A Point Mutation to Gαi Selectively Blocks GoLoco Motif Binding

Heterotrimeric G-protein Gα subunits and GoLoco motif proteins are key members of a conserved set of regulatory proteins that influence invertebrate asymmetric cell division and vertebrate neuroepithelium and epithelial progenitor differentiation. GoLoco motif proteins bind selectively to the inhibitory subclass (Gαi) of Gα subunits, and thus it is assumed that a Gαi·GoLoco motif protein complex plays a direct functional role in microtubule dynamics underlying spindle orientation and metaphase chromosomal segregation during cell division. To address this hypothesis directly, we rationally identified a point mutation to Gαi subunits that renders a selective loss-of-function for GoLoco motif binding, namely an asparagine-to-isoleucine substitution in the αD-αE loop of the Gα helical domain. This GoLoco-insensitivity (“GLi”) mutation prevented Gαi1 association with all human GoLoco motif proteins and abrogated interaction between the Caenorhabditis elegans Gα subunit GOA-1 and the GPR-1 GoLoco motif. In contrast, the GLi mutation did not perturb any other biochemical or signaling properties of Gαi subunits, including nucleotide binding, intrinsic and RGS protein-accelerated GTP hydrolysis, and interactions with Gβγ dimers, adenylyl cyclase, and seven transmembrane-domain receptors. GoLoco insensitivity rendered Gαi subunits unable to recruit GoLoco motif proteins such as GPSM2/LGN and GPSM3 to the plasma membrane, and abrogated the exaggerated mitotic spindle rocking normally seen upon ectopic expression of wild type Gαi subunits in kidney epithelial cells. This GLi mutation should prove valuable in establishing the physiological roles of Gαi·GoLoco motif protein complexes in microtubule dynamics and spindle function during cell division as well as to delineate potential roles for GoLoco motifs in receptor-mediated signal transduction.

norhabditis elegans embryos and Drosophila melanogaster embryonic neuroblasts have identified initial steps of this process as generation of cell polarity and segregation of various cell fate determinants to different sides of the polarized cell (8); the mitotic spindle is then positioned to facilitate appropriate distribution of determinants to daughter cells during chromosomal segregation and cytokinesis. An integral part of the cellular machinery underlying accurate spindle positioning is the involvement of heterotrimeric G-protein G␣ and G␤␥ subunits in a manner considered independent of 7TMR activation and instead involving RIC-8 (a cytosolic guanine nucleotide exchange factor), GoLoco motif 3 proteins (such as GPSM2/ LGN, Pins, and GPR-1/2 that act as GDP dissociation inhibitors), and GTPase-accelerating proteins ("GAPs"; i.e. RGS proteins) (6 -13). Vertebrate neuroepithelial progenitors use the same cellular machinery to modulate mitotic spindle orientation controlling the balance between asymmetric cell divisions that drive differentiation and planar divisions that favor maintenance and expansion of the neuroepithelial architecture (14 -16). Similarly, an analogous mechanism appears to operate in the stratification and differentiation of mammalian skin (17).
An essential feature of the various emerging models of G-protein nucleotide cycling in mitotic spindle positioning is the requirement for a G␣ i ⅐GoLoco motif complex. For example, in our working model of C. elegans asymmetric cell division controlled by the G␣ subunits GOA-1 and GPA-16 (18,19), it is the G␣⅐GDP/GPR-1/2 complex that activates the generation of astral microtubule (MT) force on mitotic spindle poles, whereas in a competing model (3,12,20), the G␣⅐GDP/GoLoco motif complex is required for the nucleotide exchange ("GEF") activity for RIC-8, thereby generating G␣⅐GTP as the presumed active form of the G-protein (12,21,22). However, it has not been formally established that the G␣/GoLoco motif interaction is required per se for the function of G␣ subunits and GoLoco motif proteins in mitotic spindle positioning. For example, both models of C. elegans asymmetric cell division have been generated primarily by correlating various genetic phenotype data, including loss of pulling forces upon RNA interference-mediated knockdown of goa-1/gpa-16 or gpr-1/2 expression (9 -11, 18, 19). These phenotypic results, although suggestive of a critical function for a G␣⅐GoLoco protein complex, might alternatively reflect separate and distinct functions of G␣ subunits and the multidomain GPR-1/2 proteins in parallel pathways culminating in MT force generation, given that both classes of proteins have other binding partners and established functions. Furthermore, it remains unresolved as to whether G␤␥ is an independent signaling entity in this system or merely a buffer of free G␣⅐GDP levels (14,23,24).
To provide a tool to address these questions, we sought to design a variant G␣ subunit that will not interact with GoLoco motifs and yet retain wild type interactions with guanine nucleotides, 7TMRs, G␤␥ subunits, G␣ effectors, and RGS proteins. Here we describe and validate a single point mutation that renders G␣ i subunits unable to bind GoLoco motif proteins, yet preserves all other aspects of G␣ function. Furthermore, we use this GoLoco-insensitivity ("GLi") mutation to demonstrate that direct G␣/GoLoco motif interaction is required for the G␣-dependent modulation of MT dynamics during mitotic spindle positioning.

