6C and 6D). In contrast, there were no signi cant difference in hypersensitivity to Bt infection among the aak-2(ok524) mutant and the muscle-speci c P myo−3 or neuron-speci c P rab−3 rescued worms ( Fig. 6C and 6D). Also, we knocked down the aak-2 gene by RNAi in the intestine, muscle, and epidermis of worms, respectively. Only the intestine-speci c knock-down of aak-2 made the worms more sensitive to BMB171/Cry5Ba infection (Fig. 6E). In contrast, epidermal-speci c or muscular-speci c aak-2 RNAi worms did not ( Fig. S11A and S11B). These results suggest that the intestine serves as the rst line of AMPK-mediated defense against BMB171/Cry5Ba attack.
AMPK activation triggers DAF-16 dependent innate immune singling pathway during Bt infection AMPK pathway is evolutionarily conserved from C. elegans to mammals and regulates many downstream pathways 58 . Here, AMPK has been shown to play a role in the defense of Bt infection in C. elegans. However, the AMPK downstream genes or pathways involved in the defense against Bt infection in C. elegans were not clear. It was reported that AAK-2 is capable of modulating the phosphorylation of the FOXO family transcription factor DAF-16, activating the DAF-16-dependent transcription when worms suffering from dietary restriction 59 . DAF-16 can regulate many genes involved in metabolism, immune responses against several pathogens, and longevity of C. elegans 60,61 . Moreover, the DAF-16 was also triggered by nematicidal Bt infection 62 , and functioned as an important modulator in defense against nematicidal PFTs in C. elegans 63 . Therefore, we speculated that AMPK may regulate the DAF-16dependent signaling pathway in defense against Bt infection. To con rm this hypothesis, we compared previously identi ed DAF-16 target genes 60 with our RNA-Seq data. We found 60 of the genes upregulated by Bt infection are also the targets of DAF-16 (Fig. 7A, Table S3 and S4) (t-test, p < 0.001). To verify this analysis, we selected 8 typical genes from these genes and determined their transcription by qPCR. The transcription of these 8 genes was signi cantly up-regulated after Bt infection in wild type N2 worms. Moreover, RNAi of daf-16 signi cantly suppressed the up-regulation of these genes induced by Bt infection (Fig. 7B), and RNAi of aak-2 also suppressed most gene upregulation, supported that these DAF-16-dependent genes are also regulated by AAK-2 during Bt infection.
Under standard growth conditions, DAF-16 is distributed predominately throughout the cytoplasm of all tissues. When activated, DAF-16 will be phosphorylated and translocated from cytoplasmic to the nucleus, then binds to the promoter region and activates the expression of target genes 60 . Therefore, we monitored the cellular translocation of DAF-16 using transgenic worms TJ356(Isdaf-16:: gfp) as a reporter, which expresses functional DAF-16:: GFP fusion protein. Our results showed after BMB171/Cry5Ba infection, most of the DAF-16:: GFP was translocated from the cytoplasm to the nucleus in the intestine, especially the front and middle parts of the intestines ( Fig. 7C and 7D). In contrast, the control strain BMB171/pHT304 failed to cause DAF-16 transfer to the nucleus at the same conditions (Fig. 7C, 7D and Fig. S12). These results demonstrated that nematicidal Bt infection can activate the DAF-16 nuclear translocation in wild-type worms.
To test whether the AMPK activity is required for the activation of DAF-16 during Bt infection, we monitored the cellular translocation of DAF-16 when the aak-2 gene was silenced by RNAi in the reporter worm TJ356 (Isdaf-16::gfp). The observations showed that the DAF-16 nuclear translocation induced by Bt is signi cantly diminished by RNAi aak-2 gene ( Fig. 6C and 6D). To test whether DAF-16 activation can trigger the transcription of the downstream immune-related effectors, we tested the expression of sod-3, a well-known direct target of DAF-16 during Bt infection 64 ; sod-3 gene is also one of the most signi cant up-regulated genes in Bt-infected worms (Table S3 and S4). We observed the expression of sod-3 using transgenic worm CF1553(muIs84(sod-3::GFP)) 64 . When the SOD-3::GFP reporter worms were exposed to BMB171/Cry5Ba, the expression of sod-3 was signi cantly up-regulated compared to BMB171/pHT-304. Meanwhile, the induction of sod-3 by BMB171/Cry5Ba infection was signi cantly inhibited when either daf-16 or aak-2 gene was silenced in the strain CF1553(muIs84(sod-3::GFP) ( Fig. 7E and 7F). Additionally, the aak-2 deletion also signi cantly suppressed the up-regulation of the above selected 8 immune-related genes induced by Bt infection (Fig. 7B). Taking together, we concluded that the AMPK triggers the DAF-16dependent innate immune pathway during Bt infection.

