Characterization of target proteins by mass spectrometry

Abstract

The present invention provides a method of characterizing a target protein, wherein a peptic peptide mass spectrum of the target protein is obtained and compared to a peptic peptide mass spectrum of a protein of a reference protein or reference host cell that has been determined. The comparing includes determining a similarity score for the intensity patterns of the mass spectrum and the reference mass spectrum. The characterization comprises matching the target protein to a reference protein whose reference mass spectrum has a similarity score above a predetermined threshold. The particular advantage of the present invention is that the following is taken into account: for the spectra of two highly similar reference proteins, even if the same digestive peptide is used for digestion, the intensity patterns of the two digestive peptides may differ due to the different relative intensities of the two digestive peptides.

Description

Technical Field

[0001] The present invention relates to the field of protein identification, and more particularly to a method for characterizing a target protein by mass spectrometry. Background Art

[0002] Routine identification of proteins by peptide mass fingerprinting (PMF) of protein digests has been around for 25 years. In 1993, five groups of scientists published the basic elements of this approach: (1) Henzel, WJ, Billeci, TM, Stults, JT, Wong, SC, Grimley, C., and Watanabe, C. (1993). Proc Natl Acad Sci USA 90, 5011-5; (2) James, P., Quadroni, M., Carafoli, E., and Gonnet, G. (1993). Biochem Biophys Res Commun 195, 58-64; (3) Mann, M., Hojrup, P., and Roepstorff, P. (1993). Biol Mass Spectrom 22, 338-45; (4) Pappin, DJC, Hojrup, P. and Bleasby, AJ (1993). Curr. Biol. 3, 327-32; and (5) Yates, JR, 3rd, Speicher, S., Griffin, PR and Hunkapiller, T.

[0003] (1993). Anal Biochem 214,397-408.

[0004] This method has achieved great success and is widely used. A detailed historical review entitled "Protein Identification: The Origins of Peptide Mass Fingerprinting" was given by WJ Henzel, C. Watanabe and JS Stults (J Am Soc Mass Spectrom. 2003, 14, 931-942).

[0005] In a 2004 ASMS tutorial, John Cottrell taught: "Peptide mass fingerprinting can only be used for pure proteins or very simple mixtures. The protein is digested with an enzyme with high specificity; usually trypsin, but any specific enzyme can be used. The peptide mixture product is analyzed by mass spectrometry. This produces a set of molecular mass values ​​that can be retrieved from a protein sequence database using a search engine. For each entry in the protein database, the search engine simulates the known enzyme cleavage specificity, calculates the expected peptide mass, and compares this set of calculated mass values ​​to the set of experimental mass values. Some sort of score is used to identify the entry in the database that provides the best match, and a report is generated."

[0006] Therefore, peptide mass fingerprinting utilizes only the mass of the peptide. Intensity usually does not play a role, since it can hardly be calculated by computer virtual cutting programs. John Cottrell: "In peptide mass fingerprinting, the mass value of the peak is the most important. The peak area or intensity value is a function of the peptide's basicity, length, and some other physical and chemical parameters. There is no particular reason to assume that large peaks are of interest and small peaks are of no interest... Mass accuracy is important, but coverage is equally important. It is better to have a large number of mass values ​​with moderate accuracy than one or two mass values ​​with extremely high accuracy."

[0007] There are more than half a dozen peptide mass fingerprinting programs available on the Internet, and many more are described in the literature.

[0008] Compared to using only mass values, KCParker proposed an attempt to take into account calculated peak intensities: "Scoring methods in MALDI Peptide Mass Fingerprinting: ChemScore, and the ChemApplex Program", J AmSoc Mass Spectrom 2002, 13, 22-39. The authors describe a software program (ChemApplex) based on calculated parameters (combined protein scores) that takes into account (1) peak intensity, (2) mass accuracy of the match, and (3) a theoretical intensity factor, ChemScore, which is based on a combination of chemical factors (particularly the amino acid composition of the peptide and the amino acid sequence of the amino acids spanning the cleavage site) to estimate the probability of observing a particular peptide.

