Suppression of the dihydrolipoamide dehydrogenase gene (dld-1) protects against the toxicity of human amyloid beta in C. elegans model of Alzheimer’s disease

Declines in energy metabolism and associated mitochondrial enzymes are linked to the progression of Alzheimer’s disease (AD). Dihydrolipoamide dehydrogenase (dld) and two of its enzyme complexes namely, pyruvate dehydrogenase and α-ketoglutarate dehydrogenase are associated with AD and have a significant role in energy metabolism. Interestingly, dld gene variants are genetically linked to late-onset AD; and reduced activity of DLD-containing enzyme complexes has been observed in AD patients. To understand how energy metabolism influences AD progression, we suppressed the dld-1 gene in C. elegans expressing the human Aβ peptide. dld-1 gene suppression improved many aspects of vitality and function directly affected by Aβ pathology in C. elegans. This includes protection against paralysis, improved fecundity and improved egg hatching rates. Suppression of the dld-1 gene restores normal sensitivity to aldicarb, levamisole and serotonin, and improves chemotaxis. Suppression of dld-1 does not decrease levels of the Aβ peptide, but does reduce the formation of toxic Aβ oligomers. The mitochondrial uncoupler, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) acts synergistically with Aβ to overcome the protective effect of dld-1 gene suppression. Another metabolic toxin, phosphine, acted additively with Aβ. Our work supports the hypothesis that lowering energy metabolism may protect against Aβ pathogenicity, but that this may increase susceptibility to other metabolic disturbances.


Introduction
One of the main pathological hallmarks of AD that underlies the neuronal dysfunction and dementia is extracellular accumulation of amyloid beta (Aβ) plaques resulting from protein misfolding [1]. In addition to the accumulation of Aβ, neuroimaging studies of AD brains found induced aerobic glycolysis in areas severely affected by the disease at the preclinical stage, but reduced glucose metabolism and diminished activities of mitochondrial enzymes at latter stages of the disease [2][3][4][5]. The cause and effect relationships between these observations is unclear as impairment of energy metabolism may induce protein misfolding, leading to formation of Aβ plaques, but the opposite may also be true as production and accumulation of Aβ may also damage energy metabolism [6][7][8][9][10][11][12].
The decrease in energy metabolism with age has been interpreted in two opposing ways; as a main cause of AD, or as a protective response against the symptoms of the disease. The first interpretation is mostly supported by studies conducted at a late stage of the disease on postmortem brains, making it difficult to assign causality [13][14][15][16][17]. In contrast, recent studies on mixed stage AD samples support that the down-regulation of energy metabolism is a protective factor, leading to the hypothesis that a decrease in nutrient and oxygen supply minimizes neural activity, thereby decreasing the repair burden [18,19]. This is supported by results from a transgenic mouse model of AD in which upregulation of aerobic respiration is clearly harmful [20].
The difficulty in understanding the role of metabolic decline on AD relates to the inaccessibility of AD affected brains during progression of the disease. This situation makes it difficult to distinguish cause from consequence and necessitates reliance on AD disease models.
The presence of Aβ in mice and neuronal cell lines causes mitochondrial dysfunction by inducing mitochondrial DNA damage, disrupting mitochondrial redox potential and inducing oxidative stress [21][22][23][24]. Exposure to 2-deoxy-d-glucose (2DOG), which is an analogue of glucose that inhibits both glycolysis and oxidative phosphorylation, causes a decrease in Aβ toxicity [25,26]. Both of these results could be interpreted as lower rates of glucose metabolism having a protective effect. Furthermore, induction of oxidative phosphorylation with dichloroacetic acid (an activator of pyruvate dehydrogenase (PDH)) increases neuronal sensitivity toward Aβ [27]. These findings reinforce the interpretation that there is a positive correlation between the rate of energy metabolism and Aβ toxicity.
Increased risk of late-onset AD is genetically linked to the human dld locus [28]. The DLD enzyme is a subunit of three ketoacid dehydrogenase complexes, each of which contributes to energy metabolism, pyruvate dehydrogenase PDH, α -ketoglutarate dehydrogenase complex (KGDH) and branched chain ketoacid dehydrogenase complex (BCKDH) [29,30]. A decrease in the activity of PDH and KGDH is associated with neurodegeneration as reduced activities of these complexes are observed in post-mortem brain tissues and fibroblasts of patients with either Alzheimer's or Parkinson's disease [31][32][33][34][35][36]. Repressed activity of the KGDH complex, in particular by creating mice that are heterozygous for a knockout mutation of one of its subunits (E2), decreases glucose utilization in the cortex, which mimics the situation in cortical neurodegenerative diseases [37,38]. Targeted disruption of DLD can also reduce the activities of KGDH and PDH in mice [39], which in the case of PDH blocks the connection between glycolysis and the TCA cycle. [40,41]. Thus, a direct link between DLD activity and AD progression is a distinct possibility.
To explore the relationship between metabolism and AD, we suppressed dld-1 in the nematode C. elegans that express human Aβ. C. elegans is well-suited for such studies as it has been used extensively to study the genetics of aging and associated age-related diseases such as AD. Well-developed models exist in which human Aβ is expressed in either muscle or neuronal cells of C. elegans [42,43]. These strains can be used to monitor the influence of metabolic changes on the aggregation of Aβ peptide either directly or through altered behaviour resulting from cellular malfunction. Assays that have been developed include progressive paralysis, a decrease in chemotactic ability and altered response to neurotransmitters. These assays allow convenient visual monitoring of the molecular processes leading up to Aβ aggregation. A decrease in Aβ mediated pathology in response to suppression of dld-1 supports the notion that decreased energy metabolism is neuroprotective.

