Integration of Fourier Transform Infrared Spectroscopy, Fluorescence Spectroscopy, Steady-state Kinetics and Molecular Dynamics Simulations of Gαi1 Distinguishes between the GTP Hydrolysis and GDP Release Mechanism*

Background: Multiple turnover GTPase assays of Gα are dominated by nucleotide exchange. Results: FTIR elucidates single turnover rates and individual phosphate vibrations. Conclusion: Gαi1-R178S is slowed down in single turnover hydrolysis by 2 orders of magnitude, Gαi1-Asp229 and -Asp231 are key players in Ras-like/all-α domain coordination. Significance: With FTIR on Gα established, detailed information on the reaction mechanism can be obtained. Gα subunits are central molecular switches in cells. They are activated by G protein-coupled receptors that exchange GDP for GTP, similar to small GTPase activation mechanisms. Gα subunits are turned off by GTP hydrolysis. For the first time we employed time-resolved FTIR difference spectroscopy to investigate the molecular reaction mechanisms of Gαi1. FTIR spectroscopy is a powerful tool that monitors reactions label free with high spatio-temporal resolution. In contrast to common multiple turnover assays, FTIR spectroscopy depicts the single turnover GTPase reaction without nucleotide exchange/Mg2+ binding bias. Global fit analysis resulted in one apparent rate constant of 0.02 s−1 at 15 °C. Isotopic labeling was applied to assign the individual phosphate vibrations for α-, β-, and γ-GTP (1243, 1224, and 1156 cm−1, respectively), α- and β-GDP (1214 and 1134/1103 cm−1, respectively), and free phosphate (1078/991 cm−1). In contrast to Ras·GAP catalysis, the bond breakage of the β-γ-phosphate but not the Pi release is rate-limiting in the GTPase reaction. Complementary common GTPase assays were used. Reversed phase HPLC provided multiple turnover rates and tryptophan fluorescence provided nucleotide exchange rates. Experiments were complemented by molecular dynamics simulations. This broad approach provided detailed insights at atomic resolution and allows now to identify key residues of Gαi1 in GTP hydrolysis and nucleotide exchange. Mutants of the intrinsic arginine finger (Gαi1-R178S) affected exclusively the hydrolysis reaction. The effect of nucleotide binding (Gαi1-D272N) and Ras-like/all-α interface coordination (Gαi1-D229N/Gαi1-D231N) on the nucleotide exchange reaction was furthermore elucidated.

Heterotrimeric G proteins are interaction partners of G protein-coupled receptors (GPCRs) 4 and deliver external signals into the cell (1). They are switched on by exchange of GDP for GTP induced by the GPCR as exchange factor and switched off by GTP hydrolysis. The nucleotide is bound between two domains of the G␣-subunit, namely the Ras-like domain, which is similar to the G-domain of small GTPases, and the all-␣ domain. In its inactive state G␣ i1 ␤␥ exists GDP bound in its heterotrimeric form. Activation by guanosine nucleotide exchange factors, like GPCRs or non-receptor guanosine nucleotide exchange factors (2)(3)(4), leads to nucleotide exchange in the ␣-subunit. Incorporation of GTP alters the protein conformation in the switch I-III regions (5), which causes separation of the G␣ i1 and ␤␥ subunits and signal transduction, e.g. by binding of G␣ i1 to adenylate cyclase isoforms that in turn inhibit the production of cAMP from ATP (6). GTPase activity of G␣ i1 leads to hydrolysis of GTP to GDP and P i , inactivation, and reassociation with its ␤␥ subunits. Heterotrimeric G proteins are equipped with an intrinsic arginine finger (Arg 178 in G␣ i1 ) usually provided in case of small GTPases by the GAP protein, which is known to function as a key residue for catalyzing the hydrolysis reaction. Therefore intrinsic GTPase rates of G␣ i1 are rather comparable with Ras⅐GAP than to Ras. The hydrolysis mechanism taking place in G␣ i1 thereby determines the duration of its active state, which can be pathogenic when hindered, e.g. by ADP-ribosylation catalyzed by pertussis toxin (7). As for Ras, the intrinsic hydrolysis activity of G␣ i1 can be further accelerated by GTPase activating proteins (GAPs), which are called regulators of G protein signaling (RGS), e.g. RGS4 in case of G␣ i1 (8). It is generally known that GDP/GTP exchange is the rate-limiting step in multiple turnover measurements of G␣ isoforms (9 -12). Therefore, beside steady-state assays using ␥-32 P labeling (13) or malachite green (14), pre-steady-state assays are used to characterize the hydrolysis reaction of G␣ isoforms (15)(16)(17)(18). We present here for the first time single turnover measurements of G␣ i1 using time-resolved FTIR spectroscopy, an ultrasensitive method that can be applied in solution and has been successfully used for photoactivable proteins like bacteriorhodopsin (19,20), channelrhodopsin (21), and other rhodopsins (22). Adenylyltransferases (23), ATPases (24 -27), and GTPases (28 -31) can also be investigated by usage of caged nucleotides (28). The resulting photolysis and hydrolysis difference spectra depict the label-free GTP and GDP states of G␣ i1 and are able to reflect environmental changes in the sub-Å range (32) and the dynamics at a time resolution of milliseconds. Bridging the gap between single turnover and steadystate kinetics, we also applied a multiple turnover GTPase assay using reversed phase HPLC and additionally measured the nucleotide exchange rate using tryptophan fluorescence of Trp 211 at 280/340 nm as extensively described elsewhere (15,18,33) and molecular dynamics simulations. GDP release and GTP uptake is a complex mechanism determined by the movement of the all-␣ domain, which was shown by structural studies (34). The Ras-like/all-␣ interface thereby depicts a complex interaction network including both amino acids and the nucleotide. By orchestration of different methods we were able to determine the effect of point mutations and distinguish their role in hydrolysis and nucleotide exchange.

