Anal. Chem. 2007, 79, 9531-9538
Quantification of Mutant versus Wild-Type Myosin in Human Muscle Biopsies Using Nano-LC/ESI-MS Edgar Becker,† Francisco Navarro-Lo´pez,‡ Antonio Francino,‡ Bernhard Brenner,† and Theresia Kraft*,†
Molecular and Cell Physiology, Medical School Hannover, Carl-Neuberg-Strasse 1, D-30625 Hannover, Germany, and Barcelona Cardiovascular Institute, Hospital Clinic, University of Barcelona, Barcelona, Spain
A liquid chromatography/electrospray ionization mass spectrometry (nano-LC/ESI-MS) approach is described by which abundance of proteins (e.g., of β-myosin heavy chain; MW 223 kDa) carrying a point mutation can be determined in tissue samples where the mutant protein is coexpressed with its wild-type forms. After enzymatic cleavage of the extracted parent protein, mutant and wildtype species of the peptide with the locus of the point mutation were quantified. Synthetic peptides, identical to wild-type and mutant peptides but labeled with stable isotopes (13C, 15N), were added in known amounts as internal standards. The peak areas obtained by MS for the stable-isotope-labeled peptides and for the native peptides were used for quantification. To demonstrate the suitability of this approach we determined the relative abundance of β-myosin with the Arg723Gly exchange in muscle biopsies of patients with Familial Hypertrophic Cardiomyopathy (HCM). For two such patients the fraction of mutated myosin was 62%, i.e., significantly different from 50%, which is quite unexpected for an autosomal dominant disease in heterozygous patients. Correlation between abundance of mutant myosin and clinical malignancy seen for several mutations in the myosin head domain emphasizes the relevance of such quantification. The approach for quantification described here is generally applicable for quantification of proteins with single point mutations even if only small amounts of tissue are available. Many human diseases are caused by mutations which manifest in a single amino acid exchange in a protein. The dysfunction of this protein then triggers the development of the disease phenotype. One example for such a disease is Familial Hypertrophic Cardiomyopathy (HCM),1 which in most cases is caused by point mutations in sarcomeric proteins. HCM is an autosomal dominant inherited disease. Thus, in heterozygous patients mutant and wildtype forms of the affected protein are expected to be present at a 1:1 ratio. Yet, in a previous study on muscle biopsies of HCM patients carrying either mutation Val606Met or Gly584Arg in the * Corresponding author. Phone: +49-(0)511-532-2734. Fax: +49-(0)511-5324296. E-mail:
[email protected]. † Medical School Hannover. ‡ University of Barcelona. (1) Seidman, J. G.; Seidman, C. Cell 2001, 104, 557-567. 10.1021/ac701711h CCC: $37.00 Published on Web 11/17/2007
© 2007 American Chemical Society
cardiac β-myosin heavy chain (β-MHC), only 12% and 23% of myosin were found in the mutated form, respectively.2 Clinical evaluation of patients with a mutation in heart muscle myosin suggests a correlation between abundance of mutated myosin in the muscle tissue and severity of the symptoms of the respective patient.3 However, to accurately determine the ratio of mutated versus wild-type form of a large protein like β-MHC, when both forms of the protein are present within the same sample, common methods such as Western blot analysis or immunoassays are not applicable. β-MHC has a molecular mass of 223 kDa and a single amino acid exchange, causing a mass shift of about 100 Da, cannot be detected by a shift in, e.g., electrophoretic mobility. In our previous studies, capillary zone electrophoresis (CE) of myosin peptides was used as the method for quantification.2 This method, however, is not universally applicable, as for most myosin mutations baseline separation of individual peptides could not be achieved by CE. Thus, further clarification of a possible correlation between disease phenotype and abundance of mutated protein required a generally applicable method to determine the ratio of mutated protein versus the wild-type protein, when both are present within the same sample of human tissue. Such a method is of great interest since a correlation between the amount of mutated protein and severity of the disease appears to be important for clinical evaluation of the patients and for assessment of the disease prognosis. Furthermore, such a technique can provide a valuable diagnostic parameter for other inherited diseases. In recent years, many techniques have been developed which employ stable-isotope-ratio mass spectrometry (SIRMS) for comparative analysis of proteins in different samples derived from biological tissues and body fluids.4-8 Yet, these methods cannot be used to quantify the absolute or the relative abundance of a protein and its mutated, phosphorylated, or otherwise modified states when present together within the same sample. A few years ago the absolute quantification strategy, termed AQUA, was (2) Nier, V.; Schultz, I.; Brenner, B.; Forssmann, W.; Raida, M. FEBS Lett. 1999, 461, 246-252. (3) Becker, E.; Schulte, I.; Nier, V.; Ko¨hler, J.; Perrot, A.; Osterziel, K. J.; McKenna, W.; Kraft, T. Eur. Heart J. 2003, 24, 588. (4) Aebersold, R.; Mann, M. Nature 2003, 422, 198-207. (5) Sechi, S.; Oda, Y. Curr. Opin. Chem. Biol. 2003, 7, 70-77. (6) Steen, H.; Mann, M. Nat. Rev. Mol. Cell Biol. 2004, 5, 699-711. (7) Schneider, L. V.; Hall, M. P. Drug Discovery Today 2005, 10, 353-363. (8) Gingras, A. C.; Aebersold, R.; Raught, B. J. Physiol. 2005, 563, 11-21.
