Defective glycerol metabolism in aquaporin 9 (AQP9) knockout mice

Abstract

Aquaporin-9 (AQP9) is an aquaglyceroporin membrane channel shown biophysically to conduct water, glycerol, and other small solutes. Because the physiological role/s of AQP9 remain undefined and the expression sites of AQP9 remain incomplete and conflicting, we generated AQP9 knockout mice. In the absence of physiological stress, knockout mice did not display any visible behavioral or severe physical abnormalities. Immunohistochemical analyses using multiple antibodies revealed AQP9 specific labeling in hepatocytes, epididymis, vas deferens, and in epidermis of wild type mice, but a complete absence of labeling in AQP9^−/− mice. In brain, no detectable labeling was observed. Compared with control mice, plasma levels of glycerol and triglycerides were markedly increased in AQP9^−/− mice, whereas glucose, urea, free fatty acids, alkaline phosphatase, and cholesterol were not significantly different. Oral administration of glycerol to fasted mice resulted in an acute rise in blood glucose levels in both AQP9^−/− and AQP9^+/− mice, revealing no defect in utilization of exogenous glycerol as a gluconeogenic substrate and indicating a high gluconeogenic capacity in nonhepatic organs. Obese Lepr^db/Lepr^db AQP9^−/− and obese Lepr^db/Lepr^db AQP9^+/− mice showed similar body weight, whereas the glycerol levels in obese Lepr^db/Lepr^db AQP9^−/− mice were dramatically increased. Consistent with a role of AQP9 in hepatic uptake of glycerol, blood glucose levels were significantly reduced in Lepr^db/Lepr^db AQP9^−/− mice compared with Lepr^db/Lepr^db AQP9^+/− in response to 3 h of fasting. Thus, AQP9 is important for hepatic glycerol metabolism and may play a role in glycerol and glucose metabolism in diabetes mellitus.

Aquaporin 9 (AQP9) is a member of the aquaglyceroporin subfamily of aquaporins and shares the highest amino acid sequence homology with AQP3, AQP7 (1), and AQP10 (2). In addition to water, the aquaglyceroporins transport small uncharged molecules like glycerol, urea, purines, and pyrimidines (3, 4), but their physiological function(s) remains unknown. AQP9 expression has been reported in many tissues, using different immunohistochemical and molecular approaches. In the liver, AQP9 is expressed in hepatocytes within the sinusoidal surfaces of hepatocyte plates, where, during starvation, it is speculated to function in glycerol uptake from the bloodstream for gluconeogenesis (5, 6). In addition, AQP9 has been proposed to be involved in urea elimination from hepatocytes (3). AQP9 gene expression in the liver is down-regulated by insulin, potentially via an “insulin responsive element” in the AQP9 promoter.

AQP9 expression has been reported in other tissues including the male reproductive tract, where it localizes to the efferent ductule epithelium, epididymis, and vas deferens and may be involved in sperm maturation, concentration, and storage, respectively (7). The expression of AQP9 has also been shown to be regulated by testosterone and estrogen (8, 9). AQP9 expression has been reported in the plasma membranes of Leydig cells in rat testis (10), in rat spleen white pulp (11), in rat brain (11–13), in rat spinal cord (14), in the apical membrane of the trophectoderm of the mouse blastocyst (15), in syncytiotrophoblast of human term placenta (16), and in the peripheral leukocytes (3). The function of AQP9 in these locations remains unknown.

It is important to emphasize that, despite an abundance of data from different studies, there are still major discrepancies between the reported expression sites, especially within the brain. In speculation, these differences may be due to either heterogeneity of expression in different species, existence of different splice variants, or may represent artifacts related to specificity of the anti-AQP9 antibodies used in immunohistochemical and immunoblotting methods.