Materials and Methods
G␣ i1 -WT and mutant proteins were expressed, isolated, and characterized by various biophysical and biochemical methods (Fig. 1).
Michaelis-Menten multiple turnover kinetics were monitored via reversed phase HPLC. The kinetics include both the catalytic reaction and the dissociation/association kinetics depicted as time per turnover. The isolated single turnover hydrolysis reaction was obtained via FTIR spectroscopy as half-life values of the global fit. The isolated nucleotide exchange kinetics were investigated via tryptophan fluorescence spectroscopy and depicted as half-life values of the intensity change. Experiments were accompanied by molecular dynamics simulations to decode the molecular reactions at the atomic level.
Cloning-The gene for the human GNAI1 (UniProtKB accession number P63096-1; kind gift from C. Wetzel, University of Regensburg, Germany (39)) was amplified by polymerase chain reaction using the oligonucleotide primers GCGC-CCATGGGCTGCACGCTGAGC and GCGCGGATCCTTA-AAAGAGACCACAATCTTTTAG (restriction sites for NcoI and BamHI are underlined). Resulting fragments were cut with NcoI and BamHI and ligated into the vector pET27bmod (kind gift from M. Engelhard, MPI Dortmund, Germany (40)) with a N-terminal ϫ10 histidine tag and tobacco etch virus (TEV) site. The plasmid was transformed into Escherichia coli DH5␣ for amplification. G␣ i1 mutants R178S, D229N, D231N, and D272N were created by overlap PCR using appropriate primers. Integrity of each construct was confirmed by sequencing. cDNA encoding human RGS4 was acquired from the Missouri cDNA Resource Center (Rolla, MO), tagged with a N-terminal ϫ10 histidine tag and TEV site, and also ligated into the vector pET27bmod. Amplification was performed similar to G␣ i1 .
Protein Expression-The plasmid encoding G␣ i1 was transformed into E. coli Rosetta 2 (DE3) (Novagen, Merck, Darmstadt, Germany) and incubated overnight at 37°C on LB agar plates containing 0.2% (w/v) glucose as well as 50 g/ml of kanamycin and 20 g/ml of chloramphenicol for plasmid and strain selection. A preculture (LB medium, 50 g/ml of kanamycin, 20 g/ml of chloramphenicol, 0.2% (w/v) glucose) was inoculated and incubated overnight at 37°C and 160 rpm. The plasmid encoding RGS4 was transformed into E. coli BL21(DE3) under identical conditions using only kanamycin for plasmid selection. For the main culture, 18 liters of LB medium supplemented with 50 g/ml kanamycin and 0.2% glucose were inoculated with the preculture and grown at 37°C, 100 rpm, and 20 liters/min airflow in a Biostat C20-3 Fermenter (Sartorius, Göttingen, Germany). At an A 600 of 0.5-0.6 the culture was cooled to 18°C and protein expression was induced by addition of 0.25 mM isopropyl 1-thio-␤-D-galactopyranoside. After 15-18 h the cells were harvested by centrifugation at 5000 ϫ g and 4°C, suspended in buffer A (20 mM Tris, pH 8, 300 mM NaCl, 1 mM MgCl 2 , 0.5 mM EDTA, 5 mM D-norleucine for G␣ i1 or 50 mM Tris, pH 8, 150 mM NaCl, 0.5 mM EDTA, 5 mM D-norleucine for RGS4), flash frozen, and stored at Ϫ80°C.
