The Pleckstrin Homology (PH) Domain of the Arf Exchange Factor Brag2 Is an Allosteric Binding Site*

Background: Brag2 is a PH domain-containing Arf guanine nucleotide exchange factor (GEF) that regulates cell adhesion. Results: PIP2 association with the PH domain stimulated Brag2 activity. Regulation was dependent on the N terminus of Arf and independent of the N-terminal myristate. Conclusion: PIP2 binding to the PH domain allosterically modifies Brag2 activity. Significance: A novel regulatory mechanism for GEFs was identified. Brag2, a Sec7 domain (sec7d)-containing guanine nucleotide exchange factor, regulates cell adhesion and tumor cell invasion. Brag2 catalyzes nucleotide exchange, converting Arf·GDP to Arf·GTP. Brag2 contains a pleckstrin homology (PH) domain, and its nucleotide exchange activity is stimulated by phosphatidylinositol 4,5-bisphosphate (PIP2). Here we determined kinetic parameters for Brag2 and examined the basis for regulation by phosphoinositides. Using myristoylated Arf1·GDP as a substrate, the kcat was 1.8 ± 0.1/s as determined by single turnover kinetics, and the Km was 0.20 ± 0.07 μm as determined by substrate saturation kinetics. PIP2 decreased the Km and increased the kcat of the reaction. The effect of PIP2 required the PH domain of Brag2 and the N terminus of Arf and was largely independent of Arf myristoylation. Structural analysis indicated that the linker between the sec7d and the PH domain in Brag2 may directly contact Arf. In support, we found that a Brag2 fragment containing the sec7d and the linker was more active than sec7d alone. We conclude that Brag2 is allosterically regulated by PIP2 binding to the PH domain and that activity depends on the interdomain linker. Thus, the PH domain and the interdomain linker of Brag2 may be targets for selectively regulating the activity of Brag2.


Brag2, a Sec7 domain (sec7d)-containing guanine nucleotide exchange factor, regulates cell adhesion and tumor cell invasion.
Brag2 catalyzes nucleotide exchange, converting Arf⅐GDP to Arf⅐GTP. Brag2 contains a pleckstrin homology (PH) domain, and its nucleotide exchange activity is stimulated by phosphatidylinositol 4,5-bisphosphate (PIP 2 ). Here we determined kinetic parameters for Brag2 and examined the basis for regulation by phosphoinositides. Using myristoylated Arf1⅐GDP as a substrate, the k cat was 1.8 ؎ 0.1/s as determined by single turnover kinetics, and the K m was 0.20 ؎ 0.07 M as determined by substrate saturation kinetics. PIP 2 decreased the K m and increased the k cat of the reaction. The effect of PIP 2 required the PH domain of Brag2 and the N terminus of Arf and was largely independent of Arf myristoylation. Structural analysis indicated that the linker between the sec7d and the PH domain in Brag2 may directly contact Arf. In support, we found that a Brag2 fragment containing the sec7d and the linker was more active than sec7d alone. We conclude that Brag2 is allosterically regulated by PIP 2 binding to the PH domain and that activity depends on the interdomain linker. Thus, the PH domain and the interdomain linker of Brag2 may be targets for selectively regulating the activity of Brag2.
Arf-directed guanine nucleotide exchange factors (Arf-GEFs) 2 catalyze the exchange of nucleotide on Arf family GTPbinding proteins (1)(2)(3). 15 ArfGEFs have been identified in the human genome and divided into five classes: Big1/2 and Golgi specific Brefeldin A resistant guanine nucleotide exchange fac-tor, ARNO/Grp1/cytohesin, EFA6, Brag, and Fbx8. The Arf-GEFs of the Big1/2 and Golgi specific Brefeldin A resistant guanine nucleotide exchange factor family, common to fungi, plants, and metazoa, regulate membrane traffic. The other classes of ArfGEFs are found only in metazoans. The Brag family of ArfGEFs, including Brag2, has been implicated in peripheral membrane traffic and in cell adhesion and migration (4,5). Brag2 has been reported to signal through Arf6 to drive breast cancer invasion (4,6,7).
