Role of UDP-Glycosyltransferase (ugt) Genes in Detoxification and Glycosylation of 1-Hydroxyphenazine (1-HP) in Caenorhabditis elegans

Caenorhabditis elegans is a useful model organism to study the xenobiotic detoxification pathways of various natural and synthetic toxins, but the mechanisms of phase II detoxification are understudied. 1-Hydroxyphenazine (1-HP), a toxin produced by the bacterium Pseudomonas aeruginosa, kills C. elegans. We previously showed that C. elegans detoxifies 1-HP by adding one, two, or three glucose molecules in N2 worms. Our current study evaluates the roles that some UDP-glycosyltransferase (ugt) genes play in 1-HP detoxification. We show that ugt-23 and ugt-49 knockout mutants are more sensitive to 1-HP than reference strains N2 or PD1074. Our data also show that ugt-23 knockout mutants produce reduced amounts of the trisaccharide sugars, while the ugt-49 knockout mutants produce reduced amounts of all 1-HP derivatives except for the glucopyranosyl product compared to the reference strains. We characterized the structure of the trisaccharide sugar phenazines made by C. elegans and showed that one of the sugar modifications contains an N-acetylglucosamine (GlcNAc) in place of glucose. This implies broad specificity regarding UGT function and the role of genes other than ogt-1 in adding GlcNAc, at least in small-molecule detoxification.


■ INTRODUCTION
Caenorhabditis elegans are bacterivores found in soil and decaying organic matter. 1 As they feed on bacteria in their environment, they are exposed to numerous pathogens and xenobiotics. 2To combat exposure, worms have developed three main strategies for defense.One is avoidance, where they can sense potentially hostile environments and avoid going to them. 3 Another is the presence of a strong cuticle and pharyngeal grinder to physically prevent pathogens from entering the worm. 4Finally, if pathogens can enter the worm, several mechanisms are activated, constituting the innate immune response.
Along with the pathogen response, the C. elegans innate immune system is activated upon xenobiotic exposure.Xenobiotics are defined as substances foreign to a body or ecological system.In nature, C. elegans feed on various bacteria, many of which produce compounds that are toxic to the worms.As a result, C. elegans has developed a wide array of detoxification enzymes. 5Xenobiotic metabolism is canonically divided into three phases. 6Phase I is the addition of reactive moieties such as hydroxyl groups to the parent xenobiotic.Phase II is the conjugation of either the phase I modified xenobiotic or parent xenobiotic to a water-soluble molecule to facilitate excretion.Phase III is the transport of these metabolized compounds out of the cell. 7In this study, we focus on steps involved in the phase II detoxification of one such xenobiotic: 1-hydroxyphenazine (1-HP).
1-Hydroxyphenazine (1-HP) is a small molecule produced by many Pseudomonas species, including Pseudomonas aeruginosa, a Gram-negative bacterium that causes disease in higher eukaryotes such as humans. 8,9In humans, 1-HP is known to affect ciliary function, especially in patients with cystic fibrosis. 91-HP is one of three related metabolites, along with pyocyanin and phenazine-1-carboxylic acid, produced by P. aeruginosa that are toxic to C. elegans. 101-HP is thought to cause chronic toxicity in C. elegans by generating α-synucleinand polyglutamine-induced protein misfolding and exacerbating α-synuclein-induced dopaminergic neurodegeneration. 11C. elegans modifies 1-HP by adding one, two, or three glucose moieties, with phosphorylation also observed in the endometabolome. 12In this study, we sought to more thoroughly characterize the metabolized 1-HP derivatives produced by C. elegans and to conduct preliminary experiments on the ugt gene family, which is likely involved in at least some of these modifications.