Analytical Chemistry, Vol. 79, No. 24, December 15, 2007 9531
introduced.9-11 This technique allows absolute protein quantification by (i) proteolysis of the parent protein and by (ii) adding to the digest known amounts of a synthetic stable-isotope-labeled internal standard peptide mimicking the peptide of interest. In the work presented here we describe a new approach in which synthetic stable-isotope-labeled internal standard peptides are used for relative quantification of a protein carrying a point mutation versus its wild-type form, both present within the same sample. As an example we determined the fraction of myosin heavy chain molecules with the mutation Arg723Gly in small muscle biopsies of two HCM patients. The aim of the present study was to determine, based on the abundance of mutated and wild-type peptide in the samples measured by MS, the ratio of mutated/ total β-MHC in each patient’s biopsy. The question was whether both forms of the protein, mutated and wild-type, are present at a 1:1 ratio in the muscle tissue. The quantification procedure involved selective extraction of myosin from the muscle biopsies, followed by a Lys-C digest. Peptides were analyzed using nano-LC coupled electrospray mass spectrometry (nano-LC/ESI-MS). For quantification, known amounts of synthetic peptides of wild-type and mutant sequence containing a stable-isotope-labeled leucine (13C, 15N) were added to the peptide mixture as internal standards (IS). In muscle samples of two severely affected HCM patients with the β-MHC mutation Arg723Gly, the fraction of mutated myosin was found to be nearly twice as large as the fraction of wild-type myosin. This further supports our hypothesis that phenotype and severity of clinical symptoms in HCM patients, among other factors, may also depend on the amount of mutated myosin expressed and incorporated into the myofilaments. The method presented here is generally applicable to quantification of proteins with point mutations if peptides with wild-type and mutant sequence suitable for quantification by MS can be obtained. The method allows to address questions regarding abundance of mutated proteins in small samples of human or animal tissue. EXPERIMENTAL SECTION Selective Extraction of Myosin from Muscle Biopsies of HCM Patients. Muscle tissue was obtained in open biopsy from the soleus muscle of two HCM patients who had previously been clinically characterized and genotyped.12 Both patients were from one family and carried mutation Arg723Gly in the cardiac β-MHC, which in humans is also expressed in slow skeletal muscle fibers such as in M. soleus. Informed consent of the patients was obtained according to approved Ethics Committee protocols of the involved clinical center. The preparation and permeabilization of the muscle tissue by Triton-X-100, freezing procedure, and storage in liquid nitrogen were described previously.13,14 For efficient myosin extraction, the frozen muscle tissue was first crushed into small pieces. In this study we used 70-100 mg (wet weight) of M. soleus tissue which (9) Stemmann, O.; Zou, H.; Gerber, S. A.; Gygi, S. P.; Kirschner, M. W. Cell 2001, 107, 715-726. (10) Gerber, S. A.; Rush, J.; Stemman, O.; Kirschner, M. W.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6940-6945. (11) Kirkpatrick, D. S.; Gerber, S. A.; Gygi, S. P. Methods 2005, 35, 265-273. (12) Enjuto, M.; Francino, A.; Navarro-Lopez, F.; Viles, D.; Pare, J. C.; Ballesta, A. M. J. Mol. Cell. Cardiol. 2000, 32, 2307-2313. (13) Kraft, T.; Messerli, M.; Rothen-Rutishauser, B.; Perriard, J. C.; Wallimann, T.; Brenner, B. Biophys. J. 1995, 69, 1246-1258.