EXPERIMENTAL PROCEDURES
Materials-All peptides were synthesized using Fmoc (N-(9fluorenyl)methoxycarbonyl) group protection, high pressure liquid chromatography-purified, and validated by mass spectrometry at the Tufts University Core Facility (Medford, MA). Fluorescent guanine nucleotides were from Invitrogen. Anti-KT3 antibody MMS-125P was from Covance (Berkeley, CA). Unless elsewhere specified, all additional reagents were of the highest quality obtainable from Sigma or Fisher.
Dissociation of Superior Cervical Ganglion and cDNA Microinjection-Detailed methods of preparing rat superior cervical ganglion (SCG) neurons and cDNA microinjection were described previously (27). In brief, adult male Wistar rats were anesthetized by CO 2 inhalation and decapitated as approved by the Institutional Animal Care and Use Committee. Superior cervical ganglia were digested within modified Earle's balanced salt solution containing 0.6 mg/ml collagenase, 0.3 mg/ml trypsin, and 0.05 mg/ml DNase I for 1 h in a shaking water bath at 36°C under an atmosphere of 5% CO 2 , 95% O 2 . The dissociated cells were then washed and plated on poly-Llysine-coated tissue culture dishes containing minimum essential medium and 10% (v/v) fetal calf serum. After cDNA injection, the neurons were incubated overnight at 37°C, and electrophysiology experiments were performed the next day. For some experiments, neurons were incubated overnight with 500 ng/ml Bordetella pertussis toxin (PTX, List Biological Laboratories, Campbell, CA). Microinjection of cDNA was performed with an Eppendorf FemtoJet microjector and 5171 micromanipulator (Eppendorf, Madison, WI) using custom designed software. Constructs containing inserts coding for mGluR2, G␣ i1 (N149I, C352G), G␣ i2 (N150I, C353G), G␣ i3 (N149I, C352G), G␣ i3 (C352G), G␤ 1 , and G␥ 2 were stored at Ϫ20°C as 0.6 -1 g/l stock solutions in TE buffer (10 mM Tris, 1 mM EDTA, pH 8) and injected at a final concentration of 0.1 g/l.
Electrophysiolgical Studies-The method of recording whole cell Ca 2ϩ currents from rat SCG neurons with an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA) was described in detail previously (27,45). Patch electrodes were fire-polished to final resistances of ϳ2 megohms when filled with internal solution. Uncompensated series resistance was Ͻ5 megohms and electronically compensated ϳ80%. Voltage protocol generation and data acquisition were performed using the customdesigned software S5. Current traces were filtered at 2 kHz and digitized at 10 kHz. All recordings were performed at room temperature (21-24°C).
GPSM3 Membrane Recruitment Experiments-Similar to the transmembrane domain-anchored G␣ subunits described in Ref. 48, a pcDNA3.1-based mammalian expression vector was generated to encode CFP-TM-G␣ oi subunits consisting of (starting at the N terminus) a signal peptide, enhanced CFP (49), the N-terminal 103 amino acids of the rat -opioid receptor, the N-terminal 33 amino acids of human G␣ oA , and amino acids 34 -354 of human G␣ i1 . A pcDNA3.1-based mammalian expression vector was generated to encode YFP-GPSM3 consisting of the venus variant of enhanced YFP (30), a c-Myc epitope tag, a hexahistidine tag, and human GPSM3 fused inframe (derived from pcDNA3.1mycHis human GPSM3 (26)).
Human embryonic kidney 293 cells (ATCC; Manassas, VA) were propagated in plastic flasks and seeded onto polylysinecoated glass coverslips according to the supplier's protocol. Cells were transfected using polyethyleneimine and were used for experiments 12-48 h later. Coverslips bearing transfected cells were imaged using a Leica (Bannockburn, IL) SP2 scanning confocal microscope and a 63ϫ, 1.4 NA objective; cells were excited using 458 nm (for CFP) or 514 nm (for YFP) laser lines. Images were acquired and analyzed by an experimenter who was blinded to the transfection condition. A 5-m profile drawn normal to, and centered on, the plasma membrane was obtained for each cell.
MDCK Cell and Spindle Rocking Experiments-MDCK II cells were cultured, transfected, and processed for imaging as described (32). MDCK cells were transfected with either wild type or N149I G␣ i1 -YFP. Time-lapse images of Hoechst 33342stained chromosomal DNA condensation and segregation were recorded as described (32). The angles of the long axis of the metaphase chromosomal array in each frame were measured using Metamorph software (Molecular Devices, Sunnyvale, CA). The absolute angle changes of at least 100 sets of adjacent frames were binned into Ͻ5°, 5-10°, and Ͼ10°groups.
Statistics and Curve Fitting-Unless otherwise indicated, data analysis and curve fitting were performed using PRISM version 4.0 (GraphPad; San Diego). All data are representative of three or more independent experiments.
Structural Analysis of the GoLoco Motif/G␣ Interaction-The crystallographic structure of G␣ i1 ⅐GDP bound to the GoLoco motif of RGS14 has been determined at 2.7 Å resolution (PDB code 1KJY (39)), and more recently at 2.2 Å resolution (PDB code 2OM2 (50)). The two structures share the same overall global architecture; however, there are appreciable differences between the two structural models (50). In this study, we confined our analysis to the 2OM2 structure, as it has higher overall resolution and better refinement statistics. Similarly, subtle but discrete differences exist between the two asymmetric units in both 2OM2 and 1KJY structures; for this reason, we have generally confined our analysis to the A and B chains of 2OM2 as the refinement of this asymmetric unit appeared superior. PyMol (DeLano Scientific; Palo Alto, CA) was used for analysis of structures and the generation of images. Amino acid interaction data were derived using SPACE(CMA) (51) and plotted using MATLAB (The MathWorks, Natick, MA).

