- Email: [email protected]
Detecting monoclonal immunoglobulins in human serum using mass spectrometry
Methods 81 (2015) 56–65
Contents lists available at ScienceDirect
Methods journal homepage: www.elsevier.com/locate/ymeth
Detecting monoclonal immunoglobulins in human serum using mass spectrometry John R. Mills ⇑, David R. Barnidge, David L. Murray ⇑ Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN 55905, United States
a r t i c l e
i n f o
Article history: Received 3 February 2015 Received in revised form 15 April 2015 Accepted 18 April 2015 Available online 24 April 2015 Keywords: Monoclonal gammopathy Immunoglobulin Multiple myeloma Mass spectrometry Clinical diagnostics
a b s t r a c t Established guidelines from the International Myeloma Working Group recommend diagnostic screening for patients suspected of plasma cell proliferative disease using protein electrophoresis (PEL), free light chain measurements and immunofixation electrophoresis (IFE) of serum and urine in certain cases. Plasma cell proliferative disorders are generally classified as monoclonal gammopathies given most are associated with the excess secretion of a monoclonal immunoglobulin or M-protein. In clinical practice, the M-protein is detected in a patients’ serum by the appearance of a distinct protein band migrating within regions typically occupied by immunoglobulins. Given each M-protein is comprised by a sequence of amino acids pre-defined by somatic recombination unique to each clonal plasma cell, the molecular mass of the M-protein can act as a surrogate marker. We established a mass spectrometry based method to assign molecular mass to the immunoglobulin light chain of the M-protein and used this to detect the presence of M-proteins. Our method first enriches serum for immunoglobulins, followed by reduction to separate light chains from heavy chains, followed by microflow LC-ESI-Q-TOF MS. The multiply charged light chain ions are converted to their molecular mass and reconstructed peak area calculations are used for quantification. Using this method, we term ‘‘monoclonal immunoglobulin Rapid Accurate Molecular Mass’’ or miRAMM, the presence of M-proteins can be reliably detected with superior sensitivity compared to current gel-based PEL and IFE techniques. Ó 2015 Published by Elsevier Inc.
1. Introduction B-cells play a central role in the adaptive immune system expressing a diverse repertoire of immunoglobulins (Igs). Secreted Igs protect against foreign antigens through several mechanisms including; binding to foreign antigens (i.e. bacteria, viruses), recruiting other immune cells, or directly neutralizing foreign substances [1]. Igs are homodimers composed of two heavy chains (HCs) and two identical light chains (LCs). Each LC pairs with a HC and each HC pairs with another HC through disulfide bonds (see Fig. 1). Both the HC and the LC have two distinct regions called the constant and variable region. The C-terminal half of each HC and LC encompasses the constant region which contains an epitope that defines the chain’s isotype. The HC can have 5 different isotypes; IgG, IgA, IgM, IgD or IgE, while the LC can have 2; kappa or lambda. The N-terminal half of each HC and LC contains the ⇑ Corresponding authors at: Department of Laboratory Medicine and Pathology, Mayo Clinic, 200 1st St SW, Rochester, MN 55905, United States. Fax: +1 (507) 284 1927. E-mail addresses: [email protected] (J.R. Mills), [email protected] (D.L. Murray). http://dx.doi.org/10.1016/j.ymeth.2015.04.020 1046-2023/Ó 2015 Published by Elsevier Inc.
variable region which includes the amino acid sequences that bind antigens. A single B-cell will create HC and LC variable regions by rearranging Ig specific gene segments [1]. If both heavy and light chain rearrangements generate functional sequences the expressed Ig is assembled and expressed on the cell surface of the B-cell. If this process is successful then the cell undergoes allelic exclusion to ensure each cell only expresses a single Ig [2]. Therefore, each clonal cell is uniquely defined by a particular gene rearrangement that encodes a specific Ig with a defined molecular mass. When a B-cell engages an antigen the cell is stimulated to proliferate which induces clonal expansion. B-cell clones that acquire mutations that improve antigen binding affinity will be selectively expanded in the germinal center. Those cells which differentiate into Memory B-cells may undergo class switching at the IgH locus which generates Igs encoding different constant regions prior to commitment to becoming long-lived plasma cells (PCs). These long-lived PCs reside in the bone marrow and secrete vast quantities of high-affinity antigen specific Igs. Each PC will express a defined Ig with a molecular mass which is identical to its original precursor cell. Humans produce a vast diversity of PCs – each with their own randomly generated Ig (upwards of 1012 possible unique
J.R. Mills et al. / Methods 81 (2015) 56–65
57
Fig. 1. Ig structure indicating domains and sites of covalent modification. Structure of a typical IgG immunoglobulin. Two identical HCs and LCs are connected by disulfide linkages. Furthermore, within each chain there are also disulfide bonds which are critical to maintain structure. The variable regions of both the HC and LC form the bivalent antigen binding sites. The class of the HC and LC are defined by the constant regions which have distinct amino acid residues depending on the class. Carbohydrate posttranslation modifications on the HC generate different Ig glycoforms which can be functionally important.
sequences) creating what is often referred to as the normal polyclonal Ig repertoire [3]. However, PCs can become malignant resulting in an abnormal expansion of clonal cells as is the case for the plasma cell proliferative disorders, multiple myeloma (MM), AL amyloidosis, and Waldenström’s macroglobulinemia. When this occurs a single monoclonal Ig can be detected above the normal polyclonal Ig repertoire. If there is clinical suspicion of a plasma cell proliferative disorder, the patient’s serum and urine is typically tested for the presence of a monoclonal Ig or M-protein. M-proteins are typically detected using a combination of protein gel electrophoresis (PEL) and immunofixation (IFE) [4]. These techniques use the differential charge and size of Igs to separate them from other serum proteins on a gel media, usually agarose or cellulose acetate. Typically a low ionic strength buffer with an alkaline pH is used, which is sufficient to provide most proteins with a negative charge. After an electric field is applied the serum proteins migrate into distinct fractions knows as the albumin, a1 (composed predominantly of a1-antitrypsin), a2 (mostly a2-macroglobulin and haptoglobin), b (transferrin and C3) and c (mostly Igs) regions. After separation, proteins in the gel are stained and the intensity of the staining is used to determine the relative contribution of a given fraction to the total serum protein content. In most healthy individuals the c region contains the polyclonal Igs found in serum and is a broad and diffuse fraction driven by the vast diversity of Ig amino acid sequences. When an excessive amount of a specific monoclonal Ig is present a characteristic distinct monoclonal band or spike (M-spike) is typically observed in the c fraction [5]. In instances where the presence of an M-protein is questionable, immunofixation electrophoresis (IFE) can be helpful. IFE is similar in principle to PEL with an additional in-gel immunoprecipitation step. Several identical aliquots of the patient’s serum are separated by agarose gel electrophoresis in different wells. After separation one well serves as a reference well and is immediately fixed and whereas the other wells are incubated with antisera against the HC isotypes (IgA, IgG, IgM) and LC isotypes (j or k). The interaction of the antisera with the Ig in the gel induces a precipitate. Distinct bands are noted for each well and the relationship between HC and LC is established based upon similar monoclonal migration distances (see Fig. 5). IFE has a limit of detection that is 10-fold lower than PEL; however, it is important to note IFE is not quantitative and is
only useful for qualitative assessment. Capillary electrophoresis (CE) can be used as an alternative to gel electrophoresis [6]. CE uses small-bore (10–100 lM) fused silica capillary tubes and microfluidics to separate proteins which are then visualized using a UV detector. CE offers superior speed and automation capabilities. The equivalent to gel based IFE is immunosubtraction CE. In the case of the latter, serum is incubated with antisera for each of the HC and LC, then removing the complexes prior to running CE. The results are compared back to the serum sample without antisera to make a qualitative conclusion about the type of Ig present. Despite these technical improvements these methods have inherent limitations due to their laborious setup, poor resolution, limited sensitivities, and occasionally ambiguous results. The most recent generation of mass spectrometers provide higher resolution, greater mass measurement accuracy, and greater sensitivity than previous generations. These new instruments (i.e. Orbitrap and Time-of-Flight (TOF) mass spectrometers) are also smaller and more robust and have begun to play a role in clinical diagnostics measuring the accurate molecular mass of intact proteins (i.e. IGF1, hemoglobin) [7–10]. Increases in the linear dynamic range, as well as increases in resolution and mass measurement accuracy, have made these instruments a sensible alternative to characterizing proteins as compared to traditional protein gel electrophoresis based methods. This was recently demonstrated for intact Igs where the molecular masses of therapeutic monoclonal Igs were determined with high accuracy [11– 13]. Accordingly, manufacturers of therapeutic monoclonal Igs are using the newest generation of mass spectrometers to evaluate product quality via accurate molecular mass determinations of recombinant Igs. Our group recently demonstrated how microflow liquid chromatography coupled with electrospray ionization (ESI) and Q-TOF MS (microLC-ESI-Q-TOF MS) could be used to identify and monitor a monoclonal immunoglobulin in a patient’s serum and urine [14–16]. Instead of tracking the intact monoclonal Ig, the disulfide bonds holding the HC and LC together were reduced using DTT and the LC was the primary analyte monitored by MS. Once reduced, LCs were further separated from HCs using liquid chromatography and these LCs were then injected into the mass spectrometer where their masses are obtained. There are several analytical advantages to measuring LCs rather than the HC or the
58
J.R. Mills et al. / Methods 81 (2015) 56–65
intact Ig. First, LC masses range from 22,000 to 25,000 Da compared to intact Igs which are typically in a mass range of 150,000 Da. Smaller proteins ionize much more readily, a critical component of MS, making detection of light chains more sensitive compared to intact Igs [17]. Second, Igs are commonly post-translationally modified predominantly through glycosylation which is typically restricted to the HC portion of the Ig [18]. Using microLC-ESI-Q-TOF the molecular mass of a patient’s myeloma-associated monoclonal LC was used to track their disease over time [14,15]. This work established that monoclonal LCs corresponding to M-proteins in either serum or urine could be identified, quantified, and isotyped using this methodology. Relative to PEL, IFE and FLC, this approach promises more analytical sensitivity and specificity because of high accuracy mass measurements. Here we address the methods used to determine the accurate molecular mass of a monoclonal immunoglobulin in human serum using mass spectrometry. These methods have the potential to alter how clinical laboratories screen and monitor for the presence of monoclonal gammopathies. 2. Technical considerations Below we will discuss steps that are critical in developing a MS based method to monitor Igs in human serum. The analysis of large proteins by MS presents distinct challenges compared to small non-proteinaceous molecules most often analyzed by MS in the clinical laboratory. 2.1. Immunoglobulin purification Current methods for monitoring Igs by MS require at least one purification step depending on the depth of analysis. Human serum is highly proteinaceous containing 6.3–7.9 g/dL of protein, 2/3 of which is albumin. In order to improve downstream analytical performance (including maintaining column lifetime and sensitivity) it is necessary to remove albumin and other contaminating non-immunoglobulin proteins from serum. There has been a long standing need for methods to purify antibodies from different sources as a prerequisite for use in biomedical research, diagnostic methodologies and more recently as therapeutics. There are several strategies to purify antibodies and each has been advantages and disadvantages depending on the downstream application. When evaluating Ig therapeutics, maintaining native confirmation is critical whereas for structural determinations and quantitation it may be acceptable to purify antibodies under harsher conditions (acid, high salt etc.). Methodologies for purification can be separated into physiochemical and affinity approaches. Physiochemical separation of antibodies includes the use of size exclusion chromatography (SEC), ion exchange chromatography (IEC), ammonium sulfate precipitation, caprylic acid precipitation, thiophilic adsorption and newer chemistries such as Melon-Gel™ resin. Affinity purification can be split into two groups – proteins that bind specific classes of Igs (i.e. Protein A, G, L or anti-Ig antibodies) or antigens that bind to a specific Ig (i.e. immobilized p53 peptides to enrichment for anti-p53 Igs) [19]. Plasma cell proliferative disorders can be associated with any of the HC and LC Ig classes. Therefore the ideal purification scheme can purify all classes of Igs. Furthermore, it is important to consider reproducibility, scale, robustness, cost and the time required for a given purification scheme. The pharmaceutical industry has driven demand for robust and cost-effective methods to purify the IgG class since all therapeutic monoclonal antibodies are exclusively IgG [20]. Many preparative Ig purification schemes are designed to purify large quantities of monoclonal Igs with lengthy clean up steps. However, in the clinical laboratory, most assays utilize