Protein Isolation-Frozen cells were thawed, supplemented with 0.3 mM PMSF, 5 mM ␤-mercaptoethanol, DNase (G␣ i1 containing cells were additionally supplemented with 0.1 mM GDP), and disrupted using a microfluidizer M-110L (Microflu-FIGURE 1. G␣ i1 is switched on by the exchange of GDP for GTP (k off /k on ), then GTP hydrolysis proceeds (k hyd ) and P i is released. Multiple turnover kinetics were measured via HPLC, which cannot distinguish between the three processes. Nucleotide exchange kinetics (k off /k on ) were monitored via tryptophan fluorescence spectroscopy. Single turnover kinetics (k hyd ) were measured via time-resolved FTIR difference spectroscopy.
idics Corp., Newton, MA) at 800 bar. To spin down cell fragments and not disrupted cells, the suspension was centrifuged for 45 min at 45,000 ϫ g and 4°C. RGS4 containing cells were centrifuged with an additional low-speed step for 15 min at 18,000 ϫ g and 4°C followed by high-speed centrifugation for 45 min at 75,000 ϫ g and 4°C. The supernatant was applied to a 25-ml nickel-nitrilotriacetic acid superflow (Qiagen, Hilden, Germany) column, equilibrated with buffer B (buffer A ϩ 0.3 mM PMSF ϩ 5 mM ␤-mercaptoethanol ϩ 20 mM imidazole), using a Ä KTApurifier 100 system (GE Healthcare Life Sciences, Freiburg, Germany) at 6°C with a flow rate of 1-2 ml/min. After a washing step with buffer C (buffer B ϩ 4 mM MgCl 2 , 400 mM KCl, 1 mM ATP) for 8 -10 column volumes and a subsequent step with buffer B ϩ 50 mM imidazole for another 8 -10 column volumes, the proteins were eluted with buffer B ϩ 200 mM imidazole. The fractions containing G␣ i1 or RGS4 were selected after SDS-PAGE, pooled, supplemented with 5 mM DTT, and concentrated to 5 ml using a 10,000 MWCO concentrator (Amicon Ultra-15, Merck Millipore, Darmstadt, Germany). For gel filtration chromatography, the pool was applied to an illustra HiLoad 26/600 Superdex 200 pg column (GE Healthcare Life Sciences, Freiburg, Germany) equilibrated with buffer D (20 mM Tris, pH 8, 300 mM NaCl, 1 mM MgCl 2 , 2 mM DTT, 0.1 mM GDP for G␣ i1 or 50 mM Hepes, pH 8, 100 mM KCl, 2 mM DTT for RGS4). Peak fractions were analyzed by SDS-PAGE. Purest fractions containing G␣ i1 or RGS4 were mixed 1:2 with buffer E (20 mM Tris, pH 8, 1 mM MgCl 2 , 0.1 mM GDP for G␣ i1 or 50 mM Hepes, pH 8, 100 mM KCl for RGS4), pooled, and concentrated to ϳ20 mg/ml for G␣ i1 or ϳ10 mg/ml for RGS4 using a 10,000 MWCO concentrator. Protein concentration was determined using Bradford reagent as triplicate. The concentrated pool was aliquoted, flash frozen, and stored at Ϫ80°C until utilization. Coomassie-stained gels after SDS-PAGE of purified proteins are depicted in Fig. 2.
Multiple Turnover GTPase Measurements-For determination of the GTPase activity under multiple turnover conditions, the samples contained 10 M G␣ i1 in 20 mM Tris, pH 8, 150 mM NaCl, 0.5 mM MgCl 2 , and 0.1 mM DTT. After tempering for 5 min at 30°C, 0.1 mM GTP (G␣ i1 -WT, -R178S, -D229N, -D231N) or 2.5 mM GTP (G␣ i1 -D272N) was added and immediately the first aliquot of the sample was analyzed by reversed phase HPLC at 254 nm (Beckman Coulter System Gold, Pasadena CA) (mobile phase: 50 mM P i , pH 6.5, 5 mM tetrabutylammonium bromide, 7.5% AcN; stationary phase: ODS-Hypersil C18 column). After a 10-min incubation at 30°C, a second aliquot was analyzed by HPLC. The amount of GTP was chosen to guarantee a substantial excess of GTP during the whole time of the measurements. Evaluation of the data were done by integration of the GDP and GTP peaks followed by normalization (sum of the areas A GDP ϩ A GTP ϭ 1). Determination of the time for one turnover per molecule G␣ i1 , including exchange of GDP for GTP and GTP hydrolysis, was done according to the calculation of turnover times formula, with time points t 0 and t 1 [min], the areas of the GTP peak at the time points t 0 and t 1 A GTP , t0 and A GTP , t1 (normalized area), and the concentration of G␣ i1 and GTP c(G␣ i1 /GTP). Calculated values were averaged and the standard deviation was calculated from three experiments for each wild type and mutant protein.
Monitoring the Nucleotide Exchange Rate Using Fluorescence Spectroscopy-The nucleotide exchange rate of wild type and mutant G␣ i1 was measured via tryptophan fluorescence of Trp 211 (41, 42) using a Jasco FP 6500 Spectrofluorometer (Easton, MD). 500 nM wild type or mutant protein was supplemented with 20 mM Tris, pH 8, 150 mM NaCl, 1 mM MgCl 2 , and 1 mM DTT and tempered 5 min to 30°C. After monitoring the fluorescence baseline for another 5 min, nucleotide exchange was initiated by the addition of 2.5 M GTP␥S and the reaction was monitored with exc ϭ 280 nm and em ϭ 340 nm for at least 60 min. Mixing was ensured by constant stirring with a stirring bar and the temperature was controlled by an external water bath.
