Carbonic anhydrase III (Car3) is not required for fatty acid synthesis and does not protect against high-fat diet induced obesity in mice

Carbonic anhydrases are a family of enzymes that catalyze the reversible condensation of water and carbon dioxide to carbonic acid, which spontaneously dissociates to bicarbonate. Carbonic anhydrase III (Car3) is nutritionally regulated at both the mRNA and protein level. It is highly enriched in tissues that synthesize and/or store fat: liver, white adipose tissue, brown adipose tissue, and skeletal muscle. Previous characterization of Car3 knockout mice focused on mice fed standard diets, not high-fat diets that significantly alter the tissues that highly express Car3. We observed lower protein levels of Car3 in high-fat diet fed mice treated with niclosamide, a drug published to improve fatty liver symptoms in mice. However, it is unknown if Car3 is simply a biomarker reflecting lipid accumulation or whether it has a functional role in regulating lipid metabolism. We focused our in vitro studies toward metabolic pathways that require bicarbonate. To further determine the role of Car3 in metabolism, we measured de novo fatty acid synthesis with in vitro radiolabeled experiments and examined metabolic biomarkers in Car3 knockout and wild type mice fed high-fat diet. Specifically, we analyzed body weight, body composition, metabolic rate, insulin resistance, serum and tissue triglycerides. Our results indicate that Car3 is not required for de novo lipogenesis, and Car3 knockout mice fed high-fat diet do not have significant differences in responses to various diets to wild type mice.


Introduction
Carbonic anhydrases (CAs) are isozymes that catalyze the carboxylation of water into carbonic acid, which spontaneously dissociates into bicarbonate and protons (H 2 O + CO 2 H 2 CO 3 ! HCO 3 -+ H + ) [1,2]. There are 16 unique isozymes in the vertebrate carbonic anhydrase gene family, or the α-CA family [3]. Carbonic anhydrases are involved in a wide variety of To determine whether Car3 has a functional role in lipid content we characterized the previously described Car3 -/-mice [49]. No obvious deleterious phenotypes have been reported in previous studies on Car3 knockout mice; however, the effect of specialized diets on these knockout mice was not previously researched [19,20,22,49,50]. In this study, we investigate Car3 function in de novo lipogenesis by measuring fatty acid synthesis in isolated mouse hepatocytes, and during induced obesity through high-fat diet by measuring subsequent changes in body weight, dyslipidemia, insulin resistance, food intake, and energy expenditure.

2D-DIGE protocol and western blot confirmation
Animal care. 20-week old diet-induced obesity (DIO) mice were purchased from Jackson Labs, with mice starting on the high-fat diet (D12492, Research Diets) at 6 weeks. At 20 weeks of age, mice were treated with 10mg/kg niclosamide or DMSO vehicle control for 12 days by daily intraperitoneal (IP) injections. Mice were harvested to obtain liver tissue. Mice were maintained at the University of North Carolina (UNC) under protocols specifically approved for this study by the UNC Institutional Animal Care and Use Committee (IACUC). Amino acid sequence alignment of cytosolic carbonic anhydrases reveals key enzyme amino acids changes that affect activity. CA III (Car3) amino acid alignment with carbonic anhydrase family members Car1 and Car2. Car3 shares a 57%/56% identity and 74%/75% similarity with Car2 and Car1 respectively. Car3 has a 69% identity with either Car2 or Car1. Amino acids highlighted in yellow are identical between the three CAs. Amino acids highlighted in orange are shared between Car3 and either Car1 or Car2. The active site of all carbonic anhydrases contain a Zn 2+ and three histidine residues (highlighted in blue) [12]. Two key amino acid differences (highlighted in red), at Lys64 and Phe198, are responsible for the substantially lower rate of regeneration of the active form for carbonic anhydrase 3 compared to the other two CAs [16,17]. A lysine at amino acid 64 results in a less effective proton shuttle, [13,15] and Phe198 causes a steric hindrance at the active site [14]. https://doi.org/10.1371/journal.pone.0176502.g001