Rational Design of a Loss-of-Function Point Mutation in G␣ i to Prevent GoLoco Motif
Interaction-A major feature of the G␣ i1 ⅐GDP/RGS14 GoLoco motif complex (39,50) consists of the GoLoco motif N terminus forming an ␣-helix that binds in the pocket formed by the ␣2 helix ("switch II") and the ␣3 helix of the Ras-like domain of G␣ i1 (Fig. 1A). The invariant glutamine residue (Gln 515 ) of the GoLoco motif (D/E)QR triad terminates this helical portion of the GoLoco motif. The GoLoco motif peptide continues to transit across the surface of G␣ i1 to contact the all ␣-helical domain of G␣ i1 . The conserved Asp 514 -Gln 515 -Arg 516 triad is responsible for turning the GoLoco motif peptide and positioning Arg 516 into the nucleotide binding pocket of G␣ i1 so that the Arg 516 side chain is able to make direct contact with the ␣and ␤-phosphates of GDP, thus stabilizing the bound nucleotide and conferring GDI activity (39). The C-terminal segment of the GoLoco motif makes a sharp hydrogen-bonded turn (Lys 521 -Glu 522 -Asp 523 -Leu 524 ) as it enters the central grove between the ␣Aand ␣B-helices of the G-protein helical domain where it makes an extensive net- work of contacts. The RGS14 GoLoco motif culminates in a short 3 10 helix at amino acids Glu 528 -Phe 529 -Leu 530 . Thus, the main points of contact between G␣ and the GoLoco motif are in the switch II/␣3 helix pocket, switch I, the helical domain (␣A and ␣B helices), the phosphate binding "P-loop," and in the ␣D/␣E loop of the helical domain. A comprehensive analysis of these G␣/RGS14 GoLoco motif interactions is reported in a contact map in supplemental Fig. S2.
Based on these structural data (39,50), it is not intuitively obvious how to create a loss-of-function point mutation(s) in a G␣ i subunit to abrogate GoLoco motif binding yet retain wild type nucleotide binding and hydrolysis, receptor coupling, and effector activation. Many of the GoLoco motif contact sites on G␣ are also used by other regulatory proteins (2,39); this is especially true for the main sites of GoLoco motif contact, e.g. switch-II/␣3 helix, which is essential for G␤␥, effector, and RGS domain interactions (2). Initial structure/function studies on the molecular determinants of the G␣/GoLoco interaction indicated that the ␣D/␣E loop of the G␣ i1 helical domain was essential for binding (52). In testing various G␣ i1 /G␣ s chimeric proteins, amino acids 144 -151 in the ␣D-␣E loop were identified by Natochin et al. (52) as one determinant responsible for the inability of G␣ s to interact with GoLoco motifs. The three residues that differ between G␣ i1 and G␣ s in the ␣D-␣E loop are Arg 144 to Asn, Asn 149 to Ile, and Ser 151 to Cys (illustrated in Fig. 2A). Examination of the G␣ i1 ⅐GDP/RGS14 GoLoco motif structure indicates that, of these three residues, only Asn 149 of G␣ i1 makes contact with the GoLoco motif ( Fig. 1B and supplemental Fig. S1). Asn 149 is a G␣ i/o class-specific residue ( Fig. 2A) and therefore also a good candidate for a GoLoco motif interaction loss-offunction mutation. We hypothesized that a single point mutation of N149I would be sufficient to selectively abrogate GoLoco motif binding to G␣ i1 and create a GoLocoinsensitive (GLi) G␣ subunit. We tested this hypothesis using multiple independent techniques.
Asn 149 to Ile Mutation in G␣ il Prevents Interaction with GoLoco Motifs in Vitro-Mutation of the G␣ i1 amino acid Asn 149 to isoleucine dramatically attenuated GDPdependent binding of G␣ i1 to the GoLoco motif of RGS14, as assessed by SPR spectroscopy (Fig. 2B). We verified this attenuated affinity with an independent measurement of binding that employed fluorescence anisotropy; the N149I mutation reduced the calculated binding affinity of G␣ i1 ⅐GDP for the RGS14 GoLoco motif by 300-fold (Fig. 2C). The signature biochemical activity of GoLoco motifs is GDI activity (7). We therefore also tested GDI activity of the RGS14 GoLoco motif on wild type and N149I G␣ i1 (Fig. 2, D and E). We were unable to observe significant GoLoco motifmediated GDI activity using G␣ i1 (N149I), whereas wild type G␣ i1 was a substrate for RGS14 GoLoco motif GDI activity in the nanomolar range (Fig. 2, D and E), as observed previously (25).
