The O-GlcNAc transferase OGT is a conserved and essential regulator of the cellular and organismal response to hypertonic stress

The conserved O-GlcNAc transferase OGT O-GlcNAcylates serine and threonine residues of intracellular proteins to regulate their function. OGT is required for viability in mammalian cells, but its specific roles in cellular physiology are poorly understood. Here we describe a conserved requirement for OGT in an essential aspect of cell physiology: the hypertonic stress response. Through a forward genetic screen in Caenorhabditis elegans, we discovered OGT is acutely required for osmoprotective protein expression and adaptation to hypertonic stress. Gene expression analysis shows that ogt-1 functions through a post-transcriptional mechanism. Human OGT partially rescues the C. elegans phenotypes, suggesting that the osmoregulatory functions of OGT are ancient. Intriguingly, expression of O-GlcNAcylation-deficient forms of human or worm OGT rescue the hypertonic stress response phenotype. However, expression of an OGT protein lacking the tetracopeptide repeat (TPR) domain does not rescue. Our findings are among the first to demonstrate a specific physiological role for OGT at the organismal level and demonstrate that OGT engages in important molecular functions outside of its well described roles in post-translational O-GlcNAcylation of intracellular proteins.


Introduction
6 non-viable under a specific physiological condition, hypertonic stress. Through an unbiased screen, we identified ogt-1 as being required for expression of the osmosensitive gpdh-1p::GFP reporter. We found that under hypertonic stress conditions, ogt-1 is required for accumulation of GPDH-1 protein, but not gpdh-1 mRNA. Additionally, ogt-1 mutants are unable to grow and reproduce following exposure to mild hypertonic environments. Finally, we demonstrate that expression of human OGT can rescue the C. elegans hypertonic stress phenotype. The ability of either human or C. elegans OGT to rescue is independent of O-GlcNAcylation catalytic activity. These results demonstrate for the first time a specific role for OGT in the essential process of osmoregulation and suggest that this function is conserved across >700 million years of evolution.