Data were fitted in OriginPro 9 (OriginLab Corp., Northampton, MA) using a monoexponential formula with a linear correction, which accounts for bleaching, with the rate constant k, the amplitude coefficient a, the slope m, and the offset n. Half-life values were calculated using (t1 ⁄ 2 ϭ ln(2)/k). Half-life values were averaged and the standard deviation was calculated from three experiments for each wild type and mutant protein.
Nucleotide Exchange of Wild Type and Mutant G␣ i1 to Caged GTP-The exchange of bound GDP to photolabile pHPcgGTP or NPEcgGTP was performed in the presence of alkaline phosphatase coupled to agarose beads, which is unable to hydrolyze caged compounds. Phosphatase beads were washed 5 times in buffer 1 (50 mM Tris, pH 7.5, 100 M ZnSO 4 ) to remove free phosphatase. Each washing step was followed by centrifugation at 10,000 ϫ g and the supernatant was checked for free phosphatase using a colorimetric assay with para-nitrophenylphosphate (43). 5 mg of wild type or mutant G␣ i1 were supplemented with 50 mM Tris, pH 7.5, 10 M ZnSO 4 and a 2ϫ molar excess of the caged nucleotide. Hydrolysis of free and proteinbound GDP to guanosine was monitored via HPLC. After 3 h at room temperature Ͼ95% of GDP was hydrolyzed. Samples were centrifuged at 10,000 ϫ g for 2 min and the supernatant was re-buffered through a Nap5 column (GE Healthcare Life Sciences) that was equilibrated with 10 mM Hepes, pH 7.5, 7.5 mM NaCl, 0.25 mM MgCl 2 , 1 mM DTT at 7°C. Protein fractions were pooled and concentrated in a 10,000 MWCO concentrator (Amicon Ultra-0.5, Merck Millipore, Darmstadt, Germany). Concentrations were determined using Bradford reagent as triplicate and the nucleotide exchange rate of bound caged nucleotide was again determined via HPLC (Ͼ95%). Samples were aliquoted into 107.5 g portions (5 mM final concentration in FTIR measurements), flash frozen in liquid nitrogen, and stored at Ϫ80°C. Subsequently samples were lyophilized for 3 h at Ϫ55°C/0.05 mbar in a Christ Alpha-1-2 LDPlus Lyophilizer (Martin Christ GmbH, Osterode am Harz, Germany) and stored light protected in parafilm and aluminum foil at Ϫ20°C.
FTIR Measurements on G␣ i1 -FTIR measurements were performed using 5 mM G␣ i1 ⅐cgGTP in 200 mM Hepes, pH 7.5, 150 mM NaCl, 5 mM MgCl 2 , 20 mM DTT, 0.1% (v/v) ethylene glycol at 15°C. RGS4 catalyzed measurements were performed by the addition of 5 mM RGS4 to establish a 1:1 complex with G␣ i1 . Sample preparation was done under red light to protect the photolabile caged group. Composition of the required residual buffer depends on the protein concentration of the samples after nucleotide exchange to achieve the above named ion concentrations. FTIR samples were prepared between two CaF 2 windows (Ø 2 cm, 2 mm thickness, one of them with a 10-m deepened area 1 cm in diameter). One lyophilisate of G␣ i1 ⅐cgGTP was dissolved in 0.5 l of the appropriate residual buffer at the center of the deepened window and subsequently covered with the second window, whose rim had been lubricated with a thin ϳ1 mm wide silicon grease film. The windows were fixed in a metal cuvette and mounted in the spectrometer (Bruker IFS 66v/S or Vertex 80 v (Bruker, Ettlingen, Germany)). After sample equilibration, background spectra were taken (400 scans) and photolysis of the caged compounds was carried out with an LPX 240 XeCl excimer laser (Lambda Physics, Göttingen, Germany) by 12 flashes within 24 ms (pHPcgGTP) or 40 flashes within 80 ms (NPEcgGTP) at 308 nm (100 -200 mJ/flash, 20 ns pulse duration) (28). Measurements were performed in the rapid-scan mode of the spectrometer for 30 min (G␣ i1 -WT, -D229N, -D231N, -D272N) or 3 h (G␣ i1 -R178S) using a liquid nitrogen-cooled mercury cadmium telluride detector. Data between 1800 and 950 cm Ϫ1 was collected with a spectral resolution of 4 cm Ϫ1 using an aperture of 5 mm in the double-sided forward-backward data acquisition mode with a scanner speed of 120 kHz. Data were analyzed via global fit (44). The absorbance change (⌬A(,t)) was fit-ted with a sum of exponential functions n describing the apparent rate constants k 1 and amplitudes a 1 of the hydrolysis reaction and the amplitudes a 0 of the photolysis reaction for every wavenumber .
In the figures disappearing bands face downward and appearing bands face upward. Data were averaged over at least 3 measurements. Half-lives were calculated as arithmetic means, variation was calculated as standard deviations.