[0009] Publication US2007 / 0092926 A1 (MA Alterman and BAKornilayev; 2007) describes the application of "unique proteolytic peptides" to peptide mass fingerprinting. The term "distinctive proteolytic peptide" or "unique proteolytic peptide" includes a compound consisting of amino acid subunits connected by peptide bonds formed by proteolytic cleavage of proteins by proteases, which is different from any other proteolytic peptides generated by digestion of other proteins using the same protease. When referring to SwissProt or NCBI databases, the distinctiveness or uniqueness preferably relates to the entire genome, and most preferably to the human genome. The boundaries of the peptide can be determined by predicting the cleavage site of the protease.

[0010] Qualitative and quantitative analysis of target proteins as described in publication US2005 / 0153380 A1 (NP Everett et al., 2005) is based on "characteristic peptides" for the same purpose. The target protein is extracted and concentrated from the mixture. Examples of suitable methods for concentration include the use of solid phase support resins (e.g., ion exchangers, affinity gels and other resins that adsorb proteins). The term "characteristic peptide mass" refers to the mass of peptides generated by one or more specific protein targets, which can be used to identify protein targets. Peptide masses that are easily ionized and have high mass resolution and accuracy and are derived from a specified peptide mass fingerprint are considered to be members of a set of characteristic diagnostic peptide masses for a specified target. For each protein, the pattern is unique and therefore special.

[0011] Considering the development (e.g., clone selection) and production (e.g., Fill

[0012] &Finish Rapid identification tests in operations) require fast and reliable analysis with short turnaround times to accelerate decision making and reduce costs. During the production of biopharmaceuticals, several product molecules with high similarity may appear, increasing complexity. For example, different therapeutic antibodies of the same class (e.g., IgG1κ) cannot be reliably distinguished based on sequence coverage obtained from commonly used peptide mass fingerprinting, and currently require LC-MS or dedicated ELISAs for specific identification. A rapid and reliable method that can be specifically applied to a collection of highly related molecules (monoclonal antibodies such as Herceptin or Adalimumab) will be described. Summary of the invention

[0013] In a first aspect, the present invention provides a method for characterizing a target protein, comprising the following steps: providing a library of determined reference mass spectra of reference proteins, wherein for an enzymatic digestion product of a reference protein, each reference mass spectrum is obtained, and for all reference proteins, the enzymatic digestion conditions are substantially the same; enzymatically digesting the target protein under the same conditions as for the reference protein; obtaining a mass spectrum of the target protein after enzymatic digestion; determining a similarity score of the intensity pattern of the mass spectrum and the reference mass spectrum; and characterizing the target protein by matching the target protein with a reference protein, the similarity score of the reference mass spectrum of the reference protein being above a predetermined threshold. The target protein is typically a protein from a sample to be characterized.

[0014] The method compares the mass sites and intensities of the measured mass spectra with the mass sites and intensities of the reference mass spectra of the reference proteins measured in the library for similarity. A particular advantage of the method is that it takes into account the fact that for spectra of two highly similar reference proteins, even if digested with the same digest peptide, the intensity patterns of the two may be different due to the different relative intensities of the digest peptides of the two. The similarity score is preferably determined by a cosine similarity metric or a cross-correlation.

[0015] Enzyme digestion preferably uses at least one of trypsin, cysteine ​​protease (IdeS) and Lys-C. Before enzymatic digestion, the target protein and the reference protein are subjected to the same denaturation by a reducing agent and by a denaturant. Denaturation and enzymatic digestion can be performed within 30 minutes, or even shorter than 15 minutes. Preferably, dithiothreitol (DTT) with trifluoroethanol (TFE) is used as a denaturing solution.

[0016] The mass spectrum of the digested target protein and the reference mass spectrum are preferably obtained by a MALDI-TOF mass spectrometer. More preferably, the MALDI-TOF mass spectrometer includes a reflector. The MALDI-TOF mass spectrometer can be a desktop instrument with a footprint of less than one square meter and a height of less than 1.5 meters.

[0017] In a preferred embodiment, the reference protein has a different amino acid sequence, and the enzymatic digestion product of the reference protein comprises a digestion peptide having a characteristic mass of the corresponding reference protein. The target protein is identified by matching with one of the reference proteins. For example, the target protein can be extracted from a complex sample mixture by affinity capture. The complex mixture can be one of urine, blood plasma, serum, cerebrospinal fluid, and lysed tissue cells. Preferably, an antibody is used for affinity capture.

