VLDL Induces Adipocyte Differentiation in ApoE-Dependent Manner
Objective— To clarify the role of very low density lipoprotein (VLDL) and apolipoprotein E (apoE) in adipogenesis, we studied newly developed hyperlipidemic obese (ob/ob;apoE−/−) mice. Because hydrolysis of VLDL is believed to be the major source of adipogenic free fatty acids, a higher plasma level of VLDL in these mice should exaggerate obesity.
Methods and Results— When fed a high-fat, high-cholesterol diet, ob/ob;apoE−/− mice did not show increased body weight or an increased amount of adipose tissue in spite of increased plasma VLDL levels, whereas ob/ob mice showed an increased body weight and amount of adipose tissue, suggesting that there is a novel apoE-dependent pathway for adipogenesis. In vitro experiments using bone marrow stromal cells and 3T3-L1 cells confirmed this notion. ApoE-deficient VLDL did not induce adipogenesis, whereas normal VLDL induced adipogenesis in these cells. The incubation of apoE-deficient VLDL with recombinant human apoE restored its adipogenic activity. Tetrahydrolipstatin, a lipoprotein lipase inhibitor, did not affect the adipogenic activity of VLDL, suggesting that hydrolysis of VLDL did not play a major role in its effects. In fact, lipid components of VLDL or free fatty acids induced only partial adipogenesis.
Conclusions— Our findings indicate that VLDL induces adipogenesis in an apoE-dependent manner both in vitro and in vivo.
Free fatty acids (FFAs) generated by the hydrolysis of VLDL by lipoprotein lipase (LPL) have been believed to the major driving force in adipogenesis and the development of obesity. FFAs can regulate various steps of adipogenesis, including the activation of early signals, such as peroxisome proliferator-activated receptor-γ (PPARγ), induction of adipocyte-specific genes, and maturation to heterogeneous adipocytes containing different sizes of lipid droplets or metabolic activities.1,2⇓ LPL, secreted from adipocytes and other cell types, is the rate-limiting enzyme for this process.3,4⇓ Heparan sulfate proteoglycan, the VLDL receptor, and LDL receptor-related protein (LRP) act as cell surface anchors for LPL.5–7⇓⇓ Deficiency of this enzyme induces severe hypertriglyceridemia in humans and mice.8,9⇓ Homozygous knockout mice are lipodystrophic and die of hypertriglyceridemia soon after birth.8
However, the importance of LPL in adipogenesis has been challenged by recent investigations studying various new mouse models and humans. Adipose tissue-specific knockout of LPL resulted in normal adiposity, which was mainly due to increased de novo lipogenesis in adipocytes.10 Crebbp heterozygous mice show lipodystrophy with a change in the expression of a series of genes related to lipid metabolism, suggesting that adipogenesis is a complex process regulated by different steps.11 Rodents deficient in leptin or leptin receptors show both obesity and increased activity of fatty acid synthase, suggesting the importance of intrinsic lipogenesis in obesity.12 Human LPL deficiency and apoC-II deficiency result in impaired hydrolysis of VLDL and chylomicron but are usually associated with normal adiposity,9,13,14⇓⇓ suggesting the importance of LPL-independent pathways in adipogenesis in humans.
A possible LPL-independent mechanism for adipogenesis is de novo synthesis of FFA from glucose taken up via glucose transporter 4.15 This is the main pathway in adipogenesis of cultured cells. Recently, this pathway has also been suggested to play an essential role in vivo. Another possible mechanism is receptor-mediated uptake of VLDL by adipocytes. ApoC-I specifically inhibits VLDL binding to apoE receptors.16 Therefore, human apoC-I transgenic mice exhibit hyperlipidemia, whereas they are protected from obesity and insulin resistance.17,18⇓ VLDL receptor-deficient mice do not show increased body weight when fed a high-fat, high-calorie diet.19 Both animal models suggest that receptor-mediated uptake of VLDL plays an important role in diet-induced obesity.
In the present study, we used apoE-deficient obese (ob/ob;apoE−/−) mice and studied the role of apoE, the ligand for receptors of VLDL, in the development of obesity. These obese mice show high plasma levels of chylomicron, VLDL, and their remnants that are due to the impaired clearance of these lipoproteins by receptor-mediated mechanisms.20 In spite of higher plasma lipid levels, when fed a high-fat, high-cholesterol diet, ob/ob;apoE−/− mice are less obese than are ob/ob mice, suggesting an important role of apoE in the development of obesity. In vitro experiments confirm this notion.
