Short-Term Starvation of Immune Deficient Drosophila Improves Survival to Gram-Negative Bacterial Infections

Background Primary immunodeficiencies are inborn errors of immunity that lead to life threatening conditions. These predispositions describe human immunity in natura and highlight the important function of components of the Toll-IL-1- receptor-nuclear factor kappa B (TIR-NF-κB) pathway. Since the TIR-NF-κB circuit is a conserved component of the host defence in higher animals, genetically tractable models may contribute ideas for clinical interventions. Methodology/Principal Findings We used immunodeficient fruit flies (Drosophila melanogaster) to address questions pertaining to survival following bacterial infection. We describe here that flies lacking the NF-κB protein Relish, indispensable for countering Gram-negative bacteria, had a greatly improved survival to such infections when subject to dietary short-term starvation (STS) prior to immune challenge. STS induced the release of Nitric Oxide (NO), a potent molecule against pathogens in flies, mice and humans. Administering the NO Synthase-inhibitory arginine analog N-Nitro-L-Arginine-Methyl-Ester (L-NAME) but not its inactive enantiomer D-NAME increased once again sensitivity to infection to levels expected for relish mutants. Surprisingly, NO signalling required the NF-κB protein Dif, usually needed for responses against Gram-positive bacteria. Conclusions/Significance Our results show that NO release through STS may reflect an evolutionary conserved process. Moreover, STS could be explored to address immune phenotypes related to infection and may offer ways to boost natural immunity.


Introduction
Disorders in the TIR-NF-kB pathway are primary immunodeficiencies that have been shown to predispose individuals to pneumococcal and to a lesser extend staphylococcal infections, as well as a selective tendency toward herpes simplex virus encephalitis [1,2,3]. This pathway has been shown to have remarkable similarities with those used in Drosophila immunity and makes the fruit fly a powerful genetically tractable model organism for the study of the first line host defence to infection [reviewed in 4].
In flies, Toll and Imd (for immune deficiency) have been shown to be the major pathways countering infection [reviewed in 5]. Toll signalling culminates in the translocation of the NF-kB homologue Dif to the nucleus following Gram-positive bacterial or fungal challenge [6]. Imd is deployed primarily against Gramnegative bacteria via the NF-kB homologue Relish [7]. This pathway is modulated in larvae by NO [8]. The working hypothesis is that ingestion of bacteria induces NO Synthase (NOS) in the gut. NO released from the gut signals to blood cells, which induce Relish-dependent responses in the fat body (the insect analogue of the liver), the major site of antimicrobial peptide production [8]. According to this model there are two interconnected and sequential phases of the NO-controlled pathway: an NF-kB-independent (blood cells) and an NF-kB-dependent module (fat body). Considering these data, together with the fact that a) NO is a potent antimicrobial agent in a variety of organisms [9], and b) that dietary restriction (DR) activates the endothelial production of NO in mice [10], we investigated whether we could take advantage of the NF-kB-independent phase of NO function to improve the survival of infected Relish-deficient flies if, prior to bacterial challenge, we controlled their access to nutrients.
Our rationale stemmed from the hypothesis that DR-mediated NO production may be an evolutionary conserved process and such a diet regime could be used to boost an immunocompromised immune system (such as one lacking Relish or more generally a TIR-NF-kB component) by elevating NO levels. The intriguing implication if this hypothesis was correct would be that a DR protocol could be used in humans to manipulate natural immunity and boost host defences without the need for a sophisticated clinical setting.
Our results showed that indeed NO release following a protocol of food restriction is an evolutionary conserved process. Shortterm starvation (STS) positively influenced the survival of relish (rel) Newly eclosed flies had either been fed on media supplemented with the NOS inhibitor L-NAME (red line) or its inactive analogue D-NAME (blue line) for 48 hours before the STS regimen mutant flies following Gram-negative bacterial infection. Bacterial load was markedly decreased in rel STS flies compared to their fed counterparts. These phenotypes were reversible when a known NO-inhibitor (L-NAME) was used.
Our results revealed that in wild type flies there was a Relishdependent positive feedback loop that enhanced NO production following infection, through NOS upregulation. In the absence of Relish, STS stimulated the Toll pathway where Dif activated Cytochrome Oxidase C (CCO), which in turn elevated NO levels without NOS upregulation.