ogt-1
In C. elegans, hypertonic stress rapidly and specifically upregulates expression of the osmolyte biosynthesis gene gpdh-1, which we visualized with a gpdh-1p::GFP transcriptional reporter (10,11). To optimize this reporter for genetic screening, we added a col-12p:dsRed reporter, whose expression is not affected by hypertonic stress and serves as an internal control for non-specific effects on gene expression (21). This dual reporter strain (drIs4) expresses only dsRed under isotonic conditions and both dsRed and GFP under hypertonic conditions, with very few animals exhibiting an intermediate phenotype ( Figure 1A -C). A gpdh-1p::GPDH-1::GFP translational reporter (kbIs6) is also upregulated by hypertonic stress, but exhibits more variability than the drIs4 transcriptional reporter ( Figure 1D -F).
Taking advantage of the binary nature of GFP activation by hypertonic stress in the drIs4 strain, we designed an unbiased F2 forward genetic screen for mutants that fail to activate GFP expression during hypertonic stress, but exhibit no effects on RFP ( Nio (no induction of osmolyte biosynthesis gene expression) mutants; Figure 2A). From this screen of ~120,000 haploid genomes, we identified two recessive alleles, dr15 and dr20, that genetically fail to complement each other. Whole genome sequencing and bioinformatics revealed that each allele contained a distinct nonsense mutation in the gene encoding the O-GlcNAc transferase ogt-1 (Table S1 -3, Figure 2B). Two independently isolated ogt-1 deletion alleles, ok430 and ok1474, as well as wild type worms exposed to ogt-1(RNAi) also exhibited a Nio phenotype, and ok430 and ok1474 8 failed to complement the dr15 and dr20 alleles. (Figure 2B -D, S1A -B, Table S2).
CRISPR reversion of the dr20 Q600STOP mutation back to wild type, as well as transgenic overexpression of wild type ogt-1 in the dr20 mutant, was sufficient to rescue the ogt-1 Nio phenotype, indicating that other ENU induced mutations in the background do not contribute to the Nio phenotype ( Figure 2E and S1C). Finally, we found that knock down of ogt-1 during post-developmental stages with ogt-1(RNAi) was sufficient to cause a Nio phenotype, suggesting that OGT-1 is not required for the establishment of developmental structures necessary for responding to hypertonic stress ( Figure 2F).
The function of ogt-1 in the response to hypertonic stress is specific because inhibition of ogt-1 did not affect either heat shock or endoplasmic reticulum stress inducible reporter expression ( Figure S2). In conclusion, these results suggest that ogt-1 is specifically and acutely required for hypertonic stress-induced upregulation of gpdh-1 expression.
OGT-1 is required for osmosensitive protein expression, but not osmosensitive transcription.
Since ogt-1 is required for induction of the gpdh-1p::GFP transgenic reporter by hypertonic stress, we hypothesized that endogenous osmosensitive mRNAs would not be upregulated in an ogt-1 mutant. To test this, we used qPCR to measure the expression levels of several previously described mRNAs that are induced by osmotic stress (11). Surprisingly, we found that osmosensitive mRNA expression was still upregulated in ogt-1 mutants ( Figure 3A and S3A). Consistent with this, we also observed that GFP mRNA derived from the gpdh-1p::GFP reporter was upregulated by hypertonic stress in ogt-1 mutants even though GFP protein levels were strongly reduced ( Figure 3B, 2D, S1A). These data unexpectedly suggest that OGT-1 regulates osmosensitive gene expression at a post-transcriptional level. 9 To further examine if OGT-1 affects the coupling between hypertonic stress induced mRNA and protein expression, we measured GPDH-1 protein levels in ogt-1 mutants.
However, the mRNA from this translational reporter was still induced to wild type levels ( Figure 3D). Importantly the requirement for ogt-1 in the hypertonic stress response is not transgene dependent because ogt-1 is also required for the hypertonic induction of CRISPR/Cas9 engineered GPDH-1::GFP fusion protein, which we confirmed to be functional ( Figure 3E and S4C). In conclusion, these results suggest that OGT-1 functions downstream of osmosensitive mRNA upregulation, but upstream of osmosensitive protein expression.
Physiological and genetic adaptation to hypertonic stress requires ogt-1 C. elegans upregulate osmosensitive genes, including gpdh-1, to survive and adapt to hypertonic challenges. We found that loss of ogt-1 had no effect on acute survival during hypertonic stress ( Figure S4A) (10). However, loss of ogt-1 blocked the ability of animals to adapt, grow, and reproduce under mild hypertonic stress ( Figure 4A and S4B). This adaptation phenotype was rescued by CRISPR reversion of the dr20 Q600STOP mutation to wild type ( Figure S4C). Interestingly, the adaptation phenotype of ogt-1 mutants must extend beyond its effects on gpdh-1, since the adaptation phenotype of a gpdh-1 presumptive null mutant is not as severe as that observed in an ogt-1 mutant ( Figure 4A).
In addition to physiological exposures, adaptation to hypertonic stress can also be induced genetically via loss of function mutations in several hypodermis expressed secreted extracellular matrix (ECM) proteins (11,12). These mutants exhibit maximal induction of gpdh-1 mRNA and accumulation of glycerol. As a result these mutants are constitutively adapted to survive normally lethal levels of hypertonic stress (11,12). To test if ogt-1 is required for genetic adaptation to hypertonic stress, we introduced an ogt-1 mutation into osm-8(dr9) or osm-11(n1604) mutants. Both osm-8 and osm-11 mutants exhibit constitutively elevated gpdh-1p::GFP expression under isotonic conditions. However, gpdh-1p::GFP levels were significantly reduced in osm-8;ogt-1 and osm-11;ogt-1 double mutants ( Figure 4B and S4D). Consistent with this observation, the ability of osm-8 mutants to survive a lethal hypertonic stress was suppressed in the osm-8;ogt-1 double mutants ( Figure 4C) (12). These data suggest that ogt-1 is required for both physiological and genetic adaptation to hypertonic stress caused by loss of the ECM proteins OSM-8 and OSM-11.
Non-canonical activity of ogt-1 in the hypodermis regulates gpdh-1 induction by hypertonic stress through a functionally conserved mechanism.
In C. elegans, functional endogenously GFP tagged ogt-1 is ubiquitously expressed in the nucleus, consistent with previous observations ( Figure S4C and S5A) (22). Therefore, we used tissue specific promoters to test which tissues require ogt-1 expression for gpdh-1 induction by hypertonic stress. The expression of ogt-1 from either its native promoter or a hypodermal specific promoter was sufficient to rescue gpdh-1 induction by hypertonic stress in ogt-1 LOF mutants. However, expression of ogt-1 in intestine, muscle, or neurons did not rescue ( Figure 5A -B). Since gpdh-1 is 11 induced by hypertonic stress in the hypodermis, these results suggest that ogt-1 acts cell autonomously in the hypoderm to regulate osmosensitive protein expression.
In agreement with the results from the catalytically dead human OGT, C. elegans expressing catalytically impaired alleles of endogenous ogt-1 induced gpdh-1p::GFP during hypertonic stress and had normal adaptation to hypertonic stress ( Figure 5E -G).
In conclusion, a non-catalytic function of OGT-1 in the hypodermis is required for osmosensitive protein induction by hypertonic stress. Importantly, this non-catalytic stress-responsive function is conserved from C. elegans to human OGT.