9532
Analytical Chemistry, Vol. 79, No. 24, December 15, 2007
allowed for more than 60 parallel quantifications. The crushed tissue was incubated for 15 min at 4 °C in myosin extraction buffer (modified Hasselbach-Schneider solution15 containing 2.5 mM ATP, 500 mM NaCl, 10 mM Hepes, 5 mM MgCl2, 35 mM DTT). Extraction was terminated by centrifugation of the sample, and the supernatant with the extracted myosin was collected. To improve yield the pellet was resuspended in extraction buffer for a second extraction period of again 15 min, followed by centrifugation. Supernatants of the first and second extract were diluted with 10 volumes of double-distilled water to decrease ionic strength to 1/10 of the extraction buffer. At this low ionic strength myosin precipitates. Myosin precipitation was allowed for 48 h at 4 °C. After precipitation the supernatant was discarded, and the myosin sediment was resuspended in 100 µL of 0.1% TFA (J. T. Baker, Griesheim, Germany) and lyophilized. This procedure yielded an extract of myosin heavy chain together with the two light chains and only traces of other muscle proteins, as shown earlier by SDS gel electrophoresis.2 Digestion of Myosin and Prefractionation of the Peptides Prior to Nano-LC/ESI-MS Analysis. Myosin was digested as described previously.2 Briefly, the myosin extracts were solubilized in 200 µL of 0.1 M NH4HCO3 containing 5 µg of endoproteinase Lys-C (Roche, Mannheim, Germany) and incubated in a thermomixer for 24 h at 37 °C and 600 rpm. After the first digestion, the content of the test tube was lyophilized and redissolved in 200 µL of new digest solution (5 µg of Lys-C in 200 µL of 0.1 M NH4HCO3), incubated for another 24 h, and subsequently lyophilized. For a control experiment with triple digestion of β-MHC, the digestion procedure was repeated once more. To reduce the number of peptides in the sample prior to quantification by nano-LC/ESI-MS, the Lys-C digests were fractionated via reversed-phase HPLC (RP-HPLC) (Merck/Hitachi, Darmstadt, Germany) using a C-18 monomeric reversed-phase column (Vydac 238TP52, 250 mm × 2.1 mm, 5 µm, 300 Å; MZ Analysentechnik GmbH, Mainz, Germany). The lyophilized myosin digest was dissolved in 20 µL of 0.1% TFA (v/v) and 20% ACN (v/v) (J. T. Baker, Griesheim, Germany) in water. A linear gradient from 15% solution B (40% ACN and 0.085% TFA in H2O (v/v)) and 85% solvent A (0.1% TFA in H2O (v/v)) to 85% solution B within 160 min was generated. The column temperature was maintained at 35 °C. The flow rate was 200 µL/min, and 1 min fractions were collected in siliconized reaction tubes. The difference between mutant and wild-type β-MHC for the mutation examined here is an arginine to glycine exchange (Table 1). This causes a loss of a positive charge, an increase in hydrophobicity, and a molecular mass difference of 99.08 Da, resulting in different elution times by RP-HPLC. Synthetic, stable-isotope-labeled peptides (Coring System Diagnostix GmbH, Gernsheim, Germany) with wild-type and mutant sequence (Table 1), respectively, were used to determine the retention time for the mutant and wildtype peptide before and after each run. Part of a typical RP-HPLC chromatogram is shown in Figure 1. The fractions including both mutant and wild-type peptide (horizontal arrow in Figure 1B) were pooled. The pooled material was lyophilized, resuspended in (14) Kirschner, S. E.; Becker, E.; Antognozzi, M.; Kubis, H. P.; Francino, A.; Navarro-Lopez, F.; Bit-Avragim, N.; Perrot, A.; Mirrakhimov, M. M.; Osterziel, K. J.; McKenna, W. J.; Brenner, B.; Kraft, T. Am. J. Physiol.: Heart Circ. Physiol. 2005, 288, H1242-H1251. (15) Hasselbach, W.; Schneider, G. Biochem. Z. 1951, 321, 462-475.
Table 1. Sequence, Molecular Mass, and m/z Ratio of the +5-Fold Charged Ions of the Native Form of Wild-Type and Mutant Peptides and of the Stable-Isotope-Labeled IS Peptides sequence and target peptide ions [M + 5H+] for quantificationa
deconvoluted av mass [Da] of native peptides
wild-type peptide
GFPNRILYGDFRQRYRIL*NPAAIPEGQFIDSRK native ion for quantification: m/z 782.89 isotope-labeled ion for quantification: m/z 783.89
3909.47
mutant peptide
GFPNRILYGDFRQRYGIL*NPAAIPEGQFIDSRK native ion for quantification: m/z 763.07 isotope-labeled ion for quantification: m/z 764.07
3810.33
a L* indicates the leucine which in the internal standard peptides was replaced by an isotopically enriched leucine (13C, 15N). The target peptide ions were analyzed using extracted ion chromatograms with a width of m/z ( 0.7.
Figure 1. Comparison of single and double Lys-C digests. Corresponding segments of RP-HPLC runs of digest samples after the first (A) and second (B) step of a double Lys-C digest of a β-MHC extract. The equivalent segment of another HPLC run of synthetic stableisotope-labeled IS peptides alone is superimposed to illustrate identification and retention times of wild-type (WT) and mutant (MT) myosin peptides. The horizontal arrow indicates the fraction of the eluate collected for further analysis by nano-LC/ESI-MS.