STS improves survival of rel-deficient flies to Gramnegative bacteria
In order to determine the effect of food restriction upon NO production in Drosophila, we placed rel or dif flies in nutrient-free agar vials for twenty-four hours. Throughout this period flies had free access to water. We term this short-term starvation (STS) as opposed to dietary restriction (DR) since there was no dilution of the food medium usually applied in DR [reviewed in 11]. Instead, flies were subject to a starvation regime. Furthermore, as it has been estimated that Drosophila only consumes one to two microlitres of food per 24 hours [12], our protocol was a shortterm restriction of nutrients rather than DR feeding. Two groups of rel or dif flies were used. Prior to bacterial challenge, one was fed ad libitum (AL) while the other was subjected to STS for 24 h. They were then infected with Gram-negative (rel) or Gram-positive bacteria (dif). After infection flies were transferred to fresh food and surviving individuals counted daily (see materials and methods). STS rel flies showed a greatly improved pattern of survival in comparison to AL rel flies following infection with Escherichia coli ( Fig 1A) and Erwinia carotovora ( Fig 1B). With both pathogens, 80 to 90% of AL rel flies died within seven days of infection. STS flies however, improved their survival rate considerably with only 40% of STS rel flies dying within the same period. In contrast, STS dif flies infected with the Gram-positive bacterium Enterococcus faecalis showed similarly high levels of susceptibility to infection as their AL siblings and died rapidly after septic injury ( Fig 1C).
In Drosophila, NO is active only against Gram-negative bacteria [8]. Since our results also demonstrated that STS specifically countered Gram-negative bacterial infections we pursued our study with rel STS and AL flies to investigate whether our results might also be NOdependent. When both STS and AL rel flies were transferred to food containing L-NAME, an inhibitor of NO Synthase (NOS), the two groups exhibited comparable patterns of survival that reflected the susceptibility of rel mutant flies to Gram-negative bacteria (Fig 1D,  1E). However, this was not the case when we used the inactive enantiomer D-NAME, indicating specificity in the L-NAME mediated inhibition observed (Fig 1D, 1E). To formally prove that NO release was induced during STS we measured NO levels in STS and AL animals. STS flies had a 20% increase in the levels of NO in comparison to fed animals, regardless of their genotype ( Fig 1F). These results indicated that NO induction following food restriction is an evolutionary conserved process [see also 10].

NO release following STS reduces bacterial load
Production of NO in mice and humans can be directly correlated with the ability of the host to limit microbial proliferation [13] We observed that improvement in disease susceptibility of STS rel flies also correlated with containment of infection. This was determined by assaying the proliferation of bacteria in each fly (CFU/fly) (Fig. 2). For this purpose we used a CFP-E. coli strain or a YFP-Erwinia carotovora strain (see materials and methods). During 96 h of observation the mean bacterial load was constrained at roughly the same level in STS rel animals during both infections (Fig 2A, 2B). In contrast, proliferation was seen in AL rel individuals where the initial mean bacterial load increased 5-fold over the 96 hour period (Fig 2A, 2B). This difference in bacterial proliferation was not observed however, in food containing L-NAME ( Fig 2C). Again, the inactive enantiomer D-NAME had no effect in influencing bacterial load ( Fig 2C). These results demonstrated that NO was at the centre of infection containment.