To test the universality of this GoLoco-insensitivity point mutation, we analyzed the binding of N149I G␣ i1 to all human GoLoco motifs using SPR. A graphical representation of all known GoLoco motif proteins, as well as the purified protein constructs used in these SPR analyses, is presented in supplemental Fig. S2. 4 Wild type G␣ i1 (at 1 M) exhibited robust, GDP-selective binding to all known GoLoco motifs (Fig. 3, A-G); at a 10-fold higher concentration, N149I G␣ i1 did not demonstrate any binding to GPSM2 (Fig. 3B) nor to PCP-2 ( Fig.  FIGURE 2. Asparagine 149 is an evolutionarily conserved G␣ i/o class-specific amino acid that is crucial for GoLoco motif interaction. A, multiple sequence alignment of the ␣D/␣E loops within the all ␣-helical domains of human (Hs) G␣ subunits. Sequences are grouped into the four classical G␣ subclasses (s, i/o, q, and 12/13). C. elegans (Ce) and D. melanogaster (Dm) G␣ subunits known to interact with GoLoco motifs are also included in the alignment. SwissProt or GenBank TM accession numbers are as follows: human G␣ s , P63092; human G␣ olf , P38405; human G␣ T1 , P11488; human G␣ T2 , P19087; human G␣ T3 , NP_001095856; human G␣ Z , P19086; C. elegans GOA-1, P51875; D. melanogaster G␣o, P16378; human G␣ i1 , P63096; human G␣ i2 , P04899; human G␣ i3 , P08754; C. elegans GPA-16, Q60XS3; D. melanogaster G␣i, P20353; human G␣ o , P59215; human G␣ q , P50148; human G␣ 14 , O95837; human G␣ 15 , P30679; human G␣ 12 , Q03113; human G␣ 13 , Q14344. B, surface plasmon resonance was used to measure interactions between antibody-immobilized RGS14 GoLoco motif GST fusion protein and either GDP-bound or GDP⅐AlF 4 Ϫ bound G␣ subunits. Injections of either 1 M G␣ i1 (wild type) or 10 M G␣ i1 (N149I) were used. Binding curves were obtained by subtracting nonspecific binding to GST alone surfaces. C, affinity of wild type and N149I G␣ i1 proteins for the RGS14 GoLoco motif was measured using fluorescence anisotropy. 5 nM FITC-RGS14(GoLoco motif) peptide was mixed with increasing amounts of G␣ i1 proteins in the presence of either GDP or GDP⅐AlF 4 Ϫ , and equilibrium fluorescence anisotropy was measured (expressed as millipolarization units (mP) as described in Ref. 42). Dissociation constants were determined by nonlinear regression: wild type G␣ i1 ⅐GDP (14 Ϯ 1 nM). Dissociation constants for wild type G␣ i1 ⅐GDP⅐AlF 4 Ϫ , N149I G␣ i1 ⅐GDP, and N149I G␣ i1 ⅐GDP⅐AlF 4 Ϫ could not be accurately determined as calculated values were Ͻ50% of S.E.M. D, GDI effect of RGS14 GoLoco motif binding on G␣ i1 ⅐GDP was quantified using fluorescence spectroscopy. 3D). Low but measurable levels of N149I G␣ i1 binding were observed on immobilized surfaces of GPSM1 (Fig. 3A), GPSM3 (Fig. 3C), and the GoLoco motifs of RGS12 (Fig. 3E), Rap1GAP1a (Fig. 3F), and Rap1GAP1b (Fig. 3G). We used fluorescence anisotropy to further quantify the binding of GPSM2(GL2) and RGS12GL to G␣ i1 (N149I). Calculated K D values for G␣ i1 (N149I) versus wild type suggest that the GLi mutation reduces affinity for GoLoco motifs by at least 700-and 200-fold, respectively (Fig. 3, H and I).
In Vitro Biochemical Properties of GoLoco-insensitive G␣ Subunits-Mutations in G␣ subunits can alter nucleotide binding, nucleotide hydrolysis, and interaction with regulatory proteins (53). We therefore wanted to test whether the GLi mutation may have also altered the nucleotide binding and/or hydrolysis properties of G␣ subunits. Using [ 35 S]GTP␥S binding, we observed that wild type and N149I G␣ i1 have equivalent nucleotide exchange rates (Fig. 4A). The nucleotide exchange rate as assayed by [ 35 S]GTP␥S binding is an indirect measure of spontaneous GDP release, and thus an index of G␣ affinity for GDP (54). Similarly, we used single turnover [ 32 P]GTP hydrolysis assays to measure the catalytic rate of GTPase activity for wild type and N149I G␣ i1 (Fig. 4B). We observed no significant difference in the ability of wild type or N149I G␣ i1 to hydrolyze GTP. Finally, we constructed the GLi mutation (N150I) in the C. elegans G␣ o -like G-protein GOA-1, known to functionally interact with GoLoco motifs to regulate asymmetric cell division in the one-cell embryo (9). We measured RGS-7 5 -mediated acceleration of GTP hydrolysis by GOA-1. We observed that the N150I mutation had no appreciable effect on the ability of RGS-7 to stimulate the GTPase activity of GOA-1 in a dosedependent fashion (Fig. 4C). Additionally, we verified that GOA-1(N150I) is indeed resistant to GoLoco motif-mediated GDI activity (Fig. 4D), suggesting that the Asn to Ile mutation can be transferred across species, consistent with the conserved evolutionary relationships among metazoan G-proteins and GoLoco motif proteins (6,7).