Discussion
Through an unbiased forward genetic screen for mutants that disrupt osmosensitive expression of a gpdh-1::GFP reporter in C. elegans, we identified multiple alleles of the O-GlcNAc transferase OGT-1. This ogt-1 phenotype is specific to hypertonic stress since ogt-1 is not required for upregulation of other stress responsive reporters. ogt-1 likely functions as a key signaling component of the hypertonic stress response, since post-developmental knockdown of ogt-1 was sufficient to produce the Nio phenotype. ogt-1-dependent signaling in the hypertonic stress response appears to occur primarily in the hypodermis, a known osmosensitive tissue in C. elegans (10,12).
The mechanism by which ogt-1 regulates hypertonicity induced gene expression was unexpected. ogt-1 mutants exhibited normal hypertonicity induced upregulation of stress response mRNAs. However, levels of the encoded proteins were significantly reduced, suggesting ogt-1 acts via a post-transcriptional mechanism(s). ogt-1 mutants are completely unable to adapt and reproduce in hypertonic environments and this correlates with an inability of ogt-1 mutants to properly upregulate the translation of osmoprotective proteins such as gpdh-1. Intriguingly, we demonstrate that the function of ogt-1 in the hypertonic stress response does not require O-GlcNAcylation catalytic activity. Both wild type and catalytically inactive human OGT can rescue the C. elegans ogt-1 Nio phenotypes, suggesting that this non-catalytic function is also conserved with humans ( Figure 6). C. elegans is the primary genetic model system for studies of ogt-1 because it is the only organism in which loss of ogt-1 is viable (25,26). This has allowed many previous studies to parse the roles of ogt-1 in lifespan, metabolism, innate immunity, behavior, neuron function, stress responses, cell fate, and autophagy (18,22,25,(27)(28)(29)(30)(31)(32)(33)(34)(35)(36). Importantly, most of these studies utilized global ogt-1 knockdown, which eliminates both O-GlcNAcylation-dependent and -independent functions of ogt-1. The missense alleles generated here will provide powerful tools for differentiating between these functions.

