Ras–Guanine Nucleotide Complexes: A UV Spectral Deconvolution Method to Analyze Protein Concentration, Nucleotide Stoichiometry, and Purity

Abstract

The many members of the Ras superfamily are small GTPases that serve as molecular switches. These proteins bind the guanine nucleotides GTP and GDP with picomolar affinities, thereby stabilizing on- and off-signaling states, respectively. Quantitative in vitro Ras studies require accurate determination of total protein, its fractional occupancy with guanine nucleotide, and spectroscopic purity. Yet the high nucleotide affinity of Ras and the overlapping UV spectra of the protein and bound nucleotide make such determinations challenging. Here we describe a generalizable UV spectral deconvolution method to analyze the total protein concentration, total nucleotide stoichiometry, and purity of Ras complexes.

Graphical abstract

The Ras superfamily comprises a set of more than 150 homologous, small GTPase signaling proteins in the human genome [1, 2] that each typically bind one molecule of activating GTP or inactivating GDP with picomolar to nanomolar affinities [3–5]. Quantitative in vitro studies of Ras-nucleotide complexes often require rapid analysis of basic complex parameters. Herein we describe a UV spectral deconvolution method for analyzing Ras in complex with one or a mixture of guanine nucleotides (GXP), in order to quantify Ras total protein concentration, total guanine nucleotide stoichiometry, and spectroscopic purity. Such analysis can reveal the presence of apo Ras that is inactive, is prone to aggregation, and may complicate biophysical and biochemical studies. In addition, such analysis can uncover and quantify incomplete removal of free nucleotide following nucleotide loading that will interfere with HPLC (or other) analysis of the bound nucleotide species. Finally, such analysis may reveal chromophoric or scattering contaminants that may interfere with experiments in unpredictable ways. We expect this method will be generalizable to most or all members of the large Ras superfamily of small GTPases.

The method employs wavelength range 250 – 300 nm for UV spectral deconvolution. This range adequately covers the peak absorbance regions of both the protein and nucleotide components, and is also suitable for detecting interfering contaminants including aromatic chromophores and scattering species. The detailed protocol for the analysis (see Supplemental Protocols 1 – 3) requires only small quantities of GXP-Ras complex (as little as 10 μg per sample if employing a microvolume spectrophotometer). The method is faster than colorimetric methods such as Lowry, BCA or Bradford, and is less sensitive to interference from common components of in vitro physiological buffers such as K⁺, Mg²⁺, glutathione or other reducing agents [6, 7]. Moreover, the method is more robust than a simple dual wavelength measurement (for example A₂₈₀ / A₂₅₂) that is limited in its accuracy and precision by low information content, and by its inability to detect contaminants that change the shape of the UV spectrum.

Briefly, the present UV spectral deconvolution method begins by generating distinct reference spectra for the guanine nucleotide and protein components, then uses best fit analysis to sum these reference spectra in the optimal proportions to generate a convoluted model spectrum that matches, as best possible, the measured spectrum of the purified GXP-Ras complex. Figure 1A presents a control experiment that deconvolutes the spectrum of an equimolar mixture of free nucleotide (GTP) and a control protein (CheW) into its component nucleotide and protein spectra. Here the deconvolution procedure (Supplemental Protocol 3) employs reference spectra obtained for pure GTP and purified CheW. The CheW protein was chosen as a control because it does not bind nucleotides but is similar in size to Ras and, like Ras, lacks Trp residues [8].

Figure 1: Control Spectral Deconvolution of a GTP(+)CheW Mixture, and Reference Spectra for Deconvolution of GXP–HRas Complexes.