Brag2, also called GEP100 and IQSEC1, is a Ϸ100-kDa protein that contains IQ-like, proline-rich, Sec7, and pleckstrin homology (PH) domains (7) (see Fig. 1B). Brag2 activates Arf6 to regulate cell-substrate and cell-cell adhesion (4 -6, 8, 9). Activity has been found to be stimulated by nonphosphorylated peptides from AMPA receptor (10) and phosphopeptides from epidermal growth factor receptor (4). The phosphopeptides from epidermal growth factor receptor are reported to bind to the PH domain of Brag2 to regulate its activity. These observations have been used to explain the effect of epidermal growth factor receptor on cancer cell invasion. Stimulation of Brag2 increases Arf6⅐GTP levels, which drive the cellular changes responsible for movement of the cancer cells into the normal tissue. Brag2 activity is also regulated by phosphoinositides that presumably bind the PH domain (9). Together, these findings suggest that the PH domain may represent a regulatory motif.
PH domain-mediated regulation of one subtype of ArfGEFs, cytohesin/Grp/ARNO, has been characterized. ARNO GEF activity is autoinhibited by the linker region between the Sec7 and PH domains and a C-terminal amphipathic helix containing a polybasic motif, which physically block the Arf binding site. Binding of Arf6⅐GTP and phosphoinositides to the PH domain has two functions. One is to recruit ARNO to the membrane surface on which it is active, and the second is to induce a conformational change in the PH domain that relieves autoinhibition. Phosphorylation of serines and threonine within a polybasic motif of cytohesin-1 by protein kinase C (PKC) also relieves autoinhibition (11,12). Brag2 does not have a sequence similar to that in ARNO responsible for autoinhibition; therefore, the findings from ARNO may not extrapolate to Brag2. Thus, we aimed to dissect the molecular mechanism control-ling Brag2 activity with the ultimate goal of using it as a therapeutic target for anticancer cell invasion therapy.
Here we first determined the effect of PIP 2 on the fundamental enzymatic parameters k cat and K m for Brag2. PIP 2 increased the ratio k cat /K m from 2.6 ϫ 10 5 to 8.8 ϫ 10 6 M Ϫ1 s Ϫ1 . We then analyzed structural requirements for GEF activity. Unique to Brag2, the linker between the Sec7 and PH domains had a positive effect on activity, and the effect of PIP 2 required the N terminus of Arf. These unique features of Brag2 can be used to selectively inhibit Brag2 activity.

Expression Vectors
A mammalian expression vector for Brag2-myc-His and a bacterial expression vector for GST-Brag2-His were kindly provided by Joel Moss (National Heart, Lung, and Blood Institute). Bacterial expression vectors for His-Brag2 Sec7-PH (amino acids 499 -863), His-Brag2 Sec7-linker (amino acids 499 -740), and His-Brag2 Sec7 (amino acids 499 -700) were generated by standard PCR methods using pET19 as an expression vector (EMD Biosciences). Mutants in the PH domain of His-Brag2 Sec7-PH (K753S,K756S and R762S) were generated using the QuikChange II site-directed mutagenesis kit (Agilent Technologies).

Brag2 GEF Activity
The conversion of Arf⅐GDP to Arf⅐GTP was followed in one of three ways.
Fixed Time Point Assay for Determination of C 50 -Brag2catalyzed GTP␥S binding to Arf⅐GDP was measured using nucleotide exchange buffer (25 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM dithiothreitol, 2 mM MgCl 2 , 1 mM EDTA, 1 mM ATP, 5 M GTP␥S, and [ 35 S]GTP␥S (for specific activity of ϳ10,000 cpm/pmol) as described (9,18,20,21). High [MgCl 2 ] was used in this reaction to slow down the spontaneous nucleotide exchange. The reactions also contained 0.5 mM LUVs and 0.5 M Arf⅐GDP with different concentrations of Brag2. The reactions were incubated at 30°C for 3 min and terminated with 2 ml of ice-cold 20 mM Tris, pH 8.0, 100 mM NaCl, 10 mM MgCl 2 , and 1 mM dithiothreitol. Protein-bound nucleotide was trapped on nitrocellulose, and the bound radioactivity was quantified by liquid scintillation counting.
Substrate Saturation Experiments-Brag2 GEF activity was determined under conditions satisfying the steady state assumption using a FluorMax3 spectrophotometer (Jobin Yvon Horiba, Edison, NJ). The conversion of Arf1⅐GDP to Arf1⅐GTP was monitored by fluorescence (excitation, 297 nm; emission, 340 nm). Arf1⅐GTP has a greater emission than Arf1⅐GDP; therefore, the conversion results in an increase in fluorescent signal. The reaction contained 25 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM dithiothreitol, 2 mM MgCl 2 , 1 mM EDTA, and 0.5 mM LUV with or without 1% PIP 2 . When GTP␥S was the substrate being varied, 0.1 nM His-Brag2 Sec7-PH , 5 M myrArf1 (saturating concentration), and 0.5-100 M GTP␥S were included. When Arf1⅐GDP or Arf6⅐GDP was the substrate being varied, 0.1 (for LUVs containing 1% PIP 2 ) or 0.5 nM (for LUVs lacking PIP 2 ) His-Brag2 Sec7-PH , 100 M GTP␥S, and different concentrations of myrArf1 or myrArf6 were included.