sample volumes in the range of lLs and pre-analytical steps that take minutes and hours rather than days. Listed below are several methodologies that have the potential to be incorporated into the clinical laboratory to purify Igs prior to analysis by mass spectrometry. Ammonium sulfate precipitation is a common technique to purify and concentrate Igs from serum. This technique works by ‘‘salting-out’’ Igs selectively. Igs precipitate out of solution at lower salt concentrations compared with many other serum proteins (including albumin) due to poor solubility. At 40–50% of ammonium sulfate saturation most Igs will precipitate whereas albumin remains in solution. The precipitate containing the Igs can be re-dissolved in buffer and then further processed. This method is sufficient as a first pass step in the purification of Igs. It is robust and has proven clinical utility when quantitating tryptic peptides [21] using mass spectrometry. However, the high concentration of sulfate ions can result in sulfate adducts complicating the mass spectrum and reducing sensitivity [22]. Thiophilic adsorption chromatography (TAC) was originally described in 1985 [23]. All Igs are thiophilic proteins (affinity for sulfur containing molecules) while most other proteins including albumin are not [24]. Immobilized thioether-sulfones can be utilized to selectively bind Igs in the presence of certain salts (i.e. potassium sulfate). Subsequently the use of thiophilic agarose beads has become well established as a robust and cost-effective method to purify Igs and immunoglobulin fragments from a variety of sources including human serum [25]. A major benefit being Igs can be eluted from beads by dilution into physiological buffers containing low salt concentrations at neutral pH. Thiophilic adsorption has a unique property in that it has high capacity to adsorb the three major classes of Igs (IgG, IgA, and IgM) including their subclasses. The use of high concentrations of sulfate ions necessitates desalting prior to analysis by MS. Traditionally this methodology had been utilized for selective depletion of Igs from serum. Protein A, G and L are bacterial proteins with defined antibody binding properties. Recombinant versions of these proteins are commercially available and offer wide applications in purification, immobilization and detection of Igs. Protein A and G bind the fragment crystallizable (Fc) region of most Igs and also within the fragment antibody-binding (Fab) region (Fig. 1). However, the binding affinities vary significantly between Ig isotypes and between subclasses (i.e. IgG3). Protein A and G bind IgG very strongly but IgM binding is poor due to inaccessibility of binding sites on IgM and its pentameric structure [26]. Protein L has selective binding affinity for kappa LCs and therefore for can be useful for purifying all HC Ig isotypes containing kappa LC but fails to bind most lambda LCs [27]. Melon Gel™ is advantageous as it is simple, fast, and inexpensive (Pierce Biotechnology). It is based upon selective charge attraction of albumin and other unwanted proteins to the resin and charge repulsion of Igs. However, it is important to recognize that Melon Gel™ was developed for the purification of IgG and the recovery of IgA and IgM is lower. The CaptureSelectÒ Antibody Toolbox affinity matrix (developed by Bio Affinity Company) offers a variety of products with specific binding affinity for all HC and LC isotypes and subclasses. These products are recombinant single-chain camelid Igs that lack the heavy and light chains found in other animal Igs including human and they are produced in Saccharomyces cerevisiae thus are free of any trace animal Igs. These affinity beads can be utilized to purify all human Igs from serum regardless of type. However, the cost of these reagents is greater than the non-affinity based resins and incubation times are longer. In developing a methodology for clinical use, robustness, cost and preparation time are paramount. All of the methods above
J.R. Mills et al. / Methods 81 (2015) 56–65
59
Fig. 2. Overview of ‘‘monoclonal immunoglobulin Rapid Accurate Molecular Mass’’ or miRAMM. Igs are first purified from human serum and subsequently reduced with DTT which is sufficient to break inter-chain disulfide bonds. The dissociated mixture of HCs and LCs is then separated by RPLC using C4 column chromatography prior to analysis by TOF-MS. The resulting spectra of multiply charged ions is deconvoluted into the accurate molecular mass spectra. Peaks above the polyclonal LCs are indicative of monoclonal Igs. k and j depict the two different light chain populations which correspond to the lambda and kappa LCs, respectively.
have potential to meet these criteria, however, optimization is necessary and there have not been comprehensive studies today to confirm their suitability to purify a diverse range of Igs which would be encountered when screening and monitoring monoclonal gammopathies. The pre-analytical preparation of samples is critical to downstream mass spectrometry analysis; even the best mass spectrometry will fail to produce reproducible mass measurements in the absence of high quality samples. Initial analysis utilizing miRAMM focused on using Melon Gel given its cost benefit and simplicity of use. More extensive analysis will be necessary to determine if this method will be amendable across thousands of samples with diverse matrices and Ig subtypes. 2.2. Reverse-phase liquid chromatography There are several types of liquid chromatography which can be used to separate Igs. These include; reverse-phase liquid chromatography (RPLC), affinity chromatography, SEC, IEC, or hydrophilic interaction chromatography. Overall, RPLC is the most widely used in-line chromatographic technique since the mobile phase buffers are amenable to electrospray ionization. However, RPLC has several disadvantages that need to be overcome in order to make it a viable approach for the analysis of HC and LC Igs. RPLC columns have a limited protein capacity therefore care needs to be taken to eliminate overloading by using the appropriate amount of patient serum. An overloaded RPLC column will suffer from carryover, poor resolution, and irreproducible retention times. RPLC systems with large dead volumes also run the risk of having unnecessary protein absorption resulting in additional carryover. The use of inert tubing with small diameters such as PEEK-Sil can reduce protein adsorption in the liquid chromatography system. Carryover can also be problematic in hydrophobic stationary phase columns. The use of columns with shorter alkyl chains (C3, C4, and C8) can reduce peak tailing, peak broadening and increase recovery. Incorporating small amounts of formic acid (0.1%) can improve peak shape by improving solvation without suppression of ionization [28]. In order to successfully separate larger intact proteins it is critical to use porous packing material with larger pore sizes as the rate of analyte diffusion is dependent on pore size
to analyte size ratio. The use of 300 Å pore sizes has been demonstrated to provide superior separation for Igs [29]. Further improvements in separation performance for large proteins have been made by the use of columns with fused-core technology. The core of the column is composed of fused-silica which has poor penetration by analytes. Therefore this reduces the potential for diffusion providing for superior peak shape for intact proteins. Heating a column to at least 50 °C using an oven heater is essential to avoid carryover and improve peak shape [30]. Also, the addition of 10% or more 2-propanol to the acetonitrile in the organic phase will aid in reducing carryover. The use of liquid chromatography is an essential component in resolving monoclonal Igs present within the polyclonal Ig repertoire found in human serum. By separating Igs based on hydrophobic interactions specific monoclonal Igs can be separated away from a portion of the polyclonal Ig repertoire which increases the limit of detection – in this case defined as the ability to detect a given monoclonal Ig above the polyclonal background. 2.3. Mass spectrometry The development of ESI made it possible to convert ions in solution to the gas phase, allowing mass spectrometers with limited m/z range to analyze large proteins [31,32]. In the case of Igs, ESI produces ions with charge states upwards of +45 for HC +25 for LC bring the ions into the range of 600–2500 m/z. ESI-TOF and ESI-Q-TOF mass spectrometers are ideally suited to determine the accurate molecular mass of intact Ig [33] as well as HC and LC in this m/z range. As an example, ESI-Q-TOF pyroglutamate modifications were identified on recombinant intact monoclonal Igs using analysis after partially reducing Igs to release the light chain. Deconvoluted spectra demonstrated an 18 Da mass shift consistent with a N-terminal glutamate to pyroglutamate conversion which was confirmed by bottom-up proteomics (discussed later on) [34]. Orbitrap mass analyzers have also been used to provide accurate molecular mass of monoclonal Igs providing higher resolution and better mass accuracy than TOF analyzers [11]. Comparisons of the two instrument platforms have been published using the recombinant therapeutic rituximab as the model
60
J.R. Mills et al. / Methods 81 (2015) 56–65
monoclonal Ig [35,36]. Although these limited studies were performed on purified monoclonal Ig in the absence of polyclonal background they did demonstrate the capability of high resolution MS to identify the accurate molecular mass of monoclonal Ig with resolutions below <5 ppm. Furthermore, as little as 1 ng of material was readily measureable. Work by Barnidge and co-workers demonstrated similar resolution and sensitivities for the recombinant monoclonal Igs adalimumab spiked into normal human serum matrix [37,38]. 3. Method description Here we outline a cost effective, robust mass spectrometry-based method previously developed [14,38] and subsequently refined for identifying monoclonal Igs in human serum that provides superior sensitivity compared to current clinical gel-based assays PEL and IFE. This methodology ‘‘monoclonal immunoglobulin Rapid Accurate Molecular Mass’’ or miRAMM utilizes microflow liquid chromatography-ESI-TOF MS to measure the accurate molecular mass of the LC portion of intact monoclonal Igs which serves as a surrogate marker of the intact protein (Fig. 2) [16]. Igs are purified off-line, reduced to release the LCs from HCs and then separated using microflow liquid chromatography prior to ESI-TOF MS. The mass spectra of multiply charged light chain ions generated during ESI are converted to their molecular mass to produce spectra of the distribution of the entire LC repertoire in human serum enabling accurate molecular mass measurements for monoclonal Igs and quantitation using reconstructed peak area. In healthy adults mass spectra show two distinct normally distributed peaks (corresponding to lambda and kappa polyclonal LCs [32] – discussed below) with a smooth distribution of clones (Fig. 3). The presence of a peak above the polyclonal background is consistent with the presence of a monoclonal Igs – similarly to current gel based methodologies with the added benefit of improved resolution and accurate mass measurements within 1 Da. 3.1. Reagents Waste samples were collected from the Clinical Immunology Laboratory protein electrophoresis assay. Ammonium bicarbonate, dithiothreitol (DTT) and formic acid were purchased from Sigma– Aldrich (St. Louis, MO). Water and acetonitrile were purchased from Honeywell Burdick and Jackson (Muskegon, MI). Capture Select™ Affinity Matrices were purchased from Life technologies. Melon Gel was purchased from Thermo-fisher Scientific, Waltham, MA. 3.2. Sample preparation 20 lL of patient serum was added to 200 lL of Melon Gel (Thermo-fisher Scientific, Waltham, MA) and mixed for 5 min on
an orbital shaker. The resin was then centrifuged for 1 min at 5000 RPM. 20 lL of supernatant was then mixed with 20 lL of 50 mM ammonium bicarbonate (pH 8.0) and 10 lL of 200 mM DTT. This was then incubated at 55C for 30 min prior to injection. Alternatively, 4 lL of patient serum was diluted with 200 lL of PBS. Then 30 lL of a 1:1 mix of Capture Select™ LC-Kappa Affinity Matrix and Capture Select™ LC-Lambda Affinity Matrix as a 50% slurry in PBS was added. The samples were mixed for 45 min on an orbital shaker at room temperature. Beads were then washed 2 with PBS and 1 with HPLC grade water and then resuspended in a solution of 5% acetic acid with 50 mM TCEP (PierceÒ Biotechnology, Rockford, IL). The samples were then incubated at RT for 15 min, centrifuged at 5000 rpm for 1 min and injected into the mass spectrometer. For dilutions studies, serum containing M-proteins was diluted with pooled normal human serum (Fisher BioReagents). 3.3. Liquid chromatography An Eksigent Ekspert 200 microLC system (Redwood City, CA) coupled with a Poroshell 300SB-C3 analytical column from Agilent (1.0 75 mm Poroshell 300SB-C3; 5 lM particle size) was used for separation. A 2 lL sample injection was used with a 25 lL/min flow rate. The total run time is 24 min. The column temperature was maintained at 60 °C using a column heater. The aqueous solvent (Mobile phase A) consisted of water with 0.1% formic acid, and the organic phase (Mobile phase B) was 90% acetonitrile, 10% 2-propanol and 0.1% formic acid. A gradient was used to elute light chains starting at 80% A/20% B, held for 0.5 min, ramped down to 70% A/30% B over 1 min, then ramped down to 60% A/40% B over 4 min, further ramped down to 5% A/95% B over 5 min, held for 2.5 min and then ramped up to 80% A/20% B for 1 min and the re-equilibrated at 80% A/20% B for 1 min. 3.4. Mass spectrometry An ABSciex TripleTOFÒ 5600 quadrupole time-of-flight mass spectrometer (ABSciex, Vaughan, ON, Canada) operated in ESI positive mode with a Turbo V dual-ion source with an automated calibrant delivery system (CDS) was used to acquire mass spectra. Source conditions were ISVF, 5500; temp, 50; CUR, 45; GS1, 35; GS2, 30. TOF MS scans were acquired from m/z 600–2500 with an acquisition time of 200 ms. The instrument was calibrated every five injections through the CDS using the calibration solution supplied by the manufacturer. 3.5. Data analysis Analyst TF v1.6 was used for instrument control. Data were viewed using Analyst TF v1.6 and PeakView v1.2.0.3. BioAnalyst software provided with Analyst TF was used to reconstruct the multiply charged ion peaks into a mass graph. Deconvolution
Fig. 3. Normal polyclonal LC distribution. Multiply charged LC ions and the corresponding deconvoluted mass spectra demonstrating a representative polyclonal LC mass distribution found for healthy adults. The spectrum shows multiply charged ion pairs labeled with their corresponding charge states. Each pair of unresolved multiply charged ions has a normal distribution and is spaced according to their m/z ratio.
