The NADPH Metabolic Network Regulates Human αB-crystallin Cardiomyopathy and Reductive Stress in Drosophila melanogaster

Dominant mutations in the alpha-B crystallin (CryAB) gene are responsible for a number of inherited human disorders, including cardiomyopathy, skeletal muscle myopathy, and cataracts. The cellular mechanisms of disease pathology for these disorders are not well understood. Among recent advances is that the disease state can be linked to a disturbance in the oxidation/reduction environment of the cell. In a mouse model, cardiomyopathy caused by the dominant CryABR120G missense mutation was suppressed by mutation of the gene that encodes glucose 6-phosphate dehydrogenase (G6PD), one of the cell's primary sources of reducing equivalents in the form of NADPH. Here, we report the development of a Drosophila model for cellular dysfunction caused by this CryAB mutation. With this model, we confirmed the link between G6PD and mutant CryAB pathology by finding that reduction of G6PD expression suppressed the phenotype while overexpression enhanced it. Moreover, we find that expression of mutant CryAB in the Drosophila heart impaired cardiac function and increased heart tube dimensions, similar to the effects produced in mice and humans, and that reduction of G6PD ameliorated these effects. Finally, to determine whether CryAB pathology responds generally to NADPH levels we tested mutants or RNAi-mediated knockdowns of phosphogluconate dehydrogenase (PGD), isocitrate dehydrogenase (IDH), and malic enzyme (MEN), the other major enzymatic sources of NADPH, and we found that all are capable of suppressing CryABR120G pathology, confirming the link between NADP/H metabolism and CryAB.


Introduction
The maintenance and integrity of specialized functional structures such as sarcomeres, the basic unit of contractile force in striated muscles, are inextricably linked to the cellular machinery of molecular chaperones and protein quality control pathways. Evidence for this notion is provided by the identification of myopathic mutations in genes that encode proteins with chaperone function, such as CryAB and Bag3, and whose products have been localized to Z-discs. Moreover, an increasing number of genes encoding Z-disc associated proteins, such as desmin, ZASP, myotilin and filamin C are linked to myofibrillar diseases [1]. The Z-disc, which is formed by a complex network of diverse proteins, defines the structural boundaries of sarcomeres and integrates the actin filaments of neighboring contractile units. Major morphological and cellular hallmarks that define myofibrillar disorders include disintegration of the Z-disc lattice network, mitochondrial disruption, and ectopic protein aggregates.
The autosomal dominant R120G mutation in the aB-crystallin gene (CryAB R120G ) manifests adult-onset cataracts, skeletal muscle weakness and heart failure [2]. CryAB, a small molecular weight heat shock protein, is expressed constitutively in the lens and in non-lenticular tissues associated with high rates of oxidative metabolism, such as heart and type I and type II skeletal muscle fibers. A primary function of CryAB in these tissues is to prevent aggregation of intermediate filament proteins such as desmin, a characteristic subcellular phenotype of desmin-related myopathies [3]. Earlier studies by several laboratories supported a loss-offunction mechanism for the CryAB R120G mutation, based on alterations of its secondary and quaternary structures, decreased interactions for client substrates with intermediate filaments and reduced stability on heat denaturation in vitro [4,5]. Subsequently, expression of either the mouse or human CryAB R120G allele in the mouse heart produced cardiomyopathy, heart failure and shortened lifespan, all of which phenocopy the disease condition in humans [6,7]. Additional missense, truncation and autosomal recessive mutations have underscored the dominant inheritance patterns of disease-causing CryAB expression but their underlying molecular mechanism(s) have remained elusive.
An important concept in the pathogenesis of neurodegenerative and, perhaps, myofibrillar disease is that misfolding and aggregation of destabilized mutant proteins trigger 'toxic' gain of function etiologies [8]. It has been hypothesized that the accumulation of amyloidogenic aggregates is preceded by the appearance of soluble, oligomeric species whose toxicities arise from multiple, non-exclusive mechanism(s) mediated by (1) flexible hydrophobic surfaces that promote aberrant interactions and sequestration [9], (2) poor clearance mechanisms that disrupt central protein quality control and propagate folding defects [10,11], and (3) perhaps by compromised lipid integrity as suggested by in vitro model membrane systems [12]. Although CryAB R120G does not appear to result from a ''classical'' amyloid, recent studies have shown that aggregates of mutant CryAB are recognized by the anti-oligomer A11 [13,14] providing a link with well-known amyloid protein aggregation diseases such as Alzheimer's and Huntington's diseases.
Our recent work on a mouse model of the inherited human CryAB R120G cardiomyopathy has provided the first persuasive case for pathology resulting from ''reductive'', as opposed to oxidative, stress in disease pathogenesis. The molecular events and damaging effects of reactive oxygen species (ROS) on biological molecules and systems are well known. When ROS are generated in excess of a cell's capacity to neutralize them, by enzymatic and nonenzymatic antioxidant pathways, the cell experiences oxidative stress. A large number of disease states have been attributed, in whole or in part, to damage from ROS or oxidative stress. However, there has been little appreciation for the possibility that an excess of reducing equivalents might also cause problems for the cell. Reductive stress can be induced by supplying strong reducing compounds, such as dithiothreitol, to cells in culture, but only recently has it been shown by Rajasekaran and coworkers that reductive stress may occur as a pathological state consequent to expression of a mutant protein. CryAB R120G -induced cardiomyopathy was accompanied by a significant shift towards a more reduced intracellular environment, as measured by the glutathione redox couple [7]. The dominance of CryAB R120G , and its unexpected link to an excessively reduced environment, suggested a 'toxic' gain of function mechanism.
To investigate the reductive stress hypothesis of CryAB R120G pathology, we have developed the first Drosophila melanogaster model of human CryAB R120G toxicity in multiple organs and tissues. Because there is a strong evolutionary conservation of key developmental and metabolic pathways between humans and Drosophila [15][16][17][18], studies of human conditions in Drosophila have been very productive. This has been especially true in the fields of heritable developmental defects, including congenital heart diseases, aging-related conditions and neurodegenerative diseases [19][20][21][22]. Recently, with advanced microscopic technology [23], Drosophila has been used to model human cardiac physiology and aging [24][25][26][27]. Interestingly, flies with a dilated heart have been reported in certain genetic backgrounds [28]. Strikingly, cardiac dilation is observed in patients with protein aggregation cardiomyopathy and in mice that over-express CryAB R120G [7].
In this study, we demonstrate that CryAB R120G pathologies in the fly heart and eye are regulated by key enzymes that reduce nicotinamide adenine dinucleotide phosphate (NADP) to NADPH. As was the case in the mouse heart, the deleterious effects of expressing CryAB R120G were strongly ameliorated by knockdown or mutation of the gene encoding G6PD. We further exploited the fly model and report that reduced function of other major generators of NADPH also strongly suppressed the CryAB R120G phenotype, implicating the entire cellular NADP/ NADPH network in CryAB R120G pathology.