Materials used in the present study can be accessed at the online data supplement (http://atvb.ahajournals.org).
Breeding pairs of C57BL/6J, apoE−/− (C57BL/6J),21,22⇓ and heterozygous ob/ob (C57BL/6J)23 mice were obtained from The Jackson Laboratory (Bar Harbor, Me), housed under specific pathogen-free conditions, and maintained in static microisolator cages. Heterozygous ob/ob mice were interbred with apoE−/− mice to generate ob/ob;apoE−/− mice. All procedures were performed in accordance with the guidelines for animal welfare of Tokyo Medical and Dental University, and the protocol was approved by the Animal Welfare Committee of the institute. Animals were fed a normal chow diet until 8 weeks of age, and then they were fed either a high-fat, high-cholesterol diet containing 12.5% fat and 1.25% cholesterol (CLEA Japan, Inc)24 or a normal chow diet ad libitum for 4 weeks. ApoE-deficient VLDL was prepared from the blood of mice fed the high-fat, high-cholesterol diet for 4 weeks. The adipose tissue was fixed in 4% formaldehyde and used for histological analysis. Thick sections were stained with Elastica van Gieson’s stain and photographed with a Leica DMRX. The area of adipocytes was determined in 3 mice of each genotype by using NIH-image software.
3T3-L1 cells25 at passages 3 to 9 were used in all studies. Cells were cultured for 2 days before the medium was changed to test medium containing 10 μg/mL insulin, 100 μg/mL VLDL, or both. The control medium was DMEM supplemented with 10% FBS. After incubation for the indicated period, the cells were stained with oil red O or used for total RNA preparation.
Bone Marrow Stromal Cells
Bone marrow cells were harvested from the femur of male apoE−/− mice at 4 to 5 weeks of age under aseptic conditions as described before,26 and then they were cultured in DMEM supplemented with 10% FBS. Every fourth day, half of the medium was replaced with fresh medium. Adherent fibroblast-like cells (passage 3) were used as bone marrow stromal cells.
Very Low Density Lipoprotein
Human and mouse VLDL (density <1.006 g/mL) was isolated from freshly prepared plasma by ultracentrifugation as reported previously.27 In the experiment presented in Figure 5, VLDL prepared from the plasma of apoE−/− mice was incubated with recombinant human apoE for 1 hour at 37°C28 and immediately added to the cell culture. ApoE-deficient human VLDL was prepared by treating VLDL with trypsin in Tris buffer (pH 7.4) for 2 hours at 37°C as described previously.29
Total cellular RNA was prepared with TRIZOL reagent, and real-time polymerase chain reaction (PCR) was performed with the use of a LightCycler (Roche Molecular Biochemicals) according to the manufacturer’s instructions. The primers for adipocyte fatty acid binding protein (aP2), leptin, resistin, PPARγ, and β-actin can be accessed at the online supplement. The results are expressed as the copy number ratio of the target mRNA to the β-actin mRNA.
Electrophoretic Mobility Shift Assay
3T3-L1 cells were cultured with 100 μg/mL VLDL in the presence or absence of insulin for 2 weeks. Then the nuclear extract was prepared, and electrophoretic mobility shift assays and supershift assays with anti-PPARγ-1/-2 rabbit polyclonal antibody were conducted as reported previously.30 The results were visualized and analyzed quantitatively by using a BAS 2500 image analyzer (FUJIFILM).
Heparin-releasable LPL activity (5 U heparin/mL PBS for 15 minutes at 37°C) of 3T3-L1 cells was determined as described previously by using [carboxyl-14C]triolein as the substrate.31
Statistical significance was determined by Student t tests.
Ob/ob;apoE−/− mice showed the characteristics of both apoE−/− mice and ob/ob mice. When fed a normal chow diet, ob/ob;apoE−/− mice had significantly higher plasma levels of cholesterol and triglycerides (2139.0±515.1 and 407.3±127.2 mg/dL, respectively, at 12 weeks of age; n=14) than did apoE−/− mice (788.1±203.9 and 179.1±77.4 mg/dL, respectively; n=13) or ob/ob mice (171.4±20.9 and 77.2±14.0 mg/dL, respectively; n=5). High-performance liquid chromatography showed that most of these lipids were distributed in chylomicron, VLDL, and their remnants (data not shown). Plasma levels of LDL and HDL were minimal.