J.R. Mills et al. / Methods 81 (2015) 56–65
61
Fig. 4. miRAMM demonstrates a linear dynamic range. (A) The multiply charged LC ions of rituximab and the corresponding deconvoluted mass spectra demonstrating a representative of rituximab’s LC mass. (B) Rituximab was spiked into pooled human serum at 6 different concentrations spanning 0.0005 g/dL to 0.025 g/dL. The peak area corresponding to rituximab was measured using the deconvoluted mass spectra. Using linear regression analysis, the slope and intercept were calculated and use to calculate concentrations. (C) Representative deconvoluted mass spectra used to generate the plot in (B) are shown.
was performed between 22,000 Da and 26,000 Da as a protonated ion with a step size of 1 Da, a S/N threshold of 10 and 10 iterations for molecular mass calculations. 3.6. Protein electrophoresis and immunofixation Assays were performed according to protocols in the Clinical Immunology Laboratory [4], Department of Laboratory Medicine and Pathology, Mayo Clinic: IFE was performed using Sebia 9IF gels (Sebia, Norcross GA). 4. Results and discussion Initial work developing miRAMM documented the ability of this technique to quantitate a monoclonal therapy, adalimumab, in the presence of a normal concentrations of polyclonal Igs [38]. Serial dilutions into normal human serum demonstrated as little as 0.005 g/dL of adalimumab was detectable above the polyclonal distribution of Ig found in healthy adult patients. Linearity was established from 0.005 to 5 g/dL. Subsequently we have performed similar experiments to establish if these observations would hold true for additional monoclonal therapies. Rituximab (an anti-CD20 monoclonal IgG1 Kappa clone) was spiked into pooled human serum ranging from 0.0005 to 0.025 g/dL. Analysis of rituximab using miRAMM generated a charge state distribution of ions consistent with previous reports and the converted molecular mass for pooled serum spiked with 0.025 g/dL rituximab displayed a single peak with the observed molecular mass of 23,035 Da (Fig. 4A). This molecular mass agrees with the average molecular mass calculated from the known amino acid sequence for the kappa LC of rituximab (23,035.4 Da) containing a pyroglutamic acid modification at the N-terminus as well as maintaining its LC intrachain disulfide bonds. Rituximab was identifiable above the polyclonal background at concentrations as low as 0.0005 g/dL (Fig. 4C) and demonstrated a linear
response across the range of concentrations (Fig. 4B). These results demonstrated miRAMM can detect monoclonal Ig below the limit of detection of PEL (0.05 g/dL) and IFE (0.005 g/dL) [39]. To demonstrate the improved functional sensitivity of miRAMM, a human serum sample containing an IgA kappa (IgAK) M-protein of 2 g/dL was run by IFE and the same sample was analyzed by miRAMM (Fig. 5). A distinct immunoprecipitated band present in the A (IgA anti-sera) and K (j anti-sera) lanes indicate an M-protein of the IgAK isotype. The same sample analyzed by miRAMM demonstrates the presence of a monoclonal LC with an accurate molecular mass of 23,334 Da. The asterisk indicates the presence of a sulfate adduct with a mass of 23,430 Da (23,334 Da + 96 Da). The same sample was then diluted 1:1000 into pooled serum from healthy individuals and re-analyzed by IFE and miRAMM. In the case of IFE the M-protein is no longer detectable, however, by miRAMM we can still detect a monoclonal LC at 23,333 Da which is within 1 Da of the monoclonal LC in the undiluted sample (Fig. 5B). Similarly to PEL and IFE, the limit of detection of a given monoclonal LC by miRAMM is partly determined by the physical properties of the LC. In the case of PEL and IFE, the migration of the LC dictates its limit of detection. A LC that co-migrates with other proteins during PEL charge separation will be more difficult to detect due to interference from co-migrating protein. On the other hand, if the LC has a distinct migration pattern such that the LC migrates well ahead of or behind the polyclonal LC population where other proteins do not migrate the ability to detect that LC is increased. A similar phenomenon exists for miRAMM. Given the polyclonal background has a Gaussian distribution of masses, if the mass of the LC is found at the peak of the distribution then the sensitivity maybe reduced relative to LC with a mass corresponding to the tails of the Gaussian distribution, however, miRAMM has the advantage of two dimensional separation in that the both retention times and mass/charge are used to define to monoclonal LC thus providing miRAMM with additional sensitivity compared to
62
J.R. Mills et al. / Methods 81 (2015) 56–65
Fig. 5. miRAMM can offer superior sensitivity to detect M-proteins compared with IFE. IFE results from the analysis of a serum sample from a patient with multiple myeloma. (A) IFE results from the analysis of a serum sample from a patient with multiple myeloma before and after a 1:1000 dilution into pool human serum. A distinct IgAK M-Protein is present in the ‘‘A’’ lane (IgA HC) lane and the ‘‘K’’ (kappa LC) lane is found in the neat sample. However, in the diluted sample the IFE is read as negative without a distinct banding pattern indicating an M-protein. (B) The corresponding deconvoluted mass spectrum showing predominant monoclonal LC with a molecular mass of 23,334 Da, which is within the expected mass range for a LC. The * denotes a likely salt adduct with a mass shift of 96 Da – consistent with the mass of a sulfate ion. The k and j symbols designate the polyclonal lambda and kappa LC distributions, respectively. In the diluted sample the M-Protein is still detectable although at 1000-fold lower intensity.
Fig. 6. miRAMM can differentiate monoclonal therapies from endogenous M-proteins. Serum samples were collected before and after treatment from a patient with a MGUS receiving rituximab therapy for an autoimmune condition. The top panel represents the multiply charged LC ions and the corresponding deconvoluted mass spectra demonstrating a single monoclonal LC with a retention window of 5 min and a mass of 23,614 Da. The bottom panel is serum from the patient 2 days later after receiving an infusion with rituximab. The appearance of a second monoclonal LC has a mass of 23,035 Da which is identical to that of rituximab and had a retention time of 5.3 min consistent with rituximab.
IFE/PEL. In addition, k and j have different mass distributions which are distinct from each other. Lastly, miRAMM has superior resolution (see Fig. 6). Initial work using patient samples demonstrated that accurate molecular mass of light chains is a viable approach to monitor patients over years. Using serial patient samples it was demonstrated that the accurate molecular mass of the disease associated monoclonal light chains deviated <1 Da over a 7-year period which is within the expected mass measurement error of the mass
spectrometer. Importantly, residual disease was detectable even in periods of remission [38]. A follow up study looked for monoclonal light chains in urine samples with detectable monoclonal proteins by IFE. miRAMM identified 112 of 115 urine samples as positive – demonstrating a high level of concordance with current clinical laboratory practice but with negligible reagent costs and minimal sample processing [37]. miRAMM was only tested on neat urine samples without concentrating the sample. Interestingly, the three samples missed by miRAMM all were exceptionally dilute
J.R. Mills et al. / Methods 81 (2015) 56–65
63
Fig. 7. CaptureSelect enrichment improves recovery of IgA Igs. Representative serum samples positive for an IgG and IgA M-protein were analyzed by miRAMM using MelonGel™ enrichment (upper panel) or CaptureSelect enrichment (bottom panel). Both the multiply charged LC ions and the corresponding deconvoluted mass spectra demonstrating a representative the accurate molecular mass of the M-protein LC are shown.