Drosophila heart dysfunction caused by human CryAB R120G
To extend our tests of the reductive stress hypothesis of CyrAB R120G pathology, we used the Gal4-UAS modular expression system [29,30] to permit expression of the human CryAB R120G allele in various cell types. We first generated transformants carrying either a wild-type UAS-CryAB + or UAS-CryAB R120G construct. Next, to determine whether CryAB R120G expression would affect cardiac function in Drosophila, we drove its expression in the fly heart with a Hand-Gal4 driver. We confirmed cardiacspecific expression from Hand-Gal4 by examining GFP fluorescence using a UAS-CryAB R120G -GFP fusion construct. Visual inspection of fluorescent micrographs confirmed cardiomyocyte restricted expression of the construct. Remarkably, the marked CryAB R120G appeared targeted to a repetitive myofibrillar component of the cardiac fibers of flies, as found in higher organisms [31], likely the Z-discs ( Figure S1).
To investigate the effects of human wild type or mutant CryAB expression on the simple, linear Drosophila cardiac tube ( Figure 1A) we imaged surgically exposed beating hearts and tracked wall movements of semi-intact flies using direct immersion DIC optics in conjunction with a high speed digital video camera [32]. We characterized the effects of CryAB + and CryAB R120G on the contractile performance and general morphology of Drosophila hearts.
Heart period, which is defined as the length of time between the ends of two consecutive diastolic intervals, and arrhythmia indices, a quantitative measure that reflects cardiac rhythmicity and permits exploration of heart rhythm irregularities, were calculated for ,45 three week old semi-intact Drosophila from each line. Cardiac diameters were measured directly from individual video frames at peak diastolic and systolic time points at multiple locations along the linear portion of abdominal segment three of each heart tube. These measurements revealed that expression

