Rho GTPases regulate the assembly of cellular actin structures and are activated by GEFs (guanine-nucleotide-exchange factors) and rendered inactive by GAPs (GTPase-activating proteins). Using the Rho GTPases Cdc42, Rac1 and RhoA, and the GTPase-binding portions of the effector proteins p21-activated kinase and Rhophilin1, we have developed split luciferase assays for detecting both GEF and GAP regulation of these GTPases. The system relies on purifying split luciferase fusion proteins of the GTPases and effectors from bacteria, and our results show that the assays replicate GEF and GAP specificities at nanomolar concentrations for several previously characterized Rho family GEFs (Dbl, Vav2, Trio and Asef) and GAPs [p190, Cdc42 GAP and PTPL1-associated RhoGAP]. The assay detected activities associated with purified recombinant GEFs and GAPs, cell lysates expressing exogenous proteins, and immunoprecipitates of endogenous Vav1 and p190. The results demonstrate that the split luciferase system provides an effective sensitive alternative to radioactivity-based assays for detecting GTPase regulatory protein activities and is adaptable to a variety of assay conditions.
Low-molecular-mass GTPases are central regulatory elements of signalling cascades and are responsible for diverse cellular activities, including cell motility, proliferation and vesicle transport. Rho GTPases are a family of approximately 20 low-molecular-mass GTPases, including Cdc42 (cell division cycle 42), Rac1 and RhoA, and have a key role in actin cytoskeletal rearrangement. With a few exceptions, the regulation of Rho GTPases involves activation of a GTPase by displacement of GDP from the GTPase's nucleotide-binding pocket and its replacement with GTP. When loaded with GTP, GTPases activate cellular signalling cascades by binding with high affinity to appropriate downstream effector proteins . Thus downstream effector proteins of GTPases are able to discriminate between the GTP- and GDP-loaded conformation of GTPases , and, in the case of Rho GTPases, the effectors are often involved in actin nucleation and polymerization. Eventually, inactivation of the GTPase requires hydrolysis of GTP, restoring the GTPase to the inactive GDP-loaded state [3,4].
The nucleotide status of a GTPase is functionally controlled by two major categories of cellular proteins: GEFs (guanine-nucleotide-exchange factors) and GAPs (GTPase-activating proteins). GEFs activate GTPases by ejecting bound GDP from the GTPase, providing GTP with access to the GTPase nucleotide-binding pocket, and leading to activation of the GTPase. In many Rho-family GEFs, exchange is catalysed through a conserved DH (Dbl homology) domain, and, depending on its GTPase specificity, the DH domain can potentially activate one or several Rho GTPase substrates. GAPs on the other hand inactivate the GTPase by stimulating the hydrolysis of GTP to GDP, but like GEFs, the activity of a protein containing a GAP domain can be specific for one or possibly several GTPases .
Establishing the specific GTPase interactions for particular GEFs or GAPs is essential for delineating signalling pathways regulated by these proteins, and two general methods used to characterize these interactions have relied on using either recombinant proteins or cell-based assays. In recombinant protein-based assays, a GTPase is ‘loaded’ with either a radiolabelled or fluorescence-labelled nucleotide. GEF activity is detected by displacement of labelled GDP or incorporation of labelled GTP into a GTPase [6,7]. GAP activity has generally been detected through the cleavage of a radioactive terminal phosphate from radiolabelled GTP . For both GEFs and GAPs, radiolabelled nucleotide assays are extremely sensitive, requiring only nanomolar protein concentrations, but are expensive, hazardous and produce radioactive waste. On the other hand, fluorescent nucleotide-based assays are relatively safe, but require high nanomolar to micromolar protein concentrations [9,10]. Moreover, recent evidence suggests that fluorescent nucleotide analogues, specifically Mant-GDP [where Mant is 2′(3′)-O-(N′-methylanthraniloyl)] and Mant-GTP, may alter the hydrolytic properties of a GTPase and its respective interactions with GEFs and GAPs .
In cell-based assays, a GEF or GAP may be exogenously expressed or suppressed in cultured cells, and the relative increase or decrease in active GTPase is detected using affinity-precipitation assays. In these assays, the GTPase-binding portion of a downstream effector protein is produced as a fusion protein linked to a solid-phase resin material. The effector protein selectively binds to the GTP-loaded form of a GTPase, and comparison between a control and experimental sample is made by Western blot analysis to detect GTPase activation. Ultimately, the analysis is indirect and may be complicated by the effects of other cellular proteins. The assays are also labour intensive and difficult to perform in replicates [12,13].
The development of an environmentally friendly, cost-effective and sensitive assay would benefit the in vitro study of GTPases and their regulatory proteins. In this respect, a split luciferase-based assay could provide an effective alternative for the analysis of GTPases and their related proteins. Although several uses of split luciferases have been reported [14–17], all rely upon splitting the gene sequence for luciferase into two portions. Independently, the two fragments of luciferase have low affinity for each other and are unable to produce chemiluminescence, but the fragments can be recombinantly fused to two potentially interacting proteins. If the two proteins interact, the two fused portions of luciferase are brought into proximity and generate a functional molecule with luciferase activity .