2D-Differential gel electrophoresis (2D-DIGE) of niclosamide-treated mouse liver.
Liver lysates were prepared, and 2D-DIGE experiments performed by the University of North Carolina Systems-Proteomics Center core facility as previously described [51,52]. Protein spots were identified using peptide mass fingerprinting tandem mass spectrometry data by the Yale Mass Spectometry and Proteomics Resource Core (New Haven, CT), according to previously described methods [51][52][53].
Western blotting of mouse liver lysates. 20 week old DIO high-fat diet fed mice were treated for 12 days with either 5 mg/kg, 25 mg/kg niclosamide, or DMSO (vehicle) by daily IP injections (40 μl). On the last day of treatment, mice were sacrificed and livers harvested and frozen on dry ice. For western blot analysis, liver samples were homogenized and sonicated in 10 volumes of RIA lysis buffer (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 1 mM PMSF, 1 mM beta-glycerolphosphate, 2.5 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 1X Sigma-Aldrich P8340 protease inhibitor cocktail). Homogenates were clarified by centrifugation at 16,000xg for 10 minutes at 4˚C, and protein in the supernatant was quantified using the Bio-Rad Dc Assay (Hercules, CA). Protein samples in loading buffer (Life Technologies NP0007) were loaded at 50 μg per well on 4-12% NuPAGE gels and separated for 45 minutes at 200V, then transferred to Immobilon-FL blotting membrane at 300 mA for 1 hour. Blots were blocked with 5% milk in TBS, then probed for carbonic anhydrase III (Santa Cruz Biotechnology, sc-50715) and subsequently actin (MP Biomedicals, #69100) to control for total protein loaded. Secondary antibodies were (IRDye infrared antibodies, LI-COR Biosciences, Lincoln, NE) [51]. The Odyssey Infrared Imaging System was used for quantification of the relative intensity ratio of Car3/Actin (LI-COR Biosciences, Lincoln, NE).

Animal care for Car3 knockout mice
Car3 -/-mice were obtained from NIH (kind gift from Dr. Levine) on a mixed 129SvEv background as previously generated [49]. To outcross the Car3 knockout alleles into a C57BL/6J background more appropriate for metabolic studies [54][55][56], male progeny were backcrossed with C57BL/6J female at least 8 generations. Sibling matched males were used in all analyses, unless otherwise specified. Mice were weaned at 3-4 weeks of age and placed on standard chow (2020X, Teklad Diets, Madison WI). Mice were maintained at the University of North Carolina under protocols specifically approved for this study by the UNC IACUC.