The purposes of the present studies were: (i) to generate AQP9 gene knockout mice; (ii) to identify the major expression sites of AQP9 in mouse by exploiting the AQP9 gene knockout mice; (iii) to identify the physiological role(s) of AQP9 by examining the phenotype of AQP9 gene knockout mice; (iv) to generate and phenotype obese Lepr^db lacking AQP9 (Lepr^db is a leptin receptor mutation with absence of leptin function leading to severe obesity and type II diabetes). One specific purpose of the studies was to investigate whether there was evidence for a role of AQP9 in hepatic glycerol metabolism because glycerol is a gluconeogenic substrate potentially contributing to the elevated hepatic glucose production seen in obese type II diabetic patients. These approaches have provided further insights into the expression and function of AQP9 and evidence for a possible role of AQP9 in diabetes mellitus.

Results

Generation of AQP9 Knockout Mice.

In the AQP9 targeted allele, 55 nucleotides of exon 2 were substituted for sequence encoding a neomycin phosphotransferase expression cassette (Fig. 1A). This insertion was predicted to result in correct translation of the initial 47 aa, followed by 100 random amino acids and a stop codon encoded by vector sequence and the inverted pA signal of the neomycin phosphotransferase. The presence of the mutated AQP9 transcript was confirmed by RT-PCR of AQP9^−/− mouse liver RNA using primers prim1/prim2 (Fig. 1B), followed by sequencing of the amplification product. No PCR product was observed in an identical reaction using liver RNA from AQP9^+/+ mice. RT-PCR from liver RNA using the primer pair prim1/prim3 gave the expected 303-bp product in the AQP9^+/+ and AQP9^+/−; however, a smaller 177-bp product was amplified from cDNA from the AQP9^−/− mice. DNA sequencing showed that the smaller product is an alternatively spliced transcript, where exon 1 is spliced directly to exon 3. Because this splice variant was not amplified from AQP9^+/− cDNA, we believe that this is an infrequent alternative splicing event, and probably does not take place for the wild-type allele, but only for the allele containing the neomycin cassette in exon 2. The electronic translation of this transcript predicts a correctly translated initial 37 amino acids (exon 1) and a frameshift in exon 3 sequence, thus is unlikely to result in any form of AQP9 protein.

Fig. 1.

Generation of AQP9 knockout mice. (A) Schematic diagram of the targeting strategy. (B) AQP9 transcript analysis by RT-PCR of liver RNA and sequencing of the products. Primer pair “prim1” and “prim2” produced 412-bp-long PCR product, confirming the presence of the mutated transcript in the AQP9^−/− and AQP9^+/− mice. Primer pair “prim1” and “prim3” produced the correct wild-type 303-bp-long product in the AQP9^+/+ and AQP9^+/− mice, which is absent in the AQP9^−/− mice. However, in AQP9^−/− mice, these primers amplified 177-bp-long PCR product, which is evidence of incorrect splicing of AQP9 gene where whole exon 2 containing the neo expression cassette is lost. (C) Immunoblotting of liver and epididymis protein samples with antibody against AQP9 (Alpha Diagnostics). AQP9 in the epididymis appears to have slightly lower molecular weight than AQP9 expressed in the liver.

Immunoblotting of liver and epididymis protein extracts from AQP9^−/− and AQP9^+/+ mice revealed an absence of AQP9 protein in knockout mice (Fig. 1C). Interestingly, the protein isoform of AQP9 expressed in wild-type mouse epididymis appears to have a smaller molecular weight than the form of AQP9 expressed in the liver (Fig. 1C).

Crossing of AQP9^+/− heterozygous animals with C57BL/6 mice produced an approximately equal ratio of AQP9^+/−/AQP9^+/+ offspring (55:48 ratio). Initial heterozygote crosses produced 35 wild-type mice, 97 heterozygote offspring, and 44 AQP9^−/− mice, consistent with a 1:2:1 Mendelian pattern and indicating no increased embryonic mortality of the AQP9^+/− and AQP9^−/− mice. There were no detectable differences in physical appearance, body weight, or behavior between age-matched AQP9^−/− mice and littermate controls. AQP9^−/− females are fertile and produced litter sizes similar to female AQP9^+/− siblings (AQP9^−/−, 7.1 ± 0.5; AQP9^+/−, 6.1 ± 0.5 pups in litter, difference not statistically significant). Homozygous AQP9^−/− males are fertile, and microscopic examination of the sperm showed normal morphology and motility.