and required concentration prior to analysis by PEL/IFE. It is possible concentration of urine prior to miRAMM will be required in order to assure maximum sensitivity. In most serum samples analyzed by miRAMM, two distinct Gaussian-like distributions of light chains are present (k and j respectively in (Fig. 3) [38]. The two distinct LC polyclonal populations are postulated to be due to different molecular masses of the lambda and kappa constant region as well as unique framework 1 (FR1) usage [41]. The Gaussian distribution for each unique LC population is likely representative of the randomness of VJ recombination within the VL domain. Comparison of the relative abundance of the polyclonal kappa and lambda populations as measured by miRAMM compared well to the total LC ratios (kappa:lambda) measured by a clinical nephelometry assay highlighting another dimension of miRAMM [41]. Given the accurate molecular mass accuracy of miRAMM, it can be useful in ruling out the presence of monoclonal therapeutics which may be detected by IFE or PEL in patients being monitored for the presence of MGUS [42]. This is likely to become more prevalent and problematic in myeloma patients as high dose monoclonal therapeutics become part of the treatment regime [43]. To demonstrate this, serum samples from a patient with an IgGK (IgG HC and kappa LC) MGUS and an autoimmune condition taken before and after infusion with rituximab therapy (a monoclonal IgGK anti-CD20 therapy) was measured using miRAMM. The rituximab LC has an anticipated mass of 23,035 Da. miRAMM is useful in determining that the appearance of a second monoclonal LC is due to the presence of rituximab, identifiable by its accurate molecular mass, rather than the appearance of a new M-protein. Although only a limited number of monoclonal therapies have been evaluated by miRAMM, thus far each has a unique combination of liquid chromatography retention times and molecular masses – making this an acceptable approach to eliminating the
potential confusion monoclonal therapies pose. A more comprehensive analysis using miRAMM of the commonly prescribed monoclonal therapies is warranted. The initial design of miRAMM utilized Melon-Gel™ to enrich for Igs. Melon-Gel™ has lower recovery for M-proteins of the IgA and IgM class – thus subsets of M-proteins have demonstrated limited sensitivity. In attempt to improve upon this we have used an alternative mode of Ig enrichment using a mixture of CaptureSelect anti-j/k beads to affinity purify Igs prior to analysis. CaptureSelect affinity reagents have the benefit of binding LCs regardless of the corresponding HC. Serum samples positive for either IgG or IgA M-proteins by PEL (20 g/L monoclonal Igs) were analyzed by miRAMM using Melon-Gel™ and CaptureSelect beads simultaneously (Fig. 7A and B). Use of CaptureSelect beads enhances recoveries for IgA clones. This bias does not exist for IgG clones as both enrichment methods yield similar signal intensities. This new enrichment scheme has demonstrated the ability to isolate all Igs regardless of HC type over a wide variety of sample types (data not shown). 4.1. Current limitations One current limitation of miRAMM is the inability to detect monoclonal proteins at concentrations lower than that of the polyclonal Igs in at the same mass range and retention time. Given the vast number of distinct Igs present in serum (estimated at >106 clones/individual) and the limited ranges of molecular masses observed, in order to identify a clone with certainty the intensity of the monoclonal LC needs to be present above the polyclonal background. One method to improve sensitivity is to use bottom-up proteomics rather than using the intact mass of light chains. A commonly used protocol begins with denaturation of
64
J.R. Mills et al. / Methods 81 (2015) 56–65
the protein, reduction of disulfide bonds, alkylation to present reformation of disulfide bonds and then digestion typically with endopeptidases such as trypsin, chymotrypsin, Lys-C or Glu-C. The digestion is terminated and then the peptide mixture is separated using on-line using chromatographic methods and then analyzed by ESI-MS or MALDI-MS. Rather than reconstructing the peptide sequence of the entire Ig, a clonotypic peptide which is unique to a specific Ig (either the HC, LC or both) can be found using LC–MS/MS and used as a surrogate marker. This method has been clinically applied to quantitating IgG subclasses as well as for qualitatively monitoring residual myeloma [14,21,44]. Internal standards, typically stable isotope labeled peptides with the identical sequence of interest are used or by using an isotope-labeled Ig standard [45,46] [47]. Our group has also demonstrated that a surrogate Ig from another species can serve to normalize for variation in digestion efficiency [21]. In order for a proteotypic peptide to serve as a unique surrogate marker of the intact monoclonal protein it must be derived from the variable region of Ig – which assumes that the appropriate amino acids residues are available to generate appropriate digestion sites. It is also assumed that the peptide of interest cannot be present in other Igs. Although using peptides offers superior sensitivity the process is more expensive, more laborious, and more time consuming compared with miRAMM and more challenging in the absence of prior knowledge of the Ig sequence. miRAMM offers a work flow more conducive to the clinical laboratory – minimal reagents, minimal pre-analytical processing, and the ability generate a snapshot the entire immunoglobulin repertoire – all at a fraction of the cost of tests designed to provide the same information but employing other platforms. miRAMM was initially designed as to tool for detecting monoclonal LCs due to their small molecular mass and the absence of post-translational modifications as compared to the HC. In its current format miRAMM can resolve monoclonal IgG HCs [38]. However, it has not been thoroughly tested to determine if it is sufficient to provide HC isotype information for IgM and IgA heavy chains which may be difficult due to the heterogeneity in the carbohydrate on the Fc portion of these isotypes. Furthermore, the use of a reducing agent as currently employed by miRAMM means that the LC spectrum corresponds to both FLCs as well as LCs associated with a HC. Therefore, in its current format, miRAMM will not provide the same information which is obtained by using the FLC assay. However, it should be noted that this technology could provide isotype information as well as detection of HCs once the method is refined to include either top-down fragmentation information or utilization of alternative Ig purification methods that would be selective towards HC and/or FLCs (for instance purification with resin specific to FLCs). Until this is accomplished miRAMM best fills a clinical role in improving specificity of current detection methods using accurate mass measurements as well as providing a way to detect lower levels of monoclonal LCs. In order to provide the same information as PEL/IFE modifications to the current methodology are required.
5. Conclusion The current methodologies used to screen for monoclonal proteins in the clinical laboratory are based on PEL/IFE and subsequent evaluation of the pattern of protein banding to determine if an M-protein is present. These methods have proven robust but suffer from limited sensitivity, making it impossible to rule out low amounts of residual disease. Furthermore, interfering proteins may be present making interpretation challenging. Given that clonal plasma cells responsible for monoclonal gammopathies secrete Igs with unique amino acid sequences accurate molecular mass can
be with high precision and accuracy creating added value in specificity in serial samples from a patient and when monoclonal therapeutics are present. Furthermore, the sensitivity to detect a monoclonal Ig above the polyclonal Ig background is significantly improved compared to PEL and IFE. Identification of specific proteins using accurate molecular mass has already been implemented into clinical practice. It is routine clinical practice to refer back to a patient’s past history of PEL and IFE results when interpreting subsequent samples in an aim to provide consistent interpretation and to improve accuracy of resulting. One can easily envisage that in the future, part of the patient’s clinical record will include documentation of a patient’s monoclonal Igs with an accurate mass measurement which can serve as a reference for future analysis. The studies and results discussed here provide evidence to assert the utility of mass spectrometry as a tool to monitor an M-protein in patients with a monoclonal gammopathy. By having a more sensitive and specific MS serum based assay to detect residual M-proteins this may be amendable to replacing costly and invasive methodologies currently used to establish the absence of minimal residual disease. miRAMM represents a major shift in the way M-proteins are interpreted and will likely transform the way plasma cell disorders will be tracked in the future. Lastly, monitoring of M-proteins is not the only use of gel electrophoresis to monitor Ig banding as a part of the diagnostic work up. It is likely miRAMM may find a role as a technological improvement over other methodologies as well. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]
[36] [37] [38] [39] [41] [42]
K. Rajewsky, Nature 381 (6585) (1996) 751–758. C. Vettermann, M.S. Schlissel, Immunol. Rev. 237 (1) (2010) 22–42. M.P. Lefranc, Cold Spring Harbor Protoc. 2011 (6) (2011) 595–603. J.A. Katzmann et al., Clin. Chem. 55 (8) (2009) 1517–1522. J.A. Katzmann et al., Clin. Chem. (2014). C.R. McCudden et al., Am. J. Clin. Pathol. 129 (3) (2008) 451–458. C.E. Bystrom, S. Sheng, N.J. Clarke, Anal. Chem. 83 (23) (2011) 9005–9010. N.J. Clarke, A.N. Hoofnagle, Clin. Lab. Med. 31 (3) (2011) (p. ix–xi). A.E. Acosta-Martin et al., Anal. Chem. 85 (16) (2013) 7971–7975. H.R. Bergen 3rd., S.R. Zeldenrust, S. Naylor, Amyloid 10 (3) (2003) 190–197. P.V. Bondarenko et al., J. Am. Soc. Mass Spectrom. 20 (8) (2009) 1415–1424. S. Rosati et al., Nat. Protoc. 9 (4) (2014) 967–976. D. Murray, D. Barnidge, Crit. Rev. Clin. Lab. Sci. 50 (4–5) (2013) 91–102. D.R. Barnidge et al., J. Proteome Res. 13 (4) (2014) 1905–1910. D.R. Barnidge et al., J. Proteome Res. (2014). C.M. Botz et al., Br. J. Haematol. (2014). S.J. Gaskell, J. Mass Spectrom. 32 (7) (1997) 677–688. A. Tartakoff, P. Vassalli, J. Cell Biol. 83 (2 Pt 1) (1979) 284–299. P. Gagnon, J. Chromatogr. A 1221 (2012) 57–70. A. Beck et al., Nat. Rev. Immunol. 10 (5) (2010) 345–352. P.M. Ladwig et al., Clin. Chem. 60 (8) (2014) 1080–1088. S.K. Chowdhury et al., J. Am. Soc. Mass Spectrom. 1 (5) (1990) 382–388. J. Porath, F. Maisano, M. Belew, FEBS Lett. 185 (2) (1985) 306–310. T.W. Hutchens, J. Porath, Anal. Biochem. 159 (1) (1986) 217–226. E. Boschetti, J. Biochem. Biophys. Methods 49 (1–3) (2001) 361–389. M. Wilchek, I. Chaiken, in: P. Bailon et al. (Eds.), Affinity Chromatography, Humana Press, 2000, pp. 1–6. B. Akerstrom, L. Bjorck, J. Biol. Chem. 264 (33) (1989) 19740–19746. A. Staub et al., J. Pharm. Biomed. Anal. 55 (4) (2011) 810–822. D.S. Rehder et al., J. Chromatogr. A 1102 (1–2) (2006) 164–175. T.M. Dillon, P.V. Bondarenko, M. Speed, J. Chromatogr. A 1053 (1–2) (2004) 299–305. J.B. Fenn et al., Science 246 (4926) (1989) 64–71. M. Karas, F. Hillenkamp, Anal. Chem. 60 (20) (1988) 2299–2301. H.S. Gadgil et al., J. Am. Soc. Mass Spectrom. 17 (6) (2006) 867–872. L. Yu, R.L. Remmele Jr., B. He, Rapid Commun. Mass Spectrom. 20 (24) (2006) 3674–3680. E. Johansen, J. Albanese, C. Hunter, Analysis of intact and reduced therapeutic monoclonal antibodies using the TripleTOF™ 5600 system, 2011. Available from:. M. Samonig, C. Huber, K. Scheffler, ThermoSci Application Note 591. C.M. Botz et al., Br. J. Haematol. 167 (3) (2014) 437–438. D.R. Barnidge et al., J. Proteome Res. 13 (3) (2014) 1419–1427. D.L. Murray et al., Am. J. Clin. Pathol. 138 (4) (2012) 609–613. D.R. Barnidge et al., J. Proteome Res. 13 (11) (2014) 5198–5205. C.R. McCudden et al., Clin. Chem. 56 (12) (2010) 1897–1899.
J.R. Mills et al. / Methods 81 (2015) 56–65 [43] J.P. Laubach et al., Expert Opin. Investig. Drugs 23 (4) (2014) 445–452. [44] E.R. Remily-Wood et al., Proteomics Clin. Appl. 8 (9–10) (2014) 783–795. [45] D.R. Barnidge et al., J. Proteome Res. 3 (3) (2004) 644–652.
