Metabolism of Skin-Absorbed Resveratrol into Its Glucuronized Form in Mouse Skin

Resveratrol (RESV) is a plant polyphenol, which is thought to have beneficial metabolic effects in laboratory animals as well as in humans. Following oral administration, RESV is immediately catabolized, resulting in low bioavailability. This study compared RESV metabolites and their tissue distribution after oral uptake and skin absorption. Metabolomic analysis of various mouse tissues revealed that RESV can be absorbed and metabolized through skin. We detected sulfated and glucuronidated RESV metabolites, as well as dihydroresveratrol. These metabolites are thought to have lower pharmacological activity than RESV. Similar quantities of most RESV metabolites were observed 4 h after oral or skin administration, except that glucuronidated RESV metabolites were more abundant in skin after topical RESV application than after oral administration. This result is consistent with our finding of glucuronidated RESV metabolites in cultured skin cells. RESV applied to mouse ears significantly suppressed inflammation in the TPA inflammation model. The skin absorption route could be a complementary, potent way to achieve therapeutic effects with RESV.

Introduction avoid being bed-ridden. Information from basic research on ageing can help to mitigate this problem.
Various models of ageing have been proposed in the last century, including the radical theory of ageing [6], replicative senescence [7], and others [8]. One of the most widely accepted is a calorie restriction model. Since the initial discoveries by McCay and colleagues [9], calorie restriction is now well known to extend the lifespan of many species, including rats, fish, fruit flies, nematodes among others [10].
The calorie-restricted condition affects several key metabolites. Elevated respiration during calorie restriction is a mode of energy production that creates nicotinamide adenine dinucleotide (NAD) as a by-product and lowers levels of the reduced form, NADH [11]. NAD accumulation in a calorie-restricted condition activates NAD-dependent deacetylases, Sir2 and its mammalian homolog, Sirt1 [12], which are involved in age-related diseases [13]. Calorie restriction also turns on a gene called PNC1, which produces an enzyme that rids cells of nicotinamide, a small molecule similar to vitamin B3 that normally represses Sir2 [14]. Moreover, low-glucose or calorie-restricted conditions increase the AMP/ATP ratio, followed by activation of adenosine monophosphate-activated protein kinase (AMPK), a critical energy sensor that maintains cellular energy balance and regulates metabolic functions [15].
Thus the significance of calorie restriction in experimental longevity is well established, while its clinical application is more controversial. First, there is little evidence to suggest that calorie restriction could extend life spans of humans or primates [16]. Second, calorie restriction might improve some metabolic diseases (diabetes, obesity, etc), while it might lead to lower body mass index, which exacerbates sarcopenia, frailty, and other consequences of ageing [17]. Thus calorie restriction alone might not be a viable approach to increase human longevity or to prevent people from becoming bed-ridden. As an alternative, several substances are of interest to researchers to mimic the calorie-restricted state [18]. Among them, resveratrol (RESV) is a plant polyphenolic compound, present in grapes, berries and peanuts, that acts as an anti-oxidant and as an activator of SIRT1 and AMPK [19]. RESV improves cardiovascular health [20,21], diabetes [22], inflammation, and cancer [23] and increases longevity in laboratory animals [24].
However, there have been few randomized clinical trials to examine health benefits of RESV and results have not been strongly conclusive [25,26]. Moderate pharmacological effects of RESV in humans may result partly from its low bioavailability because of its extensive catabolism to sulfated, glucuronideconjugated, or hydrogenated forms in liver and gut [27][28][29][30][31]. While most of studies of RESV treatment employed oral administration or gastric injection, few studies used alternative administration routes (e.g. skin absorption) [32]. Other routes should be considered, especially for bed-ridden patients, for several reasons. First, oral administration of any drug can be quite difficult due to dysphagia, a frequent symptom. Second, direct delivery via topical application might be more effective than oral treatment in preventing sarcopenia or frailty, which often cause patients to become bed-ridden. Moreover, topically applied RESV might be less affected by liver and gut metabolism. However, no detailed comparison of RESV metabolism and tissue distribution has been undertaken to compare oral administration and dermal absorption of RESV.
Here we report the comparison of RESV metabolites in mice administered orally and transdermally. We investigated RESV metabolites in cultured cells from various tissues. These data suggest overlapping, but distinct, tissue-dependent, metabolic profiles of RESV. . Hydrophilic ointment was prepared as described in Japanese pharmacopeia [33]. Carboxymethylcellulose (CMC) gel was prepared by mixing 20 g sodium carboxymethylcellulose salt, 100 mL glycerin, 380 mL water. Chemicals used in LC-MS analysis were HPLC or MS grade and were obtained from Wako Pure Chemical Industries.