A) Control deconvolution of the UV spectrum of an equimolar mixture of GTP and the non-nucleotide binding protein CheW (each 100 μM). Shown are the measured spectrum of the mixture (solid line) and the best fit convoluted spectrum (dot-dash-dot line) summing the indicated, deconvoluted GTP and CheW component spectra. The measured and convoluted spectra are virtually indistinguishable. The deconvolution procedure (Supplemental Protocol 3) employed reference spectra obtained for pure GTP and pure CheW, respectively. B) Overlay of spectra measured for GTP (with and without heating), GDP, GMP, GMPPNP (with and without heating), GMPPN, and protein-subtracted GXP. The spectrum of 50 μM GTP (without heating, “GTP Reference”) was used as the nucleotide reference spectrum for deconvolutions, and as the anchor for the indicated overlays which are virtually indistinguishable. The protein-subtracted GXP spectrum was acquired by heat extraction from a washed GXP-HRas complex (Supplemental Protocol 2). All other spectra were obtained for commercially available HPLC-grade pure nucleotides in standard buffer. The GTP reference spectrum is the average of 9 replicate spectra acquired as triplicates on 3 days, while the spectra for other nucleotides are each the average of 6 replicates acquired as triplicates on 2 days. C) Comparison of protein-subtracted GXP spectra generated by high and low speed spins. Shown is the measured spectrum of a representative GXP-HRas complex (black line) and GXP spectra obtained by heat-denaturation of the complex at 95 ºC for 6 min followed by a low speed (10,000 X g, blue dotted line) or high speed (90,000 X g, red dotted line) centrifugation step to pellet and remove the precipitated protein. Notably, the low speed spin is insufficient to remove all the precipitate and thus retains excess scattering and/or absorbance at short wavelengths that is more effectively removed by the high speed spin. D) Comparison of apo HRas reference spectra obtained by GXP subtraction (Subtracted HRas Reference, solid line) and by summing the absorbance spectra of chromophoric amino acids (Synthetic HRas Reference, dashed line). Both spectra are normalized to the absorbance of 50 μM apo HRas (Supplemental Protocol 2). The Subtracted HRas Reference spectrum is an average of 3 independent reference spectra, each obtained from triplicate measurements on different days. All experimental spectra were obtained at 20 ºC on a Nanodrop One microvolume spectrophotometer (Thermofisher) in standard buffer (140 mM KCl, 15 mM NaCl, 25 mM HEPES, pH to 7.4 with 10 M KOH).

When applying the deconvolution method to GXP-Ras complexes, the guanine nucleotide reference spectrum is easily obtained but additional steps are needed to obtain the isolated Ras reference spectrum. Herein pure GTP is employed to generate the nucleotide reference spectrum, although any non-cyclic guanine nucleotide can be used since the guanine base dominates the spectrum, rendering all non-cyclic guanine nucleotides virtually indistinguishable in the relevant region of the UV spectrum with the same molar extinction coefficient at 252 nm (ε₂₅₂ = 13,700 M⁻¹cm⁻¹) [9, 10]. Figure 1B shows the overlaid spectra of five representative pure guanine nucleotides (GTP, GDP, GMP, GMPPNP, and GMPPNH) as well as an overlaid spectrum of a nucleotide mixture (GXP) released from an HRas complex by heating and ultracentrifugation as described below. Figure 1B further shows that the spectra are not significantly altered by heat-triggered hydrolysis, as illustrated by heating GTP or GMPPNP to 95º C for 6 min. However, Figure 1C also shows that a deviation is observed when the GXP mixture is released from an HRas complex by heat denaturation followed by clarification via low-speed centrifugation rather than ultracentrifugation. The low speed spin fails to fully pellet the precipitated HRas and yields a detectable scattering contamination, particularly at shorter wavelengths. It follows that the UV spectra of all these pure guanine nucleotides and GXP mixtures are operationally identical due to their shared guanine chromophore, as long as scattering contaminants are removed.