Author Summary
Cardiomyopathy is a specific form of heart disease that involves progressive restructuring of the heart muscle, resulting in reduced function and increased chance of sudden heart failure. Several mutations have been identified that cause inherited cardiomyopathy, including mutations in a gene called alpha B-crystallin. The work we report here puts this mutant gene into the fruit fly Drosophila melanogaster, where it causes heart defects that are similar to the human condition. Like cataracts reported in humans, this mutant gene also causes defective development of the fly eye. By examining defective fly hearts and eyes, and asking how genes involved in oxidation/reduction (redox) reactions influence these defects, we find that genes which control the redox environment of the cell also control the heart and eye defects. Oxidative stress, defined as an excess of harmful oxygen radicals, is a common concept in disease. Our work shows that in some circumstances, the genes that generate a more reduced cellular environment may also contribute to disease.
human CryAB + did not significantly perturb any analyzed index of cardiac function relative to control hearts ( Figure 1, Figure S2). In contrast, expression of CryAB R120G significantly affected several parameters of heart function. Arrhythmic beating patterns appeared to increase ( Figure 1C), although these trends were not statistically significant. Diastolic and systolic diameters were significantly increased in response to CryAB R120G expression, and fractional shortening of the fly heart was significantly reduced thus notably impairing Drosophila cardiac function ( Figure 1D-F). In accord with the mouse observations [7], simultaneous RNAimediated knockdown of G6PD significantly improved several indices of cardiac function and overall heart performance. Hearts expressing both CryAB R120G and RNAi targeted against Zw (the gene encoding G6PD) exhibited significantly shorter heart periods (increased heart rates; Figure 1B), significantly reduced arrhythmia indices and systolic diameters, and significantly greater fractional shortening ( Figure 1C-F). Interestingly, cardiac restricted expression of G6PD RNAi alone significantly decreased heart periods relative to those of w 1118 x HandGal4 control hearts ( Figure 1B).
Many of these differences in contracting heart tubes were qualitatively visualized via M-mode traces, which display the dynamics of cardiac contractions of representative hearts from the various genotypes ( Figure 2). These traces show the positions of the heart wall edges (Y direction) over time (X direction). M-modes from semi-intact heart preparations from control flies show fairly regular contractions. However, these traces reveal a subtly arrhythmic beating pattern in CryAB R120G cardiac tubes relative to controls. Further, the CryAB R120G hearts were dilated, and exhibited a lower extent of shortening. Co-expression of Zw RNAi increased heart rate, promoted rhythmic beating and rescued The abdominally located Drosophila heart (Hrt) is a simple linear tube that is divided into an anterior, conical chamber (CC) and three posterior compartments. A1 = abdominal segment 1; A6-A7 = abdominal segments 6 and 7; Os = ostia inflow tracts (modified from [66]). (B-F) Cardiac-restricted expression of wild type human CryAB (HG4.wtCryAB) had no significant effect on any measured parameter of cardiac performance in Drosophila. Several indices of cardiac function were influenced by the expression of mutant CryAB R120G (HG4.R120GCryAB). The diameter across the heart tubes during (D) diastole and (E) systole were significantly greater, and (F) percent fractional shortening significantly lower relative to HG4.wtCryAB flies. Knockdown of G6PD in CryAB R120G mutant hearts (HG4.R120GCryAB+G6PDR-NAi) substantially improved cardiac performance of the mutant hearts as evidenced by (B) significant decreases in heart period, (C) arrhythmic beating patterns, (E) systolic dimensions and (F) significantly increased fractional shortening relative to HG4.R120GCryAB hearts. * P#0.05, ** P#0.01, *** P#0.001, **** P#0.0001. doi:10.1371/journal.pgen.1003544.g001 percent fractional shortening in the CryAB R120G mutant hearts. Thus, overall, cardiac output of the mutant hearts is likely to be significantly enhanced by reducing the enzymatic activity of G6PD in flies as found in mouse models [7].