Expression of AQP9 Protein in Mouse Organs.

The tissue localization of AQP9 was examined using immunohistochemistry. Two different peptide-targeted antibodies against AQP9 were used: (i) AQP9A1 (Alpha Diagnostics, San Antonio, TX) and (ii) RA2674–685 (11). The following organs were examined: liver, epididymis, testis, skin, spleen, muscle, brain, spinal cord, ovaries, and intestine. In wild-type control mice, using RA2674–685 and AQP9A1 antibodies, only liver, epididymis, and skin (not previously identified expression site) showed strong immunostaining using both antibodies, whereas there was a complete absence of staining in AQP9^−/− mice. At high concentration, the AQP9A1 antibody also stained different structures in spleen, brain, and spinal cord; however, the same staining pattern was seen in AQP9^−/− mice, and we therefore believe it is caused by nonspecific binding. Consistent with these findings, RA2674–685 did not stain these tissues (not shown).

In liver from control mice, staining was restricted to the sinusoidal surfaces of hepatocyte plates (Fig. 2B). Staining was strongest around the central vein (perivenous zone), whereas the sinusoids around the portal vein (periportal zone) were stained weakly (Fig. 2B), consistent with previous observations (5). This pattern of staining was observed both in males and females (not shown). In AQP9^−/− mice, there was no staining of liver (Fig. 2C), and histological examination showed no apparent abnormalities.

Fig. 2.

Immunocytochemistry of liver from AQP9^+/+ (A and B) and AQP9^−/− (C and D) mice stained with polyclonal antibody RA2674–685 at dilution 1:100. Staining of the sinusoidal plates surrounding the central vein (CV) visible in the AQP9^+/+ mouse (A and B) is absent in the AQP9^−/− mouse (C and D).

In epididymis and vas deferens of control mice, the RA2674–685 and AQP9A1 antibodies stained microvilli of the principal cells (Fig. 3 A, B, and E). This staining was absent in AQP9^−/− mice (Figs. 3 C, D, and F). Histological analysis of epididymis from AQP9^−/− mice showed no signs of epididymal tubule dilatation or other histological abnormalities and the weight of the whole epididymis (caput, corpus, and cauda segments) did not differ statistically from control mice (2.37 ± 0.14 mg per g of body weight in AQP9^−/− and 2.50 ± 0.13 mg per g of body weight AQP9^+/−, n = 8 per group).

Fig. 3.

Immunocytochemistry of epididymis from AQP9^+/+ (A and B) and AQP9^−/− (C and D) mice, and vas deferens from AQP9^+/+ (E) and AQP9^−/− (F) mice stained with polyclonal antibody RA2674–685 at dilution 1:100. In the AQP9^+/+ mouse, AQP9 is present in the microvilli of the principal cells of the epididymis and vas deferens (A, B, and E), and the staining is completely absent in the AQP9^−/− mouse (C, D, and F).

In control mice (AQP9^+/+), AQP9 was detected in the single layer of epidermis cells in the stratum granulosum of dorsal skin (Fig. 4A). This staining was absent in the AQP9^−/− mice (Fig. 4B). Incision through the epidermis and dermis followed by a 10-day period of healing resulted in proliferation of regenerating tissue and multilayered stratum granulosum with prominent AQP9 expression (Fig. 4 C and D). No defects in wound healing were observed in AQP9^−/− mice.

Fig. 4.