To our knowledge, no experimental UV spectrum has been reported for pure apo Ras protein, and it would be difficult or impossible to obtain such a spectrum due to the high affinity of Ras for nucleotide and the instability of the apo protein [4, 11]. However, an operational apo Ras reference spectrum can be generated in one of two ways. It can be calculated by addition of the spectra of individual aromatic amino acids (Trp, Phe, Tyr, His) [12] weighted by their stoichiometries in the protein amino acid sequence to yield a synthetic apo Ras reference spectrum as shown in Figure 1D. Alternatively, it can be generated by a heat-assisted subtraction method to yield a nucleotide-subtracted Ras reference spectrum, also shown in Figure 1D. Briefly, as detailed in Supplemental Protocol 1, the GXP-Ras complex is first washed with standard buffer (140 mM KCl, 15 mM NaCl, 25 mM HEPES, pH to 7.4 with 10 M KOH) to remove any unbound nucleotides and spectroscopic contaminants. Subsequently, as described in Supplemental Protocol 2, the UV spectrum of the washed GXP-Ras complex is measured in the same buffer, then the sample is heat-denatured (95 ºC for 6 min), which quantitatively precipitates the protein and releases the bound nucleotides, followed by ultracentrifugation to remove the precipitated protein. Finally, the UV spectrum of the protein-free supernatant containing the released GXP nucleotides is measured and subtracted from the UV spectrum of the GXP-Ras complex to yield the nucleotide-subtracted Ras reference spectrum, which differs substantially from the synthetic reference spectrum (compare in Figure 1D). This deviation likely arises, at least in part, from the artifactual spectral shifts in the synthetic spectrum which is calculated from aqueous amino acid spectra, whereas the aromatic side chains in the folded Ras-nucleotide complex are fully or partially buried in the protein interior.

Once the guanine nucleotide reference spectrum and the Ras reference spectrum have both been obtained and calibrated via molar extinction coefficients to their respective concentrations (Supplemental Protocol 2), they can be used to rapidly deconvolute the UV spectra of other GXP-protein samples as detailed in Supplemental Protocol 3. Briefly, the UV spectrum of the washed GXP-protein complex is measured. Next, standard deconvolution software (A|E, Version 2.2; www.fluortools.com) [13] is employed to compute the optimized sum of the nucleotide and protein reference spectra, thereby yielding (i) a convoluted model spectrum that is a best fit match to the measured spectrum of the GXP-Ras complex, and (ii) the deconvoluted spectra of the nucleotide and protein components. The total concentrations of guanine nucleotide and Ras present in the GXP-protein complex are then calculated from the component spectra (Supplemental Protocol 3).

Figures 2A,B illustrate the deconvolution analysis for representative washed GXP-HRas complexes using a nucleotide-subtracted HRas reference spectrum obtained for the standard HRas construct. In Figure 2A, three measured GPX-HRas spectra obtained via independent trials on three different days are deconvoluted into their GXP and HRas components using a reference pure GTP spectrum and a nucleotide-subtracted HRas reference spectrum obtained via Supplemental Protocol 2 for a different HRas sample. There is excellent agreement of each measured complex spectrum with the best fit convoluted spectrum summing the nucleotide and protein reference spectra. Note also the strong reproducibility illustrated by the three independent trials, which are virtually indistinguishable, demonstrating the robustness of the procedure. Figure 2B applies the same analysis to three different washed GPX-HRas proteins: standard HRas, G12S, and D38E, where the latter two point mutants are generated in the standard background. The figure shows two independent trials for each protein carried out on different days. The results are virtually indistinguishable for standard HRas and both mutants, which is not surprising since the mutations are located on the protein surface and do not add or remove chromophoric residues [14–17]. By contrast, Figures 2C,D show that the pairing of a GTP reference spectrum with a synthetic apo HRas reference spectrum yields lower quality deconvolutions that are unable to recapitulate the measured spectra of standard GXP-HRas complexes (Figure 2C), or of mutant GXP-HRas complexes (Figure 2D). The use of a synthetic protein spectrum as a reference for deconvolution is problematic most likely because, as noted above, the water-based amino acid spectra used to build it cannot account for the spectral shifts of the protein chromophores in their local, folded protein environments [18].

Figure 2: Deconvolution of GXP–HRas Complex Spectra into their Nucleotide and Protein Components.