NO acts both as a signalling molecule and as an antimicrobial agent
Abrogation of inducible NO activity produces dramatic increases in microbial burden perpetuating the idea that NO has direct antimicrobial activity [13]. However, in Drosophila the working model is that NO is functioning as a signalling molecule dependent on the Imd-Relish pathway [8]. Surprisingly, our results suggested that the effects of NO-induction we observed after starvation were independent of Relish ( Fig 1F). Hence there was a case for a direct antimicrobial role. To distinguish between an NF-kB-and a direct NO-mediated effect we used dif-key flies [14], a strain double mutant for dif and kenny (key) the IKKc/ NEMO component homologue of the fruit fly's IkB-Kinase complex, which is essential for Relish activation [15]. STS dif-key flies did not show an improved survival pattern compared to AL dif-key mutants ( Fig 3A). Moreover, no difference was observed between the mean bacterial loads of AL and STS dif-key flies (Fig 3B) However, the STS dif-key flies were still exhibiting a moderate containment of the bacterial load over time compared to L-NAME-treated STS dif-key flies (Fig 3B). This suggested that one component of NO function was independent of NF-kB released upon infection (Fig. 3C), and that this component could be acting directly as an antimicrobial agent. Nevertheless, the differences in survival between STS rel and STS dif-key flies indicated the presence of a major NF-kB-related component in which (in the absence of Relish) Dif itself was mediating NO action.
How then might Dif be mediating NO action? NO signalling induces NF-kB-dependent production of the antimicrobial peptides (AMPs) diptericin and drosomycin [8]. To determine whether Dif was directing AMP expression through NO (thus improving survival of STS rel flies), we measured AMP gene expression using quantitative real-time PCR. As expected the AMP gene diptericin (dipt), a read-out for Imd pathway activation following Gram-negative bacterial challenge [7], was not induced after E. coli infection of AL rel flies (Fig 4A). This was also the case was enforced. In either case flies were returned to the L-or D-NAME vial after STS and infection. Graphs show mean survival (6s.e.) from four independent experiments (E) seven-day survival of STS rel flies infected with Erwinia carotovora. Newly eclosed flies had either been fed on media supplemented with the NOS inhibitor L-NAME (red line) or its inactive enantiomer D-NAME (blue line) for 48 hours before the STS regime was enforced. In either case flies were returned to the L-or D-NAME vial after STS and infection. Graphs show mean survival (6s.e.) of approximately 20 flies from four independent experiments. (F) Quantification of cellular nitric oxide in wild-type (Wn), rel, or dif flies having had free access to nutrients (AL; black bar) or after STS (white bar). In each case mean STS nitric oxide levels are normalised to the level in AL flies (of 1). Graphs show mean relative NO concentration of 15 male flies from four independent experiments (6s.e.). Asterisk indicates significance value of the result as determined by Student's t-Test (*P = ,0.05, **P = ,0.01). doi:10.1371/journal.pone.0004490.g001  for STS rel flies, indicating an absolute dependence of dipt expression on Relish (Fig 4A). Similarly, drosomycin (drs) was not upregulated in rel flies following Gram-negative challenge (Fig 4B). However, drs expression was increased approximately 70-fold following STS (and infection) in wild type flies suggesting that NO can direct Relish-mediated induction of drs (Fig 4B). Together these results indicated that the amelioration of STS rel flies survival was not dependent upon AMPs, since loss of Relish abolished AMP induction.

STS induces a positive feedback loop for NO release
We next sought to determine the aspects of NO expression and/ or signalling that promoted STS rel survival. To this end we monitored transcription of the Nitric Oxide Synthase (NOS) gene. STS induced the upregulation of NOS in both wild type and rel flies (Fig. 5A). In an effort to determine whether STS-mediated upregulation was dependent upon NF-kB, we quantified NOS expression in uninfected and E. coli infected STS flies. While NOS was upregulated 1-to 2-fold following STS (Fig. 5B) (that correlated with an approximately 0.2-fold increase in total cellular NO concentration [ Fig. 1F]), infection of these STS flies with E coli led to a 35-fold increase in NOS expression in wild-type flies that was not observed in rel flies (Fig. 5B). This suggested that while NO synthesis after STS (and subsequently, infection) had an NF-kBindependent component (the fact that we observed increased NO in both rel and dif-key flies showed this), Relish could specifically enhance NO signalling via NOS upregulation. In wild type STS flies this would constitute a positive feedback loop: infection would sequentially induce NO production, Relish activation [8] and finally, NOS up-regulation, which would lead to more NO thus completing the circle. Nevertheless, our results also indicated that STS primed the initiation of such a loop that was evident following infection of STS rel but not of AL rel flies (Fig 5A white bars), with Dif presumably fulfilling the role of Relish.