AQP9 localization in the skin. (A and B) Immunocytochemistry of normal dorsal skin from AQP9^+/+ (A) and AQP9^−/− (B) mice stained with polyclonal antibody RA2674–685 at dilution 1:100. In the AQP9^+/+ mouse, AQP9 is present in the stratum granulosum of the epidermis (A), and the staining is absent in the AQP9^−/− mouse (B). (C and D) After wounding and a 10-day period of healing, the regenerating skin shows hypertrophy of the epidermis. The multilayered stratum granulosum shows clear AQP9 staining in the apical pole (C). The wounded skin from the AQP9^−/− mouse shows no staining (D).

Phenotyping of AQP9^−/− Mice Regarding Glycerol Metabolism.

To examine whether AQP9^−/− mice exhibit defects in the metabolism of various solutes, including glycerol, the concentration of glucose, glycerol, FFA, triglycerides, urea, total cholesterol, and alkaline phosphatase (ALP) in plasma samples from AQP9^−/− and AQP9^+/− mice were determined (shown in Table 1). AQP9^−/− mice exhibited a marked increase in plasma glycerol and triglyceride levels compared with AQP9^+/− mice, revealing a deficient glycerol metabolism. There were no significant differences in the other measured parameters.

Table 1.

Plasma values: Basal conditions (mice with free access to food)

	AQP9+/−	AQP9−/−
n	8	6
Glycerol, μmol/liter	339 ± 24	530 ± 28*
Triglycerids, mmol/liter	0.53 ± 0.05	0.83 ± 0.02*
Urea, mmol/liter	7.3 ± 0.6	7.4 ± 0.4
Total cholesterol, mmol/liter	1.80 ± 0.11	2.10 ± 0.15
Free fatty acids, mmol/liter	0.47 ± 0.07	0.66 ± 0.10
ALP, units/liter	144 ± 10	142 ± 7

Values are means ± SE; n, number of mice;

*, P <0.05, AQP^−/− compared with AQP^+/−.

After 24 h of fasting of AQP9^−/− and AQP9^+/− control mice, both groups showed similar body weight loss of ≈12.5%, and a similar decrease in blood glucose levels (Table 2). However, plasma glycerol and triglycerides levels were significantly increased in the fasted AQP9^−/− mice compared with fasted control mice, whereas FFA, urea, total cholesterol and ALP levels were unchanged (Table 3).

Table 2.

Response to 24 h of starvation

	AQP9+/−	AQP9−/−
n	10	12
Start body weight, g	20.4 ± 1.0	21.9 ± 0.8
Final body weight, g	17.9 ± 0.9	19.2 ± 0.7
% Change in body weight	12.6 ± 1.0	12.4 ± 0.5
Start blood glucose, mmol/liter	6.3 ± 0.2	6.6 ± 0.2
Final blood glucose, mmol/liter	3.5 ± 0.4	3.7 ± 0.3

Values are means ± SE; n, number of mice; *, P < 0.05, AQP9^−/− compared with AQP9^+/−.

Table 3.

Plasma values: Mice starved for 24 h

	AQP9+/−	AQP9−/−
n	12	12
Glycerol, μmol/liter	353 ± 16	493 ± 21*
Triglycerids, mmol/liter	0.58 ± 0.02	0.74 ± 0.03*
Urea, mmol/liter	6.9 ± 0.3	7.0 ± 0.4
Total cholesterol, mmol/liter	2.29 ± 0.09	2.46 ± 0.14
Free fatty acids, mmol/liter	1.34 ± 0.10	1.23 ± 0.07
ALP (units/liter)	142 ± 5	143 ± 6

Values are mean ± SE; n, number of mice;

*, P < 0.05, AQP9^−/− compared with AQP9^+/−.

To investigate whether AQP9-deficient mice are able to use an acute minor load of exogenous glycerol, AQP9^−/− and AQP9^+/− mice were fasted overnight for 16 h and subsequently administered 87% glycerol orally. Blood glucose levels were measured 10 min before glycerol administration and at 0, 15, 30, 45, 60 and 120 min after glycerol administration. Blood glucose levels increased comparatively in both the AQP9^−/− and AQP9^+/− mice after glycerol administration and there was no significant difference between the genotypes (Fig. 5A). These data indicate that there is a sufficient capacity to generate glucose from acutely administered exogenous glycerol in AQP9-deficient mice.

