Identification by Site-directed Mutagenesis of Three Arginines in Uncoupling Protein That Are Essential for Nucleotide Binding and Inhibition*

Primary regulation of uncoupling protein is mediated by purine nucleotides, which bind to the protein and allosterically inhibit fatty acid-induced proton transport. To gain increased understanding of nucleotide regulation, we evaluated the role of basic amino acid residues using site-directed mutagenesis. Mutant and wild-type proteins were expressed in yeast, purified, and reconstituted into liposomes. We studied nucleotide binding as well as inhibition of fatty acid-induced proton transport in wild-type and six mutant uncoupling proteins. None of the mutations interfered with proton transport. Two lysine mutants and a histidine mutant had no effect on nucleotide binding or inhibition. Arg83 and Arg182 mutants completely lost both the ability to bind nucleotides and nucleotide inhibition. Surprisingly, the Arg276 mutant exhibited normal nucleotide binding, but completely lost nucleotide inhibition. To account for this dissociation between binding and inhibition, we propose a three-stage binding-conformational change model of nucleotide regulation of uncoupling protein. We have now identified three nucleotides by site-directed mutagenesis that are essential for nucleotide interaction with uncoupling protein.

Primary regulation of uncoupling protein is mediated by purine nucleotides, which bind to the protein and allosterically inhibit fatty acid-induced proton transport. To gain increased understanding of nucleotide regulation, we evaluated the role of basic amino acid residues using site-directed mutagenesis. Mutant and wildtype proteins were expressed in yeast, purified, and reconstituted into liposomes. We studied nucleotide binding as well as inhibition of fatty acid-induced proton transport in wild-type and six mutant uncoupling proteins. None of the mutations interfered with proton transport. Two lysine mutants and a histidine mutant had no effect on nucleotide binding or inhibition. Arg 83 and Arg 182 mutants completely lost both the ability to bind nucleotides and nucleotide inhibition. Surprisingly, the Arg 276 mutant exhibited normal nucleotide binding, but completely lost nucleotide inhibition. To account for this dissociation between binding and inhibition, we propose a three-stage binding-conformational change model of nucleotide regulation of uncoupling protein. We have now identified three nucleotides by site-directed mutagenesis that are essential for nucleotide interaction with uncoupling protein.
Uncoupling protein (UCP) 1 from brown adipose tissue mitochondria mediates proton flux from cytosol back into the matrix, thereby producing heat (1,2). UCP contains an anion transport pathway that permits the charged fatty acid anion head group, to which the membrane is normally impermeable, to flip-flop in the membrane. The cycle is completed by flip-flop of the protonated fatty acid head group in the bilayer, resulting in energy-dissipating back flux of protons into the matrix (3). The only known regulation of the uncoupling cycle is provided by purine nucleotides, which bind to UCP and allosterically inhibit transport (4).
UCP does not contain a consensus nucleotide binding sequence, and relatively little is known about the location of the nucleotide binding site. Klingenberg's laboratory (5,6) has shown by chemical labeling that the base and ribose moieties are positioned near the matrix loop connecting the fifth and sixth transmembrane domains. Lysine and arginine residues have long been suspected of participation in nucleotide regulation, and Katiyar and Shrago (7) showed that phenylglyoxal abolished GDP binding to UCP. However, none of the residues that interact with the phosphate groups have been identified.
The introduction of a high-level expression system for UCP in Saccharomyces cerevisiae (8) opened the door to a study of these interactions using site-directed mutagenesis. Thus, we were able to show that mutation of a single arginine residue (Arg 276 ) abolished GDP inhibition without affecting UCP-mediated proton transport (9). We have now extended this study to Arg 83 , Arg 182 , Lys 268 and His 214 , which are positioned within the membrane-spanning helices of UCP, and Lys 72 , which lies outside the transmembrane domain (10). The mutant proteins were overexpressed in yeast (8,9), then purified and reconstituted into liposomes. Fatty acid-induced, UCP-mediated H ϩ transport, and its sensitivity to nucleotides, were assayed by fluorescence, and GDP binding was assayed using [ 3 H]GDP.