Culture conditions
Mixed-stage cultures of C. elegans were maintained on nematode growth medium (NGM) seeded with E. coli OP50 at 20 o C, except strains CL4176 and CL2355, which were maintained at 16°C to suppress Aβ expression. Synchronised cultures for bioassays were obtained by standardized protocols described previously [43,[46][47][48]. Wild type, dld-1(wr4) mutant and Aβ transgenic worms CL4176 were all initially cultured at 16 o C for 36 hours after which the temperature was increased to 23 o C for 36 hours except for the paralysis assay for which the temperature was further increased to 25 o C to maximise expression of the Aβ transgene.
Phenotypes of the worms were monitored by visual observation under a microscope and/or quantified using the WormScan procedure [49].

dld-1 gene suppression by RNAi
The E. coli strain SJJ_LLC1.3 (Source Bioscience), which expresses double-stranded RNA of the dld-1 gene was fed to each of the four C. elegans strains to suppress expression of the dld-1 gene [50]. Briefly, the bacteria were cultured in LB medium containing 100 µg/mL ampicillin overnight with shaking at 37 o C. 300 µL of this bacterial culture was transferred to NGM plates containing 100 µg/mL ampicillin and 1 mM IPTG. The plates were incubated at 25 o C overnight to allow the bacteria to grow. Synchronised L1 worms were transferred to the bacterial plates and kept at 16 o C for 36 hours. After a further 36 hours at 25 o C the worms were ready for use in the assays described below. Mock gene suppression controls were treated in exactly the same way except that the bacterial strain (HT115) for the controls contained the plasmid vector without the dld-1 gene fragment.

Paralysis and mortality assays
Paralysis can be defined as time-dependent observable decrease in worm's muscular movement.
At severe stage it may lead to complete halt of movement and worms might be consider as dead.
Paralysis could be slow down or reversible. [51]. Mortality could be defined as acute death after certain period of time due to decline of cellular functions and organelles. Synchronised, L1 stage worms were transferred to NGM plates that had been seeded with the dld-1 RNAi E. coli or an E.
coli strain (HT115) containing an empty vector as a control. After 36 hours at 16°C, worms were upshifted to 25 o C. The worms were then scored for paralysis every second hour after an initial 24-hour period until the last worm became paralysed. For mortality assays, worms were counted as dead or alive after treatment.

Touch response assay
Touch response assays were performed at 20°C on synchronised L4 worms after inducing Aβ expression at 23 o C as described for dld-1 gene suppression. Fifteen (15) animals of each strain were selected arbitrarily and put on freshly made NGM plate. Worms were left on plates for 2 minutes to allow them to equilibrate to the new conditions. Worms were then touched on the head or tail region using a platinum wire to stimulate locomotion and body bends were then counted for 30 seconds.

Aldicarb and levamisole assays
Worms prepared as described for dld-1 gene suppression were incubated in the presence of 1 mM aldicarb, an acetylcholinesterase inhibitor [52]. In parallel with the aldicarb experiment, we also exposed worms to 0.2 mM levamisole, a cholinergic agonist [53]. The number of active worms was counted every hour until all worms became paralyzed.