Fig. 5.

Physiological responses of AQP9 knockout mice. (A) Five milligrams of 87% glycerol per g of body weight was administered orally to AQP9^−/− and control mice (AQP9^+/−) starved for 16 h, and blood glucose level measured at different time intervals. There was no difference in the glucose levels after glycerol administration in the AQP9^−/− (filled circles) and the control (open circles) mice. (B) Obese (Lepr^db/Lepr^db) AQP9^−/− mice show similar body weight gain as obese AQP9^+/− mice. n = 18 AQP9^−/−, 19 AQP9^+/−. (C) The postabsorptive (3-h fasting) blood glucose is lower in the AQP9^−/− mice compared with AQP9^+/− mice. n = 18 AQP9^−/−, 19 AQP9^+/−

AQP9 Function in Obese Lepr^db Mice.

Because glycerol is a gluconeogenic substrate potentially contributing to the elevated hepatic glucose production seen in obese type II diabetes patients, we generated mice lacking AQP9 and leptin receptor function. Mice carrying a mutation in the leptin receptor gene, Lepr^db (Taconic, Ry, Denmark), are severely obese and develop type II diabetes and increased lipolysis and glycerol production (reviewed in ref. 18). Lepr^db and AQP9 knockout mice were crossed to generate obese Lepr^db/Lepr^db AQP9^−/− mice and obese Lepr^db/Lepr^db AQP9^+/− controls. Both Lepr^db/Lepr^db AQP9^+/− and Lepr^db/Lepr^db AQP9^−/− mice developed severe obesity, with no significantly different gain of body weight. The increased bodyweight was accompanied by development of type II diabetes within 15 weeks of age, evidenced by increased blood glucose levels and increased levels of plasma glycerol and free fatty acids (Fig. 5B and Table 4). Importantly, the plasma glycerol levels in obese Lepr^db/Lepr^db AQP9^−/− mice were markedly increased compared with the obese Lepr^db/Lepr^db AQP9^+/− controls. These results further support a major role of AQP9 in glycerol metabolism.

Table 4.

Plasma values in lean and obese AQP9^+/− and AQP^−/− mice: Basal conditions (mice with free access to food)

	AQP9+/−		AQP9−/−
	Obese Leprdb/Leprdb	Lean	Obese Leprdb/Leprdb	Lean
n	17	18	12	20
Blood glucose, mmol/liter	17.1 ± 1.4	6.6 ± 0.2*	17.4 ± 2.4	6.6 ± 0.2†
Glycerol, μmol/liter	553 ± 35	351 ± 40*	939 ± 43‡	505 ± 37§
Free fatty acids, mmol/liter	1.17 ± 0.07	0.92 ± 0.06*	1.34 ± 0.11	1.04 ± 0.07†

Values are means ± SE; n, number of mice;

*, P < 0.05 for obese Lepr^db/Lepr^db AQP9^+/− vs. lean AQP9^+/−;

†, P < 0.05 for obese Lepr^db/Lepr^db AQP9^−/− vs. lean AQP9^−/−;

‡, P < 0.05 for obese Lepr^db/Lepr^db AQP9^−/− vs. obese Lepr^db/Lepr^db AQP9^+/−;

§, P < 0.05 for lean AQP9^−/− vs. lean AQP9^+/−.

To test whether AQP9 may act as an entrance port for hepatic gluconeogenesis from nonhepatic glycerol, the ability of the Lepr^db/Lepr^db AQP9 ^−/− mice to generate glucose in response to fasting was examined. The blood glucose levels measured in the absence of fasting did not differ significantly in obese Lepr^db/Lepr^db AQP9^−/− mice and control obese Lepr^db/Lepr^db mice (Table 4). However, the blood glucose levels measured after 3 h of fasting were moderately lower in the Lepr^db/Lepr^db AQP9^−/− mice (Fig. 5C). Taken together, these results support the view that AQP9 plays a role in hepatic glycerol metabolism.