CryAB R120G -induced defects in non-heart tissues
We also asked whether CryAB expression produced any deleterious effects if expressed in other tissues. When CryAB + was expressed ubiquitously (using a Tub-Gal4 driver), or in the eye only (using either ey-Gal4 or GMR-Gal4 drivers) no abnormal phenotypes were observed in 32 of 33 lines, in spite of easily detectable expression (in eight lines examined by Western blotting; not shown). We did recover a single CryAB + transformant that produced a rough eye phenotype. However, the insertion was located upstream of the escargot (esg) gene, a location where misexpression elements are known to produce rough eyes by promoting esg expression [33]. We used qRT-PCR and verified that this particular CryAB + insertion also drove overexpression of esg. We therefore attribute its phenotype to esg overexpression, and not to CryAB + .
In contrast, we identified four UAS-CryAB R120G lines that produced complete or partial lethality with ubiquitous expression (using either A5C-Gal4 or Tub-Gal4 drivers). Western blotting of protein extracts from the single line with any survivors revealed abundant CryAB R120G expression. No expression was detected in seven fully viable lines ( Figure S3). We speculate that the high frequency of non-expressing transformant lines reflects a strong selection against lines that show even slightly leaky expression. Subsequent Western blots of flies with expression limited to the eye (which allows survival) revealed that the three lines that were lethal when constitutively expressed also had abundant Gal4-induced expression of CryAB R120G (Figure 3).
When CryAB R120G expression was driven in the wing, flies with deformed and mis-shapen wings were produced ( Figure 4B), while expression in the eye throughout development (with ey-Gal4) resulted in variably rough and small eyes ( Figure 4D). A stronger and more consistent eye phenotype was observed when expression was driven in differentiating eye discs (with GMR-Gal4), characterized by irregular patterning and loss of ommatidia and pigment ( Figure 4E).
Cardiomyopathy caused by mutations in CryAB is correlated with the presence of cytoplasmic protein aggregates containing CryAB and a number of other proteins [7,14]. To determine whether human CryAB R120G in Drosophila was also found in aggregates we extracted Triton-X 100 soluble and insoluble  protein fractions from eyes expressing CryAB + and CryAB R120G and used Western blotting to determine the distribution of CryAB between these fractions (Figure 3). We found a portion of the CryAB R120G mutant protein in the insoluble pellet fraction (6.6%62.9, 23%613, 1.9%61.0 and 4.9%61.0 for lines 7B, 13A, 14A and 16A, respectively), while CryAB + protein was found virtually exclusively in the soluble fraction (with only 0.3%60.2% and 0.2%60.1% in the pellet for lines 8B and 12A, respectively). Thus, CryAB R120G in Drosophila shares the characteristic of being found, in part, in insoluble aggregates.
The eye is dispensable for normal development and viability, and is an established model for examining the cellular effects of human neurodegenerative disease genes [34,35]. Therefore, we chose to use the phenotype produced by GMR-Gal4 UAS-CryAB R120G in the eye as a model to test the reductive stress hypothesis. Eye phenotypes were quantitated by scoring into five groups, ranging from ,wildtype (category 1), to having all or nearly all of the eye strongly affected (category 5; Figure 4E).
We then tested whether a reduction of G6PD could suppress the CryAB R120G phenotype, as it does in mouse and fly hearts. Two RNAi lines that significantly reduced expression of Zw both suppressed the CryAB R120G eye phenotype (Figure 5B,C; Table  S1). To extend these observations we then asked whether overexpression of Zw would exacerbate the CryAB R120G phenotype. Four overexpression lines were tested and, unlike the heart phenotype ( Figure S2), all strongly enhanced the CryAB R120G phenotype ( Figure 5E-H; Table S1). The results shown in Figure 5 using the CryAB R120G line 16A were confirmed with the independent CryAB R120G insertion line 14A that also has a strong eye phenotype: responses to Zw knockdown and overexpression were extremely similar for the two lines (not shown). There was no phenotypic effect of altered G6PD levels on normal eyes.
We considered the possibility that the UAS-controlled RNAi constructs might titrate GAL4 and produce apparent suppression simply through reduced expression of CryAB R120G . However, since the G6PD overexpression constructs are also driven by GAL4, but they enhance the CryAB R120G phenotype, this concern appears unfounded.
The enzyme 6-phosphogluconate dehydrogenase (PGD) acts downstream of G6PD in the pentose phosphate pathway, and like G6PD, reduces NADP to NADPH. We tested whether mutations in the gene encoding PGD (Pgd) would also affect the CryAB R120G phenotype. Two deletions that removed the gene (along with neighboring genes) suppressed the CryAB R120G phenotype when tested as heterozygotes ( Figure 6B,C; Table S1). In addition, flies carrying a chromosome with mutations in Zw and Pgd, in heterozygous condition, also showed strong suppression of the CryAB R120G phenotype ( Figure 6D). Finally, RNAi directed against Pgd produced very strong suppression of the CryAB R120G phenotype ( Figure 6E; Table S1). These mutants or RNAi knockdowns, by themselves, had no effect on the phenotype of normal eyes.
These results are consistent with the reductive stress hypothesis, but it is still conceivable that G6PD and PGD influence CryAB R120G pathology via other products of the pentose phosphate pathway. To determine whether varying the enzymatic production of NADPH would affect the CryAB R120G phenotype, regardless of the source of that variation, we tested whether alteration of isocitrate dehydrogenase (IDH) or malic enzyme (MEN) levels would affect the phenotype. We observed that RNAimediated knockdown of IDH (either the putative mitochondrial (CG6439) or cytoplasmic (CG7176) form) was capable of suppressing the CryAB R120G phenotype (Figure 7B-F; Table S1). Knockdown of MEN also suppressed the CryAB R120G phenotype ( Figure 7G; Table S1). Knockdown of IDH or MEN had no effect on normal eyes. Although only one of three lines expressing RNAi against Idh (CG7176) achieved significant suppression of the CryAB R120G phenotype, it is also the case that these lines only achieved an ,20-30% reduction in Idh RNA levels. In the cases of Zw, Pgd and Men, where strong suppression was observed, ,40-50% knockdowns of RNA levels were achieved (Table S1).
To assess the redox environment of cells in response to altered dosage of genes that mediate NADP/H metabolism we carried out two series of experiments to determine the ratio of reduced to oxidized glutathione (GSH:GSSG). When RNAi-mediated knockdown of G6PD, PGD, MEN or IDH was driven in heads with GMR-Gal4 the GSH:GSSG ratio was reduced in every case ( Figure 8A), though the reduction as a result of PGD knockdown was not significant at P = 0.05 level. The results confirm our supposition that these knockdowns impair cells' ability to generate NADPH. We observed a slight, though not statistically significant, increase in the GSH:GSSG ratio when G6PD was overexpressed in heads. However, when we assayed whole larvae, with expression driven ubiquitously, G6PD overexpression produced a very large and significant increase in the GSH:GSSG ratio ( Figure 8B). These results confirm the involvement of the NADP/ H network in cellular dysfunction produced by CryAB R120G expression and strongly implicate reductive stress as the causative agent for pathology.
One concern that arose when assessing the phenotypic effect of human CryAB R120G expression was that two Drosophila lines (7B and 13A) showed only mild phenotypic effects in the eye (Table S2), despite exhibiting robust protein expression ( Figure 3). However, these two lines do show rough eye phenotypes if subjected to a strong heat shock during late larval/early pupal development, a treatment that also strongly enhances the phenotype of lines 14A and 16A (Table S2). Flies carrying GMR-Gal4 alone showed no eye phenotype when subjected to this heat shock. Moreover, one of these lines (7B) also exhibits a wing phenotype when expression is driven in the wing (Table S2).
To test whether these particular insertions might be associated with suppression of the CryAB R120G phenotype, we combined them with GMR-Gal4 and the CryAB R120G line 16A that was used in the experiments of Figures 4, 5 and 7. The combination of line 7B or 13A with 16A resulted in an eye phenotype that was reduced from that of 16A alone, indicating that these two lines do suppress the CryAB R120G eye phenotype to some degree ( Figure S4). We have not determined why these two lines show little phenotypic consequence in the eye, but we speculate that it may have to do with subtle differences in timing of gene expression. Perhaps slightly earlier onset of expression in lines 7B or 13A results in a protective response, in the same way that mild heat shocks, by inducing synthesis of heat shock proteins, can protect against subsequent, more severe heat shocks [36,37]. Alternatively, the expression of genes in the region of the 7B and 13A insertion loci might be altered to result in phenotypic suppression. In spite of these lines showing only mild effects, our experiments convincingly show that Examples of wild-type (CryAB + ) and CryAB R120G phenotypes, respectively, are shown for expression in the wing (A, B) and expression in the undifferentiated eye disc driven by eyGal4 (C, D). When CryAB R120G expression is driven late in eye development with GMR-Gal4 a strong phenotype is produced with somewhat variable expressivity. The distribution of phenotypes, and corresponding images, are shown in E. The CryAB R120G line 16A was used for these tests. doi:10.1371/journal.pgen.1003544.g004 CryAB R120G expression, but not CryAB + , can cause strong phenotypic effects in the fly heart, eyes and wings. Moreover, detailed examination of the heart and wing defects show that they are responsive to altered levels of the enzymes that reduce NADP to NADPH.