Functional Properties of the Go-Loco-insensitive G␣-Ca 2ϩ channel modulation in sympathetic neurons was used to examine the ability of GoLoco-insensitive G␣ mutants to complex with endogenous G␤␥ subunits. N-type Ca 2ϩ channels respond to both tonic and 7TMRmediated G-protein activation with a characteristic voltage-dependent modulation mediated by G␤␥ subunits (55). Ca 2ϩ channel currents in whole-cell voltage-clamped neurons were evoked with a double-pulse voltage protocol consisting of two 25-ms test pulses to ϩ10 mV separated by a depolarizing conditioning pulse to ϩ80 mV (56). In control neurons under basal conditions (Fig. 5A, open circle), the amplitude during the first test pulse (prepulse) is slightly smaller than that evoked by the second test pulse (postpulse) resulting in a mean facilitation ratio (postpulse/prepulse amplitude) greater than 1 (Fig. 5F, open bar). Basal (in the absence of agonist) facilitation has been shown to arise from tonic modulation by G␤␥ subunits (28,57). Application of norepinephrine (NE, 10 M) activates endogenous ␣ 2 -adrenergic receptors, resulting in a large inhibition of prepulse amplitude (Fig. 5A, filled circle; Fig. 5E, open bar) and changes in current kinetics (slowing) and facilitation ratio characteristic of G␤␥ modulation (55). Heterologous expression of GLi G␣ subunits (Fig. 5, B-D) abolished both tonic and agonist-mediated modulation as indicated by decreases in mean basal facilitation (Fig. 5F) and agonist-mediated inhibition of the prepulse amplitude (Fig. 5E). The decreases were comparable with those produced by heterolo- These results indicate that GLi G␣ i subunits are capable of binding constitutive G␤␥ subunits and buffering G␤␥ "released" from heterotrimers activated by receptor stimulation (58). The ability to bind G␤␥ subunits does not establish whether the GLi G␣ subunits are capable of forming functional heterotrimeric complexes. Thus, to examine directly the ability of GLi G␣ subunits to form functional heterotrimers, a reconstitution assay was employed that utilizes a PTX-resistant mutation (C-terminal cysteine to glycine mutation or "CG") to distinguish responses arising from endogenous (PTX-sensitive) versus heterologously expressed G␣-containing heterotrimers (27). Expression of the 7TMR metabotropic glutamate receptor (mGluR2) in sympathetic neurons renders Ca 2ϩ channels sensitive to application of glutamate (Fig. 6A). The expressed mGluR2 receptors couple to endogenous G i/o family heterotrimers as indicated by the near complete sensitivity of the voltagedependent inhibition to PTX pretreatment (Fig. 6B). Co-expression of a PTX-insensitive G␣ i3 mutant along with G␤ 1 and G␥ 2 reconstituted the response in PTX-treated neurons (Fig.   6C) (29). Similarly, the GoLoco-and PTX-resistant mutant G␣ i3 (N149I, C352G) was capable of reconstituting the response to glutamate following PTX treatment (Fig. 6D). Fig.  6E depicts the prepulse Ca 2ϩ current inhibition for individual neurons as well as the median response. The magnitude of inhibition varies for the PTX-resistant mutants as the stoichiometric balance of G␣ and G␤␥ influences the response (59). These results demonstrate that the G␣ i3 (N149I,C352G) double mutant is competent to form a G␣␤␥ heterotrimer that couples to mGluR2 and "releases" G␤␥ upon receptor activation.

Normal G␣ Effector Modulation by GoLoco-insensitive G␣ Subunits-
Our demonstration that GLi G␣ can functionally couple to G-proteincoupled receptors and G␤␥ effectors does not preclude the possibility that the GLi mutation is in some way deleterious to G␣ effector interactions. To address this possibility, we measured the interaction of G␣ i1 with the G␣ effector mimetic peptide KB-1753. KB-1753 interacts selectively with activated G␣ i subunits (i.e. GTP␥S and GDP/AlF 4 Ϫ forms but not GDP-bound G␣) in an effector-like conformation by binding to switch-II of G␣ i (43). We measured the binding of wild type and GLi G␣ i1 to KB-1753 using fluorescence anisotropy. Binding of both proteins to FITC-KB-1753 was selective for the activated (GDP/AlF 4 Ϫ ) conformation of G␣ (Fig. 7A) and similar in magnitude (K D values as follows: wild type, 294 Ϯ 40 nM; GLi, 311 Ϯ 40 nM).