Phosphine exposure assay
Nematodes were fumigated with phosphine at 500 ppm and 2000 ppm as described previously [54]. Briefly, a synchronised population of 48 hour old (L4) nematodes was washed with M9 buffer and approximately 80-100 nematodes were transferred to each well of 12-well tissue culture plates containing 2.5 mL of NGM agar per well pre-seeded with either empty vector strain HT115 or E. coli strain SJJ_LLC1.3. Nematodes were exposed to phosphine for 24 hours in glass fumigation chambers. The chambers were opened and worms were allowed to recover for 48 hrs in fresh air with lids on plates. The numbers of surviving nematodes were then counted.

5-HT sensitivity assay
To determine the level of Aβ-induced 5-HT hypersensitivity, serotonin (creatinine sulfate salt) was first dissolved in M9 buffer to 1 mM as described previously [43]. Synchronized worms were then washed with M9 buffer and transferred into 200 µl of the 1 mM serotonin solution in 12-well assay plates. The worms were scored as either active or paralysed after 5 minutes.

Chemotaxis assays
Chemotaxis assays was performed as described previously The response to mitochondrial superoxide-mediated oxidative stress was measured using sod-3::GFP in strain CF1553. Synchronized worms were fed with E. coli containing either empty vector or vector that expresses double stranded RNA corresponding to the dld-1 gene for 72 hours at 20 o C. Quantification of sod-3 levels was obtained using Zeiss fluorescence microscope.
To get best signals, non-worm background fluorescence was subtracted from worm fluorescence. For Aβ, the membranes were blocked overnight in 5% skim milk at 4 o C to prevent nonspecific binding of antibodies. The primary antibody staining was done using the Aβ monoclonal antibody 6E10 (Covance) at 1:1000 dilution in TBS (50mM Tris, 150mM NaCl, pH 7.6) containing 1% skim milk for 3-4 hours at room temperature following three washes with TBS-T five minutes each.
Same procedure was repeated for DLD-1 detection except 5%BSA in 1X TBST was used for anti-body detection. Anti-mouse IgG alkaline phosphatase antibody produced in goat (A3562, Sigma), and anti-rabbit IgG alkaline phosphatase antibody produced in goat (A3687, Sigma were used as secondary antibody at 1:10000 dilution in TBS containing 1% skim milk or in 1% BSA in 1X TBST. Secondary antibody staining was done for 1 hour at room temperature. After washing the membrane with TBST, the proteins were detected using BCIP/ NBT substrate system (Sigma) or BCIP/ NBT kit (002209) from Lifetechnologies dissolved in 1M Tris (pH 9.0).

Statistical analysis
Differences due to treatments, strains and RNAi gene suppression were analyzed for statistical significance using GraphPad prism 6.0d. A p value less than 0.05 was considered statistically significant.

Results
The effect of metabolic rate on Alzheimer's disease is an unresolved issue. While a decline in respiration rate is associated with both age of onset and the severity of AD, there are possible alternative explanations. It may either trigger the age-related increase in AD, or it may be a response that protects against the progression of AD. We use gene suppression of dld-1 to directly test the effect of metabolic suppression in several different C. elegans models of Aβ pathology. Specifically, we tested the effect of dld-1 gene suppression on nematodes that express Aβ either constitutively or with temperature induction in muscle cells or constitutively throughout the nervous system. The general experimental paradigm is to expose the nematodes to conditions known to result in Aβ toxicity and to determine whether genetic suppression of dld activity influences that toxicity. In our study dld-1 RNAi effectively suppress the dld-1 mRNA and subsequent protein expression (Supplementary Fig S1).