[46] J.R. Barr et al., Clin. Chem. 42 (10) (1996) 1676–1682. [47] O. Heudi et al., Anal. Chem. 80 (11) (2008) 4200–4207.
65
Contents lists available at ScienceDirect
Methods journal homepage: www.elsevier.com/locate/ymeth
Detecting monoclonal immunoglobulins in human serum using mass spectrometry John R. Mills ⇑, David R. Barnidge, David L. Murray ⇑ Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN 55905, United States
a r t i c l e
i n f o
Article history: Received 3 February 2015 Received in revised form 15 April 2015 Accepted 18 April 2015 Available online 24 April 2015 Keywords: Monoclonal gammopathy Immunoglobulin Multiple myeloma Mass spectrometry Clinical diagnostics
a b s t r a c t Established guidelines from the International Myeloma Working Group recommend diagnostic screening for patients suspected of plasma cell proliferative disease using protein electrophoresis (PEL), free light chain measurements and immunofixation electrophoresis (IFE) of serum and urine in certain cases. Plasma cell proliferative disorders are generally classified as monoclonal gammopathies given most are associated with the excess secretion of a monoclonal immunoglobulin or M-protein. In clinical practice, the M-protein is detected in a patients’ serum by the appearance of a distinct protein band migrating within regions typically occupied by immunoglobulins. Given each M-protein is comprised by a sequence of amino acids pre-defined by somatic recombination unique to each clonal plasma cell, the molecular mass of the M-protein can act as a surrogate marker. We established a mass spectrometry based method to assign molecular mass to the immunoglobulin light chain of the M-protein and used this to detect the presence of M-proteins. Our method first enriches serum for immunoglobulins, followed by reduction to separate light chains from heavy chains, followed by microflow LC-ESI-Q-TOF MS. The multiply charged light chain ions are converted to their molecular mass and reconstructed peak area calculations are used for quantification. Using this method, we term ‘‘monoclonal immunoglobulin Rapid Accurate Molecular Mass’’ or miRAMM, the presence of M-proteins can be reliably detected with superior sensitivity compared to current gel-based PEL and IFE techniques. Ó 2015 Published by Elsevier Inc.
1. Introduction B-cells play a central role in the adaptive immune system expressing a diverse repertoire of immunoglobulins (Igs). Secreted Igs protect against foreign antigens through several mechanisms including; binding to foreign antigens (i.e. bacteria, viruses), recruiting other immune cells, or directly neutralizing foreign substances [1]. Igs are homodimers composed of two heavy chains (HCs) and two identical light chains (LCs). Each LC pairs with a HC and each HC pairs with another HC through disulfide bonds (see Fig. 1). Both the HC and the LC have two distinct regions called the constant and variable region. The C-terminal half of each HC and LC encompasses the constant region which contains an epitope that defines the chain’s isotype. The HC can have 5 different isotypes; IgG, IgA, IgM, IgD or IgE, while the LC can have 2; kappa or lambda. The N-terminal half of each HC and LC contains the ⇑ Corresponding authors at: Department of Laboratory Medicine and Pathology, Mayo Clinic, 200 1st St SW, Rochester, MN 55905, United States. Fax: +1 (507) 284 1927. E-mail addresses: [email protected] (J.R. Mills), [email protected] (D.L. Murray). http://dx.doi.org/10.1016/j.ymeth.2015.04.020 1046-2023/Ó 2015 Published by Elsevier Inc.
variable region which includes the amino acid sequences that bind antigens. A single B-cell will create HC and LC variable regions by rearranging Ig specific gene segments [1]. If both heavy and light chain rearrangements generate functional sequences the expressed Ig is assembled and expressed on the cell surface of the B-cell. If this process is successful then the cell undergoes allelic exclusion to ensure each cell only expresses a single Ig [2]. Therefore, each clonal cell is uniquely defined by a particular gene rearrangement that encodes a specific Ig with a defined molecular mass. When a B-cell engages an antigen the cell is stimulated to proliferate which induces clonal expansion. B-cell clones that acquire mutations that improve antigen binding affinity will be selectively expanded in the germinal center. Those cells which differentiate into Memory B-cells may undergo class switching at the IgH locus which generates Igs encoding different constant regions prior to commitment to becoming long-lived plasma cells (PCs). These long-lived PCs reside in the bone marrow and secrete vast quantities of high-affinity antigen specific Igs. Each PC will express a defined Ig with a molecular mass which is identical to its original precursor cell. Humans produce a vast diversity of PCs – each with their own randomly generated Ig (upwards of 1012 possible unique
J.R. Mills et al. / Methods 81 (2015) 56–65
57
Fig. 1. Ig structure indicating domains and sites of covalent modification. Structure of a typical IgG immunoglobulin. Two identical HCs and LCs are connected by disulfide linkages. Furthermore, within each chain there are also disulfide bonds which are critical to maintain structure. The variable regions of both the HC and LC form the bivalent antigen binding sites. The class of the HC and LC are defined by the constant regions which have distinct amino acid residues depending on the class. Carbohydrate posttranslation modifications on the HC generate different Ig glycoforms which can be functionally important.
sequences) creating what is often referred to as the normal polyclonal Ig repertoire [3]. However, PCs can become malignant resulting in an abnormal expansion of clonal cells as is the case for the plasma cell proliferative disorders, multiple myeloma (MM), AL amyloidosis, and Waldenström’s macroglobulinemia. When this occurs a single monoclonal Ig can be detected above the normal polyclonal Ig repertoire. If there is clinical suspicion of a plasma cell proliferative disorder, the patient’s serum and urine is typically tested for the presence of a monoclonal Ig or M-protein. M-proteins are typically detected using a combination of protein gel electrophoresis (PEL) and immunofixation (IFE) [4]. These techniques use the differential charge and size of Igs to separate them from other serum proteins on a gel media, usually agarose or cellulose acetate. Typically a low ionic strength buffer with an alkaline pH is used, which is sufficient to provide most proteins with a negative charge. After an electric field is applied the serum proteins migrate into distinct fractions knows as the albumin, a1 (composed predominantly of a1-antitrypsin), a2 (mostly a2-macroglobulin and haptoglobin), b (transferrin and C3) and c (mostly Igs) regions. After separation, proteins in the gel are stained and the intensity of the staining is used to determine the relative contribution of a given fraction to the total serum protein content. In most healthy individuals the c region contains the polyclonal Igs found in serum and is a broad and diffuse fraction driven by the vast diversity of Ig amino acid sequences. When an excessive amount of a specific monoclonal Ig is present a characteristic distinct monoclonal band or spike (M-spike) is typically observed in the c fraction [5]. In instances where the presence of an M-protein is questionable, immunofixation electrophoresis (IFE) can be helpful. IFE is similar in principle to PEL with an additional in-gel immunoprecipitation step. Several identical aliquots of the patient’s serum are separated by agarose gel electrophoresis in different wells. After separation one well serves as a reference well and is immediately fixed and whereas the other wells are incubated with antisera against the HC isotypes (IgA, IgG, IgM) and LC isotypes (j or k). The interaction of the antisera with the Ig in the gel induces a precipitate. Distinct bands are noted for each well and the relationship between HC and LC is established based upon similar monoclonal migration distances (see Fig. 5). IFE has a limit of detection that is 10-fold lower than PEL; however, it is important to note IFE is not quantitative and is
only useful for qualitative assessment. Capillary electrophoresis (CE) can be used as an alternative to gel electrophoresis [6]. CE uses small-bore (10–100 lM) fused silica capillary tubes and microfluidics to separate proteins which are then visualized using a UV detector. CE offers superior speed and automation capabilities. The equivalent to gel based IFE is immunosubtraction CE. In the case of the latter, serum is incubated with antisera for each of the HC and LC, then removing the complexes prior to running CE. The results are compared back to the serum sample without antisera to make a qualitative conclusion about the type of Ig present. Despite these technical improvements these methods have inherent limitations due to their laborious setup, poor resolution, limited sensitivities, and occasionally ambiguous results. The most recent generation of mass spectrometers provide higher resolution, greater mass measurement accuracy, and greater sensitivity than previous generations. These new instruments (i.e. Orbitrap and Time-of-Flight (TOF) mass spectrometers) are also smaller and more robust and have begun to play a role in clinical diagnostics measuring the accurate molecular mass of intact proteins (i.e. IGF1, hemoglobin) [7–10]. Increases in the linear dynamic range, as well as increases in resolution and mass measurement accuracy, have made these instruments a sensible alternative to characterizing proteins as compared to traditional protein gel electrophoresis based methods. This was recently demonstrated for intact Igs where the molecular masses of therapeutic monoclonal Igs were determined with high accuracy [11– 13]. Accordingly, manufacturers of therapeutic monoclonal Igs are using the newest generation of mass spectrometers to evaluate product quality via accurate molecular mass determinations of recombinant Igs. Our group recently demonstrated how microflow liquid chromatography coupled with electrospray ionization (ESI) and Q-TOF MS (microLC-ESI-Q-TOF MS) could be used to identify and monitor a monoclonal immunoglobulin in a patient’s serum and urine [14–16]. Instead of tracking the intact monoclonal Ig, the disulfide bonds holding the HC and LC together were reduced using DTT and the LC was the primary analyte monitored by MS. Once reduced, LCs were further separated from HCs using liquid chromatography and these LCs were then injected into the mass spectrometer where their masses are obtained. There are several analytical advantages to measuring LCs rather than the HC or the
58
J.R. Mills et al. / Methods 81 (2015) 56–65