In order to generate a split luciferase system for analysing GTPases, we constructed two bacterial expression plasmids containing the two fragments of split firefly luciferase. The GTPase-binding domain of two different downstream effector proteins, PAK (p21-activated kinase) and Rhophilin1, were subcloned into a plasmid containing one of the split luciferase sequences. Three of the best-characterized Rho GTPases, Cdc42, Rac1 and RhoA, were subcloned into a plasmid containing the second split luciferase fragment. Using these proteins, we were able to measure the activity of purified Rho GEFs and GAPs at concentrations that were similar to radioactivity-based methods and consistent with previous observations regarding tested Rho GEFs and GAPs. The split luciferase system also functioned well with cell lysates expressing exogenous Rho GEFs and GAPs, and was also useful for detecting endogenous Vav1 and p190 activity in immunoprecipitates.
The following human mRNA GenBank® accession numbers correspond to sequence information used for primer design and identification of the open reading frame for deriving amino acid sequences: Rac1 (NM_006908), Cdc42 (NM_001791), RhoA (NM_001664), Dbl (AB085901), Vav2 (NM_001134398.1), Asef (AB042199), Trio (NM_007118.2), p190 (NM_001030055), PARG (PTPL1-associated RhoGAP; U90920), CdGAP (Cdc42 GAP; NM_020754), PAK (NM_001128620) and Rhophilin1 (NM_052924).
PCR and cloning
Luc1 (N-terminal fragment of luciferase) and Luc2 (C-terminal fragment of luciferase) were obtained through PCR amplification using oligonucleotides corresponding to the luciferase fragments shown in Figure 1. The Luc1 PCR product was subcloned into a modified pGEX-KG plasmid for expression as a GST (glutathione transferase)-fusion protein. The Luc2 fragment was cloned into a modified pET24a+ vector. Both the Luc1 and Luc2 sequences were followed on the 3′ end by a flexible linker, a His6 tag sequence and a multiple cloning site. cDNA corresponding to the GTPase-binding domains of human PAK (amino acids 8–162) and human Rhophilin1 (amino acids 16–154) was PCR-amplified and inserted into the multiple cloning site of the Luc1 vector. Full-length human GTPases (Rac1, RhoA and Cdc42) were cloned into the Luc2 vector.
Design and performance of GTPase/effector split luciferases
Rho GEF DH-PH (pleckstrin homology) regions and Rho GAP domains were PCR-amplified and subcloned into both a pMAL-c2 construct containing a modified multiple cloning site for expression as MBPs (maltose-binding proteins) and into a modified pCI construct for the expression of GEFs and GAPs as YFP (yellow fluorescent protein) fusions in mammalian cells. In both cases, the tag was linked to the N-terminus of the GEF or GAP fragment. The amino acid regions from most of the GEFs used in the present study contained the DH-PH region. Specifically, the cDNA sequence for Dbl spanned amino acids 472–834, TrioN spanned amino acids 919–1231, and TrioC spanned amino acids 1597–1918. The Vav2 sequence spanned amino acids 393–679, which covered the DH and PH regions and also the cysteine-rich region. Asef spanned amino acids 180–619, which contained the DH-PH region and the remaining amino acids of the C-terminal end of Asef. Rho GAP domains from the GAPs PARG (658–898), CdGAP (1–251) and p190 (1249–1502) were used in the GAP assay experiments. Mammalian expression vectors (pCDNA) expressing FLAG-tagged versions of wild-type Vav1, a truncation mutation removing the first 65 amino acids of Vav1 (ΔCH) and an activating point mutant (S67D/Y174D) were generously given by Dr Daniel Billadeau (Mayo Clinic, Rochester, MN, U.S.A.).
The Luc1 point mutant (H245D) used for homologous competition assays was produced using the Stratagene QuikChange® mutagenesis kit and the following coding oligonucleotide and its reverse complement: 5′-CGGATACTGCGATTTTAAGTGTGGTACCATTCCATGACGGTTTTGGAATGTTTACTACA 3′. The oligonucleotide contained two silent mutations (indicated by the underlined italic letters) to produce a KpnI site within the sequence to facilitate rapid screening for the mutation prior to sequencing, as well as the appropriate codon change to mutate His245 to aspartic acid (indicated by underlined bold letters) .
All fusion proteins were purified using essentially the same overall procedure; however, different buffers were used for each type of specific fusion protein. Fusion proteins were expressed in Escherichia coli BL21 (DE3) cells in LB (Luria–Bertani) medium containing antibiotic. For GST–Luc1 effectors and MBP-fusion proteins, 100 μg/ml ampicillin was used, and 35 μg/ml kanamycin was used for the Luc2–GTPases. For purification, 2 ml of a starter culture was used to inoculate a 100 ml LB flask containing antibiotic. Cultures were then incubated in a shaker for 2 h at room temperature (22°C). IPTG (isopropyl β-D-thiogalactopyranoside) was then added, and cultures were incubated with shaking at 14°C for 15 h. For GST–Luc1 effectors and MBP-fusion-protein purification, IPTG was added for a final concentration of 30 μM, whereas 70 μM IPTG was used for expression of the Luc2–GTPases.