Molecular Dynamics Simulation and Evaluation-Molecular dynamics (MD) simulations were performed starting with the G␣ i1 ⅐Mg 2ϩ ⅐GTP␥S structure of Protein Data Bank (PDB) code 1GIA (5) that depicts the truncated (⌬1-32 ⌬345-354) active state of G␣ i1 . Structure preparation was performed in Moby (45) and included correction of dihedrals, angles, and bonds according to the UA amber84 forcefield (46), protonation of ionizable side chains using the PKA,MAX,UH,JAB3 algorithm as well as replacement of the GTP␥S for a GTP molecule (total charge: Ϫ4) and initial solvation by the Vedani algorithm (47). Point mutations were realized in Moby and were followed by a short headgroup optimization. Simulation systems were set up in GROMACS 4.0.7 (48 -52). The prepared structures were thoroughly solvated in a cubic simulation cell filled with 154 mM NaCl in explicit TIP4P water. Simulations were carried out in the all atom OPLS forcefield (53) with GTP parameters from T. Rudack (54) at 310 K using the berendsen thermo-and barostat and a time step of 1 fs. Long range electrostatics were calculated using PME (cutoff 0.9 nm), short range electrostatics were calculated using a VDW cutoff of 1.4 nm. Bonds were constrained using LINCS. Systems were energy minimized and equilibrated for 25 ps with restrained protein positions followed by three free MD runs, each to a simulation time of 100 ns (total simulation time 1.5 s).
Structure analysis was performed using the GROMACS evaluation tools and the contact matrix algorithm implemented in Moby. Pictures were created using PyMOL 1.7.1.1 (Schrödinger LLC, Portland, OR) and Gnuplot 4.4 (55).

Results
FTIR Measurements of G␣ i1 -Time-resolved FTIR spectroscopy enables label-free detection of the GTP/GDP vibrations as well as determination of the apparent kinetics of the hydrolysis reaction. The protein was loaded with caged GTP and the sample was excited at 308 nm with a laser flash to remove the caged group (28) that cleaves rapidly (10 7 s Ϫ1 for pHPcgGTP (36)). The resulting difference spectrum is referred to as photolysis spectrum. Subsequently the intrinsic hydrolysis reaction in G␣ i1 takes place (Fig. 3).
The reaction (Scheme 1) is observed in FTIR. Global fit analysis of the absorbance changes revealed a monoexponential function that describes the hydrolysis (Fig. 4). No intermediate enrichment was observed in the measurements of G␣ i1 -WT. Global fit analysis of five independent G␣ i1 -WT measurements at 15°C resulted in a half-life of 32.7 Ϯ 2.5 s (k hyd ϭ 0.02 s Ϫ1 ).
Data analysis according to Equation 3 resulted in photolysis and hydrolysis spectrums that represent the transition from the pHPcgGTP to the GTP bound active state of G␣ i1 and the transition from the active GTP bound state to the inactive GDP bound state, respectively. Bands facing downward represent the educt state, bands facing upward represent the product state. Both spectra show numerous highly reproducible bands in the protein (1680 -1350 cm Ϫ1 ) and the phosphate (1350 -950 cm Ϫ1 ) region (Fig. 5). Surprisingly a band at 1784 cm Ϫ1 appeared in the photolysis and disappeared in the hydrolysis reaction, indicating a protonation of a carboxyl group from an Asp or Glu (56) in the GTP state (Fig. 5). To our knowledge this is the first time a protonation change has been observed in GTPases. For a clear cut assignment further studies with sitedirected mutations have to be performed.
Phosphate vibrations were assigned using isotopically labeled nucleotides, namely ␣-18 O 2 -pHPcgGTP, ␤-18 O 3 -pHPcgGTP, and ␥-18 O 4 -NPEcgGTP. Double difference spectra of FTIR measurements using unlabeled and labeled nucleotides showed exclusively band shifts caused by the isotopes and allow band assignments of the phosphate region. In the photolysis spectrum the bands at 1240, 1224, and 1155 cm Ϫ1 were assigned to the asymmetric stretching vibrations of ␣-, ␤-, and ␥-GTP (Fig. 6, A and B). The vibrations for ␤and ␥-GTP appear as clear bands, the ␣-band appears as a shoulder only in the photolysis spectrum but is more distinct in the hydrolysis spectrum. Band assignments of the hydrolysis reaction confirmed ␣-, ␤-, and ␥-GTP vibrations at 1243, 1224, and 1156 cm Ϫ1 . The vibrations of the product state were assigned to 1214 cm Ϫ1 for ␣-GDP, 1134 and 1103 cm Ϫ1 for ␤-GDP, and 1078 and 991 cm Ϫ1 for the cleaved free phosphate (Fig. 6, C and D). The cleaved phosphate is not protein bound, as the vibrations at 1078 and 991 cm Ϫ1 are typical for free phosphate. Proteinbound phosphate intermediates are blue-shifted, e.g. in case of Ras⅐GAP an intermediate band appears at 1192 cm Ϫ1 (57). Because a protein-bound phosphate intermediate was not observed as in case of the Ras⅐GAP catalyzed reaction, bond breakage is the rate-limiting step in the hydrolysis reaction of G␣ i1 (Fig. 4). Summarizing, the hydrolysis reaction of G␣ i1bound GTP to GDP and P i was monitored label free at atomic resolution and in the millisecond time scale. Individual asymmetric stretching modes of GTP and GDP bound to G␣ i1 and P i were assigned clear cut.