UGTs are a family of enzymes critical for the homeostatic regulation of endogenous metabolites and xenobiotic detoxification in several organisms, including humans and C. elegans. 13While humans only have 22 ugt genes, C. elegans have over 70. 14,15Not all C. elegans ugt genes have obvious human homologues, but they do have homologues in parasitic nematodes, making them potential targets to combat anthelmintic resistance. 16UGTs are the primary protein family responsible for adding glucose moieties during phase II xenobiotic detoxification in C. elegans. 7Loss or modification of UGTs has been implicated in drug hypersensitivity. 16,17hey have also been shown to be upregulated upon exposure to metals, pathogenic toxins, anthelmintics, and other small molecules. 16,18RESULTS AND DISCUSSION Quantitation of 1-HP Derivatization in ugt Mutants.In this study, we tested available ugt mutants for their involvement in 1-HP modification and susceptibility (Table 1, Supporting Information Table 1).We analyzed the phylogeny of these ugt genes and found that they covered most of the clades in the UGT family 15 (Supporting Information Figure 1).We then adapted a plate-based mortality screen from our previous work to discover strains with modified sensitivity to 1-HP exposure at the LD 50 (179 μM) concentration (Figure 1A). 23All strains were paired with N2, and PD1074 replicates were used as reference controls.PD1074 is 99.98% identical to N2 and is currently recommended as the new reference C. elegans strain because N2 has significantly diverged over time and is no longer a reliable reference. 21We found that all strains had higher mortality when exposed to 179 μM 1-HP than the bacteria control.The solvent, 1.1% dimethyl sulfoxide (DMSO), sometimes trended to higher mortality, but this was not statistically significant after performing a one-way analysis of variance (ANOVA) for each strain (Supporting Information Figure 2B).
We tested 11 different strains for 1-HP exposure (Table 1).Of these strains, N2 and PD1074 had already been tested in previous studies. 12,23In our assay, N2 worms had a mean mortality of 48.2% with a standard error of 2.6 (n = 23), while PD1074 had a mean of 49.5% with a standard error of 2.7 (n = 21).We performed a one-way ANOVA in both cases, suggesting that the mortality by 1-HP exposure was significantly greater than in both the bacteria-only and DMSO controls (P < 0.0001) (Supporting Information Figure 2A).In the other strains we tested (  Figure 2A).The strain ugt-23 had a mortality of 64.7% with a standard error of 3.7 (n = 10), and ugt-49 had a mortality of 68.7% and a standard error of 3.8 (n = 8) compared to controls (P < 0.0001).Finally, we performed a nonparametric Wilcoxon pairwise analysis for 1-HP mortality between all of the strains, followed by a Benjamini−Hochberg Correction.We found that both the ugt-23 and ugt-49 mutants had significantly increased mortality compared to N2 (P < 0.1) (Figure 1B).This suggests that these genes play a role in the glycosylation of 1-HP.It is important to note that several factors might influence these results, most notably the genetic background of the mutation and the type or extent of the mutation.To examine these factors more closely in ugt-23, we generated a CRISPR ugt-23 deletion (RB2550) through SUNY Biotech and compared this to the original ugt-23 (PH7346) obtained through the CGC.There was a slight trend toward a higher percentage of death in the RB2550, but it was not statistically significant (P = 0.22, n = 6) (Supporting Information Figure 2C).
Isolation of Glycosylated 1-HP Derivatives.We then explored whether ugt-23 and ugt-49 produced the same 1-HP glycosylated products identified previously in N2. 12 We exposed larval stage 4 (L4) worms in large-scale liquid culture for 24 h, followed by high-performance liquid chromatography−ultraviolet (HPLC−UV) analysis of the worm media.2).(B) Representative UV chromatogram of PD1074 exposed to 22.3 μM 1-HP for 24 h in an S-basal medium with 2% Escherichia coli.Each peak corresponds to either 1-HP or one of its glycosylated derivatives.The peak at 30 min is (1), the peak at 25 min is (2), the peak at 23 min is (3), the peak at 19.8 min is (4), and the peak at 19.4 min is (5).
We found that a 22.3 μM concentration allowed many worms to survive for 24 h to accumulate sufficient modified 1-HP.Using HPLC, we observed four unique peaks in all strains (Figure 2B).The peaks were isolated using semipreparative C-18 reverse phase HPLC and then analyzed by NMR and liquid chromatography−mass spectrometry (LC−MS)/MS (Figure 2A, Supporting Information Figures 3−7).We identified compounds ( 2) and (3), which had also been identified in prior literature (Figure 2A, Supporting Information Figures 3  and 4). 12However, we also identified two branch-chained trisaccharides, one with three glucose moieties (4) and one with two glucose moieties and an N-acetylglucosamine (GlcNAc) (5) (Figure 2A).