After the 15 h incubation, the cultures were centrifuged at 4°C for 5 min at 2000 g and the supernatant was removed. Lysis buffer (5 ml) was added to each pellet. GST–Luc1 effector lysis buffer contained 20 mM Tris/HCl, pH 8.0, 50 mM NaCl, 5 mM DTT (dithiothreitol) and 0.5 mM MgCl2. MBP lysis buffer contained 20 mM Tris/HCl, pH 8.0, and 50 mM NaCl. Luc2–GTPase lysis buffer contained 20 mM Tris/HCl, pH 8.0, 50 mM NaCl, 10 mM imidazole, 5.0 mM MgCl2 and 10 μM GDP. Pellets were resuspended and lysed on ice using a sonicator with a microtip and a total of six 200 J bursts of 15 s each. Cell lysates were centrifuged at 4°C for 5 min at 20000 g and the supernatant was added to 0.75 ml of packed resin (GSH–agarose for GST–Luc1 effectors, amylose resin for MBP fusions and nickel resin for Luc2–GTPases). Cell lysates and resins were incubated with constant mixing at 4°C for 30 min. After incubation, tubes containing the lysate and resin were centrifuged at 4°C for 2 min at 150 g. The supernatant was removed and the resin was resuspended in 12 ml of ice-cold lysis buffer. The resin was centrifuged again at 4°C for 2 min at 150 g. The supernatant was removed, and the wash step was repeated. After removing the second wash solution, 500 μl of elution buffer was added and mixed with the resin. GST–Luc1 effector, MBP and His–Luc2 GTPase elution buffers were the lysis buffers, but also containing 20 mM GSH, 10 mM maltose or 180 mM imidizole respectively. To maintain optimal protein activity, the elution buffer containing the eluted proteins was recovered, glycerol was added to 10% (v/v) final concentration, the proteins were aliquotted into several tubes on ice, snap-frozen in liquid N2 and stored at −80°C. A small volume was retained for determining protein concentrations. Protein concentrations were determined using gel densitometry of appropriate bands on Coomassie Blue-stained SDS/PAGE gels, using BSA as a protein standard. GST–Luc1 effector fusion proteins typically yielded concentrations of 0.05–0.1 mg/ml, His6-tagged Luc2–GTPases typically yielded concentrations of approximately 0.1–0.2 mg/ml, and various MBP-fusion proteins typically yielded concentrations between 1 and 2 mg/ml.
GTPase nucleotide loading
For the effector binding and GEF assays, Luc2–GTPases were loaded with guanosine nucleotides or nucleotide analogues as follows: 150 nM GTPase was incubated with 10 μM nucleotide in a loading buffer containing 20 mM Tris/HCl, pH 8.0, 50 mM NaCl and 1 mM EDTA. Samples were allowed to incubate for 1 min at room temperature, and a small volume of 400 mM MgCl2 was added to produce a final concentration of 10 mM MgCl2. Loaded GTPases were used immediately for effector binding or GEF assays, and assay reproducibility was best when GTPases were loaded just prior to use. For GAP assays, 0.5 μM GTPase was loaded with 100 μM GTP or GTP analogue; however, all other conditions were identical with those listed for effector and GEF assays.
Luc1–PAK or Luc1–Rhophilin1 was added to a tube containing 25 μl of luciferin reagent (25 mM Tris/HCl, pH 7.8, 5.0 mM MgSO4, 2.0 mM DTT, 0.5 mM luciferin, 0.5 mM EDTA, 100 μM ATP and 50 μM acetyl-CoA) and samples were read for 30 s as a baseline. Nucleotide-loaded GTPases were added to the tube with Luc1–effectors so that the GTPase concentration was 21 nM and the effector concentration was 7 nM. Samples were mixed and read immediately on a Turner BioSystems Modulus instrument using the luminometer module. Luminometer readings were taken with an 8 s integration time, one reading taken every 10 s. Assays described below were read similarly. It should be noted that, for the effector-binding assays, as well as the GEF and GAP assays, consistent results required careful attention to pipetting, mixing and timing of samples.
In order to compare the sensitivity of fluorescence-based assays using Mant-GTP with the split luciferase assays, the signal window was determined for various dilutions of Cdc42, Rac1 and RhoA. For determining the fluorescence-based signal window, GTPases were loaded as indicated above, except that 500 nM of GTPase was incubated with 500 nM Mant-GTP (‘top’ signal). GTPases were serially diluted in buffer containing 20 mM Tris/HCl, pH 8.0, 50 mM NaCl, 2 mM EDTA and 10 mM MgCl2. The fluorescence background produced by the Mant-GTP itself was also determined at each concentration (‘bottom’ signal). Fluorescence was detected using the Modulus UV fluorescence module, with an excitation wavelength of 365 nm and an emission wavelength filter of 410–450 nm. For assessing the split luciferase signal window, GTPases were loaded with either GMPPNP [guanosine 5′-(β,γ-imido)triphosphate; ‘top’] or GDP (‘bottom’) as described above. The GTPases were then serially diluted and added to luciferin reagent containing an equal molar amount of Luc1–effector protein and incubated for 5 min before reading on the luminometer. The signal windows for both assays were calculated using the following formula :
where mtop and mbottom are the means of the maximum and minimum signals produced, and Stop and Sbottom are the S.D.s of an assay's maximum and minimum signals.
Luc1–effector (45 nM), GDP-loaded GTPases (45 nM) and GMPPNP (1.5 μM) were added to a tube containing luciferin reagent to achieve a final volume of 35 μl. Exchange was initiated by adding 2 μl of the appropriate MBP–GEFs. For analysing the specificity of various MBP–DH-PH fragments, the final concentration of the GEFs used in the assays was 30 nM. For concentration-dependence assays, the GEFs were diluted into MBP elution buffer and then 2 μl of each GEF was added to achieve a final assay concentration of 15, 30 and 60 nM. MBP was used as a control protein in all assays.