In addition, we performed the same experiments with the G␣ i1 ⅐RGS4 1:1 complex at 5°C. Addition of RGS4 further catalyzed the hydrolysis reaction by almost 2 additional orders of magnitude (Fig. 7). As for the intrinsic measurements, global fit analysis resulted in one exponential rate, which demonstrates that again bond breakage is rate-limiting. No protein-bound FIGURE 3. Three-dimensional spectrum (global fit) of G␣ i1 -WT. The first spectrum represents the photolysis spectrum, subsequently hydrolysis takes place and was monitored. Time dependence of the bands at 1155 cm Ϫ1 (gray) and 1078 cm Ϫ1 (blue) is indicated. The absorbance change of these bands is shown in Fig. 4. SCHEME 1. Reaction scheme observed in FTIR measurements.  (Fig. 6)). Solid lines represent the monoexponential global fit, dots represent data points. FIGURE 5. Photolysis and hydrolysis spectrum of G␣ i1 . Bands facing downward in the photolysis spectrum represent the pHPcgGTP state of G␣ i1 , bands facing upward depict the GTP bound state. Bands facing downward and upward in the hydrolysis spectrum represent the educt and product state of the hydrolysis reaction that takes place in G␣ i1 , respectively. The spectral region between 1680 and 1620 cm Ϫ1 is superimposed by water absorptions and not further regarded.
phosphate intermediate was observed and the absorptions of protein-bound GTP disappeared with the same rate as the absorptions of free P i developed. RGS4 contributed numerous protein bands and altered the GTP/GDP binding modes. Assignments of these bands by isotopic labeling and site-specific mutagenesis will be part of future work.
In the following, various G␣ i1 residues were investigated to determine their role in nucleotide exchange and hydrolysis using site-directed mutagenesis. Investigations included the intrinsic arginine finger mutant G␣ i1 -R178S that is known to have a slowed down hydrolysis rate (58), as well as mutations affecting the interface of the Ras-like and the all-␣ domain (G␣ i1 -D229N/G␣ i1 -D231N), and a residue that is participating in nucleotide binding (G␣ i1 -D272N) (Fig. 8). Multiple turnover measurements via reversed phase HPLC, nucleotide exchange experiments via fluorescence spectroscopy, and single turnover measurements via time-resolved FTIR spectroscopy were used to investigate these mutants.
Steady-state Measurements-The multiple turnover GTPase reaction of wild type and mutant G␣ i1 , consisting of nucleotide exchange (k off,GDP and k on,GTP ) and the hydrolysis rate k hyd , was investigated by reversed phase HPLC at 30°C according to Equation 1. Results can be grouped into four classes. One turnover is a cycle consisting of GDP release, GTP binding, and GTP hydrolysis that took 12.7 Ϯ 0.2 min for G␣ i1 -WT. The arginine finger mutant G␣ i1 -R178S was slowed down to 29.6 Ϯ 1.6 min per turnover. G␣ i1 -D229N and -D231N shared an accelerated turnover time of 3.9 Ϯ 0.03 min and G␣ i1 -D272N was accelerated even more to 0.5 Ϯ 0.02 min per turnover (Fig. 9A). It is generally accepted that the GDP release step is rate-limiting in multiple turnover measurements of G␣-proteins (59). To further investigate the underlying rate constants we additionally performed nucleotide exchange and single turnover hydrolysis experiments.
Nucleotide Exchange Experiments-In contrast to multiple turnover experiments, tryptophan fluorescence spectroscopy can monitor solely the nucleotide exchange reaction from GDP to GTP␥S of G␣ i1 as Trp 211 is sensitive for binding of the third phosphate group (Fig. 10). Hydrolysis cannot proceed as GTP␥S is a non-hydrolyzable GTP analogue. The results of nucleotide exchange can be grouped into three classes. In contrast to multiple turnover measurements, the half-life value for nucleotide exchange to GTP␥S of G␣ i1 -R178S was similar to G␣ i1 -WT (12.8 Ϯ 0.9 and 10.7 Ϯ 0.2 min, respectively). Nucleotide exchange was accelerated by a factor of about 3 in G␣ i1 -D229N and -D231N to 3.3 Ϯ 0.2 and 3.7 Ϯ 0.1 min and even more accelerated in G␣ i1 -D272N (0.8 Ϯ 0.03 min) (Fig. 9B). Hence it can be concluded that the acceleration in multiple turnover measurements of G␣ i1 -D229N, -D231N, and -D272N can be explained by accelerated dissociation times for GDP and/or association times for GTP. On the other hand the decelerated multiple turnover time for G␣ i1 -R178S is not caused by nucleotide exchange, indicating that nucleotide exchange is not the rate-limiting step for this mutant.