The ability of the GoLoco-insensitive G␣ i1 mutant to inhibit the G␣ effector adenylyl cyclase was assessed by measuring the inhibition of agonist-stimulated cAMP accumulation in cells co-expressing constitutively active G␣ i1 (Q204L) subunits and the A 2A adenosine receptor. Co-transfection of G␣ i1 (Q204L) inhibited MECA-stimulated cAMP accumulation by more than 50% when compared with cells co-transfected with the vector control (Fig. 7B). Cells co-expressing G␣ i1 (N149I,Q204L) also reduced MECA-stimulated cAMP accumulation, indicating that GLi G␣ i1 retains the canonical G␣ i inhibitory function on adenylyl cyclase.  . Biochemical properties of GoLoco-insensitive G␣ subunits. A, spontaneous nucleotide exchange rates (k ex ) of wild type (black) and N149I (gray) G␣ i1 ⅐GDP were measured. A time course of specific binding of 100 nM G␣ subunit to 1 M GTP␥S was determined using an [ 35 S]GTP␥S filter-binding assay. Data were fit to single exponential functions with rate constants as follows: wild type G␣ i1 0.013 Ϯ 0.002 min Ϫ1 and N149I G␣ i1 0.009 Ϯ 0.0004 min Ϫ1 . B, spontaneous GTP hydrolysis rates (k cat ) of wild type (black) and N149I (gray) G␣ i1 were measured using [␥-32 P]GTP hydrolysis assays. A time course of 32 P i (inorganic phosphate) production was determined using activated charcoal filtration. Data were fit to single exponential functions with rate constants as follows: wild type G␣ i1 0.40 Ϯ 0.003 min Ϫ1 and N149I G␣ i1 0.30 Ϯ 0.003 min Ϫ1 . C, GTPase-accelerating protein (GAP) activity of C. elegans RGS7 on 200 nM wild type and N150I-mutated C. elegans GOA-1 was measured using 100 nM BODIPYFL-GTP and fluorescence spectroscopy. Data were fit to the four parameter logistic equation to determine EC 50  mammalian cell lines (23,32). Using MDCK cells, we measured the ability of exogenously expressed KT3 epitope-tagged wild type and GLi G␣ i1 subunits to regulate the cellular distribution of endogenous GPSM2 (Fig. 8). We consistently observed that wild type G␣ i1 expression promoted the plasma membrane recruitment of GPSM2, whereas GLi G␣ i1 had no effect on GPSM2 cellular distribution. Analogous results were observed using mRFP-tagged G␣ i1 and endogenous GPSM2 (supplemental Fig. S3). We also observed that exogenous expression of GPSM2 frequently resulted in the accumulation of GPSM2 in "vesicle-like" intracellular organelles. These structures were eliminated by co-transfection with wild type G␣ i1 , presumably by G␣-mediated recruitment of GPSM2 to the plasma membrane (supplemental Fig. S4). However, co-transfection of YFP-GPSM2 with G␣ i1 (N149I) did not alter the morphology of YFP-GPSM2-containing vesicular structures (supplemental Fig. S4).

Effect of the GoLoco-insensitivity Mutation on GoLoco Motifdependent Properties of G␣ i Subunits-To
We also examined the effect of the GLi mutation on interactions between G␣ i1 and the triple GoLoco motif protein GPSM3 (26). We used a CFP-tagged and transmembrane domain-immobilized chimeric G␣ oA/i1 subunit 6 ("CFP-TM-G␣ oi ") to demonstrate the ability of G␣ subunits to specify the membrane localization of GPSM3 (48). Expression of YFP- 6 The chimeric G␣ oA/i1 subunit, comprising the N-terminal 33 amino acids of G␣ oA and the remainder of the polypeptide sequence from G␣ i1 , was originally created to facilitate kinetic imaging and functional assays not described in this manuscript (G. J. Digby and N. A. Lambert, unpublished data). Of the 33 G␣ oA -derived amino acids present within this G␣ chimera, 22 are identical to those found in G␣ i1 , and 8 more are conservative substitutions (i.e., only three positions represent nonconservative differences in side chain character). This 33-amino acid N-terminal region composes the flexible first ␣-helix of G␣ that does not participate in the GoLoco motif interaction (7).  tagged GPSM3 in the absence of co-expressed G␣ subunits is characterized by a uniform distribution of the GoLoco motif protein throughout the cell (Fig. 9, left panel). Expression of GPSM3-YFP in the presence of membrane-tethered, wild type CFP-TM-G␣ oi causes a redistribution of GPSM3 to the plasma membrane (Fig. 9, middle panel). In contrast, expression of GPSM3-YFP in the presence of the N149I mutant CFP-TM-G␣ oi results in a predominantly cytoplasmic distribution of GPSM3 (Fig. 9, right panel). As yet another alternative technique to monitor G␣ i /GoLoco motif interaction in cells, we used co-immunoprecipitation from lysates of cotransfected COS-7 cells; both GPSM1 and GPSM2 interacted robustly with wild type but not GLi, G␣ i1 (Fig. 10). In summary, these data are all consistent with the GLi mutation being a loss-of-function with respect to GoLoco motif binding in cells.