Discussion
Our results show that Drosophila provides a suitable model in which to study the pathology of the human CryAB R120G mutation. Expression of the mutant allele, but not the wildtype, in fly hearts, causes heart dilation and dysfunction very reminiscent of the cardiomyopathy produced in humans carrying this dominant allele [2]. We also found that a reduced level of G6PD ameliorates many of the perturbed cardiac functional parameters in CryAB R120G flies, just as it does in the CryAB R120G mouse [7]. Despite a lack of reduction in CryAB R120G diastolic diameters in response to Zw (G6PD) RNAi co-expression, systolic diameters in the double mutants were rescued and did not significantly differ from those found in fly hearts expressing wildtype human CryAB. Thus, fractional shortening was indistinguishable between wildtype CryAB-expressing and CryAB R120G +Zw RNAi-expressing hearts. Furthermore, cardiac restricted expression of Zw RNAi, either with CryAB R120G or alone, significantly increased heart rates relative to control hearts. This suggests G6PD deficiency can improve cardiac output in either mutant or non-mutant backgrounds and may be a potent modifier of cardiac function.
In Drosophila, overexpression of G6PD can extend lifespan and protect against oxidative stress [38]. In mammalian cells, overexpression of wildtype small heat shock proteins leads to increased G6PD expression and protection against oxidative stress [39]. Our finding, that reduction of G6PD can be beneficial in some circumstances, is also not without precedent. Some studies have suggested a link between G6PD deficiency and protection against cardiovascular disease in humans, although such findings have not been replicated in larger patient studies [40][41][42]. Previous work from one of our laboratories showed that G6PD reduction is highly beneficial in one specific case -when human CryAB R120G is expressed in the mouse heart [7]. We do not see a conflict in these differing outcomes, but instead conclude that the effect of modifying G6PD levels may range from beneficial to deleterious, with the outcome determined by the constellation of genetic variation present in individuals' genomes and the environmental stressors that they experience.
To generate an experimental paradigm suited to rapid genetic exploration we expressed CryAB R120G in the fly eye and found that it strongly disturbs normal development and pattern formation. The eye phenotype was also responsive to altered G6PD levels, validating it as a model for investigation of the underlying mechanism of CryAB R120G pathology. Unlike a recent report [43], we saw no abnormal eye phenotype that could be attributed to expression of the wild-type human CryAB gene. In the one case where we did observe an eye phenotype it most likely resulted from induced expression of the esg gene that lay adjacent to that particular insertion of the CryAB transgene. Our findings indicate that CryAB R120G induces cellular dysfunction in both the heart and the eye, or lethality if expressed ubiquitously, while wild-type CryAB is relatively benign.
The human disease produced by the CryAB R120G allele is sometimes called Desmin-Related Myopathy (DRM), owing to the presence of desmin in the characteristic cytoplasmic protein aggregates, and to similarities with diseases caused by mutations in the gene encoding the intermediate filament desmin [2,6,44]. Although Drosophila do not have a desmin homolog, several lines of evidence argue that the cellular dysfunction in Drosophila resulting from CryAB R120G expression is, nonetheless, a legitimate model for this disease. First, a number of other proteins have been identified within the aggregates, including (at least) another small heat shock protein and G6PD [7,14], both of which have homologs in Drosophila. Second, CryAB R120G causes the formation of cytoplasmic aggregates even when expressed in human cell types that do not express desmin [45,46], and our results show that a portion of the CryAB R120G is also found in aggregates in Drosophila. Third, the identical response of mouse and Drosophila CryAB R120G pathologies to G6PD reduction strongly suggests an identical mechanism of action.
The pyridine nucleotides NADH and NADPH are essential cofactors of oxidative and reductive enzymatic processes involved in energetics, oxidative metabolism, redox homeostasis, calcium homeostasis, macromolecular biosynthesis, mitochondrial functions, gene expression, aging and cell death [47]. In this study, we examined the effect of altered levels of the four enzymes that are primarily responsible for reducing NADP to NADPH (Figure 9): two enzymes of the pentose phosphate pathway, G6PD and PGD which together account for ,40% of NADPH levels in the adult; MEN which generates pyruvate for import into mitochondria and accounts for another ,30%; and, IDH, which accounts for ,20% of NADPH [48][49][50]. These enzymes constitute a metabolic network linked by a common substrate (NADP) and interacting regulation [50]. A major finding of our study is that, even though G6PD, PGD, MEN and IDH carry out varied metabolic reactions, alterations to any of their activities have significant consequences for the phenotypes produced by CryAB R120G expression, implying a common mechanism of action through NADP/H.
In our experiments, reduction of IDH was less effective at CryAB R120G suppression than reductions of G6PD, PGD or MEN, a result that is not entirely surprising. Alteration of either G6PD or PGD activity is likely to affect both activities coordinately since they constitute sequential steps in the tightly regulated pentose phosphate pathway, and MEN by itself produces more NADPH than any of the other three enzymes of this network. IDH produces less NADPH than either MEN or the G6PD/PGD couple. Additionally, our RNAi-mediated knockdowns of IDH were relatively ineffective. What was surprising was that knockdown of the mitochondrial NAD-dependent IDH resulted in significant suppression of the CryAB R120G phenotype. We surmise that mitochondrial metabolism affects the cytoplasmic NADP/H network.
The NADP:NADPH redox couple, and the linked glutathione redox couple (GSSG:GSH), participate in a diverse array of biological processes. Therefore, we envision a number of possible mechanisms through which CryAB R120G could alter the cellular redox potential and thereby contribute to toxicity. Most obviously, redox-sensitive sequestration of both existing and newly synthesized proteins could seriously disturb cellular regulation. The function of many proteins depends on the reduced or oxidized state of thiol-containing cysteine residues. It is conceivable that structurally flexible hydrophobic protein surfaces, which are normally buried within a folded protein, could engage in nonproductive protein-protein interactions, in part, due to alterations of intra-chain disulfide links. Partially folded proteins might, in this way, become soluble toxic intermediates [9]. Alternatively, misfolding might occur as a result of alterations in other redoxsensitive post-translational modifications, for example glutathionylation, nitrosylation, and (de)acetylation. Co-aggregation of several distinct polypeptides might cripple multiple disparate functions within the cell. As mentioned, our results also suggest that alterations in mitochondrial homeostasis and energy metabolism could affect the levels of oxidized or reduced NADP/H. The reciprocal is most certainly true as well, with normal mitochondrial function dependent upon the function of the NADP-reducing enzyme MEN. Additionally there are scores of enzymes that use NADP/H as a cofactor, and the activity of one or more of these enzymes could be affected to generate the phenotypes we observed. It will require significant further work to identify the critical determinants of NADP/H involvement in CryAB R120G pathology. Drosophila provides powerful tools for genetic screening to identify such factors and the model for CryAB R120G pathology that we describe here provides a context for carrying out such screens. We anticipate that such efforts may ultimately lead to the identification of potential targets for therapy and the promise of useful treatments for the human disorders.