[0018] In another embodiment, the target protein is a biopharmaceutical, particularly a biopharmaceutical from a product. In this case, the reference protein is a different protein isoform of the target biopharmaceutical, which is formed by at least one of alternative splicing, sequence variation, post-transcriptional modification, and stress-induced modification. The target protein is identified as one of the isoforms by matching with one of the reference proteins. The post-transcriptional modification can be at least one of glycosylation, oxidation, acetylation, and amidation.

[0019] In another embodiment, the target protein is an antibody, particularly an antibody from a product. In this case, the reference protein is a different modified form of the antibody. The target protein and the reference protein are digested into domains by enzymes, and the target protein is identified as modified by matching with one of the reference proteins. The modification can be at least one of glycosylation, oxidation, acetylation, and amidation. Preferably, IdeS is used to prepare Fc / 2 antibody domains for glycoprofiling. The reference protein can be a desired antibody, and the target protein is identified as a desired antibody by matching with the reference protein.

[0020] Preferably, the method further comprises removing the redundant mass signals present in most of the measured reference mass spectra, and providing and / or using the reduced reference mass spectrum in the determining step. The reduced set preferably includes about three to fifteen mass signals. If the reference protein has different amino acid sequences, the reduced reference mass spectrum preferably only includes the mass signals of the digested peptides as the characteristics of the corresponding reference protein. The reduced reference mass spectrum may additionally include some redundant mass signals, which are present in substantially all or a lot of reference mass spectra, and the number of these redundant mass signals is less than the number of characteristic mass signals. Multivariate statistical methods are used to determine the redundant mass signals of the measured reference mass spectra of the library. Multivariate statistical methods are preferably one of principal component analysis (PCA), hierarchical clustering (HC), receiver operating characteristic curve (ROC) and probability latent semantic analysis (pLSA).

[0021] In a second aspect, the present invention provides a method for characterizing a protein of a target host cell, comprising the following steps: providing a library of determined reference mass spectra of a reference host cell protein, wherein each reference mass spectrum comprises a mass signal of a digested peptide produced by enzymatic digestion of a protein of a reference host cell, and after the enzymatic digestion, extracting the digested peptide by affinity capture using a plurality of affinity agents; enzymatically digesting the protein of the target host cell under the same conditions used for the protein of the reference host cell, and then extracting the digested peptide by affinity capture using a plurality of affinity agents; acquiring a mass spectrum of the extracted digested peptides of the target host cell; determining a similarity score of the intensity pattern of the mass spectrum and the reference mass spectrum; and

[0022] The target host cell is characterized by matching the target host cell's proteins to a reference host cell whose reference profile has a similarity score above a predetermined threshold.

[0023] The method compares the mass sites and intensities of the measured mass spectra with the mass sites and intensities of the reference mass spectra determined in the library for similarity. A particular advantage of this method is that it takes into account the fact that the spectra of two highly similar reference mass spectra may include mass signals at the same mass sites, but with different intensity patterns, mainly due to different relative intensities of the digested peptides. The similarity score is preferably determined by a cosine similarity metric or a cross-correlation. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic workflow of the method for characterizing a target protein according to the present invention is shown.

[0025] Figure 2 A preferred workflow for rapid denaturation and digestion of target protein is shown. DETAILED DESCRIPTION

[0026] The present invention provides a preferred method for identifying a target protein by peptide mass fingerprinting (PMF) of a peptide mixture, in particular for quality assurance to confirm characteristics, wherein the peptide mixture is generated by enzymatic digestion of the target protein by a protease such as trypsin. Completely different from the commonly used peptide mass fingerprinting, the identification will be based on the similarity of the intensity pattern of the measured target protein and the intensity pattern of the determined reference protein.

[0027] Whereas known peptide mass fingerprints rely on peptide masses calculated from amino acid sequences found in the library, the new method compares the similarity of the mass and intensity of the measured mass spectra with the mass and intensity of the determined reference mass spectra of the reference proteins in the library. The specific advantage of this approach is that it takes into account the following situation: even if the same digested peptide of two highly similar reference proteins may appear in two corresponding reference mass spectra, due to different physicochemical microenvironments near the cleavage site, it will produce mass signals that are very different in their relative intensities.

[0028] Any suitable program can be used to determine the similarity of the intensity patterns between the mass spectrum of the target protein and the reference mass spectra of the reference proteins constituting the library. As an example, the similarity of the intensity patterns can be quickly determined by the cosine of the angle between the vectors constituting the intensity patterns (cosine similarity score) or by cross-correlation.