dld-1 suppression alleviates Aβ pathology in transgenic C. elegans
Transgenic expression and deposition of Aβ in body wall muscle cells of C. elegans causes severe, age-progressive paralysis. (Fig 1A). A temperature shift to 25°C was used to induce high level expression of Aβ. Fewer than 10% of the nematodes of the CL802 control strain that lacks the human Aβ transgene were paralyzed, i.e. unresponsive to prodding by 38 hours. In contrast, 100% of the worms of the CL4176 strain that does express human Aβ were unresponsive ( Fig 1A). Suppression of dld-1 in CL4176 reduced the frequency of paralysis due to Aβ expression to only ~30%, whereas suppression of the dld-1 gene did not alter the robust activity of the control strain, CL802. When we extended the time of the assay (Fig 1B), we found that CL4176 worms in which dld-1 gene expression had been suppressed did not become completely paralyzed until 144±24 hours. We repeated the test on CL2006 worms in which Aβ is expressed constitutively and found that suppression of the dld-1 gene also delayed paralysis in these worms (Supplementary Fig S2). Thus, dld-1 gene suppression prevents, to a large degree, the pathology associated with Aβ that causes paralysis.
Due to critical position in energy production cycle, dld-1 suppression could result in reduction in glycolysis and downstream energy metabolism (TCA cycle and oxidative phosphorylation). We noted that addition of 5mM 2-deoxy-d-glucose (an analogue of glucose that does not enter TCA cycle thus reduce oxidative phosphorylation) to media also reduce dld-1 mRNA and DLD protein expression and protect against Aβ-mediated paralysis in mutated worms ( Supplementary Fig S1 and S3). (B) Extended time-dependent paralysis analysis. As dld-1 RNAi significantly delayed paralysis, we repeated the analysis in a, but monitored the worms every 24 hours until the last worm become paralyzed. Kaplan Meyer survival curves were compared using a Logrank test. (C) CL4176 worms were synchronized and placed on plates either seeded with E. coli strain HT115 containing either empty vector or a dld-1 RNAi construct. After 36-hour incubation at 16 o C, the temperature was increased to 23 o C for 36 hours. Worms were collected, washed and transferred to fresh plates. Worms were touched at the head region with a platinum wire and total number of body bends were counted under the microscope at 20 o C. Results represented data from three independent trials (n=40-60 worms/ trial).

Bars = mean ± SD.
A second movement assay was performed that involved tapping the worms with a platinum wire and counting the number of body bends for 30 seconds. The worms were prepared as for the preceding assay except that the assay was carried out at room temperature (20°C) immediately after temperature induction of Aβ expression at 23°C. We found that as with the immobility assay, expression of human Aβ in the CL4176 strain resulted in a decrease in the rate of movement. Suppression of the dld-1 gene by RNAi significantly improved mobility of CL4176 worms expressing human Aβ, resulting in 6.2±1.3 rather than 2.6±0.7 body bends (p<0.0001) ( Fig 1C). The control CL802 worms that did not contain the Aβ transgene were unaffected by dld-1 suppression (9.1±0.8 rather than 8.9±0.9 body bends) (p=0.529). To check whether dld-1 inhibition restores normal ACh neurotransmission in CL2006 worms that constitutively express Aβ in muscle, we monitored paralysis in response to the cholinergic agonists, aldicarb (a potent acetylcholinesterase inhibitor) and levamisole (a cholinergic receptor agonist).  Fig 3C and 3D, worms were synchronized and placed at NGM plates with or without dld-1 RNAi at 16 o C until L4 stage appeared. Ten L4 stage worms were picked to fresh plates at 23 o C to induce transgene expression. After 24 hours at 23°C, adults were removed. Plates were shifted to 20 o C for remaining assay and eggs and larvae were counted each day for 3 days. Total number of eggs were estimated by adding the total number of un-hatched eggs and larvae present. Three independent trials were run for each experiment. Bars = mean ± SD.
Resistance of the CL2006 strain to ACh agonists is due to production and deposition of both Aβ oligomers and fibrils [60]. Exposure to aldicarb (Fig 2A) results in paralysis within 180 min, which, as expected, occurs more rapidly under dld-1 gene suppression (within 120 min, p=0.0001). Similarly, paralysis in response to levamisole is decreased from 240 min to 150 min ( Fig 2B) when the dld-1 gene is suppressed by RNAi (p=0.0001). Unlike the response in strains in which Aβ is expressed, suppression of dld-1 had no effect on the response to either aldicarb or levamisole in either wild type or dld-1 mutant worms (Supplementary Fig S4). Both nontransgenic strains became paralyzed earlier than transgenic worms that express Aβ (~120 min for both aldicarb and levamisole). Our results indicate that dld-1 suppression restores near normal ACh neurotransmission in Aβ expressing worms via a decrease in Aβ-toxicity.
Serotonin is an important biogenic amine neurotransmitter that mediates locomotion, egg laying and feeding behaviour in C. elegans. Exogenously applied serotonin causes paralysis in worms, which is exacerbated by expression of human Aβ [43]. In our study, 64±4% of worms of the control strain CL2122 were active, but this was reduced to 27±3% in the CL2355 strain that expresses Aβ throughout the nervous system (p=0.002). Suppression of the dld-1 gene did not affect the activity of the control strain CL2122, (57±7%, p=0.3), but could partially alleviate serotonin induced paralysis in CL2355, increasing the percentage of worms that were active to (49±6%, p=0.004) but not to the level of the no-Aβ control strain (p=0.03). (Fig 3B).
Serotonin and ACh neurotransmission control egg laying [62], an activity that is inhibited by neuronal expression of Aβ. Based on our findings above, we reasoned that suppression of the dld-1 gene would reverse the negative effect of Aβ expression on fecundity. Aβ expression significantly reduced egg laying in CL2355 relative to the control strain, CL2122 (157±8 vs 273±23, p=0.001). While there was no significant effect of dld-1 gene suppression by RNAi on the strain that did not express Aβ (273±23 vs 261±19, p=0.5), suppression of the dld-1 gene caused a marked improvement in fecundity in CL2355 (157±8 vs 207±11, p=0.002). The improvement in fecundity did not reach that of the matched control (p=0.01) ( Fig 3C).
Aβ expression also negatively affects egg hatching, with 61.8% of CL2355 eggs remaining un-hatched after 24 hours. In contrast, only 14.2% of CL2122 eggs remained unhatched. Inhibition of dld-1 resulted in a significant decrease in unhatched eggs after 24 hours, 30.5% ( Fig 3D). The same trend persisted over the next two days. There was no effect of dld-1 suppression on egg hatching of the control strain CL2122.