Higher-energy collisional dissociation (HCD) fragmentation of these compounds resulted in the generation of largely Btype glycosidic bond cleavages that enable the partial assignment of the molecular structure.MS/MS of compound (4) produced multiple fragment ions that support the modification of 1-HP by three hexose sugar units with the primary fragments annotated in Figure 3A.The sequential neutral loss of three hexose sugars due to glycosidic bond cleavage from the isolated molecular ion was observed, leading to a fragment identified as 1-HP (observed m/z 197.0713).Additional fragment ions corresponding to hexose units are also observed.The MS/MS of compound (5) produced multiple fragment ions, as shown in Figure 3B, which support the modification of 1-HP by two hexose sugars and one Nacetyl hexose sugar.Similar to the fragmentation observed in compound (5), the sequential neutral losses of sugar residues via glycosidic bond cleavage from the modified 1-HP  molecular ion are also observed, with unique ions now present due to the inclusion of the N-acetyl moiety.A table of assigned fragments is provided as Supporting Information Table 3.

Chemical Research in Toxicology
Nuclear Magnetic Resonance Analysis.The 1 H NMR spectrum of compound (4) (Figure 4A) contained signals corresponding to a phenazine and a sugar region with three anomeric protons, including two (H1′ and H1″) that are unusually downfield for β-anomers but consistent with being close to the aromatic phenazine.Analysis of data from 2D total correlation spectroscopy (TOCSY) and correlated spectroscopy (COSY) experiments (Supporting Information Figure 5.1) indicates that the three protons identified in the 1 H spectrum belong to three glucose moieties.The glucosyl attached to the phenazine and the glucosyl linked at C6 have chemical shifts very similar to those of the gentiobiose-phenazine described in Stupp et al. 12 The third glucosyl attached to C2 has very unusual shifts, consistent with the proximity to the phenazine, but the coupling patterns match the glucose configuration.Figure 4B illustrates the linkage position of compound (4), determined using 1D and 2D rotating frame Overhause effect spectroscopy (ROESY) data.The bottom panel shows a region from the TOCSY spectrum with H1′ along the horizontal at 5.77 ppm coupled to H2′−H6′; assignments are annotated in the top panel.This glucose is linked to the phenazine at the position shown in Figure 4A (see also Supporting Information Figure 5.2).The middle panel shows a region from the ROESY spectrum with a nuclear Overhauser effect (NOE) cross peak from H1″ at 5.1 ppm to H2′, establishing the H1″−H2 linkage.Similarly, the H1‴ proton at 4.47 ppm shows cross-peaks linking it to H6′ of the same phenazine-linked glucose.
The 1 H NMR spectrum of the minor compound (5) (Figure 5A) similarly displayed signals corresponding to phenazine and three sugar residues.However, a difference in chemical shift of H2″ (Figure 5B) suggested a β-N-acetyl-2-deoxy-glucosamine residue instead of a glucosyl residue (CASPER database), 24 consistent with the MS/MS data (Figure 3B).A combination of 1D and 2D TOCSY and 1D ROESY data (Supporting Information Figure 6) confirmed that the three residues were in the correct configuration, and the linkages were the same as in compound (4).
Because (5) contained a GlcNAc, we evaluated GlcNAc transferase mutant ogt-1 with the same assays described above.ogt-1 has previously been shown to modulate the immune response in C. elegans for S. aureus infection but not P. aeruginosa infection. 25Consistent with those findings, the ogt-1 knockout had no statistically significant difference in susceptibility to 1-HP compared with N2 and PD1074 (Figure 1B).LC−MS analysis of worm media conditioned by the ogt-1 knockout mutant challenged with 1-HP showed that (5) was still produced (Supporting Information Figure 7).