Construction of transgenic flies
P{UASP-CryAB + } and P{UASP-CryAB R120G }: A fragment DNA containing human CryAB cDNA was released from donor plasmids (pCMVHA-wtCryAB [51] or pCDH1-MCS1-R120GhCryAB (unpublished); by NcoI and DraI digestion, followed by Klenow filling in to make blunt ends. The vector pUASP [52] was digested with EcoRI or NotI respectively and filled in to make blunt ends. Ligation of respective inserts and vectors produced P{UASP-CryAB + } and P{UASP-CryAB R120G }. Each was verified by restriction enzyme mapping and sequencing, then was used in P-element mediated transformation by standard methods [53,54].
P{UASTattB-GFP-CryAB R120G} : The CryAB R120G cDNA was released from pCDH1-MCS1-R120GhCryAB by EcoRI digestion and was ligated in frame to the same site at the C terminus of GFP coding region an intermediate vector. The fusion protein was released using flanking XbaI sites and sub-cloned into pUASTattB vector provided by J. Bischof [55]. The plasmid was injected into y w P{ry +t7.2 , hsFLP}1; M{3xP3-RFP.attP}ZH-86Fb; M{vas-int.B}ZH-102D (from the Bloomington Drosophila stock center) to integrate the fusion protein construct onto chromosome 3R cytological location 86F. The X chromosome carrying FLP was removed by crossing after transformation.

Mapping of transformed P elements
Insertion sites of transformed P elements were mapped either by inverse PCR or Splinkerette PCR method [56].
Other Drosophila strains and culture conditions P{tub-GAL4}, P{ey-GAL4} and P{Act5C(FRT.y + )Gal4} flies were obtained from the Bloomington, IN, USA, Drosophila stock center (lines #5138, #5535 and #25374, respectively). After being exposed to FLP, P{Act5C(FRT.y + )Gal4} lost the y + marker and became recessive lethal. It was then kept as the balanced stock y w; P{Act5C-GAL4}/S 2 CyO. P{GMR-GAL4} flies are discussed in [57]. Fly lines carrying P{UAST-G6PD} along with control y w 2 were provided by W. C. Orr [38]. Mutations for Zw and Pgd were obtained from Bloomington stock center. RNAi fly stocks were from Vienna Drosophila RNAi Center [58]. Table S1 lists the stock numbers of all UAS-RNAi lines (Vienna Drosophila RNAi Center), Zw overexpression lines (W. C. Orr), and mutants (Bloomington Drosophila Stock Center). Flies were raised at 25uC, unless otherwise specified, on standard cornmeal-agar medium in standard 25690 mm vials.