Structural Basis of the GoLoco-insensitivity Mutation-Our data illustrate that the GLi mutation abrogates the ability of G␣ i subunits and GoLoco motif proteins to interact in vitro and in cells. Despite such an extreme loss-of-function in this one aspect of G␣ biology, the GLi G␣ subunits behave normally in all other biochemical and cellular assays we have conducted. There is ample precedent for finding such mutations within G␣ subunits; the RGS-insensitivity mutation (G183S within G␣ i1 ) was first isolated in Saccharomyces cerevisiae Gpa1 (60), shown to be transferable to mammalian G␣ i , G␣ o , and G␣ q subunits (60,61), and validated as affecting only the G␣/RGS domain interaction without affecting nucleotide, receptor, G␤␥, or effector interactions (62). To better understand the potent and highly selective nature of the GLi mutation, we re-analyzed the previously described G␣ i1 ⅐GDP/RGS14 GoLoco motif structure (50). Structural analysis of this G␣⅐GoLoco peptide complex indicates that the predominant role of Asn 149 within G␣ i1 is to directly contact Gln 515 of the RGS14 GoLoco motif. The side chain oxygen of Gln 515 forms a hydrogen bonding network with both the side chain terminal amine (distance of 3.1 Å) and the backbone amine (distance of 2.9 Å) of Asn 149 (Fig. 1B). Asn 149 appears to be an important node in a network of G␣ i1  amino acid residues, including Glu 43 , Asn 76 , Gln 79 , Ser 80 , Gln 147 , Leu 148 , and Arg 178 that act to stabilize the position of Gln 515 in the GoLoco motif ( Fig. 1C and supplemental Fig. S5). Gln 515 of the GoLoco motif is crucial in positioning Arg 516 into direct contact with GDP, and to accomplish this positioning, Gln 515 makes a number of stabilizing interactions with G␣ residues in the P-loop, ␣A helix, switch I, and the ␣D/␣E loop (supplemental Figs. S2 and S5). This network of residues, in which Asn 149 is involved, is also important in stabilizing the "seatbelt" between Glu 43 and Arg 178 hypothesized to restrain the bound nucleotide within its binding pocket (supplemental Fig. S5) (63,64). Although the only amino acid residue of the RGS14 GoLoco motif that directly contacts GDP is Arg 516 (Fig. 1B), GoLoco motif binding to G␣ i1 induces a tighter fit of GDP into the nucleotide binding pocket (39). The salt bridge interaction between the P-loop residue Glu 43 and the switch I residue Arg 178 likely stabilizes bound GDP (39,63,65) and, in cooperation with Arg 516 , accounts for the structural determinants of GDI activity. We also observed that the backbone amine of Asn 149 contacts the side chain of Ala 512 (distance of 3.8 Å); however, this interaction was not observable in all crystallographic models 7 and so may be of uncertain significance.
Through multiple experimental methods, we have demonstrated that the N149I mutation in G␣ i does not perturb in vitro biochemical nor in cellulo signal transduction properties of G␣ i subunits. Based on the structural analysis described above, substitution of the amide side chain of Asn with the aliphatic side chain of Ile would disrupt the hydrogen bonding network between the side chain nitrogen of Asn 149 and the side chain carbonyl of Gln 515 of the GoLoco motif. This is most likely responsible for the majority of the loss-of-function phenotype of the GLi mutation, as it appears that orientation of this highly conserved glutamine is critical to GoLoco motif function. The only two amino acid positions completely con- 7 The RGS14 GoLoco motif residue Ala 512 was observed to interact with Asn 149 of G␣ i1 in both asymmetric units in the PDB 2OM2 structure and the chain A/chain B asymmetric unit of the PDB 1KJY structure. served in all functional GoLoco motifs (supplemental Fig. S1) (7) are the Gln and the Arg 8 residues of the DQR triad. In light of this, we examined the role of Asn 149 in G␣ i class subunits. As described above, it has been noted that Asn 149 is involved in stabilizing the seatbelt configuration between G␣ residues Glu 43 and Arg 178 that is partially responsible for GoLoco motifand G␤␥-mediated GDI activity (39,63). However, in our studies, interaction between N149I mutant G␣ i subunits and G␤␥ subunits appeared to be normal, and this is consistent with G␤␥ subunits having GDI activity toward G␣ s despite the Ile substitution at this position in G␣ s (66). Although Asn 149 is conserved in all G␣ i/o subunits, the closely related G␣ z subunit contains a histidine at this position ( Fig. 2A). Interestingly, G␣ z is unique among G␣ subunits in that it reportedly interacts with the truncated GoLoco motif of Rap1GAP1a in a GTP-selective manner (67) unlike the canonical G␣/GoLoco motif interaction, which is GDP-selective (7). G␣ i /GoLoco Motif Interaction Is Crucial for the Modulation of Microtubule Dynamics-In mammalian cells, overexpression of either GPSM2 or wild type G␣ i1 has previously been shown to destabilize the processes of mitotic spindle orientation and metaphase chromosome segregation (32). In MDCK cells, this is characterized by an increase in the amplitude of spindle oscillations during metaphase (32). The presumed mechanism of action of G␣ i or GPSM2 overexpression on spindle oscillations is an increased recruitment of force-generating G␣ i ⅐GoLoco motif complexes to the plasma membrane (32). Independently, it has also been observed that overexpression of G␣ i3 or GPSM2 alters spindle pole positioning in mammalian cells (68). However, these observations are only suggestive of a critical function for a G␣⅐GoLoco protein complex, given that these results might alternatively reflect separate, distinct functions of G␣ and GoLoco proteins in parallel pathways.
To delineate the precise role of G␣ i /GoLoco motif interactions in ectopically induced spindle oscillations, we used MDCK cells transfected with either wild type or GLi G␣ i1 -YFP and measured simple spindle oscillations using time-lapse video microscopy. We observed that MDCK cells transfected with wild type G␣ i1 underwent vigorous mitotic spindle oscillations during mitosis, as described previously (32) (Fig. 11 and supplemental movies 1 and 2), whereas at comparable expression levels, G␣ i1 (N149I)-transfected cells did not exhibit enhanced spindle rocking relative to untransfected cells ( Fig. 11 and supplemental movies 3 and 4). To quantify these results, we measured the change of the long angle of metaphase chromosomal arrays during mitosis using image analysis. The amplitude of spindle oscillations induced by wild type G␣ i1 expression was substantially higher than that found upon GLi G␣ i1 expression (Table 1). To our knowledge, this result represents the first unambiguous demonstration that direct protein/protein interaction between G␣ i subunits and GoLoco motifs is responsible for the modulation of cortical MT dynamics controlling mitotic spindle orientation.