intact Ig. First, LC masses range from 22,000 to 25,000 Da compared to intact Igs which are typically in a mass range of 150,000 Da. Smaller proteins ionize much more readily, a critical component of MS, making detection of light chains more sensitive compared to intact Igs [17]. Second, Igs are commonly post-translationally modified predominantly through glycosylation which is typically restricted to the HC portion of the Ig [18]. Using microLC-ESI-Q-TOF the molecular mass of a patient’s myeloma-associated monoclonal LC was used to track their disease over time [14,15]. This work established that monoclonal LCs corresponding to M-proteins in either serum or urine could be identified, quantified, and isotyped using this methodology. Relative to PEL, IFE and FLC, this approach promises more analytical sensitivity and specificity because of high accuracy mass measurements. Here we address the methods used to determine the accurate molecular mass of a monoclonal immunoglobulin in human serum using mass spectrometry. These methods have the potential to alter how clinical laboratories screen and monitor for the presence of monoclonal gammopathies. 2. Technical considerations Below we will discuss steps that are critical in developing a MS based method to monitor Igs in human serum. The analysis of large proteins by MS presents distinct challenges compared to small non-proteinaceous molecules most often analyzed by MS in the clinical laboratory. 2.1. Immunoglobulin purification Current methods for monitoring Igs by MS require at least one purification step depending on the depth of analysis. Human serum is highly proteinaceous containing 6.3–7.9 g/dL of protein, 2/3 of which is albumin. In order to improve downstream analytical performance (including maintaining column lifetime and sensitivity) it is necessary to remove albumin and other contaminating non-immunoglobulin proteins from serum. There has been a long standing need for methods to purify antibodies from different sources as a prerequisite for use in biomedical research, diagnostic methodologies and more recently as therapeutics. There are several strategies to purify antibodies and each has been advantages and disadvantages depending on the downstream application. When evaluating Ig therapeutics, maintaining native confirmation is critical whereas for structural determinations and quantitation it may be acceptable to purify antibodies under harsher conditions (acid, high salt etc.). Methodologies for purification can be separated into physiochemical and affinity approaches. Physiochemical separation of antibodies includes the use of size exclusion chromatography (SEC), ion exchange chromatography (IEC), ammonium sulfate precipitation, caprylic acid precipitation, thiophilic adsorption and newer chemistries such as Melon-Gel™ resin. Affinity purification can be split into two groups – proteins that bind specific classes of Igs (i.e. Protein A, G, L or anti-Ig antibodies) or antigens that bind to a specific Ig (i.e. immobilized p53 peptides to enrichment for anti-p53 Igs) [19]. Plasma cell proliferative disorders can be associated with any of the HC and LC Ig classes. Therefore the ideal purification scheme can purify all classes of Igs. Furthermore, it is important to consider reproducibility, scale, robustness, cost and the time required for a given purification scheme. The pharmaceutical industry has driven demand for robust and cost-effective methods to purify the IgG class since all therapeutic monoclonal antibodies are exclusively IgG [20]. Many preparative Ig purification schemes are designed to purify large quantities of monoclonal Igs with lengthy clean up steps. However, in the clinical laboratory, most assays utilize
sample volumes in the range of lLs and pre-analytical steps that take minutes and hours rather than days. Listed below are several methodologies that have the potential to be incorporated into the clinical laboratory to purify Igs prior to analysis by mass spectrometry. Ammonium sulfate precipitation is a common technique to purify and concentrate Igs from serum. This technique works by ‘‘salting-out’’ Igs selectively. Igs precipitate out of solution at lower salt concentrations compared with many other serum proteins (including albumin) due to poor solubility. At 40–50% of ammonium sulfate saturation most Igs will precipitate whereas albumin remains in solution. The precipitate containing the Igs can be re-dissolved in buffer and then further processed. This method is sufficient as a first pass step in the purification of Igs. It is robust and has proven clinical utility when quantitating tryptic peptides [21] using mass spectrometry. However, the high concentration of sulfate ions can result in sulfate adducts complicating the mass spectrum and reducing sensitivity [22]. Thiophilic adsorption chromatography (TAC) was originally described in 1985 [23]. All Igs are thiophilic proteins (affinity for sulfur containing molecules) while most other proteins including albumin are not [24]. Immobilized thioether-sulfones can be utilized to selectively bind Igs in the presence of certain salts (i.e. potassium sulfate). Subsequently the use of thiophilic agarose beads has become well established as a robust and cost-effective method to purify Igs and immunoglobulin fragments from a variety of sources including human serum [25]. A major benefit being Igs can be eluted from beads by dilution into physiological buffers containing low salt concentrations at neutral pH. Thiophilic adsorption has a unique property in that it has high capacity to adsorb the three major classes of Igs (IgG, IgA, and IgM) including their subclasses. The use of high concentrations of sulfate ions necessitates desalting prior to analysis by MS. Traditionally this methodology had been utilized for selective depletion of Igs from serum. Protein A, G and L are bacterial proteins with defined antibody binding properties. Recombinant versions of these proteins are commercially available and offer wide applications in purification, immobilization and detection of Igs. Protein A and G bind the fragment crystallizable (Fc) region of most Igs and also within the fragment antibody-binding (Fab) region (Fig. 1). However, the binding affinities vary significantly between Ig isotypes and between subclasses (i.e. IgG3). Protein A and G bind IgG very strongly but IgM binding is poor due to inaccessibility of binding sites on IgM and its pentameric structure [26]. Protein L has selective binding affinity for kappa LCs and therefore for can be useful for purifying all HC Ig isotypes containing kappa LC but fails to bind most lambda LCs [27]. Melon Gel™ is advantageous as it is simple, fast, and inexpensive (Pierce Biotechnology). It is based upon selective charge attraction of albumin and other unwanted proteins to the resin and charge repulsion of Igs. However, it is important to recognize that Melon Gel™ was developed for the purification of IgG and the recovery of IgA and IgM is lower. The CaptureSelectÒ Antibody Toolbox affinity matrix (developed by Bio Affinity Company) offers a variety of products with specific binding affinity for all HC and LC isotypes and subclasses. These products are recombinant single-chain camelid Igs that lack the heavy and light chains found in other animal Igs including human and they are produced in Saccharomyces cerevisiae thus are free of any trace animal Igs. These affinity beads can be utilized to purify all human Igs from serum regardless of type. However, the cost of these reagents is greater than the non-affinity based resins and incubation times are longer. In developing a methodology for clinical use, robustness, cost and preparation time are paramount. All of the methods above
J.R. Mills et al. / Methods 81 (2015) 56–65
59
Fig. 2. Overview of ‘‘monoclonal immunoglobulin Rapid Accurate Molecular Mass’’ or miRAMM. Igs are first purified from human serum and subsequently reduced with DTT which is sufficient to break inter-chain disulfide bonds. The dissociated mixture of HCs and LCs is then separated by RPLC using C4 column chromatography prior to analysis by TOF-MS. The resulting spectra of multiply charged ions is deconvoluted into the accurate molecular mass spectra. Peaks above the polyclonal LCs are indicative of monoclonal Igs. k and j depict the two different light chain populations which correspond to the lambda and kappa LCs, respectively.
have potential to meet these criteria, however, optimization is necessary and there have not been comprehensive studies today to confirm their suitability to purify a diverse range of Igs which would be encountered when screening and monitoring monoclonal gammopathies. The pre-analytical preparation of samples is critical to downstream mass spectrometry analysis; even the best mass spectrometry will fail to produce reproducible mass measurements in the absence of high quality samples. Initial analysis utilizing miRAMM focused on using Melon Gel given its cost benefit and simplicity of use. More extensive analysis will be necessary to determine if this method will be amendable across thousands of samples with diverse matrices and Ig subtypes. 2.2. Reverse-phase liquid chromatography There are several types of liquid chromatography which can be used to separate Igs. These include; reverse-phase liquid chromatography (RPLC), affinity chromatography, SEC, IEC, or hydrophilic interaction chromatography. Overall, RPLC is the most widely used in-line chromatographic technique since the mobile phase buffers are amenable to electrospray ionization. However, RPLC has several disadvantages that need to be overcome in order to make it a viable approach for the analysis of HC and LC Igs. RPLC columns have a limited protein capacity therefore care needs to be taken to eliminate overloading by using the appropriate amount of patient serum. An overloaded RPLC column will suffer from carryover, poor resolution, and irreproducible retention times. RPLC systems with large dead volumes also run the risk of having unnecessary protein absorption resulting in additional carryover. The use of inert tubing with small diameters such as PEEK-Sil can reduce protein adsorption in the liquid chromatography system. Carryover can also be problematic in hydrophobic stationary phase columns. The use of columns with shorter alkyl chains (C3, C4, and C8) can reduce peak tailing, peak broadening and increase recovery. Incorporating small amounts of formic acid (0.1%) can improve peak shape by improving solvation without suppression of ionization [28]. In order to successfully separate larger intact proteins it is critical to use porous packing material with larger pore sizes as the rate of analyte diffusion is dependent on pore size
to analyte size ratio. The use of 300 Å pore sizes has been demonstrated to provide superior separation for Igs [29]. Further improvements in separation performance for large proteins have been made by the use of columns with fused-core technology. The core of the column is composed of fused-silica which has poor penetration by analytes. Therefore this reduces the potential for diffusion providing for superior peak shape for intact proteins. Heating a column to at least 50 °C using an oven heater is essential to avoid carryover and improve peak shape [30]. Also, the addition of 10% or more 2-propanol to the acetonitrile in the organic phase will aid in reducing carryover. The use of liquid chromatography is an essential component in resolving monoclonal Igs present within the polyclonal Ig repertoire found in human serum. By separating Igs based on hydrophobic interactions specific monoclonal Igs can be separated away from a portion of the polyclonal Ig repertoire which increases the limit of detection – in this case defined as the ability to detect a given monoclonal Ig above the polyclonal background. 2.3. Mass spectrometry The development of ESI made it possible to convert ions in solution to the gas phase, allowing mass spectrometers with limited m/z range to analyze large proteins [31,32]. In the case of Igs, ESI produces ions with charge states upwards of +45 for HC +25 for LC bring the ions into the range of 600–2500 m/z. ESI-TOF and ESI-Q-TOF mass spectrometers are ideally suited to determine the accurate molecular mass of intact Ig [33] as well as HC and LC in this m/z range. As an example, ESI-Q-TOF pyroglutamate modifications were identified on recombinant intact monoclonal Igs using analysis after partially reducing Igs to release the light chain. Deconvoluted spectra demonstrated an 18 Da mass shift consistent with a N-terminal glutamate to pyroglutamate conversion which was confirmed by bottom-up proteomics (discussed later on) [34]. Orbitrap mass analyzers have also been used to provide accurate molecular mass of monoclonal Igs providing higher resolution and better mass accuracy than TOF analyzers [11]. Comparisons of the two instrument platforms have been published using the recombinant therapeutic rituximab as the model
60
J.R. Mills et al. / Methods 81 (2015) 56–65