STS induces mitochondrial biogenesis
We next sought to determine how STS might regulate NO expression through Dif. NO generated after DR increased mitochondrial biogenesis and enhanced respiration and ATP content in various mammalian cells [10]. To investigate whether a similar phenomenon occurred in our system we monitored transcription of different mitochondrial markers (Fig 6A). Transcript levels of the Drosophila homologue of the mitochondrial transcription factor A (Tfam; master regulator of mitochondrial biogenesis) [16] and the putative cytochrome oxidase c (CCO) subunits CG10396 (COX IV), CG10664 (COX IV) and CG17280 (COX6A) were monitored by RT-PCR. We observed upregulation of all these mitochondrial markers when subject to STS (Fig 6A). Moreover, we observed an upregulation of dipt and drs following STS (Fig. 6B) that indicated STS may also activate NF-kB as observed previously [17].
Mitochondria from yeast, rat liver, and plants are capable of NOS-independent NO synthesis [reviewed in 16] when subject to hypoxia and this reaction is catalysed by CCO [18]. Since CCO has been shown to be under Dif and Relish control [ Fig 6A] and NF-kB signalling was activated by STS (as measured by AMP induction shown in Fig. 6B), this mechanism may present a way of augmenting intracellular NO without increasing the transcription levels of NOS, as would be the case in the absence of Relish (Fig 5B). This in turn could lead to a positive feedback loop with NO accumulation directing NF-kB-mediated signalling towards CCO upregulation and AMP production (Fig 6C; see below). This may be occurring in STS rel flies, augmenting their cellular NO content via Dif in the absence of any significant change in NOS expression, hence ameliorating the outlook for their survival when compared to dif-key flies.