80 µL of water with 0.1% TFA (v/v) and 20% ACN (v/v), frozen, and stored at -20 °C for subsequent analysis by nano-LC/ESIMS. Peptides for Quantification and Peptide Standards. Table 1 shows the amino acid sequence of mutant and wild-type forms of the β-MHC 708-740 peptide that are generated by Lys-C digestion. Table 1 also shows the molecular mass of both mutant and wild-type peptides in their native state. Mutant and wild-type peptides differ at position 723 where in the mutant peptide arginine is replaced by glycine. For the native peptides this exchange results in a molecular mass of 3909.47 Da for the wild-type and 3810.33 Da for the mutant peptide. A computer-aided model digestion of myosin and a search within the Swiss protein database (www.expasy.org) showed that the two peptides are specific for human β-MHC. Thus, they are well-suited for quantitative analysis by MS. For quantification, stable-isotope-labeled IS peptides with sequences of the mutant and wild-type peptides were used. These
peptides were synthesized with leucine 725 in the sequence of each peptide (Table 1) being replaced by an isotopically enriched (13C, 15N) leucine (Coring System Diagnostix GmbH, Gernsheim, Germany). Heavy isotope labeling resulted in a difference in average mass between native and stable-isotope-labeled peptides of 5 Da, i.e., 3914.5 Da for the wild-type and 3815.3 Da for the mutant stable-isotope-labeled IS peptide. For both stable-isotopelabeled peptides purity was verified by HPLC. The masses of the isotope-labeled IS peptides were measured by MALDI-TOF-MS (Figure 2A) and confirmed by nano-LC/ESI-MS (Figure 3; +5fold charged peptides). The net peptide content for both peptides was >95%. The precise concentration of the IS peptides was determined by spectrophotometric analysis (molar extinction coefficient E280nm ) 2560 M-1 cm-1). Carefully measured aliquots of 0.5-5 pmol/ 10 µL of the stable-isotope-labeled IS peptides were added to the HPLC fractions of the myosin digests to be carried through all further steps of the MS analysis. Throughout all procedures of extraction, digestion, and final preparation of the samples, including handling of the IS peptides for quantification, it was essential to use siliconized pipet tips and reaction vials (Biozym, Hessisch Oldendorf, Germany) to minimize adsorption of myosin16 and myosin peptides to plastic surfaces. In addition, for highly reproducible signals it was necessary to use solutions containing ACN (solvent: 0.1% TFA (v/v) and 20% ACN (v/v) in water). ACN was added after the Lys-C cleavage of the myosin extract to avoid adverse effects of ACN on the cleavage by Lys-C. LC/MS Conditions. The ratio of mutant versus wild-type peptides in the various digests was determined at the Proteome Factory AG, Berlin, Germany. A nano-LC/ESI-MS approach with an Agilent 1100 nano-LC system (Agilent Technologies, Waldbronn, Germany) connected to a nanoelectrospray mass spectrometer (Bruker Esquire3000plus MS, Leipzig, Germany) with a PicoTip emitter (New Objective) was used. For each run 8 µL of the prefractionated sample containing native mutant and wildtype peptides as well as the IS peptides was injected via an Agilent 1100 Micro Well-Plate sampler and enriched on a C-18 RP-HPLC column (Agilent Zorbax 300SB-C18, 5 µm, 5 mm × 0.3 mm) for 5 min and subsequently switched to a separation column (Agilent Zorbax 300SB-C18, 3.5 µm, 150 mm × 0.075 mm). The separation column was directly connected to the nano-ESI source of the MS. The flow rate on the separation column was 300 nL/min. For (16) Thedinga, E.; Karim, N.; Kraft, T.; Brenner, B. J. Muscle Res. Cell Motil. 1999, 20, 785-796.
Analytical Chemistry, Vol. 79, No. 24, December 15, 2007
9533
Figure 2. MS spectra of stable-isotope-labeled IS peptides, total ion chromatogram (TIC), and extracted ion chromatograms (EICs) of native peptide mixture spiked with IS peptides. (A) MALDI-TOF-MS spectra of wild-type (WT) and mutant (MT) stable-isotope-labeled IS peptides are shown. Heavy isotope labeling (13C, 15N) of one leucine in the sequence of each peptide results in a 5 Da mass difference between native and IS peptides. (B) TIC of a peptide mixture (fraction of a Lys-C digest as indicated by the double arrow in Figure 1) spiked with equimolar amounts of the two stable-isotope-labeled IS peptides and applied to nano-LC/ESI-MS. Also shown are the EICs of native and stable-isotope-labeled forms of the wild-type and mutant peptides of interest. Inset: expanded EICs of wild-type peptide (Nat. WT), mutant peptide (Nat. MT), and their corresponding IS peptides (IS WT and IS MT).
Figure 3. MS spectra of native and stable-isotope-labeled β-MHC peptide pairs. Sections of representative mass spectra which include (A) 5-fold charged native wild-type peptide (m/z 782.8) and wild-type IS peptide (m/z 783.9) and (B) 5-fold charged native mutant peptide (m/z 763.1) and mutant IS peptide (m/z 764.1). The spectra were recorded in high-resolution mode with a resolution of m/z 0.25. In both cases peaks of native and IS peptides are well-separated.