Single Turnover Assay-Single turnover analyses were performed using an SF-2004 stopped flow instrument (KinTek Corp., Austin, TX). MyrArf1 or myrArf6 preloaded with mant-GDP was rapidly mixed with an equal volume of His-Brag2 Sec7-PH . To load myrArf1 (or myrArf6) with mantGDP, 0.2 M myrArf1 (or 0.4 M myrArf6) was incubated at 30°C for 1-2 h in 25 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM dithiothreitol, 0.5 mM MgCl 2 , 1 mM EDTA, 5 M mantGDP, and 0.5 mM LUV with or without 1% (5 M) PIP 2 . At the end of the incubation, MgCl 2 was added to reach a final concentration of 2 mM. His-Brag2 Sec7-PH was in the same buffer as for loading except 2 mM MgCl 2 and 200 M GTP were present and mantGDP was not. The conversion of Arf⅐mantGDP to Arf⅐GTP was monitored by a FRET signal resulting from resonance energy transfer from tryptophan in Arf to the methylanthronoyl group on GDP. The excitation wavelength was set at 297 nm, and the photomultiplier tube filter cutoff was 400 nm. Arf⅐mantGDP has a FRET signal, whereas Arf⅐GTP does not; therefore, the conversion results in a decrease in fluorescent signal.

Modeling
Secondary structure prediction was done using a consensus of different prediction programs (22). A docking calculation of the putative linker helix to the complex of Brag2 Sec7 domain with myrArf was performed using the EMAP program of CHARMM (23,24). The myrArf was created by superposing the NMR structure of myrArf (Protein Data Bank code 2K5U (25)) onto the structure of (⌬17)Arf1 bound to ARNO1 Sec7 domain (Protein Data Bank code 1RE0 (26)). Homology modeling was done using Prime (Schrödinger Inc., New York, NY) to replace the ARNO1 sequence with that of Brag2. The random coil portion of the Brag2 linker was docked interactively using MacroModel (Schrödinger Inc.), and the Sec7 domain, putative linker helix, and random coil region were linked using Prime.

Miscellaneous Methods
Protein concentration was estimated using the Bio-Rad dye binding assay. Arf concentration was determined by GTP␥S binding as described (27). Graphs were prepared and parameter estimates were obtained using GraphPad Prism. Alignments based on crystal structures were performed using the Protein Structure Alignment tool in Maestro (Schrödinger Inc.). Experiments with myrArf6 are described in the supplemental Methods.

General Experimental Considerations for the Kinetics and Biochemistry of Arf and Arf-directed Guanine Nucleotide
Exchange Factors-Arfs are myristoylated proteins (referred to as myrArf here) that bind to GDP with pM affinities (21,29). Nucleotide exchange on myrArf requires that nucleotide evacuates the binding site on myrArf followed by nucleotide binding to the empty site. MyrArf is not stable without nucleotide and precipitates from solution. The function of the exchange factor is to accelerate nucleotide dissociation and stabilize the empty form of Arf. We consider the reaction a simplified Ping Pong Bi Bi (30) (Fig. 1A) based on available biochemical and crystallographic data (31)(32)(33). Bi Bi refers to two substrates and two products. Ping Pong refers to a mechanism in which the enzyme cannot bind two substrates simultaneously. One substrate binds to the enzyme; the enzyme releases the first product and is then able to bind to the second substrate. In this scheme, the first substrate, Arf⅐GDP, binds to the GEF, releasing the product, GDP, and leaving the complex of GEF-empty Arf. We consider this a second form of the enzyme to which the second substrate, GTP, binds. The product, Arf⅐GTP, is released, generating the initial form of the enzyme.
The role of a membrane surface was considered in examining Brag2 activity. Neither of the Brag2 substrates, myrArf⅐GDP and GTP, is membrane-restricted. Therefore, surface dilution does not have to be considered when examining initial rates. However, the enzyme⅐substrate complex (i.e. Brag2⅐myrArf) and the product myrArf⅐GTP are tightly associated with membranes, and myrArf⅐GTP cannot accumulate without a hydrophobic surface. We provided LUVs as the surface.