Cell-based assays using pCI2.YFP GEF constructs were conducted by electroporating 5×106 HEK (human embryonic kidney)-293 cells with 10 μg of plasmid DNA using previously described procedures . GEF assays for Vav1 activity were conducted by electroporating 4×106 HeLa cells with 40 μg of plasmid DNA at 330 V for 0.1 ms. Transfected cells were plated on 60 mm culture dishes in RPMI 1640 medium containing 5% (v/v) fetal bovine serum and 5% (v/v) bovine calf serum and grown for 18 h before being used in experiments. Prior to GEF assays, the plates were placed on ice, the medium was removed and the cells were washed once with 2 ml of ice-cold PBS. The PBS was removed, and 150 μl of ice-cold lysis buffer (20 mM Tris/HCl, pH 8.0, 0.1% Triton X-100, 5 mM DTT, 3 mM MgCl2, 1 mM EDTA, 5 μg/ml aprotinin and 5 μg/ml leupeptin) was added. Cells were scraped, and the resulting suspensions were centrifuged at 4°C for 10 min at 20000 g to produce clarified supernatants.
For experiments involving YFP-fusion proteins, supernatants were analysed for YFP-fusion protein expression by diluting 5 μl of lysate in 45 μl of lysis buffer, and total sample fluorescence was measured using the Modulus blue fluorescence module (λex=460 nm, λem=515–570 nm). Samples were normalized to have identical fluorescence values using lysates from mock-transfected cells prior to performing GEF assays. GEF assays were performed by loading 150 nM of the appropriate Luc2–GTPase with 2 μM GDP in loading buffer as indicated above. One volume of this reaction mixture was mixed with 3.5 volumes of a solution containing a modified luciferin reagent (as above, but containing 1 mg/ml BSA and 5 mM DTT), 15 nM Luc1–PAK and 0.45 μM GTP. This solution was divided into 35 μl aliquots in clear microcentrifuge tubes and kept on ice. GEF assays were initiated by adding 20 μl of lysate to a tube containing the 35 μl Luc1–PAK/Luc2–GTPase solution, mixed and assayed for luciferase activity at the indicated time points. For determining the expression of various FLAG-tagged Vav1 constructs, 15 μl of lysate was analysed using standard Western blotting procedures. Lysate protein concentrations were measured using Bradford assays and averaged 2.3±0.3 mg/ml.
IP (immunoprecipitation) experiments utilized Jurkat cells that were grown overnight in serum-free RPMI 1640 medium. In order to stimulate cells, 15×106 cells were used per IP sample, and were incubated in 1 ml of RPMI 1640 medium for 15 min at 37°C prior to stimulating the cells with 5 μg of either non-specific mouse IgG or an anti-CD3 antibody (mouse monoclonal, clone UCHT-1) for 2.5 min at 37°C. Cells were immediately placed on ice for 1 min, then centrifuged at 4°C for 2 min at 200 g. The supernatant was removed, and cells were lysed with 0.5 ml of buffer containing 20 mM Tris/HCl, pH 8.0, 50 mM NaCl, 3 mM MgCl2, 1 mM Na3VO4, 0.2% Triton X-100, 5 μg/ml aprotinin and 5 μg/ml leupeptin. Lysates were centrifuged at 4°C for 2 min at 20000 g, and the supernatants were transferred to tubes that contained 6 μl of an anti-Vav1 antibody (rabbit monoclonal EP482Y, Novus) that was prebound to 15 μl (packed volume) of an acrylamide-based Protein A/G resin (Pierce). Lysates were rotated with the resin for 1.5 h at 4°C, the resin tubes were then pelleted with brief centrifugation, the supernatant was removed, and the pellets were resuspended in 1 ml of lysis buffer. After this initial wash, the resin was washed two more times with a wash buffer containing 20 mM Tris/HCl, pH 8.0, 50 mM NaCl, 3 mM MgCl2 and 0.01% Triton X-100. After the final wash, the buffer was carefully aspirated to leave only hydrated resin and as little residual buffer as possible. Luc1–PAK and Luc2–Rac1 were prepared for the assay as described above for the YFP–GEF analysis, and 5 μl of the mixture was added to a resin-containing tube to initiate Rac1 exchange. Tubes were briefly tapped to mix the resin with the Luc2–Rac1/Luc1–PAK solution, and tubes were also briefly tapped to mix between time-point readings. The amount of total Vav1 and phosphorylated Vav1 present in the immunoprecipitates was detected by immunoblotting using a rabbit polyclonal anti-Vav1 antibody (from Daniel Billadeau) and a mouse monoclonal anti-pTyr (phospho-tyrosine) antibody (clone PY20).
For all GEF reactions, background association of Luc2–GTPases and Luc1–effectors were kept to a minimum by mixing the Luc2–GTPase solution with the Luc1–effector solution just prior to introducing a GEF.
Additional information concerning fluorescence-based assays can be found in the Supplementary Experimental section at http://www.BiochemJ.org/bj/441/bj4410869add.htm.
Prior to assaying GAP activity, an appropriate number of Luc1–effector tubes were set up, each containing 12 μl of modified luciferin reagent (see above) and Luc1–effector at 40 nM. Luc2–GTPases were then loaded with GTP (as indicated above) and diluted into modified luciferin reagent to a final concentration of 65 nM. A 12 μl portion was immediately removed from this tube and mixed with 12 μl of Luc1–effector solution to produce a zero-time reading. A small volume of MBP–GAP was added and mixed with the remaining Luc2–GTPase solution to produce a final concentration of 50 nM MBP–GAP. At the indicated time points, a 12 μl sample was removed and mixed with a tube containing 12 μl of the effector solution. After addition of Luc2–GTPase to the Luc1–effector solution, each sample was incubated for 15 min before being read on the luminometer. Relative luminescence was determined by dividing the readings taken after GAP addition by the zero-time measurement for a particular sample.