monoclonal Ig [35,36]. Although these limited studies were performed on purified monoclonal Ig in the absence of polyclonal background they did demonstrate the capability of high resolution MS to identify the accurate molecular mass of monoclonal Ig with resolutions below <5 ppm. Furthermore, as little as 1 ng of material was readily measureable. Work by Barnidge and co-workers demonstrated similar resolution and sensitivities for the recombinant monoclonal Igs adalimumab spiked into normal human serum matrix [37,38]. 3. Method description Here we outline a cost effective, robust mass spectrometry-based method previously developed [14,38] and subsequently refined for identifying monoclonal Igs in human serum that provides superior sensitivity compared to current clinical gel-based assays PEL and IFE. This methodology ‘‘monoclonal immunoglobulin Rapid Accurate Molecular Mass’’ or miRAMM utilizes microflow liquid chromatography-ESI-TOF MS to measure the accurate molecular mass of the LC portion of intact monoclonal Igs which serves as a surrogate marker of the intact protein (Fig. 2) [16]. Igs are purified off-line, reduced to release the LCs from HCs and then separated using microflow liquid chromatography prior to ESI-TOF MS. The mass spectra of multiply charged light chain ions generated during ESI are converted to their molecular mass to produce spectra of the distribution of the entire LC repertoire in human serum enabling accurate molecular mass measurements for monoclonal Igs and quantitation using reconstructed peak area. In healthy adults mass spectra show two distinct normally distributed peaks (corresponding to lambda and kappa polyclonal LCs [32] – discussed below) with a smooth distribution of clones (Fig. 3). The presence of a peak above the polyclonal background is consistent with the presence of a monoclonal Igs – similarly to current gel based methodologies with the added benefit of improved resolution and accurate mass measurements within 1 Da. 3.1. Reagents Waste samples were collected from the Clinical Immunology Laboratory protein electrophoresis assay. Ammonium bicarbonate, dithiothreitol (DTT) and formic acid were purchased from Sigma– Aldrich (St. Louis, MO). Water and acetonitrile were purchased from Honeywell Burdick and Jackson (Muskegon, MI). Capture Select™ Affinity Matrices were purchased from Life technologies. Melon Gel was purchased from Thermo-fisher Scientific, Waltham, MA. 3.2. Sample preparation 20 lL of patient serum was added to 200 lL of Melon Gel (Thermo-fisher Scientific, Waltham, MA) and mixed for 5 min on
an orbital shaker. The resin was then centrifuged for 1 min at 5000 RPM. 20 lL of supernatant was then mixed with 20 lL of 50 mM ammonium bicarbonate (pH 8.0) and 10 lL of 200 mM DTT. This was then incubated at 55C for 30 min prior to injection. Alternatively, 4 lL of patient serum was diluted with 200 lL of PBS. Then 30 lL of a 1:1 mix of Capture Select™ LC-Kappa Affinity Matrix and Capture Select™ LC-Lambda Affinity Matrix as a 50% slurry in PBS was added. The samples were mixed for 45 min on an orbital shaker at room temperature. Beads were then washed 2 with PBS and 1 with HPLC grade water and then resuspended in a solution of 5% acetic acid with 50 mM TCEP (PierceÒ Biotechnology, Rockford, IL). The samples were then incubated at RT for 15 min, centrifuged at 5000 rpm for 1 min and injected into the mass spectrometer. For dilutions studies, serum containing M-proteins was diluted with pooled normal human serum (Fisher BioReagents). 3.3. Liquid chromatography An Eksigent Ekspert 200 microLC system (Redwood City, CA) coupled with a Poroshell 300SB-C3 analytical column from Agilent (1.0 75 mm Poroshell 300SB-C3; 5 lM particle size) was used for separation. A 2 lL sample injection was used with a 25 lL/min flow rate. The total run time is 24 min. The column temperature was maintained at 60 °C using a column heater. The aqueous solvent (Mobile phase A) consisted of water with 0.1% formic acid, and the organic phase (Mobile phase B) was 90% acetonitrile, 10% 2-propanol and 0.1% formic acid. A gradient was used to elute light chains starting at 80% A/20% B, held for 0.5 min, ramped down to 70% A/30% B over 1 min, then ramped down to 60% A/40% B over 4 min, further ramped down to 5% A/95% B over 5 min, held for 2.5 min and then ramped up to 80% A/20% B for 1 min and the re-equilibrated at 80% A/20% B for 1 min. 3.4. Mass spectrometry An ABSciex TripleTOFÒ 5600 quadrupole time-of-flight mass spectrometer (ABSciex, Vaughan, ON, Canada) operated in ESI positive mode with a Turbo V dual-ion source with an automated calibrant delivery system (CDS) was used to acquire mass spectra. Source conditions were ISVF, 5500; temp, 50; CUR, 45; GS1, 35; GS2, 30. TOF MS scans were acquired from m/z 600–2500 with an acquisition time of 200 ms. The instrument was calibrated every five injections through the CDS using the calibration solution supplied by the manufacturer. 3.5. Data analysis Analyst TF v1.6 was used for instrument control. Data were viewed using Analyst TF v1.6 and PeakView v1.2.0.3. BioAnalyst software provided with Analyst TF was used to reconstruct the multiply charged ion peaks into a mass graph. Deconvolution
Fig. 3. Normal polyclonal LC distribution. Multiply charged LC ions and the corresponding deconvoluted mass spectra demonstrating a representative polyclonal LC mass distribution found for healthy adults. The spectrum shows multiply charged ion pairs labeled with their corresponding charge states. Each pair of unresolved multiply charged ions has a normal distribution and is spaced according to their m/z ratio.
J.R. Mills et al. / Methods 81 (2015) 56–65
61
Fig. 4. miRAMM demonstrates a linear dynamic range. (A) The multiply charged LC ions of rituximab and the corresponding deconvoluted mass spectra demonstrating a representative of rituximab’s LC mass. (B) Rituximab was spiked into pooled human serum at 6 different concentrations spanning 0.0005 g/dL to 0.025 g/dL. The peak area corresponding to rituximab was measured using the deconvoluted mass spectra. Using linear regression analysis, the slope and intercept were calculated and use to calculate concentrations. (C) Representative deconvoluted mass spectra used to generate the plot in (B) are shown.
was performed between 22,000 Da and 26,000 Da as a protonated ion with a step size of 1 Da, a S/N threshold of 10 and 10 iterations for molecular mass calculations. 3.6. Protein electrophoresis and immunofixation Assays were performed according to protocols in the Clinical Immunology Laboratory [4], Department of Laboratory Medicine and Pathology, Mayo Clinic: IFE was performed using Sebia 9IF gels (Sebia, Norcross GA). 4. Results and discussion Initial work developing miRAMM documented the ability of this technique to quantitate a monoclonal therapy, adalimumab, in the presence of a normal concentrations of polyclonal Igs [38]. Serial dilutions into normal human serum demonstrated as little as 0.005 g/dL of adalimumab was detectable above the polyclonal distribution of Ig found in healthy adult patients. Linearity was established from 0.005 to 5 g/dL. Subsequently we have performed similar experiments to establish if these observations would hold true for additional monoclonal therapies. Rituximab (an anti-CD20 monoclonal IgG1 Kappa clone) was spiked into pooled human serum ranging from 0.0005 to 0.025 g/dL. Analysis of rituximab using miRAMM generated a charge state distribution of ions consistent with previous reports and the converted molecular mass for pooled serum spiked with 0.025 g/dL rituximab displayed a single peak with the observed molecular mass of 23,035 Da (Fig. 4A). This molecular mass agrees with the average molecular mass calculated from the known amino acid sequence for the kappa LC of rituximab (23,035.4 Da) containing a pyroglutamic acid modification at the N-terminus as well as maintaining its LC intrachain disulfide bonds. Rituximab was identifiable above the polyclonal background at concentrations as low as 0.0005 g/dL (Fig. 4C) and demonstrated a linear
response across the range of concentrations (Fig. 4B). These results demonstrated miRAMM can detect monoclonal Ig below the limit of detection of PEL (0.05 g/dL) and IFE (0.005 g/dL) [39]. To demonstrate the improved functional sensitivity of miRAMM, a human serum sample containing an IgA kappa (IgAK) M-protein of 2 g/dL was run by IFE and the same sample was analyzed by miRAMM (Fig. 5). A distinct immunoprecipitated band present in the A (IgA anti-sera) and K (j anti-sera) lanes indicate an M-protein of the IgAK isotype. The same sample analyzed by miRAMM demonstrates the presence of a monoclonal LC with an accurate molecular mass of 23,334 Da. The asterisk indicates the presence of a sulfate adduct with a mass of 23,430 Da (23,334 Da + 96 Da). The same sample was then diluted 1:1000 into pooled serum from healthy individuals and re-analyzed by IFE and miRAMM. In the case of IFE the M-protein is no longer detectable, however, by miRAMM we can still detect a monoclonal LC at 23,333 Da which is within 1 Da of the monoclonal LC in the undiluted sample (Fig. 5B). Similarly to PEL and IFE, the limit of detection of a given monoclonal LC by miRAMM is partly determined by the physical properties of the LC. In the case of PEL and IFE, the migration of the LC dictates its limit of detection. A LC that co-migrates with other proteins during PEL charge separation will be more difficult to detect due to interference from co-migrating protein. On the other hand, if the LC has a distinct migration pattern such that the LC migrates well ahead of or behind the polyclonal LC population where other proteins do not migrate the ability to detect that LC is increased. A similar phenomenon exists for miRAMM. Given the polyclonal background has a Gaussian distribution of masses, if the mass of the LC is found at the peak of the distribution then the sensitivity maybe reduced relative to LC with a mass corresponding to the tails of the Gaussian distribution, however, miRAMM has the advantage of two dimensional separation in that the both retention times and mass/charge are used to define to monoclonal LC thus providing miRAMM with additional sensitivity compared to
62
J.R. Mills et al. / Methods 81 (2015) 56–65
Fig. 5. miRAMM can offer superior sensitivity to detect M-proteins compared with IFE. IFE results from the analysis of a serum sample from a patient with multiple myeloma. (A) IFE results from the analysis of a serum sample from a patient with multiple myeloma before and after a 1:1000 dilution into pool human serum. A distinct IgAK M-Protein is present in the ‘‘A’’ lane (IgA HC) lane and the ‘‘K’’ (kappa LC) lane is found in the neat sample. However, in the diluted sample the IFE is read as negative without a distinct banding pattern indicating an M-protein. (B) The corresponding deconvoluted mass spectrum showing predominant monoclonal LC with a molecular mass of 23,334 Da, which is within the expected mass range for a LC. The * denotes a likely salt adduct with a mass shift of 96 Da – consistent with the mass of a sulfate ion. The k and j symbols designate the polyclonal lambda and kappa LC distributions, respectively. In the diluted sample the M-Protein is still detectable although at 1000-fold lower intensity.
Fig. 6. miRAMM can differentiate monoclonal therapies from endogenous M-proteins. Serum samples were collected before and after treatment from a patient with a MGUS receiving rituximab therapy for an autoimmune condition. The top panel represents the multiply charged LC ions and the corresponding deconvoluted mass spectra demonstrating a single monoclonal LC with a retention window of 5 min and a mass of 23,614 Da. The bottom panel is serum from the patient 2 days later after receiving an infusion with rituximab. The appearance of a second monoclonal LC has a mass of 23,035 Da which is identical to that of rituximab and had a retention time of 5.3 min consistent with rituximab.
IFE/PEL. In addition, k and j have different mass distributions which are distinct from each other. Lastly, miRAMM has superior resolution (see Fig. 6). Initial work using patient samples demonstrated that accurate molecular mass of light chains is a viable approach to monitor patients over years. Using serial patient samples it was demonstrated that the accurate molecular mass of the disease associated monoclonal light chains deviated <1 Da over a 7-year period which is within the expected mass measurement error of the mass
spectrometer. Importantly, residual disease was detectable even in periods of remission [38]. A follow up study looked for monoclonal light chains in urine samples with detectable monoclonal proteins by IFE. miRAMM identified 112 of 115 urine samples as positive – demonstrating a high level of concordance with current clinical laboratory practice but with negligible reagent costs and minimal sample processing [37]. miRAMM was only tested on neat urine samples without concentrating the sample. Interestingly, the three samples missed by miRAMM all were exceptionally dilute
J.R. Mills et al. / Methods 81 (2015) 56–65
63
Fig. 7. CaptureSelect enrichment improves recovery of IgA Igs. Representative serum samples positive for an IgG and IgA M-protein were analyzed by miRAMM using MelonGel™ enrichment (upper panel) or CaptureSelect enrichment (bottom panel). Both the multiply charged LC ions and the corresponding deconvoluted mass spectra demonstrating a representative the accurate molecular mass of the M-protein LC are shown.