The experiments presented here used myrArf1⅐GDP and mutants of Arf1 as substrates for Brag2. MyrArf6 was also used for substrate saturation and single turnover experiments with results that were very similar to those obtained with myrArf1. Because Brag2 has been reported to be an Arf6 exchange factor, we provide the data for myrArf6 in supplemental Results and Discussion, Figs. S1-S3, and Tables S1-S3 to provide a documented comparison with Arf1. The data do not add significantly to the test of the hypothesis that PIP 2 binding to the PH domain regulates GEF activity and, therefore, are not included in the main text.
Brag2 Fragment Comprising Sec7 and PH Domains (His-Brag2 Sec7-PH ) as a Model for Studying Regulation of Brag2-We sought a recombinant form of Brag2 to examine phosphoinositide dependence. Although we have prepared full-length recombinant Brag2 suitable for some biochemical analyses, at this time, we are not able to obtain a sufficiently pure, homogeneous, stable full-length Brag2 of adequate concentration for the planned kinetic experiments. We were able to prepare a protein comprising the Sec7 and PH domains with a His 10 tag fused to the N terminus (His-Brag2 Sec7-PH ). We first determined whether His-Brag2 Sec7-PH was activated by PIP 2 to a similar extent as full-length Brag2. Two preparations of full-length Brag2 were used for the comparison: (i) Brag2 expressed in bacteria as a GST fusion protein that also contained a His 6 tag (GST-Brag2-His) and (ii) Brag2 with myc and His 6 tags on the C terminus (Brag2-myc-His) expressed in and purified from mammalian cells (see Fig. 1B for schematic of recombinant Brag2 proteins used in these experiments). The concentration of the full-length Brag2 was estimated by comparing the intensity of staining with standards run on the same polyacrylamide gel.
To determine the relative effect of PIP 2 on exchange factor activity, the Brag2 recombinant proteins were titrated into reactions containing myrArf1⅐GDP (0.5 M), GTP␥S (5 M), and LUVs with or without PIP 2 ( Fig. 1C and Table 1). The concentration of Brag2 that resulted in 50% exchange of nucleotide on Arf (we call this concentration the C 50 ), which is roughly proportional to the inverse of enzymatic power (28), was determined. All preparations of Brag2, including His-Brag2 Sec7-PH , had 15-20-fold more specific activity in the presence of PIP 2 than in its absence. His-Brag2 Sec7-PH was 6 -10fold more active than either full-length Brag2. Given the similar effects of PIP 2 on the activity of full-length Brag2 and His-Brag2 Sec7-PH , His-Brag2 Sec7-PH was used for our subsequent work aimed at understanding the mechanisms by which PIP 2 binding to Brag2 stimulates GEF activity.
Effect of PIP 2 on Kinetic Parameters of His-Brag2 Sec7-PH -We determined the effect of PIP 2 on kinetic parameters for His-Brag2 Sec7-PH using myrArf1⅐GDP as the substrate. The analysis for myrArf1⅐GDP is simplified if analyzed with saturating concentrations of the second substrate, GTP (we used GTP␥S, an analog of GTP that is slowly hydrolyzed) (see supplemental Appendix for equations used for analysis). To establish the necessary concentration of GTP␥S, we determined the enzymatic parameters with GTP␥S as the varied substrate (Fig. 2). Arf1⅐GDP was fixed at a saturating concentration (Ͼ20 K m ; see next paragraph). The reaction was followed continuously by measuring tryptophan fluorescence, which increases when Arf switches from the GDP-to GTP-bound forms (27). Initial rates were estimated and plotted against the concentration of GTP␥S, and the data were fit to a Michaelis-Menten equation. The K m,GTP␥S was 1 M. The k cat (calculated from the V max ) was 8/s ( Table 2). 100 M GTP␥S was used for subsequent experiments examining the effect of PIP 2 on exchange.
The effect of PIP 2 on the kinetic parameters using myrArf1⅐GDP as the varied substrate was determined ( Fig. 3 and Table 2). In these experiments, the dependence of the initial velocity of the exchange reaction on myrArf1⅐GDP concentration was determined, and the results were analyzed using the Michaelis-Menten equation. In the presence of PIP 2 , the K m,Arf1⅐GDP was 0.2 M, and k cat was 8/s (calculated from the V max ) ( Fig. 3 and Table 2). In the absence of PIP 2 , the K m was 2.2 M, 11-fold greater than in the presence of PIP 2 . More His-Brag2 Sec7-PH was used for experiments in the absence of PIP 2 than in the presence, so the V max was greater than the V max determined in the presence of PIP 2 . The calculated k cat , which is V max /[His-Brag2 Sec7-PH ], was 6.4/s, which is similar to that determined in the presence of PIP 2 .