dld-1 suppression reduces Aβ protein oligomerization without affecting Aβ peptide levels
A reduction in Aβ toxicity in our study could result from either a decrease in overall Aβ peptide levels or a decrease in the formation of toxic oligomers [63]. We did not observe any significant change in Aβ mRNA levels after dld-1 suppression, indicating that gene expression was not affected ( Fig 4A). We then assessed whether dld-1 suppression affected either the total amount of Aβ peptide produced or the degree of Aβ oligomerization. We found no change in the overall level of Aβ peptide due to dld-1 suppression (Fig 4B and 4C). However, there was a significant decrease in the proportion of Aβ peptide in the form of ~19kDa oligomers and a corresponding increase in ~4kDa monomers (0.31±0.13 vs 0.71±0.023, p=0.045) when the dld-1 gene was suppressed in strain CL4176 (Fig 4B and 4D). In contrast, there was no significant change in oligomers of 12kDa, 16kDa or 23kDa.

dld-1 indicates treatments in which the dld-1 gene has been suppressed by RNAi. The control using an anti-actin antibody (ab14128) is shown below to indicate the relative amount of protein loaded onto each lane. (C) Densitometry of overall protein bands appeared on western blot of each column to estimate differences in Aβ protein expression after dld-1 suppression. (D) Quantification of the intensity of 4kDa monomer and 19kDa
oligomer bands with and without dld-1 suppression from three independent trials. Quantification was carried out using GelQuantNET software. Bars= mean±SD.