Western blotting
Assay for CryAB R120G protein expression: The UASP-CryAB R120G lines were crossed to flies carrying the tub-GAL4 driver. One male and one female were taken from seven crosses with surviving progeny. Each fly was homogenized in 100 ml of 1X sample buffer. Eight ml of lysate was loaded in each lane of a 12% SDS-PAGE gel, separated by electrophoresis and then examined by Western blotting to detect the CryAB protein. A lysate of mammalian cells expressing CryAB R120G was included as a positive control. Rabbit antiserum (1:5000 dilution) against human CryAB protein was the primary antibody [7].
Assay for CryAB protein solubility: To determine whether CryAB expressed in eyes existed in a soluble or insoluble form we modified the procedure described by Carbone et al. [59]. Ten heads from females expressing CryAB (wildtype or mutant) were collected and homogenized for 10 minutes on ice in 100 ml lysis buffer (10 mM Tris pH 7.5, 5 mM EDTA, 1% NP40, 0.5% deoxycholate, 150 mM NaCl and 1% Triton X-100). After incubation in lysis buffer for 30 minutes on ice, samples were frozen at 220uC overnight, then thawed out. Cuticle and debris in the lysate were separated by brief centrifugation at 1,000 rpm for 1 minute. Supernatant was collected. Soluble and insoluble fractions were further separated by centrifugation at 14,000 rpm for 15 minutes. After collecting the soluble fraction, the insoluble fraction was washed three times with 200 ml lysis buffer each, and then solubilized in 40 ml 1X sample buffer for Western. The soluble fraction was TCA precipitated and washed before being resuspended in 40 ml 1X sample buffer. 2.5 head's worth of soluble or insoluble fraction (10 ml) was loaded into each lane.
Western blotting was carried out following standard procedures using the Odyssey Western Blot Kit (Li-cor). Rabbit antiserum against Human CryAB and mouse anti-b-tubulin (Developmental Studies Hybridoma Bank clone E7) were used as primary antibodies. After incubation with fluorescent anti-rabbit and anti-mouse secondary antibodies, the membrane was scanned on an infrared Odyssey scanner by Li-cor. The Western signal was quantified on the Li-cor scanner and the results from four independent experiments were averaged. The insoluble CryAB is reported as mean percent of total CryAB 6 standard error.

Scoring eye phenotypes
Stocks that carried the GMR-Gal4 driver and UAS-CryAB elements in homozygous condition were generated. Males from this stock were crossed to females from control lines (y w 2 or w 1118 ), or from lines carrying modifying elements or mutations. The eyes of daughters from these crosses were scored for the severity of the eye phenotype by assigning the eye to one of five categories ( Figure 1E). All transgenes were hemizygous in the scored females.
Mann-Whitney tests for significance were performed using GraphPad Prism software.

Quantitative RT-PCR
Total RNA from 15-25 female fly heads was harvested using Tri reagent and protocol (Sigma-Aldrich). cDNA was synthesized from total RNA using RevertAid First Strand cDNA Synthesis Kit (Fermentas). Quantitative PCR of the cDNA was carried out on an iQ-PCR machine (Bio-Rad) using Maxima SYBR green/ Fluorescein qPCR Master Mix (Fermentas). Relative copy number was calculated against a set of common standard templates for each PCR reaction. For each cDNA sample, the relative copy numbers of gene of interest (X) and ribosomal protein L32 (RPL) were both obtained. Abundance of X in the sample was calculated by dividing the copy number of X by that of RPL. The average of three independent experiments was used to represent the abundance of X in a given genotype.

Drosophila cardiac performance analysis
Two independent wildtype UAS-CryAB + controls and two UAS-CryAB R120G mutant fly lines, as well as UAS-CryAB R120G combined with UAS-Zw or UAS-Zw RNAi were crossed to Hand-Gal4 (II) driver flies (Hand is a direct target of Tinman and GATA factors during Drosophila cardiogenesis and hematopoiesis, [60]). As an additional control, Hand-Gal4 (II) driver-flies were crossed to w 1118 flies. The progeny were raised at 25uC on standard cornmeal-agar medium. All flies were transferred to fresh food every 2-3 days. At three weeks of age, 45-50 female offspring from each cross were anaesthetized and dissected. All procedures were done at room temperature (18-22uC) as previously described [17,24,32,61,62]. Briefly, each head, ventral thorax and ventral abdominal cuticle was removed exposing the abdomen [62]. All internal organs and abdominal fat were removed leaving only the heart and associated muscles for each fly. Dissections were performed in oxygenated adult hemolymph. The semi-intact preparations were allowed to equilibrate with oxygenation for 20-30 min prior to filming. Analysis of heart morphology and physiology was performed using high speed movies of the semi-intact Drosophila preparations. 30 sec. movies were taken at rates of 100-200 frames/sec. using a Hamamatsu EM-CCD digital camera on a Leica DM LFSA microscope with a 106 immersion lens. All images were acquired and contrast enhanced using Simple PCI imaging software (Compix, Inc.). M-modes were generated and determination of cardiac parameters, including heart periods, diastolic and systolic diameters, fractional shortening and arrhythmicity indices for each group was performed using a MatLab-based image analysis program [63]. The ''arrhythmicity index'', which is defined as the standard deviation of the heart period normalized to the median of each fly allowed us to quantify the average severity of arrhythmic beating patterns for each line. One-way ANOVAs of genotype as a function of each measured cardiac parameter, with Bonferroni multiple comparison tests, were employed to determine if significant differences among all Drosophila lines were present. P values,0.05 were considered significant.