peptide separation an increasing, shallow ACN gradient was used. The gradient started at 22% ACN with 0.1% (v/v) formic acid in H2O and ended at 90% ACN (ACN gradient: 0 min, 22%; 10 min, 27.1%; 27 min, 30.5%; 30 min, 90%; 35 min, 90%). The shallow gradient was necessary to separate wild-type and mutant peptides. The MS spectra were recorded in maximum resolution mode at 1650 m/z s-1 within a MS scan range of m/z 600-800. The ICC target was set to 20 000 with a maximum accumulation time of 100 ms. Four average MS spectra were taken at positive polarity with the following parameters: capillary voltages, 1600 V; dry temperature, 190 °C. The target peptide ions were quantified based on extracted ion chromatograms with a width of m/z ( 0.7 using the software Bruker Daltonics QuantAnalysis 1.6. 9534 Analytical Chemistry, Vol. 79, No. 24, December 15, 2007
RESULTS AND DISCUSSION Extraction of Myosin from Human Muscle Tissue. The aim of the present study was to determine the relative abundance (fraction) of mutated β-MHC in muscle biopsies from HCM patients. To selectively isolate myosin from the muscle tissue, a well-established procedure yielding myosin in solution was used instead of gel electrophoretic separation. First, chemical permeabilization of muscle fiber membranes was used to allow all soluble proteins, including any free myosin not integrated in the contractile apparatus, to diffuse out of the muscle fibers. Treatment of the permeabilized fibers with a high salt solution results in disaggregation of the myosin filaments into free myosin molecules. MgATP, also contained in the high salt solution, causes dissociation of the myosin from actin filaments and diffusion of the free myosin out of the muscle fibers.2,15 This treatment thus provides quite selective extraction of myosin only, whereas more rigorous extraction would have resulted in increasing contamination by other sarcomeric proteins. A final precipitation step of the extracted myosin at low salt concentration further reduced contaminating proteins.2 The efficiency and purity of the extraction procedure was assessed by SDS gel electrophoresis in comparison to unextracted samples. This revealed that approximately 6070% of the whole myosin was extracted. As shown previously2 the extract contained essentially no actin and only traces of other sarcomeric proteins. Since the myosin extraction from the muscle tissue was quite specific, additional purification steps were not necessary. The aim of the present study was not to determine the absolute abundance of mutated myosin in the muscle biopsies so that it was not essential to take into account the precise amount of tissue from which myosin was extracted. Instead, the goal was to determine the fraction of total β-MHC incorporated into the contractile apparatus in its mutated form. Thus, we first permeabilized the muscle tissue to remove soluble proteins before extraction of the myosin incorporated into the contractile ap-
paratus. In other diseases, however, mutant proteins may also form insoluble aggregates so that suitable extraction methods will be required to determine the ratio of mutated/wild-type protein in the different cellular compartments. Peptide Sample Preparation for Nano-LC/ESI-MS Analysis. To generate a peptide that includes the locus of the point mutation and which is of suitable size to distinguish and quantify mutant versus wild-type forms by mass spectrometry we used the endoproteinase Lys-C for digestion of the full length myosin (223 kDa). In a Lys-C digest (i) the peptide containing the mutated amino acid is of a suitable size (33 amino acids; Table 1), and (ii) the cleavage sites for Lys-C are not located at the site of the mutation. To optimize enzymatic cleavage, the Lys-C digestion was done in two steps (cf., the Experimental Section). Quantification of mutant and wild-type peptides was not carried out within the entire mixture of 197 different peptides which are predicted by a computer-simulated Lys-C cleavage of the sequence of β-MHC. Instead, prior to analysis by nano-LC/ESI-MS, the digest was prefractionated by RP-HPLC. The advantage of the prefractionation step was that it reduced the number of peptides present in the sample for nano-LC/ESI-MS by approximately 90%. With only about 20 different peptide species in the sample it was possible to increase the amount of mutated and wild-type peptides loaded onto the nano-LC/ESI-MS 10-fold without exceeding the peptide binding capacity of the LC column connected to the ESIMS system, thus improving the signal/noise ratio. The width of the RP-HPLC fraction collected for subsequent quantification by nano-LC/ESI-MS was set such that both wild-type and mutant peptide were well-contained to avoid errors in the relative abundance of mutant versus wild-type peptide due to possible differences in handling of samples. It should be noted that it was not possible to directly determine the ratio of wild-type to mutant peptide by UV detection because baseline separation from peaks of other peptides could not be achieved. Very similar HPLC chromatograms of peptide mixtures of single and double digestions (Figure 1) suggest reproducible cleavage by Lys-C. To further exclude that our results are affected by differentially incomplete generation of mutant and wild-type peptide by Lys-C, we compared the ratio of mutant versus wildtype peptides in double versus triple digestion of β-MHC. With a third digestion period essentially the same fraction of mutant peptide was found as in double digestions (63.4% ( 9.0% vs 64.4% ( 3.0% for double and triple digestion, respectively). Calibration Curves and Quantification of Mutant and Wild-Type Myosin. In an initial test with synthetic mutant and wild-type peptides we noticed that addition of equal amounts of the synthetic peptides to the myosin digest, i.e., to a peptide matrix, did not result in same peak areas in the mass spectrometry that were seen when analyzing synthetic peptides alone. Thus, to determine the relative abundance of mutant and wild-type myosin peptides within the fraction of the HPLC chromatogram used for mass spectrometry (cf., Figure 1) the nano-LC/ESI-MS needed calibration by suitable internal standard peptides; i.e., peptides that coelute from the LC column at exactly the same time as the native peptides. We therefore used IS peptides in which one leucine was replaced by a stable-isotope-enriched leucine (13C, 15N), resulting in a 5 Da mass difference between native and IS peptides (Table 1, Figure 2A).