Discussion

AQP9 knockout mice were generated to identify the main expression sites and the physiological role of AQP9 that thus far have remained undefined. Our data determined that AQP9 is highly expressed in liver, which represents its main expression site. Plasma levels of glycerol and triglycerides were markedly increased in AQP9^−/− mice, revealing a role of AQP9 in glycerol metabolism. Obese Lepr^db/Lepr^db AQP9^−/− double knockout mice had a dramatic increase in plasma glycerol levels compared with obese Lepr^db/Lepr^db AQP9^+/− control mice, which had higher plasma glycerol levels than non-obese control mice. This result further supports a role of AQP9 in glycerol metabolism. Moreover the blood glucose levels measured after 3 h of fasting were moderately lower in the Lepr^db/Lepr^db AQP9^−/− mice compared with Lepr^db/Lepr^db AQP9^+/− mice, supporting the view that Lepr^db/Lepr^db AQP9^−/− have a deficiency in generating glucose in response to fasting due to lack of AQP9 as an entrance port for glycerol for hepatic gluconeogenesis. Taken together, these results strongly support the view that AQP9 is important for hepatic glycerol metabolism and provide further evidence for a role of AQP9 in diabetes mellitus.

Defective Glycerol Metabolism in AQP9 Gene Knockout Mice.

AQP9^−/− mice have markedly increased plasma glycerol and triglyceride levels, revealing a defect in glycerol metabolism. The abundant expression in the liver of normal mice strongly suggests that the increased plasma glycerol levels in AQP9^−/− mice are caused by an absence of hepatic AQP9 and an impaired uptake of glycerol through the hepatocyte plasma membrane. Although the increased plasma glycerol level in AQP9 null mice is likely to be mediated by absence of AQP9, it cannot be excluded that the uptake and release of glycerol by other organs expressing aquaglyceroporins, e.g., kidney cortex (AQP3 and AQP7), fat tissue (AQP7), or intestine (AQP3 and AQP10), may also influence the final plasma glycerol level. Importantly, it cannot be excluded that the AQP9^−/− mice exhibit compensatory regulation of glycerol metabolism, which may also be the reason why they tolerate 24-h starvation.

Hypoglycemic (fasted) AQP9^−/− mice were capable of efficiently using orally administered glycerol for acute production of glucose. It is likely that administered glycerol is converted to glucose in secondary gluconeogenic organs, e.g., the kidney or the intestine (19), compensating for the assumed defect in glycerol transport into the hepatocytes. Thus, our data reveal that the capacity to metabolize exogenous glycerol via an AQP9-independent pathway is substantial, although the relative contribution is difficult to estimate due to potential compensatory mechanisms in AQP9^−/− mice. It is also possible that, in the present setting, the rate of glycerol uptake was not a rate-limiting step in hepatic gluconeogenesis.

The increased mass of adipose tissue in the Lepr^db/Lepr^db mice is associated with increased lipolysis resulting in the observed increased plasma glycerol levels. This increased glycerol availability could potentially lead to increased conversion to glucose in liver or other gluconeogenic organs and play a role in the development of type II diabetes. The obese Lepr^db/Lepr^db AQP9^−/− mice show a dramatically increased plasma glycerol level compared with obese Lepr^db/Lepr^db AQP9^+/− controls (consistent with the hypothesis of impaired glycerol utilization by the liver). To further support the hypothesis that AQP9 is involved in glycerol uptake and gluconeogenesis, Lepr^db/Lepr^db AQP9^−/− and Lepr^db/Lepr^db AQP9^+/− control mice were fasted for 3 h. Without fasting, the mice showed similar blood glucose levels, possibly due to the glucose ingestion (Table 4). After a 3-h fasting period, the Lepr^db/Lepr^db AQP9^−/− mice exhibited lower postprandial plasma glucose levels compared with those of Lepr^db/Lepr^db AQP9^+/− control mice (Fig. 5C), suggesting that the absence of AQP9 in the hepatocyte membrane reduced the capacity for glycerol entrance for gluconeogenesis, leading to reduced plasma glucose levels, which may indicate an improvement of the diabetic state. This result is consistent with a role of AQP9 as a glycerol-importing pathway. Further studies are required on these findings, but the development of specific blockers of aquaglyceroporins could potentially be very useful in elucidating whether pharmacological blockade of AQP9 may be of potential benefit in type II diabetes. Moreover, further experiments will be necessary to clarify whether insulin production and insulin action are improved in the obese Lepr^db/Lepr^db AQP9^−/− mice compared with Lepr^db/Lepr^db AQP9^+/− control mice.