(A,C) Overlay of 3 independent deconvolution trials (different colors) carried out on spectra measured on 3 different days. Each trial carried out triplicate measurements on a new aliquot of a representative, washed GXP-HRas complex. Shown for each trial are the measured complex spectrum (solid line), the best fit convoluted spectrum (dot-dash-dot line) and the deconvoluted component nucleotide and protein spectra (dotted and dashed lines, respectively). The deconvolution procedure (Supplemental Protocol 3) utilized a GTP Reference spectrum and either a Subtracted (A) or Synthetic (C) HRas Reference spectrum. The 3 trials are virtually indistinguishable. (B,D) Overlay of 3 deconvolutions carried out on spectra measured for 3 different washed GXP-HRas complexes: standard HRas and the point mutants G12S and D38E (black, blue and red, respectively). Shown for each complex are the measured spectrum (solid line), the best fit convoluted spectrum (dot-dash-dot line) and the deconvoluted component nucleotide and protein spectra (dotted and dashed lines, respectively). The deconvolution procedure (Supplemental Protocol 3) utilized a GTP Reference spectrum and either a Subtracted (B) or Synthetic (D) HRas Reference spectrum. The spectra shown for the standard GXP-HRas complex are the average of 9 replicates acquired as 3 triplicates on 3 different days, while the G12S and D38E mutant GXP-HRas complexes are the average of 6 replicates acquired as 2 triplicates on different days. The 3 proteins yielded virtually indistinguishable results. For each panel, spectra were measured under the same conditions as Figure 1 and were overlaid using the measured GXP–HRas complex spectrum obtained for standard HRas (black) as the anchor.

To validate the method, we first generated washed standard GXP-HRas complexes and measured their protein and nucleotide concentrations by UV deconvolution as described in Supplemental Protocols 1-3. Then for each washed standard GXP-HRas complex we obtained an independent measurement of the protein concentration by amino acid analysis (AAA), or of the nucleotide concentration by high performance liquid chromatography (HPLC). Excellent agreement was obtained between the protein concentrations obtained by UV deconvolution and AAA. For 3 washed standard GXP-HRas complexes, on average the HRas protein concentration obtained by UV deconvolution was negligibly higher (by 1.7% ± 1.6%, mean ± standard deviation, n = 3) than that obtained by AAA (Supplemental Data Part I). Thus the two measurements were the same within error. This finding supports the accuracy and precision of the HRas protein concentration measurement provided by the UV deconvolution method. Further, the finding supports the accuracy of the molar extinction coefficient used to determine the HRas concentration in Supplemental Protocol 3 (ε₂₈₀ = 19,370 M⁻¹cm⁻¹, calculated for the standard HRas primary structure by the ExPASy ProtParam tool [18]).

Similarly, for 6 washed standard GXP-HRas complexes each measured in triplicate, the protein concentration obtained by UV deconvolution was negligibly higher (by 1.8% ± 3.4%, mean ± SD, n = 6) than the nucleotide concentration measured by HPLC, indicating the two concentrations were within error. These quantitative HPLC measurements of the total nucleotide concentration (Supplemental Data Part II) analyzed the nucleotide population extracted from the complex by heat denaturation as described above. The close correspondence between the protein and nucleotide concentrations in the washed GXP-HRas complex is consistent with the known 1:1 stoichiometry of guanine nucleotide binding to HRas and other Ras isoforms [19], as well as the picomolar to nanomolar nucleotide affinity [3–5] that accounts for the known ability of Ras isoforms to retain bound nucleotide during purification and washing [20–26].

We also compared the protein and nucleotide concentrations measured by UV deconvolution for 8 washed standard GXP-HRas complexes, each analyzed in triplicate. This comparison revealed modest variability in the nucleotide concentration measured by UV deconvolution. On average, the measured nucleotide concentration was 9% ± 8% higher (mean ± SD, n = 6) than the protein concentration, but was still within measurement error. This variability likely arises from incomplete removal of the protein precipitate following heat extraction of the nucleotides, since the suspended precipitate is expected to scatter strongly at the 252 nm nucleotide measurement wavelength due to the λ⁻⁴ dependence of scattering on wavelength, thereby artificially increasing the apparent nucleotide absorbance and concentration. Figure 1C illustrates an extreme example of such scattering.