We also determined the enzymatic parameters using single turnover experiments. In these experiments, the complex Arf1⅐mantGDP was used as a substrate, which was detected as resonance energy transfer from the tryptophans in Arf1 to the methylanthronoyl group on GDP. Nucleotide dissociation was detected as the loss of resonance energy transfer. The rate of nucleotide dissociation with increasing concentrations of His-Brag2 Sec7-PH was measured using a stopped flow instrument. At saturating concentrations of His-Brag2 Sec7-PH , the observed rate (k obs ) is equal to the k cat ; thus, the single turnover approach has the advantage that k cat is determined directly. The concentration of His-Brag2 Sec7-PH at which the observed rate is 1 ⁄ 2 of the k cat (we call this the Brag2 50 ) is between 1 ⁄ 2K m and K m . We  found that Brag2 50 for Arf1 was 15-fold greater in the absence than in the presence of PIP 2 . The value of k cat in the absence of PIP 2 was 1 ⁄ 3 the k cat determined in the presence of PIP 2 ( Fig. 4 and Table 3), consistent with the notion that PIP 2 acts as an allosteric modifier of Brag2. The efficiency of an enzyme (also called enzymatic power) is expressed as the ratio k cat /K m . To obtain estimates of k cat and K m , we used both substrate saturation and single turnover experiments. We are most confident in the K m determined by substrate saturation because the Arf concentration was determined by titrating GTP binding sites. The concentration of Brag2 was estimated using a dye binding assay and, therefore, may not accurately represent molar mass. We are most confident in the k cat determined from single turnover studies because this is a direct measurement. In substrate saturation studies, the k cat is calculated from V max ϭ k cat ⅐[Brag2]. An error in estimating Brag2 concentration would be propagated to the calculation for k cat . Using the values in which we are most confident, we calculate a k cat /K m of 2.6 Ϯ 0.9 ϫ 10 5 M Ϫ1 s Ϫ1 in the absence of PIP 2 and 8.8 Ϯ 3.1 ϫ 10 6 M Ϫ1 s Ϫ1 in the presence of PIP 2 . There is a 34-fold increase in activity due to PIP 2 .
The PH Domain of His-Brag2 Sec7-PH Is Necessary for Robust GEF Activity-The activities of recombinant proteins comprising the Sec7 (His-Brag2 Sec7 ), Sec7-linker (His-Brag2 Sec7-linker ), and Sec7-PH (His-Brag2 Sec7-PH ) domains (schematically represented in Fig. 5A) were compared to examine the role of the PH domain in Brag2 GEF activity. MyrArf1⅐GDP was used as the substrate, and LUVs containing PIP 2 were present to stabilize the product of the reaction, myrArf1⅐GTP␥S. In the experiment presented in Fig. 5B, more than 30% of the myrArf1 exchanged nucleotide when incubated with 0.14 nM His-Brag2 Sec7-PH , whereas 100 nM Sec7-linker (His-Brag2 Sec7-linker ) induced 20% exchange, and 100 nM Sec7 alone (His-Brag2 Sec7 ) induced 10% exchange. Based on these rates, His-Brag2 Sec7-PH had more than 700-fold greater activity than either His-Brag2 Sec7-linker or His-Brag2 Sec7 . In the absence of PIP 2 , His-Brag2 Sec7-PH had more than 50-fold the activity of either His-Brag2 Sec7-linker or His-Brag2 Sec7 (not shown). These results indicate that the PH domain is a critical regulator of Brag2 activity.  Table 2. Representative experiments of two to four are shown. A, myrArf1⅐GDP as substrate in the presence of PIP 2 . B, myrArf1⅐GDP as substrate in the absence of PIP 2 .