Interestingly, according to the model of metabolic zonation of the liver, gluconeogenesis predominantly takes place in the hepatocytes surrounding the portal vein (periportal) that express little AQP9 (reviewed in ref. 20), whereas relatively less gluconeogenesis takes place in the hepatocytes surrounding the central vein (perivenous), where highest AQP9 levels are detected. Thus, it cannot be excluded that AQP9 may have additional functions in the mouse liver other than glycerol transport.

Expression of AQP9 in Skin and Male Reproductive Tract.

Our studies report the presence of AQP9 in skin, specifically in the epidermis outermost layer stratum granulosum. In addition to AQP9, another aquaglyceroporin, AQP3, is present in the epidermis at a different site: the basal cell layer (21). Previous studies have reported that mice deficient in AQP3 have a defective skin hydration, decreased skin elasticity, and impaired wound healing (22). It is therefore plausible that AQP9, together with AQP3, is involved in maintaining skin hydration. However, no apparent defects in wound healing were observed in AQP9 knockout mice.

Male AQP9^−/− mice are fertile and spermatozoa show normal motility. Further studies exploiting AQP9 knockout mice will be helpful in determining the role of AQP9 in fluid transport.

Materials and Methods

Generation of AQP9 Knockout Mice.

The genomic sequence of the mouse AQP9 gene was obtained from www.ensembl.org (transcript ID ENSMUST00000074465). A 3.2-kb DNA fragment (targeting vector arm 1) was amplified using genomic DNA from 129S1/Sv strain of mice and primer pair 5′-tttccgcggCTCAGGTCTCATGCAATGTCAGC-3′ and 5′-ttaccgcggACGGCAGTTGTGATGGCTCTTTA (lowercase indicates the restriction enzyme sequence; uppercase indicates the anchor to genomic DNA sequence). The 2.8 kb fragment (targeting vector arm 2) was amplified using primer pair 5′-cttctcgagcccgggGCTTGAGCAATAGAGCCACATCC-3′ and 5′-cctctcgagCAGTAGTCAGTGCCACTCTGCAAC-3′. A targeting construct was created by inserting the 2.8-kb vector arm 2 into the XhoI site of pKO Scrambler 1903, and the 3.2-kb vector arm 1 into the SacII site. The targeting construct was linearized by using NotI and electroporated into CJ7 embryonic stem cells derived from 129S1/Sv mice (23). G418-resistant colonies were selected and expanded. Clones with homologous recombination were detected by Southern blot analysis using a probe flanking the 3.2-kb arm and confirmed by PCR. Six clones with the correct recombination event were obtained and one clone was used for injection into 20 B6D2F2 (24) mouse blastocysts. The chimeric males were bred with C57BL/6 females and agouti offspring (indicating germ-line transmission of the manipulated 129S1/Sv ES cells) were tested for the presence of the disrupted AQP9 allele by PCR using genomic tail DNA and 3 primers: GTGCTACTTCCATTTGTCACGTCCT, GCCACTAGCCATGTGTTGGTATTTC, and AACTGGGGATAGTGGGATTCAAAGA. The expected PCR product for the mutated allele is 503 bp, and the expected PCR product for the wild-type allele is 735 bp. Heterozygous mice were further bred to obtain homozygous AQP9 KO mice on a mixed genetic C57BL6/129S1/Sv background.