Animals and diets
Male 7-8-week HR-1 hairless mice (Japan SLC) and female 8-week C57BL/6 mice (Charles River Laboratories Japan) were kept on a 12 h light-dark cycle and given free access to food (F-2) and tap water. Animal experiments were approved by the Animal Research Committee, Graduate School of Medicine, Kyoto University, Japan.
Mouse oral and skin resveratrol administration RESV was dissolved in ethanol (50 mg/mL) and stored at 220˚C. For oral administration, 1 mg RESV in ethanol was diluted in saline (total volume 200 mL). RESV was dosed about 40 mg/kg body weight, similar to the dosage used in human experiments [34]. For skin treatment, 1 mg RESV in ethanol was directly applied on the dorsal skin or mixed with 230 mg of base (0.4% RESV) and then swabbed onto the dorsal skin (around 2 cm 2 ) without using anesthesia. Mice were separated from each other and could not lick RESV from their dorsal skin.

Resveratrol treatment in cell culture for metabolomic analysis
HepG2 (human hepatoblastoma cells) [35], HaCaT (human keratinocytes) [36] and C2C12 (mouse myoblasts) [37] were used to examine RESV metabolism. Cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) with 1% penicillin/streptomycin and 2 mM L-glutamine with some modifications as follows; DMEM (1.0 g/L glucose) with 10% fetal calf serum (FCS) for HepG2; DMEM (4.5 g/L glucose) with 10% FCS for HaCaT; DMEM (4.5 g/L glucose) with 20% FCS for C2C12. C2C12 cells were differentiated into myotubes by switching medium into DMEM containing 2% horse serum. Cells were plated on a 10 cm dish and allowed to adhere for 18 h. Cells were treated with 20 or 200 mM RESV. After 4 h incubation, cells were washed with PBS twice, followed by addition of 2 mL 50% ice-cold methanol. Collected cells were transferred into tubes on ice and further processed as described in metabolome sample preparation section.

Mouse tissue sample preparation for metabolomic analysis
Mice were anesthetized with pentobarbital and blood was collected from the right ventricle. For metabolite extraction 0.2 mL of blood were mixed with 1.8 mL 55% methanol, pre-chilled to 240˚C and kept until further processing as described in the metabolome sample preparation section. In addition to blood, metabolome samples were prepared from liver, dorsal skin, leg skin, and hind limb muscle.
Around 100 mg of each mouse tissue was dissected and weighed. Tissues were minced and homogenized for 3 min in 500 mL 50% methanol on ice using a BioMasher II (Nippi, Tokyo, Japan). Next, 1 mL 50% methanol and internal standards (10 nmol 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and piperazine-N,N'-bis(2-ethanesulfonic acid) (PIPES)) were added to the samples. HEPES and PIPES were used as internal standards to correct for the variation of sample preparation and different detection efficiency. After brief vortexing, samples were centrifuged at 500 g for 10 min at 4˚C. Supernatant was recovered for further processing as described in the metabolome sample preparation section (without the repeated addition of internal standards and vortexing).

Metabolome sample preparation
Metabolome sample preparation was previously described [38]. After addition of internal standards (10 nmol HEPES and PIPES) to metabolome samples and brief vortexing, samples were transferred to Amicon Ultra 10 kDa cut-off filters (Millipore, Billerica, MA, USA) to remove proteins. Following sample concentration by vacuum evaporation, each sample was resuspended in 40 mL 50% acetonitrile and 1 mL was used for liquid chromatography-mass spectrometry (LC-MS) analysis.