In Figure 2B the total ion chromatogram (TIC) of an RP-HPLC fraction collected for quantification during the prefractionation step is shown together with the extracted ion chromatograms (EIC) of mutant and wild-type forms of native and internal standard peptides. The inset shows the EICs on an enlarged scale, i.e., the EIC of the +5-fold charged ions of native wild-type peptide with m/z 782.89, the native mutant peptide with m/z 763.07, as well as the EIC of the stable-isotope-labeled IS peptides with m/z 783.89 for the wild-type form and m/z 764.07 for the mutant (Table 1). Representative mass spectra of the +5-fold charged ions ([M + 5H+]) of stable-isotope-labeled and native peptide pairs are shown in Figure 3. For both pairs of peptides, wild-type native and IS peptides (Figure 3A) and mutant native and IS peptides (Figure 3B), respectively, the peaks are 1.0-1.1 m/z apart and sufficiently resolved for quantitative analysis. The remaining overlap between wild-type native and IS peptide peak (Figure 3A) results in an uncertainty of at most 2-3% for integrated peak areas. Quantitative analysis was carried out using Bruker Daltonics QuantAnalysis 1.6 software. Analysis of the average mass spectra showed that equimolar amounts of the two IS peptides do not generate ion traces with equal peak areas. This resulted in different slopes when calibration curves were generated with the IS peptides (Figure 4). Several factors could contribute to the different peak areas and thus different slopes of calibration curves for the two IS peptides: For instance, the difference in sequence (one amino acid exchanged) between mutant and wild-type peptide, or because of different retention times of mutant and wild-type peptide on the LC column, different peptides of the Lys-C digest coelute with the mutant and wild-type peptides. As a consequence, different peptides will compete with mutant and wild-type peptide in the ESI process. Control experiments were performed with LC/ESI-MS to clarify this point. Mutant and wild-type synthetic peptides were analyzed (1) in 0.1% TFA/20% ACN (v/v) alone or (2) in 0.1% TFA/20% ACN (v/v) containing the respective peptide fraction of the β-MHC digest. We found that when the calibration curves were generated with IS peptides in TFA/ACN alone, peak areas for known amounts of wild-type and mutant peptides differed significantly from each other (data not shown). This difference most likely is due to the single amino acid exchange between the two peptides, presumably causing different ionization efficiency of the two peptides. When equal amounts of IS peptides were analyzed in the Lys-C digest fraction, calibration curves of the IS peptides were not only different between mutant and wild-type peptide (Figure 4) but were also different from the calibration curves of the two IS peptides recorded in TFA/ACN alone. This indicates that not only the difference in sequence between mutant and wild-type peptide but also ion suppression effects strongly influence the peak areas and thus the calibration curves. From the distinct slopes of the IS calibration curves generated in the presence of the specific peptide matrix we conclude that relative quantification by just using the integrated peak areas of the native peptide ion signal, i.e., without IS, would be erroneous. It is also clear from these observations that for reliable quantification of mutant and wild-type peptide, calibration curves must be generated with IS peptides added to the particular peptide fraction of the Lys-C digest instead of added to TFA/ACN alone. This emphasizes the importance of compensating for matrix effects by Analytical Chemistry, Vol. 79, No. 24, December 15, 2007
9535
Figure 4. Quantification of native wild-type (A) and mutant (B) β-MHC peptides by synthetic stable-isotope-labeled IS peptides. Open circles and solid lines are calibration curves generated with the IS peptides. Amounts of 0.5-5 pmol of IS peptides were added to 10 µL samples of the HPLC fraction of the Lys-C digest. Integrated peak areas of the +5-fold charged ions are plotted vs the added amounts of IS peptides. Solid lines are linear least-squares fits to the open symbols (R2wild-type ) 0.988, R2mutant ) 0.990). With the use of the two calibration curves, the amounts of native mutant and wild-type peptides in each sample were determined from the integrated peak areas of their +5-fold charged ions (filled circles). Note that for both stable-isotope-labeled peptides calibration data are well fit by a linear function that extrapolates essentially through the origin. The slope in (A), however, is about 1.5 times larger than the slope in (B).
matching the matrix in calibration standards to that in samples, as is common in regulated biopharmaceutical analysis.17,18 Thus, ideal internal standards for quantification of the native peptides are synthetic peptides that have different mass but the same physicochemical properties as the native peptides so that they elute from HPLC at exactly the same time. This requires synthesis of peptides that differ from the native peptides only by incorporation of amino acids labeled with stable-isotopes such as 13C or 15N atoms, which do not cause a shift in retention time in HPLC. As a consequence, we based our quantification of native and mutant peptides on precisely measured amounts of stable-isotopelabeled mutant and wild-type IS peptides added to the sample before processing the peptide fraction. As indicated in Figure 4, using different amounts of IS peptides we obtained linear calibration curves over at least one order of magnitude as the dynamic range for determination of the amount of native peptide in the sample fraction. The graph shows examples of peak areas for native wild-type and mutant peptides in the sample fractions which are well within the range of the calibration curves. For the peptides used here the y-axis intercepts of the linear calibration curves, on average, were not significantly different from zero (Figure 4). Therefore, the relative amount of mutant versus wild-type β-MHC could also be quantified from single samples to which both IS peptides were added (“single-point quantification”), as long as the peak areas of the native peptides were within the same order of magnitude as the added IS standards (linear range of calibration curves). From either calibration curves or, in single-point quantification, from the peak areas of native and IS peptides the fractions (F) of mutant and wild-type (WT) β-MHC within the sample were calculated: Fmutated ) amount mutated/(amount mutated + amount WT) FWT ) amount WT/(amount mutated + amount WT) (17) Karnes, H. T.; Shiu, G.; Shah, V. P. Pharm. Res. 1991, 8, 421-426. (18) Shah, V. P.; Midha, K. K.; Findlay, J. W.; Hill, H. M.; Hulse, J. D.; McGilveray, I. J.; McKay, G.; Miller, K. J.; Patnaik, R. N.; Powell, M. L.; Tonelli, A.; Viswanathan, C. T.; Yacobi, A. Pharm. Res. 2000, 17, 1551-1557.