Discussion
Pathogen recognition and triggering the host innate immune system is critical to understanding pathogen-host interaction 65 . It is generally accepted that the microbial infection can be recognized by host PTI and ETI patterns which are related to PRR. However, due to the limitations of PRR ligands, these conserve innate immune pathways are not enough to distinguish pathogens from some probiotic or symbiotic bacteria. But the ETI hypothesis is an important supplement to PTI and proposes that the host responds to pathogens by monitoring bacterial virulence factors 9,66 . Moreover, cellular surveillance systems, including the ribosome, proteasome, and mitochondria, monitor core cellular physiology activities were considered as a novel pattern for hosts to distinguish pathogens from other microorganisms 18 . However, how host cells using these systems to sense pathogens are unclear. There are several ways for the host to recognize the pathogens by monitoring core cell processes, including the DNA damage 14 , the inhibition of translation 15 , and the UPR mt during the mitochondrial damage state 67 , etc. Here, we revealed that the leakage of potassium caused by pathogens leads to mitochondrial stress, and then host cells could sense pathogens via mitochondria mediated intracellular energy imbalance. Therefore, our work provides a novel insight for the host cell to detect pathogens through cellular surveillance systems.
Previous works have shown several pathogens could attack and disrupt host mitochondria 16,68,69 . In our study, damage to the mitochondria was also detected when worms were infected by nematicidal Bt, including MF, a decrease in mitochondrial membrane potential, and the changes in mtDNA content (Fig. 2). The damages to the mitochondria during Bt infection were mainly caused by its key virulent factor, Cry5Ba toxin ( Fig. 2 and S2). Although Cry5Ba could be translocated into epithelial cells, we did not nd evidence of colocalization with mitochondria (Fig. 3A), indicating Cry5Ba may lead to mitochondrial dysfunction indirectly. Next, we con rmed that the leakage of intracellular potassium caused by Cry5Ba led to the mitochondrial damage directly (Fig. 3B). This model is different from extensively studied toxins produced by P. aeruginosa 16 or L. monocytogenes 68 .
Cellular energy imbalance is an important sign of mitochondrial disorder. In fact, several PFTs such as streptolysin O, Vibrio cholera cytolysin, Staphylococcus aureus α-toxin, and Escherichia coli hemolysin A, could also cause a decrease in intracellular ATP levels in a non-virally transformed human keratinocyte cell line 70 . Our results showed that Bt infection could cause a Cry-dependent cellular energy imbalance, which was widespread among nematicidal Bt infection (Fig. 1B, Fig. S3 and Fig. S4). Both mitochondrial dysfunction and cytosolic energy imbalances were restored in a Cry5Ba receptor mutant (bre-5), indicating toxin from nematicidal Bt is critical for these phenomena. Conversely, the mitochondrial division inhibitor mdivi-1, which can inhibit mitochondrial fragmentation but not the function of PFTs, can protect cells from AMP/ATP ratio imbalances. Besides, the recovery of intracellular potassium concentration caused by Cry5Ba can also alleviate energy imbalances (Fig. 3C-3D). Thus, the pores caused by the toxin are the primal reason for the changes in intracellular energy, and potassium leakage is the direct reason.
Previous research indicated that cytosolic energy imbalance and calcium content are the important AMPK triggering patterns 71 . We found Bt infection did not induce the increase of cytoplasm calcium level of worms (Fig. S6). Rather, nematicidal Bt infection led to mitochondria damage, which then gave rise to the dramatic increase of AMP/ATP ratio and AMPK activation via phosphorylation of α-catalysis subunit protein AAK-2 ( Fig. 2A). Next, we found that the inhibition of mitochondrial fragmentation recovered the mtDNA content, restored abnormal AMP/ATP content, and inhibited the activation of AMPK (Fig. 2). These data suggest that MF was the reason why the AMP/ATP content changed when the mitochondria had been damaged. We also found that Bt infection led to a decrease in mitochondrial membrane potential, which is well known to be a consequence of mitochondrial ATP 72 . Besides, our work revealed that restoring potassium can not only restore mitochondrial damage, but also restore energy imbalance and AMPK activation. Together, our work revealed Bt infection increased the rate of AMP/ATP and cytosolic energy imbalance, which ultimately triggers AMPK via AAK-2 phosphorylation.