The GAP concentration assays and GMPPNP assays were performed similarly, except only a 4 min time point was taken for each experiment. For the concentration assay, GAPs were used at the concentrations indicated in Figure 4(D). For the GTP-analogue experiments, GTPases were loaded with either GTP or GMPPNP as indicated above, and Rac1 and Cdc42 (60 nM) were incubated with 30 nM MBP, p190 or Cdgap. RhoA (60 nM) was incubated with 7.5 nM MBP or PARG.
Cell-based assays for GAP activity were performed by transfecting HEK-293 cells with pCI2.YFP GAP constructs as described for the GEF assays. Lysates were also prepared as described above.
For measuring GAP activity, however, Luc2–GTPases were loaded with GTP instead of GDP, and 10 μl of lysate was incubated with 5 μl of loaded GTPase (300 nM). The lysate and GTPase were incubated for 2 min before adding 20 μl of the luciferin/Luc1–effector solution described in the GEF assay subsection. Samples were then incubated for an additional 10 min before luminescence was measured.
IP experiments were set up as described for the Vav1 IP experiments, except stimulated cells were exposed to 0.2 mM pervanadate for 5 min at 37°C. Pervanadate was generated as described previously . After stimulation, cell lysates, resin incubations and washes were performed as described above; however, lysates were rotated with 2 μg of anti-p190 antibody (mouse monoclonal, clone D2D6) prebound to the resin. For the GAP assays, 10 μl of GTP-loaded Luc2–GTPase (150 nM) was added to the appropriate IP resin. GTPases were incubated for 8 min with the resin at room temperature. During the incubation, tubes were gently tapped every 2 min to facilitate mixing of the resin with the GTPase solution. After the incubation, 2 μl of the reaction mixture was removed and mixed with 6 μl of the appropriate Luc1–effector/luciferin solution. The reaction was incubated for 5 min before taking readings on the luminometer.
Competition assays were carried out by mixing a constant concentration of Luc1–PAK with either His6–PAK or Luc1H245D–PAK as competitors ranging in final concentrations from 1 to 640 nM in luciferin buffer (35 mM Tris/HCl, pH 7.8, 7.0 mM MgSO4, 3.0 mM DTT, 0.7 mM luciferin, 0.7 mM EDTA, 140 μM ATP, 70 μM acetyl-CoA and 1 mg/ml BSA). The assay was initiated by adding a small volume of GMPPNP-loaded Cdc42, resulting in a final concentration of 45 nM for Luc1–PAK and 2.5 nM for Luc2–Cdc42. Reactions were monitored for 3.5 h and a plateau in luciferase activity occurred after 3 h of incubation. Percentage activity was determined from the 3.5 h incubation time point. GST was used to test for non-specific inhibition in the assay, however, no inhibition was detected with the highest concentration tested (50 μM), therefore non-specific binding was neglected from data analysis. Data was analysed using GraphPad Prism software, and fitted using one-site competitive binding non-linear regression analysis to determine IC50 values.
The split firefly luciferase used in the present paper was previously optimized as two overlapping fragments, with one larger fragment spanning the first 416 amino acids of firefly luciferase, and a second smaller fragment spanning amino acids 398–550 . These fragments have minimal luciferase activity unless fused to two interacting proteins. We hypothesized that a split luciferase assay could provide an ideal reporter system for detecting whether the nucleotide-binding pocket of a GTPase was loaded with either GDP or GTP (Figure 1A). In this system, one fragment of split luciferase would be fused to a GTPase, and the other fragment would be fused to an appropriate downstream effector capable of selectively binding the GTP-loaded form of a particular GTPase.
To test the capabilities of a split luciferase GTPase/effector system, two bacterial expression plasmids were generated. One plasmid contained amino acids 2–416 of firefly luciferase (denoted Luc1), a flexible linker sequence, a His6 tag and a multiple cloning site for subcloning the GTPase-binding regions of known Rho family effectors. The GTPase-binding region of several Rho-family effector proteins, including PAK, Rhotekin, Diaphanous, ROCK (Rho-associated kinase) and Rhophilin1, were PCR-amplified and inserted into the multiple cloning site; however, the binding regions of PAK and Rhophilin1 were the best expressed, most active, fragments in our initial assays. Additionally, initial purifications using the His6 tag to isolate the effector fragments produced functional, but highly impure, protein preparations; therefore the effector sequences were subcloned into a GST expression construct to facilitate purification (Figure 1B). A second plasmid was also designed that contained amino acids 398–550 of firefly luciferase (denoted Luc2), a flexible linker, a His6 tag and a multiple cloning site into which we inserted cDNA sequences corresponding to Cdc42, Rac1 and RhoA. Unlike the effectors, purification of the GTPases using the His6 tag produced proteins of sufficient quantity and purity for use in the split luciferase assays (Figure 1C).