LC-MS analysis
LC-MS data were obtained using a Paradigm MS4 HPLC system (Michrom Bioresources, Auburn, CA, USA) coupled to an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA), as described previously [38]. Briefly, metabolites were separated on a ZIC-pHILIC column (Merck SeQuant, Sweden; 15062.1 mm, 5 mm particle size) using acetonitrile (A) and 10 mM ammonium carbonate buffer, pH 9.3 (B) as the mobile phase, with gradient elution from 80% A to 20% A in 30 min, at a flow rate of 100 mL min 21 . Peak areas of metabolites of interest were measured using the MZmine 2.10 software [39] and normalized using the peak areas of HEPES and PIPES, as described in S2 Table. Metabolite identification and data analysis Peaks were assigned to metabolites based upon their retention times and m/z values. Compounds were confirmed with standards (STD) or with analysis of MS/ MS spectra (MS/MS), if no standard was available. In cases where no standard compound or MS/MS spectrum was available, compounds were identified by generating chemical formulae using m/z values in MZmine 2.10 module [40] and given MS status. Data were analyzed using the scatter plot function built into MZmine 2.10 software [39], as well as by exporting the data to a spreadsheet.

TPA-induced ear edema
The ear edema model was employed as previously described [41]. Briefly, 1 mg 12-O-Tetradecanoylphorbol-13-acetate (TPA) was applied to the right ear of female C57BL/6 mice. Edema was evaluated by measuring ear thickness and ear weight. Thirty minutes after TPA treatment, 1 mg RESV was applied to the inner and outer surfaces of the right ear, or was supplied orally. Ear thickness was measured with a micrometer before and 24 h after TPA treatment. After sacrificing mice with CO 2 , ear punch biopsies (6 mm diameter) were weighed.

Statistical Analysis
Data are presented as means ¡ standard deviations (SD) of three or four samples. Statistical analysis was performed using an unpaired Student's t-test, or one-way ANOVA followed by Dunnett's test or Tukey's test for comparison of multiple treatment groups. Statistical significance was established at P,0.05.

Resveratrol metabolites detected after resveratrol treatments
In a series of mouse tissues and cell culture experiments we were able to detect RESV and various RESV metabolites from RESV-treated samples (Table 1, S1 Table, S1-S5 Figures). We confirmed the presence of RESV-SULF (307.027 m/z), a sulfate-conjugated RESV, in RESV-treated mice using a standard and MS/MS analysis (S1 Figure). Similarly, we confirmed the presence of dihydroresveratrolsulfate (DH-RESV-SULF) by MS/MS analysis (S2 Figure). DH-RESV is a metabolite of RESV, formed by hydrogenation of the double bond of RESV by intestinal bacteria [28]. Retention times of peaks of free RESV and DH-RESV coincided with those of sulfated forms, so that quantification of free RESV and DH-RESV was obstructed in samples containing sulfated metabolites (S1 Figure,  [42]. These results suggest that isomerization of RESV occurs in mouse or cultured cells either before or after glucuronidation. In addition, using MS/MS or MS analysis we identified further metabolites of RESV (RESV-sulfoglucuronides, RESV-disulfate, RESV-diglucuronide) and DH-RESV (DH-RESV-glucuronide, DH-RESV-sulfoglucuronides) (S4-S5 Figures). Our mass spectrometry data for RESV and DH-RESV metabolites correspond to MS/MS data in previous studies (S1 Table).

Distribution of resveratrol and its metabolites in the mouse tissues
To analyze the distribution of RESV and its metabolites in mouse tissues, we administered 1 mg trans-RESV to mice via oral and dermal routes and collected tissues 4 h later for LC-MS analysis. In this experiment, we identified 132 distinct physiological metabolites (123 verified with standards) in mouse tissues (blood, liver, muscle, leg skin, dorsal skin) (data deposited in the MetaboLights repository, as described in the Supplemental materials section). Metabolites included nucleotides, amino acids, sugar-phosphates, vitamins, organic acids, etc. The majority of these identified metabolites (103/132578%) were detectable in both negative and positive ionization modes. Significant metabolic changes after RESV-treatment both via oral and skin routes were not observed. RESV-SULF and DH-RESV-SULF were detected mainly in liver and blood after both, oral-and skin-administration of RESV. RESV-SULF (Fig. 1A) and DH-RESV-SULF (Fig. 1D), were barely detectable in dorsal and leg skin, and muscle, regardless of RESV administration route. DH-RESV-GLUC (Fig. 1E) was also detected in blood and liver as well as at lower levels in skin and muscle after oral and skin administration of RESV. Thus RESV-SULF, DH-RESV-SULF and DH-RESV-GLUC showed comparable tissue distribution profiles regardless of administration route. These data indicate that RESV is absorbed orally and transdermally with similar efficiency and that its metabolites are detectable in a tissue-specific manner.
We observed strikingly distinct profiles for trans-RESV-3-O-GLUC and cis-RESV-3-O-GLUC depending upon administration route ( Fig. 1B and 1C). After oral administration, trans-RESV-3-O-GLUC was detected mainly in blood and liver, however, after dermal application of RESV, we observed higher levels of trans-RESV-3-O-GLUC in dorsal skin, as well as in blood and liver. Interestingly, after dermal application of RESV, higher levels of cis-RESV-3-O-GLUC were detected in dorsal skin compared to other tissues. In contrast, after oral administration more or less equal levels of cis-RESV-3-O-GLUC were found in blood, liver, dorsal skin and leg skin. We noticed that in the case of transdermal administration, trans-RESV-3-O-GLUC and cis-RESV-3-O-GLUC were detected at higher levels in dorsal skin where RESV was applied, but at lower levels in leg skin, distant from dorsal skin. These data suggest that skin cells may be able to directly metabolize RESV into RESV glucuronides, which has not been reported previously.