Single Turnover Hydrolysis Measurements Using FTIR-In contrast to multiple turnover measurements, time-resolved FTIR spectroscopy can determine the hydrolysis reaction label free under actual single turnover conditions with high spatiotemporal resolution. The half-life value was obtained from the  results of the global fit procedure (44). The measured kinetics can again be grouped into three classes. G␣ i1 -WT and G␣ i1 -D231N had similar half-life values of 32.7 Ϯ 2.5 and 27.8 Ϯ 2.6 s, respectively. The single turnover hydrolysis reaction of G␣ i1 -D229N and -D272N was slightly slowed down to 50.2 Ϯ 4.5 or 49.8 Ϯ 4.1 s. The hydrolysis reaction of G␣ i1 -R178S was noticeably slowed down by 2 orders of magnitude to 3400 Ϯ 400 s (Fig.  9C). Thereby the deceleration of G␣ i1 -R178S in multiple turnover measurements can be explained solely by the slowed down single turnover hydrolysis reaction. Due to the change of the rate-limiting step, the slowdown in the multiple turnover assay appears to be only about 2-fold, whereas the single turnover FTIR assay yield the true slowdown by 2 orders of magnitude.
Molecular Dynamics Simulations-To further examine the molecular interactions taking place in the interface between the Ras-like and all-␣ domain of G␣ i1 , molecular dynamics simulations were performed to elucidate the role of Asp 229 and Asp 231 at atomic detail. Simulations of wild type G␣ i1 and mutants G␣ i1 -D229N and -D231N were performed for 100 ns each. Subsequently contact matrix analysis was carried out for every simulation. Contacts were sampled in time windows of 1 ns and the interaction partners of Asp/Asn 229 and Asp/Asn 231 were depicted in Fig. 11. Polar contacts of the side chain groups are indicated by black bars.
Asp 229 formed a stable interdomain contact to the all-␣ domain through Arg 242 that bound Gln 147 in wild type G␣ i1 (Fig. 11A). When mutated to Asn 229 , this interdomain contact triad was interrupted after the first 30 ns. Thereby the contact loss between Asn 229 and Arg 242 happened simultaneously to the contact loss of Arg 242 to Glu 43 and Gln 147 . Hence Asp 229 seems to position Arg 242 allosterically, so that Arg 242 forms an interdomain contact that tightly binds and stabilizes the Raslike and the all-␣ domain.
Similar to Asp 229 , Asp 231 formed an interdomain contact to Arg 144 within a 100-ns MD simulation (Fig. 11B). It is notable, that the contact Asp 231 -Arg 144 does neither exist in the starting structure generated from PDB code 1GIA, nor in the original crystal structure, but formed de novo in the simulation. The initial contact to Lys 277 persisted through the simulation time. When mutated to Asn 231 the initial contact to Lys 277 was weakened, but the contact to Arg 144 was completely lost.
Summarizing the results from multiple and single turnover measurements, nucleotide exchange experiments and simulation data, we understand the effects of the point mutations in G␣ i1 . G␣ i1 -R178S was slowed down in multiple turnover measurements, but even slower in single turnover measurements that depict only the hydrolysis reaction itself. Its nucleotide exchange ability appeared unaltered. Slowdown of multiple turnover is solely from hydrolysis in this case. Thus Arg 178 only participates in the hydrolysis reaction as described elsewhere (5,58).
The interface mutations D229N and D231N were both accelerated in multiple turnover measurements. Investigations of the nucleotide exchange reaction showed that the exchange time for both mutants was accelerated. Simulation data suggest an allosteric (Asp 229 ) and direct (Asp 231 ) interdomain binding mode of both amino acids. Mutations affecting the guanosine binding moiety Asp 272 resulted in accelerated half-life values in multiple turnover measurements that can be originated to an accelerated nucleotide exchange behavior as shown via fluorescence spectroscopy.

Discussion
Molecular mechanisms that take place in G␣ i1 have been investigated by numerous studies including structural (5, 34, 60), computational (61), and biochemical (13,14,33) assays. In particular, multiple turnover GTPase assays like malachite green or radiometric phosphate tests using [␥-32 P]GTP are widely used even though it is commonly known that GDP/GTP exchange is the rate-limiting step in intrinsic multiple turnover measurements (9 -12) and thereby determines k obs . Pre-steadystate measurements using GTP or [␥-32 P]GTP pre-loaded G␣ subunits that are triggered via Mg 2ϩ addition are also able to depict single turnover conditions, but the percentage of nucleotide loading (G␣⅐GTP versus G␣⅐GDP) and the altered GTP binding affinity due to Mg 2ϩ -binding (62) may cause systematic errors such as side reactions like nucleotide exchange. However, in FTIR measurements the percentage of loaded cgGTP versus GDP does not influence the kinetics due to the method of phototriggered difference spectroscopy. Additionally we checked the loading rate via HPLC (always Ͼ95% cgGTP) and   removed non protein-bound nucleotides. The determined single turnover rate for wild type G␣ i1 measured via FTIR spectroscopy (0.02 s Ϫ1 at 15°C) is in good agreement to the literature (0.03 s Ϫ1 at 30°C (58) and 0.03 s Ϫ1 to 0.04 s Ϫ1 at 20°C (13,64)). In addition the ensemble of methods enables a classification of effects caused by point mutations in high detail. Effects caused by the intrinsic arginine finger mutant G␣ i1 -R178S were quantified correctly in single turnover measurements (2 orders of magnitude) but not in multiple turnover measurements (factor of 2). The unaltered nucleotide exchange rate of G␣ i1 -R178X mutants has already been described elsewhere (5).