Ob/ob;apoE−/− mice had body weights similar to those of ob/ob mice, and they were heavier than wild-type (C57BL/6J) or apoE−/− mice from 4 weeks of age. There was no difference in body weight between ob/ob;apoE−/− and ob/ob mice from 4 to 12 weeks of age (at 12 weeks of age, 42.5±4.3 g [n=12] versus 43.7±1.5 g [n=5], respectively; P=NS) in spite of the fact that ob/ob;apoE−/− mice had a higher plasma level of VLDL than did ob/ob mice, indicating that apoE-deficient VLDL does not contribute to the obesity in these mice.
Effects of High-Fat, High-Cholesterol Diet on Body Weight
When fed a high-fat, high-cholesterol diet for 4 weeks from 8 weeks of age, ob/ob;apoE−/− mice failed to gain body weight, whereas the body weight of ob/ob mice increased by 30%. At least 40% of the difference in body weight between the 2 strains was attributable to the difference in the mass of subcutaneous and abdominal adipose tissue recovered under a dissecting microscope (Table). The difference was not due to a difference of food intake; the difference was maintained when the amount of food intake was restricted so that the 2 stains of mice ate the same amount of food (data not shown).
Histological analysis showed that the adipose tissue of ob/ob;apoE−/− mice consisted of adipocytes that were significantly smaller than those of ob/ob mice (Figure 1). The effect of the high-fat, high-cholesterol diet was also observed in apoE−/− mice. They did not accumulate adipose tissue on the high-fat, high-cholesterol diet, whereas wild-type mice did (Table). The adipocytes of apoE−/− mice were smaller than those of wild-type mice (Figure 1). These findings suggest that there is an apoE-dependent pathway for adipogenesis in these animals.
Mechanisms Involved in VLDL-Induced Adipogenesis
To understand the underlying mechanisms of VLDL-induced adipogenesis, we cultured 3T3-L1 cells in the presence of VLDL. VLDL induced the expression of adipocyte-specific genes (aP2, leptin, and resistin) in a dose- and time-dependent fashion in the presence of insulin (Figure 2A and 2B and online Figure I, available at http://atvb.ahajournals.org). 3T3-L1 cells became fully differentiated adipocytes filled with large lipid droplets after a 14-day incubation with VLDL (Figure 2C). 3T3-L1 cells did not differentiate to adipocytes in the absence of insulin. The combination of insulin, isobutylmethylxanthine, and dexamethasone similarly induced adipogenesis, as reported previously.25 Insulin alone induced aP2 gene expression to a lesser extent than when it was combined with the other agents (Figure 2B) and barely induced lipid-containing mature adipocytes within 2 weeks (Figure 2C).
With this model, we first studied the effect of VLDL on PPARγ, one of the key transcriptional factors for adipogenesis. VLDL increased the amount of PPARγ mRNA in a dose-dependent manner (Figure 3A). An electrophoretic mobility shift assay showed that the formation of a PPARγ-retinoid X receptor-α (RXRα) heterodimer increased 5-fold in cells stimulated with both VLDL and insulin, whereas it increased 2-fold in cells stimulated with insulin alone (Figure 3Ba). The supershift induced by incubation with anti-PPARγ confirmed that these bands contained PPARγ (Figure 3Bb). These findings suggest that VLDL induces adipocyte differentiation via PPARγ.
We then studied the role of FFA generated by LPL activity. 3T3-L1 cells were incubated with VLDL in the presence of tetrahydrolipstatin (THL), a catalytic-site inhibitor of LPL,32 or with lipid components of VLDL. THL (100 μmol/L) did not suppress the VLDL-induced aP2 gene expression, whereas 1 μmol/L THL completely inhibited the LPL activity produced by 3T3-L1 cells (online Figure II, available at http://atvb.ahajournals.org). A lipid extract of VLDL and oleic acid induced the expression of adipocyte-specific genes but did not generate adipocytes filled with large lipid droplets such as generated by VLDL (Figure 4 and online Figure III, available at http://atvb.ahajournals.org). Linoleic acid, triglycerides, and cholesterol only barely induced the expression of adipocyte-specific genes. These findings suggest that VLDL particles play an important role in the development of mature adipocytes.