The ability of the Luc1–effector proteins to recognize the nucleotide status of the purified Luc2–GTPases was tested by loading GTPases with either GDP or GTP[S] (guanosine 5′-[γ-thio]triphosphate) and then mixing them with Luc1–PAK or Luc1–Rhophilin. GTP[S] was used in these assays instead of GTP to inhibit intrinsic GTPase hydrolysis rates. Previous observations regarding PAK specificity indicated that it should selectively bind GTP-loaded Rac1 and Cdc42 . In agreement with these observations, incubating the Luc1–PAK fusion protein with GTP-loaded Luc1–Rac1 and Cdc42 produced a 100-fold increase in luciferase activity in comparison with the GDP-loaded forms (Figure 1D). RhoA is not known to interact well with PAK, and, in our assays, both the GTP- and GDP-loaded Luc2–RhoA did not produce significant luciferase activity with Luc1–PAK. Instead, the RhoA nucleotide status was detected with Luc1–Rhophilin and produced an approximately 3–4-fold increase in luciferase activity over ~5 min. GTP- or GDP-loaded Luc2–Cdc42 did not produce significant luciferase activity with Luc1–Rhophilin. GTP-loaded Luc2–Rac1 also showed an unexpected significant increase in luminescence in comparison with GDP-loaded Rac1. Previous far-Western assays indicated that Rhophilin1 was RhoA-specific , and our detected interaction with Rac1 could be because the split luciferase assay maintains proteins in their native conformation, and suggests that Rhophilin1 is reminiscent of other Rac1/RhoA effectors such as Citron . This interaction was not investigated further, as the goal was to simply generate Luc1–effectors for detecting GTP-loaded Cdc42, RhoA and Rac1. Therefore, although the dynamic range of Luc1–Rhophilin was less than that of Luc1–PAK, it was clearly capable of detecting GTP-loaded RhoA and was used in subsequent assays to detect RhoA-specific GEF and GAP activity.
To more directly compare the sensitivity of split luciferase GTPase/effectors to conventional, fluorescence-based assays using N′-methylanthraniloyl-labelled nucleotides, signal windows were determined for Cdc42, Rac1 and RhoA using both assays. Signal windows assess the reliability of assay data by comparing the average of a signal maximum and minimum for a particular assay, and values above two are indicative of a reliable assay . Signal windows were determined over several concentrations of GTPase (Figure 1E). The fluorescence-based assay was ~20-fold less sensitive; however, all three GTPases displayed similar signal window threshold limits (~250 nM). The split luciferase assay was more sensitive; however, the signal windows varied with the individual GTPases, probably due to the differences in effector interactions (i.e. Rac1 and Cdc42 interactions with PAK compared with RhoA and Rhophilin1). Nevertheless, the lower limits for reliability (signal window <2) using the split luciferase assay were found to be 1 nM for Cdc42, 3 nM for Rac1 and 10 nM for RhoA (results not shown).
Since the Luc1–effectors successfully detected the nucleotide status of the Luc2–GTPases, the proteins were tested as an assay system for detecting Rho GEF activity. In this assay, the Luc2–GTPases were preloaded with GDP and then added to a tube containing the appropriate Luc1–effector, excess GTP analogue (GMPPNP) and a Rho GEF. An increase in luminescence would represent nucleotide exchange, resulting in GTPase/effector binding, luciferase reformation and the production of light (Figure 2A). To test the selectivity of the assay, MBP-fusion proteins of the catalytic DH-PH domains of several known Rho GEFs were generated (Figure 2B). The DH-PH domains were chosen on the basis of known GTPase specificities. Dbl and Vav2 were chosen as broad-range GEFs, potentially capable of activating all three GTPases [1,10,27,28]. The two distinct and separate DH-PH regions found in the Rho GEF Trio were chosen as DH-PH fragments capable of activating single specific GTPases. For Trio, the N-terminal DH-PH fragment is a known Rac1 exchange factor (designated TrioN), whereas the more C-terminal DH-PH functions as a known RhoA exchange factor (designated as TrioC) . Lastly, the DH-PH fragment of Asef was chosen as a Cdc42-specific exchange factor [30–32]. Results from our assays (Figure 2C) agreed with previous observations [10,27,29,30]. Dbl, Vav2 and Asef exchanged Cdc42; Dbl, Vav2 and TrioN exchanged Rac1; and Dbl, Vav2 and TrioC exchanged RhoA. Therefore the split luciferase assay successfully demonstrated all GEFs exchanging their appropriate individual GTPase substrates. Notably, the assays were conducted with proteins at low-to-mid nanomolar concentrations. Vav2 and Cdc42, TrioN and Rac1, and TrioC and RhoA were also used to test the concentration-dependence of the assay. All three assays showed clear concentration-dependence when GEF concentrations were at 15, 30 and 60 nM (Figure 2D). Taken together, the results from both the specificity and concentration assays demonstrate that the GEF assay results were reproducible with relatively low error. Additionally, the sensitivity of the split luciferase GEF assay was compared with the Mant-GTP-based assays (Supplementary Figure S1 at http://www.BiochemJ.org/bj/441/bj4410869add.htm). At concentrations comparable with those used in the split luciferase assay, Vav2, TrioN and TrioC were unable to produce measureable exchange with Cdc42, Rac1 or RhoA using the fluorescence-based assay. A 5-fold increase in the concentration of GEF and GTPase made the GEF activity apparent, however.