[0029] The reference spectra of the reference protein are preferably reduced to a smaller set (subset spectra) of characteristic mass signals with intensity, which makes the match have better specificity. The reduced set preferably includes about three to fifteen peptides. There are different methods that can be used to generate such subset spectra. These methods include typical statistical methods for spectra classification, such as those named with principal component analysis (PCA), hierarchical clustering (HC), and receiver operating characteristic curves (ROC). If the reference protein is an antibody, characteristic digestion peptides can be determined from complementary determining regions (CDRs). If the library is amplified by adding other reference mass spectra, the selection of specific mass signals must be repeated. Some redundant peptide mass signals can be added for quality control.

[0030] The method is preferably carried out using a rapid protein denaturation and digestion method, usually within 30 minutes, especially within 15 minutes. The resulting peptide mixture is prepared with a matrix solution to generate sample spots on the mass spectrometer sample stage for use in a matrix-assisted laser desorption ionization (MALDI)-time of flight mass spectrometer. A reasonably priced benchtop time of flight mass spectrometer can be used for spectrum acquisition.

[0031] Based on the difference in relative intensity patterns in reference mass spectra of reference protein digested peptides, the method of the present invention can be used in particular for quality control of antibody production, or clone selection workflow during drug development, for example, for screening glycan profiles in complete Fc-domains or for clones to appropriately carry target sequences. For antibody characterization, the technical solution, particularly the technical solution based on trypsin / Lys-C digestion, can be used to obtain results from providing antibodies to automatic identification within 15 minutes.

[0032] The target and reference proteins can be rapidly denatured using a mixture of reducing agents and denaturants (e.g., dithiothreitol and trifluoroethanol) and rapidly digested by enzymes such as trypsin or sequential or parallel double digestions (e.g., trypsin / Lys-C). Matrix-assisted laser desorption ionization (MALDI) can be used to prepare peptide mixtures with a matrix material. HCCA (α-cyano-4-hydroxyphenylacrylic acid) can be used as a matrix material.

[0033] However, the method can also be applied to target proteins present in complex mixtures of substances. In this case, the target protein must be extracted and purified. For example, in order to avoid time-consuming methods such as HPLC, rapid purification methods must be applied. Examples are solid phase extraction, column affinity capture or magnetic bead affinity capture, etc.

[0034] Example of a rapid protein identification test embodiment: The digestion time including denaturation, reduction and proteolytic digestion is reduced to 15 minutes, and the subsequent MALDI sample preparation includes a simple and accurate (on-target) purification step. The MALDI peptide mass fingerprints obtained from all tested antibody digests are of high quality (70% sequence coverage) and can be used for identification tests based essentially only on peptide differences, namely peptides derived from the variable N-terminus of the antibody of ~120 residues; in addition to the six common peptides, 4 to 13 peptides are used for these spectra. The spectra of these antibodies enable them to be distinguished based on the cosine similarity score (CSS), with CSS>0.9 as the acceptance criterion, and specific mismatches will result in CCS values ​​of 0.2 to 0.6. In addition, the IDs provided by the software can be visually confirmed by butterfly plots.

[0035] Examples of specific steps for clone selection: Digestion and sample preparation time for IdeS digestion and MALDI sample preparation is about 30 minutes. Major glycans such as G0F, G1F, G2F and G3F are analyzed by direct analysis of the Fc-domain of the monoclonal antibody together with the Fc C-terminus in the appropriate state. Spectrum acquisition and processing are completed within 10 seconds / sample. Different attributes are recorded in the software, such as the matching of the glycan spectrum with a reference spectrum with a determined score, or the test with G0F as the base peak glycan. The automation and parallelization of sample processing in the clone selection workflow enables the evaluation of hundreds of samples per day in a highly automated and significantly accelerated clone selection based on major Fc-glycans.