Quantitation of 1-HP Derivatization in ugt Knockout Mutants.We then quantified the HPLC−UV data to observe if there was a reduction in the amounts of 1-HP derivatives for the ugt-23 and ugt-49 knockout mutants.We normalized the data to the sum of all of the 1-HP-related compounds for each replicate.This ensured that the ratio obtained was independent of any variation due to the amount of 1-HP the worms were exposed to or the number of worms per replicate.We found that the ugt-23 knockout mutant produced decreased amounts of both trisaccharide sugars (4) and ( 5), while the ugt-49 knockout mutant had reduced amounts of compounds (3), (4), and (5) (n = 7) (Figure 6).These results show the involvement of ugt genes in 1-HP detoxification, suggesting that they have broad specificity and that multiple ugt genes are involved in detoxifying a xenobiotic in C. elegans.Prior research has implicated multiple ugt genes responsible for the glycosylation of other small-molecule toxins such as indole. 26The workflow outlined in this study can be used to test the role of ugt genes in modifying other smallmolecule xenobiotics.Future studies could validate whether broad specificity is seen in response to xenobiotics or whether this is a phenomenon specific to 1-HP.
Furthermore, our results implicate the addition of GlcNAc in detoxification in C. elegans, a result which, to our knowledge, was not previously observed.Our data also suggest that genes other than ogt-1 are responsible for adding GlcNAc in the 1-HP glycosylation pathway.This might be due to the broad specificity of ugt genes or because GlcNAc serves a particular purpose in detoxification.Using this workflow, it would be interesting to see if GlcNAc-modified products are also observed for other xenobiotics.
■ METHODS Mortality Assay.All 11 strains of C. elegans were grown and maintained on 10 cm nematode growth medium (NGM) agar plates seeded with a Luria−Bertani (LB)-cultured OP50 strain of E. coli at 22 °C.Knockout mutants were paired with an N2, and PD1074 replicates for each strain.10 cm NGM plates with C. elegans were bleached and grown to L1 arrest, and then L1 arrested worms were transferred to new 10 cm plates and allowed to grow to L4. Upon reaching L4, ∼15 worms were assigned either to control 6 cm plates or 6 cm plates with NGM and 179 μM 1-HP, the LD 50 value of PD1074. 22Worms were incubated on experimental plates for 7 h.After 7 h, fluorescent beads were added to the worms, and the uptake of these beads was used as a marker to differentiate between live and dead worms. 27We note that this method of scoring produces a nonzero background level of reported mortality (e.g., Supporting Information Figure 2) because worms that burrow into agar or crawl off the plate are scored as dead.
Lifespan Timing Assay.Worms were observed to determine how long they took to go from the egg to L4 in two different ways.The time to L4 for N2, PD1074, and the ugt-1, ugt-23, ugt-49, ugt-60, and ugt-62 knockout mutants was measured by initially spot bleaching a single adult and following a single egg, observing them until they reached L4.The time to L4 for the ugt-6, ugt-9, ugt-66, and ogt-1 knockout mutants was measured by bleach synchronization of a plate of worms and placing the resulting eggs on a 10 cm plate.The plates were observed every 4−8 h until most of the population on the plate could reliably be identified as L4.
Large-Scale Growth of C. elegans.Worms were grown on largescale culture plates (LSCPs) to generate worms for subsequent experiments.LSCPs were poured according to previously described protocols. 28LSCPs are hand-washed, sterilized, and wiped with 70% ethanol.The nematode growth medium (NGM) was prepared as a mixture of MYOB, agarose, and bacto-peptone in a ratio of 5.9:10:10 g L −1 .Poured plates were seeded with the HTS115 strain of E. coli prepared in K-media at a concentration of 0.5 g mL −1 bacteria generated according to the IBAT method. 29Worms were chunked onto the LSCPs and then grown for 7−10 days, depending on the strain, before being washed with M9 for subsequent experiments.After washing, worms were bleach synchronized and then grown to L1 arrest in M9.Upon reaching L1 arrest, they were transferred to an Sbasal medium (∼30,000 worms mL −1 ) and incubated with 2% E. coli OP50 until they reached L4.After the worms had reached L4, they were incubated with either 1.1% DMSO or 22.3 μM 1-HP.Worms were incubated for 24 h and then centrifuged.The supernatant was collected for subsequent experiments.