Discussion
Involved in many responses to various exogenous and endogenous signals are the evolutionary conserved NF-kB transcription factors. Recent studies have shown that in mice, DR acts as a trigger for NO activation [10]. We show here that this procedure is conserved in Drosophila. Moreover, it can be used to ameliorate the outlook of Relish immunodeficient flies following Gram-negative infection. Surprisingly our data show that in the absence of Relish, which is the main mediator of NO effects in wild type flies [8], the Toll-specific NF-kB regulator Dif, is recruited to direct NO synthesis. To our knowledge, this is the first direct in vivo evidence of a role for Dif in promoting NO function.
Experiments presented here as well as those of a recent study, have shown that STS induces the Toll pathway [17]; [Fig 6B], which in turn can activate CCO [this study, see also 10]. In the absence of Relish, STS may be inducing Dif-directed CCOmediated NO production. This could lead to mitochondrial biogenesis as judged by the up-regulation of several related markers (Fig 6A). Our results suggest that the link between food restriction-mediated production of NO and mitochondrial biogenesis is evolutionary conserved. Finally, in wild type flies infection leads to Relish-mediated enhancement of NOS expression, with the NO produced signalling to the IMD pathway to augment AMP expression, that in turn enhances NOS expression and NO production (Fig 6C).
In mice, intra-hepatic synthesis of NO through the expression of a NOS-2 transgene, leads to protection from liver inflammatory injury [19]. Our results strongly parallel this and indicate that the protective mechanism in flies is NF-kB dependent. However, NO was not strictly a signalling molecule but acted also directly as an antimicrobial agent (Fig. 2). Recent results concerning primary immunodeficiencies in humans indicate overlapping host defences in natura [1]. The overlapping action of NO and AMPs observed in Drosophila underlie the evolutionary implications of such a notion.
Manipulating NO synthesis through diet may have considerable consequences for developing strategies to protect immunodeficient individuals and future studies with a graded nutritional supply [20] should reveal, which parts of the fly diet may be restricted in order to activate NO in rel flies. It is highly significant to this study that nutritional status has also been shown to modify immune outcome in humans by preferentially altering the differentiation of CD4 + naive T helper (Th) 0 [21]. Differentiated Th1 cells primarily stimulate cell-mediated immunity against intracellular pathogens by activating macrophages and CD8 + cytotoxic T lymphocytes that leads to the production of gamma interferon (IFN-c) [reviewed in 22]. Th2 cells, on the other hand, favour the Bcell-dependent humoral immune response against extracellular pathogens and the production of interleukin 4 (IL-4) [23]. When healthy human volunteers were starved overnight, levels of IL-4 increased. Food intake on the other hand resulted in increased levels of IFN-c [21]. The mechanism for the differential modification of the Th1-Th2 balance in response to calorific input remains enigmatic, however, it is tempting to speculate it might also involve NO. In mammals NO negatively influences Th1 cell development by limiting IL-12 production from macrophages [24], which may therefore lead to the preferential differentiation of Th2 cells. The results of our study suggest it may be appropriate to trial an STS-based protocol in immunocompromised humans.
In conclusion, we have presented evidence of a protocol for the improvement of survival as well as the containment of bacterial load following infection with Gram-negative bacteria in immunocompromised Drosophila. This non-invasive procedure relied on the potential of priming natural elements of the immune system with the ability to respond better to infection. This protocol involved short-term starvation (STS) prior to septic injury. STS mediated the release of NO principally through an NF-kB-dependent mechanism although there was a minor component of NO that was independent of NF-kB. More experiments are needed to fully understand this latter mechanism.

Fly stocks
The following stocks were used in this study: Wn, an isogenised wild type strain, which was the strain used as a background to generate the rel 20E mutant [7], rel 20E [7], dif [6] and dif-key [14].

Short term starvation (STS)
Day old flies were transferred to nutrient-free vial (2% agar, 25 mM Methyl-4-hydroxybenzoate, 0.5% propionic acid) and incubated at 25uC for 24 h. Fold change in the expression of genes used as markers for mitochondrial biogenesis following STS compared to AL conditions. Measurements by real-time quantitative PCR were corrected with rp49 expression as in [26]. The genes used were tfam, master regulator of mitochondrial biogenesis; CG13096 and CG10664, cox4; CG17280, cox6A. Note that the induction observed was dependent on NF-kB since it was not observed when both Toll-Dif and the Imd-Relish pathway were deficient (Dif-Key) (B) AMP gene expression is induced upon starvation. Fold-change in Diptericin (black bars) or Drosomycin (white bars) expression in wild type (Wn) or rel flies following STS. Induction of AMPs implies that NF-kB signalling is robustly activated following STS. In all cases in (A) and (B) mean expression levels (6s.e.) from three independent experiments are shown. (C) Putative mode of regulation of NO expression in Drosophila. NO is synthesised via overlapping and interconnected pathways. The main component of this mode is achieved through the direct upregulation of NOS mediated by the NF-kB protein Relish after infection by Gram-negative bacteria. NO can then upregulate the IMD pathway in a positive feedback loop. A second (minor) component of this would be the direct (NF-kB-independent) production of NO as an antimicrobial agent against Gram-negative bacteria. Finally, STS-mediated NO expression is catalysed by two-independent means. Principally through the up-regulation of Relish, leading to NOS upregulation as described above, or secondly, in a NOS-independent fashion. In this case starvation-induced NF-kB (either Relish and/or Dif) upregulation leads to CCO upregulation and NOS-independent NO production. doi:10.1371/journal.pone.0004490.g006