9536
Analytical Chemistry, Vol. 79, No. 24, December 15, 2007
Application to Biopsies of Two HCM Patients. With the method described here we determined the fraction of mutant β-MHC in biopsies of two HCM patients with the Arg723Gly mutation. Quantification was based on calibration curves such as those shown in Figure 4. For the first patient 61.6% ( 6.5% (n ) 36 samples; mean ( SD) and for the second patient 62.4% ( 6.9% (n ) 31 samples) of total myosin was found to contain the mutation. The precision of the method was in the range of 1011% relative standard deviation determined from the 31 and 36 quantifications, respectively. We also compared single-point quantifications with the quantifications based on calibration curves. For average values of myosin in the mutated form we obtained 61.6% ( 6.5% versus 63.5% ( 8.6% (36 samples tested with both types of approaches) and 62.4% ( 6.9% versus 60.9% ( 6.7% (31 samples) by quantification via calibration curves versus singlepoint quantification, respectively. Both patients have significantly (p < 0.001) more than 50% mutated β-MHC which is quite surprising for heterozygous patients with an autosomal dominant disease. On the other hand, a deviation from the expected 1:1 ratio of mutant versus wildtype myosin is consistent with our previous CE quantification study, where we found only 12% ( 6% and 13% ( 2% mutated myosin for two unrelated patients with mutation Val606Met and only 23% ( 1% mutated myosin for a patient with mutation Gly584Arg.2 Furthermore, in a preliminary quantification of mutated myosin in an HCM family with mutation Ile736Thr19 a fraction of 39.3% ( 5.5% mutated myosin was found.20 Moreover, in the family with mutation Ile736Thr this fraction was very similar for all three members of which tissue samples were available to us (details will be published elsewhere). Hence, in all cases studied the fraction of mutated β-MHC is very similar for HCM patients with the same mutation but quite different for different (19) Perrot, A.; Schmidt-Traub, H.; Hoffmann, B.; Prager, M.; Bit-Avragim, N.; Rudenko, R. I.; Usupbaeva, D. A.; Kabaeva, Z.; Imanov, B.; Mirrakhimov, M. M.; Dietz, R.; Wycisk, A.; Tendera, M.; Gessner, R.; Osterziel, K. J. J. Mol. Med. 2005, 83, 468-477. (20) Tripathi, S.; Becker, E.; Navarro-Lopez, F.; Brenner, B.; Kraft, T. J. Muscle Res. Cell Motil. 2006, 27, 538.
mutations. Based on this observation we conclude that the deviation from a 50% fraction of mutated myosin is similar among individuals with the same mutation, i.e., seems to depend on the specific mutation. The large deviation from the expected 1:1 ratio between mutated and wild-type β-MHC, however, most likely reflects the biological situation in the patients and does not result from the protein purification and quantification procedure. Mutant and wildtype proteins are within the same sample from the very beginning and therefore are treated identically throughout the whole procedure. The exchange of only one out of 1935 amino acids of the myosin molecule is not expected to affect the extraction properties of the protein, in particular as the mutation is located in the myosin head domain and not in the tail which mediates aggregation of the myosin filaments in each sarcomere. Thus, extraction of myosin molecules by disaggregation of myosin filaments at high salt concentration should not be affected by the mutation. Accordingly, we did not find significant differences in wild-type/mutant ratio between first and second myosin extracts (cf., the Experimental Section). As the ratio of mutated versus wild-type myosin was the parameter of interest, it appeared not essential to achieve quantitative extraction of myosin from the tissue sample. To track losses and particularly changes in the relative abundance of wild-type and mutated peptides during the prefractionation and lyophilization steps, we subjected precisely measured amounts of stable-isotope-labeled IS peptides to HPLC runs as in the prefractionation step. This resulted in a slightly smaller peak for the wild-type IS peptide (93% ( 2%) compared to the mutant IS peak. If this deviation/loss of wild-type peptide during the HPLC prefractionation is taken into account when calculating the fraction of mutated myosin in the muscle samples, the fraction of mutated myosin is somewhat reduced (60% and 61% for the first and second patient, respectively, both values being well within the range of standard deviation). Nevertheless, the quantification method described here as well as the control experiments do not account for possible differences in proteolytic efficiency for mutated and wild-type peptides. The single amino acid that is exchanged between mutant and wild-type β-MHC is not part of the Lys-C cleavage site. Instead, Lys-C cleavage occurred at sites homologous between the two forms of the protein. Thus, preferential cleavage of the mutated form of the protein versus the wild-type form is highly unlikely. Incidentally, this is confirmed by our data from β-MHC-mRNA quantifications of the M. soleus biopsies of the two HCM patients20 with the Arg723Gly mutation. For both patients the β-MHC-mRNA shows a similar deviation from the expected 1:1 ratio between mutant and wild-type β-MHC-mRNA as observed here for the protein level (details will be published elsewhere). Diagnostic Application. As shown here and also in our previous study,2 the fraction of total β-MHC being mutated myosin in muscle biopsies from HCM patients varies quite widely from 12% in our previous study up to near 62% of total myosin found for mutation Arg723Gly. Importantly, for all mutations studied so far the fraction of mutated β-MHC seems to correlate with the clinical severity of the disease of the respective patients.2,3,20 Thus, quantification of the fraction of mutated protein appears to be a suitable approach for assessing the severity and, possibly, the