The precise mechanism of G␣ i ⅐GoLoco motif complex-mediated regulation of cortical MT dynamics and spindle positioning during cell division is not clear. The use of our newly described GLi G␣ i mutant in various model systems of symmetric and asymmetric cell division should help to clarify some of the molecular mechanisms of these processes. In particular, there are several important questions that remain unresolved. First, what is the nature of the G␣ i nucleotide binding/hydrolysis cycle that occurs during cell division? What is the active G␣ i species (G␣⅐GDP, G␣⅐GDP/GoLoco complex, or G␣⅐GTP), and what is the hierarchy of participating G␣ regulatory proteins such as G␤␥, RIC-8, RGS proteins, and GoLoco motif proteins? A consensus within the field appears to be that G␣ i ⅐GDP/GoLoco motif complexes represent the "active" species during MT dynamics in cell division (69). However, the order in which the nucleotide binding and hydrolysis cycle of G␣ i progresses has not been resolved. A recent paper has described RIC-8 as being able to act as a GEF on GoLoco motifliganded G␣ i subunits, thereby implying that GoLoco motif- 8 The GDP-binding arginine residue is a lysine in Rap1GAP2b/c GoLoco motifs (40). However, these GoLoco motifs have no functional activity (40).

TABLE 1 Ectopic expression of wild type, but not GoLoco-insensitive, G␣ il -YFP destabilizes spindle orientation
Time-lapse images of Hoechst 33342-stained chromosomes during metaphase of MDCK II cells expressing G␣ i1 (wild type)-YFP or G␣ i1 (N149I)-YFP were recorded as described (32). Consecutive fluorescence images were taken every 3 s during mitosis (see supplemental movies 1-4 for example). The angles of the long axis of the chromosomal arrays in each frame (see Fig. 11 for examples) were measured using Metamorph software. The absolute angle changes of at least 100 sets of adjacent frames were binned into three categories: Ͻ5°, between 5 and 10°, and Ͼ10°. Data are the mean (S.E.M.) values obtained from three independent time-lapse analyses for either wild type or N149I G␣ i1 -YFP-expressing cells.

Relative frequency of observation (%)
Wild type G␣ il -YFP N149I G␣ il -YFP 0-5°45 (3) 94 (2) 5-10°38 (4) 6 (2) Ͼ10°17 (2) 0 bound G␣ i may be the physiological substrate for RIC-8 (22). However, this same paper also demonstrated that the GoLoco motif is a noncompetitive inhibitor of RIC-8 GEF activity (22), in concordance with earlier results observed within the nematode system (9). Second, what is the direct mechanism by which G␣ i subunits modulate MT dynamics? This latter question is beginning to be understood. It appears that G␣ i proteins act to relieve intramolecular auto-inhibition of "Pins-like" GoLoco motif proteins (e.g. GPSM1/AGS3 and GPSM2/LGN). G␣ i ⅐GDP binding to the GoLoco motifs of these multidomain proteins is believed to act as a conformational switch, allowing the subsequent binding of members of the nuclear mitotic apparatus/mushroom body defect (NuMA/MUD) family of proteins (21,32,70). The nuclear mitotic apparatus/mushroom body defect (NuMA/ MUD) proteins are MT-binding and -regulating proteins; thus, their association with G␣ i ⅐GoLoco motif protein complexes at the cell cortex most likely modulates the dynamics of plus-end astral MTs (71)(72)(73).
Conclusion-Our data presented here describe a single point mutation in G␣ i/o subunits that selectively abrogates the ability of G␣ and GoLoco motifs to interact in vitro and in a cellular context. This Asn to Ile mutation in the ␣D/␣E loop of the helical domain of G␣ prevents the conserved GoLoco motif glutamine residue from properly orienting the GDP-binding arginine of the GoLoco motif. We have demonstrated the utility of this mutant in interrogating the role of G␣ proteins in the modulation of mitotic spindle orientation. We anticipate the widespread use of this GoLoco-insensitivity mutation in both cell culture and in in vivo settings to address the physiological roles of G␣⅐GoLoco motif complex formation in diverse cell division processes, akin to how the RGS-insensitivity mutation of G␣ subunits has been used to identify the physiological roles of endogenous RGS proteins in 7TMR signaling strength and duration (62,74). A particularly important ancillary use of the GoLoco-insensitivity mutation will be to delineate a potential role for GoLoco motif proteins in 7TMR-mediated signal transduction. Gain-of-function studies suggest that GPSM1/AGS3 can modulate the cellular levels of G␣ subunits and thus indirectly affect 7TMR signal transduction (75); other studies have suggested that 7TMR signaling in vivo may be modulated by GPSM1 and GPSM2 function (76 -78). Application of the GLi mutant to such studies will surely provide biochemical and structural insights into the biological function of this important class of G␣ regulatory proteins.