Glucoside Collection and Analysis (HPLC−UV).After the supernatant was separated from the worms, it was centrifuged again at 20,800 relative centrifugal force (RCF) for 10 min to separate the bacteria from the supernatant.The resultant volume was lyophilized and extracted in an appropriate volume of methanol (200 and 600 μL, depending on the starting volume of the supernatant).It was then centrifuged at 20,800 RCF for 30 min.Following centrifugation, the supernatant was concentrated to ∼100 μL with 90 μL injected into HPLC−UV and 10 μL separated for LC−MS.
The supernatant was analyzed on an Agilent 1200 Series HPLC system with a diode array collector, and fractions were collected manually upon observation of a peak.Absorbance was measured at 260 nm.For worm media separation, 5% methanol (A) and 95% 5 mM phosphate buffer pH 7.2 (B) were held isocratic for 4 min, increasing to 95% A and 5% B over 30 min, and then held for 5 min, followed by a re-equilibration of the column.The separation was carried out at a flow rate of 2 mL min −1 in an Agilent SB C-18 column (9.4 mm × 250 mm, 5 μM).
After initial fractionation, further separation of the fraction containing compounds 4 and 5 was carried out.For that separation, 5% methanol (A) and 95% five mM phosphate buffer pH 7.2 (B) were held isocratic for 4 min, increasing to 50% A and 50% B over 17 min.The gradient was slowed, and the ratio was increased to 67% A and 33% B by 28 min before ramping it up to 95% A and 5% B by 30 min and then holding constant for 5 min.A re-equilibration of the column followed this.The column used for this separation was the same as that for the initial worm media separation.
Glucoside Analysis (LC−MS/MS).Samples aliquoted during glucoside collection were analyzed using a Thermo Fisher Scientific Q Exactive HF Orbitrap mass spectrometer coupled to a Vanquish UPLC instrument with inline UV detection.Chromatographic separation was performed with an Agilent ZORBAX Eclipse XDB-C-18 column (2.1 mm × 150 mm, 1.8 μm) over 30 min, starting with 95% H 2 O (A) and 5% methanol (B) held isocratic for 2.5 min, then increased to 70% B by 22 min and 100% B at 22.5 min, and held for 4 min before re-equilibration at 5% B for 3 min prior the next injection.The sample queue was randomized with injection blanks included to monitor for sample carryover.All samples were analyzed by positive mode electrospray ionization (ESI).Full MS scans were performed at a specified resolution of 30,000 (m/z 200) from 150 to 2000 m/z with an AGC target of 3e6 and a maximum IT of 200 ms.Corresponding UV traces were collected at 260 nm.
Target compounds were isolated with a 4.0 m/z quadrupole window to perform structural elucidation by higher-energy collisional dissociation (HCD).A normalized collision energy (NCE) of 15 V was applied, and fragment ions were detected with a specified resolution of 15,000, AGC target of 2e5, and a maximum of IT 100 ms.MS data were analyzed with the Thermo Qual Browser and manually interpreted.
Glucoside Analysis (NMR).Pooled fractions were dried, resuspended in 60 μL of D 2 O with 0.15 mM DSS as an internal standard, dried with a speed vac, and resuspended twice in 60 μL of D 2 O to perform buffer exchange to remove excess H 2 O before being transferred into 1.7 mm NMR tubes.1D 1 H, 2D COSY, 2D TOCSY, selective 1D TOCSY, and selective 1D ROESY spectra were collected where appropriate on a Bruker 800 MHz NEO spectrometer using a 1.7 mm cryoprobe.Spectra were processed and analyzed with MestReNova 14.1.2(Mestrelab Research).
Statistical Analysis.Analysis was performed using JMP, a publicly available statistical software.A Wilcoxon test, followed by a Wilcoxon pairwise analysis and a Benjamini−Hochberg correction, was performed for the mortality assays to determine the significance between strains.Tukey's HSD test was performed to determine the significance within each strain for 1-HP exposure.A Wilcoxon test followed by Wilcoxon pairwise analysis was performed on the scaled absorbance data.