Each of the three arginine mutations caused complete loss of GDP inhibition of transport without affecting transport itself. This finding implies an extensive nucleotide interaction domain, because the three arginines are located on three different transmembrane segments. A surprising finding was that the Arg 276 mutant, despite complete loss of sensitivity to GDP inhibition (9), retained normal binding affinity for GDP. We propose a three-stage binding-conformational change model in which Arg 276 interacts with nucleotide after it is tightly bound, causing a conformational change that prevents transport of the fatty acid anion head group.
Site-directed Mutagenesis-A M13mp19 plasmid containing the rat UCP cDNA fragment was used for preparing the site-directed mutants by the method of Kunkel (11). Oligonucleotides were designed to alter UCP codons for Arg 83 (AGG) to Gln (CAG), Arg 182 (AGA) to Thr (ACA), Lys 72 (AAA) to Gln (CAA), Lys 268 (AAA) to Gln (CAA), and His 214 (CAT) to Gln (CAA). The SacI/SphI fragments from M13mp19 containing wild-type or mutated forms of UCP cDNA were subcloned into SacI/ SphI-cut pCGS110 Escherichia coli/S. cerevisiae shuttle vector. The cDNAs are under transcriptional control of the inducible GalI promoter. The S. cerevisiae strain JB516 (MATa, ura3, ade1, leu2, his4, gal ϩ ) was transformed by electroporation with the shuttle vector constructs and plated on uracil-lacking selective plates. The resulting yeast transformants were grown at 30°C in selective medium, and overexpression of UCP was induced by addition of 0.2% galactose (8).
Purification and Reconstitution of Mutant UCPs-Yeast cells were * This work was supported in part by Grant GM31086 (to K. D. G.) from the National Institutes of Health, Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ This work was in partial fulfillment of requirements for the Ph.D degree.
Assay of H ϩ Transport Mediated by Mutant UCPs-SPQ fluorescence is quenched by TES anion, but not by TES zwitterion, allowing measurements of changes in total acid concentration (14). SPQ (2 mM) trapped inside liposomes containing the mutant UCPs was used for assaying transport and GDP sensitivity of the transport, as before (9). The intravesicular medium contained TES (28.8 mM), EGTA (0.6 mM), and SO 4 (84.4 mM), pH 7.2, all TEA salts, and 1 mM GDP. The external medium, also at pH 7.2, contained TES (28.8 mM), and EGTA (0.6 mM) as TEA salts, and SO 4 (84.4 mM) as K ϩ salt. Proton efflux from the liposomes was initiated by adding 30 M laurate followed by 0.1 M valinomycin. Each proteoliposome preparation was calibrated to obtain the corresponding quench constant, and volume of the liposomes was estimated from the probe distribution. Protein content of liposomes was estimated by the Amido Black method (15).
Assay of GDP Binding to Mutant UCPs-We used a modified anionexchange method developed by Klingenberg et al. (16). A Pasteur pipette tip was sealed with packed glass wool and then filled with 30 mg of wet Dowex 1-X8, 200 -400 mesh (chloride form). 75 l of sample containing 6 -9 g of UCP was applied on top of the resin. Experimental medium contained [ 3 H]GDP in concentrations from 1 to 15 M and otherwise was the same composition as internal medium, described above. The sample was immediately chased with 2 ϫ 100 l of distilled H 2 O through the column with a bulb. The entire eluate was collected and subjected to scintillation counting. For control, 100 M nonradioactive GDP was included with each concentration of the radiolabeled nucleotide.

Effects of Arg and Lys Mutations on GDP Inhibition of Proton
Transport- Fig. 1 contains a typical GDP concentration dependence for inhibition of wild-type UCP (solid circles) and Arg 182 3 Thr mutant UCP (open circles). As shown in Table I, each of the arginine mutations completely abolished GDP inhibition, whereas the lysine and histidine mutants retained full GDP sensitivity. The K i for GDP inhibition of proton flux in wild-type and native UCP under these conditions is 11 M (9) and was not affected in the mutants exhibiting GDP inhibition.