13
Our studies reveal a critical and previously unappreciated condition-specific role of OGT-1 in adaptation to hypertonic stress. This phenotype is completely penetrant and one of the strongest ogt-1 phenotypes described to date. Although their ability to survive acute hypertonic stress is unaffected, ogt-1 mutants are unable to adapt and reproduce following extremely mild shifts in extracellular osmolarity (250 mM NaCl). Such conditions have minimal effects on the ability of wild type animals to adapt and reproduce (10). This suggests a critical physiological role of ogt-1 in the ability of C. elegans to survive in the wild, since worms are continuously exposed to fluctuating environmental salinity in their native ecosystems (37). Given that both C. elegans and human OGT are able to rescue the Nio phenotype of ogt-1 mutants in C. elegans, we speculate that OGT plays an ancient and conserved physiological function in response to environmental and physiological perturbations in osmotic homeostasis.
In mammals, OGT is essential for cell division, a physiological process that involves tight regulation of cell volume (26,(38)(39)(40). Therefore, we speculate that OGT may be required for mammalian cell division for the same reason it is essential for adaptation to hypertonic stress in C. elegans: it plays a critical role in cell volume regulation. One reason mammalian cells may be unable to divide without OGT is because they cannot properly regulate cell volume during cell division. In C. elegans, unlike in mammals, ogt-1 is not an essential gene. We hypothesize that the osmotic homogeneity of standard C. elegans lab culture conditions allows ogt-1 mutants to survive and propagate normally. However, under hypertonic conditions, ogt-1 becomes an essential gene in C. elegans, like it is in humans. It will be interesting to explore the roles and requirements of OGT in cell volume regulation in mammalian cells and tissues.
14 Knockout of OGT in mammalian cells leads to a rapid loss in cellular viability (26). This phenotype is largely thought to be due to loss of O-GlcNAcylation activity.
However, data from human cells suggest that O-GlcNAcylation activity may not be the essential function of OGT. For example, exposure of mammalian cells to the O-GlcNAc inhibitor Ac4-5SGlcNAc largely blocks O-GlcNAcylation, but cellular viability and division are unaffected (41). Additionally, cells and humans carrying inherited catalytic point mutations in OGT associated with intellectual disability are viable (42). Our data show that the role of OGT-1 in the C. elegans hypertonic stress response is also independent of catalytic activity. Such catalytically-independent roles of OGT-1 have also been described in the context of synaptic regulation and cell adhesion (17,18). If the evolutionarily critical role of OGT in mammalian cells is related to its ability to regulate cell volume, our data suggest that such functions are independent of O-GlcNAcylation activity. These non-catalytic functions of OGT and the protein domains that regulate these functions are largely unexplored. The C. elegans Nio phenotype may provide a powerful genetic system for identifying new functional domains important for OGT function via targeted and unbiased genetic screening strategies.
Cell volume regulation during environmental stress requires upregulation of osmoprotective proteins, including those that regulate osmolyte accumulation. In almost all cases, these genes are upregulated at the transcriptional level (11). Our findings are the first evidence that this pathway is also under post-transcriptional control. OGT-1 is required for the accumulation of GPDH-1 protein during hypertonic stress, but not for the upregulation of gpdh-1 mRNA. Interestingly, this is not a complete elimination of GPDH-1 protein induction and even if it were, gpdh-1 null mutants still retain significant hypertonic adaptation potential, whereas ogt-1 mutants are completely adaptation deficient. These results suggest that OGT-1 regulation of the hypertonic stress response is likely to extend beyond its effects on post-transcriptional GPDH-1 induction.
The nature of these additional targets and/or mechanisms is currently unknown. can also catalyze a unique type of proteolysis event, this activity is not thought to occur in C. elegans (46). Regardless, the K957M mutation also eliminates the known proteolytic activity of ogt-1, suggesting that this function is also not required in the hypertonic stress response (24). Future studies, utilizing both targeted ogt-1 deletion alleles and unbiased genetic screens for new ogt-1 missense mutations with a Nio phenotype, should help resolve this question. Second, the precise post-transcriptional mechanism under OGT-1-dependent control remains unknown. Such mechanisms could include mRNA cleavage and polyadenylation site usage, mRNA nuclear export, selective interactions between ribosomes and stress-induced mRNAs, or regulated proteolysis of stress-induced proteins such as GPDH-1. While most of these potential mechanisms await testing, we find that inhibition of either autophagic or proteasomemediated proteolysis does not appear to be involved ( Figure S6). Finally, it remains unclear which genes ogt-1 coordinates with to regulate hypertonic stress signaling.
Future studies analyzing new Nio mutants should shed light on these interactions.
In conclusion, our unbiased genetic screening approaches in C. elegans have revealed a previously unappreciated requirement for non-canonical OGT signaling in a critical and conserved aspect of cell physiology. The primary function of OGT has long been assumed to be due to its catalytic O-GlcNAcylation activity. However, as we and others have shown, OGT also has critical and conserved non-catalytic functions that warrant further study (17,18). It is vital that future studies involving OGT utilize point mutants that differentiate canonical from non-canonical functions rather than OGT knockouts, which ablate both. As our studies have shown, such approaches could reveal new roles for this key protein in unexpected aspects of cell physiology.
The homozygous genotype of every strain was confirmed either by DNA sequencing of the mutant lesion, restriction digest, or a loss of function phenotype.