Figure 1 .
Figure 1.Plate-based screen for the susceptibility to 1-HP.(A) Schematic describing the method for the plate-based assay.Worms were incubated for 7 h with at least six replicates per strain.(B) Box and whisker plot with quartiles showing mortality of various strains at 179 μM 1-HP.Horizontal lines with * indicate significantly (P < 0.1) increased mortality compared to N2 with an α of 0.1 after a Wilcoxon pairwise comparison, followed by a Benjamini−Hochberg Correction.The number of replicates (n) for each strain is provided below each plot.
Molecular compositions were determined by accurate mass measurement, and tentative molecular structures were established using tandem mass spectrometry (MS/MS) (Figure 3A,B).The observed m/z of compound (4) was 683.2305 (theoretical value 683.2294, mass measurement error = 1.61 ppm) (Figure 2A, Supporting Information Table 2).The molecular formula was established as C 30 H 38 N 2 O 16 .The observed m/z of compound (5) was 724.255 [M + H + ]

Figure 3 .
Figure 3. Tandem mass spectrometry (MS/MS) data for compounds (4) and (5).Tentative structural insets are based on NMR results discussed in detail below.(A) MS/MS Spectra for compound (4) with fragment ions corresponding to the loss of three hexose sugars.(B) MS/MS Spectra for compound (5) with fragment ions corresponding to the loss of two hexose and an N-acetyl hexose sugar.

Figure 4 .
Figure 4. NMR spectra for trisaccharide compound (4): (A) Structure of compound (4) and 1D 1 H spectrum with β-glucosyl anomeric protons annotated.(B) Top panel shows a 1D 1 H of the glucosyl residue attached to the phenazine.The bottom panel is a region of a 2D TOCSY showing the protons coupled to H1′.The middle panel is a region of a 2D ROESY showing NOEs between H1″ and H1‴ and the respective protons in the linkage positions.

Figure 5 .
Figure 5. 1 H NMR spectrum supporting the N-acetyl-glucosamine containing compound (5): (A) Structure of compound 5. (B) Middle panel shows the 1D proton of 5 and the three β-anomeric protons like those in compound (4) (top panel).The bottom panel is a region from a 2D COSY of the mixture of compound (4) (primary compound) and compound (5) (minor compound), indicating the chemical shift positions of H2″.

Figure 6 .
Figure 6.Box and whisker plot showing the relative amounts of 1-HP and its derivatives after a 24 h incubation at 22.3 μM 1-HP with 2% E. coli based on UV absorbance data.All replicates (n = 7) were paired, and data were normalized by dividing the absorbance for each compound at 260 nm by the sum of the absorbances of 1-HP and all its derivatives for each run (abs x/[abs z + abs y + abs x + abs w + abs v]).* Indicates a significant difference in relative amounts of compound compared to the relative amount of the same compound in PD1074 after Wilcoxon pairwise analysis (α = 0.05).(A) Relative amounts of glycosylated 1-HP derivatives for the ugt-23 mutant compared to PD1074.Compounds 4 and 5 are reduced in this strain.(B) Relative amounts of glycosylated 1-HP derivatives for the ugt-49 mutant compared to PD1074.Compounds 3, 4, and 5 are reduced in this strain.

Table 1 ,
SupportingInformation Table1), we found that all of the strains except ugt-23 and ugt-49 had a mortality percentage between 35 and 54% compared to controls (P < 0.01) (Supporting Information

Table 1 .
Information on Strains Used in This Study a Ref 22. C. elegans Gene Knockout Project at the Oklahoma Medical Research Foundation.e CRISPR/Cas9 Methodology for the Generation of Knockout Deletions in C. elegans.f C. elegans Reverse Genetics Core Facility at the University of British Columbia, International C. elegans Gene Knockout Consortium.
g SUNY Biotech.hNote:ogt-1 is included in this table as this strain was used for toxicity experiments as well based on results described in Figure2.