Upon reconstitution, UCP is distributed randomly in the bilayer (17). It transports anions in both directions, but 1 mM GDP is present in the liposomal interior, so that 50% of the wild-type UCP is always inhibited. Loss of GDP regulation should therefore free up both orientations of the protein and cause doubling of transport activity. This expectation was met in the Arg 83 3 Gln mutant, but not in the other two Arg mutants.
Undecanesulfonate anion is transported by UCP and is a competitive inhibitor of laurate-induced proton flux (3). 300 M undecanesulfonate inhibited proton flux in all mutants to the level of proton leak, consistent with normal behavior of the fatty acid anion transport pathway in the mutant UCPs.
Effects of Arg and Lys Mutations on GDP Binding-The concentration dependence of GDP binding yielded linear massaction plots, from which K d and B max values were obtained. These parameters are contained in Table II. In comparing the data in Tables I and II, there is agreement between GDP binding and inhibition in each case except one. Thus, the Arg 83 3 Gln and Arg 182 3 Thr mutants lost both GDP inhibition and GDP binding, whereas the Lys 72 3 Gln and Lys 268 3 Gln mutants retained both functions. The Arg 276 3 Leu mutant is unique in that it binds GDP normally but cannot be inhibited by GDP. Retention of normal GDP binding by the Arg 276 3 Leu mutant has been confirmed in isolated yeast mitochondria (18). The protocol used to assay binding captures only the tightly bound state (19); thus, GDP is tightly bound in the Arg 276 mutant.
The dissociation constants, K d , of the mutant UCPs capable of binding GDP (Table II) are higher than published values pertaining to native rat UCP (20); however, the K d values are close to the observed K i for GDP inhibition of UCP-mediated transport, which is generally 11 M (9). This agreement may reflect our use of identical conditions for H ϩ flux and binding measurements.
Lysine 268 is located within the third domain of the tripartite structure among residues 238 -283, which have long been identified with nucleotide interactions (6), and Lys 72 is located  [GDP]. The data shown were obtained with wild-type (q) and Arg 182 3 Thr (E) mutant in the presence of 30 M laurate. The K i for GDP inhibition of wild-type UCP was 11 M. The difference between the maximum flux ("100% flux") and the minimum flux ("0% flux") corresponds to rates shown in Table I.  Tables I and II, neither residue affected GDP binding or GDP inhibition. The Role of His 214 in pH Regulation of Nucleotide Inhibition-The pH dependence of nucleoside triphosphate binding to UCP has a pK a of 7.2, which suggested involvement of a histidine residue (21). If His 214 plays this role, the His 214 3 Gln mutation should exhibit an increased K i for GTP inhibition at neutral pH. In fact, this mutation had no effect on the K i for GTP inhibition at pH 7.2 or 7.6 when compared with the wild-type (three experiments, data not shown).

DISCUSSION
Because of their polyanionic character, nucleotides are expected to bind to positively charged amino acid residues in UCP. Arg 276 was the first such residue to be identified (9), and we now show that Arg 83 and Arg 182 are also essential for normal nucleotide binding and inhibition. None of the mutations affected H ϩ transport, confirming that fatty acid anion transport and nucleotide inhibition are separate features of UCP and implying two distinct binding sites (4,7).
Structural Implications for UCP and UCPH-The location of the three essential arginines implies an extensive interaction region involving at least three of the six transmembrane helices in UCP (Fig. 2). These helices very likely form the walls of a dead-end aqueous pocket for nucleotides, which are known to enter and exit from the cytosolic side and to interact with residues located on the matrix loop between helices 5 and 6 (6). Moreover, each of the three arginines is conserved in UCPH, suggesting that this protein is also regulated by nucleotides. UCPH, whose primary amino acid sequence is 56% identical with that of UCP, encodes a protein thought to be involved in prevention of obesity and diabetes (22,23).