and required concentration prior to analysis by PEL/IFE. It is possible concentration of urine prior to miRAMM will be required in order to assure maximum sensitivity. In most serum samples analyzed by miRAMM, two distinct Gaussian-like distributions of light chains are present (k and j respectively in (Fig. 3) [38]. The two distinct LC polyclonal populations are postulated to be due to different molecular masses of the lambda and kappa constant region as well as unique framework 1 (FR1) usage [41]. The Gaussian distribution for each unique LC population is likely representative of the randomness of VJ recombination within the VL domain. Comparison of the relative abundance of the polyclonal kappa and lambda populations as measured by miRAMM compared well to the total LC ratios (kappa:lambda) measured by a clinical nephelometry assay highlighting another dimension of miRAMM [41]. Given the accurate molecular mass accuracy of miRAMM, it can be useful in ruling out the presence of monoclonal therapeutics which may be detected by IFE or PEL in patients being monitored for the presence of MGUS [42]. This is likely to become more prevalent and problematic in myeloma patients as high dose monoclonal therapeutics become part of the treatment regime [43]. To demonstrate this, serum samples from a patient with an IgGK (IgG HC and kappa LC) MGUS and an autoimmune condition taken before and after infusion with rituximab therapy (a monoclonal IgGK anti-CD20 therapy) was measured using miRAMM. The rituximab LC has an anticipated mass of 23,035 Da. miRAMM is useful in determining that the appearance of a second monoclonal LC is due to the presence of rituximab, identifiable by its accurate molecular mass, rather than the appearance of a new M-protein. Although only a limited number of monoclonal therapies have been evaluated by miRAMM, thus far each has a unique combination of liquid chromatography retention times and molecular masses – making this an acceptable approach to eliminating the
potential confusion monoclonal therapies pose. A more comprehensive analysis using miRAMM of the commonly prescribed monoclonal therapies is warranted. The initial design of miRAMM utilized Melon-Gel™ to enrich for Igs. Melon-Gel™ has lower recovery for M-proteins of the IgA and IgM class – thus subsets of M-proteins have demonstrated limited sensitivity. In attempt to improve upon this we have used an alternative mode of Ig enrichment using a mixture of CaptureSelect anti-j/k beads to affinity purify Igs prior to analysis. CaptureSelect affinity reagents have the benefit of binding LCs regardless of the corresponding HC. Serum samples positive for either IgG or IgA M-proteins by PEL (20 g/L monoclonal Igs) were analyzed by miRAMM using Melon-Gel™ and CaptureSelect beads simultaneously (Fig. 7A and B). Use of CaptureSelect beads enhances recoveries for IgA clones. This bias does not exist for IgG clones as both enrichment methods yield similar signal intensities. This new enrichment scheme has demonstrated the ability to isolate all Igs regardless of HC type over a wide variety of sample types (data not shown). 4.1. Current limitations One current limitation of miRAMM is the inability to detect monoclonal proteins at concentrations lower than that of the polyclonal Igs in at the same mass range and retention time. Given the vast number of distinct Igs present in serum (estimated at >106 clones/individual) and the limited ranges of molecular masses observed, in order to identify a clone with certainty the intensity of the monoclonal LC needs to be present above the polyclonal background. One method to improve sensitivity is to use bottom-up proteomics rather than using the intact mass of light chains. A commonly used protocol begins with denaturation of
64
J.R. Mills et al. / Methods 81 (2015) 56–65
the protein, reduction of disulfide bonds, alkylation to present reformation of disulfide bonds and then digestion typically with endopeptidases such as trypsin, chymotrypsin, Lys-C or Glu-C. The digestion is terminated and then the peptide mixture is separated using on-line using chromatographic methods and then analyzed by ESI-MS or MALDI-MS. Rather than reconstructing the peptide sequence of the entire Ig, a clonotypic peptide which is unique to a specific Ig (either the HC, LC or both) can be found using LC–MS/MS and used as a surrogate marker. This method has been clinically applied to quantitating IgG subclasses as well as for qualitatively monitoring residual myeloma [14,21,44]. Internal standards, typically stable isotope labeled peptides with the identical sequence of interest are used or by using an isotope-labeled Ig standard [45,46] [47]. Our group has also demonstrated that a surrogate Ig from another species can serve to normalize for variation in digestion efficiency [21]. In order for a proteotypic peptide to serve as a unique surrogate marker of the intact monoclonal protein it must be derived from the variable region of Ig – which assumes that the appropriate amino acids residues are available to generate appropriate digestion sites. It is also assumed that the peptide of interest cannot be present in other Igs. Although using peptides offers superior sensitivity the process is more expensive, more laborious, and more time consuming compared with miRAMM and more challenging in the absence of prior knowledge of the Ig sequence. miRAMM offers a work flow more conducive to the clinical laboratory – minimal reagents, minimal pre-analytical processing, and the ability generate a snapshot the entire immunoglobulin repertoire – all at a fraction of the cost of tests designed to provide the same information but employing other platforms. miRAMM was initially designed as to tool for detecting monoclonal LCs due to their small molecular mass and the absence of post-translational modifications as compared to the HC. In its current format miRAMM can resolve monoclonal IgG HCs [38]. However, it has not been thoroughly tested to determine if it is sufficient to provide HC isotype information for IgM and IgA heavy chains which may be difficult due to the heterogeneity in the carbohydrate on the Fc portion of these isotypes. Furthermore, the use of a reducing agent as currently employed by miRAMM means that the LC spectrum corresponds to both FLCs as well as LCs associated with a HC. Therefore, in its current format, miRAMM will not provide the same information which is obtained by using the FLC assay. However, it should be noted that this technology could provide isotype information as well as detection of HCs once the method is refined to include either top-down fragmentation information or utilization of alternative Ig purification methods that would be selective towards HC and/or FLCs (for instance purification with resin specific to FLCs). Until this is accomplished miRAMM best fills a clinical role in improving specificity of current detection methods using accurate mass measurements as well as providing a way to detect lower levels of monoclonal LCs. In order to provide the same information as PEL/IFE modifications to the current methodology are required.
5. Conclusion The current methodologies used to screen for monoclonal proteins in the clinical laboratory are based on PEL/IFE and subsequent evaluation of the pattern of protein banding to determine if an M-protein is present. These methods have proven robust but suffer from limited sensitivity, making it impossible to rule out low amounts of residual disease. Furthermore, interfering proteins may be present making interpretation challenging. Given that clonal plasma cells responsible for monoclonal gammopathies secrete Igs with unique amino acid sequences accurate molecular mass can
be with high precision and accuracy creating added value in specificity in serial samples from a patient and when monoclonal therapeutics are present. Furthermore, the sensitivity to detect a monoclonal Ig above the polyclonal Ig background is significantly improved compared to PEL and IFE. Identification of specific proteins using accurate molecular mass has already been implemented into clinical practice. It is routine clinical practice to refer back to a patient’s past history of PEL and IFE results when interpreting subsequent samples in an aim to provide consistent interpretation and to improve accuracy of resulting. One can easily envisage that in the future, part of the patient’s clinical record will include documentation of a patient’s monoclonal Igs with an accurate mass measurement which can serve as a reference for future analysis. The studies and results discussed here provide evidence to assert the utility of mass spectrometry as a tool to monitor an M-protein in patients with a monoclonal gammopathy. By having a more sensitive and specific MS serum based assay to detect residual M-proteins this may be amendable to replacing costly and invasive methodologies currently used to establish the absence of minimal residual disease. miRAMM represents a major shift in the way M-proteins are interpreted and will likely transform the way plasma cell disorders will be tracked in the future. Lastly, monitoring of M-proteins is not the only use of gel electrophoresis to monitor Ig banding as a part of the diagnostic work up. It is likely miRAMM may find a role as a technological improvement over other methodologies as well. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]
[36] [37] [38] [39] [41] [42]
K. Rajewsky, Nature 381 (6585) (1996) 751–758. C. Vettermann, M.S. Schlissel, Immunol. Rev. 237 (1) (2010) 22–42. M.P. Lefranc, Cold Spring Harbor Protoc. 2011 (6) (2011) 595–603. J.A. Katzmann et al., Clin. Chem. 55 (8) (2009) 1517–1522. J.A. Katzmann et al., Clin. Chem. (2014). C.R. McCudden et al., Am. J. Clin. Pathol. 129 (3) (2008) 451–458. C.E. Bystrom, S. Sheng, N.J. Clarke, Anal. Chem. 83 (23) (2011) 9005–9010. N.J. Clarke, A.N. Hoofnagle, Clin. Lab. Med. 31 (3) (2011) (p. ix–xi). A.E. Acosta-Martin et al., Anal. Chem. 85 (16) (2013) 7971–7975. H.R. Bergen 3rd., S.R. Zeldenrust, S. Naylor, Amyloid 10 (3) (2003) 190–197. P.V. Bondarenko et al., J. Am. Soc. Mass Spectrom. 20 (8) (2009) 1415–1424. S. Rosati et al., Nat. Protoc. 9 (4) (2014) 967–976. D. Murray, D. Barnidge, Crit. Rev. Clin. Lab. Sci. 50 (4–5) (2013) 91–102. D.R. Barnidge et al., J. Proteome Res. 13 (4) (2014) 1905–1910. D.R. Barnidge et al., J. Proteome Res. (2014). C.M. Botz et al., Br. J. Haematol. (2014). S.J. Gaskell, J. Mass Spectrom. 32 (7) (1997) 677–688. A. Tartakoff, P. Vassalli, J. Cell Biol. 83 (2 Pt 1) (1979) 284–299. P. Gagnon, J. Chromatogr. A 1221 (2012) 57–70. A. Beck et al., Nat. Rev. Immunol. 10 (5) (2010) 345–352. P.M. Ladwig et al., Clin. Chem. 60 (8) (2014) 1080–1088. S.K. Chowdhury et al., J. Am. Soc. Mass Spectrom. 1 (5) (1990) 382–388. J. Porath, F. Maisano, M. Belew, FEBS Lett. 185 (2) (1985) 306–310. T.W. Hutchens, J. Porath, Anal. Biochem. 159 (1) (1986) 217–226. E. Boschetti, J. Biochem. Biophys. Methods 49 (1–3) (2001) 361–389. M. Wilchek, I. Chaiken, in: P. Bailon et al. (Eds.), Affinity Chromatography, Humana Press, 2000, pp. 1–6. B. Akerstrom, L. Bjorck, J. Biol. Chem. 264 (33) (1989) 19740–19746. A. Staub et al., J. Pharm. Biomed. Anal. 55 (4) (2011) 810–822. D.S. Rehder et al., J. Chromatogr. A 1102 (1–2) (2006) 164–175. T.M. Dillon, P.V. Bondarenko, M. Speed, J. Chromatogr. A 1053 (1–2) (2004) 299–305. J.B. Fenn et al., Science 246 (4926) (1989) 64–71. M. Karas, F. Hillenkamp, Anal. Chem. 60 (20) (1988) 2299–2301. H.S. Gadgil et al., J. Am. Soc. Mass Spectrom. 17 (6) (2006) 867–872. L. Yu, R.L. Remmele Jr., B. He, Rapid Commun. Mass Spectrom. 20 (24) (2006) 3674–3680. E. Johansen, J. Albanese, C. Hunter, Analysis of intact and reduced therapeutic monoclonal antibodies using the TripleTOF™ 5600 system, 2011. Available from:
J.R. Mills et al. / Methods 81 (2015) 56–65 [43] J.P. Laubach et al., Expert Opin. Investig. Drugs 23 (4) (2014) 445–452. [44] E.R. Remily-Wood et al., Proteomics Clin. Appl. 8 (9–10) (2014) 783–795. [45] D.R. Barnidge et al., J. Proteome Res. 3 (3) (2004) 644–652.
[46] J.R. Barr et al., Clin. Chem. 42 (10) (1996) 1676–1682. [47] O. Heudi et al., Anal. Chem. 80 (11) (2008) 4200–4207.
65