Accumulating evidence suggests that AMPK is not only a crucial evolutionarily conserved cellular energy sensor but also plays an important role in host-pathogen interactions 49 . Our work proved the activation of AMPK is helpful for C. elegans to defense nematicidal Bt (Fig. 5). What's more, the AMPK activity in intestinal cells is particularly important for C. elegans against nematicidal Bt (Fig. 6). As we know, due to the lack of professional immune cells, the C. elegans intestine is considered to be the rst line to identify and defense against pathogens 73 . Therefore, we can infer that the host cell monitors the energy imbalance caused by pathogens through mitochondria, and then activate a series of responses against pathogens. Our work highlighted the importance of the connection between the cellular mitochondria surveillance systems and AMPK through energy changes during pathogen infection.
Additionally, how mitochondria disorder triggering downstream innate immunity signaling is largely unknown yet. Accumulating evidences showed that AMPK is also an important trigger of FOXO family transcription factors. DAF-16 regulated by the AMPK pathway is related to defense response against nutrition deprivation and functions to extend lifespan 48,59 . Here, we found that the AMPK activation mediated by mitochondrial damage regulates the DAF-16 to increase the transcription of certain DAF-16dependent innate immune effectors, which helps C. elegans defend against pathogen infection (Fig. 7). The different sensitivity in several worms against Bt infection indicating that the AMPK is not the only way to activation of DAF-16 (Fig. S13A). At the same time, AMPK may regulate other DAF-16-independent innate immune pathways during Bt infection. MAPK pathway is also reported to be a downstream component of the AMPK signaling pathway in mammal cells 74 . The xenophagy for defense against pathogens infection was observably activated by the AMPK-dependent p38 MAPK pathway 75,76 . Under the infection of nematicidal Bt, the up-regulated of 3 typical genes of p38 MAPK pathway (pmk-1, tir-1 and sek-1) together with the activation of PMK-1 demonstrated that AMPK also triggered the p38-MAPK innate immune pathways apart from DAF-16 during Bt infection ( Fig. S13B and S13C).
In conclusion, our works demonstrated that host cells can directly sense mitochondrial-mediated intracellular energy imbalance to monitor pathogens, and activate downstream innate immune responses to defense pathogens through activation of the AMPK pathway. As modeled in Fig. 8, nematicidal Bt infection in C. elegans caused potassium leakage, leading to the mitochondria damage and the energy imbalance of intestinal epithelial cells, the latter of which subsequently triggered AMPK via phosphorylation of AAK-2. Then, the AMPK modulates DAF-16-dependent and p38-MAPK-dependent innate immune pathway to defend against nematicidal Bt infection. These ndings revealed that AMPK senses the changing of cellular energy imbalance and triggers the innate immune responses during pathogen infection. Consider that the mitochondria surveillance systems and AMPK are conserved components from worms to mammals, our study suggests that disrupting mitochondrial homeostasis to activate the immune system through AMPK-dependent pathways may widely existing in animals, which may provide new strategies for immunotherapy of multiple diseases.

Materials And Method
Caenorhabditis elegans and bacterial strains. Caenorhabditis elegans strains used in this work were kindly provided by the Caenorhabditis Genetics Center (CGC) or the National Bioresource Project (NBRP) and listed in Table S1. All C. elegans strains were maintained on nematode growth media (NGM, 0.3% NaCl, 0.25% tryptone, and 1.5% agar) using E. coli OP50 as food under standard conditions 77 . The bacterial strains and plasmids used in this study are also listed in Construction of gene aak-2 rescued worms. A 2.1 kb fragment of the intestine-speci c vha-6 promoter (P vha−6 ), a 2.5 kb fragment of the muscle-speci c myo-3 promoter (P myo−3 ), a 1.3 Kb fragment of the neuron-speci c rab-3 promoter fragment (P rab−3 ) and a 3.0 Kb fragment of aak-2 promoter fragment (P aak−2 ) were generated by PCR from total DNA of C. elegans, then inserted into pPD49.26 to gain 4 recombinant plasmids of pPD49.26-P. A full-length a 1.8 Kb of the aak-2 cDNA and a 1.6Kb of the 3-UTR of aak-2 was generated by overlap PCR and inserted into pPD49.26-P at the downstream of the promoters respectively. The recombinant plasmids which contained gene aak-2 with the tissue-speci c promoters, P aak−2 ::aak-2, P vha−6 ::aak-2, P myo−3 ::aak-2 and P rab−3 ::aak-2, were injected into the gonads of aak-2(ok524) worms to generate independent transgenic lines by standard germline transformation techniques 80 . All of the 4 fusion genes were injected at concentrations ranging 100 ng/µl with P lin−44 ::GFP at 100 ng/µl as a co-injection marker. The gene tba-1 was used as an internal reference 82 . The primers used for PCR are listed in Table S1. The experiments were conducted in triplicate, and the results were expressed as 2 −△△Ct .