Genetic methods
ENU mutagenesis and mutant isolation: L4 stage drIs4 animals (P0) were mutagenized in 0.6 mM N-ethyl-N-nitrosourea (ENU) diluted in M9 for 4 hours at 20°C. One day after ENU mutagenesis, F1 mutagenized eggs were isolated by hypochlorite solution and hatched on NGM plates overnight. Starved ENU mutagenized F1 drIs4 L1 animals were washed twice in 1 x M9 and seeded onto 3 -16 10 cm OP50 NGM plates. F2 synchronized larvae were obtained via hypochlorite synchronization and seeded onto OP50 NGM plates. Day one adult F2 drIs4 animals were transferred to 250 mM NaCl OP50 NGM plates for 18 hours. As controls, unmutagenized drIs4 day 1 adults were also transferred to 50 mM NaCl and 250 mM NaCl OP50 NGM plates for 18 hours. After 18 hours, RFP and GFP fluorescence intensity, time of flight (TOF), and extinction (EXT) were acquired for each animal using a COPAS Biosort (Union Biometrica, Holliston, MA). Using the unmutagenized 50 mM NaCl NGM data as a reference, gate and sort regions for animals exposed to 250 mM NaCl were defined that isolated rare mutant animals with GFP and RFP levels similar to the population of unmutagenized drIs4 animals on 50 mM NaCl.
These mutants were termed nio mutants (no induction of osmolyte biosynthesis gene expression). Individual nio mutant hermaphrodites were selfed and their F3 and F4 progeny re-tested to confirm the Nio phenotype.
Backcrossing and single gene recessive determination: Each nio mutant was backcrossed to drIs4 males three times. F1 progeny from these backcrosses were tested on 250 mM NaCl for 18 hours as day 1 adults. As expected for a recessive mutant, 100% of the crossed progeny were WT (non-nio). F1 heterozygous hermaphrodites from these crosses were selfed and their progeny (F2) were tested on 250 mM NaCl for 18 hours as day 1 20 adults. As expected for a single gene recessive mutation, ~25% of progeny exhibited the Nio phenotype (Table S1).
Complementation testing: nio/+ males were crossed with hermaphrodites homozygous for the mutation being complementation tested. The F1 progeny from this cross were put on 250 mM NaCl OP50 NGM plates for 18 hours and screened for complementation.
Crosses in which ~50% of these F1 progeny were WT failed to complement (i.e. were alleles of the same gene). Crosses in which 100% of these F1 progeny were WT complemented (i.e. represented alleles of different genes). Each