Role of ApoE in VLDL-Induced Adipogenesis In Vitro
To confirm the above notion, we then studied the effect of apoE-deficient VLDL on adipogenesis. VLDL prepared from the plasma of wild-type mice induced both the expression of aP2 and the accumulation of lipid in bone marrow stromal cells. In contrast, apoE-deficient VLDL prepared from apoE−/− mice failed to induce the differentiation of bone marrow stromal cells to adipocytes (Figure 5A and 5Ba to 5Bc). The adipogenic activity of apoE-deficient VLDL was partially restored by preincubation of apoE-deficient VLDL with recombinant human apoE (Figure 5A and 5Bd). ApoE alone did not induce adipogenesis (data not shown). Furthermore, selective removal of apoE from human VLDL by trypsin treatment abolished the adipogenic activity of human VLDL (Figure 5Ba, 5Be, and 5Bf). These findings show that apoE is required for VLDL-induced adipogenesis.
In the present study, we show that a novel apoE-dependent mechanism for VLDL-induced adipogenesis plays a role both in vitro and in vivo. When fed a high-fat, high-cholesterol diet, ob/ob;apoE−/− mice did not show increased body weight or an increased amount of adipose tissue in spite of an increased plasma VLDL level, whereas ob/ob mice showed a significantly increased body weight and amount of adipose tissue. These findings suggest the existence of an apoE-dependent pathway of VLDL-induced adipogenesis. In vitro experiments using apoE-deficient VLDL confirmed that VLDL induces adipogenesis in an apoE-dependent manner.
There are 2 potential mechanisms for the apoE-dependent adipogenesis. The first one is that apoE is required for hydrolysis of triglycerides by LPL. A few previous studies have shown that apoE accelerates the hydrolysis of triglycerides in VLDL by LPL.6,33⇓ However, there are also reports showing that apoE inhibits LPL activity27 and that LPL activity in apoE−/− mice is not impaired.34 Thus, it is controversial whether apoE is indispensable for LPL activity or not. We found that apoE−/− mice have normal plasma LPL activity (data not shown). In vitro, VLDL induces adipogenesis even in the presence of an LPL inhibitor. Taken together, these facts indicate that LPL activity is not likely involved in the apoE-dependent adipogenesis. However, the possible role of LPL as a linker between VLDL and its receptor has not been ruled out.35
The second and more likely mechanism is that VLDL is taken up by the cell via apoE receptors. Adipocytes express apoE receptors, including the VLDL receptor, LRP, and the LDL receptor.28,36–38⇓⇓⇓ An in vivo study using apoC-I transgenic mice and VLDL receptor-deficient mice supported the role of these receptors. Both of these types of mice were protected from obesity.18,19⇓ We attempted to confirm the role of the VLDL receptor in adipogenesis by using bone marrow stromal cells prepared from VLDL receptor-deficient mice. Contrary to our expectations, bone marrow stromal cells from these mice differentiated to adipocytes even without VLDL, probably because of the increased intrinsic lipogenesis in these cells (data not shown). It has also reported that VLDL receptor-deficient mice exhibit reduced LPL activity.39 Thus, it remains to be clarified whether VLDL-induced adipogenesis is mediated by internalization via the VLDL receptor. The involvement of other apoE receptors has not been investigated well. Insulin activates LRP in primary adipocytes isolated from rat epididymal fat pads28 and stimulates translocation of LRP to the membrane in 3T3-L1 cells,40 suggesting a possible involvement of LRP in adipogenesis. The LDL receptor mediates the uptake of VLDL by cells. It also binds LPL, increasing the hydrolysis of VLDL.41 Thus, the LDL receptor may be involved in adipogenesis. Although it remains to be determined which receptors, if any, are involved in VLDL-induced adipogenesis, the present study provides further evidence that receptor-mediated uptake of VLDL may play a role in the development of obesity.