TABLE 3 Kinetic parameters determined from single turnover experiments
Single turnover kinetics were examined as described under "Experimental Procedures" and in Fig. 5 To test the idea that PIP 2 binding to the PH domain regulates GEF activity, recombinant proteins with changes in residues predicted to bind PIP 2 were examined. The Brag2 PH domain does not align well with typical PIP 2 or phosphatidylinositol 1,4,5-trisphosphate binding PH domains, such as from ARNO (see alignment in Fig. 5C). However, a crystal structure of the PH domain of Brag2 is available (MMDB accession number 89889). Assuming the loop between strands ␤1 and ␤2 (in Fig.  5C, loop residues are indicated by a "C" and ␤ strand residues are indicated by an "E" above the sequence) contains the PIP 2 binding site (double underlined region in sequence), mutations were introduced into the PH domain of His-Brag2 Sec7-PH (Fig.  5C, highlighted in blue). Two constructs were generated, one with serine substitutions of lysines at positions 753 and 756 (K753S,K756S) and another with a serine substitution of arginine at position 762 (R762S). Both mutants had GEF activity when assayed in the presence of 5 M PIP 2 . His-(R762S)Brag2 Sec7-PH had less activity than wild type His-Brag2 Sec7-PH (Fig. 5D and Table 4).
The mutants were used to correlate PIP 2 binding to Brag2 activation. We first determined binding to LUVs containing variable concentrations of PIP 2 to determine relative affinities. The mutant His-(K753S,K756S)Brag2 Sec7-PH bound PIP 2 in LUVs less tightly than the wild type protein (Fig. 5E). We did not detect binding of His-(R762S)Brag2 Sec7-PH to PIP 2 -containing LUVs. The effect of the mutations on binding to LUVs correlated with the effect of PIP 2 on Brag2 activity. PIP 2 -stimulated activity was detected for His-(K753S,K756S)Brag2 Sec7-PH , but the PIP 2 dependence was shifted to the right. At 5 M PIP 2 (the condition used to determine whether the proteins had activity in Fig. 5D), His-(K753S,K756S)Brag2 Sec7-PH had ϳ80% the  activity of wild type protein. PIP 2 had no effect on the activity of His-(R762S)Brag2 Sec7-PH (Fig. 5F). These findings indicate that the PH domain and PIP 2 binding to the PH domain are required for maximal Brag2 activity.

PH Domain and Interdomain Linker Promote Activity of Sec7 Domain in His-Brag2 Sec7
-PH -We next tested the idea that PIP 2 binding to the PH domain relieved autoinhibition. The function of the PH domain in controlling autoinhibitory motifs within ARNO was uncovered using recombinant Arf lacking the N terminus ((⌬17)Arf1) (11), which does not require a hydrophobic surface for nucleotide exchange (21). We used a similar approach to examine the possibility of autoinhibition in Brag2.
The predictions for the autoinhibition model are that His-Brag Sec7 should be more active than His-Brag2 Sec7-PH using (⌬17)Arf1 as a substrate in the absence of phosphoinositides and that the activity of His-Brag2 Sec7-PH would be increased by PIP 2 and possibly by a soluble PIP 2 analog. We compared activities of His-Brag2 Sec7 , His-Brag2 Sec7-linker , and His-Brag2 Sec7-PH (Fig. 5A shows a schematic of the proteins) using (⌬17)Arf1 as a substrate. In contrast to the predictions of the autoinhibition model, we found that His-Brag2 Sec7-PH and His-Brag2 Sec7-linker were more active than His-Brag2 Sec7 (Fig. 6C and Table 5) and that LUVs with PIP 2 had no effect on the activity of any of these recombinant Brag2 proteins when using (⌬17)Arf1 as a substrate (Fig. 6C). Also in contrast to the prediction of the autoinhibition mechanism, a soluble analog of PIP 2 did not affect activity (Fig. 6D). We conclude that PIP 2mediated regulation of Brag2 activity does not involve rearrangements that relieve inhibition. Instead, the linker contributes to Brag2 activity.
The result that Brag2 is not regulated in the same manner as ARNO is consistent with the lack of homology in the regions responsible for autoinhibition in ARNO (Figs. 5C and 6A). In the sequence shown in Fig. 5C, the polybasic motif in ARNO that is necessary for autoinhibition (11) is underlined. There is little similarity to Brag2. In Fig. 6A, the linker following the Sec7 is underlined. This motif, which contributes to autoinhibition in ARNO (11), also has little similarity to Brag2.
The greater activity of the Brag2 recombinant proteins with the linker could be due to interaction of the linker with Arf1. The Sec7 domain comprises 10 ␣ helices, A through J, with a prominent hydrophobic groove in which Arf binds. Motifs immediately C-terminal of helix J are near the hydrophobic groove and have been previously reported to interact with Arf1 (38). A consensus of secondary structure prediction programs predicted that the first part of the Brag2 linker region, residues 704 -716, should form a helix (Fig. 6B). This putative helix is amphipathic. A docking calculation of the putative helix to the Brag2⅐myrArf complex was performed, testing all possible locations of the helix placed immediately C-terminal of Brag2 helix J. Many possible docking sites were predicted, including one that also interacted with switch 1 of myrArf (Fig. 6B). Such an interaction could explain why the Brag2 constructs containing the linker are more active; however, further experiments would be needed to prove this linker/myrArf interaction.