COPAS Biosort Acquisition and Analysis
Day one adults from a synchronized egg lay or hypochlorite preparation were seeded on 50 or 250 mM NaCl OP50 or the indicated RNAi NGM plates. After 18 hours, the GFP and RFP fluorescence intensity, time of flight (TOF), and extinction (EXT) of each animal was acquired with the COPAS Biosort. Events in which the RFP intensity of adult animals (TOF 400-1200) was <20 (dead worms or other objects) were excluded from the analysis.
The GFP fluorescence intensity of each animal was normalized to its RFP fluorescence intensity or TOF. To determine the fold induction of GFP for each animal, each GFP/RFP or GFP/TOF was divided by the average GFP/RFP or GFP/TOF of that strain exposed to 50 mM NaCl. The relative fold induction was determined by setting the fold induction of drIs4 exposed to 250 mM NaCl to 1. Each graphed point represents the quantified signal from a single animal.

Molecular Biology and Transgenics
Reporter strains: The drIs4 strain was made by injecting wild type animals with gpdh-1p::GFP (20ng/µL) and col-12p::dsRed2 (100ng/µL) to generate the extrchromosomal array drEx73, which was integrated using UV bombardment, followed by isolation of animals exhibiting 100% RFP fluorescence. The resulting strain was outcrossed five times to wild type to generate the homozygous integrated transgene drIs4. kbIs6 was generated from a Gene Gun bombardment of unc-119(ed3) animals with a gpdh-1p::gpdh-1::GFP plasmid and an unc-119(+) rescue plasmid (pMM051). The resulting strain was outcrossed five times to generate kbIs6. drIs4 is integrated on LGIV. The integration site for kbIs6 is unmapped.
Transgene rescue: All the primers used to generate the rescue constructs can be found in Table S5. The genomic ogt-1 rescue construct (used in the drEx465 extrachromosomal CRISPR/Cas9 genomic editing: CRISPR allele generation was preformed using the single-stranded oliodeoxynucelodite donors (ssODN) method (48). gRNA and repair 23 template sequences are found in Table S5. For identification of the dr20 allele, we performed RFLP analysis using the MboI restriction enzyme, which cuts the WT allele, but not dr20. For identification of the dr86, dr89, dr90, and dr91 alleles, we performed RFLP analysis using the DdeI restriction enzyme, which cuts the mutant alleles, but not WT. To make the gpdh-1::GFP CRISPR strain, we used a previously described double stranded DNA (dsDNA) asymmetric-hybrid donor method (48). To make the ogt-1::GFP CRISPR strain, we used a dsDNA donor method using Sp9 modificied primers (49). Homozygous CRISPR/Cas9 generated alleles were isolated by selfing heterozygotes to ensure that complex alleles were not obtained.
mRNA isolation, cDNA synthesis, and qPCR: Day one animals were plated on 50 mM or 250 mM NaCl OP50 NGM plates for 24 hours. Unless noted otherwise, after 24 hours, 35 animals were picked into 50 µL Trizol for mRNA isolation. RNA isolation followed a combined Trizol/RNeasy column purification method as previously described (11). cDNA was synthesized from total RNA using the SuperScript VILO Master Mix. SYBER Green master mix, 2.5 ng cDNA, and the primers listed in Table S5 were used for each qPCR reaction. qPCR reactions were carried out using an Applied Biosystems 7300 Real Time PCR machine. act-2 primers were used as a control for all qPCR reactions. At least three biological replicates of each qPCR reaction were carried with three technical replicates per biological replicate. qPCR data was analyzed through ∆∆Ct analysis with all samples normalized to act-2. Data are represented as fold induction of RNA on 250 mM NaCl relative to on 50 mM NaCl.
Western Blots: Cell lysates were prepared from hypochlorite synchronized day 1 adult animals exposed to 50 mM or 250 mM NaCl plates for 18 hours. 3-5 non-starved 10 cm plates were concentrated into a 100 µL mixture. NuPage LDS Sample Buffer (4X) and NuPAGE Sample Reducing Agent (10X) were added and the sample was frozen and thawed three times at -80°C and 37°C. Prior to get loading, the sample was heated to 100°C for 10 minutes and cleared by centrifugation at 4°C, 12,000 x g for 15 minutes. The