Detection of GEF activity using GTPase/effector split luciferases
As an additional test of the assay's sensitivity, we examined whether mammalian cell lysates expressing exogenous GEF constructs could be used in conjunction with the split luciferase assay. To perform this assay, YFP-tagged GEF expression constructs were generated using the same DH-PH sequences identified in Figure 2, and these constructs were then transfected into HEK-293 cells. Small volumes (~20 μl) of lysates produced from these cells were capable of stimulating exchange, in a manner consistent with the individual GEF's expected specificity: Dbl, Vav2 and Asef exchanged Cdc42; Dbl, Vav2 and TrioN exchanged Rac1; and Dbl and TrioC exchanged RhoA. Notably, Vav2 had only slight activity towards RhoA in this experiment (Figure 3A). Furthermore, the split luciferase assay could reliably discriminate between lysates from HeLa cells transfected with wild-type Vav1 or GEF-activating mutants (ΔCH and S67D/Y174D) . Addition of as little as 20 μl of cell lysate to a tube containing Luc1–PAK and Luc2–Rac1 revealed the activity of both GEF-active mutants, whereas vector-control-transfected cells and lysate from cells expressing the autoinhibited wild-type Vav1 protein showed minimal luciferase activity (Figure 3B). Lastly, we also used a Vav1-specific antibody to immunoprecipitate endogenous Vav1 from unstimulated and anti-CD3-stimulated Jurkat cells (Figure 3C). Vav1 immunoprecipitates from anti-CD3-stimulated cells demonstrated enhanced GEF activity in comparison with both antibody control and the Vav1 immunoprecipitates from unstimulated cells. Western blot analysis of the immunoprecipitated Vav1 from the experiment indicated similar amounts of Vav1 in both samples, and an increased amount of tyrosine-phosphorylated Vav1 in the stimulated sample. The results are consistent with the model that tyrosine phosphorylation of Vav1 leads to its activation [34,35]. An important technical aspect of this assay was that the anti-Vav1 antibody used for the IP is specific to the C-terminal tail of Vav1, and therefore distant to the DH-PH region of the protein. Initial experiments using an anti-Vav1 antibody that bound more proximal to the DH-PH region prevented measureable Vav1 GEF activity in control experiments (results not shown).
Detection of GEF activity from cell lysates and immunoprecipitates using GTPase/effector split luciferases
We also used the GTPase/effector split luciferases to detect GAP activity. GAPs accelerate the cleavage of the terminal phosphate of GTP bound to a GTPase, converting GTP into GDP and inactivating the GTPase. In these assays, the Luc2–GTPases were initially loaded with GTP, and then the GAP was added. After specific time points, an aliquot of the reaction was removed and mixed with the Luc2–effector, and the amount of luciferase activity was measured (Figure 4A). A loss in luminescence would represent GTP hydrolysis as the GTPase/effector binding and luciferase reformation would be prevented.
Detection of GAP activity using GTPase/effector split luciferases
MBP fusion proteins of the GAP domain of three previously characterized Rho GAPs were generated, and included CdGAP, p190 and PARG (Figure 4B). These GAPs have different activities toward Cdc42, Rac1 and RhoA, and previous observations indicated that p190 stimulates the GTP hydrolysis activity of all three GTPases , CdGAP has a preference for Cdc42 and Rac1 [37,38], and PARG is highly active on RhoA, but less active on Cdc42 and Rac1 [39,40]. The results from the split luciferase assays (Figure 4C) agreed well with previous observations [36,37,39]. The concentration-dependence of the GAP activity was also tested for the three GAP domains. In this case, a single time point at 4 min was used to measure the GAP activity of CdGAP towards Cdc42, p190 towards Rac1, and PARG towards RhoA. All three showed clear dose-dependence at the three concentrations of GAP tested (Figure 4D). In particular, PARG displayed significant GAP activity towards RhoA, even at a 3 nM concentration. Assays were also performed with GTP and GMPPNP-loaded GTPases to verify that the decrease in luminescence was due to hydrolysis of GTP, and not simply due to a GAP binding to the Luc2–GTPase, preventing its interaction with the effector (Figure 4E). At the concentrations indicated, CdGAP, p190 and PARG were unable to produce a significant decrease in luciferase activity when the GTPase was loaded with GMPPNP.
As with the GEFs, YFP-tagged expression constructs were produced for CdGAP, p190 and PARG (Figure 5A). Activity was detected in these assays with small volumes (10 μl) of lysate from HEK-293 cells expressing the various GAPs. PARG was not used in this assay as it was poorly expressed in comparison with CdGAP and p190, with cells having approximately 15-fold less fluorescence associated with cell lysates. Additionally, p190 was immunoprecipitated from unstimulated and pervanadate-stimulated Jurkat cells to determine if endogenous p190 activity could be detected in immunoprecipitates and whether tyrosine phosphorylation had any effect on GAP activity (Figure 5B). Immunoprecipitated p190 had detectable GAP activity towards Cdc42 and Rac1, and more robust activity towards RhoA in this assay. Pervandate stimulation produced a tyrosine-phosphorylated form of p190 (Figure 5B, lower panel), but had no effect on p190 activity, indicating that tyrosine phosphorylation of p190 does not directly affect its activity. Immunoprecipitated GAP activity towards RhoA was confirmed by incubating the IP resin with GMPPNP-loaded RhoA, demonstrating that the immunoprecipitated p190 was specifically stimulating GTP hydrolysis.