How does dld-1 gene suppression protect against the toxicity of Aβ?
DLD-1 is a core enzyme of oxidative respiration and the dld-1(wr4) mutation is known to inhibit energy metabolism (Zuryn et al., 2008). FCCP is also a disruptor of mitochondrial energy metabolism, but it acts in quite a different manner. Both likely disrupt the generation of ATP, but by opposite mechanisms. DLD-1 disruption slows the flow of metabolites through the TCA cycle and the electron transport chain, whereas FCCP accelerates the same flows, but in a futile manner. Exposure to FCCP decreases Aβ production [64,65], which implies that ATP depletion is more important to the protection than is the mechanism that causes the decrease in ATP. To determine which of these two options, metabolite flux or ATP generation is more likely responsible for the protection against Aβ-induced toxicity, we measured Aβ-mediated toxicity in combination with either dld-1 gene suppression or FCCP exposure or both. In the absence of FCCP, both wild type and dld-1 mutant strains exhibited 100% survival whether or not dld-1 was suppressed. The Aβ transgenic worms exhibited moderate mortality that was significantly alleviated by RNAi suppression of dld-1 (20.5±4.2 vs 8.2±5.3, p=0.004).
Exposure to FCCP had an effect very different to that of dld-1 gene suppression. The dose of FCCP that was used caused negligible mortality on its own, but rather than protect against Aβ-mediated mortality, it produced an apparent, but not quite significant increase in mortality from 20.5%±4.2 to 30.8%±12.3 (p=0.107). This outcome does not allow us to attribute protection against Aβ to a decrease in ATP generation. Mortality due to Aβ expression combined with FCCP increased greatly when dld-1 gene expression was suppressed by RNAi (30.8±12.3 vs 92.9±2.8, p<0.0001). Mortality was not restricted to the transgenic strain, however, but also was observed in wild-type N2 (4.2±3.5 vs 95.7±4.1, p<0.0001) and the dld-1(wr4) mutant (3.7±3.3 vs 83.2±16.8, p<0.0001). The most likely explanation is that the decrease in metabolite flux due to a decrease in DLD-1 containing metabolic complexes, together with futile pumping of protons across the inner mitochondrial membrane caused by FCCP, results in a crisis of energy metabolism that affects all three strains equivalently, and that this is largely independent of whether Aβ peptide is expressed.  Our results indicate that suppression of dld-1 gene expression causes a decrease in ROS, provide another plausible mechanism to explain the protective effect of dld-1 gene suppression against Aβ-toxicity. The dld-1(wr4) mutation that is used in the current study confers resistance against phosphine toxicity [71] a phenotype that can also be achieved by dld-1 gene suppression [45].
Phosphine is a fumigant that induces ROS production and lipid peroxidation but causes decreased respiration rates, as well as a reduction in mitochondrial membrane potential and ATP levels [72]. Thus phosphine, like Aβ, impairs mitochondrial function causing phenotypes that are countered by dld-1 inhibition. Due to these similarities, we investigated interactions between dld-1, phosphine and Aβ.
We found that exposing Aβ expressing transgenic worms to 500 ppm phosphine (the LC 50 of wildtype C. elegans) for 24 hours following by 48 hours of recovery at room temperature increased the mortality of the wildtype N2 strain to the same degree as the Aβ expressing strain CL4176. Thus, mortality of N2 increased from 0% to 48.5±10.7% in response to 500ppm phosphine and mortality of CL4176 increased from 22.9±7.8% to 75.6±9.9%. The resistance phenotype of the dld-1(wr4) mutant was unaffected by RNAi-mediated suppression of the dld-1 gene, indicating that the mutation and gene suppression confer resistance to phosphine by the same mechanism. The phosphine toxicity and Aβ toxicity is simply additive, regardless of whether or not the dld-1 gene is suppressed. This indicates that while both phenotypes are affected by the DLD-1 enzyme, they are mediated independently without any interaction. Similar findings were observed when we treated the worms at a higher phosphine concentration of 2000ppm ( Supplementary Fig S5).