Recent studies have revealed multiple pathways for adipogenesis. In addition to classic LPL-dependent hydrolysis of VLDL, de novo synthesis of fat in adipocytes and energy expenditure also regulate adipogenesis.42,43⇓ The lipid composition of VLDL also affects adipogenesis (Figure 4). In fact, the apoE-dependent pathway is not the only adipogenic pathway working in our mouse model. When mice were fed a chow diet, ob/ob;apoE−/− mice gained more body weight and adipose tissue mass than did apoE−/− mice in spite of the apoE deficiency. Although a high plasma VLDL level and normal LPL activity in ob/ob;apoE−/− mice favor the role of de novo synthesis of fat in the adipose tissue of these mice, the relative importance of the apoE-dependent and other adipogenic pathways remains to be studied under physiological and pathological conditions.
In summary, our in vivo findings showed that apoE can play an important role in adipogenesis. Our in vitro findings revealed that VLDL can induce adipogenesis in an apoE-dependent manner without LPL activity, providing one possible but not the only possible explanation for our findings with ob/ob;apoE−/− mice. This novel mechanism may play an important role in the development of obesity in humans.
This work was supported by grants from Ministry of Culture, Science, and Education and the Organization for Pharmaceutical Safety and Research (OPSR) to K. Shimokado. T. Chiba was the recipient of a postdoctoral fellowship from OPSR.
- Received April 7, 2003.
- Accepted June 13, 2003.
- ↵Olefsky JM. The effects of spontaneous obesity on insulin binding, glucose transport, and glucose oxidation of isolated rat adipocytes. J Clin Invest. 1976; 57: 842–851.
- ↵de la Llera M, Glick JM, Rothblat G. Mechanism of triglyceride accumulation in rat preadipocyte cultures exposed to very low density lipoprotein. J Lipid Res. 1981; 22: 245–253.
- ↵Saxena U, Klein MG, Goldberg IJ. Identification and characterization of the endothelial cell surface lipoprotein lipase receptor. J Biol Chem. 1991; 266: 17516–17521.
- ↵Takahashi S, Suzuki J, Kohno M, Oida K, Tamai T, Miyabo S, Yamamoto T, Nakai T. Enhancement of the binding of triglyceride-rich lipoproteins to the very low density lipoprotein receptor by apolipoprotein E and lipoprotein lipase. J Biol Chem. 1995; 270: 15747–15754.
- ↵Obunike JC, Lutz EP, Li Z, Paka L, Katopodis T, Strickland DK, Kozarsky KF, Pillarisetti S, Goldberg IJ. Transcytosis of lipoprotein lipase across cultured endothelial cells requires both heparan sulfate proteoglycans and the very low density lipoprotein receptor. J Biol Chem. 2001; 276: 8934–8941.
- ↵Weinstock PH, Bisgaier CL, Aalto-Setala K, Radner H, Ramakrishnan R, Levak-Frank S, Essenburg AD, Zechner R, Breslow JL. Severe hypertriglyceridemia, reduced high density lipoprotein, and neonatal death in lipoprotein lipase knockout mice: mild hypertriglyceridemia with impaired very low density lipoprotein clearance in heterozygotes. J Clin Invest. 1995; 96: 2555–2568.
- ↵Weinstock PH, Levak-Frank S, Hudgins LC, Radner H, Friedman JM, Zechner R, Breslow JL. Lipoprotein lipase controls fatty acid entry into adipose tissue, but fat mass is preserved by endogenous synthesis in mice deficient in adipose tissue lipoprotein lipase. Proc Natl Acad Sci U S A. 1997; 94: 10261–10266.
- ↵Rolland V, Dugail I, Le Liepvre X, Lavau M. Evidence of increased glyceraldehyde-3-phosphate dehydrogenase and fatty acid synthetase promoter activities in transiently transfected adipocytes from genetically obese rats. J Biol Chem. 1995; 270: 1102–1106.
- ↵de Graaf J, Hoffer MJ, Stuyt PM, Frants RR, Stalenhoef AF. Familial chylomicronemia caused by a novel type of mutation in the APOE-CI-CIV-CII gene cluster encompassing both the APOCII gene and the first APOCIV gene mutation: APOCII-CIV(Nijmegen). Biochem Biophys Res Commun. 2000; 273: 1084–1087.
- ↵Jong MC, van Dijk KW, Dahlmans VE, Van der Boom H, Kobayashi K, Oka K, Siest G, Chan L, Hofker MH, Havekes LM. Reversal of hyperlipidaemia in apolipoprotein C1 transgenic mice by adenovirus-mediated gene delivery of the low-density-lipoprotein receptor, but not by the very-low-density-lipoprotein receptor. Biochem J. 1999; 338: 281–287.