prognosis of the disease also for other carriers of the mutation within a family. When adapting our approach to other mutations, suitable proteases have to be used which generate peptides of sufficiently small size and with cleavage sites that do not coincide with the site of the mutation. Instead of solution-based hydrolysis, as an alternative possibility for fast and efficient digestion of a protein in solution, on-column digestion might be considered,21 which requires immobilization of the respective protease on a solid support. In addition, in some diseases where mutated proteins form insoluble aggregates, specific extraction methods will be required if the ratio of mutated/wild-type protein is to be determined in different compartments. One important parameter by which suitability for broader application of a quantification technique of mutated proteins in basic or clinical research will be judged is the minimum sample size necessary for analysis. Once linearity of the calibration curve is established by IS peptides added to the peptide mixture, approximately 1-2 mg of muscle tissue is sufficient for analysis of the fraction of mutated β-MHC in the sample. Some preliminary trials showed that with triplequadrupole MS and multiple reaction monitoring (MRM) quantitation of even smaller samples should be possible. CONCLUSIONS In this study we present a method to determine the relative abundance of mutated and wild-type forms of a protein (>220 kDa) present, side by side, in human muscle biopsies. Mutated and wild-type form differ by only a single amino acid, resulting in a molecular mass difference of 99 Da. For quantification by nanoLC/ESI-MS the protein was Lys-C digested. Stable-isotope-labeled synthetic mutated and wild-type peptides were suitable internal standards. For reliable and reproducible quantification it was essential to add the IS peptides directly to the peptide matrix. The different slopes of the calibration curves for the IS peptides when added to the protein digest indicate that not only for absolute but even for relative quantification of wild-type versus mutated peptides, the IS peptides need to be present in the peptide sample subjected to the MS quantification. The approach described here should be readily adaptable not only to all myosin mutations but also to point mutations in other proteins causing HCM or other diseases. The method is of particular interest for large proteins where wild-type and mutant species cannot be quantified by electrophoretic and immunochemical assays. The quantification method is not limited to the protein extraction protocol used here that is specific for muscle myosin. Instead, it could also be applied to proteins isolated by polyacrylamide gel electrophoresis followed by in-gel proteolytic digestion or to protein mixtures separated by liquid chromatography. A key requirement is that appropriate enzymatic cleavage sites are found which yield a pair of mutated and wild-type peptides suitable for quantification by MS, i.e., peptides of suitable size compared to the mass difference between mutant and wild-type peptide, and the sequence of which is unique for the protein of interest. ACKNOWLEDGMENT We are indebted to the HCM patients for their donation of muscle tissue. The authors thank Dr. K.-J. Osterziel, A. Perrot, (21) Slysz, G. W.; Schriemer, D. C. Rapid Commun. Mass Spectrom. 2003, 17, 1044-1050.
Analytical Chemistry, Vol. 79, No. 24, December 15, 2007
9537
and N. Bit-Avragim (Charite´-Universita¨tsmedizin Berlin, Germany) for recruiting the HCM patients with mutation Ile736Thr and S. Tripathi (Molecular and Cell Physiology, Medical School Hannover, Germany) for providing the mRNA quantification data. Drs. H. John and M. Walden (IPF PharmaCeuticals GmbH, Hannover, Germany) are acknowledged for support with preliminary quantification of mutation Ile736Thr and Dr. A. Pich for help with a MALDI-TOF-MS control experiment. The authors thank Dr. Ch. Scheler and Dr. S. Wienkoop (Proteome Factory AG, Berlin, Germany), Dr. A. Pich (Toxicology, Medical School Hannover,
9538
Analytical Chemistry, Vol. 79, No. 24, December 15, 2007
Germany), and Dr. J. L. Hodgkinson, (University of Oxford, U.K.) for critical reading of the manuscript. The work was supported by a Grant of the Deutsche Forschungsgemeinschaft to T.K. (KR 1187/5-3, 4).
Received for review August 13, 2007. Accepted October 4, 2007. AC701711H