Proposed Three-stage Mechanism of Nucleotide Regulation-Klingenberg and co-workers (19,21,24,25) have elaborated an elegant kinetic model of the interactions of nucleotide phosphates with UCP. Access to the phosphate binding region is controlled by protonation of Glu 190 , presumably by releasing a salt bridge that occludes the binding site. Huang and Klingenberg (19) proposed a two-stage mechanism in which nucleotides are first loosely bound and then the protein undergoes a slow, spontaneous conformational change to a "tight" state in which nucleotides are tightly bound and H ϩ transport is inhibited. Our results show, however, that the tightly bound state is not sufficient for transport inhibition. Thus, the tight binding exhibited by the Arg 276 mutant is ineffective in inhibiting H ϩ transport. Thus, the two-stage model (19) requires modifications, as described below and in Fig. 3.
Loose Conformation-The nucleotide enters the aqueous pocket in UCP, and its sugar-base moiety binds to the matrix loop connecting the fifth and sixth ␣-helices (5,6). This interaction may be responsible for discrimination between purine and pyrimidine nucleotides (20,26,27). The nucleotide ␤-phosphate binds to Arg 182 , and the sum of these interactions results in a loose binding conformation.
Tight Conformation-The switch to a tight binding conformation is initiated by protonation of Glu 190 , making Arg 83 available for binding to the second charge on the ␤-phosphate of diphosphates and the ␥-phosphate of triphosphates. A histidine binds to the second charge on the ␥-phosphate of triphosphates (24). The tightly bound conformation prevents trypsin cleavage of the C terminus of UCP at Lys 292 (19). This step is necessary, but not sufficient, for inhibition of transport.
Inhibited Conformation-The nucleotide is now bound stably and positioned so that the ␣-phosphate can bind to Arg 276 . This  (29). Amino acids at the cytosolic and matrix ends of the ␣-helices are numbered based on models of Klingenberg (1) and Winkler et al. (25). Lys 72 and Lys 268 were mutated without effect (this study). The remaining designated residues have been shown to participate in nucleotide binding to UCP. Thr 259 , Thr 264 , and Cys 253 interact with the sugar-base moiety of the nucleotide (25). Arg 83 , Arg 182 , and Arg 276 are essential for nucleotide binding and/or inhibition (this study), as is Glu 190 (25). Each of the four transmembrane residues, Arg 83 , Arg 182 , Arg 276 , and Glu 190 , is conserved in UCPH, as is Cys 253 . UCPH contains substitutions at position 259 (Gln instead of Thr) and 264 (Arg instead of Thr).

FIG. 3. Model of the phosphate binding pocket in UCP.
A, loose conformation. The nucleotide sugar-base moiety binds loosely to the loop between transmembrane helices 5 and 6, and the terminal phosphate binds to Arg 182 on helix 4. Binding to Arg 83 is prevented by a salt bridge from Glu 190 to a neighboring Arg or Lys. The figure tentatively identifies this residue as Arg 83 itself. B, tight conformation/inhibited conformation. Protonation of Glu 190 opens the phosphate binding cleft between helices 2 and 4 and frees Arg 83 to bind to the terminal phosphate of nucleoside diphosphates, resulting in a tightly bound state. Tight binding pulls the bound nucleotide, together with its sugar-base attachments at the N terminus of helix 6, into a new position, causing a demonstrable conformational change at the C terminus of helix 6 (19). Tight binding is not sufficient to inhibit transport. This occurs when Arg 276 binds to the ␣-phosphate, causing a further conformational change that occludes the intramembrane binding site for the anionic head group of fatty acids. final interaction induces a conformational change that occludes or removes the internal fatty acid binding site (3,28), causing inhibition of transport. This extension of Huang and Klingenberg's model (19) is consistent with our finding that mutation of Arg 276 to Leu did not affect the K d for binding but completely prevented nucleotide inhibition of transport.