Resveratrol metabolism in cell culture
While the tissue distribution of most RESV metabolites after oral and skin administration of RESV is quite similar, distinct profiles of trans-RESV-3-O-GLUC and cis-RESV-3-O-GLUC in skin indicate that tissues other than liver and gut, (e.g. skin) may also be involved in RESV metabolism. To assess whether RESV metabolism is cell type-dependent, we analyzed RESV metabolism in liver, skin, and muscle cell cultures. We treated HepG2 (human hepatocyte), HaCaT First, RESV-SULF was detected in HepG2 hepatocytes as previously reported [43], but not in the other two cell lines ( Fig. 2A). Next, we could not detect DH-RESV and its metabolites in our tested cell cultures (data not shown), suggesting that mammalian cells cannot metabolize RESV into DH-RESV.

Resveratrol absorption efficiency through skin using different bases
Although we observed that RESV was absorbed through skin using ethanol as the solvent (Fig. 1), this would not be practical for clinical applications. We evaluated 3 different bases for efficiency of RESV skin absorption in mouse tissues. As RESV and its metabolites were barely detectable in muscle in our preliminary study, using bases (data not shown), we measured RESV metabolites in blood, liver and dorsal skin 4 h after transdermal treatment (Fig. 3). Bases used in this experiment were hydrophilic ointment, macrogol gel, and CMC gel. Hydrophilic ointment is a petrolatum-modified base. Macrogol is a mixture of polyethylene glycol 400 and 4000. CMC gel is mix of carboxymethylcellulose sodium, glycerol, and water. Macrogol and CMC gel are water-soluble bases. RESV dissolved in ethanol was mixed with these bases. Bases with ethanol alone were used as the control. Tryptophan levels of tissue samples are provided as a control (Fig. 3F).
First, using RESV dissolved in ethanol, RESV metabolites were robustly detected ( Figs. 1 and 3). Samples treated with RESV mixed with each base showed similar, but distinct levels of RESV metabolites (Fig. 3).
In blood samples (Fig. 3, left column), the highest levels of RESV-SULF (Fig. 3A), trans-RESV-3-O-GLUC (Fig. 3B), and cis-RESV-3-O-GLUC (Fig. 3C) were detected in mice treated with RESV dissolved in ethanol. Blood samples after RESV treatment with hydrophilic ointment showed the smallest peaks for all RESV metabolites compared to the other 2 bases and RESV dissolved in ethanol. In liver samples (Fig. 3, middle column) RESV and DH-RESV metabolite peaks from RESV in hydrophilic ointment were also the smallest compared to other bases.
In dorsal skin samples (Fig. 3, right columns), RESV-SULF, trans-RESV-3-O-GLUC, and cis-RESV-3-O-GLUC were found most abundantly when RESV was dissolved in ethanol and less abundantly when RESV was mixed with macrogol. DH-RESV-SULF and DH-RESV-GLUC were detected most abundantly in RESV mixed with macrogol samples among the 3 tested bases.
Collectively, these results suggest that among the bases used in this study, water soluble bases, macrogol or CMC gel, exhibited efficiencies of transdermal RESV absorption and penetration, comparable to that of RESV dissolved in ethanol.
Using human blood metabolome analysis methods described previously [44], we evaluated RESV metabolites in blood from a couple of human volunteers after RESV oral or skin administration using different bases. RESV metabolites in human blood could be detected after oral, but not skin administration (data not shown).