The Amino Terminus of Arf Is Necessary for Regulated Activity-The lack of effect of PIP 2 on the activity of His-Brag2 Sec7-PH when using (⌬17)Arf1 as a substrate was evidence that regulation was not through autoinhibition. Alternative mechanisms include PIP 2 concentrating Brag2 together with myrArf1⅐GDP on a membrane or PIP 2 stimulating interaction of the N terminus of Arf1 directly with Brag2. In initial experiments to distinguish between these possibilities, we studied the Arf mutant (L8K)Arf1 as a substrate. This mutant is not myristoylated, eliminating the possibility that the myristate could account for the observed effects. In addition, (L8K)Arf1 binds GTP independently of lipids, and neither (L8K)Arf1⅐GDP nor (L8K)Arf1⅐GTP binds to lipids (13,14). Therefore, effects of the membrane on activity could be separated from effects of the membrane on product accumulation.
The exchange on (L8K)Arf1 catalyzed by His-Brag2 Sec7-PH , His-Brag2 Sec7-linker , and His-Brag2 Sec7 in the absence of LUVs with PIP 2 was examined. There was little exchange using any of these forms of Brag2 compared with the rate observed with (⌬17)Arf1 (Fig. 6E). Activity of His-Brag2 Sec7-PH but not of His-Brag2 Sec7-linker or His-Brag2 Sec7 increased at least 40-fold by including LUVs with PIP 2 and was about 3-fold more efficient with (L8K)Arf1 than with (⌬17)Arf1 ( Fig. 6E and Table 6). Soluble PIP 2 had no effect on activity (Fig. 6F). In short, the N terminus of Arf is required for activation of Brag2 by PIP 2 .
PIP 2 -dependent Recruitment of Brag2 and Arf to Membranes Does Not Completely Account for Regulation-Both Brag2 and Arf1 have been reported to bind PIP 2 . To further test the idea that concentrating the two proteins on a surface may account for at least part of the effect of PIP 2 on Brag2 activity, we directly measured binding of both proteins to LUVs. In these experiments, LUVs containing increasing concentrations of PIP 2 were incubated with either His-Brag2 Sec7-PH , myrArf1, or both. The LUVs were separated from bulk solution, and the bound proteins were measured. His-Brag2 Sec7-PH bound to the LUVs with a K d of 2 M for PIP 2 (Fig. 7, A and B). In contrast, PIP 2 had little effect on myrArf1 binding to LUVs (Fig. 7, C and D). When His-Brag2 Sec7-PH and myrArf1 were incubated together, both proteins bound efficiently to the LUVs in the absence of PIP 2 . PIP 2 had a small effect on binding when the two proteins were incubated together, but the increase was not sufficient to account for the 30 -40-fold change in activity (Fig. 7, A-D). (⌬17)Arf1 and (L8K)Arf1 were also examined. Neither bound efficiently to LUVs, and neither PIP 2 nor Brag2 increased their association with LUVs. We conclude that PIP 2 does not increase activity by concentrating Brag2 with myrArf1 on a membrane surface.
Potential Regulatory Mechanisms of PIP 2 -The regulatory effect of PIP 2 required the N terminus of Arf and a membrane surface. These requirements and the computer modeling have led us to consider three roles of PIP 2 . First, PIP 2 may bind to Arf through lysines 10, 15, and 16 in the N terminus to stabilize nucleotide-free Arf associated with Brag2. Previously, PIP 2 was found to stabilize nucleotide-free Arf when Mg 2ϩ was buffered to Ϸ1 M (39). The second role of PIP 2 may be to bind to the PH domain to control the conformation and orientation of the linker domain. The N terminus of Arf may be important to see this effect; modeling supports the idea that the linker between the PH domain and Sec7 domain may interact with switch 1 of Arf. The linker may also bind to the N terminus of Arf. Past the helix that may interact with switch 1, the residues in the linker are predicted to be in a random coil structure. In the structural model in Fig. 6B, the possibility of contact between the linker and the N-terminal region of Arf is shown. This structural model is hypothetical as the current data only imply the possibility of physical interaction between the linker and Arf. We are currently devising tests of this idea. The third effect of PIP 2 could be to anchor the enzyme⅐substrate complex to the membrane where the transition state (i.e. nucleotide-free Arf⅐Brag2) may be stabilized.