Metabolic characterization of mice on high fat diet
At 5-6 weeks of age, mice either remained on standard chow or were placed on a high-fat diet [54,58,59]. The standard chow caloric consumption is 24 kcal% protein, 16 kcal% fat, and 60 kcal% carbohydrate, and the energy density is 3.1 kcal/g food (2020X, Teklad Diets, Madison WI). The high-fat diet caloric consumption is 20 kcal% protein, 60 kcal% fat, and 20 kcal% carbohydrate, and the energy density is 5.24 kcal/g food (D12492, Research Diets, New Brunswick, NJ).
Metabolic chambers and body composition. Mouse body composition, energy expenditure, activity, and food intake were measured at the Animal Metabolism Phenotyping Core at UNC under the supervision of Dr Kunjie Hua. Body composition was determined by MRI at 11 weeks of age (EchoMRI, Houston, TX) [60]. Energy expenditure, activity, and food intake were measured at 13-14 weeks of age using CaloCages (PhenoMaster, TSE systems, Chesterfield, MO). Data were collected for 2 days; the first day was considered the acclimation period and was excluded from data analysis [61]. Mice had unlimited access to food and water for the entire duration in metabolic chambers.
Serum measurements. Fasted mice were starved overnight (15-16 hours) [62] and measurements were taken at the beginning of the light cycle. To better synchronize the feeding schedule for fed measurements, food was removed from the cages during the light cycle, and mice were fed ad libitum during the dark cycle. Fed measurements were taken at the beginning of the light cycle after three nights of entrainment as described [63]. Glucose was measured with a drop of whole blood using a glucometer (Bayer Contour, Leverkusen, Germany). Whole blood was collected from mice using Microvette1 tubes (CB300Z, Sarstedt, Nümbrecht, Germany) under fasting and fed conditions. To extract serum from whole blood, the blood was allowed to clot for 30-90 min at room temperature and then centrifuged at 2000 x g for 20 min (4˚C), per provided instructions from Microvette. Insulin levels were determined using a mouse insulin ELISA kit (Crystal Chem Inc, Downers Grove, IL). Serum triglycerides (Wako Diagnostics, Richmond, VA), non-esterfied fatty acids (Wako Diagnostics, Richmond, VA), cholesterol (Stanbio, Boerne, TX) and glycerol (Sigma-Aldrich, St. Louis, MO) were determined with enzymatic colorimetric assays.
Oral glucose tolerance test (GTT). Sixteen-week-old mice were fasted overnight for 14-15 hours followed by oral gavage of 2 mg/g body weight of glucose. Blood glucose was measured at 0, 15, 30, 60, and 120 minute time points using a glucometer, as described previously described [64,65].
Tissue triglycerides. At 18 weeks of age, mice were sacrificed and tissues were weighed and collected for further analysis. Liver triglycerides were measured from the left lobe, while muscle triglycerides were measured from calf muscle using a protocol adapted from [66]. Tissues were homogenized on ice using the BeadBug (Benchmark Scientific, Edison, NJ) with 50-100 mg of tissue in 10 or 20 volumes of 5% NP-40 in water (10 volumes for liver samples and 30 volumes for muscle samples). Samples were boiled @ 100˚C for 5 min, then cooled to room temperature twice. The samples were then centrifuged at 16,000 x g for 2 min to pellet denatured DNA and proteins. Supernatants were then used to measure triglycerides using a colorimetric assay (Sigma-Aldrich, St. Louis, MO).
Body temperature and thermoregulation. Body temperatures from mice fed standard or high-fat diet were measured by rectal probe under fasting (15-18 hours) and fed (ad libitum) conditions (MicroTherma 2, ThermoWorks, American Fork, UT) [67]. Adult female mice, 8-13 weeks of age, were fasted overnight and exposed to 4˚C for 5 hours [68].

Statistics
All data and statistics were analyzed using GraphPad Prism 7. Data are expressed as mean ± SD. Significance was determined using Tukey's multiple comparisons for all figures except for body weights, energy expenditures, RER, GTT area under the curve (AUC), and tissue triglycerides. For GTT AUC and tissue triglycerides, significance was determined using unpaired T-test. A significance threshold of P < 0.05 was applied. For body weight, energy expenditure, and RER, discovery was determined by multiple T-tests using the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli, with Q = 1% [69]. Each row was analyzed individually, without assuming consistent SD.

Niclosamide treated mice have lower Car3 protein expression
In previous unpublished experiments, we observed that mice with fatty liver induced by dietinduced obesity (DIO) and treated with niclosamide had lower liver triglycerides and improved insulin sensitivity compared to untreated mice, which has been independently confirmed [47]. In an effort to identify which proteins might change in response to niclosamide treatment, we used 2D-DIGE analyses to compare liver protein expression from high-fat DIO treated with niclosamide or vehicle. Spots were then analyzed for differences in expression. One of the most abundant differentially expressed spot on the 2D-DIGE gel was excised and identified as Car3 (Fig 2A). In order to verify that Car3 protein expression changed with niclosamide treatment, we performed western blots on liver lysates from additional DIO mice treated with vehicle, 5 mg/kg or 25 mg/kg body weight doses of niclosamide ( Fig 2B). The western blots confirmed the 2D-DIGE results with high and low doses of niclosamide leading to lowered Car3 protein levels (standardized to actin) compared to the vehicle-treated mice (mean relative Car3 intensities of 1.56, 2.16 and 2.54 respectively).