Effect of RESV after application to skin on ear edema model
To evaluate beneficial effects of percutaneous administration of RESV, we used a mouse model of acute skin inflammation after TPA treatment. Anti-inflammatory activity of RESV after oral or skin administration in TPA-induced ear edema was measured as the difference in ear thickness and weight between the two groups. Before TPA treatment ear thickness of mice was 0.22 mm in each group (Fig. 4A). Ear thickness increased to 0.30 mm within 24 h after topical TPA treatment. RESV dermal application showed a significant decrease in ear thickness (0.21 mm) compared to the control group, as observed in a previous study [41]. In contrast, oral RESV administration did not show a significant change of ear thickness (0.26 mm). Ear weight was significantly lower in the transdermal RESV group (6.3 mg) compared to TPA-treated mice (9.4 mg), whereas oral RESV treatment had less effect (8.1 mg) (Fig. 4B). These data suggest that skin treatment with RESV inhibited TPA-induced ear edema, but that oral RESV treatment did not.

Discussion
We analyzed the metabolism of RESV after oral or dermal administration. Metabolomics is a recently developed tool to detect and semi-quantify small metabolites [45,46]. Indeed, our metabolome data in mice include various RESV metabolites, such as RESV-SULF, trans-RESV-3-O-GLUC, cis-RESV-3-O-GLUC, DH-RESV, etc., some of which have been previously described in mouse and mammalian studies [31,47,48].
We determined that RESV can be efficiently absorbed through skin in mice. One of the issues regarding human treatment with RESV is effective delivery of RESV. Numerous studies have examined oral administration in mice [49], rats [50], pigs [47], and humans [25]. Little is known about metabolism of transdermally-absorbed RESV, although some reports have described the existence of glucuronosyltransferases (UGTs) in the skin [51], dermal absorption of RESV in vitro [32,52], and its antioxidant activity in mouse [53,54] and human skin [55].
In this study we analyzed and compared the tissue distribution of RESV metabolites between oral and dermal administration. Using this approach, we identified hydrophilic bases that may be suitable for efficient transdermal absorption of RESV in mice, with possible future clinical applications. Since skin permeability of chemicals is different between humans and mice [56], absorption efficiency of RESV in human skin is not clear. Expression patterns of transporter proteins and kidney excretion of xenobiotics are also different between them [57]. Also in our small human trial we could not detect RESV and its metabolites in blood after dermal application. These results illustrate the need to explore human applications.
Unexpectedly, we noticed that distribution profiles of RESV-SULF, trans-RESV-3-O-GLUC, cis-RESV-3-O-GLUC, and even DH-RESV after dermal application are quite similar to those observed after oral administration. Our metabolomic approach clearly detected DH-RESV metabolites in various tissues of mice, but not in three cell culture lines tested. These data imply that dermally absorbed RESV might also be metabolized into DH-RESV by gut bacteria (Fig. 5). After oral RESV administration, part of RESV is absorbed and metabolized by the liver into sulfated and glucuronidated forms. Unabsorbed RESV can be metabolized by gut bacteria into DH-RESV (Fig. 5 arrow 1), which then can be absorbed and transformed into DH-RESV-SULF and DH-RESV-GLUC by the liver (Fig. 5 dashed arrow 2). Enterohepatic circulation can enhance levels of RESV, DH-RESV and their metabolites. After topical application of RESV, RESV is absorbed and transformed into RESV glucuronides by skin cells (Fig. 5 arrow 3), followed by entry into the circulatory system. In liver, RESV can be modified into RESV-SULF and RESV glucuronides. Once excreted into the gut, the double bond in RESV and its metabolites can be reduced by gut bacteria, resulting in DH-RESV and its metabolites (Fig. 5 arrow 4). Enterohepatic circulation should be crucial for the absorption of DH-RESV and its metabolites into the body (Fig. 5  arrow 5).