Detection of GAP activity in cell lysates and immunoprecipitates
Although the method shown in Figure 4(A) successfully detects GAP activity, the initial method envisioned for the assay was to load Luc2–GTPases with GTP, add the GTPase to the appropriate Luc1–effector, incubate the pair until the luciferase activity was maximized, and then add an appropriate GAP. However, pre-bound GTPases/effector complexes were relatively unresponsive to the addition of a GAP in comparison with unbound GTPases (Figure 6A). To investigate whether an inherent association of the Luc1 and Luc2 fragments played a role in binding GTPase–effector complexes and inhibited their GAP-catalysed dissociation, we more closely examined the binding between Luc2–Cdc42 and Luc1–PAK by generating a Luc1–PAK fusion protein that contained an inactivating point mutation in the Luc1 fragment. His245 was previously identified as a critical amino acid for firefly luciferase activity, and it is partly responsible for binding of luciferin in the active site through two hydrogen bonds . The crystal structure for firefly luciferase (PDB code 1LCI) also indicated that His245 is not involved in any direct contacts between amino acids that could be associated with the binding of the two luciferase fragments (Figure 6B); therefore, we assumed that this mutation would not alter any possible binding activity associated with the split luciferase fragments. Site-directed mutation of Luc1–PAK produced a fusion protein (denoted Luc1H245D–PAK) that was purified to a level similar to the wild-type fusion protein (Figure 6C), but was almost completely devoid of measurable activity (Figure 6D). Competition assays were then performed to test the ability of Luc1H245D–PAK to inhibit Luc1–PAK binding to GMPPNP-loaded Luc2–Cdc42 in comparison with a His6-tagged version of the PAK PBD (protein-binding domain) that was devoid of the Luc1 fragment (Figure 6E). The His6-tagged PAK fragment bound with an IC50 of 50±3.3 nM, whereas the Luc1H245D–PAK bound with an IC50 of 57.2±4.2 nM, indicating that Luc1/Luc2 fragment binding is not significant in this assay. These results provide evidence that the two luciferase fragments do not directly bind to each other, and are consistent with the crystal structure of firefly luciferase (PDB code 1LCI), where the C-terminal region (amino acids 440–550) forms an independently folded ‘lid’ domain that is attached to the main portion of the enzyme through a flexible linker sequence composed of amino acids 436–440 . Split luciferase activity probably requires positioning of the C-terminal region found on the Luc2 fragment in proximity of the active site found on the Luc1 fragment to facilitate the catalytic activity of the enzyme. Overall, the results suggest that GAPs do not efficiently hydrolyse effector-bound GTPases, and the appropriate assay for measuring robust GAP activity using the split luciferase system required pre-mixing of GAPs and Luc2–GTPases prior to mixing with Luc1–effectors. The results may also indicate that the role of GAPs in GTPase regulation is to ‘shut off’ free, unbound, GTPases to prevent their dissociation from a localized activation site within cells rather than directly act on GTPase–effector complexes, or that specific GTPase–effector pairs are de-activated by specific GAPs, reminiscent of a model suggested for plant Rho GAP activity .
GAP-stimulated GTPase hydrolysis is less detectable using pre-formed Luc1–effector/Luc2–GTPase complexes
The split luciferase assay described in the present paper allowed us to measure three aspects of RhoGTPase regulation and interactions: GTPase–effector interactions, GEF activity and GAP activity. Similar to radioactivity-based assays, we were able to detect these activities in the mid-to-low nanomolar concentration range. The experiments testing GEF and GAP specificities were consistent with previously reported activities for the Rho GEFs and GAPs tested [10,27,29,30,36,37,39]. Taken together, the results demonstrate that the split luciferase GTPase–effector system is versatile, and it can be used for a variety of in vitro G-protein-related applications, including identification of novel GEF and GAP specificities, protein structure/function analysis, and characterization of small-molecule inhibitors/activators that may affect Rho-family proteins [28,43]. The sensitivity of the assay could also be useful for in vitro experiments involving the reconstitution of signalling pathways. Additionally, the ability of the assay to detect GEF and GAP activity from cell lysates and immunoprecipitates demonstrates that the assay can allow the detection of GEF activation following extracellular stimulation, or exogenous expression (or suppression) of an activating protein, or investigation of the functional consequences of specific mutations. In the case of immunoprecipitated proteins, it should be noted that careful consideration should be given to the epitope bound by an antibody, unless a peptide could be used to elute the protein from the resin. In terms of cellular analysis of GEF and GAP activity, we believe that the assay represents a significant improvement in comparison with pulldown/affinity assays where fusion proteins of effector fragments are bound to solid-phase support materials and incubated with cell lysates to bind active GTPases prior to detecting them using Western blots. These assays tend to be labour-intensive and time-consuming, and are difficult to perform with replicates.
We have currently focused on the split luciferase assay as a method to qualitatively assess GEF and GAP activities. Since the assay allows for real-time measurement of GEF activity, it is possible that the assay could be expanded to quantify kinetic parameters for GEFs. However, the system has several complications to direct determination of kinetic parameters, including the indirect nature of the measurement, the variable response of different GTPases with effectors, and the inherent catalytic properties of firefly luciferase. These parameters would require consideration before an overall rate equation could be developed to explore GEF/GAP kinetics.
Although the method developed in the present study was used to analyse Rac1, RhoA and Cdc42, it could be used for a wider array of GTPases and their effectors. For instance, other PAK-binding GTPases, such as Rac2, Rac3, TC10 and TCL, and their potential GEFs and GAPs, could be readily investigated by subcloning these GTPases into the Luc2 expression plasmid . Furthermore, cloning of additional effector regions could make the assay usable for additional GTPases. For example, subcloning of the RalGDS (Ral guanine nucleotide dissociation stimulator) GTPase-binding region could be used to generate a Rap GTPase assay . Overall, the split luciferase GTPase/effector system compares favourably with conventional fluorescence-based assays and affinity-precipitation assays, and it does not have the hazards and wastes associated with traditional radioactive assays.
cell division cycle 42
human embryonic kidney
N-terminal fragment of luciferase
C-terminal fragment of luciferase
yellow fluorescent protein
Erik Anderson and Michael Hamann designed and performed experiments, generated the Figures and wrote the paper.
We thank Daniel D. Billadeau, Mayo Clinic, Rochester, MN, U.S.A., for providing an abundance of reagents, discussion and editorial assistance in the preparation of the paper prior to submission.
This work was funded through start-up funds provided by the Biology Department, Bemidji State University, and Inter Faculty Organization professional improvement funds.