A Unique Aspect of the Arf Nucleotide Exchange Reaction Can Explain the Effect of Membrane on Reaction Rate-A unique aspect of nucleotide exchange on Arf is that, although both substrates (myrArf1⅐GDP and GTP) are soluble, the enzyme⅐substrate complex and one product (myrArf1⅐GTP) are membrane-associated. Therefore, when using myrArf1 as a substrate, a hydrophobic surface is required for product accumulation. MyrArf1 also has the property that in the presence of lipids its affinity for GTP is higher than its affinity for GDP, whereas the affinity of (⌬17)Arf1 for GTP and GDP is similar (21). We found that catalysis on Arf mutants that do not associate with the membrane is 1-2% that of myrArf1, which has high affinity for membranes. Two factors may explain the difference in reaction rates. First, the equilibrium of myrArf1⅐GDP and myrArf1⅐GTP lies more strongly in the direction of myrArf1⅐GTP in the presence of membranes than do the equilibria between (⌬17)Arf1⅐GDP and (⌬17)Arf1⅐GTP or between (L8K)Arf1⅐GDP and (L8K)Arf1⅐GTP. Based on the Haldane relationship (K eq ϭ V forward ⅐K m,reverse /V reverse ⅐K m,forward ), at the very least, the relative forward reaction is more efficient than the reverse when using myrArf1, and it is plausible that the absolute rates may be different. The second explanation for the difference in rates is that the myristoylated N terminus of Arf may anchor the enzyme⅐substrate complex in the membrane with the consequence of accelerating step 1 or step 2 in the reaction scheme shown in Fig. 1A.
An Additional Activator or Regulator May Affect Brag2 Activity-We examined truncated Brag2 constructs to identify the critical role of the PH domain in catalytic regulation. However, other domains of Brag2 may also regulate its activity. The preparations of full-length Brag2 that we examined had less activity than His-Brag2 Sec7-PH . Although we cannot exclude that a fraction of the full-length proteins was not active due to improper folding, the result is also consistent with the idea that a motif outside of the Sec7-PH domains has an autoinhibitory function. A number of ligands could relieve such inhibition. The activity of Brag2 has been reported to be increased by binding to peptides from AMPA receptor (10) and phosphorylated peptides from epidermal growth factor receptor (4). Although these bind to the PH or Sec7-PH domains, they could affect activity by a mechanism distinct from the effect of PIP 2 and could involve other domains of Brag2. Other binding partners might also contribute to Arf specificity by restricting Brag2 localization or by changing the qualitative interaction with the substrate.
Explanation for the k cat Effect of PIP 2 Observed in Single Turnover but Not Substrate Saturation Experiments-In single turnover experiments, PIP 2 was found to affect the K m and k cat of the reaction, whereas in substrate saturation experiments, only the K m was affected. The trivial explanation for the differ-

TABLE 5
Effect of interdomain linker on activity using (⌬17)Arf1 C 50 values were determined as described under "Experimental Procedures." Averages Ϯ S.E. for six experiments in which exchange factor activity was measured in the absence of LUVs are presented.

Allosteric Regulation of Brag2 Sec7 Domain
ence is that the relatively large error associated with parameters determined from substrate saturation experiments disguised the 3-fold difference that was apparent in the more precise and accurate determination by single turnover experiments. Excluding this explanation, the difference could be related to the particular steps of the reaction being measured. In single turnover studies, a single round of GDP release was measured. An effect on the first or second step of the reaction (see Figs. 1A and 8) would account for a change in both K m and k cat . In the case of substrate saturation, GTP binding and Arf⅐GTP release also determine reaction rate. If one of these steps were ratelimiting, a change in step 1 or step 2 might affect the K m for the first substrate without affecting the V max and calculated k cat .
Other explanations include hysteretic effects on Brag2 during the catalytic cycle. For example, the reaction scheme in Fig. 8 shows the transition state toward release of product (E*) that relaxes slowly toward the ground state. Single turnover studies measure steps 1-3 in this scheme. Substrate saturation may primarily measure the cycle of steps 7, 3, 4, and 5. Future work will focus on determining specific reaction steps affected by ligand binding to the PH domain. Summary-Our results support a model in which the PH domain of Brag2 is an allosteric binding site regulating catalysis, whereas the linker between the Sec7 and PH domains contributes to activity. Together, these represent a regulatory mechanism unique to Brag2.