Nucleotide measurements. The AMP, ADP and ATP were extracted by the adapting perchloric acid method 83 , 300-400 worms were washed with ice-cold M9 buffer, and resuspended in 20 µl of M9 buffer, then added 40µL of ice-cold 8% (v/v) HClO 4 on ice. The worms were then homogenized in liquid nitrogen, and the homogenates were disrupted using ultrasonic vibrations and neutralized with 1 N KOH. Next, the suspension was centrifuged (10,000g for 3min at 4°C), and the supernatant was passed through a 0. were normalized to genomic DNA using primer pairs speci c for ama-1(Forward: TGGAACTCTGGAGTCACACC, Reverse: CATCCTCCTTCATTGAACGG). Synchronized worms were treated and collected as described above. Collecting worms were homogenized in liquid nitrogen and lysed in a standard buffer containing proteinase K for 1 hour at 65℃. qPCR procedure was implemented as described above.
Calcium and Potassium imaging. Fluo-4 AM (Molecular Probes) and ION Potassium Green-2 AM (APG-2 AM, Abcam) are used to detect the concentration of Ca 2+ and K + respectively, according to the published methods 89, 90 . The treated C. elegans were loaded with 100 µM Fluo-4 AM for 1 h in M9 buffer. C. elegans were washed more than 3 times and then photographed by Olympus BX63 epi uorescence microscope.
For K + , treated C. elegans were labeled in 5µM APG-2 AM in each buffer of different potassium concentrations for 1 hour. The labeled C. elegans need to be cultured in the corresponding buffer added OP50 for 1 hour to remove the free dye in the intestine. The washed C. elegans were photographed under the same conditions. The excitation and emission wavelengths of Fluo-4 AM were 489 nm and 508 nm.
For APG-2 AM, they are 526nm and 550nm, respectively. At least three repeats were performed for each condition and at least 20 animals were measured per treatment. The average optical density of each worm was measured using Image Pro Plus v6.0.
Bt infection assay. The Bt infection assays were based on the published protocol 91 . which contains two different procedures.The plate assays: Bt strains were grown in LB medium containing 50µg/ml spectinomycin at 28°C overnight and seeded 400µl to fresh NGM plate. Then these plates were placed at 28°C for 3 days. Synchronized nematodes were grown to L4 stage in NGM plate containing OP50.
Nematodes were then washed off and cleaned by M9 buffer. 2000-3000 worms were added to the Bt plate and treated at 25°C for 4-6h. The liquid assay: Synchronized L1 worms were grown to L4 stage on OP50 plate at 20°C. Then 300-500 worms were added on a 24-well plate (total volume 50 µl) which containing 400µl M9 buffer. Then add different mixture of crystal toxins and spores to each well (total volume 50ul). The plates were placed at 25°C for 4-6h.
The growth assays. The L1 growth assays were carried out in different doses of the mixture of crystal Cry5Ba and spores as described before 91 , E. coli OP50 was added at an optical density of 0.2-0.25 OD 600 and 20-30 synchronized L1 larvae were used per well. After 3 days at 20°C, photographed at least 60 worms in different doses on a microscope, and calculated the average area for each condition using the software NIH Image J 1.33, normalized the average area at each toxin concentration to the average area of the no toxin control. The size of the worms in the absence of toxin was set at 100%. Each experiment was independently replicated at least three times.
The survival assays. The survival assays were conducted in different doses of the mixture of crystal Cry5Ba and spores as described before 91 . C. elegans N2 and these mutant worms were exposed to the mixture of crystal Cry5Ba and spores in S media in 96-well plates to quantitatively score the survival.