[0036] The present invention further provides a preferred method for characterizing proteins of target host cells. Here, after trypsin digestion, anti-polypeptide antibodies are used to extract digested peptides of proteins of target host cells. Typically, stable isotope labeled standard peptides (SIS peptides) can be additionally added after digestion, so that SIS peptides and native peptides obtained by trypsin digestion can be extracted together with the anti-polypeptide antibodies. Currently, multiple antibodies (polyclonal antibodies or monoclonal antibody libraries) and SIS peptides can be used together in the analysis to extract 5, 10 or more peptides that are specific for host cell protein (HCP) analysis. Currently, spectra can be defined with SIS peptide mass signals and intensities at levels associated with each host cell protein. Then, the intensity of the native peptide can be determined and accurately quantified based on the peak intensity ratio of the native peptide / SIS peptide and the deviation from the automatically detected target HPC level.

Claims

1. A method for characterizing a target protein, comprising the following steps: Providing a library of determined reference mass spectra of reference proteins, wherein each reference mass spectrum is obtained for an enzymatic digestion product of one reference protein, and the conditions of the enzymatic digestion are substantially the same for all reference proteins; enzymatically digesting the target protein under the same conditions as used for the reference protein; Obtaining mass spectra of target proteins after enzymatic digestion; determining a similarity score between the intensity patterns of the mass spectrum and the reference mass spectrum; as well as characterizing the target protein by matching the target protein to a reference protein whose reference mass spectrum has a similarity score above a predetermined threshold, wherein redundant mass signals present in most of the determined reference mass spectra are removed, and a reduced reference mass spectrum is provided and / or used in the determining step, and wherein the reduced reference mass spectrum further comprises some redundant mass signals, which are substantially present in all reference mass spectra, and the number of the redundant mass signals is less than the number of the characteristic mass signals.

2. The method according to claim 1, wherein the enzymatic digestion uses at least one of trypsin, IdeS and Lys-C.

3. The method according to claim 2, wherein prior to the enzymatic digestion, the target protein and the reference protein are identically denatured by a reducing agent and by a denaturing agent.

4. The method of claim 1, wherein the reference proteins have different amino acid sequences, the enzymatic digestion products of the reference proteins comprise digested peptides each having a characteristic mass of the corresponding reference protein, and the target protein is identified by matching with one of the reference proteins. The method of claim 1 , wherein the target protein is extracted from a complex sample mixture by affinity capture. The method of claim 1 , wherein the target protein is a biopharmaceutical.

7. The method of claim 6, wherein the reference protein is a different protein isoform of the biopharmaceutical, the isoforms being formed by at least one of alternative splicing, sequence variation, post-transcriptional modification, and stress-induced modification, and the target protein is identified as one of the isoforms by matching with one of the reference proteins. The method of claim 1 , wherein the target protein is an antibody.

9. The method of claim 8, wherein the reference protein is a different modified form of the antibody, the target protein and the reference protein are enzymatically digested into domains, and the target protein is identified as modified by matching with one of the reference proteins.

10. The method of claim 9, wherein a reference protein is the antibody, and the target protein is identified as the antibody by matching with the reference protein.

11. The method of claim 1, wherein the reference protein has a different amino acid sequence, the enzymatic digestion product of the reference protein comprises digested peptides having characteristic masses of the corresponding reference protein, and the reduced reference mass spectrum comprises only characteristic mass signals.

12. The method of claim 1, wherein the similarity score is determined by a cosine similarity metric or a cross-correlation.

13. A method for characterizing a protein of a target host cell, comprising the following steps: providing a library of determined reference mass spectra of proteins of a reference host cell, wherein each reference mass spectrum comprises mass signals of digested peptides generated by enzymatic digestion of a plurality of proteins of a type of reference host cell, and after the enzymatic digestion, the digested peptides are extracted by affinity capture using a plurality of affinity agents; enzymatically digesting the target host cell protein under the same conditions used for the reference host cell protein, and then extracting the digested peptides by affinity capture using the plurality of affinity agents; acquiring a mass spectrum of the extracted digested peptides of the target host cell; determining a similarity score between the intensity patterns of the mass spectrum and the reference mass spectrum; as well as characterizing the target host cell by matching a protein of the target host cell to a reference host cell of a certain type, the reference host cell having a reference profile with a similarity score above a predetermined threshold, wherein redundant mass signals present in most of the determined reference mass spectra are removed, and a reduced reference mass spectrum is provided and / or used in the determining step, and wherein the reduced reference mass spectrum further comprises some redundant mass signals, which are substantially present in all reference mass spectra, and the number of the redundant mass signals is less than the number of the characteristic mass signals.

Images