Car3 is not required for de novo fatty acid synthesis
Given the correlation between Car3 expression and fatty liver we wondered whether Car3 plays a role in lipid homeostasis. Given that bicarbonate is required for lipid synthesis, we tested the hypothesis that Car3 might be required for lipogenesis, using the radiolabeled-acetate incorporation assay to compare de novo fatty acid synthesis of wild type and Car3-/-primary mouse hepatocytes. Because the rate-limiting step of fatty acid synthesis requires bicarbonate (Fig 3A), we were concerned that traditional cell culture incubations utilizing a bicarbonate-based buffering system would mask any potential differences in [1,2-14 C]acetate incorporation into lipid. To account for this, we used L-15 media, a cell culture media with a bicarbonate-free buffering system, to culture and assay the hepatocytes. We also performed the assay in the presence and absence of carbon dioxide during incubation. For both with or without CO 2 , the incorporation of [1,2-14 C]acetate into lipids was statistically indistinguishable between wild type and Car3 -/-cells. (Fig 3B) The incorporation of [ 14 C]acetate into fatty acids was 65-67% higher in the hepatocytes incubated with 5% CO 2 , as expected, since that increases free bicarbonate in the media available for the first step in fatty acid synthesis ( Fig  3A). These results indicate that Car3 expression has no significant effect on de novo lipogenesis in primary liver cells. Liver is the primary site of de novo lipogenesis in mammals.

Weight gain, body composition, and chow consumption is not affected by Car3
To determine if Car3 affects body weight or composition, wild type and Car3 -/-mice were put on a high-fat diet (HFD) or standard chow at 5 weeks of age and measured once a week for 10  weeks. Body weights of wild type and Car3 -/-mice did not differ significantly at any age for either standard chow or high-fat diet (Fig 4A). After 8-9 weeks on the diets, food consumption was monitored over a pre-acclimated 24-hour period in a metabolic chamber (Fig 4B). For both diets, Car3 did not affect food consumption. Wild type and Car3 -/-mice ate the same amount of standard chow or HFD during both the light and dark cycle. To see if fat and lean mass were individually affected, we measured body composition on standard chow and highfat diet. The lean masses of wild type and Car3 -/-mice were not statistically different, with the overall mean at 22.85 ± 1.058g (Fig 4C). Similarly, fat masses of wild type and Car3 -/-mice were not significantly different, with overall mean fat mass on standard chow at 1.031 ± 0.3721 and on high-fat diet at 7.729 ± 2.358 (Fig 4C). Mice on HFD had an increased contribution of fat mass compared to the mice on standard chow, as expected. These results suggest that Car3 is not affecting weight gain or fat and lean mass composition under ad libitum feeding conditions.
Car3 has no significant effect on total energy expenditure or energy substrate utilization with standard chow or HFD fed mice To determine if Car3 affects fuel source utilization, we measured total energy expenditure, energy substrate usage, and physical movement activity in metabolic chambers. Wild type and Car3 -/-mice on both diets showed no difference in activity or energy expenditure (Fig 5A). To measure contribution of lipids or carbohydrates to energy usage, respiratory exchange ratios (RER) were calculated for the wild type and Car3 -/-mice fed standard chow and HFD; an RER equal to 0.7 indicating fatty acid oxidation is the main source of energy, whereas an RER equal to 1.0 or greater indicates carbohydrates are the primary fuel source. Wild type and Car3 -/-mice have similar RER values throughout the 24-hour monitoring period, and the light and dark cycle values were not statistically different for either standard chow or HFD-fed mice (Fig 5B). During both the light and dark cycle, there were also no significant differences in movement activity/average beam breaks for wild type and Car3 -/-mice fed standard chow or HFD (Fig 5C).
Car3 has no effect on HFD-induced insulin resistance Mice fed high-fat diets can have elevated fasting blood glucose and insulin reflecting lower insulin sensitivity [58,70]. To test the effect of Car3 on insulin sensitivity, we performed glucose tolerance tests (GTT) and collected fasting blood glucose and insulin on mice fed either standard chow or high-fat diet. Since liver Car3 mRNA expression is highest in the fed state, to maximize possible differences between wild type and knockout mice, we also obtained fed blood measurements [25]. Both wild type and Car3 -/-mice fed standard chow had lower blood glucose and insulin in the fasting state than in the fed state (as expected); similarly HFDfed mice had elevated fasting blood glucose and insulin compared to standard chow fed mice (Fig 6A and 6B), as also expected. Glucose tolerance tests in mice fed standard chow and highfat diet were not statistically significant between wild type and Car3 -/-mice in individual glucose measurements or area under the curve calculations ( Fig 6C).