RT-PCR.

Liver RNA was isolated by using the RNeasy kit (Qiagen, Valencia, CA). RT-PCR was performed by using SuperScript II and primer pair prim1: 5′-AAAGGGGAACTTGAACCACTCCA-3′ and prim2: 5′-GTGCTACTTCCATTTGTCACGTCCT-3′ (the expected product for the disrupted AQP9 allele 412 bp) or the primer pair prim 1 and prim 3: 5′-GAGAAGGACCGAGCCAAGAAGAA-3′ (the expected product for the wild-type allele is 303 bp).

Blood/Plasma Measurements.

Method A: blood glucose measurements were performed using an Accu-Chek Sensor (Roche, Mannheim, Germany) and a drop of saphenous vein blood. Method B: for other tests, mice were anesthetized, their chest cavity opened, and blood samples collected from the right ventricle of the heart into heparinized tubes. The blood cells were removed by centrifugation for 10 min at 4,000 × g, and plasma glycerol, triglycerides, urea, free fatty acids, and total cholesterol analysis using standard laboratory tests. Total ALP was measured using a Vitros 950 analyzer (Ortho Chemical Diagnostics, Copenhagen, Denmark).

Immunostaining.

Six-week-old AQP9^+/+, AQP9^+/−, and AQP9^−/− siblings were perfusion fixed through the heart using 3% PFA/0.1 M cacodylate buffer. Tissues were extracted and postfixed for 1 h in the same fixative, and washed three times for 10 min in 0.1 M cacodylate buffer. Tissues were embedded in paraffin, 2-μm sections were cut on a rotary microtome (Leica Microsystems, Herlev, Denmark) and subjected to immunolabeling, as described (17). Primary antibodies used were (i) AQP9A-1 (Alpha Diagnostics) and (ii) RA2674–685 as described (11).

Protein Sample Preparation and Immunoblotting.

Organs were homogenized in dissection buffer (0.3 M sucrose/25 mM imidazole/1 mM EDTA, pH 7.2, containing 8.5 μM leupeptin, and 1 mM phenylmethylsulfonyl fluoride) using an ultra-turrax T8 homogenizer (IKA Labortechnik, Staufen, Germany). For semiquantitative immunoblotting using antibody AQP9A-1 (Alpha Diagnostics) the homogenate was centrifuged at 4,000 × g for 15 min at 4°C to remove whole cells, nuclei, and mitochondria. Gel samples were prepared from the supernatant by addition of Laemmli sample buffer to a 0.5% SDS final concentration. The immunoblotting procedure was performed as described (11).

Skin Wounding and Regeneration.

An ≈4-cm² area on the mouse back was shaved to synchronize the hair follicle cycle. Two weeks later, the skin was shaved again and, under general anesthesia, a 1-cm-long wound was generated through the epidermis and dermis using scissors. The wound was closed using metal stitches and allowed to regenerate for 10 days. The shaven area of skin was subsequently excised and immersion fixed in 3% PFA, 0.1 M cacodylate buffer for 24 h and processed for immunostaining as described above.

Oral Glycerol Administration to Fasted Mice.

Mice were fasted for 16 h (from 1600 until 0800 the following day), and 5 mg of 87% glycerol per g body weight was administered orally by spontaneous feeding. The blood glucose level was measured at stated time intervals before and after glycerol administration using an Accu-Chek Sensor (Roche) and a drop of saphenous vein blood.

Presentation of Data and Statistical Analyses.

Quantitative data are presented as means ± SE. Statistical comparisons were accomplished by unpaired t test (equal variances). P values < 0.05 were considered statistically significant.

Abbreviations

AQP9

aquaporin 9

ALP

alkaline phosphatase.