Discussion
Imaging of the AD brain, consistently reveals a decrease in glucose metabolism in areas most strongly affected by disease pathology. While the effect of reduced glucose metabolism on AD progression has not been determined, the deposition of Aβ and reduction in glucose metabolism both begin decades before any clinical manifestation of the disease, indicating a possible causal relationship between the two [73].
To assess the role of energy metabolism in AD-associated Aβ proteotoxicity in worms, we used RNAi to suppress the activity of the dld-1 gene, which is known to suppress aerobic respiration [74,75]. The DLD enzyme is also a subunit of two mitochondrial enzyme complexes, pyruvate dehydrogenase (PDH) and alpha-ketoglutarate dehydrogenase (KGDH). These complexes contribute to the oxidative respiration of glucose and are implicated in AD as well [31,76]. Our results show that suppression of the dld-1 gene significantly alleviates the symptoms associated with Aβ expression in either muscles or neurons of C. elegans.
We find that suppression of the dld-1 gene does not affect either the Aβ transgene mRNA levels or the levels of Aβ peptide. Suppression of dld-1 does, however, significantly inhibit the oligomerization of Aβ. Accumulation of Aβ oligomers is thought to be a major culprit in AD progression [77,78], whereas monomers actually help to maintain glucose homeostasis and are not toxic [79,80]. Our findings suggest that dld-1 suppression reduces Aβ oligomerization thus resulting in reduced paralysis, better movement rates, and improved behavioral phenotypes as observed previously [43,60,81,82].
Recent studies have observed that diet rich in carbohydrates accelerates neurodegeneration by increasing Aβ oligomers [83,84]. Suppression of the dld-1 gene causes a decrease in glucose catabolism, which would appear to have the opposite effect of a carbohydrate-rich diet. In fact, dietary restriction has been shown to delay Aβ-mediated proteotoxicity and extend lifespan in C. elegans by activating the DAF-16/ FOXO, AAk-2/AMPK and SIR-2.1/sirtuin pathways [85][86][87][88][89]. Both mutation and RNAi-mediated suppression of the dld-1 gene result in phosphine resistance and an extended lifespan (Kim and Sun, 2007;Cheng et al., 2003;Schlipalius et al., 2012), as well as inhibition of Aβ oligomerization and protection against Aβ-mediated toxicity as we have shown here.
Based on our understanding of the relationship between the dld-1 gene and phosphine toxicity/resistance, we carried out several additional assays designed to compare the mechanisms of action of dld-1 and Aβ. When we exposed worms to FCCP, we found that a dose that had no effect on either wildtype or dld-1(wr4) mutant worms and only a modest effect on Aβ expressing worms was highly toxic when the dld-1 gene was subjected to RNAi-mediated suppression. The basis of the interaction between FCCP and dld-1 gene suppression is unknown. FCCP does, however, deplete the mitochondrial proton gradient that is utilized for ATP synthesis, whereas the DLD enzyme generates NADH that delivers electrons to the electron transport chain that generates the proton gradient. It may be that simultaneous depletion of the proton gradient by FCCP as well as the source of electrons (NADH) by suppressing the dld-1 gene, results in a cellular energetic catastrophe.
As protonophores, mitochondrial un-couplers also reduce the pH of the mitochondrial matrix. At low pH, the dehydrogenase activity of DLD is inhibited and the reverse activity (diaphorase) is induced [90,91]. This would have the same effect on cellular energy metabolism as described in the previous paragraph and indeed, both mechanisms may contribute to the synergistic increase in mortality that is observed.
Aβ proteotoxicity and oxidative stress are positively correlated [67,92,93] and DLD inhibition is known to reduce ROS generation [68,94,95]. The DLD enzyme and mitochondrial electron transport chain (ETC) are both major sources of ROS generation [69,96], so ROS production could either be direct (less ROS emanating from DLD) or indirect (less NADH feeding electrons to the ETC). We found reduced levels of the mitochondrial superoxide detoxifying enzyme SOD-3 after dld-1 gene suppression, suggesting that suppression of dld-1 does indeed decrease the burden of ROS in C. elegans. We also exposed worms to the fumigant phosphine, a mitochondrial poison that causes oxidative stress and inhibits respiration, likely by targeting the DLD enzyme [45,72]. When combined, Aβ expression and exposure to phosphine cause an additive increase in mortality.
Suppression of the dld-1 gene protects against both Aβ and phosphine individually and provides the same degree of protection against the two stressors in combination. One interpretation of these results is that the two stressors act through the same mechanism(s), with each stressor simply increasing the magnitude of the insult.
The similarities that we observe between the toxicity of Aβ and phosphine are worth noting. Both cause suppression of energy metabolism and yet are protected by suppression of the dld-1 gene, a manipulation that likewise suppresses energy metabolism. The toxicity of both Aβ and phosphine is synergistically exacerbated by co-exposure to the mitochondrial uncoupler, FCCP. However, co-exposure to Aβ and phosphine does not result in an additive rather than synergistically increased toxicity, an outcome that is consistent with Aβ and phosphine acting by the same mechanism.
Our findings establish a neuroprotective effect resulting from suppression of the dld-1 gene that encodes a core enzyme of mitochondrial energy metabolism in C. elegans strains that express human Aβ. Although it is not clear how dld-1 suppression regulated the Aβ oligomerization in our study, one depiction could be the change in calcium influx. There are some observations that induced calcium influx promote Aβ oligomerization and vice versa [91,97]. DLD complex could be involved in regulating calcium homeostasis as induced H 2 O 2 levels were found to increase internal calcium stores [98]. We find no evidence that metabolic suppression through DLD is a risk factor for Aβ proteotoxicity. Our results do not distinguish between two possibilities; that neuroprotection is a direct effect of metabolic suppression or that it is an indirect effect resulting from decreased ROS generation. Regardless of the mechanism, our results are consistent with the hypothesis that a decrease in mitochondrial energy metabolism protects against Aβ pathogenicity, which in humans could delay clinical dementia resulting from AD. Figure S1 Fig S1: dld-1 mutation, or suppression by RNAi causes a decrease in dld-1 transcript and protein levels. Synchronized L1 stage worms of the Aβ expressing strain CL4176 were fed E. coli that expressed dld-1 dsRNA for 36 hours at 16 o C. The temperature was raised to 25 o C for 36 hours to enhance Aβ expression. Temperature was also increased to 25°C in control worms (N2 and dld-1(wr4)) that do not express Aβ peptide. (A) Results of real-time quantitative PCR from three independent trials showing a significant decrease in dld-1 mRNA levels in dld-1 mutated and suppressed worms. (B) Western blot of protein extracted from cell lysate. Anti-lipoamide dehydrogenase antibody (ab133551) was used to detect DLD-1 protein, whereas anti-γ-tubulin antibody (ab50721) was used as reference control. Quantification of the DLD-1 bands from western blots using GelQuantNET software. Graphs and western blots represent the results from three independent experiments. Errors bars = mean±SD.