Car3 knockouts have normal serum and tissue lipid/fatty acid biomarker levels
To see if Car3 plays a role in increased liver triglycerides found in mice with fatty liver, we measured serum and tissue lipids (fatty acids and triglycerides). As stated previously, fed serum measurements were also obtained due to Car3 mRNA expression being highest in the liver in the fed state. Fasted and fed serum triglycerides, non-esterfied fatty acids (NEFA), cholesterol, and glycerol levels were all measured and found not statistically different between wild type and Car3 -/-mice on either diet (Fig 7A-7D). Liver and muscle triglyceride content also showed no difference between wild type and Car3 -/-mice (Fig 7E and 7F).

Is Car3 required for lipid uptake/mobilization during thermogenesis?
Given the abundance of Car3 in adipose and skeletal muscle and increase in fatty livers, this was somewhat reminiscent of the expression of CD36 protein, which also increases in fatty liver and is abundant in adipose tissue [71,72]. One of the most profound phenotypes of CD36 knockout mice is the inability to maintain core body temperature under cold challenge due to suspected inability to uptake lipid and subsequently hydrolyze it for energy/heat, particularly in brown fat [73]. Given the similarity in expression of CD36 and Car3 proteins in lipidstoring tissues, we subjected Car3 -/-mice to cold temperature challenge. Car3 -/-mice are not impaired in their ability to maintain core body temperature in the cold (Fig 8).