Fluorescent Imaging of Drosophila cardiac tubes
Fluorescent imaging of Drosophila heart tubes were performed according to [64]. The UAS-GFP-CryAB R120G fly line was crossed to Hand-Gal4 (II) driver flies. The progeny were aged to 1 and 3 weeks. Beating hearts of semi-intact Drosophila were placed in artificial Drosophila hemolymph containing 10 mM EGTA. Cardiac tubes were examined to ensure contractions were inhibited. Hearts were fixed in 16PBS containing 4% formaldehyde at room temperature for 20 minutes with gentle shaking. Washing of hearts was performed three times for ten minutes with PBSTx (PBS containing 0.1% Triton-X-100) at room temperature with continual shaking. After washing, the hearts were incubated with Alexa584-phalloidin in PBSTx (1:1000) for 20 minutes with continual agitation. Washing of the hearts was again carried out three times for ten minutes with PBSTx at room temperature. The hearts were rinsed in 100 ml of PBS for 10 minutes. The specimens were mounted on microscope slides and viewed at 10-256 magnification using a Zeiss Imager Z1 fluorescent microscope equipped with an Apotome sliding module.

Glutathione assay of adult fly heads
Female flies from transgenic or control lines were crossed to w 1118 ; GMR-Gal4 males. 30-60 heads were collected from flies of the appropriate genotype for each cross and then immediately put in 60-120 ml (2 ml per head) 5% SSA (5-sulfosalicylic acid; Sigma #S-7408) and homogenized with a small eppendorf dounce pestle. Samples were then frozen at 280u and maintained frozen until assayed as described [65].

Larval glutathione assay
Female flies from UAS-RNAi, UAS-G6PD or control stocks were collected and crossed to y w; tub-Gal4/T(2;3)SM6, Cy; TM6,Tb males. Resulting third instar Tb + larvae were collected and washed with 0.7% NaCl solution. Each collection was weighed and frozen on dry ice. The collected larvae were then stored at 280uC until ready for testing. At least 30 mg of larvae from the same genotype were pooled for each sample. Larvae were homogenized in 1X GSH MES buffer from the Glutathione Assay Kit by Cayman Chemical Company (10 ml/mg sample). The protocol provided with the kit was followed. Total GSH and GSSG measurements were calculated based on the Kinetic Method with minimal time course of 30 minutes.

Data processing
At least three independent samples were assayed for each genotype. The GSH/GSSG ratio for each sample was calculated independently. Statistical analysis of GSH/GSSG ratios was conducted using Graphpad Prism. Figure S1 Human CryAB R120G -GFP fusion protein discretely associates with sarcomeric components of Drosophila cardiomyo-cytes. Cardiac-specific expression from Hand-Gal4(II).UAS-CryAB R120G -GFP flies results in fluorescently labeled and repetitive myofibrillar components of Drosophila cardiac fibers. As found in higher organisms, co-localization of GFP with a-actinin antibodies (not shown) suggests human CryAB likely associates with Drosophila Z-discs. (TIF) Figure S2 Physiological analysis of cardiac tubes from multiple UAS-controlled wildtype (9A and 29A) and mutant (14A and 16A) CryAB lines, and of mutant CryAB lines with Zw overexpression (+G6PD) or RNAi-mediated knockdown (+G6PDRNAi). The data suggest that CryAB R120G is deleterious to several indices of cardiac performance in the two independent mutant fly lines relative to non-mutant CryAB controls. Also, overexpression of Zw does not appear to exacerbate the mutant phenotype while Zw knockdown substantially improves it. Below the column graphs of each analyzed cardiac parameter are tables summarizing the results of a one way ANOVA followed by a Bonferroni's Multiple Comparison Test. Only comparisons with differences that reach statistical significance are shown. * P#0.05, ** P#0.01, *** P#0.001, **** P#0.0001. (TIF) Figure S3 UAS-CryAB R120G expression: Western blot of viable and semi-viable transgenic lines carrying tubGal4 and UAS-CryAB R120G transgenes. Only the semi-viable line 13A exhibited any expression, which was seen in males and females. The positive control (left lane) was a cell lysate of mammalian cells that expressed CryAB R120G . Numbers beneath each lane indicate independent transformed lines. (TIF) Figure S4 Suppression of CryAB R120G -16A eye phenotype by coexpression from CryAB R120G lines 7B and 13A. Flies that carried GMR-Gal4, CryAB R120G -16A, and either CryAB R120G -7B or CryAB R120G -13A were scored for eye phenotype. Both combina-tions showed a significant reduction in the severity of the eye phenotype compared to flies carrying only GMR-Gal4 and CryAB R120G -16A ( Figure 5A). (TIF)