Concentrations of each toxin were set-up in triplicate for each assay with approximately 20-30 synchronized L4 worms per well. The determination of whether worms were alive is according to the worms' movement. The crawling worms were marked as alive. Non-crawling worms should be gently touched with a platinum pick to observe their movement. The survival rate of each well was scored after incubating at 20°C for 6 days. Each experiment was independently replicated at least three times.
Lifespan assays. The life span assay was performed at 20°C on NGM plates. Each bacterium must be fully spread on the NGM plate to prevent worms from avoiding or escaping the bacteria lawn.
Approximately 60 L4-stage worms were incubated on NGM plate seeded with OP50 or other pathogens. Every plate contained 0.05 mg/ml of 5-uorodeoxyuridine (FUDR) to prevent eggs from hatching. Five plates were tested for each strain in each experiment. Each experiment was independently replicated at least three times. The determination of whether worms were alive was as described for the survival assay of C. elegans. The surviving worms on each plate were counted at 20 °C every 12 h. Statistical analyses were assessed by Kaplan-Meier survival analysis followed by a log rank test. Statistical signi cance was assessed by Kaplan-Meier survival analysis followed by a log rank test.  39 , were fed with rhodamine-labeled crystal protein Cry5Ba (a pore-forming toxin produced by Bt) 4 hours, then these worms were placed on 2% agarose pads, the signals of rhodamine-labeled and GFP-labelled were observed using the using confocal microscope at 100×magni cation. At least 2-3 independent biological repeats were carried out for each experiment.
DAF-16 nuclear localization assay. After 2 h of treatment with Bt, the synchronized L4 TJ356 worms (transgenic animals expressing DAF-16::GFP) were immediately placed in M9 buffer and onto microscope slides. GFP localization was observed using a uorescent microscope (Olympus BX31, Japan) at 40×magni cation. DAF-16 localization was categorized as cytosolic localization, intermediate localization and nuclear localization. The number of worms with each level of nuclear translocation was counted. The worms were exposed to non-nematicidal Bt BMB171/pH304 were used as negative controls; while the worms were exposed to heat shock for the same periods at 30°C were used as a positive control. P values were calculated using SPSS ver13.0 (SPSS, Chicago, IL). Western Blotting Analysis. After feeding Nematicidal Bt or Non-nematicidal Bt, C. elegans N2 and different mutants were washed three times with ice-cold M9 buffer and were homogenized in liquid nitrogen. Then homogenates were harvested in lysis buffer (50 mM HEPES, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl2, and 1 mM EGTA) with protein inhibitors (0.2mM Na 3 VO 4 , 1mM NaF).
The lysate samples were subjected to SDS-PAGE using 10% (wt/vol) polyacrylamide gels and transferred to PVDF membranes (Life technologies). The trans blotted membrane was washed three times with PBS containing 0.05% Tween 20 (PBST). After blocking with PBST containing 5% BSA, the membrane was probed with the primary antibody (Phospho-AMPKα(Thr172) Rabbit mAb, Cell Signaling, # 2535, or β-actin Antibody, Proteintech, # 66009) and washed three times. The membrane was then probed with HRPcoupled secondary antibody and washed. Finally, the membranes were exposed using a chemiluminescent substrate (SuperSignal West Pico, Thermo Scienti c).

Con ict of interest Con ict of interest
The authors declare that they have no con ict of interest.  (E) The relative AMP/ATP ratio was measured in each condition. In each assay, the mean and SD of three independent replicates are shown. Values differences were calculated by one-way ANOVA. Triple asterisks ( * * * ) are set at p< 0.001, two asterisks ( * * ) are set at p< 0.01. a single asterisk ( * ) is set at p<0.05.     The action model of mitochondria surveillance systems recognition the pathogens and activation innate immune responses by AMPK in C. elegans. After nematicidal Bt spores and crystals mixture was fed to the worms, the Cry5Ba toxin assembled to trigger pore formation and transport into the intestinal epithelial cells of C. elegans, and disrupts mitochondria indirectly that lead to MF, a decrease in mitochondrial membrane potential, and the changes in mtDNA content, resulting in intercellular energy imbalance. The AMPK senses Bt-caused intercellular ATP changing and activated the AMPK by phosphorylating AAK-2. The activated AAK-2 modulates conversationally DAF-16-dependent and p38-MAPK dependent innate immune pathways to defense pathogens. Solid arrows represent relationship has been con rmed in this study. The dotted arrow represents the relationship is still unclear.
Supplementary Files