Discussion
Despite the abundance of Car3 in key metabolic tissues including white and brown fat, liver, and skeletal muscle its function remains unknown. The relatively high abundance of Car3 these tissues-in fats Car3 is nearly 30% of soluble protein and 8% in slow-twitch muscle [23, 49]-led to extensive analysis of Car3 -/-muscle function but showed relatively mild muscle phenotypes with normal "growth, development and life span" [49].
Although the exact function of Car3 remains unknown, many researchers postulate Car3 function is not as a carbon dioxide hydratase given its relatively miniscule enzymatic activity compared to Car1 and Car2 (0.16% with purified protein in vitro [10]). Additional evidences suggesting Car3 may not function to produce bicarbonate in vivo are observations that the most potent Car3 inhibitor is carbonate ion (10μM K I ), which is in equilibrium with bicarbonate itself [10]. These investigators stated "it is difficult to explain why Nature preserved . . . such a "bad" catalyst" that is also most potently inhibited by a proposed equilibrium product.
Even though evidence suggests Car3 may have no functional role in producing bicarbonate, we still explored potential roles for Car3 in lipid homeostasis because Car3 protein is highly enriched in tissues that can accumulate or synthesize lipid, and because bicarbonate is essential for fatty acid synthesis [18]. The three prior studies characterizing Car3 knockout mice did not investigate phenotypes related to lipid synthesis or storage nor did these studies feed high-fat diets to find any phenotypes [20,49,50].
Although Car3 may not drive bicarbonate production in vivo, no one previously used Car3 -/-mice to demonstrate that Car3 is not required for de novo lipogenesis in vivo (Fig 3). Given the abundance of Car3 in lipid containing tissues, we decided to focus on the response of Car3 -/-mice fed a high-fat diet known to alter lipid homeostasis. We measured key metabolic indicators including: weight gain, body composition, chow consumption, energy expenditure, insulin resistance, and dyslipidemia. Car3 deficiency had no effect on the biomarkers we tested, suggesting that Car3 is not responsible for the synthesis of fat nor does it affect the mobilization of fat in these tissues.
We identified Car3 as a protein increased in the liver in a mouse model fed high-fat diet but lowered with niclosamide (Fig 2), a drug shown to improve fatty liver and T2D-like symptoms [47]. Since Car3 liver protein expression was higher in DIO mice fed HFD, and lower in niclosamide treated HFD-fed DIO mice, we hypothesized that Car3 was involved in the development of fatty liver in this disease model. We saw no difference in liver triglycerides of wild type and Car3 -/-mice fed HFD, which suggests that Car3 is not required for accumulation of fat in the liver. While Car3 could be a biomarker for fatty liver disease in DIO mice, it is also possible that the effect of decreased Car3 seen in niclosamide-treated mouse livers may simply reflect lower fat content in the tissue, without regulating its decrease.
The high abundance of Car3 in selected tissues and relative lack of enzymatic activity suggest perhaps Car3 plays a non-enzymatic role. There is evidence that Car3 acts as a protectant against oxidative damage; cells transfected with Car3 are protected from apoptosis induced by hydrogen peroxide [21]. Car3 can also undergo the posttranslational modification by S-glutathionylation at two cysteine residues, Cys186 and Cys181 not found in Car1/2 [74][75][76]. Sglutathionylation of proteins is an early response to oxidative damage [77]. S-glutathionylation of Car3 increases with increased oxidative damage that occurs naturally as in aging [78] or with muscle damage [22]. Car3 contains numerous other additional cysteine residues not found in Car1/2. The concept of Car3 as an abundant "scavenging" protein to protect against oxidative damage is becoming an area of more research focus [79].
Interestingly, Car3 protein expression is altered in SOD1, copper zinc super oxide dismutase 1, providing another connection to oxidative stress responses in tissues. In SOD1 -/-mouse knockouts Car3 protein was identified as an abundant oxidized protein in liver that decreases over time specifically in knockouts [80]. Similarly, in the 'toxic milk' mouse mutant, Car3 protein is also greatly reduced in the liver; Car3 was shown to bind copper ions in addition to zinc [81]. (Toxic milk protein is an ATP7B transporter that helps regulate copper levels in the body.) Thus Car3 seems to have physiological connections to both metallic ions and oxidative stress responses. While our study has not demonstrated a role for Car3 in lipid metabolism, future characterization of Car3 -/-mice for roles in oxidative stress responses or for functions related to binding metal ions in cells may be more fruitful.

Conclusion
In conclusion, we attempted to determine the role of Car3 in lipid metabolism/homeostasis after we discovered Niclosamide decreased Car3 protein in the livers of HFD fed mice under similar conditions that improve fatty liver and insulin sensitivity/diabetic symptoms in mice [46,47]. Since Niclosamide decreases liver triglycerides and improves insulin sensitivity, we hypothesized that perhaps Car3 had a functional role in promoting fatty acid synthesis or lipid accumulation. We determined using mouse genetics that Car3 is not required for de novo lipogenesis. Additionally, we found that Car3 knockout and wild type mice fed either a standard chow or a HFD are not statistically different in weight, body composition, metabolic rate, insulin sensitivity, serum lipids, or tissue triglycerides. While Car3 is an abundant protein in many key metabolic tissues, the physiological role of Car3 in these tissues still remains to be determined.