Microscopy
Worms were anesthetized (10mM levamisole) and mounted on agar plates for widefield fluorescence microscopy. Images were collected on a Leica MZ16FA stereo dissecting scope with a DFC345 FX camera (Leica Microsystems, Wetzlar, Germany). Unless noted, images within an experiment were collected using the same exposure and zoom settings.
Unless noted, images depict merged GFP and RFP channels of age matched day 2 adult animals exposed to 50 or 250 mM NaCl for 18 hours.

25
Embryos from a hypochlorite preparation were freeze-cracked on a superfrost slide, fixed with paraformaldehyde, blocked with bovine serum albumin (BSA), incubated with a 1:400 dilution of α-O-GlcNAc monoclonal antibody (RL2) overnight, and incubated with 1:400 dilution of 1:400 goat α-mouse IgG, IgM (H+L) Secondary Antibody, Alexa Fluor 488 for 4 -6 hours. The antibodies used are listed in Table S7. Washes with PBS or antibody buffer were carried out between each incubation step. Images were collected on a Leica MZ16FA stereo dissecting scope with a DFC345 FX camera (Leica Microsystems, Wetzlar, Germany). Images were taken with the same zoom settings, but different exposure settings.

C. elegans Assays
Adaptation Assay: Day one adult animals were transferred to five 50 mM NaCl OP50 NGM plates and five 200 mM NaCl OP50 plates. ~25 animals were transferred to each plate. After 24 hours, 20 animals from each 50 mM or 200 mM plate were transferred to 600 mM NaCl OP50 NGM plates. Animals were scored for movement after 24 hours on the 600 mM NaCl OP50 NGM plates. To be counted as moving, the animal had to move greater than half a body length. Animals that were not moving were lightly tapped on the nose to confirm that they were paralyzed or dead. Survival assays and osmotic stress resistance (Osr) assays were performed as previously described (11).

Statistical Analysis
Comparisons of means were analyzed with either a two-tailed Students t-test

Declaration of Interests
The authors declare no competing interests.   (D) qPCR of gpdh-1 mRNA from day 2 adult animals expressing the kbIs6 translational fusion exposed to 50 or 250 mM NaCl NGM plates for 24 hours. Strains include WT and ogt-1(dr34). The ogt-1(dr34) allele is the dr20 point mutation introduced using  (A) Wide-field fluorescence microscopy of day 2 adult animals expressing drIs4 exposed to 50 or 250 mM NaCl NGM plates for 18 hours. Strains express an ogt-1 cDNA from the indicated tissue-specific promoter. Images depict the GFP channel only for clarity.

Figure legends
The RFP signal was unaffected in these rescue strains (not shown). Scale bar = 100 microns.   represented as fold induction of normalized GFP/TOF ratio of animals exposed to heat shock conditions relative to animals exposed to control conditions. Each point represents the quantified signal from a single animal. Data are expressed as mean ± S.D. **** -p<0.0001 (One-way ANOVA with post hoc Dunnett's test). N ≥ 158 for each group.
(D) COPAS Biosort quantification of GFP and TOF signal from animals in (C). Data are represented as fold induction of normalized GFP/TOF ratio of animals exposed to DTT plates relative to animals exposed to control plates. Each point represents the quantified signal from a single animal. Data are expressed as mean ± S.D. **** -p<0.0001 (Oneway ANOVA with post hoc Dunnett's test). N ≥ 48 for each group.  (D) COPAS Biosort quantification of GFP and RFP signal in day 2 adult drIs4 animals drIs4 exposed to 50 mM NaCl NGM plates. Data are represented as the fold induction of normalized GFP/RFP ratio on 50 mM NaCl NGM plates, with osm-11(n1604) set to 1.