Supplementary information
Real-time quantitative PCR and western blotting were used to confirm the efficacy of RNAimediated dld-1 gene suppression. We assessed the dld-1 mRNA and protein levels in the Aβ expressing strain CL4176 before and after dld-1 suppression, compared to the wild type strain N2 and the dld-1 mutant dld-1(wr4). When the dld-1 gene was suppressed by RNAi, both transcript and protein decreased to the levels in dld-1(wr4) mutant. Suppression of dld-1 was found to delay the Aβ-associated paralysis in constitutively Aβ expressing worms just as occurred with the temperature inducible worms (Figure 1). Addiction of 5mM 2DOG in media was found to delay the Aβ-associated paralysis in Aβ expressing worms just as occurred with the worms with suppressed dld-1 expression (Figure 1). However, delayed paralysis was more profound in dld-1 suppressed worms than treated with 2DOG.     o  n  n  e  w  a  l  d  ,  B  r  a  i  n  [  U  -1  3  C  ]  g  l  u  c  o  s  e  m  e  t  a  b  o  l  i  s  m  i  n  m  i  c  e  w  i  t  h   d  e  c  r  e  a  s  e  d  a  l  p  h  a  -k  e  t  o  g  l  u  t  a  r  a  t  e  d  e  h  y  d  r  o  g  e  n  a  s  e  c  o  m  p  l  e  x  a  c  t  i  v  i  t  y  ,  J  o  u  r  n  a  l  o  f  n  e  u  r  o  s  c  i  e  n  c  e  r  e  s  e  a  r  c  h  ,  8  9 ( 2 0 1 1 ) 1 9 9 7 -2 0 0 7 .   S  a  a  d  a  ,  I  .  A  p  t  o  w  i  t  z  e  r  ,  G  .  L  i  n  k  ,  O  .  N  .  E  l  p  e  l  e  g  ,  A  T  P  s  y  n  t  h  e  s  i  s  i  n  l  i  p  o  a  m  i  d  e  d  e  h  y  d  r  o  g  e  n  a  s  e  d  e  f  i  c  i  e  n  c  y  ,   B  i  o  c  h  e  m  B  i  o  p  h  y  s  R  e  s  C  o  m  m  u  n  ,  2  6  9  (  2  0  0  0  )  3  8  2  -3  8  6  .   [  7  6  ]  P  .  B  u  b  b  e  r  ,  V  .  H  a  r  o  u  t  u  n  i  a  n  ,  G  .  F  i  s  c  h  ,  J  .  P  .  B  l  a  s  s  ,  G  .  E  .  G  i  b  s  o  n  ,  M  i  t  o  c  h  o  n  d  r  i  a  l  a  b  n  o  r  m  a  l  i  t  i  e  s  i  n   A  l  z  h  e  i  m  e  r  b  r  a  i  n  :  m  e  c  h  a  n  i  s  t  i  c  i  m  p  l  i  c  a  t  i  o  n  s  ,  A  n  n  a  l  s  o  f  n  e  u  r  o  l  o  g  y  ,  5  7  (  2  0  0