Single turnover FTIR spectroscopy unravels for the first time the rate-limiting step of the intrinsic GTP hydrolysis in G␣ i1 . Analogue experiments with small GTPases revealed that in some cases the bond breakage and in others the P i release is rate-limiting (57). For G␣ i1 no protein-bound cleaved phosphate intermediate could be observed, thus bond breakage is the rate-limiting step in this reaction. This is surprising due to the tight coordination of the nucleotide by G␣ i1 . The narrow protein environment is still able to release free phosphate to the periphery, probably through a small channel located near the ␥-phosphate. The measured IR bands for GTP and GDP are very sensitive to changes in the protein environment and depict for the first time the coordination of the natural nucleotides GDP and GTP in G␣ i1 in contrast to GTP analogues, which were described to have poor affinities for G␣ i1 (65). After we have successfully assigned the ␣-, ␤-, ␥-, and the free phosphate vibrations it will be possible to assess the effect of point mutations in the binding pocket of G␣ i1 in the future supported by theoretical IR spectra calculation from QM/MM simulations as performed for the small GTPase Ras (66) to further decode the experimental spectra. In addition to the phosphate bands, various bands caused by the protein itself were nicely resolved, which will enable investigations of the hydrolysis mechanism taking place in G␣ proteins with improved spatio-temporal resolution. The observed band at 1784 cm Ϫ1 is the first protonation change observed in GTPases to our knowledge. In fact, heterotrimeric G proteins have been speculated to function as pH sensors (67) and a protonation change close to the surface of G␣ i1 could function as a key player in this reaction.
In addition to the intrinsic GTPase reaction of G␣ i1 we were also able to measure the hydrolysis reaction catalyzed by RGS4 via FTIR spectroscopy. Hydrolysis was thereby accelerated by almost 2 orders of magnitude (Fig. 7). As for intrinsic G␣ i1 , bond breakage is the rate-limiting step.
We were able to show with our orchestration of different methods that two point mutations in the G␣ i1 Ras-like/all-␣ interface (G␣ i1 -D229N/G␣ i1 -D231N) are able to weaken the coordination in the protein domain interface. Our measurements together with MD simulations demonstrate the importance of the amino acid triad Asp 229 -Arg 242 -Gln 147 for the interface coordination in G␣ i1 . Asp 229 holds Arg 242 in a position to bridge the interface to Gln 147 . Investigations on the mutant G␣ i1 -R242A confirmed its role (nucleotide exchange: 3.09 Ϯ 0.21 min/single turnover hydrolysis: 32.5 Ϯ 3.5 s) and resulted in similar values as G␣ i1 -D229N. In agreement, accelerated nucleotide exchange for G␣ i1 -R242A has recently been described for the analogue R243H in G␣ o (68). Our findings on the other amino acid in the domain interface, Asp 231 , suggest a direct binding mode of the side chain of Asp 231 across the interface to Arg 144 . This contact is not observable in any of the deposited structures of G␣ i1 in the Protein Data Bank, except structure 4PAQ where the side chain of Arg 144 is slightly tilted toward Asp 231 with occupancy of 0.46 (69). In contrast to tightly packed crystal structures, the dynamics of G␣ i1 in our experiments and in our simulations are much more comparable with physiological conditions, so we hereby demonstrate the importance of the salt bridge Asp 231 -Arg 144 , which is not observable in the crystal structures. Our findings are summarized in an advanced interface binding model of G␣ i1 as shown in Fig. 12.
In summary, we were able to measure the isolated rates of nucleotide exchange and GTP hydrolysis, which both contribute to the signaling state of G␣ i1 . In addition we identified the individual phosphate vibrations of GTP, GDP, and P i during the hydrolysis reaction of G␣ i1 . We demonstrated the importance of the intrinsic arginine finger for the hydrolysis reaction and the relevance of Asp 272 for nucleotide binding. Furthermore, we identified novel key players in the coordination of the Raslike/all-␣ interface. Asp 229 stabilizes the interface allosterically via Arg 242 and Asp 231 forms a direct H-bond to Arg 144 . Orchestration of our methods will further elucidate the molecular mechanisms taking place in G␣ i1 in the future.