- ↵Jong MC, Voshol PJ, Muurling M, Dahlmans VE, Romijn JA, Pijl H, Havekes LM. Protection from obesity and insulin resistance in mice overexpressing human apolipoprotein C1. Diabetes. 2001; 50: 2779–2785.
- ↵Goudriaan JR, Tacken PJ, Dahlmans VE, Gijbels MJ, van Dijk KW, Havekes LM, Jong MC. Protection from obesity in mice lacking the VLDL receptor. Arterioscler Thromb Vasc Biol. 2001; 21: 1488–1493.
- ↵Zhang SH, Reddick RL, Piedrahita JA, Maeda N. Spontaneous hypercholesterolemia and arterial lesions in mice lacking apolipoprotein E. Science. 1992; 258: 468–471.
- ↵Jong MC, Dahlmans VE, Hofker MH, Havekes LM. Nascent very-low-density lipoprotein triacylglycerol hydrolysis by lipoprotein lipase is inhibited by apolipoprotein E in a dose-dependent manner. Biochem J. 1997; 328: 745–750.
- ↵Descamps O, Bilheimer D, Herz J. Insulin stimulates receptor-mediated uptake of apoE-enriched lipoproteins and activated alpha 2-macroglobulin in adipocytes. J Biol Chem. 1993; 268: 974–981.
- ↵Gianturco SH, Gotto AM Jr, Hwang SL, Karlin JB, Lin AH, Prasad SC, Bradley WA. Apolipoprotein E mediates uptake of Sf 100–400 hypertriglyceridemic very low density lipoproteins by the low density lipoprotein receptor pathway in normal human fibroblasts. J Biol Chem. 1983; 258: 4526–4533.
- ↵Nilsson-Ehle P, Schotz MC. A stable, radioactive substrate emulsion for assay of lipoprotein lipase. J Lipid Res. 1976; 17: 536–541.
- ↵Medh JD, Fry GL, Bowen SL, Ruben S, Wong H, Chappell DA. Lipoprotein lipase- and hepatic triglyceride lipase-promoted very low density lipoprotein degradation proceeds via an apolipoprotein E-dependent mechanism. J Lipid Res. 2000; 41: 1858–1871.
- ↵Zsigmond E, Fuke Y, Li L, Kobayashi K, Chan L. Resistance of chylomicron and VLDL remnants to post-heparin lipolysis in apoE-deficient mice: the role of apoE in lipoprotein lipase-mediated lipolysis in vivo and in vitro. J Lipid Res. 1998; 39: 1852–1861.
- ↵Merkel M, Kako Y, Radner H, Cho IS, Ramasamy R, Brunzell JD, Goldberg IJ, Breslow JL. Catalytically inactive lipoprotein lipase expression in muscle of transgenic mice increases very low density lipoprotein uptake: direct evidence that lipoprotein lipase bridging occurs in vivo. Proc Natl Acad Sci U S A. 1998; 95: 13841–13846.
- ↵Takahashi S, Kawarabayasi Y, Nakai T, Sakai J, Yamamoto T. Rabbit very low density lipoprotein receptor: a low density lipoprotein receptor-like protein with distinct ligand specificity. Proc Natl Acad Sci U S A. 1992; 89: 9252–9256.
- ↵Le Lay S, Krief S, Farnier C, Lefrere I, Le Liepvre X, Bazin R, Ferre P, Dugail I. Cholesterol, a cell size-dependent signal that regulates glucose metabolism and gene expression in adipocytes. J Biol Chem. 2001; 276: 16904–16910.
- ↵Yagyu H, Lutz EP, Kako Y, Marks S, Hu Y, Choi SY, Bensadoun A, Goldberg IJ. VLDL receptor deficient mice have reduced lipoprotein lipase activity: possible causes of hypertriglyceridemia and reduced body mass with VLDL receptor deficiency. J Biol Chem. 2002; 277: 10037–10043.
- ↵Salinelli S, Lo JY, Mims MP, Zsigmond E, Smith LC, Chan L. Structure-function relationship of lipoprotein lipase-mediated enhancement of very low density lipoprotein binding and catabolism by the low density lipoprotein receptor: functional importance of a properly folded surface loop covering the catalytic center. J Biol Chem. 1996; 271: 21906–21913.