Overall, the validation findings highlight both the key advantages and limitations of the UV deconvolution method. The method provides accurate (within ~2%) and precise measurement of the HRas protein concentration for washed GXP-HRas complexes (see above). As predicted by the known 1:1 nucleotide:protein stoichiometry and high affinity of the wild type guanine nucleotide binding site, the site remains fully nucleotide-loaded, within measurement error, during the washing procedure that provides >10⁵-fold net dilution. While the measured protein concentration is highly accurate, the measured nucleotide concentration and nucleotide:protein mole ratio are more approximate and subject to more variability (see above). Despite these limitations, replicate analysis can reveal major systematic differences between samples arising from incomplete removal of free nucleotide, or the presence of damaged protein lacking bound nucleotide. The method can also reveal the presence of absorbing or scattering contaminants as recognized by poor fitting with reference spectra during deconvolution. When a more accurate or precise nucleotide quantitation is desired, it is advisable to carry out HPLC analysis (Supplemental Data Part II) of the nucleotide population extracted from the GXP-HRas complex by heating and ultracentrifugation of the precipitated protein (Supplemental Protocol 2).

Finally, it is anticipated that the UV deconvolution method applied here to GXP-HRas complexes will be generally applicable to washed nucleotide complexes of other Ras superfamily members. We conclude with a discussion of additional advantages and avoidable caveats of the method. First, a major advantage of the subtracted Ras reference spectrum (Supplemental Protocol 2) is that it correctly accounts for spectral shifts that Ras protein chromophores experience in the folded protein environment of the GXP-Ras complex, because it is generated by subtracting the GXP spectrum from the measured spectrum of the native GXP-Ras complex. Second, the subtracted Ras reference spectrum conveniently includes an intrinsic correction for any spectral shift experienced by the guanine chromophore in the nucleotide-Ras complex, since the subtraction utilizes a guanine nucleotide spectrum measured in aqueous buffer, thereby leaving any guanine spectral shift behind in the subtracted Ras spectrum. Third, heating generates limited but measurable hydrolysis of endogenous guanine nucleotide polyphosphates (GDP or GTP), or extensive hydrolysis of a GTP analog, GMPPNP, commonly used to activate Ras proteins (Hannan, Swisher and Falke, data not shown); yet such hydrolysis causes little or no change to the GXP spectrum since the spectra of different non-cyclic guanine nucleotides are virtually indistinguishable in the relevant spectral range (Figure 1B and [9]. Fourth, it is important to employ adequate g-force and avoid disturbing visible or invisible pelleted material when centrifuging washed or heat-denatured samples, since the pelleted material will include protein aggregates and/or other particles that may scatter light and artifactually increases the UV absorbance, especially at the low wavelength end of the spectrum (Figure 1C). Fifth, to avoid circular analysis, the subtracted Ras reference spectrum should not be used to analyze the same sample of GXP-Ras complex used to generate the reference spectrum, but instead should be used to analyze independent samples. Sixth, when applying the procedure to mutants of the same Ras protein, care should be taken that the subtracted Ras reference spectrum employed is valid for the mutants of interest. Surface mutations that do not alter the protein conformation or the sidechain chromophore population will often have little impact on the subtracted Ras reference spectrum and the deconvolution, as illustrated for two HRas mutants (Figures 1D, and 2A,B). However, this assumption should be tested by generating a subtracted reference spectrum for each mutant to ascertain whether it matches or differs from the subtracted reference spectrum of standard HRas.

Abbreviations:

GTP

guanosine-5’-triphosphate

GDP

guanosine-5’-diphosphate

GMP

guanosine-5’-monophosphate

GMPPNP

GMP-PNP, guanosine-5’-[β,γ-imido]triphosphate

GMPPNH₂

guanosine-5’-[β-amino]-diphosphate

GXP

mixture of guanine nucleotides

HRas

H isoform of human Ras protein