Our comparison of oral and dermal administration of RESV in vivo identified distinct profiles of RESV metabolite distribution; the glucuronidation of RESV in skin after dermal application, but not its sulfation. These data are consistent with our observations in vitro. RESV glucuronides, but not RESV-SULF, are clearly detected in cultured skin cells, while both forms are found in cultured liver cells. These findings suggest that skin itself metabolizes RESV into the glucuronidated, but not the sulfated forms. Sulfation and glucuronidation of RESV are catalyzed by sulfotransferase SULT1 [58] and UDP-glucuronosyltransferases [59], respectively. It is also well known that RESV-SULF is generated by various cell types, not only liver, but also in breast [60] and brain cancer cells [61], while data concerning cell types producing RESV glucuronides are still scarce [62]. This is the first report of RESV glucuronide formation by skin cells.
While rapid absorption and metabolism of RESV after oral dosing was previously reported [63], permeability and metabolism of RESV through skin were largely unknown. In addition, topical application of RESV on mouse ears actually inhibited ear edema, whereas oral administration showed a smaller effect. This result suggests that skin application of RESV is effective to prevent acute inflammation in skin.
Glucuronidation of xenobiotic molecules is generally thought to serve in detoxification [64]. On the other hand, mammalian tissues contain glucuronidases that hydrolyze glucuronic acid from substrates [65]. Tissue glucuronidases might regenerate intact RESV from RESV glucuronides. In this way, RESV glucuronides might serve as a RESV reserve in the body. A similar situation might be expected in the case of RESV sulfates, which could be hydrolyzed to RESV by tissue sulfatases [30].
In conclusion, our comparative analysis in vivo and in vitro between oral and skin administration will be valuable for future clinical applications of RESV.
Supporting Information S1 Figure Table); possible chemical structures are shown. In mouse and cell tissue samples we detect single, early eluting peaks for the 227.071 m/z (RESV) and 307.027 m/z (RESV-SULF). Since we cannot exclude the possibility that cis-and trans-forms, as well as resveratrol-sulfates sulfated at different positions are coeluting, we designate those peaks as resveratrol (RESV) and resveratrol-sulfate (RESV-SULF). MS/MS fragmentation patterns of mouse liver samples match that of trans-Resveratrol-3-O-sulfate, the major fragment being 227.071 m/z (RESV). As shown in (A) and (B), retention times of trans-Resveratrol and trans-Resveratrol-3-O-sulfate are almost identical. Furthermore, injection of 0.5 pmol of trans-Resveratrol or trans-Resveratrol-3-O-sulfate results in similar signal intensity for the 227.071 m/z ion. In case of trans-Resveratrol-3-O-sulfate injection, 227.071 m/z ion is detected at much lower intensity than the 307.027 m/z ion. This complicates quantification of small resveratrol amounts, which could be present in the samples. RESV and RESV-SULF peak areas correlate, except for dorsal skin where RESV was applied (S6 A Figure).  Table). Possible chemical structures are shown. In mouse samples we detect a single, early eluting peak for at 309.044 m/z (DH-RESV-SULF). Since we cannot exclude the possibility that DH-RESV sulfates, sulfated at different positions, are coeluting, thus we designate the peak as dihydroresveratrol-sulfate (DH-RESV-SULF). Similarly to spectra for RESV and RESV-SULF (S1 Figure), the MS/MS fragmentation pattern of 309.044 m/z peak (DH-RESV-SULF) in mouse liver shows the major fragment at 229.087 m/z (DH-RESV). As shown in (A) and (B), retention times of DH-RESV and DH-RESV-SULF are almost identical. Furthermore, as in RESV and RESV-SULF (S1 Figure), the 229.087 m/z ion is detected at much lower intensity than the 309.044 m/z ion. This complicates quantification of traces of DH-RESV that could be present in the samples. DH-RESV and DH-RESV-SULF peak areas correlate in all tissues (S6 B Figure).  Figure. Comparison of the RESV and DH-RESV normalized peak areas with their sulfated counterparts. (A) RESV and RESV-SULF normalized peak areas correlate in RESV-treated mouse tissues, except for dorsal skin, where RESV was applied directly. This is most likely due to the RESV-SULF fragmentation pattern (S1 Figure). (B) DH-RESV and DH-RESV-SULF normalized peak areas correlate in RESV-treated mouse tissues. This is most likely due to the RESV-SULF fragmentation pattern (S2 Figure). doi:10.1371/journal.pone.0115359.s006 (TIF) S1